Multicomponent Separation Performance of MFI-type Zeolite Membranes

2007:088

M A S T E R ' S T H E S I S

Multicomponent Separation Performance of MFI-type Zeolite Membranes

Irina Trusheva

Luleå University of Technology D Master thesis

Chemical Technology

Department of Chemical Engineering and Geosciences

Division of Chemical Technology

Luleå University of Technology

Department of Chemical Engineering and Geosciences Division of Chemical Technology

Multicomponent separation performance of MFI-type zeolite membranes

Irina Trusheva

Luleå, June 2007

ABSTRACT

Zeolite membranes are of great interest due to potential ability to separate many industri- ally important molecules with high selectivity. As well as high selectivity, high flux and durabil- ity are essential for practical application of zeolite membranes.

Application of zeolite membranes in hydrogen production can be important for the future of hydrogen as a fuel source. Using hydrogen in fuel cells as well as in internal combustion en- gines gives possibility to eventually break the link between transport sector and CO 2 emissions and improve energy efficiency. The production of hydrogen by steam reforming could be per- formed more efficiently by the use of a membrane reactor with continuous removal of CO 2 . Thus the development of membranes which can separate CO 2 from synthesis gas is very important.

This thesis work has been devoted to investigation of separation performances of MFI- type zeolite membranes for the mixtures related to the steam reforming process. Silicalite-1 and ZSM-5 membranes were tested for separation of the following mixtures: CO 2 /H 2 , H 2 O/H 2 , and CO 2 /H 2 O/H 2 . Prior to the separation, membranes have been characterized by n-hexane po- rosimetry in order to evaluate membrane quality. The presence of defects in the pore structure can significantly affect separation performances of membranes.

The separation experiments with a binary mixture of CO 2 and H 2 showed that the mem- branes of both types were slightly hydrogen selective. The highest CO 2 /H 2 separation factors were observed at 22 ⁰ C and were 0.7 for silicalite-1 and 0.8 for ZSM-5. The hydrogen selectiv- ity could be explained by significant Knudsen diffusion through the support as well as very weak CO 2 adsorption.

The H 2 O/H 2 separation showed the highest selectivity towards water at 22 ⁰ C. The separa- tion factors were 2.2 and 4.0, for silicalite-1 and ZSM-5 respectively. Strong adsorption of water in zeolite pores significantly limits permeation of hydrogen and thus provides effective separa- tion. ZSM-5 membrane has a higher H 2 O/H 2 separation factor than silicalite-1. This was ex- plained by more pronounced affinity of water to ZSM-5 membrane.

When the feed was a ternary mixture of CO 2 , H 2 and H 2 O, the CO 2 /H 2 separation factors at 22 ⁰ C were 2.2 and 3.7 for silicalite-1 membranes S1 and S2 and 4.2 for ZSM-5 membrane.

The H 2 O/H 2 separation factors at 22 ⁰ C were 2.1 and 4.6 for silicalite-1 membranes S1 and S2

and 4.1 for ZSM-5 membrane. It was seen that at low temperature the CO 2 /H 2 separation factors

in a ternary mixture are higher than that observed in a binary mixture of CO 2 and H 2 . Increased

CO 2 /H 2 separation factors were obtained due to significant decrease in hydrogen permeance be-

cause of blocking effect of water. However, the CO 2 permeance was not decreased significantly

by the adsorbed water, due to its ability to adsorb in the zeolite pores. H 2 O/H 2 separation factors are not affected by presence of CO 2 .

Results show that for all three mixtures membranes of both types were selective at low temperature, and the selectivity decreased dramatically when the temperature was increased. At temperatures above 100 ⁰ C all membranes were selective towards hydrogen.

Additionally, separation of ethanol and hydrogen was investigated using both types of membranes. Results show that membranes were very selective towards ethanol at low tempera- ture. The highest separation factors were observed at around 21 ⁰ C, and were 19.1 for ZSM-5 and 16.6 for silicalite-1. Adsorption of ethanol in the zeolite pores drastically decreases hydrogen permeance at low temperature. This may be a promising start to adapt zeolite membranes for ap- plication in ethanol production or separation processes.

Separation of hexane isomers such as n-hexane and 2,2-dimethyl-butane was performed using silicalite-1 membrane. The highest selectivity towards n-hexane was achieved at 230 ⁰ C and was 130.

KEYWORDS: silicalite-1, ZSM-5, membrane, separation, diffusion, selectivity.

ACKNOWLEDGEMENTS

First of all, I would like to thank Professor Jonas Hedlund and professor Sergey Tretya- kov for giving me the opportunity to do my Master thesis at the division of Chemical technology.

I extend my deepest gratitude to my supervisor, Lic. Eng. Jonas Lindmark for his guid- ance, help and great patience during this work.

I also thank Charlotte Andersson, Ivan Carabante and the rest of the people at the divi- sion of Chemical technology for helping me in my work.

Special thanks to Elizaveta Potapova for your undying enthusiasm and friendly advices.

The Swedish Institute is sincerely acknowledged for financial support.

Finally, I wish to thank my family, Nicolay Sviazov and all my friends for your encour-

agement and support.

CONTENTS

1 INTRODUCTION ………. 5

1.1 Zeolites ………. 5

1.1.1 Structure ……… 5

1.1.2 Properties of zeolites ………. 6

1.1.3 Application ……… 7

1.2 Zeolite membranes ………... 8

1.2.1 Structure and properties……… 8

1.2.2 Separation by zeolite membranes ………. 9

1.3 Diffusion……….. ……… 11

1.4 Separation equipment ………... ………... 13

1.5 Membrane preparation ………. 15

1.6 Objective of this work ………. 15

2 EXPERIMENTAL ………. 16

2.1 n-Hexane porosimetry……….. 16

2.2 Mixture separation measurements ………... 18

3 RESULTS AND DISCUSSION ……… 20

3.1 n-Hexane porosimetry …..………... 20

3.2 Mixture separation measurements ……….. 21

3.2.1 CO

/H

separation ……… 21

3.2.2 H

O/H

separation ……… 23

3.2.3 CO

/H

/H

O separation ……… 24

3.2.4 C

H

OH /H

separation ……… 27

3.2.5 n-Hexane/2,2-DMB separation ………. 28

4 CONCLUSIONS ………... 30

REFERENCES ………. 31

APPENDIX A.1 Calibration data……… ……….. 33

APPENDIX A.2 Mixture separation data……… 34

1 INTRODUCTION

1.1 Zeolites

1.1.1 Structure

Zeolites are highly crystalline aluminosilicates with a very regular microporous structure.

The three-dimensional framework is made up of SiO 4 4- and AlO 4 5- tetrahedra, each of which con- tains a silicon or aluminum atom in the center [1]. The tetrahedra are joined together in various well-defined arrangements through shared oxygen atoms. In this way, an open crystal lattice with pores of molecular dimensions is formed [2].

Depending on zeolite structure, the pore size varies from 0.3 to 1.3 nm [3]. Access to the pores is limited by openings consisting of a ring of oxygen atoms of joint tetrahedra. The rings can contain 4, 5, 6, 8, 10 and 12 oxygen atoms [4].

Alumina is trivalent, and AlO 4 5- in the framework thus gives a net negative charge.

Charge balancing cations are required to maintain electrical neutrality. It can be an alkaline, al- kaline earth or rare metal ion. The zeolite framework can also contain water molecules which are obtained from synthesis solution.

A general formula for chemical composition of zeolite can be expressed as:

M x/n [(AlO 2 ) x (SiO 2 ) y ] ⋅ ω H 2 O, where M – charge balancing cation;

n - the cation valence;

x/y – a silicon/aluminum ratio for zeolite.

ω - number of water molecules.

The bracketed term presents crystallographic unit cell.

Over 172 different zeolite frameworks are known today. MFI-type structure is well stud- ied and is the most widely used in practice. Zeolites of this structure are of great interest due to the suitable pore size, similar to the dimension of many industrially important molecules, and the high thermal and chemical stability. Silicalite-1 and ZSM-5 are well known crystalline solids of MFI-type.

MFI- type zeolites have a two-dimensional pore structure composed of the zig-zag chan-

nels running in y-direction, which are cross-linked by the straight channels running in x-direction

(Figure 1.1 a). The straight elliptical channels have a 0.53×0.56 nm cross-section and zig-zag channels have 0.51×0.55 nm cross-section (Figure1.1 b) [5]. Both are defined by 10-member oxygen rings.

a) b)

Y

X Z

Figure 1.1: a) schematic of MFI-type structure; b) cross-section of the straight and zig- zag channels [5].

The framework of silicalite-1 is essentially the same as for ZSM-5. The difference be- tween them is that ZSM-5 contains both silica and alumina tetrahedra but silicalite-1 is free of aluminum content. Consequently, by definition silicalite-1 is not a true zeolite, but since its pores size lies in the range of 0.3-2 nm it can be referred to as a microporous material. The Si/Al ratio ranges typically between 10 and 100 for ZSM-5, and it must be larger than 100 for silicalite-1 [1]. Aluminum charge introduces active sites available for exchangeable cations. This is why ZSM-5 can exist in different forms, such as the H ⁺ , Na ⁺ , Ba ²⁺ and the Cu ²⁺ form etc.

1.1.2 Properties of zeolites

The key parameters which determine the properties of a zeolite are structure, Si/Al ratio, framework cations and deposited compounds, particularly their size, location and coordination [3].

The well-defined pore structure makes zeolites size-and shape selective and allows to act as molecular sieves. There are three main types of selectivity based on limited pore size of zeo- lites: reactant and product selectivity assumes only molecules of appropriate size and shape can enter and leave the zeolite pores, respectively. In the case of transition state selectivity, only a certain reaction pathway may be possible within the pores [6].

Small zeolite pores also provide a large specific surface area, which is essential in adsorp-

tion and catalytic reactions.

The framework composition determines the hydrophilicity and acidity of the pore struc- ture. High aluminum content gives the zeolite a hydrophilic nature and high affinity for water and other polar molecules. On the contrary, materials with low aluminum content, such as pure silica, have a hydrophobic nature.

As described above, due to the aluminum content in the framework, zeolites possess ion- exchange capacity. Changing the cations provides a widely used means of modifying zeolite properties. For instance, substitution of the exchangeable cation for a hydrogen ion gives the acid form of the zeolite [6].

Exchangeable cations have different size and bond energy and thus can change the effec- tive pore size, shape and the nature of adsorption in the pores [7].

1.1.3 Application

The advantageous properties of zeolites, such as molecular adsorption, catalytic activity and ion-exchange capacity, explain the widespread use of zeolites in industry [8].

The molecular adsorption properties have opened up a range of applications in drying, purification and separation processes. The acidic properties, selectivity based on uniform pore structure, and a high specific surface area make zeolites extremely useful as a selective catalyst material for industrial processes. The ion-exchange ability of the zeolites renders application in water softening, detergent production and waste and sewage treatment. Cations within pores are loosely-bound to the zeolite framework and can be readily exchanged with other cations when in aqueous solution.

The crystalline zeolite structure offers the opportunity to prepare thin, highly selective membranes. Due to a well-defined pore structure of molecular size and preferential adsorption properties, membranes are suited for separation processes. Zeolite membranes are also attractive for the use in membrane reactors, because in addition to catalytic activity and high selectivity, these materials can operate under severe conditions. In this way, in the membrane reactor both reaction and separation can occur simultaneously in a continuous way.

Zeolite membranes are of great interest for application in hydrogen production and puri-

fication processes. Hydrogen is mainly produced from fossil fuels by reforming processes, such

as natural gas reforming and coal gasification, with subsequent Water-gas shift reaction. Typi-

cally, the Water-gas shift reaction is used to remove CO from synthesis gas after reforming proc-

ess: CO(g) + H 2 O(g) ↔ CO 2 (g) + H 2 (g). Selective removal of CO 2 during reaction is thermody-

namically favorable. Application of a CO 2 selective membrane in reactor may possibly increase

the CO conversion and, consequently, purity of hydrogen, and also increase the yield of hydro-

gen. Most likely, many other industrial reactions can be carried out more efficiently by means of membrane reactor.

An innovative application of zeolite membranes is in sensor detection systems. At pre- sent, chemical sensors are being intensively developed to detect and monitor emissions from cars, power plants and other combustion processes [9].

1.2 Zeolite membranes

1.2.1 Structure and properties

Zeolite membranes are microporous inorganic materials, capable of separating compo- nents in a gas and liquid mixtures with high selectivity. Inorganic membranes are generally me- chanically strong, thermally and chemically resistant. These materials can withstand chlorine, organic solvents, and other harsh chemicals better than organic membranes.

An important feature of any membrane is a high selectivity, combined with a high flux [8]. Flux means the throughput, and selectivity means ability of the membrane to separate com- ponents. A membrane with a high selectivity but with a low flux is not attractive for practical application because the membrane area must be very large to handle a certain stream. As well as good separation performances, reproducibility is another essential factor to put membranes in large scale application.

Zeolite membranes can be manufactured both as self-supported zeolite film and a thin zeolite film on a porous support, such as α-alumina, γ-alumina, glass, stainless steel. α-Alumina is widely used as support material for membranes due to its inert properties and low tendency to leach aluminum into synthesis solution [10].

A self-supported membrane combines the separating zeolite layer and the bulk support as a uniform structure. Since the flux through the membrane is inversely proportional to the mem- brane thickness, the membrane should be as thin as possible to obtain high flux. However, a thin stand-alone membrane lacks mechanical strength [10].

The practical approach to the contradiction is supported membranes, which possess a higher mechanical strength as compared with self-supported membranes. The pores of the sup- port must be sufficiently small in order to obtain a continuous zeolite layer. If the pores at the support/zeolite layer interface are sufficiently large, a thick zeolite film is required to close the pores of the support properly [4]. This leads to a decrease of the flux through the zeolite layer.

On the other hand, small pores of the support introduce high mass transport resistance within the

bulk support [11].

In order to avoid flux restriction both from the zeolite layer and support, an asymmetrical support is used. An asymmetrical support consists of two or more layers with different pore size.

This structure is called a graded support. Thin top layer has fine pore size and thick bottom layer has a coarse pore size [10]. The thin layer serves as intermediate between zeolite film and the thick support and allows to obtain thin zeolite layer without limitation of mass transport through the support.

There are several kinds of defects in zeolite membranes, such as cracks, open grain boundaries and nonclosed film [12]. Defect free zeolite films are essential to obtain highly selec- tive membranes and effective separation. Really, it is not possible to eliminate every defect and non-zeolite pore, because the zeolite films are polycrystalline [13]. Nevertheless, it is important to minimize the defect concentration because flux through the defects can reduce separation per- formances of the membranes.

Furthermore, another important property to be taken into account is that the difference in thermal expansion factor between the support and the zeolite layer should be as small as possible to minimize the risk of crack formation while heating. Self-supported membranes do not suffer from this problem [4].

Membranes can undergo modification in order to obtain the selectivity required for a par- ticular separation [3]. Impregnation and ion exchange are used to change properties of zeolite membranes [6]. Impregnation of the zeolite pores by calcium compounds allows for improve- ment of the CO 2 adsorption and could make a membrane more selective for CO 2 . As mentioned above, membranes can be prepared in different forms by means of ion-exchange. By adjusting a Si/Al ratio it is possible to change polarity of the membrane and thus the selectivity with respect to either polar or nonpolar molecules.

In this work, separation performances of MFI-type membranes, such as silicate–1 and NaZSM – 5, were studied.

1.2 Separation by the zeolite membranes

High permeance and high selectivity of zeolite membranes provide effective separation performance. Separation is based on transport of mixture components through the membrane.

The driving force for flow through the membrane can be created by difference in total pressure, partial pressure and/or concentration [12]. The mixture is fed to the membrane and separated into two streams: permeate and retentate. Permeate is the portion of the feed that passes through the membrane and the retentate is the portion of the feed rejected by the membrane.

Separation factor for a binary mixture of compounds i and j with the molar fraction x is

defined as:

α i,j =

j feed i

permeate j

i

x x x x

⎟ ⎟

⎠

⎞

⎜ ⎜

⎝

⎛

⎟ ⎟

⎠

⎞

⎜ ⎜

⎝

⎛

(1.1)

This expression describes ability of the membrane to separate two components in a mix- ture under certain conditions. Zeolite membranes can also separate multi-component mixtures;

however, it is difficult to predict the behavior of membranes in separation of this kind.

Separation in the zeolite membrane can occur by three different mechanisms: molecular sieving, preferential adsorption and diffusion [4, 15].

The molecular sieving mechanism assumes that only molecules of an appropriate size can permeate through the membrane. If the molecular diameter is larger than the zeolite pore size, the molecules can not enter the pores, and, сonsequently are rejected by it. When molecular siev- ing is dominant, the requirement of low defect concentration is especially strict [16]. Separation by molecular sieving is mainly observed at higher temperatures.

If both components are smaller than the pore size, molecular sieving will not occur. Dif- ference in interaction of permeating molecules with the membrane surface can also provide a separation effect. Preferential adsorption means that presence of stronger adsorbing components suppresses permeation of weaker adsorbing components. Molecules with high adsorption strength can be transported through the membrane more effectively. Adsorption is a temperature dependent process; therefore high selectivity is often achieved at relatively low temperature.

If one of the components in a mixture is a condensable vapor, capillary condensation can occur at a relatively low temperature. Condensed gases diffuse through the pores and block dif- fusion of the other components. As a result of capillary condensation, especially in small pores, the pores can be completely closed by condensed gas. If the other gases do not dissolve in a con- densable component, high separation factors can be achieved [17].

The third mechanism is based on difference in diffusivity of molecules. Molecules with higher diffusivity pass through the membrane faster and thus are separated from slow molecules.

Thus, difference in molecular size, diffusivity and adsorption strength of mixture compo- nents are key factors in mixture separation. Permeation properties and separation performances of the zeolite membranes are affected by many factors, some of these are presented below [1, 18, 19]:

- physical and chemical properties of the diffusing components;

- structure of the zeolite layer (pore size distribution, defects, thickness);

- characteristics of the zeolite in the membrane (type, crystal size, shape, orientation);

- properties of the porous support;

- operation conditions: temperature and pressure; character of driving force: pressure gra- dient or sweep gas usage; mixture composition.

1.3 Diffusion

Mixture separation by zeolite membranes is based on transport of components through the membrane [1,17]. Due to the polycrystalline structure of a zeolite membrane, molecules can pass both through zeolite pores and gaps between crystals [13].

Mass transport through the zeolite membrane follows an adsorption-diffusion model and can de divided into the following steps [7,20]:

1) Diffusion of components from the bulk feed to external surface of zeolite layer;

2) Adsorption to external surface;

3) Mass transport from the external surface into the zeolite pores;

4) Diffusion along the internal surface of pores;

5) Mass transport out of the pores to zeolite layer/support interface;

6) Desorption;

7) Diffusion through the support pores into the bulk permeate.

The adsorption step depends on conditions such as temperature, pressure and affinity of components to the material of the zeolite layer. The strongest adsorbing molecules can hinder the adsorption of weakly adsorbing molecules, and thus competitive adsorption occurs, predomi- nantly at low temperature. At higher temperatures adsorption decreases.

Mass transport from the external surface into zeolite pores can be limited by a molecular sieving effect. Smaller molecules more readily diffuse into the pores than larger molecules which have difficulty to enter the zeolite pores. [20].

Generally, diffusion in porous media can occur as surface diffusion, Knudsen diffusion and molecular diffusion, depending on the pore size, kinetic diameter of molecules and the op- eration conditions. Kinetic diameter, temperature and pressure are factors determining the mean free path of the molecules. The mean free path is defined as the average distance traveled by a molecule between two collisions. The mean free path increases with decreasing kinetic diameter and pressure, and increasing temperature.

Molecular or bulk diffusion is dominant when the pore diameter is large relative to the

mean free path of the molecules. In this case, collisions between neighboring molecules take

place frequently. Molecular diffusion mainly occurs at high pressure. As pore size is lowered, the

mean free path is becoming larger than the pore diameter and the number of collisions with the

pore wall increases. This is termed Knudsen diffusion. In Knudsen transport, the mobility of the components is kinetically determined [7].

The Knudsen diffusivity is given by [7]:

D

=

M kT d 2

3 (1.2)

where d is pore diameter, M is molecular mass, T is temperature, and k is Boltzmann’s constant. Thus, Knudsen diffusivity increases with decreasing molecular weight of the gas and increasing temperature. Knudsen diffusion is prevalent at low pressure [1].

When separation occurs by the Knudsen mechanism, preferential diffusion of the lighter gas molecules through the membrane is observed [8, 21].

Separation factor in terms of Knudsen diffusion can be derived in the following way:

As described earlier, separation factor for a binary mixture of compounds i and j with the molar fraction x is expressed by formula (1.1):

α i,j =

j feed i

permeate j

i

x x x x

⎟ ⎟

⎠

⎞

⎜ ⎜

⎝

⎛

⎟ ⎟

⎠

⎞

⎜ ⎜

⎝

⎛

(1.3)

A partial pressure of components can be expressed as x

=

tot i

P

P and then separation factor

can be written in the terms of partial pressure of compounds in feed and in the permeate Pi and Pj:

α i,j =

tot feed j tot

i permeate tot

j tot

i

P P P

P P

P P

P

⎟⎟

⎟ ⎟

⎟

⎠

⎞

⎜⎜

⎜ ⎜

⎜

⎝

⎛

⎟⎟

⎟ ⎟

⎟

⎠

⎞

⎜⎜

⎜ ⎜

⎜

⎝

⎛

=

j feed i

permeate j

i

P P P P

⎟ ⎟

⎠

⎞

⎜ ⎜

⎝

⎛

⎟ ⎟

⎠

⎞

⎜ ⎜

⎝

⎛

(1.4)

For an ideal gas we can write P i =

v RT F

(1.5)

J i = A F

(1.6)

P _i = v RTA J

(1.7) Substituting expression (1.7) into (1.4) gives:

α i,j =

J feed i

permeate j

i

P P vRTA J

vRTA J

⎟⎟ ⎠

⎜⎜ ⎞

⎝

⎛

⎟ ⎟

⎠

⎞

⎜ ⎜

⎝

⎛

=

j feed i

permeate j

i

P P J J

⎟ ⎟

⎠

⎞

⎜ ⎜

⎝

⎛

⎟ ⎟

⎠

⎞

⎜ ⎜

⎝

⎛

(1.8)

The Knudsen flux can be written as:

J

= - D

,

P

,

K (1.9) where D

,

is the Knudsen diffusivity;

Then separation factor becomes:

α i,j =

j feed i

feed j j kn

feed i i kn

P P

K P D

K P D

⎟ ⎟

⎠

⎞

⎜ ⎜

⎝

⎛

⎟ ⎟

⎠

⎞

⎜ ⎜

⎝

⎛

, ,

, ,

(1.10)

Rearranging and inserting the expression for Knudsen diffusivity (1.2) gives us

α i,j =

j kn

i kn

D D

, ,

=

j i

M kT d

M kT d

2 3

2

3 =

i j

M

M (1.11)

Selectivity in the Knudsen diffusion is inversely proportional to the square root of the molecular weight of the components. When the difference in molecular weight between compo- nents is small, separation is not efficient by Knudsen diffusion.

At very small pore size, the interaction of diffusing molecules with pore surface becomes stronger and surface diffusion is dominant. Surface diffusion assumes physically adsorbed mole- cules diffuse along the surface of the pores by jumping from site to site, driven by the chemical potential gradient within the pores.

As detailed in previous section, zeolite has a microporous structure; therefore diffusion

in zeolite pores is attributed to both surface diffusion and Knudsen diffusion.

The concentration gradient and temperature both determine the diffusion rate. Mass trans- fer can be described by Fick’s law:

J A = - D AB

dz dC

, (1.12) where J A is the molar flux of component A relative to molar average velocity,

dz

dC

is the concentration gradient, D AB is the diffusion coefficient for component A diffusing through com- ponent B [22].

At low temperature, the adsorption is high, and permeation of components can occur due to a concentration gradient within the pores. As temperature increases, adsorption decreases and the amount of adsorbed species begins to decrease. Adsorbed molecules progress in the pores as long as there is sufficient surface coverage and concentration gradient. As soon as the amount of adsorbed species in the pores components is not significant anymore, surface diffusion declines [8]. Thus, at higher temperature, transport along the pores is due to activated Knudsen diffusion.

Diffusion through the support, in particular a graded support, can be described in the terms of Knudsen and Poiseuille diffusion. As described earlier, a graded support can be com- posed of two layers S1 and S2. Thin top layer S1 has small pores (100 nm) and thick bottom layer has larger pores (3 μm). In layer S1 Knudsen diffusion mainly occurs since mean free path of diffusing gases is similar to the pores of layer S1. In layer S2 Poiseuille diffusion is becoming dominant because pores of layer S2 are much larger than the mean free path for all gases.

The transport in non-zeolite pores and defects may occur by Knudsen diffusion or viscous flow and provides additional flux through the membrane [1]. Viscous flow commonly occurs in large pores and defects at high total pressure gradient through the membrane and does not result in separation [21].

Multicomponent diffusion through the pores can result in reducing permeances of species through the membrane as compared to single gas permeances. Single gas permeances for He, CO 2 , and H 2 for silicalite-1 and ZSM-5 membranes are given in the Table 1 [3].

Table 1 - Single gas permeances [10 ^-7 mol/(s m ² Pa)] [3].

Membrane He CO 2 H 2

silicalite-1 83 134 216

NaZSM-5 58 126 157

1.4 Membrane preparation

MFI-type membranes were prepared by the seeded growth method. Graded porous α- alumina discs with a diameter 25 mm were used as supports. The surface of the supports were masked and seeded with colloidal zeolite crystals. Seeded supports were immersed into an ap- propriate synthesis solution and a film was grown by hydrothermal synthesis. After synthesis, the membranes were rinsed and calcined at 500 ⁰ C. The calcined membranes were kept in a dessica- tor. The membranes have a 500 nm thick zeolite film.

1.5 Objective of this work The objective of this work was:

- To investigate the separation performances of MFI-type zeolite membranes for mixtures related to the steam reforming process;

- To see if it is reasonable to use zeolite membranes for ethanol separation.

2 EXPERIMENTAL

2.1 n-Hexane Porosimetry

All membranes used in the separation measurements were characterized using n-hexane porosimetry. n-Hexane porosimetry is a technique used to investigate the quality of the mem- branes and determine non-zeolite pore size distribution [23].

The method is based on the phenomenon of capillary condensation of a vapor, present as a component of a gas mixture, in pores and uses absolute pressure gradient as driving force for diffusion through the membrane [10]. The feed to membrane contains a noncondensable carrier gas and some condensable compound. Helium is used as noncondensable carrier gas, and some hydrocarbons, such as p-xylene or n-hexane, are used as condensable compound.

Consistent increase of partial pressure of n-hexane in gas mixture causes blocking of pores of a certain size with condensed vapor. The hexane reduces permeation of helium through the membrane. The principle of the method is that the permeance of helium is measured as func- tion of the partial pressure of n-hexane in order to estimate pore size distribution within the zeo- lite film. According to the Kelvin equation and Horwath-Kawazoe equation, each relative pres- sure of n-hexane can be related to a pore/defect size as shown in Table 2 [23]. A principal scheme of the porosimetry unit is shown in Figure 2.1 [24].

Table 2 – Relative pressure of n-hexane and the corresponding diameter of pores [24].

P/P0 0.01 0.025 0.25 0.85 0.99

D, nm 1.08 1.27 2.65 9.18 100

Figure 2.1 – Porosimetry unit

The membranes were mounted in a stainless steel cell and heated at a rate of 1 ⁰ C/min to 300 ⁰ C and kept at this temperature for 6 hours in a flow of pure helium to remove traces of wa- ter and adsorbed species in the zeolite pores. Graphite gaskets with the inner diameter of 17 and 19 mm were used for sealing. After drying, the membranes were cooled to room temperature.

Two mass flow controllers were used to adjust partial pressure of n-hexane in the feed to the membrane. A condenser, connected on the permeate side of the membrane, removed most of the n-hexane from the permeate stream. The flow of helium was measured using a flowmeter (ADM 1000, J&W Scientific).

The permeate side was kept at atmospheric pressure and pressure difference over the membrane was 1 bar. The measurements were carried out at room temperature. The helium flow through the membrane and the pressure were recorded at different values of the partial pressure of n-hexane: 0, 0.01, 0.025, 0.25, 0.85, 0.99. At each value the system was allowed to equili- brate.

Initially, only pure helium (P/P 0 = 0) permeates through the membrane. It provides a high

helium permeance. Then, the pure helium stream is mixed with a flow of helium saturated with

n-hexane, and the partial pressure of n-hexane in a gas stream is successively increased. At low

activity (P/Po = 0.01) n-hexane adsorbs in the smallest pores and helium can enter only the larger

pores, because activity of n-hexane is still not enough to close the larger pores. At higher activity

(P/Po = 0.25) it continues to close the larger pores for helium flow. Thus, if the membrane con-

tains very little defects, significant helium permeance drop is observed because all zeolite pores

are completely filled with condensed vapour. If helium can still permeate, though to small extent,

at increased partial pressure of n-hexane, this means that defects and non-zeolite pores are pre- sent in the membrane [24].

2.2 Mixture separation measurements

A sketch of the experimental set up used in mixture separation measurements is presented in Figure 2.2 [24]

MFC

Gas chromatograph

Cell

Sweep gas

Furnace

Thermocouple Retentate

Permeate Saturator

Feed MFC

MFC

Figure 2.2 – Experimental set up for separation measurements

The set up includes following main units: stainless steel cell contained in furnace; mass- flow controllers; saturators; a gas chromatograph equipped with detectors.

The membrane is mounted in the stainless steel cell, which is contained in a furnace, and the temperature is controlled electronically. The temperature at the membrane cell is monitored by means of a thermocouple.

Mass-flow controllers are used to adjust the feed to the system. Two mass-flow control- lers are used for preparing the feed and one for the sweep gas.

Saturators are used to obtain liquid components in a vapour phase. Thermostat bath is used for controlling the temperature of the saturator.

Separation of components occurs in cell. Mechanism of separation is presented in Wicke-

Kallenbach set up (Figure 2.3) [24].

Retentate Feed Sweep

gas

Permeate

Membrane

Figure 2.3 – a Wicke-Kallenbach experimental set up

In this experimental set up, a partial pressure gradient is used as a driving force for diffu- sion through the membrane. The mixture was introduced to the feed side of the membrane, while helium was used as a sweep gas at the permeate side. Sweep gas serves for flushing permeating species from the permeate side of the membrane in order to maintain the partial pressure gradient through the membrane [14].

Analysis of permeate and retentate streams is based on a gas chromatography concept.

The on-line gas chromatograph consists of the following main components: carrier gas supply and flow controller, sample injection valve, the chromatographic column and column oven, the detector system [25], see Figure 2.4.

Figure 2.4 – Schematic of the on-line gas chromatograph [25]

The principle of a gas chromatography is that components are injected into a column con-

taining stationary phase, transported by carrier gas and detected as series of peaks when compo-

nents leave the column. The carrier gas must be chemically inert, for instance He. The stationary

phase is used to selectively inhibit motion of components and making each one reach the detec-

tor at different time [25]. Two types of columns, packed and capillary are commonly used in gas chromatography.

The rate at which the components transport through the column depends on the adsorp- tion strength, which in turn, depends on the physical and chemical properties of components and the stationary phase material. Moreover, the temperature of the column is a factor determining the rate of progression along column, because of molecular adsorption of components on the sta- tionary phase is temperature dependent. The higher the column temperature, the faster compo- nents move through the column. However, the faster components move through the column, the less components interact with the stationary phase, and the less effective separation is achieved.

Thus, the column temperature is selected to find a compromise between the length of the analy- sis and the level of separation.

In order to keep a desired temperature, the chromatographic column is contained in an oven.

The retention time is the time which it takes after a sample injection for the components to reach the detector [25]. The detector is located at the end of column and used to monitor the outlet stream from the column. Thus, the time at which each component reaches the end of col- umn and the amount of that component can be determined. The detector response is a series of peaks along the time line. Generally, components are identified by the order in which they elute from the column and by the retention time.

The number of peaks correlates with the number of components in a sample. For each component there is corresponding peak. The size (area) of peak is proportional to the amount of the component in a feed stream.

A number of detectors are used in a gas chromatography. The most common are the flame ionization detector (FID) and the thermal conductivity detector (TCD).

The TCD mechanism is based on the capability to detect components whose thermal con- ductivity is substantially different from that of the carrier gas. Helium is the most common used carrier gas. The FID is an ion detector which uses an air-hydrogen flame to produce carbon ions and electrons. As components elute from the column they pass through the flame and are burned, producing ions. The ions produce an electric current, which is the signal output of the detector.

The greater the concentration of the component, the more ions are produced, and the greater the current. The FID is essentially selective toward combustible carbon containing components.

In this work separations of following mixtures were studied:

- Carbon dioxide/hydrogen (CO 2 /H 2 );

- Water /hydrogen (H 2 O/H 2 );

- Carbon dioxide/ water/hydrogen (CO 2 / H 2 O /H 2 );

- Ethanol/ hydrogen (C 2 H 5 OH /H 2 );

- n-hexane/ 2,2-DMB (C 6 H 14 /2,2-dimethyl-butane).

The membranes were mounted in a stainless steel cell, sealed with graphite gaskets and dried at 300 ⁰ C overnight under a flow of pure helium. The rate of heating was 1 ⁰ C/min.

The pressure was atmospheric on both side of the membrane. Before starting the experi- ment the membrane should be equilibrated with the feed at room temperature. Measurements were taken over the temperatures range of 25–400 ⁰ C in order to follow how the temperature af- fects the membrane selectivity.

In separation measurements with a mixture of CO 2 /H 2, the feed to the membrane con- sisted of 50 kPa carbon dioxide and 50 kPa hydrogen with the total volumetric flowrate of 1000 ml/min. The flowrate of sweep gas was 1400 ml/min.

For the system H 2 O/H 2 , 500 ml/min of hydrogen and 500 ml/min of helium were fed to water saturator kept at 20 ⁰ C by a thermostat bath. Thus, the feed to the membrane was 49 kPa hydrogen, 49 kPa helium and 2 kPa water. The total volumetric flowrate was 1000 ml/min. The flowrate of sweep gas was 1400 ml/min.

For the system CO 2 /H 2 /H 2 O , 500 ml/min of hydrogen and 500 ml/min of carbon dioxide were fed to water saturator. The feed to the membrane was 49 kPa carbon dioxide, 49 kPa hy- drogen and 2 kPa water, the total flowrate was 1000 ml/min. The flowrate of sweep gas was 1400 ml/min.

For the system C 2 H 5 OH/H 2, 500 ml/min of hydrogen and 500 ml/min of helium were fed to saturator filled with ethanol at 20 ⁰ C. Thus, the feed to the membrane was 47 kPa hydrogen, 47 kPa helium and 6 kPa ethanol. The total volumetric flowrate was 1000 ml/min. The flowrate of sweep gas was 1400 ml/min.

Separation of n-hexane and 2.2-DMB was also carried out in order to make sure that the

experimental unit works properly. Separations of hexane isomers have been studied before and

the behavior of the components is well known. Helium was fed to two saturators, containing 2.2-

DMB and n-hexane at 20 ⁰ C. The feed to the membrane was 11 kPa n-hexane and 11 kPa 2.2-

DMB with a helium balance to 100 kPa. The flowrate of sweep gas was 200 ml/min.

3 RESULTS AND DISCUSSION

3.1 Porosimetry

In order to evaluate the membrane quality, each membrane was first tested in n-hexane porosimetry experiment. The permeance of helium through the membrane was measured at dif- ferent values of partial pressure of n-hexane. A significant drop in helium permeance was ob- served in the range of partial pressure between 0 and 0.025. This means that even at low activity, n-hexane blocks zeolite pores for helium flow. The rest of helium flow through the membrane is due to all larger pores and defects.

Figure 3.1 shows n-hexane porosimetry data for silicalite-1 and ZSM-5 membranes. The porosimetry pattern for a silicalite-1 membrane of known high quality from earlier work by Hed- lund et al. [26] has been added for comparison. For the silicalite-1 membrane, the helium per- meance drop is 98.55 % (S1) and 99.9 % (S2). The corresponding value for ZSM-5 (Z1) is 98.8

%. It is clear that S2 membrane exhibits the highest quality; only 0.1 % of the helium permeance is through the non-zeolite pores. S1 and Z1 membranes are of slightly lower quality than the ref- erence membrane (R) and can be expected to exhibit slightly poorer separation performance than S2 membrane due to a larger amount of defects.

1,E-09 1,E-08 1,E-07 1,E-06 1,E-05

0,0 0,2 0,4 0,6 0,8 1,0

Relative pressure

H e p e rm ea nc e [m o l/ m 2 s Pa]

S1 S2 Z1 R

Figure 3.1 – n-Hexane porosimetry data for silicalite-1 and ZSM -5 membranes

3. 2 Mixture separation measurements

3.2.1 CO

/H

separation

Silicalite-1 and ZSM-5 were tested in CO 2 /H 2 separation experiment. Molecular sieving is not expected to play any role in the system, because of close kinetic diameter of both mole- cules (0.29 and 0.33 nm, respectively for H 2 and CO 2 ), and also both of them are considerably smaller than the pore size of MFI-type membranes (0.55 mm). Thus, it can be supposed that CO 2

and H 2 can be separated due to preferential adsorption of CO 2 at low temperature. Figure 3.2 shows the CO 2 /H 2 separation factors for silicalite-1 and ZSM-5 as a function of temperature in the range 22-400 ⁰ C. It was seen that the membranes of both types were actually slightly hydro- gen selective. The highest CO 2 /H 2 separation factors are observed at 22 ⁰ C and are 0.7 for sili- calite-1 and 0.8 for ZSM-5. The hydrogen selectivity could be explained by Knudsen diffusion through the support which is becoming even more pronounced at the higher temperatures, and, consequently, separation factors approach 0.3.

The CO 2 /H 2 separation factors are almost the same for membranes of both types, a slightly higher separation factor for ZSM-5 membrane probably can be explained by better qual- ity of ZSM-5 membrane as compared to silicalite-1 membrane used in the experiment.

Figure 3.3 shows the permeance of CO 2 and H 2 in the temperature range 22-400 ⁰ C. The trend for CO 2 permeance can be explained in terms of surface diffusion. At low temperature there is a slight increase in CO 2 flow through the membrane due to surface diffusion of adsorbed molecules. At 35 ⁰ C permeance reaches maximum and then starts to decrease because of de- creasing amount of adsorbed molecules at the higher temperature.

The permeance of H 2 mainly occurs due to Knudsen diffusion. CO 2 is a weakly adsorbing

component, so even at low temperature the adsorbed CO 2 has only a minor effect on the H 2 per-

meance. Hydrogen permeance increases consistently with temperature due to both increased

Knudsen diffusion and decreased CO 2 adsorption.

α CO

/H

0,0 0,2 0,4 0,6 0,8 1,0

0 100 200 300 400

Temperature,

C

S e p a ra ti o n f a ct o r

S1 Z

Figure 3.2 – Binary CO 2 / H 2 selectivity for silicalite-1 and ZSM-5.

P H

0 10 20 30 40 50 60

0 100 200 300 400

Temperature,

C

P e rm ea nc e [ /1 0

mo l/ (s m

Pa) ] ^S1 _Z1

a)

P CO

0 5 10 15 20 25

0 100 200 300 400

Temperature,

C P er m ea nc e [ /1 0

mo l/ (s m

Pa )]

S1 Z1

b)

Figure 3.3 – Gas permeances for the mixture CO 2 / H 2 as a function of temperature for

silicalite-1 and ZSM-5: a) H 2 permeance; b) CO 2 permeance.

3.1.3 H

O/H

separation

H 2 O is a polar molecule and strong adsorption of water in the zeolite pores, especially at low temperature, is expected to limit permeation of hydrogen, and in this way provide effective separation. Separation of H 2 O and H 2 at low temperature is thus explained mainly by the signifi- cant difference in adsorption strength of the two components. Figure 3.4 shows the H 2 O/H 2 sepa- ration factor for silicalite-1 and ZSM-5 as a function of temperature in the range 22-400 ⁰ C. At 22 ⁰ C, separation factors are 2.2 and 4.0, for silicalite-1 and ZSM-5 respectively. At 25 ⁰ C, membranes are still selective towards water, the separation factors are 1.0 and 1.2; and then, at even higher temperature, transition to selectivity to hydrogen is observed. At 350 ⁰ C, the separa- tion factors are 0.2 for the silicalite-1 and 0.05 for ZSM-5. An explanation for this behavior could be that adsorption effect is decreasing with increasing temperature, resulting in a domi- nance of Knudsen diffusion at higher temperatures. There is still no clear explanation for separa- tion factor of 0.05. This is probably achieved due to influence of a graded alumina support within which Knudsen diffusion in combination with Poiseuille diffusion occurs.

The higher separation factor for ZSM-5 is explained by the hydrophilic nature of ZSM-5 membrane which provides strong affinity of water to the membrane material.

Figure 3.5 shows the permeance of H 2 O and H 2 in the temperature range 22-400 ⁰ C. The permeance of water is high at 22 ⁰ C and then decreases quickly in the temperature range from 25 to 93 ⁰ C. Hydrogen permeance is extremely low at 22 ⁰ C, because the pores are blocked by wa- ter, and increases rapidly as water leaves the pores in the range 30-100 ⁰ C. At temperature above 100 ⁰ C, permeance of both components is almost constant.

α H

O/H

0 1 2 3 4 5

0 100 200 300 400

Temperature,

C

Sep ar a ti on f a c tor

S1 Z

Figure 3.4 – Binary H 2 O/ H 2 selectivity for silicalite-1 and ZSM-5.

P H

0 10 20 30 40 50

0 100 200 300 400

Temperature,

C

Perm eanc e [ /1 0

mo l/ (s m

Pa) ] ^S1

Z1

a)

P H

O

0 5 10 15 20 25

0 100 200 300 400

Temperature,

C P er m e anc e [ /1 0

mo l/ (s m

Pa )]

S1 Z1

b)

Figure 3.5 – Gas permeances in the mixture H 2 O/ H 2 as a function of temperature for sili- calite-1 and ZSM-5: a) H 2 permeance; b) H 2 O permeance.

3.1.4 CO

/H

/H

O separation

Multi-component separation experiments with a feed containing CO 2 , H 2 and H 2 O were also performed. It is quite difficult to predict the separation performance for the ternary mixture.

However, it is possible to make a guess about the behavior of a membrane from data obtained for

binary separation, such as CO 2 /H 2 and H 2 O/H 2 separation. CO 2 /H 2 /H 2 O separation at low tem-

perature is mainly based on competitive adsorption of components which differ in adsorption

strength. As discussed in previous section, in a binary mixture of CO 2 and H 2 , CO 2 adsorption

does not suppress the permeance of hydrogen significantly even at low temperature. In presence

of water which is a polar molecule, much less pores are available for hydrogen and thus hydro-

gen permeance is becoming lower at low temperatures.

Figure 3.6 shows CO 2 /H 2 and H 2 O/H 2 separation factors for silicalite-1 and ZSM-5 mem- branes as a function of temperature in the range 22-400 ⁰ C. The CO 2 /H 2 separation factors at 22

0 C are 2.2 and 3.7 for silicalite-1 membranes S1 and S2 and 4.2 for ZSM-5 membrane. The H 2 O/H 2 separation factors at 22 ⁰ C are 2.1 and 4.6 for silicalite-1 membranes S1 and S2 and 4.1 for ZSM-5 membrane. All membranes change selectivity towards hydrogen when temperature approaches 40 ⁰ C.

At low temperature, the CO 2 /H 2 separation factors in a ternary mixture are higher than that observed in a binary mixture of CO 2 and H 2 . Increased CO 2 /H 2 separation factors were achieved due to significant decrease in hydrogen permeance because of blocking effect of water, whereas CO 2 permeance was only slightly decreased by the adsorbed water due to competitive adsorption of CO 2 and H 2 O. Effect of water adsorption was stronger for ZSM-5 membrane which was more selective towards both H 2 O and CO 2 than silicalite-1 membranes.

The S2 membrane has a higher separation performance than the S1 membrane which was expected due to significant difference in quality of the membranes.

Comparing results obtained in separation of a binary mixture of H 2 O/H 2 using silicalite-1 S1 and ZSM-5 with that in a ternary mixture of CO 2 / H 2 O/H 2 we can conclude that the H 2 O/H 2

separation factors are not affected by presence of CO 2 due to adsorption strength of water is hig- her than that of CO 2.

Figure 3.7 shows the permeance of H 2 , CO 2 and H 2 O in the temperature range 22-400 ⁰ C.

The permeance of hydrogen is very low at low temperature and increases significantly at high

temperature. It is quite explainable, that hydrogen flow through the membrane is lower as com-

pare with that in CO 2 /H 2 separation experiment until the amount of adsorbed water starts to de-

crease. The hydrogen permeance at low temperature is around 6.7 · 10 ^-7 mol/(s m ² Pa) for sili-

calite-1 and 2.6 · 10 ^-7 mol/(s m ² Pa) for ZSM-5 membranes. This is much lower than in a binary

mixture where the hydrogen permeance at low temperature is around 23 · 10 ^-7 mol/(s m ² Pa) for

silicalite-1 and 27 · 10 ^-7 mol/(s m ² Pa) for ZSM-5 membranes due to strong adsorption of water

in zeolite pores. The hydrogen permeance at high temperature is approximately 64 · 10 ^-7 mol/(s

m ² Pa) for silicalite-1 and 55 · 10 ^-7 mol/(s m ² Pa) for ZSM-5 membranes. This is about factor 3.5

lower than observed in the single gas measurements at 25 ⁰ C, which can be caused by significant

back diffusion of helium from sweep to feed side (Table 1). The permeance of CO 2 is also af-

fected by water but not much due to competitive adsorption. CO 2 permeance reaches its maxi-

mum value at around 70 ⁰ C and then decreases consistently.

α CO

/H

0 1 2 3 4 5

0 100 200 300 400

Temperature,

C

S e p a ra ti o n fa cto r

S1 S2 Z

a)

α H

O/H

0 1 2 3 4 5

0 100 200 300 400

Temperature,

C

Separ at ion f ac tor

S1 S2 Z

b)

Figure 3.6 – a) CO 2 /H 2 selectivity and b) H 2 O/H 2 selectivity in the mixture CO 2 /H 2 /H 2 O for silicalite-1 and ZSM-5.

P CO

5 10 15 20 25

0 100 200 300 400

Temperature,

C P e rm ea nc e, [/ 10

mo l/ (s m

Pa )]

S1 S2 Z

a)

b) P H

0 20 40 60 80

0 100 200 300 400

Temperature,

C Pe rm e a nc e , [/1 0

mo l/ (s m

Pa )]

S1 S2 Z

P H

O

0 10 20 30

0 100 200 300 400

Temperature,

C Per m eanc e, [ /10

mo l/ (s m

Pa)

S1 S2 Z

c)

Figure 3.7 – Gas permeances in mixture CO 2 /H 2 /H 2 O for silicalite-1 and ZSM-5: a) CO 2

permeance; b) H 2 permeance; c) H 2 O permeance.

3.1.5 C

H

OH/H

separation

Separation of ethanol and hydrogen at low temperature is based on preferential adsorp- tion of ethanol. In comparison to the water molecule, ethanol is a larger and less polar molecule.

Figure 3.8 shows the C 2 H 5 OH/H 2 separation factors for silicalite-1 and ZSM-5 as a function of

temperature in the range 21-400 ⁰ C. The highest separation factors are observed at around 21 ⁰ C,

and are 19.1 for ZSM-5 and 16.6 for silicalite-1. At 50 ⁰ C, membranes are still selective towards

ethanol, and the separation factor is around 8.0 for both types of membranes, and then, at the

higher temperature, transition to selectivity for hydrogen is observed. At 390 ⁰ C, the membranes

are very hydrogen selective. ZSM-5 membrane has a higher C 2 H 5 OH/H 2 separation factor than

silicalite-1 membrane. This is probably due to better quality and also the smaller effective pore size of ZSM-5 membrane.

Figure 3.9 shows the permeance of ethanol and hydrogen in the temperature range 21-400 ⁰ C.

The hydrogen permeance is suppressed significantly by ethanol below 100 ⁰ C and then starts to increase. Hydrogen permeance at elevated temperature for silicalite-1 is much higher than for ZSM-5 membrane. It is probably caused by coke formation at high temperature which results in reducing permeance through the membrane.

α Eth/H

0 5 10 15 20

0 100 200 300 400

Temperature,

C

Separat ion f a c tor

S1 Z

Figure 3.8 – C 2 H 5 OH/H 2 selectivity for silicalite-1 and ZSM-5.

P H

0 20 40 60 80

0 100 200 300 400

Temperature,

C P er m ea nc e, [ /10

mo l/ (s m

Pa )]

S1 Z

a)

P Eth

0 4 8 12 16

0 100 200 300 400

Temperature,

C P e rm ea nc e / 1 0

mo l/ (s m

Pa)

S1

b) Z

Figure 3.9 – Gas permeances in the mixture Ethanol /H 2 as a function of temperature for silicalite-1 and ZSM-5: a) C 2 H 5 OH permeance; b) H 2 permeance.

3.1.6 n-Hexane/2,2-DMB separation

n-Hexane and 2,2-dimethyl-butene are hexane isomers, one of which, n-hexane is linear, and 2,2-dimethyl-butane is di-branched. The branched structure of 2,2-DMB molecule results in a larger kinetic diameter of molecule (0.62) as compared to n-hexane molecule (0.43). The di- ameter of both components is very close to the MFI pore size. Separation of the components mainly occurs by the molecular sieving mechanism at higher temperature. The linear molecules of n-hexane diffuse through the membrane more easily than the branched molecules of 2,2-DMB and the highest separation factor for n-hexane/2,2-DMB, 130, was observed at 230 ⁰ C. These re- sults can be compared with the previous data reported by Jonas et al. [27]. They reported the n- hexane/2,2-DMB separation factor of 75 at 100 ⁰ C and 227 at 390 ⁰ C for similar membrane.

α n-hex/2,2-DMB

0 40 80 120 160

0 100 200 300 400

Temperature,

C

Separ at ion f ac to r

S2

Figure 3.10 – n-hexane/2,2-DMB selectivity for silicalite-1

P n-hexane

0 2 4 6 8 10

0 100 200 300 400

Temperature,

C Perm eanc e [ /1 0

mo l/( s m

Pa)]

a) S1

P 2,2-DMB

0 4 8 12 16

0 100 200 300 400

Temperature,

C Perm eanc e [ /10

mo l/ s m

pa)]

b) S1

Figure 3.11 – Gas permeances in the mixture n-hexane/2,2-DMB as a function of tem-

perature for silicalite-1: a) n-hexane permeance; b) 2,2-DMB permeance.

4 CONCLUSIONS

MFI-type zeolite membranes, silicalite-1 and ZSM-5, were tested for separation of differ- ent mixtures: CO 2 /H 2 , H 2 O/H 2 , and CO 2 / H 2 O/H 2 , C 2 H 5 OH/H 2 , and n-hexane/2,2-DMB

.

The separation experiments with a binary mixture of CO 2 and H 2 showed that the mem- branes of both types were slightly hydrogen selective. The highest CO 2 /H 2 separation factors were observed at 22 ⁰ C and were 0.7 for silicalite-1 and 0.8 for ZSM-5. The hydrogen selectiv- ity could be explained by significant Knudsen diffusion through the support as well as very weak CO 2 adsorption.

The H 2 O/H 2 separation showed the highest selectivity towards water at 22 ⁰ C. The separa- tion factors were 2.2 and 4.0, for silicalite-1 and ZSM-5 respectively. ZSM-5 membrane has a higher H 2 O/H 2 separation factor than silicalite-1. This was explained by more pronounced affin- ity of water to ZSM-5 membrane.

When the feed was a ternary mixture of CO 2 , H 2 and H 2 O, the CO 2 /H 2 separation factors at 22 ⁰ C were 2.2 and 3.7 for silicalite-1 membranes S1 and S2 and 4.2 for ZSM-5 membrane.

The H 2 O/H 2 separation factors at 22 ⁰ C were 2.1 and 4.6 for silicalite-1 membranes S1 and S2 and 4.1 for ZSM-5 membrane. It was seen that at low temperature the CO 2 /H 2 separation factors in a ternary mixture are higher than that observed in a binary mixture of CO 2 and H 2 . Increased CO 2 /H 2 separation factors were achieved due to significant decrease in hydrogen permeance be- cause of blocking effect of water. H 2 O/H 2 separation factors are not affected by presence of CO 2 .

Separation of ethanol and hydrogen has shown that membranes of both types were very selective towards ethanol at low temperature. The highest separation factors were observed at around 21 ⁰ C, and were 19.1 for ZSM-5 and 16.6 for silicalite-1.

Separation of hexane isomers such as n-hexane and 2,2-dimethyl-butane using silicalite-1 membrane has shown that the highest selectivity towards n-hexane was achieved at 230 ⁰ C and was 130.

The main conclusion of this work was that water efficiently blocks the permeation of H 2

at low temperature and thus provides efficient separation.

REFERENCES

[1] J.D.F. Ramsay and S. Kallus, "Zeolite membranes" in Recent advances in gas separation by microporous ceramic membranes. Elsevier Science, Amsterdam, 2000.

[2] D.M.Ruthven. Principles of adsorption and adsorption processes. A Wiley Interscience Publication, the United States of America, 1984.

[3] J. Lindmark. Tailoring of MFI membranes for enhanced selectivity. Licentiate Thesis.

Luleå University of Technology, Luleå, 2006.

[4] M. Gualtieri. Synthesis and Characterization of Zeolite films and Membranes. Doctoral Thesis. Luleå University of Technology, Luleå, 2006.

[5] R. Krishna, R. Baur. Modelling issues in zeolite based separation processes. Separation and Purification Technology, 33 (2003) 213-254.

[6] C.N. Satterfield. Heterogeneous catalysis in industrial practice. Krieger Publishing Com- pany, Malabar, Florida, second ed., 1996.

[7] T.C. Bowen, R.D. Noble, and J.L. Falconer. Fundamentals and applications of pervapora- tion through zeolite membranes. Journal of Membrane Science, 245(2004) 1-33.

[8] C. Algueri, P. Bernardo, G. Golemme, G. Barbieri, and E. Drioli. Permeation properties of a thin silicalite-1 (MFI) membranes. Journal of Membrane Science, 222 (2003)181- 190.

[9] S. Aguado, A.C. Polo, M.P.Bernal, J. Coronas, and J. Santamaria. Removal of pollutants from indoor air using zeolite membranes. Journal of Membrane Science, 240 (2004) 159- 166.

[10] H.P. Hsien. Inorganic membranes for separation and reaction. Elsevier Science, Amster- dam, 1996.

[11] F. Jareman, and J. Hedlund. Single gas permeance ratios in MFI membranes: Effects of material properties and experimental conditions. Microporous and Mesoporous Materi- als, 82 (2005) 201-207.

[12] F. Jareman. MFI-Molecular Sieve Membranes Synthesis Characterization and Modelling.

Licentiate Thesis. Luleå University of Technology, Luleå, 2002.

[13] M. Nomura, T. Yamaguchi, and S. Nakao. Transport phenomena through intercrystalline and intracrystalline pathways of silicalite zeolite membranes. Journal of Membrane Sci- ence, 187 (2001) 203-212.

[14] V. Sebastian, I. Kumakiri, R. Bredesen, and M. Menendez. Zeolite membrane for CO 2

removal: Operating at high pressure. Journal of Membrane Science, 292 (2007) 92-97.