Selectivity in Catalysis - American Chemical Society

Chapter 21

Preparation, Characterization, and Catalysis of a Modified ZSM-5 Zeolite

Downloaded by UNIV OF SYDNEY on May 3, 2015 | http://pubs.acs.org Publication Date: May 5, 1993 | doi: 10.1021/bk-1993-0517.ch021

Anne M. Gaffney1, C. A. Jones1, John A. Sofranko1, Chihji Tsiao2, and Cecil Dybowski2 1 A R C O Chemical Company, 3801 West Chester Pike, Newtown Square, P A 19031 2Center for Catalytic Science and Technology, University of Delaware, Newark, D E 19716

Protonated and modified ZSM-5 zeolites were prepared, characterized and used in the conversion of methanol to gasoline and in hydrocarbon cracking. The modified ZSM-5 was prepared by heating Na-ZSM-5 with a protonated, sterically hindered amine salt such that only cationic sites external to the zeolite channels underwent exchange. Its constraint index number was significantly lower than that of the unmodified H-ZSM-5, reflecting external protonic sites. Also, its sodium content was higher due to retention of internal sodium. NMR spectrometry of129Xegas sorbed in the modified and slightly coked zeolite demonstrated coking occurred primarily on the surface; whereas, the H-ZSM-5 zeolite was shown to deposit coke internally. In the conversion of methanol to gasoline, the modified zeolite was significantly less reactive. However, both zeolites had comparable activity in hydrocarbon cracking reactions carried out at high space velocities, i.e., under diffusion limited conditions.

Catalyst Preparation A modified Na, H-ZSM-5 was prepared which had protonated, external sites and cationic, internal sites. Initially, the Na-ZSM-5 was prepared according to standard procedures (i). A solution of 200 g of NaSi0 in 150 ml of water was mixed with a solution of 75 g of Na CI, 6.7 g of Al (S0 ) »16 H 0, 19.6 g of H S0 in 340 ml of water for 5 minutes. To this mixture was added 25 g of (npropyl) Ν · Β Γ in 100 ml of water. After 50 minutes of mixing, a portion was charged to an autoclave and heated at 149°C for 16 hours. The zeolite was filtered off and washed thoroughly with water; dried at 110°C; and calcined at 550°C for 16 hours. To 10 g of the Na-ZSM-5, was added 500 ml of a 1 molar aqueous solution of (n-butyl) N«HCI. This mixture was heated at 80°C and 2

2

2

4

3

2

4

4

3

0097-6156/93/0517-0316$06.00/0 © 1993 American Chemical Society

In Selectivity in Catalysis; Davis, M., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1993.

21. GAFFNEY ET AL.

Catalysis of a Modified ZSM-5 Zeolite

317

stirred for 2 hours. The solids were filtered, washed and recontacted with a fresh 500 ml of solution. This procedure was repeated a total of 5 times. The modified Na,H-ZSM-5 was dried at 110°C and calcined at 550°C for 16 hours. Elemental analysis of the modified zeolite indicated that it contained 0.32 wt % Na as compared to 4.3 wt % Na for Na-ZSM-5 and < 0.02 wt % Na for H-ZSM-5.


Catalyst Coking Catalyst coking studies were carried out on the protonated and modified ZSM-5 zeolites. Approximately 0.8 g of zeolite was heated to 823 Κ followed by exposure to 2-butene at 100 ml/min for a specific time to obtain samples with varying degrees of coking, as shown in Table I. To achieve a similar wt % coke, the modified Na, H-ZSM-5 zeolite required a 33% longer exposure time to 2-butene. The coke content of the zeolite samples was measured on a Perkin-Elmer 240B CHN instrument which uses a combustion method to convert the sample elements to simple gases (C0 , H 0 and N ). The sample is first oxidized in a pure oxygen environment; the resulting gases are then controlled to exact conditions of pressure, temperature and volume. Finally, the product gases are separated. Then, under steady-state conditions, the gases are measured as a function of thermal conductivity. The results are accurate to + 0.5%, absolute. 2

Table I.

2

2

Treatment and Coke Contents of Protonated H-ZSM-5 and Modified Na, H-ZSM-5 Samples Exposure to 2-Butene (min)

Coke Content Wt%

A. Na,H-ZSM-5 Fresh Slightly coked Heavily coked

0 1.6 48

0 1 12

B. H-ZSM-5 Fresh Slightly coked Heavily coked

0 1.2 36

0 1 12

Sample

129

Xe NMR Spectrometry

NMR spectrometry of Xenon-129 adsorbed in coked samples of the totally protonated H-ZSM-5 zeolite and the modified Na, H-ZSM-5 showed variations attributable to differences in coke distribution. X e NMR spectrometry is extremely useful for probing microporous materials. Ito et al.fi/ demonstrated, for example, that NMR spectrometry of adsorbed xenon in coke-fouled H-Y zeolite could probe the deposits after coking and the nature of the internal surfaces after decoking. The NMR results in this study are consistent with a distribution of coke restricted by size selectivity of the acidifying medium. 129


318

SELECTIVITY IN CATALYSIS

1 2 9

All coked and uncoked samples were pretreated for X e NMR spectrometry by the following procedure: Approximately 0.4 g of material was placed in an NMR tube with a resealable coaxial Teflon stopcock. The sample was outgassed to a pressure of 1.0 χ 10" Torr at 295 K. It was slowly heated to 673 Κ over 6 hours and maintained at that temperature for 12 hours under dynamic vacuum. After cooling to 295 K, the sample was exposed to xenon gas (Air Products and Chemicals, 99.99%) for 15 minutes to allow equilibrium to be reached. Subsequently, the tube was sealed and removed from the manifold. Xenon uptake isotherms were measured at 295 Κ during this successive adsorption process. They show Langmuir-type dependences of uptake on pressure in the range up to 700 Torr, as can be seen in Figure 1. X e NMR spectra were recorded with a Bruker WM-250 NMR spectrometer at 69.19 M Hz. Each spectrum is the accumulation of 500 transients with a relaxation delay of 0.5 seconds. The chemical shifts are reported with respect to bulk xenon gas extrapolated to P=0, using a secondary standard of xenon adsorbed in Na-Y zeolite at 590 Ton-ft/. Positive chemical shifts are to higher frequency than the reference. Figures 1A and 1B show the adsorption isotherms of xenon on the Na, H-ZSM-5 and H-ZSM-5 zeolites, respectively. From the comparison, one sees that xenon uptake decreases slightly (about 10%) with coke content in the Na, H-ZSM-5 with a low (1%) coke content, on zeolite H-ZSM-5, and decreases only slightly more with heavy coking (12%). Figures 2A and 2B give the xenon-129 chemical shift as a function of the uptake, p, of xenon per gram of dry zeolite for the Na, H-ZSM-5 and the H-ZSM-5 zeolites, respectively. One can see that the chemical shift in these materials obey the truncated form of Fraissard's equation^-*; : 5


1 2 9

o(p) = o + otf,

(1)

0

129

where σ(ρ) is the observed X e chemical shift at a given uptake. σ is the extrapolated chemical shift at infinitesimally low uptake, a term characteristic of the interactions between the xenon atom and the structure in which it is adsorbed, is the first-order coefficient in a virial expansion of the chemical shift and is a measure of the effects of xenon-xenon interactions. From the graphs one obtains the Fraissard parameters for xenon under various conditions of treatment of the ZSM-5 zeolites; these are listed in Table II. Figure 2A shows that, for the Na, H-ZSM-5 zeolite, the plots are nearly parallel, whatever the coke level. The intercepts, σ , do depend on the pretreatment conditions and increase with coke content. From Figure 2B, a much different picture emerges for xenon in H-ZSM-5. Again the plots obey Equation 1. However, coking of H-ZSM-5 alters the slope of the profile, while leaving the intercept at low coverage nearly the same. See Table II. 0

0

Discussion on

129

Xe NMR Results

Demarquay and Fraissard^ have shown that the intercept, σ , can be interpreted in terms of a "mean free path" of xenon in the zeolite and have related this NMR-derived mean free path to structural parameters such as channel diameter by a Monte Carlo simulation procedure. They empirically determined the intercept to depend on mean free path as 0

σ = Α/(Β + λ), 0


(2)


319


GAFFNEY ET AL.

0

50 100 150 200 250 300 350 400 450

500

X e n o n P r e s s u r e (torr) Figure 1. (a) Xenon adsorption isotherms (at 297 K) of the size-selectively modified Na,H-ZSM-5 zeolites having different coke contents: •-uncoked; A--1 wt % coke; B--12 wt % coke, (b) Xenon adsorption isotherms (at 297 K) of fully protonated H-ZSM-5 zeolites having different coke contents: • -uncoked; A - 1 wt % coke; B--12 wt % coke. (Reproduced with permission from ref. 16. Copyright 1991 Academic Press Inc.



320

Table II. Fralssard Parameter, Effective Channel Diameters, and Decrease of Void Volume for Samples after Various Treatments Sample

σ

0

(ppm)

1

Dchannel(A)

a

V/Vo

(ppm-g/M mol)


A.

Na, H-ZSM-5 Fresh Slightly Coked Heavily Coked

101.310.7 104.6±1.3 107.7±0.6

23.1 10.5 23.211.1 23.510.5

7.3 7.1 7.0

1.00 0.98

B. H-ZSM-5 Fresh Slightly Coked Heavily Coked

103.1 11.1 103.410.5 102.910.8

18.710.8 31.310.6 39.211.0

7.2 7.2 7.2

0.60 0.48

a

—

b

b

—

b

b

= Calculated using Equations 1 and 2 with A = 499, Β = 2.054, andD = 4.4À. = These quantities have associated uncertainties of l 0.05. xenon

b

where λ is the mean free path in the structure. For an ideal infinitely long straight channel, λ is related to the channel diameter by the equation λ = D

channe

| - D

xenon

·

(3)

Cheung and Fu (7) have derived an equation of similar form from a simple physical model. Recently, a relationship between the extrapolated chemical shift at low coverage, σ , and the average pore size of zeolites determined by argon uptake as analyzed by the Horrath-Kawazoe Method^ has been shown. Such derived parameters as the mean free path and average pore size seem to scale with other measures of the structure, but they do not necessarily correspond quantitatively to measures such as X-ray-derived interatomic distances. However, changes in σ for Na,H-ZSM-5 do indicate slight changes in structural features that affect the xenon NMR response. The data may be interpreted, for example, as a slight change in effective channel diameter. Using the A and Β parameters of Demarquay and Fraissard, the effective channel diameters for the samples of ZSM-5 from Equation 3 may be estimated. As seen in Table II, the effective channel diameter of the modified Na,H-ZSM-5 (as calculated from the Xe NMR parameter, σ ) is reduced by a few tenths of an Angstrom by deposition of coke even at the 1 wt.% level. If one models the ZSM-5 structure as an infinite cylinder, a uniform reduction of the channel diameter by 5% (from 7.3 to 7.0 A) is sufficient to account for a reduction of approximately 9-10% in volume. The channels available to xenon in coked H-ZSM-5 appear to be essentially of the same structure, no matter how heavy the coking, as indicted by the essentially constant value of σ for these samples (and the similarly constant value of D |). Thus, the local environment of xenon atoms is, at most, only 0

0

129

0

0

cnanne


21. GAFFNEY ET AL.


321

slightly changed by coking in this sample. This is in stark contrast to the 45% loss in uptake upon coking this sample, even at levels as low as 1 wt.%. Perhaps the most startling observation in these experiments is th change in th slope, σ of the curves for H-ZSM-5 with extent of coking. Ito et al.fi; observed similar changes in slope for coked H-Y zeolites. Generally, σ reflects the effects of two-body collisions between xenon atoms. The change in slope can be interpreted as a change in free volume upon coking. One may estimate effective relative internal volume accessible to xenon gas per gram of material from the slopesfijl ν

1


V/Vo =

uncoked / σ coked,

(4)

1

where V is the internal volume of the coked sample and Vo is the internal volume of the uncoked material. From the σ values in Table MB, one would conclude that H-ZSM-5 has lost about 40% of the internal volume per gram due to coking, even at 1 wt.% coke. The isotherms for xenon uptake in Na,H-ZSM-5 show a 910% reduction in capacity not detectable in the slopes of the chemical-shift profiles. See Table IIA. For H-ZSM-5, the expected loss of volume due only to channel filling by the coke, assuming a density of 1.8-2.1 g cm" for amorphous carbonft/ and an internal volume for ZSM-5 in the range of 0.10-0.19 cm g~* (10,11) may be estimated. For a 1 wt.% loading the coke would occupy 2.5-5.7% of the internal volume, significantly less than the 45% loss observed. The fact that the loss in xenon-accessible volume calculated either from NMR parameters of Figure 2 or from uptake measurements of Figure 1 is larger than expected by a simple model of channel filling by coke indicates that much of the volume not accessible to xenon is not filled with coke in H-ZSM-5. The internal Bronsted sites in protonated H-ZSM-5 zeolites have been proposed to be located at intersections between the straight and sinusoidal channel^ Other investigators indicate that, at low coking, the carbonaceous material resides primarily on the external surfaces of crystallites with each "coke molecule" covering one active s\\e(io). At higher coking levels, highly alkylated aromatic material formed at the channel intersections blocking poresfuj. The modified Na,H-ZSM-5 contains Bronsted sites only external to the channels. Carbonaceous residues will have formed primarily on the outer surfaces only and, perhaps near some channel mouths, as shown in Figure 3. During coking the channels remain effectively accessible to xenon. As indicated by the NMR results, with a slight effective reduction of the channel size, even at 12 wt.% coke. At this level of coking, a calculation similar to th one above would predict a blockage of 30-50% of th Na,H-ZSM-5 channels, if all the coke is formed inside the pores. Thus, most of the coke resides on the external surfaces of the crystallites, with only a minor amount (about 10%) blocking channels in the material. One model consistent with all these results is shown in Figure 3, in which a few isolated, internal sites of Na,H-ZSM-5 are coked, but with no substantial blockage of large internal volumes, leaving most of the internal volume accessible to xenon. Most of the coke must, therefore, reside on the external surfaces of the Na,H-ZSM-5. For H-ZSM-5, on the other hand, substantial internal volume is blocked from access by the xenon, even at 1 wt.% coking. The inaccessible volume is larger than the volume of coke deposited. Therefore, coke deposited in this material limits access of xenon to areas not 1

3

3



322


- τ

1

0.2

1

1

0.6

1

1.0

1

1

1.4

1

1

1.8

1

1

1

0.2

1

1

0.6

1

1

1.0

1

1.4

1

1 —

1.8

_ 1

Xenon Uptake (mmole

g )

129

Figure 2. (a) Xenon chemical shift as a function of xenon uptake for the modified Na,H-ZSM-5 zeolites with different coke contents: •--uncoked; A--1 wt % coke; B--12 wt % coke, (b) Xenon chemical shift as a function of xenon uptake for fully protonated H-ZSM-5 zeolites having different coke contents: •--uncoked; A--1 wt%coke; •~12wt%coke. (Reproduced with permission from ref. 16. Copyright 1991 Academic Press Inc.)

J

U

U

J _

. J

LI

U

J

_

Γ C.J L> L

J

• • • • η

Π (a)

r

Ί

· Π

nil rl

Γ

(b)

Figure 3. Schematic depiction of coke distributions in zeolite ZSM-5. (a) Lightly coked Na,H-ZSM-5; (b) Heavily coked Na,H-ZSM-5; (c) Lightly coked H-ZSM-5; (d) Heavily coked H-ZSM-5. (Reproduced with permission from ref. 16. Copyright 1991 Academic Press Inc.) In Selectivity in Catalysis; Davis, M., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1993.

21.

GAFFNEYETAL.

323


filled with coke. Thus, the coke deposits in H-ZSM-5 occur in sufficiently large volumes to block off xenon movement. These deposits could be either at the surface, at intersections, or in the channels. Comparisons of the xenon results for this material to that of Na,H-ZSM-5, where the sites acidified are at the surface, leads one to conclude that the material is not totally deposited at the surface.


Catalytic Activity

The catalytic activities of the modified Na,H-ZSM-5 and the fully protonated H-ZSM-5 were measured for several reactions. The constraint index experiment was carried according to the literature (u). A 1:1 molar mixture of nhexane and 3-methylpentane was fed over the catalyst bed at 300°C at a rate of 1 ml/hr with a helium diluent at 12.5 ml/min. The effluent stream was sampled after 20 minutes by syringe and analyzed by gas chromatography. The constraint index is given by Equation 5. Lo9io (Fraction of n-hexane remaining) Constraint Index =

(5) LoQio (Fraction of 3-methylpentane remaining)

The fully protonated H-ZSM-5 had a constraint index of 10.8 in agreement with the literature values. The modified Na, H-ZSM-5 had a constraint index of 1.1 which is typical of a larger pore zeolite with a pore diameter greater than 6Â. The modified zeolite reacts with the substrates primarily at external catalytic sites and in a non-discriminating manner. Size and shape selectivity are not major factors since few catalytic sites are located internally. Catalytic cracking of 2-butene over the modified Na, H-ZSM and the fully protonated H-ZSM-5 at 550°C and 16,000 hr GHSV was carried out. A dilute, fixed bed of 3 wt % zeolite in c

Selectivity in Catalysis - American Chemical Society

Recommend Documents