- Email: [email protected]
A 27Al MAS, MQMAS and off-resonance nutation NMR study of aluminium containing silica-based sol-gel materials
SOLID STATE Nuclear Magnetic Resonance
ELSEVIER
Solid State Nuclear Magnetic Resonance
9 (1997) 203-217
A 27A1MAS, MQMAS and off-resonance nutation NMR study of aluminium containing silica-based sol-gel materials ’ M.P.J. Peeters a,2, A.P.M. Kentgens b3* b N.S.R.
a Philips CFT, Pro5 Holstlaan 4, 5656AA Eindhouen, Netherlands Centerfor Molecular Design, Synthesis and Structure, National HF NMR Facility, University of Nijmegen, 6525ED Nijmegen, Netherlands Received 6 December
1996; accepted 4 February 1997
Abstract Aluminium containing hybrid materials were prepared via the sol-gel method using aluminium set-butoxide complexed with ethylacetoacetate (Al(OBti”),EAA or Al(OBu”),/EAA mixtures). As silanes, phenyltrimethoxysilane (PhTMS) or phenyltriethoxysilane (PhTES), 3-glycidoxypropyl trimethoxysilane (Glymo) and tetraethylorthosilicate (TEOS) were used. After room temperature drying of the samples the 27A1single pulse excitation (SPE) magic angle spinning (MAS) NMR shows that octahedral (5 ppm) and tetrahedral (55 ppm) coordinated aluminium species are present in the materials. The relative amount of these two species depends on the preparation method. However, the Al(IV)/Al(VI) ratio is lower than 3 (typically 2.3) in all materials, indicating the presence of a small amount of an aluminate phase. Annealing of the samples at 100, 150 and 200°C results in the formation of an extra signal at 30 ppm (peak maximum measured at 11.7 T). Based on the resonance frequency this signal is generally assigned to a pentahedrally coordinated aluminium species. Hydration/dehydration processes of annealed samples were studied with *‘Al SPE MAS NMR, multiple-quantum MAS NMR (MQMAS) and off-resonance nutation NMR. Upon hydration of the annealed sample the signal intensity around 30 ppm decreases in intensity and at the same time the intensity of the signal around 55 ppm increases by the same amount (tetrahedrally coordinated aluminium). The MQMAS spectra reveal that the signal around 30 ppm is not caused by a fivefold-coordinated aluminium species but mainly by tetrahedrally coordinated aluminium species in a distorted environment, experiencing large quadrupole induced shifts and small chemical shifts due to conformational changes in the polymeric network. From the MQMAS NMR spectra it can be concluded that the linebroadening observed in the 27A1MAS NMR spectra is due to both a distribution in isotropic chemical shifts and a distribution in quadrupole coupling constants (C,,, = e*qQ/h). Hydration of the sample results in a decrease of the average Cs_ for the tetrahedrally coordinated aluminium from 6 to 4 MHz, whereas the average C,,, of the octahedrally coordinated aluminium is hardly influenced (4 MHz). These MQMAS results are confirmed by off-resonance nutation experiments. 0 1997 Elsevier Science B.V. Keywords: Solid-state NMR; MQMAS; Off-resonance nutation; Combined inorganic-organic
* Corresponding author. E-mail: [email protected]. ’ Dedicated to Giinter Engelhardt. ’ E-mail: [email protected]. 0926-2040/97/$17.00 0 1997 Elsevier Science B.V. All rights reserved PII SO926-2040(97)00060-X
materials;
Al-coordination
204
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
1. Introduction Aluminosilicate gels are used in catalysis, optics and as multicomponent precursors for the preparation of ceramics (e.g., Mullite, Corderiete) [l-6]. Preparation of these materials via the sol-gel process opens the possibility to synthesise (on a molecular scale) homogeneous glass-like materials via a low temperature process. Aluminosilicates have, compared to pure silicates, an increased thermal and chemical stability, a lower thermal expansion coefficient, and an increased mechanical toughness. A great number of chemical precursors can be used for the sol-gel preparation of aluminosilicates alkoxides, oxides, salts). Often (e.g. Al(OBu”),/EAA mixtures or Al(OBu”),EAA (complex, no free set-butanol) are used (EAA = Ethylacetoacetate, C,H,,O,). The EAA (P-diketone) is added to the aluminium alkoxide to reduce the reactivity of aluminium alkoxides towards water
and to prevent immediate precipitation [7]. Complexation of aluminiumalkoxide with EAA is a consequence of the fact that the oxidation state (Z) of the metal is lower than the coordination number (Nl. In that case coordination expansion by oligomerisation and/or charge transfer complex formation is preferred 171. Several structures (Fig. 1) have been proposed for mono P-diketone complexed aluminium alkoxides [8], depending on the size of the alkoxide group and/or the temperature. Structure (4) for example was confirmed by a single crystal study of Al(OPr’),(AcAc). Very bulky alkoxy groups lead to the monomeric structure (1) [9]. The 27A1 NMR spectrum of Al(OBu”),EAA shows the presence of 4-5 and 6-coordinated aluminium. This indicates that in Al(OBu”),EAA several species are present, for example a mixture of (11, (2) and (3) or of (31 and (4) (Fig. 1). Several papers deal with the characterisation of aluminium containing materials prepared via the ,’
‘\
(1) OR
0
OR
0--_. /
b R
Fig. 1. Possible structure indicated in parenthesis.
of ethylacetoacetate
(EAA)
complexed
aluminium
alkoxides
[S]; the coordination
(IV)
,x _---OR
number
of aluminium
is
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
sol-gel method using Al(OBu”),EAA [7-121. Babonneau et al. [8] used Al(OBu”),EAA for the preparation of a-cordierite (Mg:Al:Si:H,O:EtOH = 2:4:5:32:52). IR measurements revealed that when Al(OBu”),EAA reacts with D,O, hydrolysis of butoxy groups occurred (strong decrease of the band at 1060 cm- ’ >. A small amount of free EAA could be detected in the sol (bands at 1710 and 1740 cm-’ > although the characteristic bands of EAA bonded to aluminium (1630, 1610, 152.5, 1420, 1370, 1300, 1175, 630 and 510 cm-‘) remained intense. A broad band at 400 to 700 cm-’ indicated the formation of an Al-O-Al network [8]. In a second publication Bonhomme-Coury et al. [ll] studied the a-cordierite preparation with 29Si and *‘Al liquid and solid-state NMR. Hydrolysis of the precursor solution resulted in the appearance of a sharp signal at 50.2 ppm in the 27A1 NMR spectra (symmetric tetrahedral site, intensity 7%) and a asymmetric signal at 55 ppm (distorted tetrahedral sites, intensity 58%), indicating tetrahedral coordination of 65% of the aluminium. The amount of tetrahedrally coordinated aluminium increased with reaction time. It was proposed that the Al-OBuS bond in Al(OBu”),EAA hydrolyses first and that the partly hydrolysed aluminium is incorporated in the silicon network via the following reaction: Al-OH + Si( OEt), + Al-O&(
OEt), + EtOH
( 1)
The aluminium bonded to silicon is expected to be in a fourfold-coordination (tetrahedral), which explains the increase of the amount of tetrahedrally coordinated aluminium with time. These results are in line with a publication of Heimich et al. [5]. They studied the preparation of Mullite gels (stoichiometry: 3 Al,O, * 2 Si$+> starting from TEOS and Al(OBu”),EAA. The Al NMR spectra showed narrow lines at 51 and 58 ppm. These are assigned to tetrahedrally coordinated aluminium atoms with silicon in the second coordination sphere. The average coordination number of aluminium in the sol increased with increasing reaction time. Since each aluminium introduced into the silica network introduces a charge deficiency, part of the aluminium is expected to compensate for this. Charge compensation can be obtained by Al(H,O)~‘, with aluminium in an octahedral coordination. Irwin et al. [13] have shown that upon addition of sodium to the
205
sol (in the form of NalAl(OR),]) for charge compensation, the amount of octahedrally coordinated (nonnetwork) aluminium decreased or disappeared from the samples after drying at 40°C. This suggests that fourfold-coordinated aluminium is incorporated in the silicon network, while octahedral aluminium is necessary for charge compensation. Drying of the aluminosilicate gels at moderate temperatures ( < 200°C) results in the formation of a new signal at approximately 30 ppm in the *‘Al MAS NMR spectra [ 13-151. The chemical shift of this new resonance is indicative of five-coordinated aluminium [14,16]. Since the first observations of fivefold-coordinated aluminium by Dupree et al. in 1985 [ 161, fivefold-coordinated aluminium species have been observed in aluminas, zeolites [17,18], and aluminium containing ceramic precursors [19]. In many of these materials fivefold-coordinated aluminium is formed after a severe chemical or thermal treatment. This can lead to some very distorted sites, e.g., in the case of dehydroxylation of Pyrophillite [Al,Si,O,,(OH),] at 800°C [20] a C,,, of 10.5 MHz was found [21]. It is not generally true, however, that fivefold-coordinated aluminium has the largest quadrupole coupling constant, e.g., in X-Al,O, the fivefold-coordinated aluminium appeared to have the smallest C,,, 1221. Ray and Samoson [17] found a relatively small C,,, for fivefold-coordinated aluminium in steamed zeolites, compared to a very distorted tetrahedrally coordinated aluminium species they encountered in the same material. The presence of fivefold-coordinated aluminium in aluminosilicates has also been reported. Hatakeyama and Maekawa [15] report that the intensity of the signal at 30 ppm increases in intensity with temperature in their aluminosilicate samples. Hydration of the samples (autoclave, 50 h, 70°C) lead to a decrease of the 30 ppm signal. Fivefold-coordinated aluminium is said to be formed by dehydration of the non-network octahedrally coordinated aluminium. This process is reversed upon hydration. The fivefold-coordinated aluminium disappears upon crystallisation of the A1,03-SiO, gels. De Witte et al. 1141 studied aluminium-rich amorphous aluminosilicates treated at low temperatures. They also assign the resonance observed at 30 ppm in the *‘Al MAS NMR spectra to fivefold-coordinated aluminium. The aluminium in a fivefold-coordination is
M.P.J. Peeters, A.P.M. Kentgens/Solid
206
State Nuclear Magnetic Resonance 9 (1997) 203-217
an intermediate, metastable species, typically disappearing from the spectra when the sample is treated at temperatures exceeding 980°C and crystallisation has occurred. Calcination of the amorphous aluminosilicate sample at 400°C leads to a strong increase of the 30 ppm signal at the expense of tetrahedrally coordinated aluminium. Synthesis of the materials in alkaline media results in a decrease of the amount of octahedral and fivefold-coordinated aluminium. The aluminium containing silica-based sol-gel materials used in this study were prepared using the Al(OBu”),EAA or Al(OBu”), components as the aluminium containing precursor. The resulting dried hybrid materials were studied with “Al NMR. The annealing temperature and the degree of hydration were varied. *‘Al MAS NMR, 27A1 MQMAS NMR and off-resonance nutation NMR were performed to characterise the different aluminium species present in the dried materials.
2. Experimental
2.1. Sample preparation The samples were prepared by mixing tetraethylorthosilicate (TEOS) with a trifunctional alkoxysilane (T). To this mixture (absolute) ethanol was added followed by a 1: 1 (W/W> acidic water/ethanol mixture (HCl, [Hf] = 5 * 10-3). The trifunctional siloxane (T) was either PhTMS (Phenyltrimethoxysilane), PhTES (Phenyltriethoxysilane) or Glymo (3-glycidoxypropyl-trimethoxysilane). The last step was the addition of the aluminium set-butoxide compound, either Al(OBu”),EAA or Al(OBu”), complexed with EAA (ethylacetoacetate). TEOS was obtained from Merck, the trifunctional siloxanes were obtained from Hiils, the Al(OBu”), and A1(OBuS),EAA from Gelest. All reagents were reagent grade and used without further purification. The general composition of the samples was: xT,yTEOS,zA1(OBu”),EAA(orAl(OBu’)3), 4EtOH,(4 x+y+z=
-x
- z)H,O
1
Hydrolysis with water was performed for 2 h at 50°C in closed vessels. After 2 h the solvent was allowed to evaporate at room temperature. For an-
nealing an additional heat treatment was applied in a conventional oven for 2 h at a temperature of 100, 150 and 200°C respectively.
2.2. 2?41 NMR “Al NMR measurements were performed on a Bruker AM-500 operating at 130.32 MHz. Short radio frequency pulses @-field of 25 kHz, tip angle < 25” for aluminium species with a large C,o> and a recycle delay of 2 seconds were used to obtain quantitative results. A MAS speed of N 13.0 kHz was used. No proton decoupling was applied. Unless otherwise indicated, the chemical shift of the resonances is characterised by the peak maximum measured at 11.7 T. MQMAS experiments were recorded using the two-pulse sequence [23-261. Experiments were performed using a home-built MAS probe equipped with a Jakobsen stator. The spinning frequency was 13 kHz (Si,N, or zirconia rotors). The rf-field strength was N 65 kHz. A 5 ps preparation and a 2 ,us mixing pulse was used. 128 (t,)*512 (t,) points with a dwell time of 5 pus in both directions were acquired. Phase cycling over four times six phases was employed to select triple-quantum coherence during the evolution period; 3168 scans were acquired per experiment with a recycle delay of 1 s. TPPI was used to obtain pure-absorption mode lineshapes. It should be noted, however, that due to the different signal intensities from the 0 + + 3(t,) -+ - l(t,> and the 0 + - 3(t,) -+ - l(t,) transfer pathways some phase errors remain in the spectra [27]. Off-resonance nutation spectra were recorded under slow MAS conditions (ca. 2 kHz), in order to be able to discriminate between otherwise overlapping peaks. The spinning speed was chosen so low that the pulse duration did not exceed about l/4 of a rotor period. Under these conditions the sample can be considered static during the evolution period of the nutation experiment. A Bruker 7 mm MAS multinuclear resonance probehead was used; the sample was restricted to the centre of the spinner (using Teflon spacers) to ensure good rf-homogeneity. A frequency-stepped adiabatic half-passage was applied as preparation period [28]. 128 Spectra were recorded with 1 ps pulse length increments and resonance offsets be-
M.P.J. Peeters, A.P.M. Kentgens/Solid
207
State Nuclear Magnetic Resonance 9 (1997) 203-217
3. Results
tween 120 and 260 kHz. An exponential filter with a line-broadening of 800 Hz was applied in both directions. Magnitude spectra were obtained whose sum projections along the F,-axis are the spectra shown in the figures.
3.1. 27Al MAS NMR 27A1 MAS NMR spectra were recorded for 0.20 Al(OBu”),EAA/0.80 TEOS and 0.20 Al(OBu”),EAA/0.40 PhTES/0.40 TEOS samples after room temperature drying. A signal for tetrahedrally coordinated aluminium (Al(W), 55 ppm) and octahedrally coordinated aluminium (Al(W), 0 ppm) can be detected. Unless otherwise indicated the chemical shifts mentioned are the peak maxima observed at 11.7 T. Both 27A1resonances are broad and a-symmetric, indicative of a distribution in quadrupo lar coupling constants and chemical shifts. The relative amount of these two species depends on the gel preparation method. In Fig. 2 the percentage of Al(W) en Al(W) in the samples after drying are displayed as a function of the Al(OBu”),EAA addition moment. Prehydrolysis of TEOS, prior to the addition of the aluminium compound, a frequently used technique to get a better match between the hydrolysis rates of the different metal-alkoxide com-
2.3. Calculations All numerical simulations were run on a PC using software written in MATLAB, C + + and Pascal. Powder averages were performed by simultaneously incrementing the Euler angles [29,30]. Off-resonance nutation spectra were calculated by numerical diagonalization of the off-resonance rotating frame Hamiltonian [31]. MQMAS spectra were calculated by simulating the line-shape in the frequency domain. Gaussian distributions in isotropic chemical shift and quadrupole coupling constants were generated and the resulting line shapes added to give the final spectrum. The distributions were calculated from - 3a to + 3a in 5 1 points, where u is the halfwidth at half height.
80 Al( IV)
n n
n
8 h > Y
60
a
0
1
2
3
4
5
6
7
AI(OBU~)~ EAA addition time (hrs) Fig. 2. Aluminium coordination number as a function of the Al(OBuS),EAA addition moment; gel composition 0.01 AI(OBuS),EAA/0.20 Glymo/0.79 TEOS/4 EtOH/3.8 HZO. The two datapoints at t = 0 are from an acid catalysed and a neutral sol. Acid catalysis (faster hydrolysis of the silanes) leads to a higher amount of tetrahedrally coordinated aluminium.
208
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
pounds, leads to a higher amount of tetrahedrally coordinated aluminium. Tetrahedrally coordinated aluminium species are ascribed to network aluminium, whereas octahedrally coordinated aluminium species are needed for charge compensation [13]. As can be seen from Fig. 2, the amount of tetrahedrally coordinated aluminium increases rapidly by lo-15% using a prehydrolysis period of 30 min. Continuation of the prehydrolysis hardly increases the amount of tetrahedrally coordinated aluminium. This indicates that a higher amount of aluminium is incorporated in the silica network if a small prehydrolysis period is used. A “0 NMR study of the hydrolysis mixture shows that during the prehydrolysis period the water concentration of the sol drops to +20% of its initial value after 20-30 min, using TEOS in the presence of 0.005 M HCl. Prolonged hydrolysis leads to condensation of the hydrolysed TEOS molecules. The water concentration remains constant at approximately 20% during this stage. To study the influence of the water concentration (during synthesis) on the aluminium coordination number, sols were prepared with varying water concentrations. The general composition of these sols was 0.20 Glymo/0.60 TEOS/0.20 Al(OBu”),EAA/4 EtOH/w H,O. The value of w was varied between 2 and 8. No prehydrolysis period was applied. Gelation of the sols was observed after +20 min hydrolysis at 50°C for all sols with w > 2. In the case of w = 2 a viscous liquid was obtained after 2 h. The results of the 27Al MAS NMR measurements are displayed in Fig. 3. The amount of tetrahedrally coordinated aluminium decreases with increasing water concentration. These results are in line with the results displayed in Fig. 2 and with the IR study of Babonneau et al. [8]. At a low water concentration (later addition of the Al(OBu”),EAA component), hydrolysis of the Al-OBu” bond is likely to occur, whereas hydrolysis of the Al-EAA complex will be negligible. This controlled hydrolysis leads to monomeric tetrahedrally coordinated aluminium species (27A1 liquid state NMR), which are capable of undergoing condensation reactions, as described in Eq. (1). Charge compensation by octahedrally coordinated Al(H,O)z+ is expected to results in an Al(W) to Al(V1) ratio of 3 [13]. In all materials prepared, however, this ratio is lower than three, a typical
65 .
I
1
2
3
4
5
6
7
8
9
w Fig. 3. Aluminium coordination number as a function of the water concentration (w). Gel composition: 0.20 Glymo/0.20 Al(OBu”),EAA/0.60 TEOS/4 EtOH/ w H,O. Lines represent the result of linear regression.
value being about 2.3. This lower value suggests the presence of a small amount of an alumina-like phase. Another important substituent in the preparation of aluminium containing hybrid materials is ethylacetoacetate (EAA). The influence of EAA on the final aluminium coordination number was investigated by preparation of sols with the following composition: 0.20 Al(OBu”),/0.40 PhTES /0.40 TEOS/3.4 H,0/4 EtOH/c EAA with c varying between 0 and 0.4 (EAA/Al = O-2). Prehydrolysis of the 0.4 TEOS/0.40 PhTES/2 EtOH/3.4 H,O mixture was performed in all cases for 20 min, to prevent the precipitation of the uncomplexed Al(OBu”), (c = 0). After 20 min the Al(OBu”),/c EAA/2 EtOH mixture was added to the prehydrolysed solution. “Al MAS NMR of the dried gels (RT) revealed the presence of 40 + 3% octahedrally coordinated aluminium in all the materials. This percentage of octahedrally coordinated aluminium is comparable to the value obtained for the gels prethe commercially available pared using Al(OBu”),EAA (Fig. 2). The influence of the EAA addition to the sol on the final aluminium coordination number is thus negligible for those samples prepared with a prehydrolysis period. 3.2. Influence
of annealing
temperature
on the 27Al
MAS NMR spectra
After drying at room temperature the samples were given an additional heat treatment for 2 h at
209
M.P.J. Peeters, A.P.M. Kentgens/ Solid State Nuclear Magnetic Resonance 9 (1997) 203-217 I
Strained 4-coordinated
,,
,_?m_-,
,“““““““““““““““““““”
0
lb0
-100
-200
(fwm) Fig. 4. “Al MAS NMR (11.7 Tesla) spectra of 0.20 Al(OBu”),EAA/0.80 TEOS samples after heating at the indicated temperature (MAS speed 13.0 Hz). The signal at 105 ppm represents an impurity in the rotor material.
for 2 h
C
b
a m,
110
90
70
50
Fig. 5. 27Al MAS NMR of 0.20 Al(OBuS),EAA/0.40 dehydrated and (c) rehydrated.
30
10 (wm)
PhTES/0.40
I
-10
I,
-30
TEOS samples
I
-50
I,
I,
-70
/,
I
-9o-sF
after heat treatment
at 200°C. (a) hydrated,
(b)
210
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
temperatures of 100, 150 up to 200°C. Annealing of the samples at temperatures of 100, 150 and 200°C leads to the formation of a signal at 30 ppm (Fig. 4). Given the position in the 27A1MAS NMR spectrum the signal can be ascribed to either a fivefold-coordinated aluminium species or to an aluminium species in a distorted tetrahedral environment, experiencing a large quadrupole interaction leading to a large quadrupole induced shift. Moreover a shoulder at 80 ppm is formed. This shoulder is assigned to tetrahedrally coordinated aluminium. Due to the heat treatment, the amount of octahedrally coordinated aluminium decreased slightly to approximately 30-40%. These changes in the Al spectra upon heat treatment can be attributed to the formation of small alumina particles in the sample containing four-, five- and sixfold-coordinated Al. The amount of octahedral Al in the samples is higher than necessary for chargecompensation of the tetrahedral Al in the aluminosilicate network. Therefore, it is not unlikely that the excess aluminium forms small alumina-particles. It should be noted, however, that for alumina treated at temperatures below 2OO”C, generally only octahedral Al is observed [32]. Quantitative 27A1 MAS NMR using an external reference ( a-A1,03, Corundum) showed that all aluminium atoms can be observed in the NMR experiment, even after heat treatment. 3.3. Influence of hydration /dehydration MAS NMR spectra
Table 1 Integration Al(OBu”),
of the “Al MAS NMR spectra of /0.40 PhTES/0.40 TEOS/c EAA samples
Sample condition
Integration 95-40
ppm
0.20
region 40- 16 ppm
16-( - 40) ppm
c=o After annealing Hydrated Dehydrated (120°C) Rehydrated
34 45 31 40
29 17 30 21
37 38 39 39
c=O.lO Hydrated Dehydrated
43 29
18 28
39 43
c = 0.20 After annealing Hydrated Dehydrated (120°C) Rehydrated
35 47 35 44
27 17 26 22
38 36 39 35
c = 0.30 Hydrated Dehydrated
(120°C)
47 36
13 25
40 39
c = 0.40 Hydrated Dehydrated
(120°C)
40 31
18 25
42 46
( 12O’C)
the
The absolute integral after the various treatments within 5%.
remains constant
on the 2%1
Upon storage of the samples in air, a decrease of the 30 ppm resonance was noted. This was attributed to hydration effects, and therefore a more systematic study of hydration-dehydration effects was undertaken. After annealing of the samples at 200°C the samples were saturated with water (hydrated). The appearance of the aluminium spectra of the annealed materials changes with the hydration state of the materials (Fig. 5). In the dry state (directly after annealing) an intense signal at 30 ppm is present in the material as described above. Addition of water to the sample leads to a strong reduction of the intensity of this signal. At the same time the intensity of the tetrahedral signal at 55 ppm increases proportionally, as shown by integration (Table 1). Dehydration of the samples results in a decrease of the signal at 55 ppm by lo- 12%, whereas the signal at 30 ppm
increases by the same amount. The intensity of the signal at 0 ppm does not change significantly (maximally +4%). The absolute integral, a measure for the number of aluminium atoms observed in the experiments remains constant (+5%), i.e., all the aluminium remains visible. Although integration of the overlapping resonances is difficult, it seems as if upon hydration of the samples the signal at 30 ppm is converted into a signal at 55 ppm. Hydration of the sample is not likely to reduce the aluminium coordination number from five to four, however. Considering this reversible interchange of intensity of the 30 ppm resonance to the 55 ppm resonance upon hydration, it is unlikely that the signal at 30 ppm originates from five-coordinated aluminium species. Ray and Samoson [17] have shown that hydration of (dehydroxylated) aluminas leads to a irreversible reconstruction of the surface favouring AI(W) at the expense of Al(W) and Al(V). In a
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
previous study of aluminas and aluminoborates we also found a not completely reversible conversion of tetra- and penta-coordinated Al to hexa-coordinated aluminium [33]. De Witte et al. [14] observed a strong increase of the Al(IV) and Al(V1) resonance upon hydration in aluminosilicate gels, but hardly any effect on the intensity of the Al(V) resonance. Taking these observations into account, we think that it is more likely that the signal at 30 ppm in the present sample is partly due to tetrahedrally coordinated aluminium atom in a strained environment. There are two possibilities for a tetrahedral aluminium to end up in the 40-30 ppm area. One possibility is that its isotropic chemical shift lies in the normal range of 50-70 ppm, but due to a large quadrupolar interaction the line is shifted strongly. This has been observed in the de-aluminated zeolites by Ray and Samoson [17]. In our samples dehydration probably leads to a closer distance between the network aluminium and the charge compensating alumina particles, which might induce large field gradients. This was also observed in the drying of zeolites, where the cation gets located close to the tetrahedral site. Another possibility for a tetrahedral aluminium to end up in the 40-30 ppm area is that the isotropic chemical shift of the resonance is very low because of a strained conformation of these sites in the aluminosilicate network. A relation between the Al-O-S1 angle and the 27A1 chemical shift has been established 1341. We found that this correlation can be extended for the mineral harkerite which has tetrahedral aluminium sites with an Al-O-S1 bond angle of 176”, resulting in a chemical shift of 44 ppm [35]. For the largest bond angle (180”) an isotropic chemical shift of 42 ppm is expected. In the hybrid aluminosilicates studied here, a large amount of strain in the network can be expected because the Al in the aluminosilicate network has to be close to the charge compensating alumina in the sample. These electrostatic forces in the material are thought to induce a strain in the material leading to Al(IV) with deviant Al-O-% angles. This might also explain the occurrence of the 80 ppm shoulder and the 30 ppm line in the Al spectra upon dehydration. Hydration probably reduces the strain in the material. That hydration results in diminished strain in the samples is also observable by 29Si MAS NMR. Here hydration of the samples prior to measuring the 29Si MAS NMR
211
spectra of annealed samples results in a slightly increased resolution. In order to gain more insight on the nature of the various observed Al-sites, we tried to obtain extra information on the 27A1 NMR resonances with the use of the newly developed multiple-quantum MAS NMR [36] and off-resonance nutation NMR [28]. Off-resonance nutation NMR was preferred above normal Nutation NMR since the quadrupolar interactions in the samples are expected to be high. 3.4. 2%1 MQMASNMR The idea of the MQMAS experiment is to refocus second-order quadrupolar broadenings in MAS NMR spectra by creating echoes of multiple-quantum transitions and the directly observable central transition. After its introduction by Medek et al. [26] and Frydman and Harwood [36], the experiment soon found its way in a number of applications. As the experiment has been discussed extensively in all these papers, we only describe the main spectral features in our representation, which is without shearing of the spectra. We performed 3Q-MAS experiments on 27A1(I = 5/2) nuclei. In these experiments the resonances are situated on the line vi = 3v, in the absence of a quadrupolar interaction. The spectra are therefore plotted in such a way that the line marked ui,, (vi = 3v,) is the diagonal in these 2D spectra. In the presence of a quadrupolar interaction the center of gravity of the resonance is shifted from the diagonal by the quadrupolar induced shift (QIS) in the direction v, = 3/4v2, indicated in the spectra by a solid-dashed line. Furthermore the line is now broadened by the second-order quadrupolar interaction along the direction v, = 19/ 12 v2, indicated by a dashed line in the spectra marked Anisotropy. Fig. 6a displays the 3Q-MAS NMR of an 0.20 Al(OBu”),EAA/0.40 PhTES/0.40 TEOS sample after annealing at 200°C. The improved resolution of the experiment is obvious. The lines around 55, 30 and 0 ppm are now well separated. The line shapes look very similar to those obtained for y-Al,O, in a previous study [37]. The lines run parallel to the isotropic shift axis at the low-field side indicating a distribution in isotropic chemical shift. At the highfield side the lines bend away from the diagonal in
212
M.P.J. Peeters, A.P.M. Kentgens/Solid
u
A
0
State Nuclear Magnetic Resonance 9 (1997) 203-217
B
(5
is0
is0
Anisotropy i
V,,,,,,,,‘~,,,,,,,,,,,,,,,,,,,i
120
100
80
60
40
20
0
-20
120
100
80
60
(PI%
2o
0
Fig. 6. (a) 27A1 3Q-MAS spectrum of a 0.20 Al(OBuS),EAA/0.40 PhTES/O.40 TEOS/4 EtOH/3.4 HCl (0.005 M) gel after heating at 200°C for 2 h (b) 2’Al 3Q-MAS spectrum of a 0.20 Al(OBuS),/0.40 PhTES/0.40 TEOS/0.30 EAA/4 EtOH/3.4 HCI (0.005 M) gel after heating at 200°C for 2 h and subsequent hydration. The line at 105 ppm is due to an Al impurity in the Si,N, rotor.
the quadrupolar induced shift direction, reflecting a distribution in quadrupolar interaction. In order to get an idea of the size of the distributions involved we decided to simulate the line shapes using Gaussian distributions in both isotropic chemical shift and quadrupole coupling constant. We realise that the results of such an analysis should be treated with caution; the excitation of sites with differing quadrupole interactions is non-uniform, and not fully described yet. So far the excitation behaviour has been described assuming a static Hamiltonian during the pulses [26,27]. Considering these results, our r-f-field strength and expected quadrupole coupling constants of several MHz lead to a 3Q-excitation that is fairly inefficient but does not vary greatly with C qcc. A further point of concern is that the shape of the distribution in isotropic shift and quadrupolar interactions need not to be Gaussian. Depending on the structural variations (i.e., variation in bond angles and distances, position of charge balancing cations etc.) in the materials, different distributions of the observable NMR parameters can be expected. Despite these restrictions we calculated MQMAS spectra with varying Gaussian distributions in
isotropic chemical shift and quadrupole coupling constants with the sole purpose to get a rough impression of the ranges that are encountered in these materials. As can be seen in Fig. 7 we were able to get a satisfactory agreement with the experimental spectrum using this approach. Small discrepancies between the experimental and theoretical line shapes are attributed to the above-mentioned shortcomings of the analysis, and/or small phase errors in the experimental spectrum. For the Al(IV) resonance we used an isotropic chemical shift distribution centered at 60 ppm with a half width at half height (HWHH) of 5 ppm, in combination with an (uncorrelated) distribution of 6 + 2 (HWHH) MHZ for C,,,. For Al(V) we found ais = 33 + 3 ppm and C,,, = 4 + 0.7 MHz. The Al(V1) line was simulated using ojsO = 7 + 4 ppm and Cqcc = 4.5 + 1.5 MHz. The asymmetry parameter was taken 1 in all cases. The calculations are not extremely sensitive to variations in 71~ but did show that 774 is large. Considering the relative insensitivity we did not take a variation in 71, into account. The 3Q-MAS spectrum of a hydrated sample (Fig. 6b) shows a diminished intensity of the Al(V)
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
easier seen by looking at the F,-projections (i.e., onto the normal MAS-axis) of the Al(W) resonance in the samples. These projections are shown, for the sample of Fig. 6a dried at 200°C in Fig. 8a for a sample dried at 120°C (Fig. 8b) and the same sample after hydration (Fig. 8~). Clearly, the resonance of Al(W) in the hydrated sample is narrow with most intensity in the range from 60-40 ppm, there is only a weak tail to higher field (lower ppm values). Dehydration strongly broadens the resonance, which is most pronounced for the sample annealed at 200°C where the line has a significant intensity down to 0
-20 0 20
40
/ Anisotropy
60
,
80 100
// 1 20
I
100
I
I
80
60
40
20
213
I
I
0
-20
Fig. 7. Simulation of the *‘Al 3Q-MAS spectrum of a 0.20 AI(OBuS),EAA/0.40 PhTES/0.40 TEOS/4 EtOH/3.4 HCl (0.005 M) gel after heating at 200°C. Parameters: Al(W): a,,, centered at 60 + 5 ppm (half width at half height), in combination with an (uncorrelated) distribution of 6+2 (HWHH) MHz for C,,,. AI(V): rrlsO = 33 * 3 ppm and Cscc = 450.7 MHz. AI( q so = 7 + 4 ppm and Cqcc = 4.5 + 1.5 MHz. The asymmetry parameter was taken 1 in all cases.
resonance. In this particular case of a sample with a small initial amount of Al(V), the resonance vanishes, but in other samples with higher initial Al(V) signal, the resonance remains present, without showing great variation of the isotropic shift and quadrupole parameters as compared to the dehydrated sample. A superficial inspection of the Al(W) and Al(N) resonances in the hydrated and the dehydrated sample indicate no great changes as the contours span the same region. For the Al(W) resonance this is indeed true, only the high-field tail of the resonance seems a little bit smaller. A closer inspection of the Al(W) resonance around 60 ppm shows, however, that its maximum is shifted closer to the diagonal, compared with the dehydrated sample, suggesting a reduced average C,,, for this site. Indeed, it was not possible to simulate this peak with a single Gaussian distribution in giisO and C,,,. A second distribution with a much smaller average C,,, = 4 + 1 MHz and giis, = 60 + 3 had to be added to the one used for the dehydrated sample. This effect is much
ppm. These results explain why we observe the interchange of signal intensity from the Al(V) region to the Al(W) region in the normal MAS spectra upon hydration. In fact there is only a small amount of Al(V) present in the materials (maximally 10%). This signal is added up to the strong high field flank of the Al(W) resonance in the dehydrated samples. Upon hydration, the Al(N) resonance narrows causing the intensity in the 40-16 ppm region to decrease. Although the intensity of the Al(V) resonance also decreases upon hydration, this is only a minor effect as there is only a small amount present. The precision of the integration procedure prohibits us to determine whether this Al(V) is converted to Al(V1) or ANIV). The Al(W) resonance is only mildly affected by the hydration/dehydration process. The observations are in accordance with the model proposed earlier. The fourfold-coordinated aluminium is incorporated in the aluminosilicate network. The remaining aluminium serves as chargecompensating cations, mostly in the form of tiny alumina particles where most of the aluminium is octahedrally coordinated. The quadrupole parameters observed for the Al(W) resonance are very similar to those observed in alumina. In a hydrated sample these particles are solvated allowing the polymer network some mobility, as is reflected in the line width of the Al(W) resonance. Upon dehydration, however, much more strain is induced in the network because of the electrostatic forces of the alumina particles on the Al in the network. This attraction of the network Al to the charge compensating alumina particles has two effects; first of all different conformations occur in the network. Indeed, an increase in the isotropic chemical shift distribution is observed.
214
M.P.J. Pee&m, A.P.M. Kentgens/Solid
180
140
100
State Nuclear Magnetic Resonance 9 (1997) 203-217
60
20
0
-40
-80
(w-N Fig. 8. Projections of the 27A1 3Q-MAS spectrum along the F2 direction of the tetrahedrally coordinated aluminium. The F, spinning sideband of the octahedrally coordinated aluminium is indicated by an asterisk. (a) 0.20 Al(OBu”),EAA/0.40 PhTES/0.40 TEOS/4 EtOH/3.4 HCl (0.005 M) gel after heating at 200°C for 2 h, (b) 0.20 AI(OBuS),/0.40 PhTES/0.40 TEOS/0.30 EAA/4 EtOH/3.4 HCI (0.005 M) gel after heating at 200°C for 2 h, hydration and subsequent dehydration at 120°C for 2 h and (c) 0.20 AI(OBu”)s/0.40 PhTES/O.40 TEOS/O.30 EAA/4 EtOH/3.4 HCl (0.005 M) gel after heating at 200°C for 2 h and hydration.
This is expected as the isotropic correlated to the Al-0-Si bond the charge-compensating cations closer to the network aluminium, ence much larger electric field witnessed by a strong increase quadrupole coupling constant. 3.5. “Al off-resonance In order parameters we tried quadrupole
chemical shift is angles. Secondly, get located much which thus experigradients. This is in the average
nutation NMR
to get a verification of the quadrupolar as estimated from the MQMAS spectra nutation spectroscopy [38-401. The interactions encountered are fairly strong
so we had to resort to off-resonance nutation spectroscopy [31]. In a study of a series of aluminosilicate glasses it was demonstrated that off-resonance nutation can be used in amorphous systems to get information about the average quadrupole parameters [41]. Due to the fact that off-resonance nutation mainly relies on line intensities to discriminate between different sites well resolved spectra can still be obtained in systems (e.g., glasses) where a distribution of quadrupolar interactions is present. Thus the average quadrupole coupling constant can be obtained but there is no information about the details (i.e., width and shape) of the distribution of these parameters. “Al off-resonance nutation spectra were
M.P.J. Peeters, A.P.M. Kentgens/Solid
Nutation
State Nuclear Magnetic Resonance 9 (1997) 203-217
Frequency
215
(Hz)
Fig. 9. Off-resonance nutation NMR spectra of the 0.20 AffOBu”),/0.20 EAA/0.40 PhTES/0.40 TEOS sample after annealing at 200°C (dry sample). Resonance offset = 190 kHz, rf field strength = 38 kHz. Simulation of the signal at 55 ppm (trace a), experimental spectrum (trace b), Simulation of the signal at 0 ppm (trace c), experimental spectrum (trace d). The uninformative signal at zero-frequency is truncated in the experimental spectra. Parameters used for the simulations are listed in Table 2.
taken for two samples: 0.20 Al(OBu”),/0.20 EAA/0.40 PhTES/0.40 TEOS after annealing at 200°C and after hydration. Resonance offsets used were between 120 and 260 kHz with an rf-field strength between 38 and 48 kHz, using a spinning speed of 2 kHz. As has been discussed before, nutation spectra should either be taken statically or at a spinning speed that is low enough to ensure that the sample can be assumed static during the nutation pulse. In practice no effects of the spinning on the nutation spectra is observed when the longest t,period does not exceed l/4 rotor-period. In the present case this is a serious drawback as the resonances partly overlap even at high spinning speeds. At low speeds this only gets worse as spinning sidebands appear as well. As has been noted in the MQMAS spectra the tail of the Al(W) resonance strongly overlaps with the (low intensity) Al(V) resonance. Therefore, it is not possible to get a reliable value for the average C,,, of Al(V) in the dehydrated sample. An example of an off-resonance nuta-
tion spectrum, together with the simulation is given in Fig. 9 for the dry sample. The full set of parameters used for the simulation of all spectra are listed in Table 2. In the dehydrated sample the signal at 60 ppm is characterised by an average quadrupole coupling constant of 5.5 MHz. Considering the accuracy of the method this agrees well with the results obtained from the MQMAS spectrum (6.0 MHz). The slightly lower value might be due to the slow spinning which has the disadvantage that the lines with the largest quadrupole interactions contribute less to the central line shape [42]. As was already mentioned, spectral overlap prohibits us to extract information about the Al(V) resonance. Good agreement is observed for the average quadrupole coupling constant derived from the MQMAS (4.5 MHz) and off-resonance nutation NMR spectrum (4.0 MHz) for the signal of the octahedral aluminium. The off-resonance nutation again giving a slightly lower value. Although the off-resonance nutation NMR spectra are not so
216
M.P. J. Peeters, A.P.M. Kentgens / Solid State Nuclear Magnetic Resonance 9 (1997) 203-217
Table 2 Off-resonance nutation NMR performed on 0.20 Al(0Bt1”)~ /0.20 EAA/0.40 PhTES/O.40 TEOS sample (dry) after annealing at 200°C and after hydration. Parameters used for the simulations of the projections along the Fl dimension Sample state
Resonance offset (kHz)
RF-strength (kHz)
6 (ppm)
$Hz)
7:)
Dehydrated Dehydrated Dehydrated Dehydrated Dehydrated Dehydrated Dehydrated Dehydrated Hydrated Hydrated Hydrated Hydrated
120 190 220 260 120 190 220 260 150 220 150 220
38 38 36 38 38 38 36 38 48 38 48 38
0 0 0 0 55 55 55 55 0 0 55 55
4.0 4.0 3.9 4.0 5.7 5.7 5.5 5.3 3.5 3.7 3.9 3.8
1 1
I 1 1
1 1 1 1 1 1 1
sensitive for the determination of v~, it is clear from the simulations that 71~is close to 1 for all sites. For the hydrated sample a reduced quadrupole interaction was found for the Al(W) resonance in the MQMAS spectra (4.0 MHz). The off-resonance nutation spectra corroborate this result, giving a C,,, of 3.9 MHz. For the octahedrally coordinated aluminium species a Cscc 3.6 MHz is obtained by off-resonance nutation NMR, again in accord with the MQMAS result of 4.0 MHz.
4. Conclusions
The effects of aluminium incorporation in TEOS based hybrid materials was studied by MAS NMR. After RT drying of the sol-gel synthesised materials, octahedral and tetrahedral aluminium exist in the materials; the formation of tetrahedrally coordinated aluminium is favoured by a low water concentration. In all materials prepared the ratio of tetrahedral to octahedrally coordinated aluminium is lower than three. Some of the octahedrally coordinated aluminium must therefore reside in an alumina rich phase. Drying of the aluminosilicate gels results in an decrease of the amount of octahedrally coordinated aluminium and the formation of a new resonance at 30 ppm. Several papers [13-151 ascribe this resonance to pentahedrally coordinated aluminium
ions, at the interface between the aluminosilicate and the aluminate region. The fivefold-coordination is stabilised by the interfacial strain, or alternatively minimises this strain. Storing the samples in air showed a change in the observed signal intensities indicating hydration effects. These effects were investigated by 27A1MAS, MQMAS and off-resonance nutation NMR. In the MAS spectra the intensity of the 30 ppm signal decreases upon hydration, and a congruent increase in intensity of the 55 ppm signal is observed. This hydration/dehydration process is fully reversible, ruling out the possibility of atomic rearrangements on a large scale. Drying at 120°C results in an increase of the intensity of the signal at 30 ppm. Drying at this temperature is expected to remove only the physisorbed water. It is therefore unlikely that the signal at 30 ppm is caused only by a fivefold-coordinated aluminium species. MQMAS experiments clearly demonstrated that most of the intensity in the 40-15 ppm region is due to the asymmetric tail of tetrahedrally coordinated aluminium species due to two causes: the average C,,, is very large, leading to large quadrupole induced shifts, and a great variation in the isotropic chemical shift. This increase in chemical shift distribution is attributed to a greater dispersion of conformations in the polymer network. This is expected when the network aluminium is electrostatically attracted to the charge-compensating alumina particles. The closer location of these charged particles increases the field-gradients felt by the network aluminium resulting in the increased C,,,. The off-resonance nutation data corroborated the qualitative analysis of the MQMAS spectra.
Acknowledgements Mr. H. Janssen, Mrs. G. Nachtegaal and Mr. J. van OS of the SON/NW0 HF-NMR Facility and Mr. M. van Bommel and Mrs. Snijkers-Hendrickx of the Philips Research Laboratories are greatly acknowledged for their various contributions to this work.
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
References 111 I. Jaymes, A. Douy, J. Eur. Ceram. Sot. 16 (2) (1996) 155.
t21I.
Jaymes, A. Douy, D. Massiot, J.P. Busnel, J. Am. Ceram. Sot. 78 (10) (1995) 2648. [31 I. Jaymes, A. Douy, D. Massiot, J.P. Coutures, J. Chim. Phys. Phys.-Chim. Biol. 92 (10) (1995) 1851. [41 Y.F. Chen, S. Vilminot, Mat. Res. Bull. 30 (3) (1995) 291. (51 T. Heinrich, F. Raether, H. Marsmann, J. Non-Cryst. Solids 168 (1994) 14. 161 U. Selvaraj, S. Komarneni, R.J. Roy, J. Am. Ceram. Sot. 73 (12) (1990) 3663. [71 C. Sanchez, J. Livage, M. Henry, F. Babonneau, J. Non. Cryst. Solids 100 (1988) 65. t81 F. Babonneau, L. Coury, J. Livage, J. Non. Cryst. Solids 121 (1990) 153. F. Babonneau, J. Livage, J. Sol-Gel I91 L. Bonhomme-Coury, Sci. Technol. 3 (1994) 157. [lOI W.L. Olson, L.J. Bauer, Mat. Res. Sot. Proc., D.E. Clark, D.K. Ulrich (Eds.) 73 (1986) pp. 187-193. [Ill L. Bonhomme-Coury, F. Babonneau, J. Livage, Chem. Mater. 5 (1993) 323. [121G.A. Pozamsky, A.V. McCormick, J. Non-Cryst. Solids 190 (1995) 212. iI31 A.D. Irwin, J.S. Holmgren, J. Jonas, J. Mater. Science 23 (1988) 2908. [141 B.M. de Witte, P.J. Grobet, J.B. Uytterhoeven, J. Phys. Chem. 99 (1995) 6961. [151 F. Hatakeyama, T.J. Maekawa, J. Ceram. Sot. Jpn. 100 (21 (1992) 163. [I61R. Dupree, I. Farnan, A.J. Forty, S. El-Ma&i, L. Bottyan, J. Phys. Colloq. c8 (1985) 113-117. 071 G.J. Ray, A. Samoson, Zeolites 13 (1993) 410. [181 D. Coster, A.L. Blumenfeld, J.J. Fripiat, J. Phys. Chem. 98 (1994) 6201. [191 A. Taylor, D. Holland, J. Non-Cryst. Solids 152 (1993) 1. DOI P.J. Sanchez-Soto, I. Sobrados, J. Sanz, J.L. Perez-Rodriguez, J. Am. Ceram. Sot. 76 (12) (1993) 3204. WI J.J. Fitzgerald, S.F. Dee, A.I. Hamza, Am. Miner. 74 (1989) 1405.
WI G. Kunath-Fandrei,
217
T.J. Bastow, J.S. Hall, C. Jager, M.E. Smith, J. Phys. Chem. 99 (1995) 15138. b31 C. Femandez, J.P. Amoureux, Solid State Nucl. Magn. Reson. 5 (1996) 315. t241 D. Massiot, B. Touzo, D. Trumeau, J.P. Coutures, J. Virlet, P. Florian, P.J. Grandinetti, Solid State Nucl. Magn. Reson. 6 (1996) 73. [251 G. Wu, D. Rovnyank, B. Sun, R.G. Griffin, Chem. Phys. Lett. 249 (1995) 210. [261 A. Medek, J.S. Harwood, L. Frydman, J. Am. Chem. Sot. 117 (1995) 12779. 1271J.P. Amoureux, C. Femandez. L. Frydman, Chem. Phys. Lett. 259 (1996) 347. I281 A.P.M. Kentgens, J. Magn. Reson. 95 (1991) 619. t291 V.B. Cheng, H.H. Suzukawa, M.J. Wolfsberg, J. Chem. Phys. 59 (19731 3992. I301 H. Conroy, J. Chem. Phys. 47 (1967) 5307. [311 A.P.M. Kentgens, J. Magn. Reson., Ser. A 104 (1993) 302. [321 S. Simon, G.J.M.P. van Moorsel, A.P.M. Kentgens, E. de Boer, Solid State Nucl. Magn. Reson. 5 (1995) 163. t331 S. Simon, A. var der Pol, E.J. Reijerse, A.P.M. Kentegens, G.J.M.P. van Moorsel, E. de Boer, J. Chem. Sot. Faraday Trans. 90 (18) (1994) 2663. t341 E. Lippmaa, A. Samoson, M. Magi, J. Am. Chem. Sot. 108 (1986) 1730. [351 P.J. Dirken, A.P.M. Kentgens, G.H. Nachtegaal, A.M.J. van der Eerden, J.B.H. Jansen, Am. Miner. 80 (1995) 39. [361 L. Frydman, J.S. Harwood, J. Am. Chem. Sot. 117 (1995) 5367. [371 H. Kraus, R. Prim, A.P.M. Kentgens, J. Phys. Chem. 100 (1996) 16336. [381 A. Samoson, E. Lippmaa, Chem. Phys. lett. 100 (1983) 205. [391 A. Samoson, E. Lippmaa, Phys. Rev. B 28 (1983) 6567. I401 A.P.M. Kentgens, J.J.M. Lemmens, F.M.M. Geurts. W.S. Veeman, J. Magn. Reson. 71 (1987) 62. I411 P.J. Dirken, G.H. Nachtegaal, A.P.M. Kentgens, Solid State Nucl. Magn. Reson. 5 (1995) 189. [421 R.K. Sato, P.F. McMillan, P. Dennison, R. Dupree, J. Phys. Chem. 95 (1991) 4483.
ELSEVIER
Solid State Nuclear Magnetic Resonance
9 (1997) 203-217
A 27A1MAS, MQMAS and off-resonance nutation NMR study of aluminium containing silica-based sol-gel materials ’ M.P.J. Peeters a,2, A.P.M. Kentgens b3* b N.S.R.
a Philips CFT, Pro5 Holstlaan 4, 5656AA Eindhouen, Netherlands Centerfor Molecular Design, Synthesis and Structure, National HF NMR Facility, University of Nijmegen, 6525ED Nijmegen, Netherlands Received 6 December
1996; accepted 4 February 1997
Abstract Aluminium containing hybrid materials were prepared via the sol-gel method using aluminium set-butoxide complexed with ethylacetoacetate (Al(OBti”),EAA or Al(OBu”),/EAA mixtures). As silanes, phenyltrimethoxysilane (PhTMS) or phenyltriethoxysilane (PhTES), 3-glycidoxypropyl trimethoxysilane (Glymo) and tetraethylorthosilicate (TEOS) were used. After room temperature drying of the samples the 27A1single pulse excitation (SPE) magic angle spinning (MAS) NMR shows that octahedral (5 ppm) and tetrahedral (55 ppm) coordinated aluminium species are present in the materials. The relative amount of these two species depends on the preparation method. However, the Al(IV)/Al(VI) ratio is lower than 3 (typically 2.3) in all materials, indicating the presence of a small amount of an aluminate phase. Annealing of the samples at 100, 150 and 200°C results in the formation of an extra signal at 30 ppm (peak maximum measured at 11.7 T). Based on the resonance frequency this signal is generally assigned to a pentahedrally coordinated aluminium species. Hydration/dehydration processes of annealed samples were studied with *‘Al SPE MAS NMR, multiple-quantum MAS NMR (MQMAS) and off-resonance nutation NMR. Upon hydration of the annealed sample the signal intensity around 30 ppm decreases in intensity and at the same time the intensity of the signal around 55 ppm increases by the same amount (tetrahedrally coordinated aluminium). The MQMAS spectra reveal that the signal around 30 ppm is not caused by a fivefold-coordinated aluminium species but mainly by tetrahedrally coordinated aluminium species in a distorted environment, experiencing large quadrupole induced shifts and small chemical shifts due to conformational changes in the polymeric network. From the MQMAS NMR spectra it can be concluded that the linebroadening observed in the 27A1MAS NMR spectra is due to both a distribution in isotropic chemical shifts and a distribution in quadrupole coupling constants (C,,, = e*qQ/h). Hydration of the sample results in a decrease of the average Cs_ for the tetrahedrally coordinated aluminium from 6 to 4 MHz, whereas the average C,,, of the octahedrally coordinated aluminium is hardly influenced (4 MHz). These MQMAS results are confirmed by off-resonance nutation experiments. 0 1997 Elsevier Science B.V. Keywords: Solid-state NMR; MQMAS; Off-resonance nutation; Combined inorganic-organic
* Corresponding author. E-mail: [email protected]. ’ Dedicated to Giinter Engelhardt. ’ E-mail: [email protected]. 0926-2040/97/$17.00 0 1997 Elsevier Science B.V. All rights reserved PII SO926-2040(97)00060-X
materials;
Al-coordination
204
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
1. Introduction Aluminosilicate gels are used in catalysis, optics and as multicomponent precursors for the preparation of ceramics (e.g., Mullite, Corderiete) [l-6]. Preparation of these materials via the sol-gel process opens the possibility to synthesise (on a molecular scale) homogeneous glass-like materials via a low temperature process. Aluminosilicates have, compared to pure silicates, an increased thermal and chemical stability, a lower thermal expansion coefficient, and an increased mechanical toughness. A great number of chemical precursors can be used for the sol-gel preparation of aluminosilicates alkoxides, oxides, salts). Often (e.g. Al(OBu”),/EAA mixtures or Al(OBu”),EAA (complex, no free set-butanol) are used (EAA = Ethylacetoacetate, C,H,,O,). The EAA (P-diketone) is added to the aluminium alkoxide to reduce the reactivity of aluminium alkoxides towards water
and to prevent immediate precipitation [7]. Complexation of aluminiumalkoxide with EAA is a consequence of the fact that the oxidation state (Z) of the metal is lower than the coordination number (Nl. In that case coordination expansion by oligomerisation and/or charge transfer complex formation is preferred 171. Several structures (Fig. 1) have been proposed for mono P-diketone complexed aluminium alkoxides [8], depending on the size of the alkoxide group and/or the temperature. Structure (4) for example was confirmed by a single crystal study of Al(OPr’),(AcAc). Very bulky alkoxy groups lead to the monomeric structure (1) [9]. The 27A1 NMR spectrum of Al(OBu”),EAA shows the presence of 4-5 and 6-coordinated aluminium. This indicates that in Al(OBu”),EAA several species are present, for example a mixture of (11, (2) and (3) or of (31 and (4) (Fig. 1). Several papers deal with the characterisation of aluminium containing materials prepared via the ,’
‘\
(1) OR
0
OR
0--_. /
b R
Fig. 1. Possible structure indicated in parenthesis.
of ethylacetoacetate
(EAA)
complexed
aluminium
alkoxides
[S]; the coordination
(IV)
,x _---OR
number
of aluminium
is
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
sol-gel method using Al(OBu”),EAA [7-121. Babonneau et al. [8] used Al(OBu”),EAA for the preparation of a-cordierite (Mg:Al:Si:H,O:EtOH = 2:4:5:32:52). IR measurements revealed that when Al(OBu”),EAA reacts with D,O, hydrolysis of butoxy groups occurred (strong decrease of the band at 1060 cm- ’ >. A small amount of free EAA could be detected in the sol (bands at 1710 and 1740 cm-’ > although the characteristic bands of EAA bonded to aluminium (1630, 1610, 152.5, 1420, 1370, 1300, 1175, 630 and 510 cm-‘) remained intense. A broad band at 400 to 700 cm-’ indicated the formation of an Al-O-Al network [8]. In a second publication Bonhomme-Coury et al. [ll] studied the a-cordierite preparation with 29Si and *‘Al liquid and solid-state NMR. Hydrolysis of the precursor solution resulted in the appearance of a sharp signal at 50.2 ppm in the 27A1 NMR spectra (symmetric tetrahedral site, intensity 7%) and a asymmetric signal at 55 ppm (distorted tetrahedral sites, intensity 58%), indicating tetrahedral coordination of 65% of the aluminium. The amount of tetrahedrally coordinated aluminium increased with reaction time. It was proposed that the Al-OBuS bond in Al(OBu”),EAA hydrolyses first and that the partly hydrolysed aluminium is incorporated in the silicon network via the following reaction: Al-OH + Si( OEt), + Al-O&(
OEt), + EtOH
( 1)
The aluminium bonded to silicon is expected to be in a fourfold-coordination (tetrahedral), which explains the increase of the amount of tetrahedrally coordinated aluminium with time. These results are in line with a publication of Heimich et al. [5]. They studied the preparation of Mullite gels (stoichiometry: 3 Al,O, * 2 Si$+> starting from TEOS and Al(OBu”),EAA. The Al NMR spectra showed narrow lines at 51 and 58 ppm. These are assigned to tetrahedrally coordinated aluminium atoms with silicon in the second coordination sphere. The average coordination number of aluminium in the sol increased with increasing reaction time. Since each aluminium introduced into the silica network introduces a charge deficiency, part of the aluminium is expected to compensate for this. Charge compensation can be obtained by Al(H,O)~‘, with aluminium in an octahedral coordination. Irwin et al. [13] have shown that upon addition of sodium to the
205
sol (in the form of NalAl(OR),]) for charge compensation, the amount of octahedrally coordinated (nonnetwork) aluminium decreased or disappeared from the samples after drying at 40°C. This suggests that fourfold-coordinated aluminium is incorporated in the silicon network, while octahedral aluminium is necessary for charge compensation. Drying of the aluminosilicate gels at moderate temperatures ( < 200°C) results in the formation of a new signal at approximately 30 ppm in the *‘Al MAS NMR spectra [ 13-151. The chemical shift of this new resonance is indicative of five-coordinated aluminium [14,16]. Since the first observations of fivefold-coordinated aluminium by Dupree et al. in 1985 [ 161, fivefold-coordinated aluminium species have been observed in aluminas, zeolites [17,18], and aluminium containing ceramic precursors [19]. In many of these materials fivefold-coordinated aluminium is formed after a severe chemical or thermal treatment. This can lead to some very distorted sites, e.g., in the case of dehydroxylation of Pyrophillite [Al,Si,O,,(OH),] at 800°C [20] a C,,, of 10.5 MHz was found [21]. It is not generally true, however, that fivefold-coordinated aluminium has the largest quadrupole coupling constant, e.g., in X-Al,O, the fivefold-coordinated aluminium appeared to have the smallest C,,, 1221. Ray and Samoson [17] found a relatively small C,,, for fivefold-coordinated aluminium in steamed zeolites, compared to a very distorted tetrahedrally coordinated aluminium species they encountered in the same material. The presence of fivefold-coordinated aluminium in aluminosilicates has also been reported. Hatakeyama and Maekawa [15] report that the intensity of the signal at 30 ppm increases in intensity with temperature in their aluminosilicate samples. Hydration of the samples (autoclave, 50 h, 70°C) lead to a decrease of the 30 ppm signal. Fivefold-coordinated aluminium is said to be formed by dehydration of the non-network octahedrally coordinated aluminium. This process is reversed upon hydration. The fivefold-coordinated aluminium disappears upon crystallisation of the A1,03-SiO, gels. De Witte et al. 1141 studied aluminium-rich amorphous aluminosilicates treated at low temperatures. They also assign the resonance observed at 30 ppm in the *‘Al MAS NMR spectra to fivefold-coordinated aluminium. The aluminium in a fivefold-coordination is
M.P.J. Peeters, A.P.M. Kentgens/Solid
206
State Nuclear Magnetic Resonance 9 (1997) 203-217
an intermediate, metastable species, typically disappearing from the spectra when the sample is treated at temperatures exceeding 980°C and crystallisation has occurred. Calcination of the amorphous aluminosilicate sample at 400°C leads to a strong increase of the 30 ppm signal at the expense of tetrahedrally coordinated aluminium. Synthesis of the materials in alkaline media results in a decrease of the amount of octahedral and fivefold-coordinated aluminium. The aluminium containing silica-based sol-gel materials used in this study were prepared using the Al(OBu”),EAA or Al(OBu”), components as the aluminium containing precursor. The resulting dried hybrid materials were studied with “Al NMR. The annealing temperature and the degree of hydration were varied. *‘Al MAS NMR, 27A1 MQMAS NMR and off-resonance nutation NMR were performed to characterise the different aluminium species present in the dried materials.
2. Experimental
2.1. Sample preparation The samples were prepared by mixing tetraethylorthosilicate (TEOS) with a trifunctional alkoxysilane (T). To this mixture (absolute) ethanol was added followed by a 1: 1 (W/W> acidic water/ethanol mixture (HCl, [Hf] = 5 * 10-3). The trifunctional siloxane (T) was either PhTMS (Phenyltrimethoxysilane), PhTES (Phenyltriethoxysilane) or Glymo (3-glycidoxypropyl-trimethoxysilane). The last step was the addition of the aluminium set-butoxide compound, either Al(OBu”),EAA or Al(OBu”), complexed with EAA (ethylacetoacetate). TEOS was obtained from Merck, the trifunctional siloxanes were obtained from Hiils, the Al(OBu”), and A1(OBuS),EAA from Gelest. All reagents were reagent grade and used without further purification. The general composition of the samples was: xT,yTEOS,zA1(OBu”),EAA(orAl(OBu’)3), 4EtOH,(4 x+y+z=
-x
- z)H,O
1
Hydrolysis with water was performed for 2 h at 50°C in closed vessels. After 2 h the solvent was allowed to evaporate at room temperature. For an-
nealing an additional heat treatment was applied in a conventional oven for 2 h at a temperature of 100, 150 and 200°C respectively.
2.2. 2?41 NMR “Al NMR measurements were performed on a Bruker AM-500 operating at 130.32 MHz. Short radio frequency pulses @-field of 25 kHz, tip angle < 25” for aluminium species with a large C,o> and a recycle delay of 2 seconds were used to obtain quantitative results. A MAS speed of N 13.0 kHz was used. No proton decoupling was applied. Unless otherwise indicated, the chemical shift of the resonances is characterised by the peak maximum measured at 11.7 T. MQMAS experiments were recorded using the two-pulse sequence [23-261. Experiments were performed using a home-built MAS probe equipped with a Jakobsen stator. The spinning frequency was 13 kHz (Si,N, or zirconia rotors). The rf-field strength was N 65 kHz. A 5 ps preparation and a 2 ,us mixing pulse was used. 128 (t,)*512 (t,) points with a dwell time of 5 pus in both directions were acquired. Phase cycling over four times six phases was employed to select triple-quantum coherence during the evolution period; 3168 scans were acquired per experiment with a recycle delay of 1 s. TPPI was used to obtain pure-absorption mode lineshapes. It should be noted, however, that due to the different signal intensities from the 0 + + 3(t,) -+ - l(t,> and the 0 + - 3(t,) -+ - l(t,) transfer pathways some phase errors remain in the spectra [27]. Off-resonance nutation spectra were recorded under slow MAS conditions (ca. 2 kHz), in order to be able to discriminate between otherwise overlapping peaks. The spinning speed was chosen so low that the pulse duration did not exceed about l/4 of a rotor period. Under these conditions the sample can be considered static during the evolution period of the nutation experiment. A Bruker 7 mm MAS multinuclear resonance probehead was used; the sample was restricted to the centre of the spinner (using Teflon spacers) to ensure good rf-homogeneity. A frequency-stepped adiabatic half-passage was applied as preparation period [28]. 128 Spectra were recorded with 1 ps pulse length increments and resonance offsets be-
M.P.J. Peeters, A.P.M. Kentgens/Solid
207
State Nuclear Magnetic Resonance 9 (1997) 203-217
3. Results
tween 120 and 260 kHz. An exponential filter with a line-broadening of 800 Hz was applied in both directions. Magnitude spectra were obtained whose sum projections along the F,-axis are the spectra shown in the figures.
3.1. 27Al MAS NMR 27A1 MAS NMR spectra were recorded for 0.20 Al(OBu”),EAA/0.80 TEOS and 0.20 Al(OBu”),EAA/0.40 PhTES/0.40 TEOS samples after room temperature drying. A signal for tetrahedrally coordinated aluminium (Al(W), 55 ppm) and octahedrally coordinated aluminium (Al(W), 0 ppm) can be detected. Unless otherwise indicated the chemical shifts mentioned are the peak maxima observed at 11.7 T. Both 27A1resonances are broad and a-symmetric, indicative of a distribution in quadrupo lar coupling constants and chemical shifts. The relative amount of these two species depends on the gel preparation method. In Fig. 2 the percentage of Al(W) en Al(W) in the samples after drying are displayed as a function of the Al(OBu”),EAA addition moment. Prehydrolysis of TEOS, prior to the addition of the aluminium compound, a frequently used technique to get a better match between the hydrolysis rates of the different metal-alkoxide com-
2.3. Calculations All numerical simulations were run on a PC using software written in MATLAB, C + + and Pascal. Powder averages were performed by simultaneously incrementing the Euler angles [29,30]. Off-resonance nutation spectra were calculated by numerical diagonalization of the off-resonance rotating frame Hamiltonian [31]. MQMAS spectra were calculated by simulating the line-shape in the frequency domain. Gaussian distributions in isotropic chemical shift and quadrupole coupling constants were generated and the resulting line shapes added to give the final spectrum. The distributions were calculated from - 3a to + 3a in 5 1 points, where u is the halfwidth at half height.
80 Al( IV)
n n
n
8 h > Y
60
a
0
1
2
3
4
5
6
7
AI(OBU~)~ EAA addition time (hrs) Fig. 2. Aluminium coordination number as a function of the Al(OBuS),EAA addition moment; gel composition 0.01 AI(OBuS),EAA/0.20 Glymo/0.79 TEOS/4 EtOH/3.8 HZO. The two datapoints at t = 0 are from an acid catalysed and a neutral sol. Acid catalysis (faster hydrolysis of the silanes) leads to a higher amount of tetrahedrally coordinated aluminium.
208
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
pounds, leads to a higher amount of tetrahedrally coordinated aluminium. Tetrahedrally coordinated aluminium species are ascribed to network aluminium, whereas octahedrally coordinated aluminium species are needed for charge compensation [13]. As can be seen from Fig. 2, the amount of tetrahedrally coordinated aluminium increases rapidly by lo-15% using a prehydrolysis period of 30 min. Continuation of the prehydrolysis hardly increases the amount of tetrahedrally coordinated aluminium. This indicates that a higher amount of aluminium is incorporated in the silica network if a small prehydrolysis period is used. A “0 NMR study of the hydrolysis mixture shows that during the prehydrolysis period the water concentration of the sol drops to +20% of its initial value after 20-30 min, using TEOS in the presence of 0.005 M HCl. Prolonged hydrolysis leads to condensation of the hydrolysed TEOS molecules. The water concentration remains constant at approximately 20% during this stage. To study the influence of the water concentration (during synthesis) on the aluminium coordination number, sols were prepared with varying water concentrations. The general composition of these sols was 0.20 Glymo/0.60 TEOS/0.20 Al(OBu”),EAA/4 EtOH/w H,O. The value of w was varied between 2 and 8. No prehydrolysis period was applied. Gelation of the sols was observed after +20 min hydrolysis at 50°C for all sols with w > 2. In the case of w = 2 a viscous liquid was obtained after 2 h. The results of the 27Al MAS NMR measurements are displayed in Fig. 3. The amount of tetrahedrally coordinated aluminium decreases with increasing water concentration. These results are in line with the results displayed in Fig. 2 and with the IR study of Babonneau et al. [8]. At a low water concentration (later addition of the Al(OBu”),EAA component), hydrolysis of the Al-OBu” bond is likely to occur, whereas hydrolysis of the Al-EAA complex will be negligible. This controlled hydrolysis leads to monomeric tetrahedrally coordinated aluminium species (27A1 liquid state NMR), which are capable of undergoing condensation reactions, as described in Eq. (1). Charge compensation by octahedrally coordinated Al(H,O)z+ is expected to results in an Al(W) to Al(V1) ratio of 3 [13]. In all materials prepared, however, this ratio is lower than three, a typical
65 .
I
1
2
3
4
5
6
7
8
9
w Fig. 3. Aluminium coordination number as a function of the water concentration (w). Gel composition: 0.20 Glymo/0.20 Al(OBu”),EAA/0.60 TEOS/4 EtOH/ w H,O. Lines represent the result of linear regression.
value being about 2.3. This lower value suggests the presence of a small amount of an alumina-like phase. Another important substituent in the preparation of aluminium containing hybrid materials is ethylacetoacetate (EAA). The influence of EAA on the final aluminium coordination number was investigated by preparation of sols with the following composition: 0.20 Al(OBu”),/0.40 PhTES /0.40 TEOS/3.4 H,0/4 EtOH/c EAA with c varying between 0 and 0.4 (EAA/Al = O-2). Prehydrolysis of the 0.4 TEOS/0.40 PhTES/2 EtOH/3.4 H,O mixture was performed in all cases for 20 min, to prevent the precipitation of the uncomplexed Al(OBu”), (c = 0). After 20 min the Al(OBu”),/c EAA/2 EtOH mixture was added to the prehydrolysed solution. “Al MAS NMR of the dried gels (RT) revealed the presence of 40 + 3% octahedrally coordinated aluminium in all the materials. This percentage of octahedrally coordinated aluminium is comparable to the value obtained for the gels prethe commercially available pared using Al(OBu”),EAA (Fig. 2). The influence of the EAA addition to the sol on the final aluminium coordination number is thus negligible for those samples prepared with a prehydrolysis period. 3.2. Influence
of annealing
temperature
on the 27Al
MAS NMR spectra
After drying at room temperature the samples were given an additional heat treatment for 2 h at
209
M.P.J. Peeters, A.P.M. Kentgens/ Solid State Nuclear Magnetic Resonance 9 (1997) 203-217 I
Strained 4-coordinated
,,
,_?m_-,
,“““““““““““““““““““”
0
lb0
-100
-200
(fwm) Fig. 4. “Al MAS NMR (11.7 Tesla) spectra of 0.20 Al(OBu”),EAA/0.80 TEOS samples after heating at the indicated temperature (MAS speed 13.0 Hz). The signal at 105 ppm represents an impurity in the rotor material.
for 2 h
C
b
a m,
110
90
70
50
Fig. 5. 27Al MAS NMR of 0.20 Al(OBuS),EAA/0.40 dehydrated and (c) rehydrated.
30
10 (wm)
PhTES/0.40
I
-10
I,
-30
TEOS samples
I
-50
I,
I,
-70
/,
I
-9o-sF
after heat treatment
at 200°C. (a) hydrated,
(b)
210
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
temperatures of 100, 150 up to 200°C. Annealing of the samples at temperatures of 100, 150 and 200°C leads to the formation of a signal at 30 ppm (Fig. 4). Given the position in the 27A1MAS NMR spectrum the signal can be ascribed to either a fivefold-coordinated aluminium species or to an aluminium species in a distorted tetrahedral environment, experiencing a large quadrupole interaction leading to a large quadrupole induced shift. Moreover a shoulder at 80 ppm is formed. This shoulder is assigned to tetrahedrally coordinated aluminium. Due to the heat treatment, the amount of octahedrally coordinated aluminium decreased slightly to approximately 30-40%. These changes in the Al spectra upon heat treatment can be attributed to the formation of small alumina particles in the sample containing four-, five- and sixfold-coordinated Al. The amount of octahedral Al in the samples is higher than necessary for chargecompensation of the tetrahedral Al in the aluminosilicate network. Therefore, it is not unlikely that the excess aluminium forms small alumina-particles. It should be noted, however, that for alumina treated at temperatures below 2OO”C, generally only octahedral Al is observed [32]. Quantitative 27A1 MAS NMR using an external reference ( a-A1,03, Corundum) showed that all aluminium atoms can be observed in the NMR experiment, even after heat treatment. 3.3. Influence of hydration /dehydration MAS NMR spectra
Table 1 Integration Al(OBu”),
of the “Al MAS NMR spectra of /0.40 PhTES/0.40 TEOS/c EAA samples
Sample condition
Integration 95-40
ppm
0.20
region 40- 16 ppm
16-( - 40) ppm
c=o After annealing Hydrated Dehydrated (120°C) Rehydrated
34 45 31 40
29 17 30 21
37 38 39 39
c=O.lO Hydrated Dehydrated
43 29
18 28
39 43
c = 0.20 After annealing Hydrated Dehydrated (120°C) Rehydrated
35 47 35 44
27 17 26 22
38 36 39 35
c = 0.30 Hydrated Dehydrated
(120°C)
47 36
13 25
40 39
c = 0.40 Hydrated Dehydrated
(120°C)
40 31
18 25
42 46
( 12O’C)
the
The absolute integral after the various treatments within 5%.
remains constant
on the 2%1
Upon storage of the samples in air, a decrease of the 30 ppm resonance was noted. This was attributed to hydration effects, and therefore a more systematic study of hydration-dehydration effects was undertaken. After annealing of the samples at 200°C the samples were saturated with water (hydrated). The appearance of the aluminium spectra of the annealed materials changes with the hydration state of the materials (Fig. 5). In the dry state (directly after annealing) an intense signal at 30 ppm is present in the material as described above. Addition of water to the sample leads to a strong reduction of the intensity of this signal. At the same time the intensity of the tetrahedral signal at 55 ppm increases proportionally, as shown by integration (Table 1). Dehydration of the samples results in a decrease of the signal at 55 ppm by lo- 12%, whereas the signal at 30 ppm
increases by the same amount. The intensity of the signal at 0 ppm does not change significantly (maximally +4%). The absolute integral, a measure for the number of aluminium atoms observed in the experiments remains constant (+5%), i.e., all the aluminium remains visible. Although integration of the overlapping resonances is difficult, it seems as if upon hydration of the samples the signal at 30 ppm is converted into a signal at 55 ppm. Hydration of the sample is not likely to reduce the aluminium coordination number from five to four, however. Considering this reversible interchange of intensity of the 30 ppm resonance to the 55 ppm resonance upon hydration, it is unlikely that the signal at 30 ppm originates from five-coordinated aluminium species. Ray and Samoson [17] have shown that hydration of (dehydroxylated) aluminas leads to a irreversible reconstruction of the surface favouring AI(W) at the expense of Al(W) and Al(V). In a
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
previous study of aluminas and aluminoborates we also found a not completely reversible conversion of tetra- and penta-coordinated Al to hexa-coordinated aluminium [33]. De Witte et al. [14] observed a strong increase of the Al(IV) and Al(V1) resonance upon hydration in aluminosilicate gels, but hardly any effect on the intensity of the Al(V) resonance. Taking these observations into account, we think that it is more likely that the signal at 30 ppm in the present sample is partly due to tetrahedrally coordinated aluminium atom in a strained environment. There are two possibilities for a tetrahedral aluminium to end up in the 40-30 ppm area. One possibility is that its isotropic chemical shift lies in the normal range of 50-70 ppm, but due to a large quadrupolar interaction the line is shifted strongly. This has been observed in the de-aluminated zeolites by Ray and Samoson [17]. In our samples dehydration probably leads to a closer distance between the network aluminium and the charge compensating alumina particles, which might induce large field gradients. This was also observed in the drying of zeolites, where the cation gets located close to the tetrahedral site. Another possibility for a tetrahedral aluminium to end up in the 40-30 ppm area is that the isotropic chemical shift of the resonance is very low because of a strained conformation of these sites in the aluminosilicate network. A relation between the Al-O-S1 angle and the 27A1 chemical shift has been established 1341. We found that this correlation can be extended for the mineral harkerite which has tetrahedral aluminium sites with an Al-O-S1 bond angle of 176”, resulting in a chemical shift of 44 ppm [35]. For the largest bond angle (180”) an isotropic chemical shift of 42 ppm is expected. In the hybrid aluminosilicates studied here, a large amount of strain in the network can be expected because the Al in the aluminosilicate network has to be close to the charge compensating alumina in the sample. These electrostatic forces in the material are thought to induce a strain in the material leading to Al(IV) with deviant Al-O-% angles. This might also explain the occurrence of the 80 ppm shoulder and the 30 ppm line in the Al spectra upon dehydration. Hydration probably reduces the strain in the material. That hydration results in diminished strain in the samples is also observable by 29Si MAS NMR. Here hydration of the samples prior to measuring the 29Si MAS NMR
211
spectra of annealed samples results in a slightly increased resolution. In order to gain more insight on the nature of the various observed Al-sites, we tried to obtain extra information on the 27A1 NMR resonances with the use of the newly developed multiple-quantum MAS NMR [36] and off-resonance nutation NMR [28]. Off-resonance nutation NMR was preferred above normal Nutation NMR since the quadrupolar interactions in the samples are expected to be high. 3.4. 2%1 MQMASNMR The idea of the MQMAS experiment is to refocus second-order quadrupolar broadenings in MAS NMR spectra by creating echoes of multiple-quantum transitions and the directly observable central transition. After its introduction by Medek et al. [26] and Frydman and Harwood [36], the experiment soon found its way in a number of applications. As the experiment has been discussed extensively in all these papers, we only describe the main spectral features in our representation, which is without shearing of the spectra. We performed 3Q-MAS experiments on 27A1(I = 5/2) nuclei. In these experiments the resonances are situated on the line vi = 3v, in the absence of a quadrupolar interaction. The spectra are therefore plotted in such a way that the line marked ui,, (vi = 3v,) is the diagonal in these 2D spectra. In the presence of a quadrupolar interaction the center of gravity of the resonance is shifted from the diagonal by the quadrupolar induced shift (QIS) in the direction v, = 3/4v2, indicated in the spectra by a solid-dashed line. Furthermore the line is now broadened by the second-order quadrupolar interaction along the direction v, = 19/ 12 v2, indicated by a dashed line in the spectra marked Anisotropy. Fig. 6a displays the 3Q-MAS NMR of an 0.20 Al(OBu”),EAA/0.40 PhTES/0.40 TEOS sample after annealing at 200°C. The improved resolution of the experiment is obvious. The lines around 55, 30 and 0 ppm are now well separated. The line shapes look very similar to those obtained for y-Al,O, in a previous study [37]. The lines run parallel to the isotropic shift axis at the low-field side indicating a distribution in isotropic chemical shift. At the highfield side the lines bend away from the diagonal in
212
M.P.J. Peeters, A.P.M. Kentgens/Solid
u
A
0
State Nuclear Magnetic Resonance 9 (1997) 203-217
B
(5
is0
is0
Anisotropy i
V,,,,,,,,‘~,,,,,,,,,,,,,,,,,,,i
120
100
80
60
40
20
0
-20
120
100
80
60
(PI%
2o
0
Fig. 6. (a) 27A1 3Q-MAS spectrum of a 0.20 Al(OBuS),EAA/0.40 PhTES/O.40 TEOS/4 EtOH/3.4 HCl (0.005 M) gel after heating at 200°C for 2 h (b) 2’Al 3Q-MAS spectrum of a 0.20 Al(OBuS),/0.40 PhTES/0.40 TEOS/0.30 EAA/4 EtOH/3.4 HCI (0.005 M) gel after heating at 200°C for 2 h and subsequent hydration. The line at 105 ppm is due to an Al impurity in the Si,N, rotor.
the quadrupolar induced shift direction, reflecting a distribution in quadrupolar interaction. In order to get an idea of the size of the distributions involved we decided to simulate the line shapes using Gaussian distributions in both isotropic chemical shift and quadrupole coupling constant. We realise that the results of such an analysis should be treated with caution; the excitation of sites with differing quadrupole interactions is non-uniform, and not fully described yet. So far the excitation behaviour has been described assuming a static Hamiltonian during the pulses [26,27]. Considering these results, our r-f-field strength and expected quadrupole coupling constants of several MHz lead to a 3Q-excitation that is fairly inefficient but does not vary greatly with C qcc. A further point of concern is that the shape of the distribution in isotropic shift and quadrupolar interactions need not to be Gaussian. Depending on the structural variations (i.e., variation in bond angles and distances, position of charge balancing cations etc.) in the materials, different distributions of the observable NMR parameters can be expected. Despite these restrictions we calculated MQMAS spectra with varying Gaussian distributions in
isotropic chemical shift and quadrupole coupling constants with the sole purpose to get a rough impression of the ranges that are encountered in these materials. As can be seen in Fig. 7 we were able to get a satisfactory agreement with the experimental spectrum using this approach. Small discrepancies between the experimental and theoretical line shapes are attributed to the above-mentioned shortcomings of the analysis, and/or small phase errors in the experimental spectrum. For the Al(IV) resonance we used an isotropic chemical shift distribution centered at 60 ppm with a half width at half height (HWHH) of 5 ppm, in combination with an (uncorrelated) distribution of 6 + 2 (HWHH) MHZ for C,,,. For Al(V) we found ais = 33 + 3 ppm and C,,, = 4 + 0.7 MHz. The Al(V1) line was simulated using ojsO = 7 + 4 ppm and Cqcc = 4.5 + 1.5 MHz. The asymmetry parameter was taken 1 in all cases. The calculations are not extremely sensitive to variations in 71~ but did show that 774 is large. Considering the relative insensitivity we did not take a variation in 71, into account. The 3Q-MAS spectrum of a hydrated sample (Fig. 6b) shows a diminished intensity of the Al(V)
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
easier seen by looking at the F,-projections (i.e., onto the normal MAS-axis) of the Al(W) resonance in the samples. These projections are shown, for the sample of Fig. 6a dried at 200°C in Fig. 8a for a sample dried at 120°C (Fig. 8b) and the same sample after hydration (Fig. 8~). Clearly, the resonance of Al(W) in the hydrated sample is narrow with most intensity in the range from 60-40 ppm, there is only a weak tail to higher field (lower ppm values). Dehydration strongly broadens the resonance, which is most pronounced for the sample annealed at 200°C where the line has a significant intensity down to 0
-20 0 20
40
/ Anisotropy
60
,
80 100
// 1 20
I
100
I
I
80
60
40
20
213
I
I
0
-20
Fig. 7. Simulation of the *‘Al 3Q-MAS spectrum of a 0.20 AI(OBuS),EAA/0.40 PhTES/0.40 TEOS/4 EtOH/3.4 HCl (0.005 M) gel after heating at 200°C. Parameters: Al(W): a,,, centered at 60 + 5 ppm (half width at half height), in combination with an (uncorrelated) distribution of 6+2 (HWHH) MHz for C,,,. AI(V): rrlsO = 33 * 3 ppm and Cscc = 450.7 MHz. AI( q so = 7 + 4 ppm and Cqcc = 4.5 + 1.5 MHz. The asymmetry parameter was taken 1 in all cases.
resonance. In this particular case of a sample with a small initial amount of Al(V), the resonance vanishes, but in other samples with higher initial Al(V) signal, the resonance remains present, without showing great variation of the isotropic shift and quadrupole parameters as compared to the dehydrated sample. A superficial inspection of the Al(W) and Al(N) resonances in the hydrated and the dehydrated sample indicate no great changes as the contours span the same region. For the Al(W) resonance this is indeed true, only the high-field tail of the resonance seems a little bit smaller. A closer inspection of the Al(W) resonance around 60 ppm shows, however, that its maximum is shifted closer to the diagonal, compared with the dehydrated sample, suggesting a reduced average C,,, for this site. Indeed, it was not possible to simulate this peak with a single Gaussian distribution in giisO and C,,,. A second distribution with a much smaller average C,,, = 4 + 1 MHz and giis, = 60 + 3 had to be added to the one used for the dehydrated sample. This effect is much
ppm. These results explain why we observe the interchange of signal intensity from the Al(V) region to the Al(W) region in the normal MAS spectra upon hydration. In fact there is only a small amount of Al(V) present in the materials (maximally 10%). This signal is added up to the strong high field flank of the Al(W) resonance in the dehydrated samples. Upon hydration, the Al(N) resonance narrows causing the intensity in the 40-16 ppm region to decrease. Although the intensity of the Al(V) resonance also decreases upon hydration, this is only a minor effect as there is only a small amount present. The precision of the integration procedure prohibits us to determine whether this Al(V) is converted to Al(V1) or ANIV). The Al(W) resonance is only mildly affected by the hydration/dehydration process. The observations are in accordance with the model proposed earlier. The fourfold-coordinated aluminium is incorporated in the aluminosilicate network. The remaining aluminium serves as chargecompensating cations, mostly in the form of tiny alumina particles where most of the aluminium is octahedrally coordinated. The quadrupole parameters observed for the Al(W) resonance are very similar to those observed in alumina. In a hydrated sample these particles are solvated allowing the polymer network some mobility, as is reflected in the line width of the Al(W) resonance. Upon dehydration, however, much more strain is induced in the network because of the electrostatic forces of the alumina particles on the Al in the network. This attraction of the network Al to the charge compensating alumina particles has two effects; first of all different conformations occur in the network. Indeed, an increase in the isotropic chemical shift distribution is observed.
214
M.P.J. Pee&m, A.P.M. Kentgens/Solid
180
140
100
State Nuclear Magnetic Resonance 9 (1997) 203-217
60
20
0
-40
-80
(w-N Fig. 8. Projections of the 27A1 3Q-MAS spectrum along the F2 direction of the tetrahedrally coordinated aluminium. The F, spinning sideband of the octahedrally coordinated aluminium is indicated by an asterisk. (a) 0.20 Al(OBu”),EAA/0.40 PhTES/0.40 TEOS/4 EtOH/3.4 HCl (0.005 M) gel after heating at 200°C for 2 h, (b) 0.20 AI(OBuS),/0.40 PhTES/0.40 TEOS/0.30 EAA/4 EtOH/3.4 HCI (0.005 M) gel after heating at 200°C for 2 h, hydration and subsequent dehydration at 120°C for 2 h and (c) 0.20 AI(OBu”)s/0.40 PhTES/O.40 TEOS/O.30 EAA/4 EtOH/3.4 HCl (0.005 M) gel after heating at 200°C for 2 h and hydration.
This is expected as the isotropic correlated to the Al-0-Si bond the charge-compensating cations closer to the network aluminium, ence much larger electric field witnessed by a strong increase quadrupole coupling constant. 3.5. “Al off-resonance In order parameters we tried quadrupole
chemical shift is angles. Secondly, get located much which thus experigradients. This is in the average
nutation NMR
to get a verification of the quadrupolar as estimated from the MQMAS spectra nutation spectroscopy [38-401. The interactions encountered are fairly strong
so we had to resort to off-resonance nutation spectroscopy [31]. In a study of a series of aluminosilicate glasses it was demonstrated that off-resonance nutation can be used in amorphous systems to get information about the average quadrupole parameters [41]. Due to the fact that off-resonance nutation mainly relies on line intensities to discriminate between different sites well resolved spectra can still be obtained in systems (e.g., glasses) where a distribution of quadrupolar interactions is present. Thus the average quadrupole coupling constant can be obtained but there is no information about the details (i.e., width and shape) of the distribution of these parameters. “Al off-resonance nutation spectra were
M.P.J. Peeters, A.P.M. Kentgens/Solid
Nutation
State Nuclear Magnetic Resonance 9 (1997) 203-217
Frequency
215
(Hz)
Fig. 9. Off-resonance nutation NMR spectra of the 0.20 AffOBu”),/0.20 EAA/0.40 PhTES/0.40 TEOS sample after annealing at 200°C (dry sample). Resonance offset = 190 kHz, rf field strength = 38 kHz. Simulation of the signal at 55 ppm (trace a), experimental spectrum (trace b), Simulation of the signal at 0 ppm (trace c), experimental spectrum (trace d). The uninformative signal at zero-frequency is truncated in the experimental spectra. Parameters used for the simulations are listed in Table 2.
taken for two samples: 0.20 Al(OBu”),/0.20 EAA/0.40 PhTES/0.40 TEOS after annealing at 200°C and after hydration. Resonance offsets used were between 120 and 260 kHz with an rf-field strength between 38 and 48 kHz, using a spinning speed of 2 kHz. As has been discussed before, nutation spectra should either be taken statically or at a spinning speed that is low enough to ensure that the sample can be assumed static during the nutation pulse. In practice no effects of the spinning on the nutation spectra is observed when the longest t,period does not exceed l/4 rotor-period. In the present case this is a serious drawback as the resonances partly overlap even at high spinning speeds. At low speeds this only gets worse as spinning sidebands appear as well. As has been noted in the MQMAS spectra the tail of the Al(W) resonance strongly overlaps with the (low intensity) Al(V) resonance. Therefore, it is not possible to get a reliable value for the average C,,, of Al(V) in the dehydrated sample. An example of an off-resonance nuta-
tion spectrum, together with the simulation is given in Fig. 9 for the dry sample. The full set of parameters used for the simulation of all spectra are listed in Table 2. In the dehydrated sample the signal at 60 ppm is characterised by an average quadrupole coupling constant of 5.5 MHz. Considering the accuracy of the method this agrees well with the results obtained from the MQMAS spectrum (6.0 MHz). The slightly lower value might be due to the slow spinning which has the disadvantage that the lines with the largest quadrupole interactions contribute less to the central line shape [42]. As was already mentioned, spectral overlap prohibits us to extract information about the Al(V) resonance. Good agreement is observed for the average quadrupole coupling constant derived from the MQMAS (4.5 MHz) and off-resonance nutation NMR spectrum (4.0 MHz) for the signal of the octahedral aluminium. The off-resonance nutation again giving a slightly lower value. Although the off-resonance nutation NMR spectra are not so
216
M.P. J. Peeters, A.P.M. Kentgens / Solid State Nuclear Magnetic Resonance 9 (1997) 203-217
Table 2 Off-resonance nutation NMR performed on 0.20 Al(0Bt1”)~ /0.20 EAA/0.40 PhTES/O.40 TEOS sample (dry) after annealing at 200°C and after hydration. Parameters used for the simulations of the projections along the Fl dimension Sample state
Resonance offset (kHz)
RF-strength (kHz)
6 (ppm)
$Hz)
7:)
Dehydrated Dehydrated Dehydrated Dehydrated Dehydrated Dehydrated Dehydrated Dehydrated Hydrated Hydrated Hydrated Hydrated
120 190 220 260 120 190 220 260 150 220 150 220
38 38 36 38 38 38 36 38 48 38 48 38
0 0 0 0 55 55 55 55 0 0 55 55
4.0 4.0 3.9 4.0 5.7 5.7 5.5 5.3 3.5 3.7 3.9 3.8
1 1
I 1 1
1 1 1 1 1 1 1
sensitive for the determination of v~, it is clear from the simulations that 71~is close to 1 for all sites. For the hydrated sample a reduced quadrupole interaction was found for the Al(W) resonance in the MQMAS spectra (4.0 MHz). The off-resonance nutation spectra corroborate this result, giving a C,,, of 3.9 MHz. For the octahedrally coordinated aluminium species a Cscc 3.6 MHz is obtained by off-resonance nutation NMR, again in accord with the MQMAS result of 4.0 MHz.
4. Conclusions
The effects of aluminium incorporation in TEOS based hybrid materials was studied by MAS NMR. After RT drying of the sol-gel synthesised materials, octahedral and tetrahedral aluminium exist in the materials; the formation of tetrahedrally coordinated aluminium is favoured by a low water concentration. In all materials prepared the ratio of tetrahedral to octahedrally coordinated aluminium is lower than three. Some of the octahedrally coordinated aluminium must therefore reside in an alumina rich phase. Drying of the aluminosilicate gels results in an decrease of the amount of octahedrally coordinated aluminium and the formation of a new resonance at 30 ppm. Several papers [13-151 ascribe this resonance to pentahedrally coordinated aluminium
ions, at the interface between the aluminosilicate and the aluminate region. The fivefold-coordination is stabilised by the interfacial strain, or alternatively minimises this strain. Storing the samples in air showed a change in the observed signal intensities indicating hydration effects. These effects were investigated by 27A1MAS, MQMAS and off-resonance nutation NMR. In the MAS spectra the intensity of the 30 ppm signal decreases upon hydration, and a congruent increase in intensity of the 55 ppm signal is observed. This hydration/dehydration process is fully reversible, ruling out the possibility of atomic rearrangements on a large scale. Drying at 120°C results in an increase of the intensity of the signal at 30 ppm. Drying at this temperature is expected to remove only the physisorbed water. It is therefore unlikely that the signal at 30 ppm is caused only by a fivefold-coordinated aluminium species. MQMAS experiments clearly demonstrated that most of the intensity in the 40-15 ppm region is due to the asymmetric tail of tetrahedrally coordinated aluminium species due to two causes: the average C,,, is very large, leading to large quadrupole induced shifts, and a great variation in the isotropic chemical shift. This increase in chemical shift distribution is attributed to a greater dispersion of conformations in the polymer network. This is expected when the network aluminium is electrostatically attracted to the charge-compensating alumina particles. The closer location of these charged particles increases the field-gradients felt by the network aluminium resulting in the increased C,,,. The off-resonance nutation data corroborated the qualitative analysis of the MQMAS spectra.
Acknowledgements Mr. H. Janssen, Mrs. G. Nachtegaal and Mr. J. van OS of the SON/NW0 HF-NMR Facility and Mr. M. van Bommel and Mrs. Snijkers-Hendrickx of the Philips Research Laboratories are greatly acknowledged for their various contributions to this work.
M.P.J. Peeters, A.P.M. Kentgens/Solid
State Nuclear Magnetic Resonance 9 (1997) 203-217
References 111 I. Jaymes, A. Douy, J. Eur. Ceram. Sot. 16 (2) (1996) 155.
t21I.
Jaymes, A. Douy, D. Massiot, J.P. Busnel, J. Am. Ceram. Sot. 78 (10) (1995) 2648. [31 I. Jaymes, A. Douy, D. Massiot, J.P. Coutures, J. Chim. Phys. Phys.-Chim. Biol. 92 (10) (1995) 1851. [41 Y.F. Chen, S. Vilminot, Mat. Res. Bull. 30 (3) (1995) 291. (51 T. Heinrich, F. Raether, H. Marsmann, J. Non-Cryst. Solids 168 (1994) 14. 161 U. Selvaraj, S. Komarneni, R.J. Roy, J. Am. Ceram. Sot. 73 (12) (1990) 3663. [71 C. Sanchez, J. Livage, M. Henry, F. Babonneau, J. Non. Cryst. Solids 100 (1988) 65. t81 F. Babonneau, L. Coury, J. Livage, J. Non. Cryst. Solids 121 (1990) 153. F. Babonneau, J. Livage, J. Sol-Gel I91 L. Bonhomme-Coury, Sci. Technol. 3 (1994) 157. [lOI W.L. Olson, L.J. Bauer, Mat. Res. Sot. Proc., D.E. Clark, D.K. Ulrich (Eds.) 73 (1986) pp. 187-193. [Ill L. Bonhomme-Coury, F. Babonneau, J. Livage, Chem. Mater. 5 (1993) 323. [121G.A. Pozamsky, A.V. McCormick, J. Non-Cryst. Solids 190 (1995) 212. iI31 A.D. Irwin, J.S. Holmgren, J. Jonas, J. Mater. Science 23 (1988) 2908. [141 B.M. de Witte, P.J. Grobet, J.B. Uytterhoeven, J. Phys. Chem. 99 (1995) 6961. [151 F. Hatakeyama, T.J. Maekawa, J. Ceram. Sot. Jpn. 100 (21 (1992) 163. [I61R. Dupree, I. Farnan, A.J. Forty, S. El-Ma&i, L. Bottyan, J. Phys. Colloq. c8 (1985) 113-117. 071 G.J. Ray, A. Samoson, Zeolites 13 (1993) 410. [181 D. Coster, A.L. Blumenfeld, J.J. Fripiat, J. Phys. Chem. 98 (1994) 6201. [191 A. Taylor, D. Holland, J. Non-Cryst. Solids 152 (1993) 1. DOI P.J. Sanchez-Soto, I. Sobrados, J. Sanz, J.L. Perez-Rodriguez, J. Am. Ceram. Sot. 76 (12) (1993) 3204. WI J.J. Fitzgerald, S.F. Dee, A.I. Hamza, Am. Miner. 74 (1989) 1405.
WI G. Kunath-Fandrei,
217
T.J. Bastow, J.S. Hall, C. Jager, M.E. Smith, J. Phys. Chem. 99 (1995) 15138. b31 C. Femandez, J.P. Amoureux, Solid State Nucl. Magn. Reson. 5 (1996) 315. t241 D. Massiot, B. Touzo, D. Trumeau, J.P. Coutures, J. Virlet, P. Florian, P.J. Grandinetti, Solid State Nucl. Magn. Reson. 6 (1996) 73. [251 G. Wu, D. Rovnyank, B. Sun, R.G. Griffin, Chem. Phys. Lett. 249 (1995) 210. [261 A. Medek, J.S. Harwood, L. Frydman, J. Am. Chem. Sot. 117 (1995) 12779. 1271J.P. Amoureux, C. Femandez. L. Frydman, Chem. Phys. Lett. 259 (1996) 347. I281 A.P.M. Kentgens, J. Magn. Reson. 95 (1991) 619. t291 V.B. Cheng, H.H. Suzukawa, M.J. Wolfsberg, J. Chem. Phys. 59 (19731 3992. I301 H. Conroy, J. Chem. Phys. 47 (1967) 5307. [311 A.P.M. Kentgens, J. Magn. Reson., Ser. A 104 (1993) 302. [321 S. Simon, G.J.M.P. van Moorsel, A.P.M. Kentgens, E. de Boer, Solid State Nucl. Magn. Reson. 5 (1995) 163. t331 S. Simon, A. var der Pol, E.J. Reijerse, A.P.M. Kentegens, G.J.M.P. van Moorsel, E. de Boer, J. Chem. Sot. Faraday Trans. 90 (18) (1994) 2663. t341 E. Lippmaa, A. Samoson, M. Magi, J. Am. Chem. Sot. 108 (1986) 1730. [351 P.J. Dirken, A.P.M. Kentgens, G.H. Nachtegaal, A.M.J. van der Eerden, J.B.H. Jansen, Am. Miner. 80 (1995) 39. [361 L. Frydman, J.S. Harwood, J. Am. Chem. Sot. 117 (1995) 5367. [371 H. Kraus, R. Prim, A.P.M. Kentgens, J. Phys. Chem. 100 (1996) 16336. [381 A. Samoson, E. Lippmaa, Chem. Phys. lett. 100 (1983) 205. [391 A. Samoson, E. Lippmaa, Phys. Rev. B 28 (1983) 6567. I401 A.P.M. Kentgens, J.J.M. Lemmens, F.M.M. Geurts. W.S. Veeman, J. Magn. Reson. 71 (1987) 62. I411 P.J. Dirken, G.H. Nachtegaal, A.P.M. Kentgens, Solid State Nucl. Magn. Reson. 5 (1995) 189. [421 R.K. Sato, P.F. McMillan, P. Dennison, R. Dupree, J. Phys. Chem. 95 (1991) 4483.