Quantum Chemical Studies of Zeolite Acidity

73

QUANTUM CHEMICAL STUDIES OF ZEOLITE ACIDITY JOACHIM SAUER Zentralinstitut fur Physikalische Chemie, Akademie der Wissenschaften der DDR, Rudower Chaussee 5, Berlin, 1199, German Democratic Republic. SUMMARY Non-empirical quantum chemical calculations on models of acidic hydroxyl groups in zeolites and related catalysts are reviewed. In the first part, structure predictions are made and properties are considered in order to assist spectroscopy in the identification of active sites. In order to characterize proton mobility in zeolites an estimate is made of the energy of activation for proton jumps between neighbouring oxygen sites of the framework. In the second part, the energy of deprotonation as a quantum chemically accessible measure of acidity is addressed. From results for a variety of models, inferences are made about the strength of acidic sites in different catalysts (structurereactivity relations). Specifically, hydrogen-bonded silanol groups on defects and hydroxyl groups on extra-framework aluminium species are considered, and the activity of catalysts such as AlPO4 and SAP0 is discussed. The extent to which spectroscopic parameters such as the 1H NMR chemical shift are suitable measures of acidity is investigated. INTRODUCTION Understanding the origin of catalytic activity on the molecular scale requires the development of a model which specifies both the active site and the elementary reaction step. The present study considers surface hydroxyls as the active sites of acidic catalysts (neglecting a possible cooperative effect of neighbouring basic sites) and the proton transfer from this site onto the substrate as the initial step. Other steps of the often complex catalytic reaction are disregarded. There is general agreement that bridged hydroxyl sites, Ib in the formulae below, are the origin of the Br~rnstedacidity of zeolitic catalysts, while surface hydroxyls, Ia, which terminate the three-dimensional zeolite framework are non-acidic. Attempts were made to create different types of bridged sites, for example Ic with T = B, Fe, by isomorphous substitution of tetrahedral framework atoms. Lattice defects such as broken Si-0-Si bonds, Id, may also contribute to the activity of a catalyst. Highly active high-silica forms of zeolites are frequently prepared by framework dealumination. At least part of the extra-framework material produced is likely to bear different Al-OH groups, e.g. IIa or 1%. In microporous Alp04 materials bridged hydroxyls of the Ib type can be created by substituting Si,H for P. In addition,

74

terminal AlOH and POH groups, IIIa, or phosphoric acid-like groups, IIIb, may be present.

/

/

Ib

Ia

Ic

Id

"\/"'

A1 ' 0

I

0

IIa

H

\A1 -

P-

Spectroscopy, predominantly IR and MAS NMR, is extensively used to identify different types of surface hydroxyl groups. Attempts were also made to use spectroscopic parameters such as OH valence stretching frequency or *H NMR chemical shift as measures of acidity. Quantum chemistry can contribute in two ways. First, computational techniques can be used to determine the local geometry of active sites and to calculate their properties to assist spectroscopy for identification. Second, the reactivity of surface sites could be studied by calculating the potential energy surface that governs the proton transfer process (Fig. 1) or at least parts of it. This is presently a too ambitious aim and a more realistic approach is to calculate the deprotonation energy of the hydroxyl groups as a measure of acidity. More generally, concepts from theoretical chemistry are available to explain acidity differences and to derive property-activity relations which allow to predict

75

physi)r

F a

sorption

chemi-

[ZO-H-S]'

C

w

t

Reaction coordinate Fig. 1. Energy profile for proton transfer reactions. the acidity of a given surface hydroxyl group from its spectroscopic parameters. The use of these concepts is not limited to results of quantum chemical calculations, but is also helpful in the analysis of the wealth of experimental data. MOLECULAR MODELS OF BR0NSTED SITES Application of quantum chemical methods to catalysts requires approximations. What theoreticians would like to do are crystal orbital calculations (ref. 1) which exploit the translatory symmetry of a crystal. This non-empirical approach involves so much computational work that it is only now that crystals as simple as 5 0 2 or A1203 can be treated. The lack of periodicity also limits the use of crystal orbital techniques for catalysts. What we cun do to make calculations on solids feasible is to choose finite models and treat them like molecules (ref. 2). As a consequence, only the intrinsic properties of individual active sites can be studied, which is a significant limitation of this approach. Fig. 2 shows models of different types of surface hydroxyls. STRUCTURE AND VIBRATIONAL PROPERTIES OF BRIDGED HYDROXYL SITES IN ZEOLITES The geometry of bridged hydroxyl groups is difficult to infer from experiments. X-ray diffraction cannot easily distinguish between Si and A1 atoms or localize

76

protons. Neutron scattering experiments are more sensitive towards protons, but the studies reported so far for zeolites (refs. 3,4) did not succeed in the accurate determination of the geometry of hydroxyl sites. A major complication is the low concentration of these sites in most catalysts. H

I

H/si\

P\

0

H

H

I

\0/Si\IH H 3

H

H 0,*,\I f j OH l

H

a

H

I

Fig. 2. The models adopted. From studies on gas phase molecules we know that non-empirical SCF calculations using "double zeta" basis sets augmented by polarization functions (DZP) yield bond lengths and angles which deviate by less than 2 pm and 3 degrees from the accurate result (e.g. see refs. 5,6). Table 1 shows the results of this type of calculation for models of terminal (L silanol) and bridged hydroxyls (3) (ref. 7). Calculations on molecules involving hydroxyl groups show that predicted OH bond lengths are too short by a fairly constant amount of 1.2 pm. Comparison with an

77

TABLE 1 Calculated bond lengths (pm) and angles (degrees) for silanol recommended geometry of bridged hydroxyl sitesa

117.4 95.8

120.4 94.6

165.0

164.6

Parameter L SiOH r (OH) L SiOAl r (SiO) r (A101

(aand model 3 and

H3SiOH.Al(OH)20SiH3 (3J Calculated Recommended 120.0 117.5 f 3 95.2 96.4 f 0.4 131.4 131.5 f 5 169.8 170.0 f 2 194.3 194.5 f 2

c Accurate

calculations including electron correlation and employing extended basis sets, cf. ref. 7.

accurate geometry for the silanol group (Table 1) confirms this general experience and indicates that the SiOH bond angle is too large by about 2.5 degrees. This knowledge of systematic errors of the chosen computational procedure is employed to make empirical corrections on the calculated structure for bridged hydroxyls.

1986, Wax et al

1971, Stevenson

(Rho)

(HY)

Recommended structure

131.52 5'

H Fig. 3. Geometry (pm, degrees) of bridged hydroxyl sites inferred from experiments (ref. 8,9) (above) and quantum chemical calculations (ref. 7) (recommended structure, below).

78

The "recommended" structure (Fig. 3) is a significant improvement over previous models which emerged from a combination of different observations (ref. 8) or were adopted in order to interpret vibrational properties (ref. 9). Due to the shallow bending potentials the bond angles show the largest uncertainty intervals. It is likely that the SiOAl angle of bridged hydroxyl sites depends strongly on the particular zeolite framework type. Note that the A1-0 bond is longer than the Si-0 bond by as much as 24.5 pm. Although our model neglects the strain imposed by the framework, this result shows that the A1-0 bond is much more sensitive to perturbations, such as adding a proton, than the Si-0 bond. From recent NMR experiments the A1-H non-bonded distance of bridging hydroxyls in zeolites HY and HZSM-5 was estimated as 238f4 pm and 248+4 pm respectively (ref. lo), while our recommended structure implies 246 pm. IR spectroscopy remains a major source of information on surface hydroxyls even after high-resolution solid-state NMR spectroscopy became a widely applied technique in characterizing catalysts. In bridged hydroxyls, the proton gives rise to three vibrational modes: (1) OH valence stretch (VOH) (2) SiOH in-plane deformation (&iOH) (3) OH out-of-plane deformation (VOH) While the latter two are more or less coupled with stretchings of the neighbouring bonds or bendings of neighbouring angles, the OH valence stretching motion is largely decoupled from all other modes. Only the OH stretching mode is easily observed and its strong absorption is a common means of identifying surface hydroxyls and to monitor their fate during modification or use of catalysts. The SiOH in-plane deformation, which is in the range of 700 to 1100 cm-1, is hidden by framework vibrations and can only be observed as a combination band with the 0-H stretching mode in the near-IR region (diffuse reflectance technique) (ref. 11). Indications of a similar combination band with the out-of-plane deformation have been found only very recently (ref. 12). A selective means to study vibrational modes involving the proton is inelastic neutron scattering (INS), but only one study of bridged hydroxyls in zeolite rho has been published so far (ref. 9). In quantum chemistry analytical derivative techniques are available which yield the complete harmonic force field including all coupling terms. Hence harmonic vibrational frequencies and normal modes are straightforwardly accessible by the standard Wilson GF technique. Table 2 shows results for modes of terminal and bridged hydroxyls (cf. refs. 13,141. The following conclusions can be reached: (1) The calculations predict a much lower frequency for the OH out-of-plane mode than for the in-plane SiOH bending. This contradicts the assignment made by Wax et al. of the peaks observed in their INS spectrum of zeolite rho (ref. 9). They

79

TABLE 2 Vibrational frequencies (cm-1) of terminal and bridged surface hydroxylsa

Mode "OH

8SiOH 'YOH

Site Modelb Calcdc Obsdd Calcdc Obsdd CalcdC ObsdC

terminal 1

4140 3745 960 795 to 835

bridped

-2 4 4090 (-50)a 4000 (-140)a 3550 to 3660 (-195 to -85)a 1200 (240)a 1180 (220)a 970 to 1055 (175 to 220)a 420 440 360 to 380

aDifference between bridged and terminal is given in parenthesis. bFig. 2. CSCF/6-31G* approximation, refs. 13-15. dref. 11. erefs. 9,12. assumed only a small splitting between in-plane and out-of-plane OH deformation but their interpretation left the strongest feature of the spectrum at about 360 cm-1 unassigned (Fig. 4). The results of quantum chemical calculations, indicated by bars in the upper part of Fig. 4, unequivocally show that this feature of the spectrum is due to the OH out-of-plane mode. Independent support for this assignment comes from recent diffuse reflectance experiments on HNaY and H-erionite (ref. 12). From the observed combination bands a fundamental frequency of about 370 to 380 cm-* is inferred for the out-of-plane deformation. (2) The results reveal that the SiOH in-plane deformation undergoes much larger shifts than the OH stretching mode and confirm an earlier suggestion (ref. 11) to look at this mode when trying to identify surface hydroxyls. (3) As is commonly found (ref. 6), the calculated frequencies are too high in absolute terms; however, shifts observed between different types of hydroxyls are well reproduced. (4) Two models, 2. and & are considered for bridged hydroxyls. The results differ notably for the OH stretching mode only. While model 4 is certainly more realistic than model & there are indications (ref. 7) that calculations on even larger models such as 3 will predict a frequency shift which is smaller than that obtained for model 4 (Table 2). Observation of combination bands by diffuse reflectance techniques faces the difficulty that the anharmonicity constants are required to get from them the fundamental wavenumbers. Usually there is not enough observed data to get anharmonicity constants from experiment, but quantum chemical calculations yield them with even better accuracy than the harmonic force constants.

80

INS

i

[arb. un.1 1

counts 600 400

200

0

cis,' n.1I

IR

[arb. u

0

. D I

0 0

0 I

I

0 0

I

I

I

I

0 0

0 0

F

-

C W C O O G

Fig. 4. Observed INS spectrum of zeolite rho (ref. 9) (upper part). The bars indicate results of quantum chemical calculations [SCF/3-21G, analytical second derivatives (ref. 6)1 for model 2 (ref. 15). The intensities are calculated as root mean square hydrogen displacement amplitudes, < x H ~ > 12. ' (1) SiOH in-plane bending; (2) SiO stretch; (3) OH out-of-plane; (4) A10 stretch; (5) SiOAl bending. The lower part of the figure shows the corresponding IR intensities. Such calculations were recently made for the H3SiOH (1)and H3SiOH.AIH3 (2J models and comparison was made with HDO and HD20+ (ref. 13). The calculated anharmonicity constants of the OH bonds in these systems are remarkably constant and vary between -76 and -84 cm-1 only, in agreement with the values observed for DOH (-83 cm-1) and surface silanols, 5 i O H (-9Wl5 cm-1). For the SiOH bending, the predicted anharmonicity is -17 cm-1 (ref. 13), which is close to the value for DOH bending in the HDO molecule (-12 cm-1). It is concluded that constant

81

anharmonicities can safely be assumed for zeolitic hydroxyl groups independent of its type or environment. ENERGY BARRIER FOR PROTON JUMPS BETWEEN NEIGHBOURING SITES A proton that compensates the negative charge of an aluminosilicate framework can be attached to any of the four oxygen atoms belonging to a given A104 tetrahedron. Hence there are four, in first approximation equivalent, bridged hydroxyl sites. The probability of proton jumps between them is of fundamental interest and characterizes the proton mobility in zeolites. Quantum chemical calculations were performed in order to estimate the height of the barrier for such proton motions (ref. 7). It was assumed that the energy is maximum when the proton is halfway along the path from one oxygen site to the other, i.e. the transition state is assumed to have C ~ symmetry. V Complete geometry optimizations were made on the equilibrium structure of model 3 (CS symmetry, cf. Table 1) and the corresponding transition structure (C2v symmetry) within the SCF approximation (DZP basis set) (ref. 7). The energy barrier obtained as a difference of the respective total energies is 68 kJ/mol. Calculations on reduced models indicate that electron correlation lowers the barrier height to about two thirds the SCF result, while basis set extension may increase it by about 10%. The final estimate of the barrier height is 49f10 kJ/mol (ref. 7). We stress that this result applies to a completely unloaded catalyst. The barrier height is likely to decrease when external molecules with proton donor and acceptor abilities such as H20 or NH3 become involved. From NMR experiments an activation energy of 26-41 kJ/mol (at about 230°C) was deduced for differently prepared H-faujasites (ref. 16). The conclusion was reached (ref. 16) that even this result for the unloaded catalysts may be affected by the presence of residual ammonium ions. Hence the experimental result is a lower limit for the situation modelled in the calculations. BRPINSTED ACIDITY: DEFINITION AND PRINCIPLES The elementary step of catalysis by Bransted acidic sites is proton transfer from the surface hydroxyl site onto the substrate S: ZO-H + S H ZO- + SH+ (1) Fig. 1 shows the corresponding energy profile. The transition from the physisorption state into the chemisorption state may be connected with a barrier (as for the proton jump to a neighbouring site, vide supra) or not. When speaking about acidity, we are looking for a reactivity parameter which characterizes the active site but does not depend on the particular substrate. One way to achieve this is to work with a "model substrate". A frequently applied example is

82

the temperature programmed desorption of NH3. A hypothetical decomposition of process (11, ZO-H t)ZO- + H+ (2a) H+ + S t)HS+ (2b) leads to the definition of acidity and basicity of gas phase molecules as the standard Gibbs free energy change of reactions 2a and 2b respectively (e.g. see ref. 17). Neglecting entropy effects that arise from distribution and concentration of sites on the surface, the intrinsic acidity of surface sites can be defined as the following Gibbs free energy change of deprotonation:

and the intrinsic entropy The thermal part of the enthalpy change, change, ASYJT), arise from conversion of three vibrational degrees of freedom of the active site into three translational degrees of freedom of the free proton. The translations of the proton make a constant contribution of

5

RT - T . Strans(H+).

Complete force field calculations for a number of gas phase species (CH30H, H20, (H20)2, H3O+) and molecular models (ref. 18) reveal that AEvib - T ASvib is smaller than 5 kJ/mol and can be neglected:

(m

These calculations also show that the zero-point energy change, AErp, is larger (about 30-45 kJ/mol), but fairly constant. Hence, both the heat of deprotonation at absolute zero temperature (cf. ref. 191,

or the deprotonation energy, AE;lp,

are suitable measures of acidity, which are

accessible from quantum chemical calculations. AG B~(T)= A H ~ ~ (+oconstant ) = AE&

+ constant

(6)

While the deprotonation energy can be directly calculated as the energy difference of ZO-and ZOH, evaluation of AHbP(O) requires much more computer time in order to get the vibrational frequencies. Note that deprotonation is a purely hypothetical process which is highly endothermic. Nevertheless, changes of deprotonation energies between different sites will run parallel to changes of reaction energies for the proton transfer (1) between these sites and the same base or substrate.

83

Since anions are formed on deprotonation, accurate calculations of the deprotonation energy, AEDP, require extended basis sets including diffuse functions and inclusion of electron correlation. But for reasonably extended models of surface hydroxyl sites we can only afford to perform SCF calculations employing DZP or even smaller basis sets (6-31G', 3-21G, 3-21+G). There are indications, however, that the errors we make are systematic and fairly constant and increments have been derived (ref. 19,201 to correct for them. Since the minimal STO-3G basis set yields much too large AEDP values, in this case AHbp(0) is estimated by scaling

1200 SiO.Al H

Si0.B H

1coo

Si 0 HOSi H

SiOH

Al 0 Al H >Ai-OH

1600 kJ molAH&,(o)

:A[-OH

Fig. 5 . Acidity scale for different types of hydroxyls in zeolitic catalysts. Original deprotonation energies taken from refs. 18-22; for .SiOH.Br sites, cf. ref. 23, for different types of hydroxyls on alumina, from refs. 24-25. AEgP(STO-3G) by a factor (ref. 20). This way the basis set dependence can be largely

eliminated from SCF results and the values obtained with different basis sets can be compared within a common acidity scale as done in Fig. 5 . Low heats of deprotonation indicate high acidity values.

84

The acidity differences observed for different hydroxyl groups are explained by the following rules, the first two of which have already been formulated by Pauling (ref. 26): (1) Acidity is the higher the larger the number of 0 atoms is over which the negative charge can be distributed (ref. 26):

\0 \

0

\\ /OH

2

>

/s\

"\

./OH

1

/ O /O i.e. a phosphoric acid-like group is more acidic than a terminal silanol group. (2) The acidity of a ZOH group is the higher the larger the electronegativity (ref. 27), x, of Z is (ref. 26): rSiOH >zAlOH 2.14 x 1.71 (7) (3) The acidity of a given ZOH group is significantly increased by a neighbouring Lewis acid (ref. 24):

H

This explains the high acidity of bridged hydroxyl sites in zeolites. (4) Generally, the acidity is the higher the larger the coordination number of the hydroxyl oxygen atom and the lower the coordination numbers of the neighbouring cations are (refs. 24,251. This is clearly seen from the calculations for different hydroxyls on alumina coordinated by one, two or three aluminium atoms in a trigonal, tetragonal or octahedral environment (ref. 25). Fig. 5 show the wide range of acidity values predicted. (5) The acidity of a given ZOH group is also increased by a neighbouring Brmsted ncid (ref. 22):

Z

H

\

P H

/'

/"\ Y

Z

*

\

0'

-

H /'

/"\ Y + H +

(9)

85

This explains the moderately acidic properties of defect sites (unclosed Si-0-Si links) in high-silica ZSM-5 catalysts with a high concentration of internal silanol groups (ref. 22). An additional rule to understand zeolite acidity differences not emerging from Fig. 5 is: (6) The acidity of a given type of hydroxyl site may be modified by composition of the bulk material. The increase of the acidity strength of bridged sites, Ib, with increasing Si/A1 ratio has been consistently explained in two ways: (i) by considering larger models, e.g. of type (T0)3Si-OH.Al(OSi)3 with an increasing number of T=Al atoms (ref. 281, or (ii) by means of mean electronegativity values (ref. 29). Note that the latter can only be applied when properties of active sites in different samples are compared. When discussing acidities of different types of hydroxyls attention must also be paid to the "levelling" effect (ref. 26), which makes the H3O+ ion the strongest acid in aqueous solution. For zeolitic catalysts this means that no sites can exist that are more acidic than bridged SiOH.Alr sites as long as there is A1 within the framework. For example, [.SiOHSiz]+ sites are much stronger acids than :SiOH.Ak sites (for tricoordinated Si+ is a stronger Lewis acid than tricoordinated A1 (ref. 24), cf. rule (3) above), but would react with the strong base [=SiOAlr]- to form the weaker pair of corresponding acids and bases:

This "levelling" effect also excludes superacidic sites in zeolitic catalysts as possible explanation of activity enhancement observed after mild steaming of HZSM-5 catalysts (ref. 30). That means that a previously suggested model of a hydroxyl group (ref. 20) for which a lower deprotonation energy was calculated than for bridged hydroxyls Ib is not realistic for zeolites. SPECIFIC CATALYSTS Quantum chemistry cannot say which sites will be present in a given catalyst. It can only make a prediction what the acidity of a given site is. From the calculations (Fig. 5) and general principles outlined above the following inferences can be made:

86

Zeolites (1) Bridged hydroxyls are the origin of strong acidity in zeolitic catalysts. (2) Hydroxyls on extra-framework alumina-type phases may contribute to the overall acidity of a catalyst. The possible acidity values cover a wide range (ref. 25, cf. Fig. 5). Note that AlOH sites that are more acidic than bridged hydroxyls, Ib, cannot occur in zeolites due to the levelling effect. Specifically, terminal =AlOH sites are weaker than terminal S i O H sites, but water coordinated to Lewis sites on aluminalike surfaces (tricoordinated Al) shows Brensted acidity almost as strong as bridged =SiOH.Alrhydroxyl groups (ref. 15). (3) Moderate acidity is connected with unclosed Si-0-Si links in form of hydrogenbonded silanol pairs, Id (ref. 22). These defects are likely to occur in high-silica ZSMB catalysts (internal silanols). (ref. 22) and have been shown to exhibit cation exchange capacity in excess to their framework A1 content provided that more basic conditions are chosen than required to exchange the bridged hydroxyl protons (ref. 31). B-modified zeolites Bridged hydroxyls involving boron, Ic, with T=B, are somewhat less acidic than bridged hydroxyls involving A1 (Ib). This is concluded from 3-21G calculations on models like 2. (cf. ref. 23). Previous estimates based on STO-3G calculations on larger models predicted virtually no effect on the acidity when replacing A1 by B (ref. 20). Other forms in which boron may be present in the catalyst involve BOH groups (ref. 20), which are about as weakly acidic as terminal silanol groups in zeolites, for we know that boric acid is as weak an acid as silicic acid. SAPOs and P-modified zeolites Bridged hydroxyls in microporous AlPO4 are only created when some phosphorus is replaced by silicon (P02+ 4 Si02(H+) 1. Assuming that the active site formed is a bridged zSiOH.Als site we conclude from the unchanged mean electronegativity

(refs.

27,29)

of

AIPO4

compared

with

Si02

1

($Xp+XAl]=2.1 l=xsi=2.14) that the sites are about as acidic as in high-silica zeolites.

The possibility that zP0H.Ak sites are formed will not create stronger acidity. Namely, if we assume that the latter are potentially more acidic than =SiOH.Alc sites the system would equilibriate by proton transfer according to: nP+-OH.Alz+ $i-O--Alz t)zP-O-Als + 3i-OH.Alz

(10)

(levelling effect, see above). Terminal +O ' H groups of type IIIa should be similarly weakly acidic as terminal S i O H groups. However when phosphorus is present as phosphoric acid ester as in

IIIb moderately acid sites may be formed. This is consistently inferred both from principle (1)above and calculations on H3P04 (Fig. 5). Moderately acidic groups of type IIIb may also be present on P-modified zeolites or silica. SPECTROSCOPIC ACIDITY MEASURES 1H NMR CHEMICAL SHIlT Use can be made of quantum chemically calculated deprotonation energies in two ways. In the previous section, from deprotonation energies for models of a variety of active sites, inferences have been made about the strength of acidic sites in different catalysts (structure-reactivity relation). In this section, from comparison with calculated 1H NMR chemical shifts for a number of active sites we learn to what extent this parameter can serve as a measure of acidity (property-reactivity relation). Table 3 summarizes the shift ranges of lines observed in 1H MAS NMR spectra of zeolitic catalysts (refs. 32-34). In addition, reference is made to specific results for H-ZSMS obtained in a 500 MHz experiment (ref. 35). The agreement of the relative shift of lines a and b, assigned to terminal, Ia, and bridged, Ib, hydroxyl groups (refs. 32-34) with the corresponding values calculated for the H3SiOH (1)and H3SiOH.AlH3 ('2) models (ref. 37) by the IGLO method (ref. 38) is encouraging, in particular when acknowledging that calculation of chemical shifts is a very demanding quantum chemical task. The shifts calculated for models 5 and 6 of isolated (IIa) and bridged (IIb) AlOH groups, respectively, are less split and fall into the range between =SOH and =SiOH.Ak groups in agreement with the assignment of line e to AlOH groups (ref. 32-34). TABLE 3 *H NMR chemical shifts, 8~ (ppm), for hydroxyl protons in gas phase molecules and surface hydroxyl groups relative to gas phase CH3OH.a Calculatedb Observed H20 1.0 H20 0.6C CH3OH 0.0 CH3OH 0.W c2H5oH 0.5 C2H50H 0.4C H3SiOH 0.9 SiOH 1.8- 2.3d (a) 1.W H2AlOH 2.2 AlOH 2.6 - 3.6d (el 2.5e (H2A10kU2 2.6 ) H3SiOBAlH3 3.0 =SiOH.AL 3.8 - 4.4d (b) 4.0e a The experimental shift of methanol relative to gaseous TMS is 0.02 ppm (ref. 36). b Ref. 37, cf. ref. 38. C Ref. 36. d Refs. 32-34. e Ref. 35. ~

1

88

The suggestion (refs 32,33,36) that the 1H NMR chemical shift loHcan serve as a measure of acidity relies on the following arguments: (1) Deprotonation is the easier the higher the net charge in the hydrogen atom in the molecule is. (2) The higher the net charge on the hydrogen atom (i.e. the lower the electron density) the less the nucleus is shielded.

1

5 0

3 2 4 0 O 00

Q 0.30 m

z 0

8 0

10

Y

I

0-

0.35

t t 20

9 . 0

7

8

6 :O la 12

13 1

,

n

l

n

l

25

l

r

l

*

30 H '

[ppm]

l

I

-

I

35

Fig. 6 . Dependence of net atomic charges on hydrogen, qH, on the 1H NMR shielding constants, ti^, (IGLO method, ref. 38)- see ref. 37 for computational details. Geometries are all optimized adopting the SCF/6-31G* approximation, except for H3P04 for which the 4-31G(*)geometry was used. (1)H20; (2) CH30H; (3) CzH50H; (4) H2NOH; (5) HOOH; ( 6 ) ClOH; (7) O=P(OH)3; (8) H2BOH; (9) H2AlOH (3 (10) H3SiOH (1);(10a) H3SiOH (3-21G(') geometry); (11) (HzA1OH)z (6);(12) H3SiOH.AlH3 0;(13) H3O+. Data in favour of argument (1) have been previously produced (ref. 39). From theoretical reasons argument (2) is more difficult to accept. Preliminary calculations (ref. 37) raise doubts about a generally valid correlation between loHand q H (Fig. 6 ) and between loHand A E D ~(ref. 37). On the other hand, within the group H20, CH30H and C2H50H the electron density decreases but the shielding increases. This is also true when comparing H2BOH and H2AIOH. Moreover, results for two different geometries of H3SiOH (points 10 and 10a in Fig. 6 ) produce evidence that the IH NMR chemical shift is very sensitive to geometry changes which are not

89

connected with notable changes of the deprotonation energy. On the other hand, comparison of terminal and bridged sites (HjSiOH with H3SiOH.AlH3 and H2AlOH with (H2AlOH)z or of H20 with H3O+ is in favour of argument (2) above and shows the expected decrease of the shielding constant with decreasing electron density (and increasing acidity). Further calculations are in progress to learn to what extent changes in loHreflect acidity changes. REFERENCES 1 C. Pisani, R. Dovesi and C. Roetti, Hartree-Fock Ab Initio Treatment of Crystalline Systems (Lecture Notes in Chemistry, Vol. 48), Springer-Verlag, Berlin, 1988. 2 J. Sauer, Chem. Rev., 89 (1989) 199. 3 Z. Jirak, S. Vratislav and V. Bosdzek, 1. Phys. Chem. Solids, 41 (1980) 1089. 4 A.K. Cheetham, M.M. Eddy and J.M. Thomas, 1. Chem. SOC.,Chem. Comm., 1984, 1337. 5 E.R. Davidson and D. Feller, Chem. Rev.,86 (1986) 681. 6 W.J. Hehre, L. Radom, P.V.R. Schleyer and J.A. Pople, A b Initio Molecular Orbital Theory, Wiley, New York, 1986. 7 J. Sauer, C. Kolmel and R. Ahlrichs, Chem. Phys. Lett., to be published. 8 R.L. Stevenson, 1. Catal., 21 (1971) 113. 9 M.J. Wax, R.R. Cavanagh, J. Rush, G.D. Stucky, L. Abrams and D.R. Corbin, 1. Phys. Chem., 90 (1986) 532. 10 D. Freude, J. Klinowski and H. Hamdan, Chem. Phys. Lett., 149 (1988) 355. 11 L.M. Kustov, V. Yu. Borokov and V.B. Kazansky, J. Catal., 72 (1981) 149; V.B. Kazansky, L.M. Kustov and V. Yu. Borokov, Zeolites, 3 (1983) 77. 12 L.M. Kustov, V.L. Zholobenko, E. Loffler, C.Peuker and V.B. Kazansky, in preparation. 13 H. Mix, JSauer, K.-P. Schroder and A. Merkel, Coll. Czech., Chem. Comm., 53 (1988)2191. 14 J. Sauer, H. Horn and R. Ahlrichs, I. Phys. Chem., in preparation. 15 J. Sauer, unpublished results. 16 D. Freude and H. Pfeifer, in Proceedings of the 5th International Conference on Zeolites, (L.V.C. Rees, Ed.), Heyden, London 1980, p.490. 17 J.E. Bartmess, J.A. Scott and R.T. McIver, 1. A m . Chem. SOC., 101 (1979) 6046. 18 J. Sauer and A. Bleiber, unpublished results. 19 W.J. Mortier, J. Sauer, J.A. Lercher and H. Noller, 1. Phys. Chem., 88 (1984) 905. 20 J. Sauer and W. Schirmer, in Innovation in Zeolite Materials Science, (PJ. Grobet et al., Eds.), Elsevier, Amsterdam, 1988, p. 323. 21 J. Sauer, 1. Phys. Chem., 91 (1987) 2315. 22 J. Sauer and A. Bleiber, Catalysis Today, 3 (1988) 485. 23 P.J. O'Malley and J. Dwyer, Chem. Phys. Lett., 143 (1988) 97. 24 H. Kawakami, S. Yoshida and T. Yonezawa, 1. Chem. SOC., Faraday Trans. 2, 80 (1984)205. 25 H. Kawakami and S. Yoshida, 1. Chem. SOC.,Faraday Trans 2,81(1985) 1117; and 1. Chem. SOC., Faraday Trans 2, 82 (1986) 1385. 26 L. Pauling, The Nature of the Chemical Bond, 3rd ed., Cornell University Press, Ithaca, 1960. 27 R.T. Sanderson, 1. Am. Chem. SOC., 105 (1983) 2259.

90

28 V.B. Kazansky, in: P.A. Jacobs et al. (Eds.), Structure and Reactivity of Modified Zeolites, Elsevier, Amsterdam, 1984, p.61. 29 W.J. Mortier, 1. Catal., 55 (1978) 138. 30 R.M. Lago, W.O. Haag, R.J. Mikovsky, D.H. Olson, S.D. Hellring, K.D. Schmitt and G.T. Kerr, in Proceedings of the 7th International Zeolite Conference, (Y. Murakami, A. Iijima and J.W. Ward, Eds.), Kodansha Ltd., Tokyo, 1986, p.677. 31 M. Hunger, J. Karger, H. Pfeifer, J. Caro, B. Zibrowius, M. Biilow and R. Mostowicz, J. Chem. SOC.,Faraday Trans 1 , 83 (1987) 3459. 32 H. Pfeifer, D. Freude and M. Hunger, Zeolites, 5 (1985) 274. 33 D. Freude, M. Hunger and H. Pfeifer, Z. Physikal. Chem. NF, 152 (1987) 171. 34 E. Brunner, H. Ernst, D. Freude, M. Hunger and H. Pfeifer, in Innovation in Zeolite Materials Science, (P.J. Grobet et al., Eds.), Elsevier, Amsterdam, 1988,

p.155.

35 G. Engelhardt, H.-G. Jerschkewitz, U.Lohse, P. Sarv, A. Samoson and E. Lippmaa, Zeolites, 7 (1987) 289. 36 J.P. Chauvel Jr. and N.S. True, Chem. Phys., 95 (1985) 435. 37 U. Fleischer, M. Schindler, W. Kutzelnigg, J. Sauer and A. Bleiber, in preparation. 38 M. Schindler and W. Kutzelnigg, J. Chem. Phys., 76 (1982) 1919. 39 J. Sauer, 1. Mol. Catal., submitted.