Diffusion of n-paraffins in offretite-erionite type zeolites

Diffusion of n-paraffins in offretite-erionite type zeolites

I N E M A N N Diffusion of n-paraffins in offretite-erionite type zeolites C~lio L. Cavalcante Jr., Mladen Ei~, Douglas M. Ruthven, and Mario L...

1MB Sizes 0 Downloads 3 Views

Recommend Documents

No documents
I N

E

M

A

N

N

Diffusion of n-paraffins in offretite-erionite type zeolites C~lio L. Cavalcante Jr., Mladen Ei~, Douglas M. Ruthven, and Mario L. Occelli*

Department of Chemical Engineering, University of New Brunswick, Fredericton, New Brunswick, Canada and the *Zeolites and Clays Research Program, Materials Science and Technology Laboratory, Geor~a Tech Research Institute, Atlanta, GA, USA Intracrystalline diffusivities are reported for a series of n-paraffins (C6 to C2o) in four different offretite-erionite intergrowths with constraint indices (CI) varying from 2.1 to 115 (a low value of CI corresponds to nearly pure offretite, whereas a high value corresponds to erionite). The intracrystalline diffusivities for each sample show a monotonic decrease with carbon number, except for one of the intermediate samples (HZSM-34), which has the highest proportion of extraframework AI. For the sorbates studied, the diffusivities decrease with the CI. Activation energies show different trends with carbon number, depending on the erionite/offretite character of the sample. The experimental evidence does not support the existence of a window effect, as suggested by Gorring (Gorring, R.L.J. Catal. 1973, 31, 13) on the basis of measurements with a zeolite of similar structure. Keywords: offretite; erionite; intergrowths; n-paraffins; intracrystallinediffusion; activation energy; window effect

INTRODUCTION The tailoring of catalyst or sorbent zeolites for specific applications has attracted the attention of many different research groups. The use of zeolites with micropores of molecular dimensions makes it possible to separate similar isomers or to enhance the selectivity of a given reaction toward a more desirable product. Among the more widely used zeolites are Y, ZSM-5, and A, which have effective pore diameters of about 8, 6.0, and 4 ]k, respectively. In the late 1960s and early 1970s, before the commercialization of ZSM-5 (which has a 10-membered ring pore opening), studies were performed on the potential use of zeolites of the offretite/erionite family for the shapeselective cracking of heavy n-alkanes into lighter products.~-a The experimental results from Miale et al. 1 showed that offretite could convert n-hexane into lighter cracked products, whereas 2-methylpentane did not react, even at higher temperatures (430°C), suggesting an intracrystalline diffusion controlled process. Experimental reaction runs with erionite 2 showed that the product distribution was richer in n-alkanes in the C~-C 6 and Cll-C12 ranges, with alkanes in the ranges of C1-C 2, C7-C 9, and over C]2 hardly present at all. Chen a reported that zeolite T (an intergrowth containing no more than 3% erionite in offretite) showed catalytic behavior intermediate Address reprint requests to Prof. Ruthven at the Department of Chemical Engineering, University of New Brunswick, P.O. Box 4400, Fredericton, NB E3B 5A3, Canada. Received 31 August 1993; revised 15 July 1994; accepted 8 December 1994 Zeolites 15:293-307, 1995 © Douglas M. Ruthven 1995

between that of pure erionite and pure offretite but somewhat closer to erionite, mainly with respect to the product distribution (rich in n-alkanes). An experimental study of the diffusion of several n-alkanes over zeolite T was reported by Gorring, 4 and these results have been quoted widely in the secondary literature. Surprisingly, instead of decreasing monotonically with carbon number, the diffusivities passed through a minimum at about C s and a maximum at Ct,), thus providing a neat explanation of the catalytic cracking data reported in the earlier studies. ]-3 An explanation for the observed behavior (the "window effect") was suggested in terms of the match between the sizes of the sorbate molecule and the erionite cage. The minimum diffusivity (-Ca) corresponds to the situation in which the diffusing molecule can just fit within a single erionite cage. As the length of the molecule increases beyond this point, part of the molecule will always be located within a window, leading to a decrease in the activation energy and a corresponding increase in diffusivity. More recently Derouane et al., 5 in analyzing the role of surface curvature when the dimensions of sorbate molecule and zeolite cage are similar, reported results that are in qualitative agreement with Gorring's data. Nitsche and Wei 6 presented a theoretical model based on Brownian motion in the zeolite pores, which is also capable of reproducing (qualitatively) the unusual trend in diffusivities reported by Gorring. However, examination of the experimental conditions used by Gorring reveals that large samples of adsorbent (several g) were used, and the uptake rates were measured over relatively large changes in sor0144-2449/95/$10.00 SSDI 0144-2449(94)00061-V

Diffusion of n-paraffins: C.L. Cavalcante Jr. et aL

bate concentration. In light of what is now known about the intrusion of extracrystalline mass transfer resistance and heat effects in such experiments, 7,s it seems probable that the reported diffusivities may reflect the effects of heat transfer and extracrystalline diffusion rather than diffusion within the zeolite crystals. The present study was undertaken to measure intracrystalline diffusivities for n-paraffins in several different zeolite intergrowths of the offretite/erionite family, using the zero-length column (ZLC) technique, which is much less sensitive to the intrusion of heat and extracrystalline mass transfer resistance. 91° Results were then compared with Gorring's data to check the validity of the proposed window effect. Ideally such measurements would be repeated with the same material (zeolite T), but this presents a significant difficulty since zeolite T is not a single welldefined crystallographic phase. We therefore decided to make the measurements with a series of distinct offretite/erionite intergrowths. In addition to a test of the trends originally reported by Gorring, this allows the effect of the introduction of periodic 8-ring blocks in the 12-ring offretite channel to be studied in greater detail.

ABBREVIATIONS The abbreviations used are: c, sorbate concentration in fluid phase; c0, sorbate concentration in feed; CI, constraint index, defined in R e f 18; D, intracrystal line diffusivity (cm2/s); Ea, diffusional energy of acti vation (kcal/mol); F,, purge volumetric flow rate (cin3/ s); K, dimensionless adsorption equilibrium constant (crystal volume basis); ~, half-width of crystal in slab diffusion model (cm); L, parameter defined in Equation (2) or (5); P, sorbate pressure (torr); Q, sorbate concentration in adsorbent (% g of sorbate/g of zeolite); R, crystal radius (cm); t, time (s); T, temperature (K); Vc, volume of the crystal (cm3); p,,, roots of Equation (3) or (6); ZLC, zero-length column; D6R, double 6-ring; QAC, quaternary ammonium cations; TMA, tetramethylammonium; BTMA, benzyl-N,N,N-triethylammonium; DABCO, 1,4-diazabicyclo(2.2.2)octane; CC, chlorine chloride; LT, long-time; ST, short time. •

.

F

linked by 8-membered ring apertures in such a way as to form parallel channels (micropores) of approximately 6.3 A in diameter which span the entire length of the crystal in the c direction• As a result of this arrangement, offretite adsorbs normal and isoparaffins as well as some cyclic molecules of up to 6.0 A in diameter (e.g., cyclohexane). In contrast, the framework of erionite contains discrete cages connected through 6- and 8-membered oxygen windows. These cages are alternately rotated by 60 ° at 15.2 A intervals. These (8-ring) windows, which circumscribe the micropores, have dimensions of about 3.6 x 5.2 A. As a result, erionite can sorb n-paraffins, but it completely excludes branched and cyclic isomers.

Intergrowth, stacking faults, and constraint index (CI) In the laboratory synthesis of offretite and erionite, intergrowths of the two zeolites causing stacking faults in the building structure, are commonly observed• 12-15 Occelli and co-workers, 11.16,17 by changing the organic cations used in the synthesis process, obtained several types of offretite/erionite crystals which showed different catalytic properties for the conversion of methanol to ethylene. They compared the different zeolite samples in terms of X-ray diffraction, electron diffraction, n.m.r., and i.r. characterization. Crystals with CI values (determined by passing an equimolar mixture of 3-methylpentane and n-hexane over the zeolite under cracking reaction conditions ~8) smaller than 8 were found to have X-ray diffraction patterns similar to natural offretite and were thus considered to be rich in offretite. Samples with CI values greater than 8 had X-ray patterns similar to natural erionite and were thus considered rich in erionite. In general, increasing CI values indicate increasing erionite character of the zeolite sample. To illustrate the difference in behavior between samples, Occelli et a1.16 showed that the selectivity for production of ethylene from methanol was higher for Table I

Diffusivity values: Comparison between long time and

short time results

Diffusivity (cm2/s) System (sample/sorbate

OFFRETITE AND ERIONITE STRUCTURES Individual structures Offretite and erionite are closely related but different zeolites. Their structures consist of columns of cancrinite cases joined together by double 6-ring (D6R) u n i t s . " T h e relationship between the two structures was only differentiated by Bennett and Gard 12 in 1967. They can be distinguished by singlecrystal X-ray or electron diffraction techniques but not easily by the standard X-ray powder diffraction techniques, r2 In offretite, the framework cages are connected through unrestricted 12-membered oxygen rings

294

Z e o l i t e s 1 5 : 2 9 3 - 3 0 7 , 1995

temperature, purge rate) HTMA-OFF/nC8 210°C, 60 ml/min HTMA-OFF/nC12 250°C, 30 ml/min H BTMA-OFF/nC8 200°C, 85 ml/min HBTMA-OFF/nC16 200°C, 85 ml/min HZSM-34/nC6 150°C, 83 ml/min HZSM-34/nC16 150°C, 100 ml/min HDABCO-I-OFF/nC6 180°C, 83 ml/min HDABCO-I-OFF/nC12 150°C, 67 ml/min

Long time

Short time

2.60 × 1 0

12

1.92 x l 0 -12

5.27 x l O

13

5.86 x l O -13

2.83 x l O - - 1 3

2.33 x l 0 -13

1.13 x l O -13

1.16 x l 0 -13

5.84 x l O

14

7.55 x l O -14

2.08 × 1 0

14

2.12 x l O

8.23 × 1 0

14

2.80 x l O -14

14

10.52 x l O -14 2.07 x l O -14

Diffusion of n-paraffins: C.L. Cavalcante Jr. et aL

samples rich in erionite (CI values between 8 and 115) than in offretite rich materials (CI values less than 8). Thus, offretite/erionite intergrowth can control the catalytic behavior of these types of zeolites.

ther details concerning the synthesis of these zeolites can be found elsewhere, a6

Diffusion measurements The ZLC methods, introduced by Eic and Ruthven 9 in the late 1980s, minimizes the intrusion of heat or extracrystalline mass transfer resistance by using very small amounts (1-2 mg) of zeolite with a relatively high purge flow. To provide a basis for comparison with the ZLC data and to investigate the effect of sorbate concentration on the apparent diffusivity, a limited n u m b e r of gravimetric sorption uptake measurements were also performed.

EXPERIMENTAL Sample preparation Zeolites of the offretite/erionite type were synthesized from hydrogels of composition A1203:12SIO2: 2 . 2 N a 2 0 : 0 . 6 4 K 2 0 : 2 . 5 ( Q A C ) 2 0 : 2 0 0 H 2 0 prepared from commercial grades of NaA102 (A1203 • Na20 • 3H20), KOH and NaOH. The quaternary ammonium cations (QAC) were obtained from tetramethy l a m m o n i u m (TMA) chlorides; benzyl-N,N,Nt r i e t h y l a m m o n i u m (BTMA) chloride, and 1,4diazabicyclo (2.2.2) octane (DABCO). ZSM-34 was synthesized using chlorine chloride (CC) with 1.5 Na2 O and 2.2 K20/mol of A1203 in the gel. 16 Heating at about 150-200°C for 7 days produced zeolites resembling intergrowths of offretite and erionite. Fur-

1.00011

°'°°~'"~:~:'q'~:

a

a

ZLC method Experimental details can be found in several references," 9,a~25 and no major modification was necessary for the present study. The zeolite sample is equilibrated at a given temperature with a flow of the sorbate at a known low concentration in an inert carrier gas. At time zero, the flow through the system is switched to the pure inert gas purge, and the decay of

r'l

30 ml/min

O

45 mllmin Full Solution

a (a)

0001

.

.

.

.

=

.

.

.

.

50

=

.

.

.

.

=

100

.

.

.

.

200

150

Figure 1 Consistency tests for ZLC results. (a) HTMA-OFF/nC12 at different flow rates (180°C, 30 and 45 ml/ min); (b) HDABCO-I-OFF/nC12 with different carrier gases (150°C, 67 ml/min).

T i m e (min)

1.000~ []

Nitrogen

X

Helium Full Solution

0.100o

0.010

0,001 -

(b)

-

-

-

-

i 50

. . . .

i 100

. . . .

i 150

. . . .

200

T i m e (min)

Zeolites 15:293-307, 1995 295

Diffusion of n-paraffins: C.L. Cavalcante Jr. et aL

Table 2 Crystal surface properties after NHg exchange and calcination in air at 500°C 11'16 Zeolite sample HTMA-OFF HBTMA-OFF HZSM-34 HDABCO-I-OFF

CI

BET area (m2/g)

Pore volume (ml/g)

Si:AI Chem.

AI(VI): AI(IV)

2.1 3.2 15.0 115.0

445 495 421 454

0.27 0.28 0.24 0.27

3.9 5.8 5.9 5.0

0.13 0.22 0.15

sorbate concentration with time is followed. The use of a flame ionization detector to follow the desorption curve makes it possible to work at very low concentrations, thus minimizing any nonlinearity effects. Typical sorbate concentrations in the gas phase during this work were in the range of 0.002% mol (0.010 tOrE). For analysis of the desorption curves, the long-time

analysis, as p r e s e n t e d p r e v i o u s l y by Eic a n d Ruthven, 9 was applied. The solution for the sorbate desorption along time (expressed in terms of the effluent concentration c), from a spherical particle of radius R, is given by Equations (1) to (3). For long times, only the first term of the series is significant, thus yielding a straight line plot of c/co versus time. The diffusivity (D) and equilibrium constant (K) can then be calculated from the slope and intercept. For spherical particles:

c

~ e x p ( - ~ D t / R ~) n =

where

L = 1Fp R2 3 Vc KD

(a)

(b)

(c)

(d)

Figure 2 SEMs: (a) HTMA-OFF; (b) HBTMA-OFF; (c) HZSM-34; (d) HDABCO-I-OFF.

296

Zeolites 15:293-307, 1995

(1)

l

(2)

Diffusion of n-paraffins: C.L. Cavalcante Jr. et aL

Table 3

Shapes and dimensions used in diffusivity calculations

Equivalent radius Zeolite sample

HTMA-OFF HBTMA-OFF HZSM-34 HDABCO-I-OFF

Particle shape

(l~m)

Diffusion model

Rectangular prism (length = 1.25 ixm, side = 0.63 ixm) Needle (length = 2.04 I~m, side = 0.42 i~m) Sphere (diameter = 0.74 ixm) Hexagonal prism (length = 1.60 ~m, side = 0.75 ixm)

0.36 0.28 0.37 0.44

Slab (~ = 0.625 ixm) Spherical Spherical Spherical

Again, D and K can be calculated from the slope and intercept of plot of In (C/Co)versus time. In a more recent work, H u f t o n and Ruthven 24 derived an expression that permits calculation of the diffusivity a n d equilibrium constant (in isotropic three-dimensional systems) from the initial region of the desorption curve. This concept has been extended to one-dimensional diffusion t h r o u g h a slab and can also be used to check the conformity of the experimental results to either spherical or slab diffusion models. 25 This approach was used to reevaluate the results for some of the more critical points of this study, and, in general, good a g r e e m e n t was f o u n d

and 6, is given by the roots of 13. cot13n + L - 1 = 0. (3) T h e c o r r e s p o n d i n g expressions for slab-shaped particles are as follows (see Ruthven et al.21).

exp (_132Dt/f2)

c Co

2L Z . : ] [13] + L(L + 1)]

=

(4)

K-D

(5)

[3, tanl3,, = L

(6)

led I

X

O

1E-12

O

T

2E 0

0~

Figure 3 Arrhenius plot showing temperature dependence of intracrystalline diffusivities of n-paraffins in

HTMA-OFF (pure offretite).

nCl2

1E-13

°n~C20

1E-14

nC16

I

I

I

I

I

I

1.8

1.9

2.0

2.1

2.2

2.3

2.4

lO00ff

Zeolites 1 5 : 2 9 3 - 3 0 7 , 1995

297

Diffusion of n-paraffins: C.L. Cavalcante Jr. et aL

between long-time (LT) and short-time (ST) approaches (see Table i). Consistency of the results obtained in these experiments was checked by varying the purge flow rate and the nature of the carrier gas. Results are shown in Figure i. Curves representing the full solution of the model (with calculated D and K) are also shown and may be seen to provide a good representation of the experimental data.

Gravimetric sorption uptake method The theory and experimental details for this frequently used method can be found in the literature. 7'8'2° Experimental runs were performed only for the system of pure offretite (HTMA-OFF)/noctane. The amount of zeolite used was roughly 17 mg, and concentration pressure steps up to 1.4 torr of pure n-octane were performed. The sample was regenerated previously at 400°C and 10 -4 torr for at least 12 h. In general, the diffusivity decreased with increasing partial pressure, and for sorbate partial pressures greater than 2 torr, the diffusion process was so slow that the equilibration time became impractically long, making it difficult to obtain reliable data. Because of the very slow diffusion, even with our fastest diffusing sample (pure offretite), we did

not attempt gravimetric measurements with the other samples.

RESULTS AND DISCUSSION Four different zeolite samples of the offretite/erionite family were studied. Two of these were considered rich in erionite (HDABCO-I-OFF and HZSM-34) and the other two rich in offretite (HTMA-OFF and HBTMA-OFF). Synthesis and physicochemical properties of this type of zeolites have been discussed in detail in the references given. 11.16,17 After decomposition of the organic cations, ion exchange with 1 M NH4NO 3 solution and calcination in air at 500°C/10 h, the four zeolites under study were submitted for analysis. Some of their properties are summarized in Table 2, with the zeolites listed in order of increasing CI value (erionite character). The two end members of this set will be referred to in this work as pure offretite and erionite, and the other two as intermediate samples, with different offretite or erionite character. The AI n.m.r, spectra of these zeolites exhibit intense asymmetric peaks at about 53 ppm associated with AI(IV) in the lattice, it The oxidative decomposition of the organic templates followed by NH4 ~

1E-I1

\

[]

150 C

x

200 C

A

250 C

. . . . . . . . . . .

i

IE-12

&

X

Figure 4 Variation of diffusivity with carbon number for n-paraffins in HTMA-OFF.

&

IE-13,

1E-14

!

6'

8'

I

1'o

Zeolites 15:293-307, 1995

I

22

CarbonNumber

298

I

24

Diffusion of n-paraffins: C.L. Cavalcante Jr. et aL

exchange does not destabilize the crystal lattice, and none of the NH 4 zeolites exhibits a peak near 0 ppm characteristic of extralatdce octahedral AI(VI). However, after decomposition of the ammonium ions in air, signals in the neighborhood of - 0 ppm are found in all samples but one (HTMA-OFF), suggesting dealumination with the formation of extralattice AI(VI). The greatest concentration of extraframework AI(VI) was observed in HZSM-34 (see Table 2). Scanning electron micrographs (SEM) for the four samples are shown in Figure 2. It can be seen that the crystals have different forms, from the rod-type HTMA-OFF to the needle shape of HBTMA-OFF. Table 3 summarizes the shapes and average dimensions for each sample. These dimensions were used in the numerical calculations. It may be observed that the average equivalent radii do not differ greatly between the samples, regardless of the differences in crystal size and shape. An important consideration for the evaluation of the diffusivities in the different samples is the diffusional path within each sample. This was brought into our analysis because diffusion through the offretite structure can occur only in one dimension, as discussed previously, whereas the diffusional path in erionite would probably be in all three directions. This difference in behavior should be reflected in the dif-

fusion model and therefore determines the equations and crystal dimensions to be used in calculating the diffusivities from the desorption curves. 21 In our calculations, we used the unidimensional slab model only for the pure offretite sample (HTMA-OFF), considering diffusion to occur only in the c direction, taken as the long axis of the cylinder. So, the diffusion dimension for the HTMA-OFF sample is taken as the half-width of the slab (0.625 p,m). For the other three samples, we assumed that the presence of even small amounts of erionite intergrowth eliminates the unidimensional character of the diffusion path and causes the whole crystal to behave approximately as an isotropic three-dimensional diffusion system. The spherical diffusion model with the equivalent radii shown in Table 3 was therefore used for these samples.

Pure offretite (HTMA-OFF) This sample, which is the only one that did not contain extraframework A1, has a CI of 2.1, and is considered to be as close as possible to pure offretite. The slab model was used to calculate the diffusivities, based on the one-dimension diffusion path of the offretite structure. Experimental runs were made with n-heptane, n-octane, n-dodecane, and n-hexadecane as sorbates.

20

15

¢)

f

10.

Figure 5 Variation of activation energies with carbon number for n-paraffins in HTMA-OFF.

== &

8

10

12

14

16

18

20

22

24

Carbon Number

Zeolites 15:293-307, 1995

299

Diffusion of n-paraffins: C.L. Cavalcante Jr. et aL

sorbate partial pressures because of the long time required for equilibration. A decreasing trend of diffusivity with increasing sorbate concentration is observed; extrapolation to zero adsorbed phase concentration yields a diffusion coefficient that agrees reasonably well with the value estimated from the ZLC results (interpolated in Figure 3). This pattern of variation of diffusivity with loading is somewhat unusual since, in most of the systems studied previously in this laboratory, we have observed the opposite trend (diffusivity increasing with loading). However, the zeolites s t u d i e d p r e v i o u s l y all had threedimensional pore networks. Diffusion in the onedimensional channel system of offretite is expected to be much more sensitive to the blocking effect of the sorbate, so such a difference seems plausible. It is nevertheless difficult to be certain that this trend is real since at higher loadings there was evidence of slow coke formation. The sample was regenerated by

The ZLC diffusivity results for these systems are shown in Figure 3, as a function of temperature. It should be noted that, for clarity of the resulting graph, when more than one experiment was performed at given conditions, only the average value is shown. However, the scatter of the data is indicated. It can be seen that for the series of n-alkanes the data show the normal monotonic decrease of diffusivity with increasing carbon number, as has been observed previously with zeolites A, NaX, and silicalite (see Figure 1 i). This is seen more clearly in Figure 4. The activation energies (Figure 5) also show a monotonic increase with carbon number, as is commonly observed for other zeolites. The gravimetric isotherm and apparent diffusivities for n-octane at 220°C are shown in Figure 6. Diffusivity values were calculated using both the initial part of the uptake curve (ST) and the final part (LT). It was not possible to perform experiments at higher

10

(a)

P (Tort)

Figure 6 Gravimetric data for HTMAOFF/nCB at 220°C. (a) isotherm; (b) diffusivities versus sorbate concentration in the adsorbent.

IE-11

ZLC (from Fig. 3)

/

[]

ST

X

LT

It It

[] IE-]2 •

(b) 1E-13 11

IE-14

i

i

i

i

2

4

6

8

Q (%g/g)

300

Zeolites 15:293-307, 1995

10

Diffusion of n-paraffins: C.L. Cavalcante Jr. et aL

exposure to oxygen at 480°C, and the points obtained after this procedure (loadings of 4.0, 5.2, and 8.1%) are quite consistent with the original data (at 2.5 and 4.2% loading).

Therefore, it is to be expected that tbr all molecules with carbon chains equal or greater than 15 ,~, the diffusivities and activation energies should be similar. n-Octane is the n-paraffin with chain length closest to the critical dimension, so, according to this argument, all sorbates with carbon n u m b e r greater than 8 should have approximately the same diffusivity. For n-hexane, as expected, diffusivities were higher, since its chain length is less than 15 ~. The results published previously for n-pentane in erionite 26 confirm that the normal decreasing monotonic trend for diffusivities versus carbon number is reestablished for carbon numbers less than 8.

Pure erionite (HDABCO-I-OFF) This sample has a CI value of 115 and is thus considered to be as close as possible to pure erionite. The isotropic spherical diffusion model was used to calculate the diffusivities, based on the three-dimensional diffusion path within the erionite structure. Experiments were p e r f o r m e d with n-hexane, n-octane, n-dodecane, and n-hexadecane as sorbates. Figure 7 shows the Arrhenius plot for the calculated intracrystalline diffusivities (D). Results from an earlier uptake study for natural erionite with n-pentane as the sorbate 26 are also shown. It can be seen that the diffusivities and activation energies are essentially the same for n-octane, n-dodecane, and n-hexadecane. Only n-hexane shows different (higher) diffusivities. These results can be related to the dimension of the cage along the diffusional path (15.1 ~), since the molecules in erionite must diffuse in a zigzag pattern.

Intermediate samples (HBTMA-OFF and HZSM-34) Diffusivities for the two intermediate samples fall, as expected, between the values for the two pure sampies, and they reflect some of the characteristic features of the main components of each sample. HBTMA-OFF, with a C1 value of 3.2,16 is considered to be an offretite, with a minor intergrowth of erionite. The diffusion results for n-hexane, n-octane,

IE-11

(Ruthvenand Derrah26)

IE-12

o o

1E-13

~

Figure 7 Arrhenius plot showing temperature dependence of intracrys-

nC6

talline diffusivities of n-paraffins in HDABCO-I-OFF (pure erionite).

IE-I,I

1.7

i

i

1.8

1.9

"

I

2.0

-

I

2.1

-

X

nC8

0

nC12

r"l

nC16

I

2.2

I

2.3

I

2.4

"

I

I

2,5

2.6

-

f

2.7

I

2.8

i

2.9

3.0

1000ff

Zeolites 15:293-307, 1995

301

Diffusion of n-paraffins: C.L. Cavalcante Jr. et al.

n-dodecane, and n-hexadecane, plotted in Figure 8, indeed show the same decrease of diffusivity with increasing carbon number, as observed with the pure offretite sample (HTMA-OFF). It may be noted, however, that the activation energies derived from Figure 8 are essentially constant, contrary to what was observed for the pure offretite sample and similar to the behavior for the pure erionite sample. It may also be observed that the range of the diffusivities for the different n-paraffins is smaller than for the pure offretite because of the presence of erionite domains within the crystal lattice. The data for the HZSM-34 sample show some surprising features. The CI value of 1516 suggests a strong erionite character with some offretite intergrowth. The diffusivity data for n-hexane, n-octane, n-dodecane, and n-hexadecane in Figure 9 show some similarities to the pure offretite as well as to the pure erionite behavior. For example, diffusivities for n-hexane and n-octane are higher than for n-dodecane and n-hexadecane (offretite-like behavior) but are essentially the same within each group, which is erionite-like behavior. Most surprising is the observation that the activation energies are in general higher

for all sorbates in HZSM-34 than in the other three samples (see Figure lOb). Thus the diffusivity for a certain sorbate in HZSM-34 will become higher than in the pure offretite sample (HTMA-OFF) at temperatures above 400°C. On the other hand, at temperatures below 150°C, the diffusivity for HZSM-34 would be lower than the diffusivity of the same sorbate in the pure erionite sample (HDABCO-I-OFF). The explanation for this anomaly is unclear, but it may be related to the presence of extraframework aluminum that was present in greater proportion in this sample. The nature and location of AI are not known, although

/o\ A10 l+, AI(OH) 2+, [AI(OH)2] 1+, [A1

All 2+

\o / and amorphous aluminum residue formation has been proposed in HY-type zeolites. 27-29 This AI(VI) may interact with C6-C16 paraffins, thus decreasing the mobility and obstructing the diffusion path of the sorbed molecules. The high activation energies for C 0, C s, and Cl2 in HZSM-34 are consistent with such

1E-12 X

/ X , ~

nC6

Figure 8 Arrhenius plot showing temperature dependence of intracrystalline diffusivities of n-paraffins in HBTMA-OFF.

IE-13

nC16 ~

X

0

1E-14 1.7

I

I

I

1.8

1.9

2.0

|

2.1 looo~

302

Zeolites 15:293-307, 1995

I

i

2.2

2.3

24 ,

Diffusion of n-paraffins: C.L. Cavalcante Jr. et aL

IE.11

1E-12

&

E

X

nCl2

c~

1E-13

nC6

nCl6 ~ nCl6

Figure 9 Arrhenius plot showing temperature dependence of intracrystalline diffusivities of n-paraffins in

nC8

HZSM-34.

IE-14

,.7

118

,19

21o

211

212

213

2.4

1000/T

an explanation, but the activation energy for C16 is similar to that in the other sorbents.

Comparison among samples T o establish easier comparisons among the four different offretite/erionite samples (CI values varying from 2.1 to 115), Figure 10 shows plots of some of our previous results against the CI values of each zeolite sample. As expected intuitively, the various paraffins show a decreasing trend of diffusivity in the sequence from offretite to erionite content (increasing CI values). Activation energies are similar for all samples, with the exception of HZSM-34 which, as noted previously, shows higher values.

THE WINDOW EFFECT The gravimetric uptake method was used by Gorring 4 to determine the diffusion rate of n-paraffins in zeolite T (minor erionite intergrowth in offretite). Rather large samples ( - 5 g) of zeolite were used. The sorption capacity was found to be 7-8% g/g (hydrocarbon/zeolite). Diffusion coefficients were calculated

assuming constant radial diffusion in a cylinder (the crystals were cylindrical, with the average length 2.9 txm and diameter 0.65 ixm). In general, diffusivities were found to be independent of concentration, up to 80% of the equilibrium saturation. Above this loading, Gorring reported that the diffusivity declined with loading, which is consistent with our own data for octane in H T M A offretite (Figure 6). Figure 11 shows explicitly the most striking result from Gorring's data: a break in the decreasing trend of diffusivity with carbon numbers between C 8 and Ca2. The curve for 300°C shows a minimum at C s and a maximum at C12. Similar experimental studies for other zeolites, such as 5A, 19 NaX, and silicalite, 2° are shown in Figure 11. They do not show this minimum/ maximum behavior but, instead, show the intuitively expected monotonic decrease of intracrystalline diffusivity with carbon number. Gorring's results for the variation of activation energy (for zeolite T) with carbon number are plotted in Figure 12, along with other comparable experimental observations for zeolites, A, NaX, and silicalite. Again, although the activation energies for the other zeolites tend to increase monotonically with carbon

Zeolites 15:293-307, 1995 303

Diffusion of n-paraffins: C.L. Cavalcante Jr. et al. IE-I1 []

X IE-12

O

nC6

X

nC8

A

nCl2

0

nC16

[]

A 1E-13

~

[] X

(a)

[]

0 o

n 0

IE-14 . . . . . . . .

i0

. . . . . . . .

! 100

1000

C.I.

Figure 10 Summary of results plotted in terms of CI values: (a) diffusivities at 200°C; (b) activation energies.

20 nC6

[]

nC8

X

nCI2

15.

nC16

0

o 10.

(b) A & X

. . . . . . . .

1

0

. . . . . . . .

I

100

1000

C.I. number, the values for zeolite T reported by Gorring decrease initially from C~ to C 5, then show a maximum in the C7-C9 region, and decrease again after C10, falling essentially to zero at C13. Gorring related his findings to the specific dimensions of the erionite lattice structure and the sorbate molecules. He suggested that the window effect could be a general phenomenon, common to diffusion in many zeolites, with the maxima/minima positions identifiable by crystal parameters. This effect has not been observed in other zeolites in the years following Gorring's report. The quantity of zeolite used in Gorring's study may very well have led to extracrystalline diffusional resistance, and there seems to have been no attempt to check for such an effect by varying the sample quantity and configuration. In Figure 13, diffusivity results from the present study for the different offretite/erionite samples, extrapolated to 300°C, are presented. There is no clear indication of any window effect. At the lowest temperature of our measurements, 150°C, the results for

304

Zeolites 15:293-307, 1995

HZSM-34 suggest a slight minimum in diffusivity at carbon number 12 (n-dodecane), as shown in Figure 14. These results may be affected by the lower precision of our measurements at lower temperatures, but even if this minimum is real, it would not agree with the window effect. Predictions according to the Gorring theory from lattice and channel dimensions sug.gest that the minimum should fall at nC s with a max1mum at nC~2. The energies of activation, for our four samples, are plotted against carbon number in Figure 15. The behavior expected from the theoretical structural explanations of the window effect would be a decreasing trend with carbon number, with a local maximum between C7 and C 9 and a drastic drop to zero after C12. Again, our results do not show this behavior. Our pure offretite sample showed the usual monotonic increase of activation energy with carbon number as observed for other zeolites. The other three samples showed essentially constant activation energies, except for a somewhat lower value for n-hexa-

Diffusion of n-paraffins: C.L. Cavalcante Jr. et aL 1E-05

1E-06.

°

1 E-O7.

~

5

o

c

(19)

1E-08,

1 E-O9.

Figure 11 Diffusion coefficients for zeolite T (from Gorring4) and for other zeolites.

5A @ 300°C (19) IE-IO

1E-II ,

IE-12,

1E-13

IE-14

I

I

I

I 18

Carbon Number

o

~

ZEOLITET (4)

o

?

o

f

x

5A (19)

o o

-'-3

Figure lZ Activation energies for zeolite T (from Gorring4) and for other zeolites (1 kcal = 4.18 kJ).

~

F o

W

o

O

.

D NoX (19)

~:~

/ Nox(19) [

2

L

l

4

i

[

6

R

I

8

I

i

10

I

I

1

12

14

J

I

16

I

l

18

I

]

20

I

22

CARBON NUMBER

Zeolites 15:293-307, 1995 305

Diffusion of n-paraffins: C.L. Cavalcante Jr. et aL IE-10

Ruthven and Derrah (26)

J

0

H-TMA-OFF

O

HBTMA-OFF

X

HZSM -34

[]

H-DABCO-I-OFF

IE-I1

o

Figure 13 Diffusion coefficients ver-

x

sus carbon number at 300°C for all samples (filled square was extrapolated from data for nC5 in natural erionite by Ruthven and DerrahZ6).

x

o

IE-12,

x IE-13

I

I

I

I

I

I

I

I

6

8

10

12

14

16

18

20

22

Carbon Number IE-11

1E-12

250oc

1E-13.

1 0o0

Figure 14 Diffusivities of n-paraffins in HZSM-34 in terms of carbon number.

IE-14 -

1

I

I

|

I

I

8

10

12

14

16

Carbon Number

306

Z e o l i t e s 15:293-307, 1995

18

Diffusion of n-paraffins: C.L. Cavalcante Jr. et al.

0

ite, as expected from their different degrees of offretite/erionite intergrowths. Our results do not show any clear evidence of a window effect as reported by Gorring for a zeolite of the same family. 4 The diffusivity values show the usual monotonic decrease with carbon number, with no indication of sudden breaks in this trend, although one of the intermediate samples did show some anomalies, probably because of the higher proportion of extraframework AI present.

HZSM-54

0

H-TMA-OFF

x

HBTMA-OFF H-DABCO-I-OFF

o

E

ACKNOWLEDGMENTS Financial support from CAPES (Brasil), NSERC (Canada), and N A T O (Collaborative Grant CRG 921274) is gratefully acknowledged.

REFERENCES

o

I i t I±~ 6

8

I I I i I 15~__2

10

12

14

16

18

i 20

22

CARBON NUMBER Figure 15 Activation energies versus carbon number for all samples.

decane in HZSM-34 (but still substantially higher than zero). It must be noted that this outlying point for the activation energy of nC]6 is still quite close to the pure erionite value. CONCLUSIONS A series of experiments with four different zeolite samples of the offretite/erionite family using the ZLC method yielded consistent diffusivity data. For one of the systems studied the ZLC diffusivity values were confirmed by gravimetric measurements. The behavior of the pure zeolite samples (offretite and erionite) was as expected. The pure offretite behaved as a unidimensional structure, with large pores available for the diffusion of the n-paraffins, and showed trends that were similar to those observed previously for zeolites 5A, NaX, and silicalite. The pure erionite yielded results that are consistent with what is known about the dimensions of the channels and framework structure. The behavior of the mixed intergrowth samples was consistent with their mixed structures, showing some patterns closer to offretite and other patterns closer to erionite character. T h e behavior of the mixed erionite rich sample (HZSM-34) showed some minor anomalies in the variation of activation energy with carbon number which have not yet been fully explained. These may be related to extraframework A1, which was present in higher proportion for this sample. In general, diffusivity values decrease with increasing erionite character of the sample, following the sequence offretite > H B T M A > HZSM-34 > erion-

1 Miale, J.N., Chen, N.Y. and Weisz, P.B.d. Catal, 1966, 6, 278 2 Chen, N.Y., Lucki, S.J, and Mower, E.B.J. CataL 1969, 13, 329 3 Chen, N.Y. in Proceedings of the 5th International Congress on Catalysis (Ed. J.W. Hightower) Elsevier, Amsterdam, 1973, Vol. 2, Paper 97 4 Gorring, R.L.J. CataL 1973, 31, 13 5 Derouane, E.G., Andre, J.-M. and Lucas, A.A. J, Catal. 1988, 110, 58 6 Nitsche, J.M. and Wei, J. AlChEJ. 1991, 37, 661 7 Ruthven, D.M. and Lee, L. K. AIChEJ. 1981, 27, 654 8 Ruthven, D.M. Principles ofAdsorption andAdsorption Processes, Wiley, New York, 1984, 189 9 Eic, M. and Ruthven, D.M. Zeolites 1988, 8, 40 10 Voogd, P., van Bekkum, H., Shavit, D. and Kouwenhoven, H.W.J. Chem. Soc. Faraday Trans. 1991, 87, 3575 11 Occelli, M.L., Ritz, G.P., lyer, P,S., Walker, R.D. and Gerstein, B.C. Zeolites 1989, 9, 104 12 Bennett, J.M. and Gard, J.A. Nature 1967, 214, 1005 13 Chen, N.Y., Schlenker, J.L., Garwood, W.E. and Kakotailo, G.T.J. Catal. 1984, 86, 24 14 Kerr, I.S., Gard, J.A., Barter, R.M. and Galabova, I. Am. Miner. 1970, 55, 441 15 Aiello, R., Barrer, R.M., Davies, J.A. and Ker, I.S. Trans. Faraday Soc. 1970, 571, 1610 16 Occelli, ML., Innes, R.A., Pollack, S.S. and Sanders, J.V. Zeolites 1987, 7, 265 17 Sanders, J.V., Occelli, M.L., Innes, R.I. and Pollack, S.S. in Proceedings of the 7th International Zeolite Conference, Tokyo, 1986, p. 429 18 Frilette, V.J., Haag, W.O. and Lago, R.M.J. CataL 1981, 67, 218 19 Eic, M. and Ruthven, D.M Zeolites 1988, 8, 472 20 Karger, J. and Ruthven, D.M. Diffusion in Zeolites and Other Microporous Solids, Wiley, New York, 1992, pp. 439 and 487 21 Ruthven, D.M., Eic, M. and Richard, E. Zeolites 1991, 11,647 22 Eic, M. and Ruthven, D.M. in Zeolites: Facts, Figures, Future (Eds. P.A. Jacobs and R.A. van Santen) Elsevier, Amsterdam, 1989, p. 897 23 Voogd, P., van Bekkum, H., Shavit, D. and Kouwenhoven, H.W.J. Chem. Soc. Faraday Trans. 1991, 8, 5575 24 HuRon, J.R. and Ruthven, D.M., Ind. Eng, Chem. Res. 1993, 32, 2379 25 Cavalcante, C.L., Jr. PhD Thesis, University of New Brunswick, 1993 26 Ruthven, D.M. and Derrah, R.I. in NaturalZeolites: Occurrence, Properties, Use (Eds. L.B. Sand and F.A. Mumpton) Pergamon Press, Oxford, 1978, p. 403 27 Lunsford, J.H. in Fluid Catalytic Cracking II, ACS Symp. Ser. 452, (Ed. M.L. Occelli) Washington, DC, 1990, p. 1 28 Anderson, M. and Klinowski, J. Zeolites 1986, 6, 455 29 Feng, C.C., Hall, J.B., Huggins, B.J. and Beyerlein, R.A.J. CataL 1993, 140, 395

Zeolites 15:293-307, 1995

307