Solid-state nuclear magnetic resonance studies of acid sites in

SOLID STATE Nuclear Magnetic Resonance

Solid State Nuclear Magnetic Resonance 3 (1994) 271-286

Solid-state nuclear magnetic resonance studies of acid sites in zeolites 1

D. Freude *, H. Ernst, I. Wolf

Abteilung Grenzflichenphysik, Uniuersitiit Leipzig, Linnbtrasse 5, D-04103 Leipzig, Germany

Received 12 April 1994; accepted 10 May 1994

Abstract

‘H Magic-angle spinning nuclear magnetic resonance (MAS NMR), echo 27A1 NMR, two-dimensional (2D) echo ‘H MAS NMR and ‘H, 27A1 and 29Si MAS NMR have been applied to study the acid sites of dehydrated zeolites. The quadrupole coupling constants Cot- determined from 2H MAS NMR spectra of Si-OH-AI sites increase with the framework aluminium content of the zeolites from 208 kHz (H-ZSM5) to 236 kHz (H-X and H-Y) due to the decreased acid strength of the bridging OH groups. The 27A1 signal of the Si-OH-AI sites, which was considered “NMR-invisible” in the past, yields Cocc = 16 MHz for zeolite H-ZSM-5. The majority of non-framework aluminium species could also be observed in dealuminated and dehydrated zeolites H-ZSM5 giving Cocc = 9 MHz. 2D echo ‘H MAS NMR spectra yield values of 16-40 ms for the lower limits of the lifetime of hydroxyl species at room temperature. A lifetime of more than 40 ms was obtained from echo 2H MAS NMR spectra for protonated sites giving a signal at ca. 6.5 ppm in partially dealuminated and weakly rehydrated samples.

Key words: ‘H, *H magic-angle spinning nuclear magnetic resonance; 27A1 nuclear magnetic resonance; Acidity; Zeolite Y; ZSMS

1. Introduction

Surface sites capable of donating protons or accepting electrons from adsorbed molecules are essential for the phenomenon of heterogeneous catalysis. The strong acidity of zeolites is generated by bridging hydroxyl groups. Their strength of acidity depends on the electronic charge of the hydrogen atom which can be measured by means of ‘H or *H MAS NMR spectroscopy of dehy-

’ This paper is dedicated to Professor Harry Pfeifer on the occasion of his 65th birthday.

* Corresponding author.

drated samples. *‘Al NMR is capable of investi- gating the nature of non-framework aluminium species, which play an important role for the Lewis acidity of partially dealuminated zeolites. NMR studies of the adsorption and reaction of probe molecules in zeolites can elucidate the acidic properties and also the nature of catalytic activity. The number of published NMR studies of zeolites increased from ca. 400 in 1984, when the first review was given by Klinowski [l], to ca. 1600 papers now. The most recent review is presented by Pfeifer and Ernst [2].

Brplnsted sites in zeolites are “bridging” Si- OH-Al hydroxyl groups of the aluminosilicate framework. They can be characterized by various

0926-2040/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved SSDI 0926-2040(94)00019-9

272 D. Freude et al. /Solid State Nuclear Magnetic Resonance 3 (1994) 271-286

NMR parameters: chemical shift, intensity, dipole broadening and quadrupole parameters (spin > l/2). The bare sites can be studied by ‘H, *H (after H-D exchange), 170 [3], 29Si [4] or *‘Al NMR measurements of dehydrated (also denoted as activated or calcined) samples.

‘H NMR studies of the geometry of Bronsted sites 14-81 yield the proton-aluminium distances in Si-OH-Al sites of various zeolites in the range 0.23-0.25 nm. Hunger et al. [4] showed that this distance depends on the size of the oxygen rings.

Vega and Luz [9] measured the *H NMR powder pattern of the zeolite RHO exchanged with ND,f ions. Heating the samples to above 400°C without evacuation results in a spectrum with a spin = 1 quadrupole powder pattern which is due to rigidly bound hydroxyl groups and is described by a quadrupole coupling constant C oc- = 251 kHz and an asymmetry parameter of the electric field gradient (EFG) tensor n = 0. Gluszak et al. [lo] discussed the two components of the *H NMR signal of a deuterium-exchanged and partially dealuminated zeolite NH,-Y which was heated under vacuum to 450°C. They ascribed the quadrupole powder pattern to hydroxyl deuterons with Cocc = 234 kHz and 77 = 0.03 and a Gaussian-shaped signal with a FWHM (full-width-at-half-maximum) of 63.1 kHz to silanol groups. The application of *H MAS NMR to the hydroxyl groups in zeolites gave an increased sensitivity compared with the static measurements. The values of Cocc of the %-OH-Al sites increase with the framework aluminium content of the zeolites from 208 kHz (H-ZSM-5) to 236 kHz (H-X and H-Y) due to the decreased acid strength of the bridging OH groups [ill.

In the past, 27Al NMR (and 29Si NMR) experiments were carried out mainly on hydrated samples. From the strongly broadened signal of the 27Al nuclei in a dehydrated zeolite Na-Y the following quadrupole parameters could be determined [7]: C,,, = 6.8 MHz and TJ = 0.5. The signal of the Al nuclei being constituents of the Si-OH-Al sites in the dehydrated zeolites was considered “NMR-invisible” in the past. This is because of the heavy distortion of the tetrahedral or octahedral symmetry around the Al atom which causes a strong electric field gradient at the site

of the nucleus and strongly broadens the 27Al NMR signal. In a recent paper [ll] we could show that the 27A1 NMR signal of the bridging Bransted sites becomes visible at 11.7 T with an echo Fourier technique yielding a quadrupole coupling constant Cocc = 16 MHz for zeolite H- ZSM-5.

Non-framework aluminium compounds, which are generated by calcination or steaming of zeolites, can act as Lewis acid sites. Aluminium atoms in non-framework positions in partially dealuminated zeolites are supposed to be the source of their enhanced catalytic activity [12,13]. Numerous 27Al MAS NMR studies were performed on partially dealuminated hydrated zeolites [2]. Three lines at ca. 60, 30 and 0 ppm were assigned to four-coordinated framework aluminium, four-coordinated non-framework aluminium and six-coordinated non-framework aluminium, respectively [13]. Kellberg et al. [14] and Rocha et al. [15] monitored the non-framework aluminium species in hydrated ultrastable zeolites Y by ‘H-*‘Al cross-polarization (CP) MAS NMR and found CP signals at ca. 0, 30 and 59.3 ppm [14] or 56 ppm [15]. However, conclusions about the nature of Lewis sites or sites of enhanced catalytic activity in zeolites cannot be drawn from 27Al NMR studies of hydrated samples. The compounds are changed upon dehydration and the 27Al NMR signal of non-framework aluminium species in activated (dehydrated) samples was “NMR-invisible”.

Evidence for the existence of Lewis acid sites can be obtained by NMR measurements of probe molecules 121, e.g. a small amount of water molecules. A ‘H MAS NMR signal at 6.5 ppm was observed by Hunger et al. [16] in the spectrum of a dealuminated zeolite Y with water loading up to 10 molecules per unit cell (u.c.). The maximum intensity of this line corresponded to a concentration of 2 H,O/u.c. [16]. While Hunger et al. [16] proposed three-coordinated silicon atoms as Lewis acid sites, Batamak et al. [17] attributed the signal to water molecules on Lewis acid Al atoms still partially bonded to the framework.

In this paper the 6.5-ppm signal in the ‘H spectrum is studied by a 2D echo MAS experi-

D. Freude et al. /Solid State Nuclear Magnetic Resonance 3 (1994) 271-286 273

ment. Echo 27Al NMR shows that in addition to the *‘A1 NMR signal of the Si-OH-Al sites also the signal of the majority of non-framework aluminium species becomes visible in the field of 11.7 T. For these signals, MAS or double resonance (DOR) cannot improve the resolution of different second-order quadrupole powder pat- terns in the *‘A NMR spectrum. However, for the first-order quadrupole broadened *H NMR signal of the dehydrated or weakly rehydrated (and deuterated) zeolites a 2D spectrum can be obtained under MAS. The 2D spectra allow the characterization of hydroxyl species by means of their chemical shift and first-order quadrupole shift.

2. Experimental

Zeolite H-Y 92 containing 8% of the original content of Na+ cations, prepared by a repeated exchange of Na-Y (Si/Al = 2.6) in an aqueous solution of NH,NO, followed by calcination, was provided by Dr. J. Meusinger (Leipzig, Germany). H-Y 85 was prepared from zeolite Y @i/Al = 2.4) containing 85% NH: and 15% Na+. This zeolite and zeolite Na-Y were provided by UOP. Zeolite H-X 50 (Si/Al = 1.4) containing 50% of the original content of sodium ions was provided by Dr. U. Lohse (Berlin, Germany). Zeolites Na-ZSM-5 and H-ZSMJ with Si/Al = 22 were synthesized by Dr. W. Schwieger (Halle, Germany) using mono-n-butylamine as template, and then exchanged with 0.5 M HCl.

Samples for ‘H and *H MAS NMR were prepared by heating g-mm-deep layers of zeolite in glass tubes of 4 or 5 mm diameter. The temperature was increased at a rate of 10 K h-i under vacuum. After maintaining the samples at 400°C and at a pressure less than lo-* Pa for 24 h, several samples were partially deuterated by con- tact with 2 kPa D,O at 230°C for 30 min, followed by evacuation. This procedure was repeated eight times followed by a final evacuation at 300°C for 12 h. Some samples were loaded with a small amount of water at room temperature. All samples were finally loaded under a pressure of 13 kPa 0, and sealed off.

Adsorption of 0, shortens the spin-lattice relaxation times Ti of ‘H and *H, so that the MAS NMR spectra could be acquired with a l-s recycle delay. Without oxygen, the proton longitudinal relaxation times Ti of the OH groups are ca. 10 s, while an average value of 26 s has been determined for the corresponding deuteron Ti [lo]. The adsorption of 0, under a pressure of 13 kPa does not modify the ‘H MAS NMR spectra measured at room temperature.

The same exchange treatment at 230°C except for the use of H,O instead of D,O, has been carried out on a zeolite H-Y 85 sample. By com- paring the ‘H MAS NMR spectrum of this sample with the spectrum of a sample which had not been treated at 23o”C, it could be shown that the deuteration did not change the concentration of the hydroxyl groups in the zeolite.

In order to obtain a signal from residual water molecules and residual ammonium ions, a zeolite H-Y 92 sample was evacuated at a maximum temperature of 230°C only and denoted as H-Y 92/230.

For 27A1 and 2ySi NMR measurements on dehydrated samples, the zeolites were evacuated as described above at 4Oo”C, then loaded with ca. 300 mg of n-octane per gram of zeolite under vacuum and sealed off. After annealing the samples at 150°C for 2 h the zeolites were fully immersed in octane. Before the measurements the glass tubes (10 mm diameter) were opened under air, and the samples were transferred into MAS sample containers.

Some 27Al MAS NMR measurements were performed on as-prepared samples and samples which had been activated and then rehydrated and kept in a desiccator for 48 h over aqueous NH&l. To achieve partial rehydration of the zeolite H-Y 92 the samples were loaded at room temperature under vacuum with known amounts of doubly distilled water, sealed and annealed at 150°C for 4 h.

NMR measurements were generally carried out at room temperature using a Bruker MSL-500 spectrometer. The 27Al pulse length was 0.5 ps, which corresponded to a a/6 pulse for non- selective excitation (a/2 pulse for selective excitation of the central transition [18]). The radio-


frequency (rf) power used corresponded to a (non-selective) nutation frequency of V~ = 167 kHz, so for all sample prf exceeded the spectral width of the central transition. If the latter value exceeded 50 kHz, a phase-cycled selective echo

sequence [I91 (~/'&.5ps- delay,, cor 50) wLs -rIgLs) was used, starting the acquisition 20 (or 50) PCS after the second pulse. For ‘H MAS NMR (v,,~ = 12.5 kHz) and *H MAS NMR (v,,~ = 5 kHz), acquisition was initiated one rotation period after the single pulse. The number of scans varied from 1000 for ‘H to 100000 for *H MAS NMR.

3. Results

Some of the ‘H MAS NMR spectra of the activated zeolites are given in Fig. 1. The different signals are denoted by a-f [2,4,5]. Line a at 1.9 ppm is due to non-acidic SiOH (or silanol) groups at framework defects and in amorphous parts of the sample. Line b at 3.7 ppm for H-X

50,3.9 ppm for H-Y 85,4.0 ppm for H-Y 92 and 4.3 ppm for H-ZSM-5 is due to bridging OH groups pointing into the large zeolite cages or into the intersections, respectively. Line c at 4.8- 4.9 ppm in the spectra of zeolites H-Y is caused by bridging OH groups pointing into the sodalite cage. Line d at 7 ppm is assigned to residual ammonium cations. Line e at ca. 2.7 ppm represents hydroxyl groups associated with non-framework aluminium species and bonded to framework oxygen atoms. Line f is caused by the adsorption of a small amount of water on Lewis acid sites in partially dealuminated zeolites Y, as observed for the first time at 6.5 ppm in Ref. [16]. The spectrum of sample H-Y 92 loaded with 8 water molecules/u.c. shows for line f two peaks at 6.9 and 7.4 ppm. These peak positions are independent of the water loading in the range 4-16 molecules/u.c. For the loading of 8 molecules/u.c., the intensity of line f corresponds to a concentration of ca. 8 hydrogen nuclei/u.c. The ‘H spectrum of the sample measured with-

A (D)

,,.I,,. I.,,,. II,.,

12 a 4 0 -4

B NC=1

E (d) (W

8, .I., . , . I ., ., . , * ,

12 a 4 0 -4

NC = 1 (b) iL_ w

12 a 4 0 -4

Fig. 1. ‘H MAS NMR spectra of samples. (A) H-Y 92; (B) H-Y 92 deuterated (after one week); (C) H-Y 92 deuterated (after eight months); (D) H-Y 92 with 8 H,O/u.c.; (E) H-Y 92/230; (F) H-Y 85. The two observable spinning sidebands in the distance of 12.5 kHz (ca. 30 and - 20 ppm) contain less than 10% of the total intensity of the spectra. Concentrations of species can be determined from the spectra after multiplication of their intensities with the normalization constant NC given for each of them in the figure.

D. Freude et al. /Solid State Nuclear Magnetic Resonance 3 (I 994) 271-286 275

out MAS shows besides a non-resolved broad signal of the species giving rise to lines a, b, c, e and f in the MAS spectrum a narrow signal at ca. 4.9 ppm due to mobile water molecules. This signal is superimposed on line c in the MAS spectrum. The spinning sidebands, not shown in Fig. 1, contain less than 10% of the total intensity of the spectra. By a comparison of the signal intensities of deuterated and non-deuterated samples an exchange degree of 75-95% of the hydrogen species has been found.

2D ‘H MAS NMR experiments were performed with a Hahn echo sequence (7r/2, t,, r, t,, t,). The time t, was varied in multiples of the rotation period (80 ps). Stack plots of the spectra are given in Fig. 2 with ppm and Hz scales for the o2 and wr dimensions, respectively. The line- shapes in the 6.1~ dimension are Lorentzian with FWHM = lo-30 Hz. The relaxation time Tph”, which determines the echo decay, can be obtained from 7’phn /s = (r x FWHM/Hz)-‘. The values of Tphn determined in this way are 20 + 4 ms (H-Y 85, line b), 20 + 4 ms (H-Y 85, line c), 40 rt 8 ms (H-Y 92, line a), 20 + 4 ms (H-Y 92, line b), 20 f 4 ms (H-Y 92, line c), 25 k 5 ms (H-Y 92, line e), 18 + 4 ms (H-Y 92, line f at 6.9 ppm) and 12 + 3 ms (H-Y 92, line f at 7.4 ppm).

Fig. 3A shows the 2H MAS NMR sideband pattern of sample H-Y 92/230. The structure of the spinning sidebands is lost in this presentation of the spectrum. Fig. 3B presents the same spectrum in a 2D contour plot created using the Bruker WINNMR software by dividing the sideband spectrum (32768 points) into single sideband files. The shape of the sidebands is plotted in the second dimension, representing the chemical shift, whereas the first (horizontal) dimension plots the order of the sidebands and represents the first-order quadrupole broadened (static) linewidth of species with the corresponding value of the chemical shift. This type of representation of a sideband spectrum by a 2D spectrum was used for the first time by Bliimich et al. [20]. z& represents the angular-dependent first-order quadrupole shift for the transitions - 1,0 and o,+ 1:

vb = +(3/8)Coc.e(3 cos2/3 - 1 + 7 sin’p cos 2a)

IO 0 (wm)

10 @pm) O

Fig. 2. Stack plots of 2D Hahn echo ‘H MAS NMR spectra of zeolites H-Y 85 (A) and H-Y 92 loaded with 8 H,O/u.c. (ES).

c oCC = e*sQ/h denotes the quadrupole coupling constant, where eQ is the electric quadrupole moment and eq the electric field gradient, i.e. the V,, component of the traceless EFG tensor in its principal axis system which is defined by the convention I V,, I 2 I VW I L I VI, I. The asymmetry parameter is defined by 71 = <V, - V&)/V,,. The angles LY and p are Eulerian angles of the rotation of the EFG principal axis system with respect to the laboratory system. The quadrupole frequency vo corresponds to the distance between the two singularities in a theoretical powder spectrum with 77 = 0. For spin = 1 nuclei vQ is equal to (3/4) Cocc.

276 D. Freude et al. /Solid. State Nuclear Magnetic Resonance 3 (1994) 271-286

W-1 Q’ *nz

Fig. 3. (A) ‘H MAS sideband pattern of sample H-Y 92/230. (B) Same spectrum in a 2D contour plot. The 2D plot was created using the Bruker WINNMR software by dividing the sideband spectrum (32768 points) into single sideband files. The shape of the sidebands is plotted in the second dimension, representing the chemical shift, whereas the first (horizontal) dimension represents the first-order quadrupole broadened linewidth of species having the corresponding value of the chemical shift. vb represents the angular-dependent first-order quadrupole shift for the both transitions - 1,0 and 0, + 1.

Fig. 4 shows contour plots of the ‘H MAS spectra of the samples H-Y 92 and H-Y 92 loaded with 8 H,O/u.c. Although from these figures the c occ values can be estimated only roughly, a MAS sideband simulation of Fenzke et al. [81 allows the exact determination of the quadrupole parameters. Mean values for the bridging hydroxyl groups, lines b and c, are listed in Table 1. From the spectra in Figs. 3 and 4 and spectra in

wm

Ref. [ll], five different species can be distinguished due to different quadrupole coupling parameters of the deuterium nuclei in the activated samples: bridging hydroxyl groups (Cocc = 208- 236 kHz, n = 0.10 f O.OS>, AlOH groups (Cocc = 67 kHz, 77 = 0.21, silanol groups (quadrupole broadened structureless signal with FWHM = 65 kHz), species giving rise to line f (weak quadrupole broadening, FWHM < 5 kHz) and

wm

100 50 0 -50

v;lkHz

-100

Fig. 4. Contour plots of the *H MAS spectra of samples H-Y 92 (A) and H-Y 92 loaded with 8 H,O/u.c (B). A sideband simulation allows the exact determination of the quadrupole parameters. Four different species can be distinguished due to different quadrupole coupling parameters: bridging hydroxyl groups (Cocc = 224 kHz, 17 = 0.13 f 0.05, see Table l), defect AlOH groups (Cocc = 67 kHz, n = 0.21, defect silanol groups (quadrupole broadened structureless signal with FWHM = 6.5 kHz), and species giving rise to signal f (weak quadrupole broadening, FWHM < 5 Wz).


Fig. 5. 14th spinning sideband of the *H MAS NMR spectrum of zeolite H-Y 92.

residual ammonium ions (weak quadrupole broadening, FWHM < 5 kHz). Fig. 5 demonstrates that in addition to the resolution of five species the signal of bridging hydroxyl groups can be resolved into lines b and c in the outer sidebands.

2D echo ‘H MAS NMR experiments (in anal- ogy to the ‘H experiments) yield small information because the echo is destroyed after some rotation periods by the quadrupole interaction. However, for line f (with weak quadrupole broad-

Fig. 6. “Al MAS NMR spectra of the as-synthesized zeohtes NH,-X 50 (A), NH,,H-Y 92 (0, NH,-Y 85 (El, H-ZSMJ (G) and the activated and then rehydrated forms of the zeolites H-X 50 (B), H-Y 92 (D), H-Y 85 (I?, H-ZSMJ (H).

ening) the echo can be observed and the echo envelope decay time is TZHah” = 40 + 10 ms.

Fig. 6 shows “A1 MAS NMR spectra of the as-synthesized zeolites and the activated and rehydrated forms of the zeolites. The framework aluminium gives rise to a narrow line at ca. 60 ppm, whereas the narrow line at ca. 0 ppm and the broad line at ca. 30 ppm, which is not significant in this representation, must be attributed to

Table 1 NMR parameters describing the hydrogen forms of the zeolites

Sample a H-X 50 H-Y 85 H-Y 92 H-ZSMJ

Si/Al 1.4 * 0.1 2.4 + 0.1 3.1 + 0.1 22+ 1

tH C.S. [int.] Line a 2.0 ppm [4.3 + 0.51 2.0 IO.5 + 0.11 ppm Line b 3.7 ppm [40 * 41 3.9 ppm [23 f 21 4.1 [21 + 21 ppm 4.3 [4 + 0.41 ppm Line c 4.8 ppm [25 f 2] 4.9 ppm [22 f 21 Line e 3.0 ppm [lo + 31 2.8 ppm [ 12 + l]

Co,,t*H) 236 + 10 kHz 236 + 10 kHz 224 f 10 kHz 208 f 10 kHz ?1 0.10 f 0.05 0.06 f 0.05 0.13 + 0.05 0.15 + 0.05

Co,,(27A1) 10.8 f 1.0 MHz 13.1 + 1.0 MHz 12.7 + 1.0 MHz 16.0 + 1.0 MHz 77 0.9 f 0.1 0.75 + 0.1 0.7 + 0.1 0.1 f 0.1

a Si/AI denotes the framework %/Al ratio determined by 29Si MAS NMR. ‘H C.S. denotes the values of the isotropic chemical shift (accuracy + 1 ppm) determined from the MAS spectrum measured at vrot = 12.5 kHz for lines a, b, c, and e. The values in brackets, e.g. [40 f 41, are the respective number of species per unit cell. Co,,(*H) and n denote the mean value of the quadrupole coupling constants of the signals of the bridging hydroxyl groups and the corresponding asymmetry parameters, respectively. Co,-,(*‘A0 and n give the values for the aluminium nuclei in the SiOHAl sites.


non-framework aluminium. The relative amounts of non-framework aluminium atoms, obtained from the relative intensities of the lines at ca. 0 and 30 ppm are: in spectra A 03) of zeolite X 50 0 k 2% (8 k 2%), in spectra C (D) of zeolite Y 92 20 t_ 4% (40 k 8%), in spectra E (F) of zeolite Y 85 0 & 2% (20 f 4%), and in spectra G (H) of zeolite H-ZSM-5 are 5 f 2% (5 f 2%). The values in the parentheses demonstrate that except for zeolite H-ZSM5 the amount of non-framework aluminium increases after activation and rehydration of the zeolites. Kellberg et al. [141 claimed that in strongly dealuminated zeolites non-framework aluminium contributes also to the signal at ca. 60 ppm. This amount should be small for weakly dealuminated zeolites and is neglected here.

The centre bands of the 27Al MAS NMR spectra of the hydrated zeolites Na-Y and Na-ZSM-5 consist of only one signal at ca. 60 ppm due to four-coordinated framework aluminium atoms. The sideband spectra in Fig. 7 show that the

signal of the central transition can be observed in the centre band and in the first spinning sidebands: 76% and 24% of the intensity of the second spinning sidebands are due to the inner and outer satellites, respectively. This corresponds to the intensity ratio of 8/5 and their broadening ratio of the static spectrum of l/2. Thus, the FWHM of the centre band is dominated by the second-order broadening of the central transition ( - l/2 tf + l/2), whereas the FWHM of the second (and next order) spinning sidebands is mainly determined by the second- order broadening of the &- 3/2 * f l/2 transitions. Values obtained from Fig. 7 and other spectra for the FWHM of the centre band (and in the parentheses of the third spinning sideband) are 600 Hz (1600 Hz) for the hydrated zeolite Na-Y, 650 Hz (1200 Hz) for the hydrated zeolite Na-ZSM-5, 2400 Hz (3000 Hz) for the octane- loaded zeolite Na-Y, 1300 Hz (satellite sidebands could not be observed) for the octane-loaded zeolite Na-ZSM-5, 2500 Hz (2500 Hz) for the

I’, _;-i’i -300 -340

(pm)

Fig. 7. *‘A MAS NMR satellite transition sideband spectra of the hydrated zeolites Na-Y (A) and Na-ZSM-5 (B). The centre band due to the central transition is truncated in both spectra A and B. The centre bands and the fourth spinning sidebands are presented in detail.


Na-Y i Na-ZSM-5 /,,

H-ZSM-5

I%

H-Y 85 A

H-X 50 II

J,Jy_J)-l 400 0 -400 400 0 -400

mm)

Fig. 8. z’ Al MAS NMR spectra of the dehydrated and octane-loaded zeolites. Aluminium atoms in negatively charged oxygen tetrahedra, compensated by sodium cations, give rise to the spectra of the sodium zeolites and to the narrow component in the spectra of the hydrogen zeolites. Therefore, the spectra of the hydrogen forms must be split into two components. The lineshape simulation for the broad components gives the values: Cocc = 5.5 + 1.0 MHz, n = 0.3 f 0.1 for Na-Y; Co, = 4.7 + 1.0 MHz, TJ = 0.5 + 0.1 for Na- ZCSM-5; Cocc = 12.7k 1.0 MHz, 77 = 0.7+ 0.1 for H-Y 92;

oc- = 16.0+ 1.0 MHz, n = O.l&O.l for H-ZSM-5; Cocc = 13.1 rt 1.0 MHz, n = 0.75+0.1 for H-Y 85; Cocc = 10.8$: 1.0 MHZ, n = 0.9kO.l for H-X 50.

dehydrated zeolite Na-Y, and 2200 Hz (2300 Hz) for the dehydrated zeolite Na-ZSM-5. The longitudinal relaxation times T, Al in the hydrated zeolites Na-Y and Na-ZSM-5 were measured with a non-selective 7r/2 pulse pair (pulse duration 2 PSI giving values of ca. 2 and 3 ms, respectively. Haase et al. 1211 measured the same value for the hydrated zeolite Na-Y using a soft-pulse technique.

The resonance position of the 27A1 NMR central transition signals of the dehydrated samples is given by the centre of gravity of the line. The second moment of the line with respect to this centre of gravity gives a value for the (high field) second-order quadrupole shift of the centre of gravity (cf. Freude and Haase 1181 p. 30). Finally, the chemical shift 6 with respect to the reference Al(H,O)i+ in aqueous solution is obtained by the addition of the resonance position of the centre of gravity and the magnitude of the quadrupole shift, giving 6 = 105 f 20 ppm for the dehydrated zeolites H-Y and 6 = 82 k 20 ppm for zeolite H-ZSM-5.

27Al NMR spectra of the dehydrated zeolites The longitudinal relaxation times T, Al in the are shown in Fig. 8. The full width of the spectra dehydrated samples were measured for the cen- varies from 24 kHz (Na-ZSM-5) to 100 kHz (H- tral transition, because a non-selective excitation ZSM-5). This means that the maximum value of all transitions cannot be achieved with a 0.5~ps available of the MAS frequency vrOt = 18 kHz pulse. A selective r/2 pulse length (correspond- does not exceed the second-order broadening of ing to a non-selective n-/6 pulse) of 10 pus was the central transition. Thus the application of chosen for the 7r/2, t,, ~r/2 pulse sequence to be

MAS is difficult for the sodium form and point- less for the hydrogen forms of dehydrated zeolites. Aluminium atoms in negatively charged oxygen tetrahedra, compensated by sodium cations, give rise to the spectra of the sodium zeolites and to the “narrow” component in the spectra of the hydrogen zeolites. Therefore, the spectra of the hydrogen forms must be split into at least two components. Moreover, different crystallographic positions of the aluminium atoms in the sodium form, and different distances to the hydroxyl protons in zeolites H-Y, should give rise to a distribution of the quadrupole parameters. The values of c occ given in the legend to Fig. 8 were determined by means of a lineshape simulation with the programme WINFIT (Bruker) assuming only one or two sets of parameters for the sodium and hydrogen forms of the zeolites, respectively. A comparison of the spectral width of the dehydrated zeolite H-ZSM-5 with the rehydrated one (see Figs. 6 and 7) yields a factor of about 100. The spectrum of an octane-loaded sample acquired one month after the filling of the MAS sample container did not show a component nar- rowed by water. This proves that the octane loading prevented the sample from rehydration while exposed to air.


selective for the central transition. The samples were dehydrated and sealed in quartz ampoules. Values of

JL 400 0 -400

Fig. 9. “4 NMR spectra of dehydrated samples H-ZSMJ before (A), and after mild hydrothermal treatment at water vapour pressures of 7 kPa (B), 13 kPa (0, and 93 kPa CD).

A

:’ B

J

430 -lob -12b

2. I I I

-80 -100 -120

(pm)

Fig. 10. 29Si MAS NMR spectra of the hydrated (A) and dehydrated (B) zeolites H-Y 92.

olites H-Y 92. Spectra of the other samples are not shown. The spectra exhibit resolved lines due to different values of n in the coordinations Si[ nAl,(4 - n)Si], which allow the determination of the framework Si/Al ratios by means of lineshape simulations [22]. The framework Si/Al ratios so obtained are 22, 3.1, 2.4 and 1.4 for zeolites H-ZSM5, H-Y 92, H-Y 85, and H-X, respectively. Note that no significant difference could be observed between spectra of hydrated and dehydrated (and octane-loaded) samples except that the resolution is decreased for the dehydrated samples. The change in the peak positions was less than 1 ppm. CP experiments on the dehydrated samples yield, in agreement with Hunger et al. [4], an increase of the lines with higher values of IZ.

4. Discussion

The characterization of solid acids by ‘H MAS NMR is now well established. In order to inter-


pret the spectra of Bronsted acid catalysts it is normally accepted that the strength of acidity of hydroxyl groups (connected to silicon or aluminium atoms whose first coordination sphere consists of oxygen atoms only) increases with decreasing electronic charge of the hydrogen atom [23], which corresponds to a larger value of the ‘H chemical shift. Since the NMR intensity of the resolved lines in the high-speed MAS spectrum is directly proportional to the respective concentration of sites, the spectrum may be interpreted as a distribution function of acidity. This approach allows an interpretation of lines a and b in the spectra. Considering, for example, the acidic hydroxyl groups, the chemical shift values of line b increase from 3.7 ppm for zeolite H-X through 3.9-4.0 ppm for zeolite H-Y to 4.3 ppm for zeolite H-ZSM-5. This corresponds to an increas- ing acid strength, whereas the concentration of acidic hydroxyl species in the respective zeolites decreases (see Table 1). In the case of the non- acidic hydroxyl groups, the intensity of line a at ca. 2 ppm is a measure of the concentration of silanol groups. However, the spectra in Fig. 1 show two more hydroxyl group signals, lines c and e, for which the value of the chemical shift cannot be explained by the simple model above. The corresponding hydroxyl groups experience an additional electrostatic interaction with other oxygen atoms in the neighbourhood. This increases the value of the chemical shift but not the acid strength. These hydroxyl protons point to other framework oxygen atoms which reduces the ac- cessibility of the hydroxyl protons to adsorbed molecules and, consequently, their ability to act as acid sites [2,4,5].

The spectrum of zeolite H-Y 92 (Fig. 1A) consists of four signals of hydroxyl species, lines a, b, c and e. The quantitative comparison of spectrum A with B in Fig. 1 verifies the decrease of the lines a-e upon deuteration. Spectrum C compared with B demonstrates that the concentration of the ‘H nuclei on silanol groups, line a, is decreasing during some months after the exchange procedure. The explanation is that these hydroxyl groups are inaccessible to deuterated water molecules during the H-D exchange. The concentration of 12 AlOH groups and 4.3 silanol

groups/u.c. in sample H-Y 92 is caused by the calcination of the ammonium form under air during the modification from Na-Y to H-Y. The 7-ppm peak in spectrum E, line d, is due to the incomplete deammonization (8.5 residual ammonium ions/u.c.) of the sample H-Y 92/230 at the maximum treatment temperature of 230°C. In contrast to zeolite H-Y 92, the spectrum of zeolite H-Y 85 (Fig. 1F) shows neither line e nor line a. This characterizes the zeolite H-Y 85 as defect-free.

Line f (Fig. 1D) is caused by a weak rehydration of dealuminated zeolites H-Y. It was found that this signal does not change its position in a wide range of water loading [161 in contrast to a signal in the spectra of weakly rehydrated zeolites H-ZSM-5, which appears in the same region of the chemical shift but shifts the position in dependence on the water loading [24]. In the first study [16] line f was already ascribed to water adsorption on Lewis acid sites. From the weak spinning sideband intensity of line f it was con- cluded that the signal would not be related to non-framework A13+ species but rather to three- coordinated and positively charged silicon atoms of the zeolite framework [16]. This explanation did not take into consideration that the spinning sideband intensities owing to the strong intra- molecular ‘H-‘H and the inter-molecular ‘H- *‘Al dipolar interactions can decrease as a result of an isotropic rotation (or rotation around one axis) of the water molecule on the adsorption site. Such a rotation seems to be very likely on the basis of the ‘H MAS NMR measurements (see below). Batamak et al. [24] attributed the line f to water molecules interacting with Lewis acid aluminium atoms still partially bonded to the framework.

We observed in this study two peaks at 6.9 and 7.4 ppm, which show the typical behaviour of line f as observed for the first time in Ref. [16]: weak spinning sidebands, peak positions independent of the water loading in the range of 4-16 molecules/u.c. and recovery of the signal by means of a Hahn echo after more than 10 ms. The weak quadrupole broadening of line f in the *H MAS NMR spectrum (Fig. 4B) demonstrates the fast isotropic rotation of the species. Hydroxyl


groups and rigid water molecules would give rise to a quadrupole broadening in the order of magnitude of 100 kHz.

In order to determine correlation times of the nuclear magnetic interactions giving rise to lines a, b, c, e and f from the wr dimension in Fig. 2 or the corresponding values of Tphn, some prelimi- nary remarks must be made. The transverse relaxation times 7’TID, TpAS and Tph”, which de- scribe the free induction decay without MAS, the decay of the rotational echo envelope and the decay of the Hahn echo envelope, respectively, are influenced by the correlation times T,’ of the nuclear magnetic interactions which cause the line broadening, if the correlation time is not much greater than the relaxation time, cf. Refs. [25], [26] and [27], respectively. In this study, the correlation time can be dominated by the residence time T, of the observed nuclei at one site in the zeolite, by the longitudinal relaxation time T 1 hetero of non-resonant nuclei (mainly 27Al> in the neighbourhood or by the fluctuation time rEFo of the EFG for the observed quadrupole nuclei. If jumps of the observed hydrogen nucleus between different positions and magnetic transitions of aluminium nuclei interacting with the hydrogen nucleus must be taken into consideration, the correlation time is given by the relation r-1 =r-1 +T-’ C

The Lquatid; given in Refs. [26] and [27] must meet the condition TZMAS or TZHahn < 7, for T, > T F’D. Otherwise the experimentally obtained re- laiation time will be equal to the correlation time Tc, because the spin coherence is irreversibly destroyed and an echo cannot be observed after 3 7,. The rotational echoes due to lines a, b, c, e and d are decaying to zero within 4 ms and can be restored after 8 ms using the Hahn echo pulse sequence. Thus the correlation time T, must be greater than the relaxation times TTrD and TZMAS and has no influence on the static or MAS spectrum. However, the values in the range 12-40 ms obtained for Tphn under MAS conditions show an influence of TV on Tyhn.

We have Tphn = TV, because T, CannOt exceed T 1A, = 20-40 ms (which was measured for the dehydrated zeolite II-Y in a fused quartz am- poule). Therefore, the conclusion can be drawn

that the residence time T, of the ‘H nuclei at one site in the zeolite is greater than the corresponding relaxation time TZHahn, which is ca. 20 ms for the bridging hydroxyl groups. Until now the mini- mum value of T, was estimated to 1 ms, because the resolved lines b and c are at a distance of ca. 500 Hz (Fig. 1A). It is remarkable that all values of T2Hahn for lines a, b, c, e and f are of the same order of magnitude. This hints that the Tl Al of the more ore less strongly dipolar coupled aluminium nuclei in the neighbourhood could be responsible for the decay of the envelope of the Hahn echo and T, can be much longer.

Line f is the evidence of zeolitic Lewis acidity. The combination of Lewis and Bronsted sites seems to be the source of the enhanced catalytic activity of mildly steamed hydrogen forms of the zeolites [12]. Hence attention was given to the nature of the species giving rise to line f [17]. The information, which has been added from this study, is that the species are not uniform. They give rise to two peaks having different chemical shifts and relaxation times T2Hahn. Because line f was observed in the spectra of dealuminated zeolite H-Y [16,17], but could not be observed in the spectra of dealuminated zeolite H-ZSM-5, the zeolitic structure must be important for the species. Two kinds of mobility must be considered for the water in the zeolite: fast isotropic reorientation with a time constant much shorter than the reciprocal width of the spectrum of a rigid water molecule, which is of the order of magnitude of 10 ps. In the static spectrum there is no line f. But the relaxation times TZHahn measured under MAS conditions are longer than 10 ms for ‘H and 40 ms for 2H. This indicates that the mean residence time of the species at one site is longer than 40 ms. In this case, the effective proton-aluminium distance between all hydrogen nuclei of one species and one aluminium nucleus in the neighbourhood is the distance between the mean position of the centre of gravity of the molecule and the aluminium nucleus [S]. The corresponding ‘H-27Al dipolar interaction determines the spinning sideband intensities of line f. From the weak sideband intensities we estimate that the distance between the mean centre of gravity of the species and one aluminium nucleus


cannot be smaller than 0.3 nm [16]. In conclusion, strong adsorption (chemisorption) of the species on one Lewis site can be excluded. However, the influence of Lewis sites on the appearance of line f cannot be excluded. Framework defects or non- framework species acting as Lewis sites can favour the sorption of species giving rise to line f and can close the windows of a sodalite cage so that the sorbed species are locked. The former ex- plains the increased chemical shift of the species, which is higher than those of bulk water. The latter account for the long residence time of the species in the neighbourhood of one site.

In addition it should be noted that neither lowering the measurement temperature to 160 K nor adsorption of fully deuterated pyridine (24 NC,D,/u.c.) remove line f [281. This demonstrates that the species are inaccessible to the pyridine molecules.

From the *H MAS NMR spectra of deuterated zeolites values of the chemical shift can be determined, which must be the same as from the ‘H MAS NMR spectra. Also the separation of the two types of bridging OH groups giving rise to lines b and c, which was not possible in the first experiments Ill], is now demonstrated in Fig. 5. That means that our line broadening a,,, due to the deviation 8-e,,, from the magic angle between the rotation axis and the external magnetic field a,,, = (3/8) Cocc (3 cos*8 - l)/vr is less than 0.02”.

*H MAS NMR reveals a new dimension for the study of surface hydroxyl groups, which can be used in two ways. Firstly, it is evident from Fig. 3B that, due to the quadrupole dimension, the identification of different hydroxyl species in the zeolites with only slightly differing positions in the &scale is clearly improved. Most signals of residual ammonium ions or adsorbed water ap- pear close to VL = 0 kHz. Secondly, in addition to the values of chemical shifts, the quadrupole coupling constants Cocc can be studied in order to characterize the hydroxyl groups. The values of c oCC of the bridging hydroxyl groups (see Table 1) increase with the framework aluminium content of the zeolite from 208 kHz (H-ZSM-5) to 235 kHz (H-X 50 and H-Y 85) due to the decreased acid strength of the bridging OH groups.

Relations between the proton chemical shift and the corresponding deuterium quadrupole coupling constant can be found in the literature. Rosenberger et al. [29] derived the equation G/ppm = c, - c2 x Cocc/kHz with the constants cr = 18.88 and c2 = 0.05 for polycrystalline trihy- drogen selenites. Sternberg and Brunner [30] have predicted for hydroxyl groups in solids values of cr = 26 and c2 = 0.095. According to this correlation our values of Cocc = 208 kHz and 235 kHz should correspond to chemical shift values of 6 = 6.24 ppm and 3.68 ppm, respectively. The absolute value is well predicted for the ‘H MAS NMR signal of zeolite H-X. However, the average chemical shift of bridging hydroxyl groups in the zeolites H-Y (lines b and c) is very close to the value of 4.3 ppm measured for zeolite H- ZSM-5, whereas from the theoretical prediction [30] a difference of 2.56 ppm should follow from our experimental values of Co,,.

The quadrupole dimension of the signals of defect hydroxyl groups shows signals reduced in the FWHM compared with those of the bridging hydroxyl groups. The signal of the AlOH groups shows two singularities. The fitting procedure yields Co,, = 67 kHz and 77 = 0.2. The signal of the silanol groups has only one maximum, the FWHM is 65 kHz and the fit yields Cocc = 75 kHz and 77 = 0.6. The simple explanation of the quadrupole coupling constant predicts a decrease

of Cocc for the weakening of the oxygen-hydrogen bond. However, bond weakening of the defect hydroxyl groups in comparison to bridging hydroxyl groups is in disagreement with the explanation of the corresponding values of the chemical shift. A motional narrowing of the spectral width due to fast jumps between oxygen atoms can also be excluded. If the spinning sideband pattern exists, the jump frequency must be small in comparison to the rotation frequency, which is small compared with the spectral width. However, a fast rotational diffusion of the hydrogen around the Si-0 or Al-O axis reduces the static linewidth by a factor of (l/2) (3 cos*p - 1). Ernst 1311 has determined an angle of p = 40 + 5” between the Si-0 and O-H axes of silanol groups. The fast rotation around the Si-0 (Al-O) axis with a Si-O-H angle in the corresponding


range reduces the static linewidth by a factor of 0.25-0.5. This effect and a distribution of an additional electrostatic interaction between defect hydroxyl groups and framework oxygen atoms can explain the quadrupole lineshape of the 2H MAS NMR spectra of defect hydroxyl groups.

MAS satellite spectroscopy has been often suc- cessfully applied to solid-state NMR studies of 27A1 nuclei. Samoson [32] showed that the second-order quadrupole shift of the various transitions can be split into an isotropic part and an angle-dependent part. The latter causes for powder under MAS condition the broadening of the centre band and spinning sidebands and this broadening depends on the spin number and the considered transition. For instance, for spin-S/2 nuclei, the second-order broadening of the signal due to the inner satellites (m = *3/2 ++ m = i-1/2) is decreased by a factor of 24/7 compared with the central transition [32] (cf. Ref. [18] p. 55). Therefore, the analysis of the spinning sideband due to the inner satellites is useful for many compounds. Jager [33] denoted it as SATRAS (satellite-transition spectroscopy) and gave various examples of successful applications. In contrast, Fig. 7 demonstrates that for hydrated zeolites Na-Y and Na-ZSM-5 the sidebands due to the inner satellites are broadened compared with the centre band of the central transition. This cannot be explained by an incorrect setting of the magic angle, because the accuracy of 0.02” was demonstrated by the 2H MAS NMR spectra above. The longitudinal relaxation time T1 Al limits the narrowing of the centre band and spinning sidebands as well, if T2MAS = T1 Al. For the zeolite Na-Y under study is TZMAS = 1 ms and Tl Al = 2 ms. Thus the longitudinal relaxation time cannot be excluded as a reason for the sideband broadening. Furthermore, the second-order quadrupole broadening and chemical shift anisotropy are of the same order of magnitude and the quadrupole tensor can be non-coincident with the chemical shift tensor. The description of the combined effect needs special considerations (cf. Ref. [181 p. 24). Skibsted et al. [34] performed a 51V MAS study and found that the spinning sideband pattern is sensitive to variations in the angle between the principal axes of the chemical shift and EFG

tensors. This could be one reason for the broadening of the satellite sidebands in our study. Another reason could be the time dependence of the EFG tensor. The portion of the EFG due to the framework is static. The time-dependent portion arises from the mobility of the cations and the adsorbed water molecules. The influence of this portion of the EFG on the central line can be averaged out, if the jump frequency of the cations or molecules is much greater than the spectral width of the central transition, but smaller than the spectral width of the satellites. The satellite sidebands would be broadened in this case. This is the reason why we measured the satellite MAS spectra on hydrated, dehydrated and octane- loaded zeolites. The highest discrepancy between the theoretically expected (175 Hz) and experimentally measured (1600 Hz) values was obtained for the hydrated zeolite Na-Y, which has the highest concentration of hydrated Na+ cations. Lower discrepancies (but still a relative broadening of the satellite sidebands) can be observed for the dehydrated or octane-loaded samples, for which the fluctuation of the EFG should be much smaller. But it must be noted that these spectra (not shown in Fig. 7) do not represent the total intensity of the satellites due to the insufficient spectral excitation by the 0.5~ps pulse and that the spinning frequency is smaller than the full width of the static spectra (Fig. 8). In conclusion, there is no final explanation for the broadening of the satellite sidebands.

Studies of Bronsted acid sites or non-framework aluminium species must be performed on activated samples. This is well known for the ‘H and 2H NMR studies, but the comparison of the spectra in Fig. 6 with those in Fig. 8 demonstrates that it is true also for the 27Al NMR investiga- tions. Even the measurement of the concentration of non-framework aluminium species by 27Al MAS NMR on hydrated samples is not represen- tative for activated zeolites, because there is a significant difference between the concentrations before and after activation, except for the zeolite H-ZSM-5. In contrast to the 27Al NMR spectra in the 29Si MAS NMR spectra no strong difference could be observed between spectra of hydrated and dehydrated octane-loaded samples. The reso-


lution of the lines due to different values of IZ in the coordinations Si[nAl,(4 - n>Si] slightly decreases for the dehydrated samples, and the vari- ation of the chemical shift of the lines is less than

1 ppm. While a detailed discussion exists about the

correlation between proton chemical shift values [2,5] or deuteron quadrupole coupling constants [29,30] with the weakening of the oxygen-hydrogen bond, the correlation between the aluminium quadrupole coupling constant of the Bronsted acid site with its acidic properties is so far un- known. In the present study first experimental Cocc values are given for the aluminium atom in a SiOHAl site. Experiments with hydrated zeolite samples could not give information about this site, because the hydroxyl proton is involved in the formation of hydroxonium ions [24]. The values of Cocc in Fig. 8 demonstrate that the EFG at the position of the aluminium atom increases by a factor of 2.3 or 3.4 for zeolites Y or ZSM-5, respectively, going from the dehydrated sodium form to the dehydrated hydrogen form of the zeolite. This result shows that the acidic proton must affect the EFG tensor at the position of the framework aluminium atoms. A comparison between the samples under study yields the higher value of Cocc for the Bronsted acid site in zeolite H-ZSM5, which gives also the maximum value of the chemical shift of line b in the ‘H MAS NMR spectrum (Table 1). The higher rotational symmetry of the EFG tensor for zeolite H-ZSMS can be explained by an isolated Bransted acid site, whereas in zeolites of the faujasite type the Bronsted acid sites in the neighbourhood or aluminium cations can destroy the symmetry of the tensor. Further experiments and also calculations of the EFG tensor are necessary in order to correlate the field gradient tensor at the site of the *‘Al nucleus in the SiOHAl group with the properties of the Bronsted acid site.

Values of the 27A1 NMR chemical shift can be accurately determined for the hydrated zeolites. The value for the four-coordinated framework aluminium is ca. 60 ppm for the hydrogen and sodium forms of zeolites Y and ZSM-5, while a narrow signal of six-coordinated non-framework

aluminium appears at ca. 0 ppm. In the case of dehydrated hydrogen zeolites one Al-O bond in the bridging groups (= Si-OH-Al = ) is weak- ened or broken, commonly denoted by dotted instead of solid line in the structure. This can be understood as an intermediate case between four- and three-coordination. Consequently the chemical shift values of these nuclei should be expected to be higher than those in hydrated samples. Our values of 6 = 105 f 20 ppm for dehydrated zeolites H-Y and 6 = 82 & 20 ppm for dehydrated zeolite H-ZSM-5 support this suggestion.

Fig. 9 shows that also the non-framework aluminium in the dehydrated sample can now be observed. The fact that no broader component can be detected in the observed spectral range of 500 kHz (2600 ppm) does not entirely exclude the existence of non-framework Al species which are still “NMR-invisible”. However, at least the major part of non-framework aluminium species (70-100%) has become observable and it has to be assigned to a signal which is narrower than that of the bridging Si-OH-Al groups. There- fore, quadrupole broadening of the non-framework aluminium species is smaller, corresponding to a higher symmetry of their positions. Further experiments and calculations of the EFG tensors are necessary in order to correlate the ‘7Al field gradient tensors with the properties of the Bronsted sites and non-framework aluminium species.

5. Conclusions

‘H MAS NMR provides a new dimension for the study of surface hydroxyl groups. Due to the quadrupole dimension, the identification of different hydroxyl species in the zeolites with only slightly differing positions in the &scale is clearly improved. More complicated sample preparation and much longer acquisition times required re- main major disadvantages of *H MAS NMR compared with ‘H MAS NMR.

The ‘H MAS NMR signal at ca. 6.5 ppm in weakly dealuminated zeolites H-Y, which does not change its position in the loading range of 4-16 water molecules/u.c., can be explained by


the water adsorption on Lewis acid sites. But chemisorption of the water species on one Lewis site must be excluded. The species are locked in the zeolite framework, which is modified by dea- lumination. Their residence time in the neighbourhood of one site is long and they are not accessible to the pyridine molecules.

Measurements of the satellite 27A1 MAS NMR spectra on hydrated, dehydrated, and octane- loaded zeolites Na-Y and Na-ZSM-5 give satellite linewidths up to one order of magnitude broader than the theoretically expected values. Several reasons for this broadening were discussed, but no conclusive explanation is available.

“A1 NMR studies of Bronsted acid sites or non-framework aluminium species must be performed on dehydrated samples. The 27Al signals of Si-OH-Al sites and of non-framework aluminium species in dehydrated (and dealuminated) zeolites, which were considered “NMR-invisible” in the past, can now be observed.

Acknowledgements

The authors wish to express their gratitude to Professor Harry Pfeifer for many stimulating dis- cussions; this paper is dedicated to him on the occasion of his 65th birthday. We wish to thank Dr. C. Doremieux-Morin (Paris), Dr. D. Fenzke (Leipzig) and Dr. J. Klinowski (Cambridge) for comments.

References

[l] J. Klinowski, Progr. Nucl. Magn. Reson. Spectrosc., 16 (1984) 237.

[2] H. Pfeifer and H. Ernst, Ann. Rep. NMR Spectrosc., 28 (1994) 91.

[3] H.K.C. Timken, G.L. Turner, J.G. Gilson, L.B. Welsh and E. Oldfield, J. Am. Chem. Sot., 108 (1986) 7231.

[4] M. Hunger, M.W. Anderson, A. Ojo and H. Pfeifer, Microporous Mater., 1 (1993) 17.

[5] D. Freude, Stud. Surf Sci. Catal., 52 (1989) 169. [6] R.L. Stevenson, J. Catal., 21 (1971) 113.

[7] D. Freude, J. Klinowski and H. Hamdan, Chem. Phys. Lett., 149 (1988) 355.

[8] D. Fenzke, M. Hunger and H. Pfeifer, J. Magn. Reson., 95 (1991) 477.

[9] A.J. Vega and Z. Luz, J. Phys. Chem., 91 (1987) 365. [lo] T.J. Gluszak, D.T. Chen, S.B. Sharma, J.A. Dumesik and

T.W. Root, Chem. Whys. Lett., 190 (1992) 36. [ll] H. Ernst, D. Freude and I. Wolf, C/rem. Phys. Lett., 212

(1993) 588. [12] R.M. Lago, W.O. Haag, R.J. Mikovsky, D.H. Olson, S.D.

Hellring, K.D. Schmitt and G.T. Kerr, Stud. Surf. Sci. Catal., 26 (1986) 677.

[13] E. Brunner, H. Ernst, D. Freude, T. Friihlich, M. Hunger and H. Pfeifer, J. Catal., 127 (1991) 34.

1141 L. Kellherg, M. Linsten and H.J. Jakobsen, Chem. Phys. Lett., 182 (1991) 120.

[15] J. Rocha, S.W. Carr and J. Klinowski, Chem. Phys. Lett., 187 (1991) 401.

[16] M. Hunger, D. Freude and H. Pfeifer, J. Chem. Sot. Faraday Trans., 87 (1991) 657.

[17] P. Batamak. C. Doremieux-Morin, R. Vincent and J.

[181

I191

DO1

WI

WI

[W

[241

I251

I261

[271

D81

Fraissard, Microporous Mater., in press. D. Freude and J. Haase, NMR Basic Principles Progr., 29 (1993) 1. A.C. Kunwar, G.L. Turner and E. Oldfield, J. Magn. Reson., 69 (1986) 124. B. Bliimich, P. Bliimler and J. Jansen, Solid State Nucl. Magn. Resort., 1 (1992) 111. J. Haase, K.D. Park, K. Guo, H.K.C. Timken and E. Oldfield, J. Phys. Chem., 95 (1991) 6996. G. Engelhardt and D. Michel, High-Resolution Solid-State NMR of Silicates and Zeolites, John Wiley & Sons, Chich- ester, 1987. U. Fleischer, W. Kutzelnigg, A. Bleiber and J. Sauer, J. Am. Chem. Sot., 115 (1993) 7833. P. Batamak, C. Dorimieux-Morin, J. Fraissard and D. Freude, .I. Phys. Chem., 95 (1991) 979. P.W. Anderson and P.R. Weiss, Reu. Mod. Phys., 73 (1953) 269. B. Schneider, D. Doscocilova, J. Jakes, Proceedings of the 20th Congress Ampere, Tallin, Springer Verlag, Berlin, Heidelberg, New York, 1976, p. 57. D. Fenzke, D. Freude, D. Miiller and H. Schmiedel, Phys. Status Solidi. (b), 50 (1972) 209. I. Wolf and D. Freude, in preparation.

[29] H. Rosenberger, G. Scheeler and Yu.N. Moskvich, Magn. Reson. Chem., 27 (1989) 50.

[30] U. Sternberg and E. Brunner, J. Magn. Magn. Reson., in press.

[31] H. Ernst, Z. Phys. Chem. (Leipzig), 268 (1987) 405. [32] A. Samoson, Chem. Phys. Lett., 119 (1985) 29. [33] Ch. Jager, NMR Basic Principles Prog., 31 (1994) 133. [34] J. Skibsted, N.Ch. Nielsen, H. Bildge and H.K. Jacobsen,

Chem. Phys. Lett., 188 (1992) 405.

Documents

Solid-state nuclear magnetic resonance studies of acid sites in