3882
J. Phys. Chem. B 2002, 106, 3882-3889
Characterization of Acidic OH Groups in Zeolites of Different Types: An Interpretation of NH3-TPD Results in the Light of Confinement Effects Bernd Hunger,*,† Matthias Heuchel,‡ Louis A. Clark,§ and Randall Q. Snurr§ Wilhelm-Ostwald-Institut fu¨ r Physikalische und Theoretische Chemie, UniVersita¨ t Leipzig, Linne´ strasse 2, D-04103 Leipzig, Germany; GKSS Forschungszentrum Geesthacht, Institut fu¨ r Chemie, D-14513 Teltow, Germany; Department of Chemical Engineering and Center for Catalysis and Surface Science, Northwestern UniVersity, EVanston, Illinois 60208 ReceiVed: July 12, 2001; In Final Form: January 10, 2002
Ammonia TPD experiments have been carried out on protonated forms of the zeolites FAU, FER, MFI, and MOR to investigate the acid strength dependence on OH group location and Al content. Effective adsorption energy distributions are derived using a regularization method and corrected to achieve a truer measure of acid strength using nonspecific interaction energies from atomistic Monte Carlo simulations. The correction energies include all interactions not associated with the OH groups and range from 18 to 37 kJ/mol, depending on the region and the zeolite structure. In the FAU structure, we find a bimodal distribution of adsorption energies in both the supercages and the sodalite cages, with the stronger acid sites being more common in the supercages. In contrast, essentially the same acid strength for different regions in MOR is seen, though the acid strength for those sites in the side pockets may be slightly lower. For FER and MFI, the TPD gives unimodal distributions and therefore prohibits distinction between different regions. In agreement with the model of Barthomeuf et al., we see an apparently linear decline (60 kJ/mol) in acid strength beyond a certain aluminum framework density. However, we see a smaller but consistent increase (16 kJ/mol) in acidity before this threshold, despite model predictions that it should be constant.
1. Introduction Zeolites with acid properties are important catalysts for numerous reactions of organic molecules.1-3 These applications require the best possible characterization of the acid properties of the zeolites. This intention needs investigation concerning the nature, number, and strength of acid sites, as well as exploration of their localization, environment, and accessibility. One of the most powerful methods for this involves the adsorption of basic probe molecules. The resulting interaction between probe molecule and Bro¨nsted or Lewis acid sites can be studied by means of different experimental techniques.4,5 A very common method applied to the characterization of acid OH groups (Si-OH-Al groups or bridging OH groups) is the temperature-programmed desorption (TPD) of ammonia.6-21 Due to its small molecular dimensions (3.70 × 3.99 × 3.11 Å3),22 ammonia is a suitable probe for all OH groups accessible through pores, channels, or windows g4 Å. The desorbed amount of ammonia gives information about the number of OH groups. A quantitative analysis of the characteristic desorption curves (calculation of the effective desorption energy; see elsewhere6,7,12,13,15,17,19,23) provides information about the acid strength and the distribution of the interaction strengths, respectively. Because in different types of zeolites the OH groups are located in pores, channels, and cages of different size and shape, a confinement effect should be considered when comparing the acid properties of zeolites on the basis of adsorption of probe molecules.24-26 This confinement effect, * E-mail:
[email protected]. Fax: +49 341 9736399. † Universita ¨ t Leipzig. ‡ GKSS Forschungszentrum. § Northwestern University.
also detectable in the effective desorption energy of NH3-TPD, should result from the nonspecific interaction of ammonia with the zeolite framework. The effect exists also in a certain zeolite, if the OH groups are localized in different segments of the pore system. A detailed interpretation of NH3-TPD therefore demands an assessment of this nonspecific contribution to the overall interaction. Up to now, this problem was seldom considered in applications of NH3-TPD to the characterization of acid properties of zeolites. Previous work has employed a simple, empirical, parametrized analytical model that includes general effects from the pore curvature as well as molecular size and polarizability.24-26 This analytical model provides efficient estimates of the adsorption energy, but more detailed methods are available. In this work, we use conventional atomistic simulations to explicitly incorporate the crystallographically known zeolite structure and realistic van der Waals (Lennard-Jones) and Coulombic interactions. Additionally, we break the nonspecific adsorption energy contributions into separate pore volumes for each zeolite structure considered. This allows us to compare the acidity of OH groups in different pore regions. The promise of being able to determine corrected region or site-specific acidities provides the motivation for the present work. Using GCMC simulations, the strength of the nonspecific interaction was calculated for ammonia using the pure siliceous form of various types of zeolites (FAU, MOR, MFI, and FER) which are also used as catalysts. In doing so, first of all the regions of the pore system were considered in which the acid OH groups could be localized. With these calculated energy values, the effective desorption energy values derived from NH3TPD experiments were corrected to obtain a much more detailed picture of the acid properties of the investigated zeolites.
10.1021/jp012688n CCC: $22.00 © 2002 American Chemical Society Published on Web 03/23/2002
Characterization of Acidic OH Groups in Zeolites
J. Phys. Chem. B, Vol. 106, No. 15, 2002 3883
TABLE 1: Zeolite Characteristics
a
zeolite code
chemical composition
Si/Al
TDAl × 100a
desorbed NH3 amount [mmol/g]
FAU/X(1) FAU/X(2) FAU/X(3) FAU/Y(1) FAU/Y(2) MOR(1) MOR(2) MOR(3) MFI(1) MFI(2) FER(1) FER(2) FER(3)
(NH4)22Na74[Al96Si96O384] (NH4)30Na42[Al72Si120O384] (NH4)31Na36[Al67Si125O384] (NH4)13Na40[Al53Si139O384] (NH4)47Na6[Al53Si139O384] (NH4)6.8Na1.2[Al8Si40O96] (NH4)3.4Na3.8[Al7.2Si40.8O96] (NH4)7.2[Al7.2Si40.8O96] H6[Al6Si90O192] H3.3[Al3.3Si92.7O192] H1.5[Al1.5Si34.5O72] H0.86[Al0.86Si35.14O72] H0.51[Al0.51Si35.49O72]
1.00 1.67 1.87 2.60 2.60 5.00 5.67 5.67 15.00 28.00 23.00 41.00 70.00
9.05 6.78 6.31 5.03 5.03 4.58 4.13 4.13 1.74 0.96 1.20 0.69 0.41
1.7 2.4 2.5 1.1 4.0 2.3 1.2 2.5 0.60 0.28 0.70 0.40 0.24
references to characterization 27 27 14 16 28 28 17 17 29 29 29
Topological density of aluminum atoms calculated according.30
TABLE 2: Potential Parameters for Ammonia-Ammonia (ss) and Ammonia-Zeolite (sz) Interactions
2. Experimental Section The TPD experiments were carried out in a flow apparatus with helium as carrier gas (50 cm3/min). For evolved gas detection, both a thermal conductivity detector (TCD) and a quadrupole mass spectrometer (Leybold, Transpector CIS System) with a capillary-coupling system were used. The zeolites (Table 1) were equilibrated with water vapor over a saturated Ca(NO3)2-solution in a desiccator. For each experiment, 50-250 mg of the granulated zeolite (0.2 to 0.4 mm) was used in a mixture with 1 g of quartz of the same grain size. At first, the samples were flushed with helium at room temperature for 1 h and then heated at 10 K/min in helium up to 373 K. After a period of 1 h at this temperature, they were further heated at 10 K/min to 700 (FAU and FER) and 873 K (MOR and MFI). During this temperature increase, the ammonia desorption (15 amu) of the ammonium forms of the zeolites was recorded. Then, the samples were cooled to temperatures between 373 and 550 K (FAU 373-460 K; MOR 550 K; MFI 515 K; and FER 500 K) and loaded with a surplus of ammonia by dosage of pulses. The loaded samples were flushed with helium until no further desorption was observed. The linear temperature program (10 K/min) was then started. For a quantitative analysis, experiments with different heating rates (2-20 K/min) were carried out. Additionally, with selected zeolites, further experiments were performed using different sample amounts and flow rates of the carrier gas. The desorbed amounts of ammonia were determined by calibration of the TCD signal and the intensity of the 15 amu response. 3. Simulation Methodology Monte Carlo simulations were carried out for ammonia in siliceous zeolite structures. Atomistic adsorption simulations were performed using grand canonical ensemble Monte Carlo (GCMC) simulations and a modified potential model from the literature. The base of atomistic simulations is the potential model; it and the structure of the molecules and sorbent are the only critical assumptions in the simulations. We assume here that the potential energy of the adsorbate-adsorbent system can be modeled as a summation of Lennard-Jones and Coulombic interactions between atomic centers.
νij ) 4ij
[( ) ( ) ] σij rij
12
-
σij rij
6
+
qiqj rij
(1)
Here, νij is the interaction potential between two centers separated by distance rij. Atomic charges are represented by qi and qj.
atom
charge
N H O Si
-1.02 0.34 -0.71 1.42
a
ss/k [K] 85.578 0.0
σss/k [Å] 3.420 0.0
sz/k [K] 87.566 36.785
σsz/k [Å] 3.113 2.653
source 31
LBmixa LBmix 36 36 31
“LBmix” indicates usage of the Lorenz-Bertholet combining rules.
Molecular geometries and sorbate-sorbate OPLS-AA potential parameters (ij, σij, qi, and qj) were taken from Rizzo et al.31 Only rotational and translational degrees of freedom were considered; the molecules were considered rigid. The zeolite was assumed to be fixed at the geometry given by the crystallographic study for each structure: FAU,32 FER,33 MOR,34 and MFI.35 Zeolite oxygen and silicon charges were taken from the work of Kyrlidis et al.36 Charge-charge interactions were calculated using the Ewald summation. Lorenz-Bertholet mixing rules were used to estimate sorbatezeolite Lennard-Jones parameters based on the oxygen potentials given by Snurr et al.37 All potential parameters are given in Table 2. This potential model does not include polarization effects, though their approximate effects are present in the adsorption heats through the Lennard-Jones interactions. Realistic polarization effects would be the next level of potential model improvement. The energy-bias GCMC algorithm of Snurr et al.37 was used in this work to sample the zeolite pore space efficiently. Simulations were performed in the absence of sorbate-sorbate interactions (infinite dilution) using approximately 40 molecules. The pore space of the zeolite was divided into distinct volumes termed “sites” using a method described earlier.38 Averages were computed over at least two million Monte Carlo steps for each ammonia-zeolite system. 4. Results 4.1. Temperature-Programmed Desorption (TPD). Figure 1 shows the ammonia desorption profile of all investigated probes in a normalized mode (initial coverage of NH3 ) 1) at a heating rate of 10 K/min. The desorbed amounts of ammonia are summarized in Table 1. NH3-TPD after ammonia adsorption on the activated faujasites and mordenites at temperatures, characteristic for the respective sample, resulted in reproducible desorption profiles, which are nearly identical with the course of ammonia desorption of the NH4+ forms of the zeolites. Thus it can assumed that during the activation no OH groups have been removed by dehydroxylation. For the H+ form of the
3884 J. Phys. Chem. B, Vol. 106, No. 15, 2002
Figure 1. Ammonia desorption profiles on zeolites of different types (10 K/min).
Hunger et al.
Figure 2. Desorption energy distribution functions of ammonia on the faujasites (A ) 3.5 × 108 min-1).
zeolites (MFI and FER), the adsorption temperature of ammonia (MFI 515 K; and FER 500 K) was chosen in a way such that only the NH3 amount, corresponding to the high-temperature peak of the NH3-TPD, was adsorbed. For the ferrierites, this amount is approximately in agreement with the proton content (see Table 1). In contrast, the desorbed amount for the MFIprobes corresponds only to about 55-74% of the amount of Si-OH-Al groups determined with 1H MAS NMR.17 Probably, the ammonia is more weakly bonded on a part of the OH groups and therefore not recorded at the selected experimental conditions. The observed desorption profiles rd(T) were analyzed by considering a first-order desorption process with a distribution function f(ETPD) of the effective desorption energy ETPD:19,39
rd(T) ) A
dθ (T) ) dt
∫EE
TPD,max
TPD,min
(
θloc(ETPD,T)exp -
)
ETPD f(ETPD)dETPD (2) RT
Here, θ is the average coverage or loading, and A is an effective preexponential factor. θloc is the coverage of adsorption sites with desorption energy ETPD, and ETPD,min and ETPD,max are the limits of the desorption energy range. The assumption of a desorption process of first order for the observed desorption rate is also valid in the cases of a diffusion-limited desorption process or if readsorption effects are decisive. As Tronconi and Forzatti40 could show with model calculations for desorption from a heterogeneous surface, the observed course of desorption can be well described with a first-order rate law containing an effective rate constant. The temperature dependence of this effective rate constant results from both the temperature dependences of an effective diffusion coefficient Deff and the heat of adsorption Qads, respectively. In this context, it could be shown in a recent experimental work for TPD of water on a NaX zeolite41 that the loading dependence of both the effective desorption energy ETPD and the activation energy for the longrange diffusion Elr (determined with (PFG) NMR investigations) are in relative good agreement with the loading dependence of the respective heat of adsorption, Qads.
Figure 3. Desorption energy distribution functions of ammonia on the mordenites (A ) 5 × 107 min-1).
The energy distribution function f(ETPD) was determined from the experimental desorption curves rd(T) by means of the program INTEG, which involves a regularization method for solving the integral eq 2.23,39 The preexponential factor A required to solve eq 2 was estimated by means of an extended integral equation using TPD profiles with different heating rates23 or by use of the dependence of the peak temperature of the desorption profile on the heating rate.15,42 The calculated distribution functions of the effective desorption energy are presented in Figure 2 for faujasites, in Figure 3 for mordenites, and in Figure 4 MFI zeolites and ferrierites. In some cases, for example, FAU/X(1), in Figure 2, the desorption profile dips below the baseline. A constraint could be applied to the numerical procedure to avoid this unphysical behavior. However, in our experience, this constraint itself can produce artifacts, such as “ghost” peaks. For a more detailed discussion of several aspects concerning the inter-relationship of the
Characterization of Acidic OH Groups in Zeolites
J. Phys. Chem. B, Vol. 106, No. 15, 2002 3885
Figure 5. Normalized desorption profiles of ammonia on FAU/Y(2) for different flow rates of helium (10 K/min, 50 mg hydrated zeolite, 1.3 mmol/g desorbed ammonia).
Figure 4. Desorption energy distribution functions of ammonia on the MFI-type zeolites (A ) 3 × 109 min-1) and ferrierites (A ) 1 × 109 min-1).
experimental data, the physical model expressed in the integral kernel, and the numerical procedure used to solve the integral eq 2, see earlier works.39,43 The energy range of the desorption energy distribution functions of the different zeolites is in good agreement with microcalorimetrical values in the literature for comparable zeolites (faujasites,44-47 mordenites,47,48 MFI47,49-51) determined from differential heats of adsorption as function of the ammonia loading. For one of the zeolites of this work, FAU/Y(2), it was shown in an earlier work18 that the effective energy of desorption as function of NH3 loading is in good agreement with the respective dependence of the microcalorimetrically determined heats of adsorption Qads. A good agreement of calorimetrically determined heats of adsorption with results from NH3-TPD on zeolites of different type was also found by other authors52,53 who applied another model based on freely occurring readsorption for the analysis of the TPD profiles. Because in their model a constant heat of adsorption was assumed when considering possible readsorption effects, it is not possible with that model to find evidence about a possible acid strength distribution of the OH groups. Consequently, the distribution functions of the desorption energy also represent an adequate measure for the quantitative characterization of the strength of interaction of ammonia with acid OH groups. The energy value of the pronounced maxima of the distribution functions could be regarded as a measure for the “average” strength of the interaction. Except for the probe FAU/X(1), all distributions have a range of about 20-50 kJ/ mol. This shows clearly the energetic heterogeneity of the interaction of ammonia with the OH-groups in the specific zeolite and allows a conclusion about the acid strength distribution. In this context, it should be noted that TPD investigations of 1-methylpyrrolidine on FAU/Y(2) gave for the stronger acid OH groups a desorption energy distribution of shape very similar to the one determined from ammonia.54 Both bases of different strength “detect,” therefore, the same distribution of the acid strength of the OH groups. The larger basicity of 1-methylpyrrolidine manifests itself only by a larger energy value for the maximum of the distribution. The energetic heterogeneity of acid OH groups in zeolites was also found in IR and NMR
Figure 6. Normalized desorption profiles of ammonia on MFI(2) for different sample amounts (10 K/min, helium flow rate 50 cm3/min).
spectroscopic investigations27,28,55-60 as well as by computational techniques using a combined quantum mechanics-interatomic potential functions approach.61,62 In the context of interpretation of TPD experiments, the problem is often discussed that the run of the desorption curve may depend on the chosen experimental conditions (flow rate of the carrier gas, probe amount), and that therefore for quantitative analysis the specific model assumptions have to be considered (see elsewhere63-65). Hence, we have carried out experiments on selected examples with different amounts of zeolite and different flow rates. Though the maximum of the desorption curve was shifted to lower temperatures in dependence on increasing flow rate, (16.7 cm3/min, Tmax ) 672 K; 83.3 cm3/min, Tmax ) 598 K), the shape of the curve however is not changed, as can be seen clearly in Figure 5 by using normalized curves of desorption (rn ) rd/rd,max, Tn ) T/Tmax). Performing NH3-TPD on this zeolite under vacuum conditions, i.e., without carrier gas, resulted in a desorption profile with nearly equal shape, as it could be shown in a recently published work.66 Therefore, diffusive influence of varying strength and possible effects of readsorption should be excluded here as responsible for the shape of the profiles. By use of larger amounts of zeolite, the maximum of the desorption curve is shifted to higher temperatures. But also in this case, the characteristic shape of the curve is not changed, as can be seen in Figure 6 for the MFI zeolite. However, since it is mainly the width of the experimental desorption profile which determines the range of the desorption energy distribution calculated with eq 2, we can assume that the choice of the experimental conditions has no influence on it. The shift of the maximum of the desorption curve as a function of flow rate and zeolite
3886 J. Phys. Chem. B, Vol. 106, No. 15, 2002
Hunger et al.
TABLE 3: GCMC Simulation Results
zeolite
channel/cage
Lennard-Jones term [kJ/mol]
electrostatic term [kJ/mol]
average interaction energy 〈U〉 [kJ/mol]
EC [kJ/mol]
siting numbers at infinite dilution [%]
EC26 [kJ/mol ]
FAU
supercage sodalite unit
-9.3 -19.1
-12.6 -0.9
-21.9 -20.0
24.4 22.5
63.9 37.1
14
MOR
12-ring pores 8-ring pores side pockets
-11.5 -14.1 -17.3
-4.7 -11.1 -17.4
-16.2 -25.2 -34.7
18.7 27.7 37.2
1.6 0.3 98.1
26
MFI
intersections straight channels zigzag channels
-13.2 -15.0 -14.8
-2.6 -7.4 -7.8
-15.8 -22.5 -22.5
18.3 25.0 25.0
2.9 45.1 52.0
18
FER
intersections 10-ring pores 8-ring pores
-14.9 -15.3 -15.8
-0.6 -2.0 -2.8
-15.5 -17.3 -18.7
18.0 19.8 21.2
15.8 23.9 60.3
amount determines therefore only the value of the effective preexponential factor A, as it was already shown for the example NH3-TPD on MFI zeolites with different zeolite amounts.17 From this follows that it is always necessary, in a quantitative analysis of nonisothermal desorption profiles, to determine the effective preexponential factors for the particular applied experimental conditions. 4.2. Monte Carlo Simulation. The calculated average interaction energies 〈U〉 of ammonia with the zeolitic framework of FAU, MOR, MFI, and FER are summarized in Table 3. Based on previous experience reproducing experimental adsorption data, these potentials should provide reasonable agreement (heats within 10-20%) with experiment. However, isosteric heats over siliceous MFI reported by Bolis et al.67 are significantly higher (about 60 kJ/mol) than those we can calculate (22 + RT kJ/ mol). This experimental work is the only measurement of ammonia heats over a siliceous zeolite in the literature. In this case, the discrepancy between simulation and experiment can be attributed to the large concentration of internal defect hydroxyl groups, for which the authors provide infrared spectroscopy evidence. Since the siting in column 7 of Table 3 was determined at infinite dilution, it should be noted that it is not directly comparable to the experimental conditions, where we expect some significant change in the siting due to the presence of acid sites and the sorbate-sorbate interactions. Overall, in pore segments where OH groups are located, they show values between 16 and 35 kJ/mol. The span and approximate values of these adsorption energies are very similar to those recently measured by Yang et al.68 for CH4 over a variety of high-silica zeolite structures. In contrast to previous work considering only van der Waals (Lennard-Jones) interactions38 or the analytical model from Derouane et al.,24-26 we see less correlation between smaller pore regions and stronger adsorption. The electrostatic contributions have a more complex dependency on the pore shape, and in some cases, even counteract the trends from the LennardJones contributions. This leads in some zeolite types to very similar total interaction energies in different region, e.g., in faujasite (about 20 kJ/mol in sodalite cages versus 22 kJ/mol in supercages) and ferrierite (about 17 kJ/mol in 10-ring pores versus 19 kJ/mol in 8-ring pores). In contrast, the interaction energies in the different pore regions of mordenite show a much wider distribution. Compared to the interaction energy in the 12-ring pores, those in the flattened 8-ring pores and side pockets are about 9 and 19 kJ/mol stronger, respectively. Some of the seemingly unpredictable dependence of the electrostatic contributions on the pore shape may be attributable to the fit of the dipolar ammonia in the various pore regions.
39
Stronger interactions will result when the positively charged hydrogen atoms on one side of the ammonia can approach the negatively charged oxygen atoms along the pore walls more closely. Conversely, less favorable, weaker interactions will result when the negatively charged nitrogen atom must occupy similar positions. In the tight, approximately spherical sodalite cages, the nitrogen must remain close to the pore wall, thereby counteracting the favorable interaction of the hydrogens with the wall and making the total electrostatic contribution small. A more detailed investigation of this type of effect has been performed for benzene adsorption in MFI.69 The simulated averaged nonspecific interaction energies 〈U〉 have been used to calculated an effective confinement contribution EC according to:
EC ) RT + HR - 〈U〉
(3)
where HR is the molar residual enthalpy of the pure component in the real gas state. For the thermodynamic conditions of the TPD experiments (P ) 1 bar, T > 300 K), HR is only a very small correction below 0.1 kJ/mol70 and can be neglected in the following discussion. Thus, the effective confinement contribution EC has the meaning of an isosteric heat at infinite dilution in the absence of OH groups. These EC values are listed in Table 3, as well as those available from the work of Derouane et al.26 Where comparison is possible, it is seen that the values calculated in this work have a smaller spread. We attribute the differences to the inclusion of electrostatic interactions and atomistic detail in our simulations. In consideration of this part of the nonspecific interaction on the effective desorption energy, it is now possible to calculate the protonation energy EP which expresses the strength of interaction of ammonia with the OH groups and is thus a measure for the “true,” i.e., intrinsic, acid strength and acid strength distribution of Si-OH-Al groups, respectively:
EP ) ETPD - EC
(4)
The validity of the decomposition of adsorption heat into specific and nonspecific contributions has recently been questioned.68 This is an important point because localization of molecules on a site could affect the nonspecific heat by changing the distribution of the adsorbates in the pores. We hope, by using region-specific corrections, that we can account for some of this effect. In the following section, the values obtained in this way for the strength of the OH groups in different zeolites are discussed, and the acid properties are compared with each other.
Characterization of Acidic OH Groups in Zeolites
Figure 7. Desorption energy distribution functions of ammonia on FAU/Y(2). Solid circles: Si-O(1)H-Al groups (supercage). Open circles: Si-O(3)H-Al groups (sodalite unit).
5. Discussion 5.1. Faujasites. High-resolution neutron powder diffraction studies have shown that in HY zeolites the oxygen atoms O(1) and O(3) are the preferred positions for the protons.71 The O(1)H groups are located in the supercage and the O(3)H groups in the sodalite unit. Due to the small proton content, the samples FAU/X(1), FAU/X(2), FAU/X(3),27 and FAU/Y(1) zeolite72 possess Si-OH-Al groups only in the supercage. Based on the energy value for the maximum of the desorption energy distribution EC of 24.4 kJ/mol (see Table 3), the nonspecific interaction portion of the effective desorption energy adds up to values between 31% (FAU/X(1)) and 23% (FAU/Y(1)). The desorption energy distribution of FAU/X(1) (Figure 2) with a Si/Al ratio of 1 shows a peak at 80 kJ/mol with only a very small width (the width at half-height is about 8 kJ/mol). From this, it follows that in this zeolite all the Si-O(1)H-Al groups have almost the same acid strength. This result is in very good agreement with IR- and NMR-spectroscopic investigations of Datka et al.27,55-57 The increase in the width of the distribution function for X zeolites at lower Al content as well as the simultaneous establishment of a distinct bimodal distribution at FAU/Y(1) (Figure 2) indicates an increase of the energetic heterogeneity of the Si-O(1)H-Al groups in the supercage. This result agrees well with extensive spectroscopic investigations of Datka et al.,27,55-57 and it is mainly discussed in the context of Al content and distribution in these zeolites. The shift of the desorption energy maximum from 80 kJ/mol for FAU/ X(1) to an energy value of 105 kJ/mol for FAU/Y(1) shows also that the mean acid strength of the OH groups increases as the aluminum content decreases. For the Y zeolite with the larger proton content (FAU/Y(2)), the OH groups are localized in the supercage as well as in the sodalite unit.72 Because the mean specific interaction energy of ammonia for both types of cavity is approximately similar (see Table 3), the effective desorption energy itself may be considered for a comparison of the acid strength of OH groups in both cavity regions. Previous FTIR-TPD investigations of ammonia on this probe20 have shown that both peaks of the pronounced bimodal desorption energy distribution (Figure 2 below) contain variable portions of ammonia desorption from OH groups localized in the supercage and the sodalite unit. Figure 7, which has already been discussed in a previous publication,20 shows the desorption energy distributions for both types of OH groups obtained from FTIR-TPD experiments. Because of the similar nonspecific interaction in both pore regions, it therefore follows that, in these zeolites, O(1)H as well as O(3)H groups must exist which differ from each other
J. Phys. Chem. B, Vol. 106, No. 15, 2002 3887 in their acid strength by a value of about 20 kJ/mol. In the supercage, about 64 % of the OH groups are more acidic, while in the sodalite units the figure is only about 38%. For the Y zeolites, with increasing proton content the maximum of the low energy peak of the desorption energy distribution does not shift in Figure 2, but decreases in relative height. The high energy peak, on the other hand, increases in relative height and shifts about 5 kJ/mol toward higher desorption energies. This result indicates an increase of the average acidity with increasing proton content which was already found by use of other probe molecules.58,73 This effect on the acid strength of the OH groups is however much less pronounced in comparison to that of the Al content of the faujasites. 5.2. Mordenites. Recent neutron powder diffraction studies have shown that in mordenite the OH groups are localized in the 12-ring pores (main or broad channels), in the 8-ring pores, and in the side pockets.74 Further, it is known that up to a sodium cation exchange degree of about 50%, the OH groups are only formed in the 12-ring pores.28,55,75 At this exchange level, the desorption energy distribution function for MOR(2) shows a distinct maximum at 129 kJ/mol (see Figure 3). About 15% of this energy value should be assigned to nonspecific interactions (Table 3). With increasing proton content, OH groups are also generated in the 8-ring pores and in the side pockets. The distribution function of the effective desorption energy has, in such cases (MOR(1) and MOR(3)), a maximum at 137-138 kJ/mol (Figure 3). The nonspecific interaction of ammonia in this pore region is 9-19 kJ/mol larger than that in the 12-ring pores (see Table 3). A stronger interaction as a consequence of confinement effects in these pore regions was also found in IRspectroscopic investigations of adsorption of nitrile compounds on mordenites.76 The consideration of these different nonspecific interaction energies for the 12-ring pores and the other smaller pore regions shows that the OH groups in both 12-ring and flattened 8-ring regions of the pore system have essentially the same acidity (EP ) 110 kJ/mol). In the side pockets, we find a somewhat lower mean acid strength (EP ) 100 kJ/mol). Since the distribution functions at larger energy values do not show any further structuring, it is not possible to discriminate or attribute the ammonia desorption specifically to the 8-ring pores or side pockets. A comparatively large OH group acidity in the 12-rings has also been found in other investigations.77 5.3. MFI-Type Zeolites and Ferrierites. Probably as result of the small content of OH groups, there exist up to now no experimental investigations concerning the localization of OH groups in MFI-type zeolites. Therefore, theoretical investigations are of increased interest. Early calculations using semiempirical quantum chemical methods78 found that T(2) and T(12) atoms were, because of energetic reasons, the preferred positions for Al atoms and therefore for the OH groups. In this case, the OH groups would be localized in the channel crossings. However, more recent calculations, using more accurate methods, make this preference much less clear.79 Classical defect energy calculations by Schro¨der et al.80 also result in a picture of a more random distribution of OH groups. With neutron powder diffraction studies on D-ferrierites, Martucci et al.81 found that in this zeolite type the OH groups are located in the 10-ring channels and near the center of the 8-ring of the ferrierite cage facing toward the channel. For both types of zeolites, the desorption energy distributions resulting from NH3-TPD show a significant maximum which is shifted with decreasing Al content of the zeolite to somewhat lower energy values (Figure 4). The distributions do not possess any discernible structural detail. Since the calculated nonspecific
3888 J. Phys. Chem. B, Vol. 106, No. 15, 2002
Hunger et al. Additionally, they show that the NH3-TPD investigation is an adequate means to describe the acid properties of zeolites. Acknowledgment. B.H. gratefully acknowledges the financial support of the Deutsche Forschungsgemeinschaft, Graduate College “Physical Chemistry of Interfaces” and the Fonds der Chemischen Industrie. R.Q.S. thanks the U.S. National Science Foundation CAREER program for financial support. The authors thank Prof. J. Datka (Jagiellonian University, Cracow) for providing some faujasite and mordenite samples and Dr. D. Rutenbeck (University of Leipzig) for providing the ferrierites. References and Notes
Figure 8. Correlation between the protonation energy EP and the topological density of aluminum atoms TDAl of the zeolites. Solid symbols: this work. Open circles: calculated with literature values of microcalorimetrically determined heats of adsorption: 1,51 2,47 3,49 4,48 5,45 6,44 7,82 and 8.53
interaction contributions (see Table 3) for all considered channels of both zeolites differ by only about 7 kJ/mol at maximum, we can assume that the mean acidity of all OH groups in MFI and FER are approximately equal. 5.4. Comparison of the Acid Strength of All Zeolites. In addition to structural influences, the acidity of Si-OH-Al groups in zeolites is particularly determined by the Al content.30 With decreasing Al content, the probability increases that an Al atom has no other Al atom in close vicinity. In the limit of such isolation, the OH groups can be expected to have very similar acid strengths. To take this influence into consideration, Barthomeuf30 derived, on the basis of a structure-dependent topological density in the second to fifth coordination sphere of the SiO4 and AlO4 tetrahedrons, a dimensionless parameter, the so-called topological density of Al atoms TDAl. Below a value of TDAl ) 2.65 × 10-2, all OH groups should (because topological and chemical effect are considered) have the same acidity independent of the zeolite structure. In Figure 8, the obtained protonation energies (eq 4), calculated at each case with the energy value corresponding to the maximum value of the desorption energy distribution function, are presented as a function of the topological density of the Al atoms in the investigated zeolites. For comparison, additional values are given in the figure which have been derived from published microcalorimetrically determined heats of adsorption. It can be seen clearly that down to the value given by Barthomeuf of TDAl ) 2.65 × 10-2, the mean acid strength of the OH groups increases linearly and afterward can be assumed nearly constant. The linear correlation (r2 ) 0.8473) gives for TDAl ) 2.65 × 10-2 a value of EP of 118 ( 6 kJ/mol. In the framework of the errors of this investigation, this value agrees relatively well with the values found here for the zeolites: MFI(1) EP ) 125 ( 10 kJ/mol; MFI(2) EP ) 116 ( 10 kJ/mol; FER(1) EP ) 117 ( 10 kJ/mol; FER(2) EP ) 113 ( 10 kJ/mol; and FER(3) EP ) 109 ( 10 kJ/mol. However, looking at the left side of Figure 8, we find a certain systematic decrease in the acid strength of the OH groups with decreasing Al content of the MFI zeolites and ferrierites, about which at present only speculations are possible. A heterogeneous Al distribution as well as contamination of the zeolites could be the reason. Altogether, the obtained results support the hypothesis set up by Barthomeuf about the influence of the topological density of the Al atoms on the acid strength of OH groups in zeolites.30
(1) Weitkamp, J.; Weiβ, U.; Ernst, S. Stud. Surf. Sci. Catal. 1995, 94, 363. (2) Venuto, P. B. Stud. Surf. Sci. Catal. 1997, 105, 811. (3) Tanabe, K.; Ho¨lderich, W. F. Appl. Catal. A: Gen. 1999, 181, 399. (4) Karge, H. G. Stud. Surf. Sci. Catal. 1991, 65, 133. (5) Weitkamp, J., Puppe, L., Eds. Catalysis and Zeolites - Fundamentals and Applications; Springer-Verlag: Berlin, 1999; Chapter 4. (6) Karge, H. G.; Dondur, V. J. Phys. Chem. 1990, 94, 765. (7) Karge, H. G.; Dondur, V.; Weitkamp, J. J. Phys. Chem. 1991, 95, 283. (8) Katada, N.; Igi, H.; Kim, J.-H.; Niwa, M. J. Phys. Chem. B 1997, 101, 5969. (9) Miyamoto, T.; Katada, N.; Kim, J.-H.; Niwa, M. J. Phys. Chem. B 1998, 102, 6738. (10) Post, J. G.; van Hooff, J. G. C. Zeolites 1984, 4, 9. (11) Topsøe, N.-Y.; Pedersen, K.; Derouane, E. G. J. Catal. 1981, 70, 41. (12) Costa, C.; Lopes, J. M.; Lemos, F.; Ribeiro, F. R. J. Mol. Catal. A: Chem. 1999, 144, 221. (13) Costa, C.; Dzikh, I. P.; Lopes, J. M.; Lemos, F.; Ribeiro, F. R. J. Mol. Catal. A: Chem. 2000, 154, 193. (14) Hoffmann, J.; Hunger, B.; Streller, U.; Stock, Th.; Dombrowski, D.; Barth, A. Zeolites 1985, 5, 31. (15) Hunger, B.; Hoffmann, J. Thermochim. Acta 1986, 106, 133. (16) Hunger, B.; Hoffmann, J.; Mothsche, P. J. Therm. Anal. 1987, 32, 2009. (17) Hunger, B.; Hoffmann, J.; Heitzsch, O.; Hunger, M. J. Therm. Anal. 1990, 36, 1379. (18) Hunger, B.; v. Szombathely, M. Z. Phys. Chem. 1995, 190, 19. (19) Hunger, B.; v. Szombathely, M.; Hoffmann, J.; Bra¨uer, P. J. Therm. Anal. 1995, 44, 293. (20) Hunger, B.; Miessner, H.; v. Szombathely, M.; Geidel, E. J. Chem. Soc., Faraday Trans. 1996, 92, 499. (21) Hunger, B.; Datka, J. J. Therm. Anal. 1998, 53, 217. (22) Webster, C. E.; Drago, R. S.; Zerner, M. C. J. Am. Chem. Soc. 1998, 120, 5509. (23) Koch, K.; Hunger, B.; Klepel, O.; Heuchel, M. J. Catal. 1997, 172, 187. (24) Derouane, E. G. Chem. Phys. Lett. 1987, 142, 200. (25) Derouane, E. G. J. Mol. Catal. A: Chem. 1998, 134, 29. (26) Derouane, E. G.; Chang, C. D. Microporous Mesoporous Mater. 2000, 35-36, 425. (27) Gil, B.; Broclawik, E.; Datka, J.; Klinowski, J. J. Phys. Chem. 1994, 98, 930. (28) Datka, J.; Gil, B.; Weglarski, J. Microporous Mesoporous Mater. 1998, 21, 75. (29) Rutenbeck, D. Thesis, University of Leipzig, 2000. (30) Barthomeuf, D. Mater. Chem. Phys. 1987, 17, 49. (31) Rizzo, R. C.; Jorgensen, W. L. J. Am. Chem. Soc. 1999, 121, 4827. (32) Hriljac, J. A.; Eddy, M. M.; Cheetham, A. K.; Donohue, J. A.; Ray, G. J. J. Solid State Chem. 1993, 106, 66. (33) Vaughan, P. A. Acta Crystallogr. 1966, 21, 983. (34) Alberti, A.; Davoli, P.; Vezzalini, G. Z. Kristallogr. 1986, 175, 249. (35) van Koningsveld, H.; Tuinstra, F.; van Bekkum, H.; Jansen, J. C. Acta Crystallogr. 1989, B45, 423. (36) Kyrlidis, A.; Cook, S. J.; Chakraborty, A. K.; Bell, A. T.; Theodorou, D. N. J. Phys. Chem. 1995, 99, 1505. (37) Snurr, R. Q.; Bell, A. T.; Theodorou, D. N. J. Phys. Chem. 1993, 97, 13742. (38) Clark, L. A.; Gupta, A.; Snurr, R. Q. J. Phys. Chem. B 1998, 102, 6720. (39) v. Szombathely, M.; Bra¨uer, P.; Jaroniec, M. J. Comput. Chem. 1992, 13, 17. (40) Tronconi, E.; Forzatti, P. Chem. Eng. Sci. 1987, 42, 2779.
Characterization of Acidic OH Groups in Zeolites (41) Kirmse, A.; Ka¨rger, J.; Stallmach, F.; Hunger, B. Appl. Catal. A: Gen. 1999, 188, 241. (42) Dawson, P. T.; Peng, Y. K. Surf. Sci. 1972, 33, 565. (43) Heuchel, M.; Jaroniec, M.; Gilpin, R. K.; Bra¨uer, P.; v. Szombathely, M. Langmuir 1993, 9, 2537. (44) Kriva´nek, M.; Jı´ru, P. Collect. Czech. Chem. Commun. 1984, 49, 2739. (45) Lohse, U.; Parlitz, B.; Patzelova´, V. J. Phys. Chem. 1989, 93, 3677. (46) Tsutsumi, K.; Mitani, Y.; Takahashi, H. Bull. Chem. Soc. Jpn. 1983, 56, 1912. (47) Kapustin, G. I.; Brueva, T. R.; Mishin, I. V. Proceedings of the 12th International Zeolite Conference, Baltimore, 1998; Treacy, M. M. J., Marcus, B. K., Bisher, M. E., Higgins, J. B., Eds.; Materials Research Society, Warrendale, 1999; p 2637. (48) Klyachko, A. L.; Bankos, I.; Brueva, T. R.; Kapustin, G. I. React. Kinet. Catal. Lett. 1985, 29, 451. (49) Parrillo, D. J.; Lee, C.; Gorte, R. J. Appl. Catal. A: Gen. 1994, 110, 67. (50) Sayed, M. B.; Auroux, A.; Vedrine, J. C. Appl. Catal. 1986, 23, 49. (51) Ducourty, B.; Occelli, M. L.; Auroux, A. Thermochim. Acta 1998, 312, 27. (52) Niwa, M.; Katada, N.; Sawa, M.; Murakami, Y. J. Phys. Chem. 1995, 99, 8812. (53) Kapustin, G. I.; Brueva, T. R.; Klyachko, A. L.; Beran, S.; Wichterlova´, B. Appl. Catal. 1988, 42, 239. (54) Hunger, B.; v. Szombathely, M. Stud. Surf. Sci. Catal. 1994, 84, 669. (55) Datka, J.; Boczar, M.; Gil, B. Colloids Surf. A: Physicochem. Engin. Aspects 1995, 105, 1. (56) Datka, J.; Gil, B. J. Catal. 1994, 145, 372. (57) Datka, J.; Boczar, M.; Gil, B. Langmuir 1993, 9, 2496. (58) Kubelkova´, L.; Beran, S.; Lercher, J. A. Zeolites 1989, 9, 539. (59) Brunner, E. J. Mol. Struct. 1995, 355, 61. (60) Brunner, E. Catal. Today 1997, 38, 361.
J. Phys. Chem. B, Vol. 106, No. 15, 2002 3889 (61) Sierka, M.; Eichler, U.; Datka, J.; Sauer, J. J. Phys. Chem. B 1998, 102, 6397. (62) Sierka, M.; Sauer, J. J. Phys. Chem. B 2001, 105, 1603. (63) Gorte, R. J. J. Catal. 1982, 75, 164. (64) Demmin, R. A.; Gorte, R. J. J. Catal. 1984, 90, 32. (65) Rieck, J. S.; Bell, A. T. J. Catal. 1984, 85, 143. (66) Trunschke, A.; Hunger, B. Topics Catal., in press. (67) Bolis, V.; Bordiga, S.; Lamberti, C.; Zecchina, A.; Carati, A.; Rivetti, F.; Spano`, G.; Petrini, G. Langmuir 1999, 15, 5753. (68) Yang, L.; Trafford, K.; Kresnawahjuesa, O.; ? Sepa, J.; Gorte, R. J.; White, D. J. Phys. Chem. B 2001, 105, 1935. (69) Clark, L. A.; Snurr, R. Q. Chem. Phys. Lett. 1999, 308, 155. (70) Smith, J. M.; van Ness, H. C.; Abbott, M. M. Introduction to Chemical Engineering Thermodynamics, 5th ed.; McGraw-Hill: New York, 1996. (71) Czjzek, M.; Jobic, H.; Fitch, A. N.; Vogt, T. J. Phys. Chem. 1992, 96, 1535. (72) Hunger, M. Solid State Nucl. Magn. Reson. 1996, 6, 1. (73) Chen, D. T.; Sharma, S. B.; Filimonov, I.; Dumesic, J. A. Catal. Lett. 1992, 12, 201. (74) Martucci, A.; Cruciani, G.; Alberti, A.; Ritter, C.; Ciambelli, P.; Rapacciuolo, M. Microporous Mesoporous Mater. 2000, 35-36, 405. (75) Datka, J.; Gil, B.; Kubacka, A. Zeolites 1997, 18, 245. (76) Marie, O.; Thibault-Starzyk, F.; Lavalley J.-C. Phys. Chem. Chem. Phys. 2000, 2, 5341. (77) Maache, M.; Janin, A.; Lavalley, J.-C.; Benazzi, E. Zeolites 1995, 15, 507. (78) Derouane, E. G.; Fripiat, J. G. Zeolites 1985, 5, 165. (79) Nachtigallov, D.; Sierka, M.; Sauer, J. Phys. Chem. Chem. Phys. 1999, 1, 2019. (80) Schro¨der, K.-P.; Sauer, J.; Leslie, M.; Catlow, C. R. A. Zeolites 1992, 12, 20. (81) Martucci, A.; Alberti, A.; Cruciani, G.; Radaelli, P.; Ciambelli, P.; Rapacciuolo, M. Microporous Mesoporous Mater. 1999, 30, 95. (82) Parrillo, D. J.; Gorte, R. J. J. Phys. Chem. 1993, 97, 8786.