Gas separation of steam and hydrogen mixtures using an ?-alumina-Alumina supported NaA membrane

(1)

Y U ' N i ' B E S l T I Y A B O K O N E - B O P H 1 R I M A N O R T H - W E S T U N I V E R S I T Y N O O R D W E S - U N I V E R S I T E I T

Gas Separation of Steam and Hydrogen Mixtures

using an a-Alumina Supported NaA Membrane

By

S. Moodley, BScEng (Chemical)

This dissertation is submitted in fulfillment of the requirements for the degree of Masters in Engineering in the School of Chemical and Minerals Engineering at North-West University.

Supervisor. Prof. H.W.J.P Neomagus (North-West University, South Africa) Co-supervisors: Prof. R.C Everson (North-West University, South Africa)

Prof. H.M Krieg (North-West University, South Africa) Potchefstroom

South Africa December 2007

(2)

EDJTOPML D iFOPMATTO]'

Zeoltlfi Ixfembrarjfi Technology Declaration

Declaration

I, Shawn Moodley, hereby declare that this dissertation entitled "Gas Separation of Steam and

Hydrogen Mixtures using an a-AIumina Supported NaA Membrane", submitted in

fulfillment of the degree of Masters in Engineering, is entirely my own effort arid has oot been submitted at any other university either in part or as a whole.

Signed at Secunda, Mpumalanga on the 10th day of December 2007,

■)Cv

(3)

EDIT0P1AL JJ FORMA . 1.03

T ZBOHIR Meffihran Techri'-'la.gy

A c nowiedgRmenfS

Acknowledgements

Thank you to my supervisors, Professors Neomagus, Krieg and Everson, for imparting your expertise and insight on the relevant subject matter. I would also like to express ray gratitude for the patience and understanding shown during the completion of this dissertation.

To the technical staff at the School of Chemical and Minerals Engineering, your willingness to help with the technical aspects of my project, was truly and greatly appreciated.

To Jaco Zah, Hertzog Bissett, Anel van Niekerk, Dewald Kapp and Dr Lourens Tiedt, my sincerest thanks to you all for your shared expertise and assistance concerning the preparation and characterisation of the cr-alumina supports and NaA membranes used during the course of this study.

To my dear mother Maya, brother Luchen and late father Sham Moodley, thank you for being my pillars of strength and my motivation, during my time away on study and for always,

To my dear Aunt Shakun and her family, words cannot suffice in expressing my gratitude for helping me to become the person that I am today. I am forever grateful for all that you have done for me.

To Olwen, who was my friend, companion and confidante while I was studying in Potchefstroom; my heartfelt thanks to you for all your helpful advice, encouragement and support.

Tliank you to all my friends who supported me during my study in Potchefstroom.

(4)

l~UijlWQJf{!f;\Tb

WN.PWl{}\f.0-\lIl.fQJl'

T_r

;ZiJJji~e :l';Emjtrflifle·iFe~WieJWgy ,~~tf9ffU!Jt

Abstract

In this study, the feasibility of a NaA zeolite membrane for the gas phase separation of steam and hydrogen mixtures was determined. The Fischer-Tropsch (FT) process, which produces high value fuels and chemicals from coal and natural gas, can be greatly improved upon by the selective removal of water from the FT reactor product stream. According to the FT reaction kinetics, the rate of reaction increases with the partial pressure of hydrogen but is adversely affected the presence of water in the reactor product stream. Chemisorbed water on the surface of the metal catalysts also enhances deactivation due to sintering and fouling. The use of a zeolite membrane reactor is well equipped to serve the purpose of in-situ water removal as it can facilitate the separation of chemical components from one another in the presence of catalytic reactions. The LTA type zeolite membrane NaA or zeolite 4A, in particular, is well suited for the separation of polar (HzO) from non-polar (Hz) molecules because of its high hydrophilicity. NaA has also been identified as an excellent candidate for selective water removal applications due its high adsorption affinity and capacity for water.

The NaA membrane used in this study was manufactured by means of the in-situ crystallisation method where the growth of crystals on the inside surface of a centrifugally casted a-alumina support was favoured. Scanning electron microscopy (SEM) analyses performed on the membrane after a double hydrothermal synthesis indicated that the surface topology was rough and that the zeolite crystals formed were not uniform in size. Overall, the membrane thickness varied between 6.5 and 8.0 flm. An evaluation of the membrane quality was made possible through permeation experiments involving SF6 and Hz. The calculated Hz/SF6 permselectivity in

this study was found to be 9.78, which despite being higher than the Knudsen diffusion selectivity of 8.54, confirmed the presence of intercrystalline defects or non-zeolitic pores in the membrane. Experiments concerning pure component and binary mixture permeation of steam and hydrogen through the supported NaA membrane were conducted over a temperature range of 115°C to 160 °c for binary hydrogen/steam mixtures, 25°C to 160°C for pure hydrogen and 130°C to 170°C for pure steam. For the permeation of pure component hydrogen, a local maximum in its permeance having a value of 224 x 10'°8 mol.m,z.s'!.Pa'! was reached at a system pressure and temperature of 6.875 bar and 75°C respectively. For the permeation of pure component steam

(5)

1fJJ?JTI'IT1Q)Mi\TL

1]!'!RFQ)I~V;~\'ITrQ}l'fr

;Zediit~ :l\!E~Iljvrll.il~lfe~hfl~ltigj '1~tJ5fftict

through NaA, the effects of capillary condensation in the pores and defects of the zeolite membrane resulted in a decrease in steam permeance as a function of absolute pressure for temperatures lower than 160

°e.

Once the effects of capillary condensation had receded, maxima in the steam permeances as a function of temperature corresponding to values of 70 x 10,08, 65 X

10,08 and 75 x 10,08 mol.m·2_{.s'I.Pa'l were found for the 182.5, 197.5 and 222.5 kPa isobars}

respectively. These observations collaborated well with the description of surface diffusion with permeation taking place in the Langmuir (strong adsorption) regime.

Permeation experiments through NaA as function of temperature were conducted for a 90 mol% steam - 10 mol% hydrogen (90-10) binary mixture as well as for a 60-40 mixture of these two. At low temperatures the permeation of hydrogen was completely suppressed by the condensed steam resulting in an almost perfect separation. The Kelvin equation was used to estimate the pore size of the defects which was found to range between 1.86 and 2.45 nm. The temperature range over which these defects in the membrane were assumed to become unblocked (i.e. assuming when the first breakthrough of hydrogen occurred), were determined to be between 140 to 148

°e

and between 128 to 130

°e

for the 90-10 and 60-40 mixtures respectively. The mixture selectivities (towards water) between 115

°e

and 130

°e

were found to be immensely high (much greater than 1000) for both the 90-10 and 60-40 mixtures, while the ideal selectivities were calculated to be less than lover the same temperature range. At 140

°e,

the selectivity towards water for the 90

10 mixture was still greater than 1000; however for the 60-40 mixture at this temperature, an inversion of selectivity towards H2 had already taken place. The breakthrough in H2 permeance

occurs at a much lower temperature when the feed mixture contains a lower concentration of water. Since the partial pressure of steam will be reduced, larger pores will become unblocked at lower temperatures according to the Kelvin equation.

Keywords: Fischer-Tropsch (FT), NaA zeolite membrane, in-situ crystallisation, double hydrothermal synthesis, a-alumina support, capillary condensation, Kelvin equation.

(6)

rDITOPjALlJiPOPMATJOn I

Zeolite Membrane Tflclinology J

Table of Contents

EDITOPIAL n WOJtMA TIG] T i

Zeolite Membrane Technology -*-' Table of Contents

2.4.2. Diffusion in Porous Media 23 2.4.2.1. The Dusty Gas Model 25 2.4.2.2. Diffusion within Micropores 26

2.4.3. Controlling Mechanisms 28 2.4.3.1. Capillary Condensation 31

2.4.3.2. Water Transport through the NaA Mjcrostructure 33

2.4.3.3. Intercrystall ine Diffusion 34

CHAPTER 3-EXPERIMENTAL 36

3.1. INTRODUCTION 36 3.2. EXPERIMENTAL APPARATUS 36

3.2.1. Design of the Membrane Separation Module 43

3.3. ALUMINA SUPPORTS 44 3.3.1. Preparation of the Supports 46

3.3.2. Characterisation of the Supports 48 3.3.2.1. Mercury Porosimetry 48 3.3.2.2. Nitrogen Permeation through the Supports 48

3.3.2.3. Scanning Electron Microscopy of the Supports 49

3.4. SUPPORTED NaA MEMBRANES 49 3.4.1. Synthesis of the Supported NaA Membrane 49

3.4.2. Characterisation of the Supported NaA Membrane 51 3.5. MEMBRANE SEPARATION OF STEAM /HYDROGEN MIXTURES 52

CHAPTER 4-RESULTS AND DISCUSSION 55

4.1. INTRODUCTION 55 4.2. CHARACTERISATION OF THE SUPPORT 55

4.2.1. General Appearance of the Support 55 4.2.2. Support Characterisation by Hg Intrusion 56 4.2.3. Support Characterisation by Nitrogen Gas Permeation 58

4.2.4. Support Characterisation by Scanning Electron Microscopy (SEM) Analysis. 60

4.3. CHARACTERISATION OF THE SUPPORTED NaA MEMBRANE 61

4.3.1. Scanning Electron Microscopy Analysis 61 4.3.2. Evaluation of the Membrane Quality 63

4.3.2.1. Motivation 63

4.3.2.2. Permeation of H2 and SF6 64

4.4. PERMEATION OF PURE HYDROGEN THROUGH NaA 64 4.5. PERMEATION OF PURE STEAM THROUGH NaA 66

(8)

JIJJFTOPLAL

JJ fpOKMA J101 i s

Zeolite Mernfjrane Technology Tr:i:,lr. of Contents

4.6. BINARY PERMEATION THROUGH NaA 70 4.6.1. Effect of Capillary Condensation 73 4.6.2. Mixture and Ideal Selectivities 76 CHAPTER 5 - CONCLUSION AND RECOMMENDATIONS 79

5.1. CONCLUSION 79 5.1.1. Experimental Apparatus 79

5.1.2. Support and Membrane Preparation 80 5.1.3. Support and Membrane Characterisation 80

5.1.4. Gas Permeation Experiments 82

5.2. RECOMMENDATIONS 84 5.2.1. Support and Membrane Preparation 84

5.2.2. Experimental 85 5.2.3. Modelling 85

BIBLIOGRAPHY 86 APPENDICES

APPENDTX A: NITROGEN PERMEATION THROUGH THE oALUMTNA SUPPORT 92

APPENDDCB: GAS PERMEATION OF H2 AND SF6 97

APPENDTX C: PURE COMPONENT PERMEATION OF H2 THROUGH NaA 101

APPENDTX D: PURE COMPONENT PERMEATION OF STEAM THROUGH NaA 104 APPENDIX E: BINARY PERMEATION OF STEAM / HYDROGEN MTXTURES THROUGH NaA 112

APPENDTX F: KELVIN EQUATION CALCULATIONS 133 APPENDLX G: HA20P ANALYSIS OF EXPERIMENTAL INSTALLATION 139

APPENDIX H: STANDARD EXPERIMENTAL PROCEDURE 143 APPENDLX I: Hg INTRUSION SUMMARY REPORT 146

(9)

List of Figures

Figure 2.1: Schematic Representation of four Types ofZeolites 12

Figure 2.2: Repeating unit of Zeolite NaA 16 Figure 2.3: Comparison of Bulk and Knudsen Gas Diffusion Mechanisms 24

Figure 2.4: Three Distinct Mechanisms by which Molecular Species are Transported

within an Adsorbent or Catalyst Particle 25 Figure 2.5: A Conceptual Model for Surface Diffusion of Adsorbed Species 27

Figure 2.6: Typical Permeance of a Single Gas through a Zeolite Membrane as a

function of Temperature 31 Figure 3.1: Graphic of the Experimental Setup used for the Gas Permeation

Experiments 40 Figure 3.2: Graphic of the Tubular Oven Housing for the Membrane Reactor Module .. 41

Figure 3.3: Graphic of the Hydrogen Analyser and Soap Flow Meters that were used to

analyse the product streams 41 Figure 3.4: Process Flow Diagram representing the Separation of Steam and Hydrogen

Mixtures using a NaA Zeolite Membrane 42 Figure 3.5: Graphic of the Membrane Module 44 Figure 3.6: Graphic of the inside of the Membrane Module depicting the Membrane

and the different seals used during the experimental program 44 Figure 3.7: Graphic of the Centrifuge Apparatus used for the manufacture of the

a-Alumina Support 47 Figure 3.8: Graphic of the Autoclave used for the synthesis of the NaA Membrane on

the inside surface of the or-Alumina Support 51 Figure 4.1: Pore Size Distribution of the <z-AJumina Support used in this Study 58

Figure 4.2: Plot of the LHS of equation 2.5 against the Average Pressure across the

Support 59 Figure 4.3: SEM Photo of the Inside Surface of the or-Alumina Support Tube 60

Figure 4.4: SEM Photo showing the formation of the NaA Zeolite Layer on the Inside

Surface of the ar-AJumina Support 62 Figure 4.5: SEM Photo showing the Top Surface of the Zeolite Membrane formed on

(10)

j:DJTOPJAL J J ij-OPMA 'JJO]

J

if

Zeolite Mftnjbrsri*1 Technology

Lisl of Figures. Tables and Symbols

the Inner Wall of the or-Alumina Support 62 Figure 4.6: Pure Component Permeation through NaA: Permeance of Hydrogen vs

Temperature 65 Figure 4.7: Pure Steam Permeance through NaA as a function of Temperature 69

Figure 4.8: Pure Component Permeation through NaA. Permeance of Steam as a

function of Absolute Pressure 70 Figure 4.9: Binary Mixture Permeation through NaA. Permeances of Steam and H2 as

a function of Temperature (50 kPa AP; 90 mol % steam, 10 mol % H2: 1053.5 ml/min

(STP) N2 sweep gas used) 72

Figure 4.10: Binary Mixture Permeation through NaA. Permeances of Steam and H2 as

a function of Temperature (50 kPa AP; 60 mol % steam, 40 mol % H2: 1053.5 ml/min

(STP)N2 sweep gas used) 73

List of Tables

Table 2.1: Physical Properties of four types of Zeolites 11 Table 2.2: Characterisation Techniques and the Information that they Provide 20

Table 3.1: Specifications of Equipment Uti lised for Gas Permeation Experiments 39 Table 3.2: Specifications of Equipment Utilised for the Support and Supported

Membrane Manufacture 45 Table 3.3: Specifications of Equipment Utilised for the Support and Supported

Membrane Characterisation 45 Table 3.4: Specifications of the Chemicals Utilised 46

Table 4.1: Results for the Characterisation of the a-Alumina Support by Hg Intrusion 57 Table 4.2: Comparison of the Support Characterisation Results obtained in this study

with that reported by Steenkamp et al. [42] and Nunes [43] 57 Table 4.3: Use of the Slope and Intercept of Figure 4.2 to determine the Support Pore

Diameter and Porosity/Tortuosity Ratio 59 Table 4.4: Comparative H2 and SF6 Permeances following Gas Permeation

Experiments involving the supported NaA Membrane 64 Table 4.5: Summary of the Defect Pore Size Ranges that were estimated using the

(11)

Table 4.6: Use of the Kelvin Equation to provide Final Estimates of the Temperature Ranges in which the end of the Capillary Condensation Regimes for the different

Steam-Hydrogen Mixtures are initialised 75 Table 4.7: Mixture and Ideal Selectivities for the different Steam-Hydrogen Mixtures

and Temperatures investigated 77

Abbreviations

FT Fischer-Tropsch WGS Water Gas Shift

SEM Scanning Electron Microscopy CTL Coal to Liquid

GTL Gas to Liquid

HAZOP Hazard and Operability PFD Process Flow Diagram MFC Mass Flow Controller NRV Non-return Valve

BPR Back-pressure Regulator APMA Ammonium Polymethacrylate

ISS Intergrowth Supporting Substance

List of Symbols

Latin symbols

dp Pore diameter

Ed Activation energy of diffusion M Molecular weight

N; Permeation flux of component i P Pressure

p; Partial pressure of component i

Api Partial pressure difference of component i Qa Heat of adsorption run kJ.mol"1 g.moi"1 mol m2.s_1 or kg Pa Pa Pa Imol'1 mis"1

(12)

r Reaction rate raol.s" or kg.

rk Kelvin radius nrn

rP Pore radius ran

T Temperature K

Xi Mole fraction of component i in the feed to tbe membrane

-Yi Mole fraction of component i in the permeate

-Bo Permeability coefficient m2

Ko Knudsen coefficient m

P, Tube side pressure kPa

Ps Shell side pressure kPa

R Universal gas constant J.mo^.K"1

Po Saturated vapour pressure kPa

* COD Capillary condensation pressure kPa

Greek symbols

a Separation factor

-OK* Knudsen diffusion limit

-« i j Permselectivity -£ / l Porosity/Tortuosity Ratio

-P Density kg.m"3

e

Contact angle radians or degrees

a Surface Tension N.m"'

n

s Permeance of component i moI.mV.Pa"1

(13)

JJ-TCXODUCTTOF

Zedim Membrane Technology Charier 1

1 Introduction

1.1. General and Motivation

Currently the world's fuel and chemical production is based predominantly on crude oil. However, as the reserves of crude oil become depleted or the price of brent crude rises, an alternative means of producing these commodities on a large scale is needed to meet the world's growing energy demands. The importance of CTL (coal to liquid) process and GTL (gas to liquid) process technologies are being recognised throughout the world for the production of high value fuels and chemicals from coal and natural gas respectively. The presently known reserves of methane and of coal exceed that of crude oil by factors of about 1.5 and 25 respectively. At the heart of both these technologies are the Fischer-Tropsch (FT) reactors where purified synthesis gas, made up of carbon monoxide and hydrogen in a specific ratio, is reacted over metal catalysts to yield a broad spectrum of hydrocarbons. In South Africa, FT technology adds value annually to more than 40 million tons of low grade coal which cannot be exported or otherwise utilized. The beneficiation of this indigenous raw material into value-added final products has enabled South African petrochemical companies to contribute substantially towards the growth of the country's economy due to job creation, skills development and the profitability of their exports. To date, South Africa has over fifty years experience in FT technology and this acumen is already in high demand in countries such as Qatar and China that are looking to capitalise on their huge methane and coal reserves by building FT plants of their own.

1.1.1. Improving the productivity of the FT process

There are currently two FT operating modes. For the high temperature (300-350 °C) process the synthesis gas is sent to the FT reactors where the hydrogen and carbon monoxide react under pressure in the presence of a fluidised, iron-based catalyst to yield a broad spectrum of hydrocarbons in the CrC5 0 range. High value chemical components are produced simultaneously

(14)

JjiTROlJUCTiOlT

Zeolite Membrane Tec; .:vMogy

0;r-pier"l

Hydrocarbons produced in the reactors are cooled in successive stages in a product recovery plant until most components become liquefied.

The low temperature (200-240 CC) process with either iron or cobalt catalysts is used for the

production of high molecular mass linear waxes as well as high quality diesel. For cobalt based FT catalysts the dominant reaction is the FT reaction itself, typically

CO + 2.15#2-» hydrocarbons + H20 (1.1)

When iron-based catalysts are used, however, the water gas shift (WGS) reaction also readily occurs

CO + H2OoC02+H2 (1.2)

At high temperatures the WGS reaction is rapid and goes to equilibrium and this also allows CO2 to be converted to FT products, via the reverse WGS followed by the FT reaction.

In order to improve the productivity of the FT process by means of minimising reactor down-time and catalyst consumption, it is imperative that the metal catalysts used, maintain high activity for long times. The FT reactions are highly exothermic and it is important to rapidly remove the heat of reaction from the catalyst particles in order to avoid overheating of the catalyst, which would otherwise result in an increased rate of deactivation due to sintering and fouling.

The presence of H2O in the reactor product stream contributes substantially to this afore mentioned adverse effect on the activity of the catalysts. The exact detail of the chemical steps occurring during FT reaction remains a contentious topic but because the carbon-oxygen bond in CO has to be broken during the process it is very probable that both carbon species and oxygenated species such as H2O are chemisorbed on the surface of the metal catalyst. The former represents "carbided" metal sites and the latter "oxidised" metal sites. The process involves rapid cycling where, at any instant, a particular metal surface atom could be in either the carbided or oxidised states. This chemical cycling should enhance sintering and the loss of active surface area. The higher the proportion of exposed surface metal atoms, the higher the likelihood of these

(15)

JlT'iROJjuCTTOIT

Zeolile Membrane Teo'Kv-i -gy Cfiapler 1

processes occurring. This could mean that a highly disperse metal may well have a high initial FT activity but could rapidly decline with time-on-stream.

According to the FT reaction kinetics, equation 1.3, the rate of reaction increases with H2 pressure and at low conversion levels the rate is solely dependent on the partial pressure of H2. This reaction rate, however, is negatively affected by the presence of H20 (steam) in the product

stream. The H20 ends up in the vapour phase, thereby lowering the partial pressure of the reactant

gases and this consequently decreases the reaction rate. For iron-based catalysts, the FT kinetic equation is given by:

r= mP^Pco (1.3)

PcO+aPH20

Based on the information reported thus far, it follows that H2O and H2 are the key components in

a complex chemical system that can improve the efficiency of the FT process if their separation from each other in the reactor product stream can be optimised.

The use of membrane reactors can be highly effective in this regard as they are special types of multifunctional reactors where a chemical, often a catalytic reaction, takes place in the presence of a membrane. In the case of equilibrium-limited reactions, the achievable conversion can be improved; in the case of consecutive reactions the selectivity towards intermediate products can be increased. The type of in-situ membrane that would best facilitate the separation of H20 from

H2 is addressed in the next section.

1.1.2. Selection of a Suitable Membrane

Due to their low temperature stability, organic polymer membranes could be used in low-temperature membrane reactor systems only at T<150 °C most of them in biotechnology. Metallic and ceramic membranes become suitable materials for membrane reactors as they could withstand the temperatures of 400-600 °C usually applied in heterogeneous catalysis. The advantage of using ceramic membranes, especially zeolites, is that they can potentially separate molecules in a continuous way. The crystalline nature of zeolites offers a unique opportunity to

(16)

J J - T ' J R O J J T J ' C T J . O F

Zeolite Membrane Techno'; -cry

Chapter 1

obtain membranes with mono-dimensional pore sizes, which are therefore capable of separating mixtures of substances based on differences in molecular size and shape, and selective adsorption, with extremely high selectivity. Zeolite membranes have high thermal, chemical and structural stability, and for this reason they can be used at high temperatures and with organic solvents, when polymeric membranes cannot operate.

The last decade has witnessed a significant progress on zeolite membranes. MFI zeolite membranes showed promising performance on organic separations, e.g., butane isomers and xylene isomers, while X-type and Y-type zeolite membrane showed good performance on carbon dioxide separations and 1,3-propanediol separations. For gas separation, most of the reports have concentrated on the MFI (Silicalite-1 and ZSM-5) zeolite membrane because MFI zeolite has a channel opening size close to several industrial important materials, e.g. butane isomers. Although high separation factors have been obtained for these mixtures, the MFI zeolite membrane has not shown good separation performance for permanent gases or permanent gas/organic gas because of the large pore size and the hydrophobic nature of the zeolite. Therefore the investigation of zeolite membranes with small pore sizes, such as A-type zeolites, has attracted much attention.

NaA zeolite, for example, has a channel opening size of 0.41 nm, which is smaller than that of the MFI zeolite (0.55 nm). However, with reference to this study, the kinetic diameters for water (0.26 nm) and hydrogen (0.29 nm) molecules are still smaller than the pore size of NaA; therefore the molecular sieving principle becomes inadmissible as a separation mechanism. Very few instances of this mechanism have in any case, been reported since it is only exclusive to zeolite membranes that have no defects or intercrystalline voids. In reality, the combination of the individual adsorption and diffusion properties of the permeating species has been found to dictate the permeation performance of zeolite membranes on most occasions. The concentration of intercrystalline voids in the membrane, which maybe due to inadequate crystal intergrowth, anisotropic crystal structures or random crystal orientation, is also understood to have a quite a significant influence on permeation. Despite recent attempts to improve on the quality of NaA membranes, the presence of intercrystalline voids was not overcome completely. It has only been reported that some preparation methods were found to yield membranes with a lower concentration of intercrystalline voids than others.

(17)

The decision to use a NaA membrane in this study, nonetheless, is based strongly on the fact that these types of zeolites have been identified as excellent candidates for selective water removal applications due to their high adsorption affinity and capacity for water. NaA zeolite is also well suited for the separation of polar (H20) from non-polar (H2) molecules because of its high

hydrophilicity.

1.2. Objective of Investigation

The primary objective of this investigation was to determine the feasibility of a NaA zeolite membrane for the gas phase separation of steam and hydrogen mixtures. The acquisition of repeatable experimental results and the critical discussion thereof was of paramount significance to this research.

1.3. Scope of Investigation

The scope of this investigation can be summarised as follows:

□ Firstly a suitable support for the growth of the NaA zeolite membrane had to be identified, manufactured and characterised.

□ Secondly, the number of hydrothermal treatments required for an effective NaA zeolite membrane was decided upon. In addition to the characterisation of the synthesised membrane a suitable probe gas had to be used to evaluate the membrane quality.

□ Finally, an existing experimental set-up was modified to facilitate the separation of steam and hydrogen mixtures. The process variables of interest were the mixture feed composition, the module temperature and the system pressure.

Chapter 1 of this dissertation addresses the importance of FT technology as one of the South Africa's major economic drivers. Improving the productivity of this process would be highly profitable, thereby leading to further economic prosperity for the country. The presence of water in the reactor product stream has a negative effect on the rate of the FT reaction as well as on the

(18)

JJ T T J ^ . O D U C T T . O J T

Zeo'h ~ Membrane Technology Chapter i

activity of the metal catalysts used. The in-situ removal of water from the FT reactors presents an ideal opportunity for the deployment of zeolite membrane technology as it is less energy intensive and can offer a high degree of separation in chemically aggressive environments.

In Chapter 2, attention is focused on published information that is relevant to the field of membrane technology. The fundamental concepts behind membrane separations, gas permeation, zeolite membranes, membrane preparation and characterisation and transport phenomena are all incorporated within this chapter. Special attention was devoted to intercrystalline diffusion, capillary condensation and the reported use of NaA membranes to selectively separate water from water vapour/gas mixtures.

Chapter 3 concerns the experimental procedures that were followed during the manufacture and characterisation of the a-alumina support and the NaA membrane. A description of the membrane separation process is included for the binary mixture permeation of steam and hydrogen. Finally, the details of the different gas permeation equipment, characterisation equipment and chemicals used are specified.

Chapter 4 accounts for the results and discussion of the experimental work performed during this study. References to other studies were frequently made that helped provide a critical assessment of the experimental results before any legitimate conclusions could be reached.

Finally, in chapter 5, a summary of conclusions are compiled and recommendations for further investigative work are made.

(19)

Literature Review

2.1. Introduction

This chapter provides an assessment of reported information on the topic of membrane technology that is relevant to the following:

• The significance of membrane separations as a complementary technology to other conventional separation technologies.

• The suitability of membrane materials for potential or current use in diverse industrial applications.

• The preparation and optimisation of zeolite membranes, specifically NaA. • The mechanisms governing mass transport through porous media (zeolites).

2.2. Membrane Separations

Membrane separation is an emerging unit operation. Important progress is still being made in the development of efficient membrane materials and the packaging thereof for industrial applications. Already, membrane separation processes have found wide application in such diverse industries as the beverage, chemical, dairy, electronic, environmental, food, medical, paper, petrochemical, petroleum, pharmaceutical and textile industries. Often, compared to other separation equipment, membrane separators are more compact, less capital intensive, and more easily operated, controlled and maintained. The replacement of more common separation operations with membrane separations has the potential to save large amounts of energy. This replacement requires the production of high mass-transfer flux, defect-free, long-life membranes on a large scale and the fabrication of the membrane into compact, economical modules of high surface area per unit volume, Seader and Henley [18].

In a membrane separation process, a feed consisting of a mixture of two or more components is partially separated by means of a semi-permeable barrier (the membrane) through which one or more species move faster than another or other species. In the most general membrane process the

(20)

J J T J i K A T U J ' J i j'.JiVJJiW

Zeolile Membrane Technology Chapter 2

feed mixture is separated into a retentate (that part of the feed that does not pass through the membrane, i.e. is retained) and a permeate (that part of the feed that does pass through the membrane). The membrane must not dissolve, disintegrate or break. The optional sweep, is a liquid or gas, used to help remove the permeate.

The membrane process is particularly useful as a separation technique whenever conventional separation methods cannot be used economically to get reasonable separation. It can also be used as a unit operation in conjunction with a conventional separation unit. For example, a membrane permeation unit can be used to break an azeotropic mixture before feeding it to a distillation column, Perry et al. [19]

In general, the use of gas- or liquid- permeable membrane separation techniques is of special benefit for separating

1. Mixtures of compounds of similar chemical and physical properties 2. Mixtures of structured or position isomers, and

3. Mixtures containing thermally unstable components.

Membrane processes in general can be classified based on the driving force and the phases of feed and permeate streams. For example, microfiltration, ultrafiltration and reverse osmosis belong to the pressure driven operations with liquid feed and permeate. The pressure driving force for conventional ultrafiltration is in the range of 1-5 x 105 Pa typically. As the pore size

decreases from nanometer to molecular level, the pressure driving force required can be as high as 100 x 105 Pa, Chiang et al. [15]

The key to an efficient and economical membrane separation process is the membrane and the manner in which it is packaged and modularised. Desirable attributes of a membrane are:

1. Good permeability 2. High selectivity

3. Chemical and mechanical compatibility with the processing environment 4. Stability, freedom from fouling and reasonable useful life

5. Amenability to fabrication and packaging

(21)

LJTJiJ* ATUj^Ji i> JiVJJiV/ Zeolii'" Membrane Technology

Char"-r2

Research and development of membrane processes deals mainly with the discovery of suitable membrane materials and their fabrication, Seader and Henley [18].

Most membranes for commercial separation processes are natural or synthetic, glassy or rubbery polymers. However, for high temperature (>200°C) or operations with chemically reactive mixtures, ceramics, metals, and carbon find applications. The mean annual growth rate of inorganic membranes is about 30% and inorganic membranes was expected to reach in 2003 ~15% of the membrane market volume. At present, microfiltration is the dominant application field of the membranes with about 2/3 of the market volume. Increasingly new applications of membranes in fuel cells and in catalytic membrane reactors are studied. The new application fields have high demands and expectations from the membrane material such as thermal stability for high-temperature applications, solvent and chemical stability, sterilization ability and biocompatibility. Despite their material variability and the highly developed module technology, organic polymer membranes can hardly fulfil the structural and functional requirements of the new application fields. Therefore, increasing research on three types of inorganic membranes has started: (I) zeolite and (ii) sol-gel based microporous membranes as well as (iii) Pd based and Perovskite like dense membranes, Caro et al. [1]

2.2.1. Gas Permeation

In gas permeation, mixtures of gases are separated by differences in permeation rates through the membrane. The driving force for each component is its partial pressure difference across the membrane. Both the permeance and permeability depend on the adsorption characteristics of the membrane for the particular gas species and the diffusivity of the species through the membrane.

Gas permeation must compete with distillation at cryogenic conditions, absorption, and pressure-swing adsorption. Some of the advantages of gas permeation are low capital investment, ease of installation, ease of operation, absence of rotating parts, high process flexibility, low weight and space requirements, and low environmental impact. In addition, if the feed gas is already at so high a pressure that a gas compressor is not needed, then no utilities are required, Seader and Henley [18].

(22)

JJTJiJ'ATTjj^Ji J^JiVjJiW

Zeolile Membrane Technology Chnp1er2

2.2.2. Zeolites

Zeolites, natural and synthesized, are used widely as adsorbents in gas and liquid separations and drying processes. Typically, zeolites are hydrated, porous crystalline aluminosilicates, which can act as sieves on a molecular scale, Raileanu et al. [14] and are routinely used as catalysts, ion-exchangers and adsorbents, Chiang et al. [15].

The framework is an assemblage of Si04 and A104 tetrahedra joined together by sharing of

oxygen atoms. The tetrahedrally coordinated atoms (T-atoms) arrange themselves into cage-like unit cells (also called secondary building structures) featuring pores of molecular dimensions. The sizes of the micropores are specific to the framework type of the unit cells, van Bekkum et al. [20]

Zeolites have unique properties that are very attractive features for a large number of industrial applications. In particular, the acid sites due to the presence of aluminium, the high specific surface and the well defined pore dimensions have imposed them as selective catalyst materials. The crystalline nature of zeolites offers the opportunity to obtain membranes with a regular tridimensional network of micropores at the molecular scale and they are therefore able to separate mixtures of substances on the basis of differences in the molecular size and shape, such as, for example, isomers, compounds with similar molecular weight and also azeotropic mixtures, Algieri et al. [21]

An ideal zeolite membrane combines the general advantages of inorganic membranes (temperature stability, solvent resistance) with a perfect shape selectivity. Due to their "molecular sieve" function, zeolite membranes can principally discriminate the components of gaseous or liquid mixtures dependent on their molecular size, Caro et al. [1]

2.2.2.1. Types of Zeolite

A brief description of each type of common zeolite is given in Table 2.1 with illustrations of these given in Figure 2.1.

(23)

LITEPATUPEPEVTEW

;

|T

Zeolite Me-jbrans Tfidinol ■■.!_. -*-Chapler 2

Type Isotypes Pore window

(free diameter)

Si/Al ratio Pores/Channels

LTA A zeolite 8-ring: 0.41nm ~1 3D Spherical 1.14nm cavities

FAU X zeolite

Y zeolite

12-ring: 0.74nm 1-1.5 (X)

1.5-3 (Y)

3D Spherical 1.18nm cavities

MOR Mordenite 12-ring: 0.70nm 5-20 2D Straight 0.70nm channels

MFI ZSM-5

Silicalite-1

10-ring: 0.60nm -30 (ZSM-5)

oo (Silicalite)

3D straight 0.60nm channels with .90nm intersection cavities

Table 2.1: Physical Properties of four types of Zeolites; van Bekkum [20].

2.2.2.2. Applications of Zeolites and Zeolite Membranes

Morigama et al.[8] reported on the current industrial operation of the first large-scale pervaporation plant. This plant produces 5301/h of solvents at less than 0.2 wt.% of water from 90 wt.% solvent at 120°C. It is equipped with 16 modules, each of which consists of 125 pieces of NaA zeolite membrane tubes.

Other applications of zeolite membranes that have been, or are currently being researched include:

• Technology for hydrogen separation and recovery from various petrochemical and chemical streams, which is one of the most important factors for improvement of the global environment because hydrogen is an important energy media. There is a huge amount of hydrogen that is vented or flared annually because it cannot be economically recovered with current technologies, Noble et al. [5]

• Gas purification processes, where the very low water uptake and, therefore, high hydrophobicities of siliceous materials like all-silica zeolite beta and silicalite are beneficial

(24)

JTEPATUPEPEVIEW

Zeolite Membrane Technology Chnplerz

for the economics of the process. All silica zeolite beta is most suitable for a selective adsorption of organic vapours from moist gas streams, Stelzer et al. [7]

8 ring - .41 nm

12 ring - .74 nm

(a) LTA (Zeolite A) (b) FAU (Zeolite X or Y)

(c) MOR (Mordenite)

(d) MFI (Pentasil Zeolite e.g. ZSM-5)

(25)

LITERATURE REVIEW

Zeoliln Membrane Technology Chnptcr 2

• Catalytic applications (viz. silicalite-1 membranes) and their potential use in membrane reactors as they can withstand severe operating conditions, Kapteijn et al. [4]

• Separation of industrially important gases based on size and shape selectivity of the zeolite. By tailoring the pore size of ion-exchange membranes viz. Zeolite A, alkanes with different molecular sizes can be separated. For example, when NaA zeolite is ion-exchanged into CaA form, the pore size of the zeolite is slightly enlarged. Thus, the linear alkanes can permeate through the zeolite channels, while the branched alkanes cannot. In industry, CaA zeolite has already been applied in the separation of n-butane and z-butane by pressure-swing adsorption, Xuetal.[24]

2.3. Membrane Preparation and Optimisation

2.3.1. General Synthesis Considerations

An ideal structure of a zeolite membrane should be a slice of a perfect zeolite crystal attached on a porous metal or ceramic support. To maximise the throughput, the zeolite layer must be very thin, limited only by the cell dimension of zeolite. Separation of a mixture may then be achieved based on the molecular sieving ability of zeolite, which allows only molecules smaller than a critical size to pass through, Chiang et al. [15]. This molecular sieving principle requires a pinhole and crack free zeolite membrane, Caro et al. [1]

The mechanisms of zeolite formation are very complex due to many factors involved as chemical reactions (temperature, alkalinity, pH, chemical composition of the reaction mixture), equilbria, solubility variations that occur through the heterogeneous synthesis mixture during the crystallization process. The zeolite formation process is thermally activated in order to achieve a high yield of crystals in a "decent" period of time. In spite of the intensive research on the synthesis of zeolites, the factors controlling the synthesis of porous crystals are still not well-defined, Raileanu et al [14]. Because of the large number of factors and degrees of freedom that are at work in zeolite layer formation the structure of the membranes may differ considerably, thus resulting in poor reproducibility, Gora et al. [16]

(26)

LITER A TTJPE PEVTEW

Zreolile Membrane TeehnoJogy Chapter 2

The traditional one step method to prepare supported zeolithic membranes consists in immersing the support in a suitable hydrothermal system, and eventually repeating the deposition if the membrane has defects. This method tends to produce almost defect-free membranes with high selectivity, but the flux is very low due to the rather thick layers, and also because a significant part of the membrane can fill the pore system of the support, Algieri et al. [21]

2.3.2. Significance of the Support

To obtain a reasonable flux, the zeolite membrane must be thin, Chiang et al. [15]. Self standing zeolite layers larger than a few square centimeters are difficult to form, and the resulting structures are fragile. Therefore, zeolite membranes are typically deposited on mechanically resistant porous supports (alumina, stainless steel etc.), Algieri et al. [21]. Other materials such as porous glass and anodic alumina have also been used, as well as a titania support in the case of titano-silicate molecular sieve ETS-4, Chiang et al. [15]. The technical problem in the development of supported zeolithic membranes lies in the large thickness (typically 50 urn) required to obtain defect-free samples, that is the presence of defects in very thin layers, Algieri et al. [21]

The flow resistance and the bonding with zeolite are the main concerns in choosing a support. The permeance of hydrogen through a 2mm thick porous disk with a 0.2 um pore is about 50 x 10"7 mol/m2 Pa.s at normal temperature and pressure. Compared to the typical permeance of 1 x

10"7 mol/m2 Pa.s found for a 60 um MFI zeolite membrane, the porous disk may contribute about

5% of the total flow resistance. The resistance of the support may become the dominant effect when the thickness of the zeolite layer is reduced. Therefore, asymmetric supports with good strength and low flow resistance are preferred, Chiang et al. [15]

A strong effect of the support on the flux and separation properties of the membrane can be expected in cases of strongly adsorbing gases with relative high pressure or low temperature if the support resistance is non-negligible (but small). This is caused by relative strong changes in occupancy at the interface of support and membrane layer with small changes in the permeate

(27)

LTTBPATTJPEPBVTBW is

Zfioiile Membrane Technology J

Chapter!

The thermal expansion coefficient of both the support and the zeolite must be considered if the membrane is to be used or calcined at high temperature. Furthermore, many zeolites are synthesisefl with the help of an organic template. The thermal removal of the template leads to shrinkage of the lattice. Different supports also alter the extent of lattice shrinkage.The difference in thermal expansion of support and zeolite may cause stress in the interface and weaken the attachment of the zeolite layer. Furthermore, since the shrinkage resulting from the removal of the template is irreversible, non-zeolithic pores may be created during this process. For the synthesis of a zeolite layer on a porous support, the challenges are the intrusion of zeolite into the porous support and the sealing of the intercrystalline void to make the zeolite layer continuous and pinhole free. Since the support is immersed in a reactive sol, its pores will be filled with the liquid. Nucleation and deposition of zeolite happen on the surface as well as inside the pores. The permeation property of the support will be altered by the deposition of zeolite inside the pores, and will be difficult to differentiate from the effect of zeolite layer itself. To avoid the penetration of sol into the porous support, different approaches, including the filling of pores with oil, the position of support just slightly above the liquid surface, the use of reverse osmosis pressure and more have been studied, Chiang et al. [15]

2.3.3. NaA Zeolite Membranes

2.3.3.1. Introduction

NaA zeolite membranes have already been commercialised, Urtiaga et al. [33] and have since achieved significant progress in pervaporative-dehydration operations, Xu et al. [26]. The first large-scale pervaporation plant using NaA membranes has been commissioned for the production of industrial solvents and has realised much success to date, Morigami et al. [8]. NaA (4A) zeolite is relatively simple to synthesise, has importance as a detergent zeolite, Slangen et al. [31] and is also known as an effective adsorbent for carbon dioxide removal from natural gas in industrial purification processes, Khodakov et al. [32]. Zeolite A membranes are considered to be effective in the separation of refinery gases, oxygen and nitrogen, and butane isomers.

(28)

LITEPATUPBPBVIBW 'f

Zeoiiifi Membrane 7edri~i'<;;\ Giapler 2

However, most of the reports on NaA zeolite membrane focus on the use of pervaporation for dehydration and few gas permeation studies are reported Xu et al. [25]

The structure of zeolite A is shown in Figure 2.2. As shown, the aluminosilicate framework of zeolite A is generated by placing truncated octahedrons (/?-cage) at eight corners of a cube and each edge of the cube is formed by joining two /?-cages by a D4R linkage. Each /?-cage encloses a cavity with a free diameter of 0.66 nm and each unit cell encloses a larger cavity known as the a-cage enclosing a free diameter of 1.14 nm. There are two interconnecting, three-dimensional channels in zeolite A: (i) connected or-cages, 1.14 nm in diameter, separated by 0.42 nm apertures, (ii) /9-cages, alternating with the or-cages separated by 0.22nm apertures. Thus, molecules smaller than 0.42 nm in diameter can diffuse easily through the micropores of the zeolite. Also, the position of sodium ions in the unit cells is important since these ions act as the sites for water sorption and transport through the membrane. For a typical zeolite, a unit cell having the compostion Na12Al12Sii2048.27H20, eight (out of 12) sodium ions are located inside

the or-cage and four ions are located in the /?-cages, Shah et al. [28]

Truncated Oclahedjnats (p-cages.)

elCages

-D£R linages

Figure 2.2: Repeating unit of Zeolite NaA, Shah et al. [28]

When fully hydrated, there are 27 water molecules in the pseudo unit cell. The a and /? cages have space for 20 and 4 water molecules, respectively. The remaining three water molecules link through the eight-rings between the dodecahedral arrangements of water. Therefore, the maximum water adsorption capacity of the dehydrated zeolite-4A is 28.51 wt.% or 15.84 mol

(29)

kg-1. The high adsorption affinity and capacity for water make zeolite-4A of interest for application in water removal/separation, Zhu et al. [36]

2.3.3.2. Synthesis Considerations

The quality of zeolite A membranes can be influenced in many ways, e.g. by optimizing the synthesis mixture (nutrients used, nutrient ratios, dilution, aging time, alkalinity) and/or the hydrothermal synthesis procedure (temperature, time, way of heating, static or rotated). Additionally, the application of crystallization seeds, multistage syntheses, or organic templates have been investigated, and it has become clear that these approaches do not improve the quality of zeolite A membranes in terms of both selectivity and permeability. The first two increase the membrane thickness, which causes an improvement of the selectivity but a decrease in permeability. The last one facilitates crystal growth in free solution, thus, causing the membrane not to be formed completely. The parameters described are related to hydrothermal in-situ synthesis. However, it is also possible to create a membrane by means of chemical bonding between existing zeolite crystals and the support with covalent or ionic bonding. So far, the zeoliteA membranes synthesized with this technique give no satisfying results. Furthermore, other techniques like electrophoresis, as yet, do not give any satisfying results, Van den Berg et al. [22]

The hydrothermal synthesis of active NaA zeolite membranes on the surface of porous tubular supports like a-alumina were reported by Shah et al. [28], Morigami et al. [8] and Kondo et al. [34] while Jafar et al. [23] prepared membranes from homogenous solution on macroporous zirconia composite supports in both sheet and tube form. The synthesis of a NaA membrane on a ceramic hollow fiber has also been described by Xu et al. [26].

Xu et al. [24] reported on the synthesis of a NaA zeolite membrane on an a-Al203 support with

the aid of nucleation seeds from a gel synthesis mixture. The influence of synthesis conditions, such as synthesis times, synthesis stages and nucleation seeds, on the formation and permeation properties of the NaA zeolite membrane was investigated. According to the formation mechanism of the zeolite membrane on the porous support, the nucleation of zeolite on the support/gel interface and in the bulk synthesis mixture are competitive processes. In order to form a

(30)

LJTJiJ<ATTj]<JiJ'JiVTJiW CliapiRr2

continuous NaA zeolite membrane, the nucleation of the zeolite in the bulk synthesis mixture must be inhibited while the nucleation of NaA zeolite on the support surface must be promoted, i.e. the amount of NaA zeolite nuclei on the support surface must be increased at the beginning of the synthesis. It has long been recognized that adding seed crystals to crystal synthesis system can increase the rate of crystallization and can improve the purity of the crystal product, Gora et al. [12]. However, seeds need not be large, but may be quite small, even undetectable to the naked eye. These crystallization catalysts are sometimes called "directing agents" and they serve the purpose of promoting the rate of formation of the desired phase.

The synthesis of NaA zeolite membranes from clear solution using both seeded and unseeded supports were investigated by Xu et al. [25]. When an unseeded support was used, the NaA zeolite began to transform into other types of zeolites e.g., NaX zeolite and hydroxy-sodalite zeolite, before a continuous NaA zeolite membrane formed. When the support was coated with nucleation seeds, not only was the formation of NaA zeolite on the support surface accelerated, but the transformation of NaA zeolite into other types of zeolites was inhibited as well. The multi-stage synthesis method to improve the quality of the membranes was also investigated, Xu et al [25] and Xu et al. [24]

Xu et al. [27] developed further, a novel synthesis method of microwave heating for zeolite membranes, in order to increase the permeance of several solitary gases. The use of microwave heating significantly promoted the formation of a thinner and more uniform NaA zeolite membrane than that synthesized by conventional heating. Gas permeance values were found to be four times higher while the permselectivities of both membranes were comparable.

Numerous papers have now appeared on the subject of microwave synthesis of zeolites. The authors invariably report of highly shortened synthesis times compared to conventional heating methods. According to Slangen et al. [31] however, this shortening of the synthesis time is obtained at a certain price: the extra care needed for the preparation of the synthesis mixture. Ageing of the synthesis mixture is an important prerequisite for fast crystallization. Sufficiently aged mixtures can yield NaA, with crystal sizes ranging from 0.1 to 0.3 urn, after 1 min in the microwave. It was also found that the synthesis of an un-aged mixture could be improved

(31)

U T l i i ' A T L i J r ' j i j<Ji"VJJiW

ZeoJilr: Mn—brane Technology Chapter 2

dramatically by adding aged synthesis mixtures to it This suggests that the rearrangement of the synthesis mixture to yield nuclei is the bottleneck in a microwave synthesis.

Pina et al. [10] proposed a new semi-continuous procedure to prepare NaA zeolite membranes over tubular supports in a single step, in a reaction vessel where the synthesis solution is rapidly removed from the membrane environment at regular intervals, and simultaneously replaced by the same volume of fresh gel. This represents a compromise between a batch process, in which there is a significant depletion of reactants over time, and a process with a continuous feed of low flowrate, where a certain variation of concentration along the support is inevitable, likely leading to inhomogeneous growth of the membrane. The aim of the investigation was to overcome the problems of unwanted synthesis in the bulk of the liquid phase and poor crystal intergrowth due to local exhaustion of reactants in the autoclave. At the same time, the procedure allows a better control of the composition of the reacting mixture and avoids heating and cooling steps between syntheses, thus improving the feasibility of the scale up.

2.3.4. Membrane Characterisation

Information on the structural, chemical and catalytic characteristics of zeolites is essential for deriving relations between their chemical and physicochemical properties on the one side and the sorptive and catalytic properties on the other.

In general, the characterisation of a zeolite has to provide information about (i) its structure and morphology, (ii) its chemical composition, (iii) its ability to adsorb and retain molecules and (iv) its ability to chemically convert these molecules. The different techniques for characterisation and the information they provide are listed in Table 2.2, Van Bekkum et al. [17]

(32)

LTTEPATuPEPEVIEW t

Zeolile Membrane Technology -*■ Chapier 2

Technique Synonym Structure Pore Size Chemical

composition Functional groups Electron microscopy HRTEM SEM EDX

Yes Yes Yes No

Nuclear magnetic resonance (Magic angle spinning)

MAS-NMR No Yes Yes Yes

Sorption of probe

molecules - No Yes No Yes

Model reactions - No Yes Yes Yes

Thermogravimetry, differential scanning

calorimetry TGA/DSC No Yes Yes Yes

Temperature programmed desorption

TPD No No Yes Yes

Vibrational

spectroscopy - Yes Yes Yes Yes

X-ray absorption

spectroscopy XAS Yes No Yes No

X-Ray diffraction XRD Yes No Yes No

X-Ray fluorescence

spectroscopy XRF No No Yes No

X-ray photoelectron

spectroscopy XPS No No Yes Yes

Table 2.2: Characterisation Techniques and the Information that they Provide; reproduced from van Bekkum et al. [17]

(33)

JJJ'JiJ'AT'U^JiJ'JiVJJiW Zeoiile Membrane Tech"' -'ogy

Chapter 2

2.3.5. Evaluation of the Membrane Quality

XRD and SEM characterisations can only indicate whether a continuous membrane forms on the support; they cannot confirm whether a defect-free zeolite membrane forms. The quality of the zeolite membranes can only be evaluated by gas permeation, Xu et al. [24].

Caro et al. [I] reported the single gas fluxes for different probe gases on supported silicalite-1 and ZSM-5 membranes. The molecular sieving pattern exhibited by the silicalite-1 represented an MFI membrane with low defect concentration. Small molecules like H2, H20, C02, CH4, 02, and

N2 can pass the molecular sieve membrane with similar, strongly T-dependent fluxes. The rather

T-independent fluxes of the bulkier molecules like SF6, i-C/iHio, or 2,2 dimethylbutane (DMB) can be taken as a direct measure for the defect concentration of the membrane under study. The silicalite-1 membranes showed a remarkable splitting of the single component fluxes over five orders of magnitude. However for the ZSM-5 membrane, the evaluation of the membrane quality was more complicated as the single component gases exhibited no clear "cut- off'. From the absence of a clear cut-off, however, it can be concluded that the mass transport was not controlled by the pore system of the ZSM-5 membrane.

Xu et al. [25] selected the permselectivity, section 2.4.1, of H2/W-C4H10 as the yardstick for the perfection of the NaA zeolite membrane, taking into consideration the pore size and the hydrophilic nature of the NaA zeolite. As the kinetic diameter of «-C4Hio (0.43 nm) is larger than

the pore size of the NaA zeolite channels (0.41 nm), 7i-C4Hi0 should not permeate through a

defect-free NaA zeolite membrane. The higher the permselectivity of H2/«-C4Hio, the more

perfect the NaA zeolite membrane.

2.4. Theory of Transport for Zeolites

2.4.1. Permeance and Selectivity

To be effective for separating a mixture of chemical components, a membrane must possess high permeance and a high permeance ratio for the two species being separated, Seader and Henley

(34)

PATUPEPEVIEW

[

g\

o!i(eMern!;rfinfiT(;ch;-i"inrrv -^

JJTJi

Zeoli

Chanter;

/"7S/. Permeance is used for temperature programmed permeation experiments, because it allows a simple compensation for the changing concentration gradient over the membrane during these experiments, Bakker et al. [3]. For a given species diffusing through a membrane of given thickness, the permeance ZJ is analogous to a mass transfer coefficient, and is related to the permeation flux, Nt according to:

N

n , = ^ - (2.D

where Apt is the partial pressure difference of species i over the membrane. Nt is defined as the molar flowrate of species i in the permeate per unit surface area of the membrane per unit time.

The ratio of the permeances of two pure gases measured at the same temperature is the ideal selectivity or permselectivity (atj), Algieri et al. [21]

n,

au=i=r (2-2)

Permselectivity values are compared with the Knudsen diffusion limit aKn(ijf, when this mechanism is predominant light gases permeate faster than heavier gases. The selectivity in the case of Knudsen diffusion is independent from the pressure and is proportional to the inverse square root of the molecular weight (M):

««u)=,f-rr- (2.3)

An experimental selectivity higher than the Knudsen limit may indicate that the mean pore size of the membrane is comparable to the molecular dimensions of the largest species, Algieri et al. [21]

For the separation of a binary gas mixture of species A and B, the selectivity factor aAiB is defined as:

(35)

IJTEPATUFE REVIEW

f g g |

Zei-Hit- Membrane Techno'~r-'

f g g |

Ch; pier 2

^^=ts>^

^ /3 = , , ' ( (2-4)

where yA and yB are the concentrations of components A and 5 in the permeate and xA and xB are the concentrations of the components in the feed, Mulder [45].

2.4.2. Diffusion in Porous Media

Many different experimental methods have been developed in the past to study zeolitic diffusion, but the results were far from consistent At least part of the inconsistency among the diffusivities can be attributed to the differences in samples. Very few zeolite samples are perfectly free of crystal fault and mtergrowth. It has been demonstrated that diffusion may be retarded or accelerated by the existence of crystal fault and mtergrowth interface. Therefore, zeolite membranes prepared by different groups may be similar in adsorption behaviour, but their diffusion characteristics could vary greatly. The inconsistency of difmsivity measurements couJd also result from the fact that diffusivity is not a physical constant, but a phenomenological parameter. There are different ways to define the proportionality between flux and driving force, and even the choice of driving force itself is subjective, Chiang et al. [15]

Within a pore we may, in general, distinguish three fundamentally different types of diffusion mechanisms, Krishna et al. [6]

• Bulk, 'free space' or free molecular diffusion, Figures 2.3 and 2.4, that are significant for large pore sizes and high system pressures; here molecule-molecule collisions dominate over molecule-wall collisions.

• Knudsen diffusion, Figures 2.3 and 2.4, becomes predominant when the mean free path of the molecular species is much larger the pore diameter and hence molecule-wall collisions become important

• Surface diffusion, Figwe 2.4, of adsorbed molecular species along the pore wall surface; this mechanism of transport becomes dominant for micropores and for strongly adsorbed species.

(36)

Micropores have diameters smaller than 2 nm; macropores have sizes greater than 50 nro and mesopores are in the size range 2-50 nm. Bulk and Knudsen diffusion mechanisms occur together and it is prudent to take both mechanisms into account rather than assume that one or the other mechanism is 'controlling'. Surface diffusion occurs in parallel to the other two mechanisms and its contribution to the total species flux may be quite significant in many cases, Kiislina et al. [6]

In addition to the above three mass transfer mechanisms, ordinary molecular diffusion could also dominate, if the pores present tortuous paths and hinder the movement of large molecules when their diameter is more than 10 % of the pore diameter, Seader mid Henley [18]

Balk gas dotation

° o

o

QfX

o

(mean frae path) «

KpudsM diffusion

°/o

o

X (mean fr&e path) >> d

(37)

3ulk diffusion

Knudsen dWasion

Micropore diffusion

vacancies

Figure 2.4: Ttwee Distinct Mechanisms by which Molecular Species are Transported within an Adsorbent or Catalyst Particle, Krishna et al. [6]

2.4.2.1. The Dusty Gas Model

It is generally agreed that the most convenient approach to modelling combined bulk and Knudsen diffusion is the dusty gas model, Krishna et al. [6]. A simplified form of the dusty gas model, equation 2.5, can be used to derive important structural parameters of meso- or macroporous membranes. The permeability coefficient Bo, Knudsen coefficient Ko and the porosity/tortuosity ratio e/r can be derived from permeation measurements of a pure gas through the membrane, Neomagus et at. [35]

Narl\n{rs/rl)RT_BQ{P!+Ps) 4 r — An

A ? M

ISRT

-(2-5)

Na (mol/m2.s) is the molar flux of component a, r, and rs are the tube side and shell side radii of the membrane, P, and Ps are the tube side and shell side pressures, AP is the pressure difference across the membrane, u. is the viscosity and Ma is the molar weight of component a.

(38)

Bo and Ko can be determined, respectively, from the slope and intercept of the plot of the LHS of equation 2.5 versus tbe average pressure over the membrane. By assuming non-interconnected and uniform cylindrical pores both the pore radius and the porosity/tortuosity ratio can be estimated with: - = — 5 - (2.6) v 2B0 %B0 2K0 rP=—± (2-8) SIT

2.4.2.2. Diffusion within Micropores

Within micropores, surface forces are dominant and an adsorbed molecule never escapes from the force field of the surface even when at the centre of the pore. Steric effects are important and diffusion is an activated process, proceeding by a sequence of jumps between regions of low potential energy (sites). Since the diffusing molecules never escape from the force field of the pore wails, the fluid within the pore can be regarded as a 'single' adsorbed' phase, Krishna et al. [6]. Diffusion within this regime primarily occurs via the surface diffusion mechanism, Figure 2.5. This and other types of diffusion mechanisms that may control or jointly control micropore permeation are discussed below.

■ Surface diffusion:

> For this situation adsorbed molecules lose the gaseous character, and transport takes place by molecules jumping between minima in the potential energy imposed by the lattice, Van den Broeke et al. [29]

(39)

LTTEP A TUP E P EYIEW

7JT-C< lie Me$n!sr?me Technology

Cfospr r,r 2

> Diffusing gas molecules bound on the surface may escape from their adsorbed state to the gaseous state if their kinetic energy is larger than their sorption energy. Surface diffusion is only applicable in a relatively low temperature range and depends on the sorption energy of the gas molecules. The activation energy for surface diffusion is also lower than the sorption energy, Lee et al. [30]

Configurational diffusion:

> During diffusion in micropores, molecules can either retain a gaseous character (activated diffusion) or adsorb on the micropore surface, Vent de Graqf[48].

> For the latter, the expression for the diffusion coefficient has the same form as that of surface diffusion but the identities of the energy barriers are different in both cases. For surface diffusion, these are energy barriers on the internal pore surface while for configurational diffusion these are obstructions that are considered to be windows separating the voids in the crystalline structure, Bwggraaf[ll].

Gas translational diffusion or activated Knudsen diffusion:

> For this situation adsorbed molecules retain their gaseous character but the molecules, still, have to overcome the potential energy barriers imposed by the zeolite lattice. Therefore the diffusion is also activated, Van den Broeke et al. [29]

> The expression for gas translational diffusion is similar to the Knudsen equation, but in this model the gas molecules are considered to move between sorption sites (cages) in a translational mode by overcoming the obstructions formed by the small channels that connect adjacent adsorption sites. The gas translational activation energy is considered to be the difference between potential energy in the sorption sites and in the channels of the zeolites, Lee et al. [30]

Knudsen diffusion:

> In this regime the gas molecules pass through the pores undergoing random collisions with the pore walls. The Knudsen diffosivity is obtained from the kinetic gas velocity and the geometric parameters associated with the membrane. Gas transport occurs in the gaseous state without the involvement of adsorption, Lee et al. [30]

> Permeance of gases will be proportional to the inverse square root of the molecular weight and/or proportional to the inverse square root of the temperature, Algieri et al. [21]

(40)

LmgPATUREF33V3EW

Zso'ifr; h/iemhnmf, Tcc: m ~ingy

C'i,ip1er 2 0 / / " \

2 0''^^p^SS^^giiy

activation energy

I. I ■ o

activation energy

^\M\J

x

Figure 2.5: A Conceptual Model for Surface Diffusion of Adsorbed Species. D/vs and £>2v are the Maxwell-Stefan surface diffusivities of components 1 and 2. £>// represents the Maxwell-Stefan counter-sorption coefficient, Krishna et al. [6]

In membrane applications for gas separation, adsorption is usually not multilayer, and often well below a roonolayer, so its well described by the Langmuir adsorption model, Lee et al. [30]. In this region of strong adsorption and high occupancies, molecules have strong interactions with each other that will affect their diffusiona! behaviour, Bakker et al. [3]. At low pressure and high temperature, the Langmuir equation may be simplified to Henrys law. In the Henrys law regime the amount of adsorbed molecules increases linearly with the applied pressure, Lee et al. [30]

2.4.3. Controlling Mechanisms

Caro et al. [1] described the separation ability of a microporous membrane by the interplay of the mixture adsorption equilibrium and the mixture diffusion. Three limiting cases of separation mechanisms are discussed below.