UNIVERSITI TEKNOLOGI MALAYSIA

Size: px
Start display at page:

Download "UNIVERSITI TEKNOLOGI MALAYSIA"

Transcription

1 UNIVERSITI TEKNOLOGI MALAYSIA PSZ 19:16 (PIND. 1/97) BORANG PENGESAHAN STATUS TESIS JUDUL: EFFECT OF CO 2 -INDUCED PLASTICIZATION AND MEMBRANE CONFIGURATION ON THE PERFORMANCE OF ASYMMETRIC POLYSULFONE HOLLOW FIBER MEMBRANE FOR CO 2 /CH 4 SEPARATION Saya SESI PENGAJIAN: 2004/2005-II NURSHAHNAWAL BINTI YAACOB (HURUF BESAR) mengaku membenarkan tesis (PSM/Sarjana/Doktor Falsafah)* ini disimpan di Perpustakaan Universiti Teknologi Malaysia dengan syarat-syarat kegunaan seperti berikut : 1. Tesis adalah hakmilik Universiti Teknologi Malaysia. 2. Perpustakaan Universiti Teknologi Malaysia dibenarkan membuat salinan untuk tujuan pengajian sahaja. 3. Perpustakaan dibenarkan membuat salinan tesis ini sebagai bahan pertukaran di antara institusi pengajian tinggi. 4. ** Sila tandakan ( ) SULIT (Mengandungi maklumat yang berdarjah keselamatan atau kepentingan Malaysia seperti yang termaktub di dalam AKTA RAHSIA RASMI 1972) TERHAD (Mengandungi maklumat TERHAD yang telah ditentukan oleh organisasi/badan di mana penyelidikan dijalankan). TIDAK TERHAD Disahkan oleh (TANDATANGAN PENULIS) (TANDATANGAN PENYELIA) Alamat Tetap: 2458 J, Taman Lumba Kuda, Jalan Sultanah, 05250, Alor Setar, Kedah. Prof. Dr. Ahmad Fauzi bin Ismail Nama Penyelia Tarikh: 21 Jun 2005 Tarikh: 21 Jun 2005 CATATAN: * Potong yang tidak berkenaan. ** Jika tesis ini SULIT atau TERHAD, sila lampirkan surat daripada pihak berkuasa/organisasi berkenaan dengan menyatakan sekali sebab dan tempoh tesis ini perlu dikelaskan sebagai SULIT atau TERHAD. Tesis dimaksudkan sebagai tesis bagi Ijazah Doktor Falsafah dan Sarjana secara penyelidikan, atau disertai bagi pengajian secara kerja kursus dan penyelidikan atau Laporan Projek Sarjana Muda (PSM).

2 I hereby declare that I have read this thesis and to my opinion this thesis is sufficient in term of scope and quality for the award of the degree of Master of Engineering (Gas). Signature : Name of Supervisor : Prof. Dr. Ahmad Fauzi bin Ismail Date : 21 st June 2005

3 BAHAGIAN A - Pengesahan Kerjasama* Adalah disahkan bahawa projek penyelidikan tesis ini telah dilaksanakan melalui kerjasama antara dengan Disahkan oleh: Tandatangan : Nama : Jawatan : (Cop rasmi) Tarikh :.. * Jika penyeiaan tesis/projek melibatkan kerjasama BAHAGIAN B Untuk Kegunaan Pejabat Sekolah Pengajian Siswazah Tesis ini telah diperiksa dan diakui oleh: Nama dan Alamat : Prof. Madya Dr. Mohamed Kheireddine Aroua Pemeriksa Luar Jabatan Kejuruteraan Kimia Universiti Malaya Kuala Lumpur Nama dan Alamat : Prof. Madya Dr. Wan Aizan Wan Abdul Rahman Pemeriksa Dalam I Fak. Kej. Kimia dan Kej. Sumber Asli Universiti Teknologi Malaysia Skudai Pemeriksa Dalam II : Nama Penyelia Lain : (jika ada) Disahkan oleh Timbalan Pendaftar di SPS: Tandatangan : Nama : GANESAN A/L ANDIMUTHU Tarikh:

4 i EFFECT OF CO 2 -INDUCED PLASTICIZATION AND MEMBRANE CONFIGURATION ON THE PERFORMANCE OF ASYMMETRIC POLYSULFONE HOLLOW FIBER MEMBRANE FOR CO 2 /CH 4 GAS SEPARATION NURSHAHNAWAL BINTI YAACOB A thesis submitted in fulfilment of the requirements for the award of the degree of Master of Engineering (Gas) Faculty of Chemical and Natural Resources Engineering Universiti Teknologi Malaysia JUNE 2005

5 ii I declare that this thesis entitled Effect of CO 2 -Induced Plasticization and Membrane Configuration on the Performance of Asymmetric Polysulfone Hollow Fiber Membrane for CO 2 /CH 4 Separation is the result of my own research except as cited in the references. The thesis has not been accepted for any degree and is not concurrently submitted in candidature of any other degree. Signature : Name : Nurshahnawal binti Yaacob Date : 21 th June 2005

6 iii To my beloved parents (Encik Yaacob bin Awang Ahmad and Puan Kartini binti Roseley), My husband (Nur Shamriman bin Abdul Rahman) And my siblings (Yusnaini Yaacob and Muhammad Akmal Yaacob) Who gave me inspiration, encouragement and valuable support toward the success of this study.

7 iv ACKNOWLEDGEMENTS I would like to take this opportunity to express my sincere appreciation to people and organizations that have directly or indirectly given contributions toward the success of this academic study. First and foremost, I would like to give special thanks to my supervisor, Prof. Dr. Ahmad Fauzi bin Ismail for his keen effort, interest, advice, continuous guidance and insightful comments throughout the course of this research. I gratefully express my thanks to the members of Membrane Research Unit and FKKKSA workshop especially Mr. Mohamad Sohaimi bin Abdullah, Mr. Ng Be Cheer, Mr. Abdul Rahman bin hassan, Mrs. Rosmawati bt. Naim, Mrs. Nor Aida bt. Zubir, Ms. Suhana bt. Jalil, Mr. Mohd. Idham bin Mustaffar, Ms. Suhaina bt. Ibrahim, Mr. Wan Mohd. Hafiz, Mrs. Norida bt. Ridzuan, Mr. Syed Mohd Saufi bin Tuan Chik, Mrs. Ramlah bin Tajuddin, Mr. Zulkifle bin Nasir and Mr. Yahya bin Khalid, who have given me a substantial moral support to finish the study. Last but not least, my thanks goes to Universiti Teknologi Malaysia for granting me generous financial support under Industrial and Technology Development fellowship award (PTP-UTM), that enabling this work to be successfully completed. Above all, I thank Allah the Almighty for His grace, mercy and guidance throughout my life.

8 v ABSTRACT The present study focuses on the effect of CO 2 -plasticization and membrane configuration on the performance of asymmetric polysulfone hollow fiber membrane for CO 2 /CH 4 separation. Heat treatment method to suppress plasticization effect and membrane module configurations in series and cascades arrangement for the CO 2 /CH 4 gas separation was investigated. The membranes were prepared using polysulfone (Udel P1700) and tested using pure CO 2 and CH 4 and CO 2 /CH 4 gas mixture. Gas permeation experiments were conducted for single, two and three-stage configurations. The produced membranes were characterized by pure gas permeation experiments, density measurement, Differential Scanning Calorimetry (DSC), Scanning Electron Microscopy (SEM) and Thermogravimetric Analysis (TGA). In pure gas permeation experiment for both untreated and treated membranes, the pressure-normalized flux of CO 2 decreases with increasing of the membrane stages. In addition, the selectivities of the asymmetric polysulfone hollow fiber membrane showed a more constant trend with increasing feed pressure. Treated membrane exhibited lower pressure-normalized flux than untreated membranes due to skin layer densification which increases the gas transport resistance which lead to the reduction in the CO 2 pressure-normalized flux values. Among all configurations studied, twostage series configuration showed the most constant trend of selectivity values. The selectivity is slightly below the intrinsic selectivity. However, three-stage cascade configuration produced the highest CO 2 /CH 4 selectivity especially when tested at low feed pressure range. Some of the selectivity even surpasses the intrinsic selectivity of polysulfone. Effect of stage cut on feed pressure showed an increasing trend with increasing of CO 2 and CH 4 feed pressure in all configurations. This is due to the increase of the permeation driving force, which causes the passage of larger amounts of more permeable gas through the membrane. This study showed that, three-stage cascade configuration exhibited the smallest stage cut values than other module configurations. Hence, cascade configuration produces higher purity of CO 2 in the permeate stream. In mixed gas permeation experiment, increasing trend of CO 2 pressure-normalized flux was also observed but exhibited lower value due to competition among the penetrant species. As a result, the selectivity and the stage cut achieved are also lower in values. As a conclusion, the results of this work served as a platform in determining the most suitable module configuration to be used for gas separation processes.

9 vi ABSTRAK Fokus utama kajian ini adalah ke atas kesan pemplastikan teraruh CO 2 dan konfigurasi membran terhadap prestasi membran gentian geronggang asimetrik polisulfona bagi pemisahan gas CO 2 /CH 4. Kaedah rawatan pemanasan untuk merencat kesan pemplastikan dan modul konfigurasi membran secara bersiri dan menirus untuk pemisahan gas CO 2 /CH 4 turut dikaji. Membran disediakan menggunakan polisulfona (Udel P1700) dan diuji menggunakan gas tulen CO 2 dan CH 4 serta campuran gas CO 2 /CH 4. Ujikaji penelapan gas dijalankan untuk konfigurasi satu, dua dan tiga tahap. Membran yang dihasilkan diciri dengan ujian penelapan gas, pengukuran ketumpatan, Mikroskop Elektron Imbasan (SEM), Permeteran Kalori Pengimbasan Kebezaan (DSC) dan Analisis Termogravimetrik (TGA). Dalam ujian penelapan gas tulen bagi kedua-dua membran yang tidak dirawat dan yang dirawat, fluk tekanan ternormal CO 2 menurun dengan peningkatan bilangan membran. Tambahan pula, kememilihan membran gentian geronggang asimetrik polisulfona menunjukkan keadaan tetap dengan peningkatan tekanan masukan. Membran yang dirawat menunjukkan penurunan dalam fluk tekanan ternormal berbanding membran yang tidak dirawat disebabkan penebalan lapisan kulit membran yang meningkatkan rintangan pengangkutan gas yang membawa kepada penurunan nilai fluk tekanan ternormal CO 2. Di antara kesemua konfigurasi, konfigurasi dua tahap secara bersiri menunjukkan nilai kememilihan yang paling tetap. Kememilihan yang terhasil adalah sedikit rendah berbanding kememilihan intrinsik polisulfona. Walaubagaimanapun, konfigurasi tiga tahap secara menirus menunjukkan kememilihan CO 2 /CH 4 yang tertinggi terutamanya apabila diuji pada julat tekanan masukan yang rendah. Terdapat juga kememilihan yang mengatasi kememilihan intrinsik polisulfona. Kesan keratan aras ke atas tekanan masukan meningkat dengan peningkatan tekanan masukan CO 2 dan CH 4 dalam semua konfigurasi. Ini adalah disebabkan peningkatan daya peresapan yang menyebabkan sejumlah besar gas yang mudah meresap merentasi membran. Kajian ini menunjukkan yang konfigurasi tiga-tahap secara menirus menunjukkan nilai keratan aras yang terkecil berbanding konfigurasi yang lain. Oleh itu, konfigurasi menirus menghasilkan ketulenan CO 2 yang tinggi dalam aliran peresapan. Bagi ujikaji gas campuran, fluk tekanan ternormal CO 2 didapati meningkat tetapi menunjukkan nilai yang lebih rendah disebabkan saingan di antara kumpulan gas. Hasilnya, nilai kememilihan dan keratan aras yang terhasil juga adalah rendah. Kesimpulannya, keputusan ujikaji ini bertindak sebagai satu cara untuk menentukan konfigurasi modul yang paling sesuai untuk kegunaan proses pemisahan gas.

10 vii TABLE OF CONTENTS CHAPTER TITLE PAGE TITLE PAGE i DECLARATION ii DEDICATIONS iii ACKNOWLEDGEMENTS iv ABSTRACT v ABSTRAK vi TABLE OF CONTENTS vii LIST OF TABLES x LIST OF FIGURES xi LIST OF SYMBOLS xviii LIST OF APPENDICES xxi 1 INTRODUCTION 1.1 Membrane for Gas Separation Problem Statement Research Objectives Scope of Thesis 7 2 LITERATURE REVIEW 2.1 Membrane Process for Gas Separation Membrane Gas Separation System Advantages of Membrane Technology Configurations of Membrane Polymeric Membranes 19

11 viii 2.2 Gas Transport Mechanism in Polymers Solution-Diffusion Model Dual-Mode Sorption Model Molecular Model Asymmetric Membrane Formation Phase Inversion Process Dry/Wet Phase Inversion Dry Phase Separation Wet Phase Separation Gas Polymer Interactions CO 2 -Induced Plasticization on the Properties of 46 Glassy Polymers Plasticization in Dense Membranes Plasticization in Asymmetric Membrane Films Plasticization in Asymmetric Hollow Fiber 57 Membranes 2.6 Glass Transition Temperature, T g Suppression of CO 2 -Plasticization Cross-Linking Method Heat Treatment Method Blending Method 64 3 METHODOLOGY 3.1 Research Design Materials Selections Polysulfone Solvents Gases Fabrication of Asymmetric Hollow Fiber Membrane Turbidimetric Titration Method Preparation of Multi Component Polymer Dope Membrane Fabrication using Dry/Wet Spinning 72 Process

12 ix Hollow Fiber Coating Hollow Fiber Potting Characterization Pure Gas Permeation Measurements Membrane Configuration Use Network Model Heat Treatment Density Measurements Differential Scanning Calorimetry (DSC) Scanning Electron Microscopy (SEM) Thermogravimetric Analysis (TGA) 92 4 RESULT AND DISSCUSSION 4.1 Spinning Solution Formulation Pure Gas Carbon Dioxide and Methane Permeation 95 Behavior in Untreated and Treated Membranes 4.3 Mixed Gas Carbon Dioxide and Methane Permeation 111 Behavior in Untreated and Treated Membranes 4.4 Effect of Heat Treatment on Fiber Density Effect of Heat Treatment Process on the Polymer 138 Glass Transition Temperature 4.6 Morphology of the Develop Membrane Before and 140 After Heat Treatment Process 4.7 Effect of Membrane Weight Loss upon Heat Treatment 143 Process 5 CONCLUSIONS AND RECOMMENDATIONS 145 LIST OF PUBLICATIONS 148 REFFERENCES 149 Appendices A-G

13 x LIST OF TABLES TABLE NO. TITLE PAGE 1.1 Principal gas separation markets, producers and membrane systems Predicted sales of membrane gas separation in the main target market Characteristics of different module types Kinetics sieving dimensions of penetrants based on zeolite sorption cutoff Summary of types of heat treatment used by previous researchers Significant suppression methods for gas-plasticization Summary of spinning conditions Properties of carbon dioxide and methane Summary of the solution composition before and after titration Selectivity comparison of untreated asymmetric polysulfone hollow fibers in series and cascade configurations Fiber density of untreated and treated asymmetric polysulfone hollow fiber membrane

14 xi LIST OF FIGURES FIGURE NO. TITLE PAGE 2.1 Typical membrane process for gas separation Shell side feed hollow fiber module Spiral wound membrane module Schematic illustration of the structure of integrally skinned asymmetric membranes Schematic presentation of mechanism for permeation of gases through membranes Typical gas sorption isotherm forms for polymeric media Polymer specific volume as a function of temperature Solution-diffusion model Models for the transport of small penetrant molecules in polymers Schematic phase diagram for ternary consisting of polymer/solvent/nonsolvent Phase separation by instantaneous and delayed demixing of a polymer solution in a ternary system Schematic phase diagram for quaternary system consisting of polymer/solvent/nonsolvent/additive Correlation between the solubility parameter of glassy polymers and the ideal separation factors for CO 2 /CH 4 systems CO 2 permeation behavior in glassy polymer Steps of the research The chemical structure of polysulfone (PSF)

15 xii Apparatus for turbidimetric titration Apparatus for polymer dope preparation The laboratory scale for dry/wet spinning process Hollow fiber potting Schematic diagram of gas permeation testing system Typical schematic diagram of the membrane module configuration arrangement Single-stage module configuration Two stage module configuration in (a) series and (b) cascade configuration Three-stage module configuration in (a) series and (b) cascade configuration Pressure-normalized flux of CO 2 and CH 4 of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressures CO 2 /CH 4 selectivity of untreated and treated asymmetric hollow fiber membrane in single-stage configuration as a function of feed pressures Pressure-normalized flux of CO 2 and CH 4 of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressures CO 2 /CH 4 selectivity of the untreated and treated asymmetric hollow fiber membrane in two-stage series configuration as a function of feed pressures Pressure-normalized flux of CO 2 and CH 4 of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure Pressure-normalized flux of CO 2 and CH 4 of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascades configuration as a function of the feed pressure

16 xiii CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascades configuration as a function of the feed pressure Effect of stage cut on feed pressure in single-stage configuration of untreated and treated membrane (a) CO 2 and (b) CH 4 Effect of stage cut on feed pressure in two-stage series configuration untreated and treated membrane (a) CO 2 and (b) CH 4 Effect of stage cut on feed pressure in three-stage series configuration untreated and treated membrane (a) CO 2 and (b) CH 4 Effect of stage cut on feed pressure in three-stage cascade configuration untreated and treated membrane (a) CO 2 and (b) CH 4 Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 10 LPM Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 20 LPM Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 30 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 10 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 20 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 30 LPM

17 xiv Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in single-stage configuration of untreated and treated membrane at 10 LPM Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in single-stage configuration of untreated and treated membrane at 20 LPM Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in single-stage configuration of untreated and treated membrane at 30 LPM Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 10 LPM Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 20 LPM Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 30 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 10 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 20 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 30 LPM Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in two-stage series configuration of untreated and treated membrane at 10 LPM Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in two-stage series configuration of untreated and treated membrane at 20 LPM Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in two-stage series configuration of untreated and treated membrane at 30 LPM

18 xv Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 10 LPM Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 20 LPM Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 30 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 10 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 20 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 30 LPM Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in three-stage series configuration of untreated and treated membrane at 10 LPM Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in three-stage series configuration of untreated and treated membrane at 20 LPM Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in three-stage series configuration of untreated and treated membrane at 30 LPM Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascades configuration as a function of the feed pressure at 10 LPM Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascades configuration as a function of the feed pressure at 20 LPM

19 xvi Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascades configuration as a function of the feed pressure at 30 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascades configuration as a function of the feed pressure at 10 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascades configuration as a function of the feed pressure at 20 LPM CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascades configuration as a function of the feed pressure at 30 LPM Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in three-stage cascades configuration of untreated and treated membrane at 10 LPM Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in three-stage cascades configuration of untreated and treated membrane at 20 LPM Effect of CO 2 /CH 4 gas mixture stage cut on feed pressure in three-stage cascades configuration of untreated and treated membrane at 30 LPM Fiber density of the untreated and treated asymmetric polysulfone hollow fiber membrane Effect of heat treatment on the glass transition temperature of asymmetric polysulfone hollow fiber membrane Scanning electron microscopy pictures of asymmetric polysulfone hollow fiber membrane before and after different heat treatments. Left: partial cross section of membrane, right: skin surface (x250). (a1 and a2: untreated membrane, b1 and b2: treated membrane at 70 C for 2 min, c1 and c2: treated membrane at 70 C for 5 min) Scanning electron microscopy pictures of asymmetric polysulfone hollow fiber membrane before and after different heat treatments. Left: partial cross section of membrane, right: skin surface (x500). (d1 and d2: untreated membrane, e1 and e2: treated membrane at 70 C for 2 min, f1 and f2: treated membrane at 70 C for 5 min)

20 xvii 4.53 Thermogravimetric analysis of asymmetric polysulfone hollow fiber membranes at different heat treatments 144

21 xviii LIST OF SYMBOLS A - Membrane surface area (cm 2 ) A 1 - Effective surface area of dense skin region A 2 - Porous surface area b - Hole affinity constant, atm -1 C - Solubility, cc (STP)/ cc polymer C A - Concentration of component A C D - Henry s law isotherm C H - Langmuir s isotherm C H - Hole saturation constant, cc (STP)/cc polym d - Average pore diameter dµ i /dx - Gradient in chemical potential of component i D A - Diffusion coefficient J i - Flux of component i J vi - Volumetric flux of i across the membrane k D - Henry s law dissolution constant, cc (STP)/cc polym atm K i - Total effective pressure-normalized flux l - Thickness of active layer l r - Pore length L - Thickness of the selective layer L i - Coefficient of proportionality LPM - Liter per minute m - Mean hydraulic radius or mean pore size m 2 /m 3 - Surface to volume ratio unit M - Molar mass (g/mol) of the monomer unit M - Molecular weight (g mol -1 ) N - Avogadro constant p - Pressure, atm

22 xix p A - External partial pressure of A p f - CO 2 average partial pressure in the feed absolute pressures p r - CO 2 average partial pressure in the reject absolute pressures p if - Partial pressure of i on the feed side of the membrane p ip - Partial pressure of i on the permeate side of the membrane p ds - Downstream pressure of the gas p us - Upstream pressure of the gas p - Mean pressure P - Pressure-normalized flux of i in the membrane polymer P A - Pressure-normalized flux of gas A in a membrane material P 1 - Intrinsic pressure-normalized flux of the membrane material P 2 - Effective pressure-normalized flux characteristic of pore P/l i - Pressure-normalized flux Q - Volumetric flow rate (cm 3 (STP) s -1 or mol s -1 ) Q i - Total gas flux for permeant i Q p - Permeate flow rate (cm/s) s - Second S - Stage cut S A - Solubility coefficient R - Gas constant T - Temperature (K) T c - Critical temperature T g - Glass transition temperature v i - Mean molecular speed of permeant i V - Total molar volume of the monomer unit (cm 3 /mol) V e - Equilibrium volume of densified glass V g - Actual glassy specific volume V w - Van der Waals volume V 0 - Volume occupied by the chains (cm 3 /mol) wt% - Weight percentage w/w - Weight per weight x - CO 2 mole fractions in the retentate stream x f - CO 2 mole fractions in the feed streams

23 xx y - CO 2 mole fraction in the permeate stream α - Selectivity Å - Angstrom ºC - Degree Celsius C/min - Degree Celsius per minute δ - Numerical factor for a particular system δ - - Negative charge δ + - Positive charge ρ - Density of the film (g/cm 3 ) p - Pressure difference (bar or cmhg) γ i - Chemical potential γ i c i - Solvent activity µm - Micrometer ρ - Viscosity (N s m -2 ) τ - Turtousity factor

24 xxi LIST OF APPENDICES APPENDIX TITLE PAGE A Spinning Process Calculation 159 B Pressure-Normalized Flux and Selectivity of Pure Gas 160 C Stage Cut of Pure Gas 183 D Experimental Results of Mixed Gas 208 E Simulation Results of Mixed Gas 220 F Skin Layer Thickness and Density Measurements 232 G Differential Scanning Calorimeter 234

25 CHAPTER 1 INTRODUCTION 1.1 Membrane for Gas Separation Membrane separation processes is a well-established technology. The process essentially involves contacting one side of a semi-permeable gas separation membrane with a feed gas mixture containing at least the gas whose enrichment is desired, along with one or more other gases. The membranes divide a separation chamber into a high-pressure side into which the feed gas mixture is fed at a lowpressure side. A pressure differential is maintained across the membrane under conditions such that at least one of the gases in the feed gas mixture selectively permeates through the membrane from the high-pressure side to the low-pressure side of the membrane. Then the gas mixture which is relatively enriched in the first group of gases and depleted in the second different group of gases is removed from the low-pressure side of the membrane. While, gas depleted in the first group of gases is removed from the high-pressure side of the membrane (Stern et al., 1997). The application of membranes to gas separation problems has grown rapidly since the installation of the first industrial plants in the early 1980s. Membranes have gained an important place in chemical process industries and are used in a broad range of applications. The key property that is exploited is the ability of a membrane to control the permeation rate of a chemical species through the membrane. Removal of CO 2 is the only natural gas separation currently practiced on a large scale (more than 200 plants have been installed). Most were installed by Grace (now Kavanaugh-GMS), Separex (UOP) and Cynara. All of these plants used

26 2 cellulose acetate membranes in hollow fiber or spiral wound module form. More recently, hollow fiber polyaramide (Medal) and polyimide (Ube) membranes have been introduced due to their higher selectivity (Baker, 2000). A list of the principal gas separation markets, producers and membrane systems is given in Table 1.1. These market estimates are based on the new membrane equipment produced each year. Currently, only eight or nine polymer materials that are listed above are used to make at least 90% of the total installed gas separation membrane base. Plasticization problems, aging phenomena, low process ability and high cost are the main reasons why only eight to nine different polymers are used in 90% of the commercial applications (Barsema, 2003). The table shows that, to date, two thirds of the total gas separation market is in the separation of hydrogen from ammonia purge gas or syngas. These are clean gas streams, generally free of components that might foul or plasticize the membrane, which means that hollow fiber modules work well. However, the growing application areas are in natural gas treatment and in refining and petrochemical plants. The gas streams often contain high levels of plasticizing and condensable vapors, which degrade membrane performance. Table 1.1: Principal gas separation markets, producers and membrane systems (Baker, 2000) Company Principal membrane material used Module type Principal markets/ Estimated annual sales Permea (Air Products) Medal (Air Liquide) Generon (MG industries) IMS (Praxair) Polysulfone Polyimide/Polyaramide Tetra bromo polycarbonate Polyimide Hollow fiber Hollow fiber Hollow fiber Hollow fiber Large gas companies; N 2 /air at US$75 million per year; and hydrogen separation at US$25 million per year. Kvaerner Cellulose acetate Spiral wound Mostly natural gas Separex (UOP) Cellulose acetate Spiral wound separations at US$30 million Cynara (Natco) Cellulose acetate Hollow fiber per year. Parker-Hannifin Polyphenylene oxide Hollow fiber Vapor/gas separation, air Ube Polyimide Hollow fiber dehydration and other at GKSS Licensees Silicone rubber Plate and frame US$25 million per year MTR Silicone rubber Spiral wound

27 3 Therefore, robust membrane modules, which are able to handle upsets, are required (Baker, 2001). Developing processes such as CO 2 separation from natural gas, Volatile Organic Compound (VOC) separation from air and nitrogen and recovery of light hydrocarbons from refinery and petrochemical plant purge gases are performed on a commercial scale and in total several hundred plants have been installed. Some predictions of the future for the membrane gas separation market are given in Table 1.2. It seems that the total market will grow, but perhaps not uniformly in all the areas that are shown. Natural gas sales have reached about USD 30 million per year and should increase rapidly, perhaps reaching USD 90 million by CO 2 removal from natural gas has been practiced using cellulose acetate membranes for more than 10 years and the introduction of more selective polyimide membranes has begun and in time is likely to make membrane processes much more competitive with amine absorption. In the area of CO 2 /CH 4 separation membranes, Table 1.2: Predicted sales of membrane gas separation in the main target market (Baker, 2001) Separation Membrane market (USD million, 2,000 dollars) Nitrogen from air Oxygen from air Hydrogen Natural gas CO 2 NGL N 2 /H 2 O Vapor/nitrogen Vapor/vapor Other Total 75 < <

28 4 natural gas sweetening, the removal of CO 2 in landfill gas recovery processes and CO 2 removal from fractured wells as well as the removal of CO 2 in enhanced oil recovery applications (EOR) are of interest (Staudt-Bickel and Koros, 1999). CO 2 produced can be injected into adjacent oil fields to enhance oil recovery (Lonsdale, 1982). In order to achieve excellent performance in membrane processes, the process reliability needs to be enhanced to make CO 2 removal technology the ultimate choice in a variety of processing conditions. It is crucial to transport the field gas through the conventional pipeline without catastrophic corrosion problems. Hence, an efficient method to reduce the composition of CO 2 gas is critically in need and membrane gas separation processes was found to be the best solution. Nowadays, there are wide varieties of acid gas removal technologies available. Membrane separation processes have been shown to be very effective for natural gas processing. An efficient separation of CO 2 and hydrogen sulfide from natural gas can be achieved by selective permeation through polymer membranes (Lee et al., 1995). However, membrane technologies have been chosen for applications that have large flows, have high CO 2 contents or are in remote areas. The removal of CO 2 from off gas and reinjection into the oil field is desirable but the recycle gas must have a CO 2 purity of at least 95%. This minimum level is necessary in order to maintain the solvent power of the CO 2 (Dortmundt and Doshi, 1999). When the CO 2 content of the feed was above 75% CO 2, the separation could be achieved in a single membrane stage. In this case, the compressor was no longer needed (Ho and Sirkar, 1992). Even though the separation of CO 2 /CH 4 using polymeric membranes is growing rapidly, the plasticization of the membrane material is always a problem. This is due to the pressure-normalized flux of the slower gas which is facilitated by the highly soluble, faster gas. This phenomenon is attributed to plasticization effects caused by the high CO 2 solubility or interactions between CO 2 and the polymer material. As the membrane is plasticized the pressure-normalized flux increases significantly but the selectivity for gaseous mixtures decreases (Barsema et al., 2003). A good asymmetric membrane for natural gas separation can be achieved if it possesses the following material and performance characteristics: (1) inherently high

29 5 selectivity for CO 2 and CH 4 gas pair and (2) immunity to plasticization induced by CO 2. The CO 2 -induced plasticization usually causes a severe deterioration of membrane separation performance in the natural gas application loss (Cao et al., 2003). An understanding of the plasticization phenomenon is crucial to develop and achieve a high performance membrane in order to make membrane separation application attractive. Therefore, a thorough investigation of CO 2 -induced plasticization phenomenon must be carried out in order to reduce the extent of plasticization phenomena in glassy polymer membranes. 1.2 Problem Statement The problem encountered in the CO 2 /CH 4 separation was the swelling of the polymer matrix by the highly sorbed CO 2, which resulted in an increase in CO 2 pressure-normalized flux. This behavior is related to the so-called plasticization effects that occur during the separation process. This phenomenon took place since CO 2 that falls into the category of acid gas such as hydrogen sulfide (H 2 S) is commonly found in natural gas streams and hydrocarbon gases at levels as high as 40% to 60% and sometimes up to 80% as hydrocarbon capacities decline, and up to a feed pressure of 60 atm. These extreme operation conditions were the result of the swelling and plasticization of most membrane materials by the CO 2 present in the feed stream (Ho and Sirkar, 1992, Scott, 1998, Staudt-Bickel and Koros, 1999). In other words, degraded the membrane performance. As a result, pressure-normalized flux of CH 4 increases. As it increases more than the pressure-normalized flux of CO 2, the selectivity decreases (Bos et al., 1998). Since plasticization is a major problem that occurs in CO 2 /CH 4 separation, it is necessary to develop a membrane that has less plasticization effect besides maintaining the separation performance at elevated pressure conditions.

30 6 1.3 Research Objectives As stated above, the separation of CO 2 from CH 4 customarily takes place in the processes of natural gas treatment, enhanced oil recovery, landfill gas and also in digester gas upgrading and flue gas recovery in order to reduce pipeline corrosions induced by CO 2 as well as to produce high-purity energy products. Polymer membrane based technology is competitive for this kind of application in view of the following facts: (1) the high pressure of feed gas is a ready-made driving force for permeation and (2) CO 2 is more permeable than CH 4 in most membranes. Thus, the enriched CH 4 as the residual stream still retains at a high pressure for other operations without a significant pressure loss (Cao et al., 2003). Many polymers used for the CO 2 /CH 4 gas separation show the typical trend of a decreasing pressurenormalized flux with increasing pressure at low feed pressures and an increasing pressure-normalized flux as the CO 2 pressure is further elevated. The increase of pressure-normalized flux with increasing pressure is cause by plasticization (Bos et al., 1998). From the literature, few researchers reported on asymmetric membrane film, in addition very few researchers report on plasticization in asymmetric hollow fiber membranes. This phenomenon is mainly due to the fact that the dense selective layer of the hollow fibers is very thin and the inception of plasticization in the hollow fibers may occur at a very low feed pressure (Wang et al., 2002a). Many researchers studied polyimide membranes. However, polyimide trade off is not attractive to be used commercially and is expensive compared to polysulfone. On the other hand, the trade-off for polysulfone is favorable for commercial applications. This can help to reduce the cost of using membrane separation systems with condensable gases such as CO 2. Developing a better membrane that has less plasticization effect besides maintaining the membrane performance is essential for the future application of membrane based gas separation. Not much attention was given to overcoming the plasticization effect in hollow fiber membranes especially on polysulfone. As such, it is necessary to reduce the effect caused by the dissolved CO 2 that will alter the

31 7 polymer matrix, resulting in an increase in pressure-normalized fluxes and a reduction of selectivities. From the discussion above, the objectives of this research are: 1. To develop and characterize asymmetric polysulfone hollow fiber membranes. 2. To study the effect of plasticization in asymmetric polysulfone hollow fiber membranes for CO 2 /CH 4 gas separation systems. 3. To determine the optimize operating conditions for suppression of plasticization in hollow fiber membranes for CO 2 /CH 4 separation system. 1.4 Scope of Thesis In order to achieve the objective as stated above, the following scopes of work are identified: 1. Fabricating asymmetric polysulfone hollow fiber membrane and membrane modules for testing purposes. 2. Designing and fabricating a high-pressure three-stage gas permeation testing system in order to determine the separation performance of the asymmetric polysulfone hollow fiber membranes. 3. Investigating the plasticization effect using pure CO 2 and CH 4 as well as CO 2 /CH 4 mixture permeation experiments. 4. Performing a mild heat treatment process below polysulfone glass transition temperature in order to suppress CO 2 -plasticization. 5. Conducting membrane characterizations using density measurement, Scanning Electron Microscopy (SEM), Differential Scanning Calorimetry (DSC) and Thermogravimetric Analysis (TGA) that can provide an indirect evidence of plasticization.

32 CHAPTER 2 LITERATURE REVIEW 2.1 Membrane Process for Gas Separation The use of membrane separation processes is a well established technology. The process essentially involves contacting one side of a semi-permeable gas separation membrane with a feed gas mixture containing at least the gas whose enrichment is desired, along with one or more gases. The stream to be separated is fed to the membrane device at an elevated pressure, where it passes across one side of a membrane. The opposite side of the membrane is held at a lower pressure. The pressure difference across the membrane provides the driving force for the diffusion of gas across the membrane. Separation is achieved because of differences in the relative transport rates of the feed components. Components that diffuse more rapidly become enriched in the low-pressure permeate stream, while the slower components are concentrated in the retentate or residue stream. Separation is achieved only if the system is not maintained in a state of equilibrium. The typical membrane process for gas separation is shown in Figure 2.1. A membrane will separate gases only if some components pass through the membrane more rapidly than others. If the membrane contains pores large enough to allow a convective flow, separation will not occur. If the size of the pores is smaller than the mean free path of the gas molecules, then convective flow is replaced by Knudsen diffusion. Low molecular weight gases are able to diffuse more rapidly than heavier ones and separation occurs. If the pores are small enough, large molecules are unable to pass through them and are excluded by the membrane. This molecular sieving is potentially useful in separating molecules of different sizes (Ho and Sirkar, 1992).

33 9 Feed Permeate Note: Driving Force Fast Permeate Gas Slow Permeate Gas Figure 2.1: Typical membrane process for gas separation Membrane Gas Separation System Gas separation by polymer membranes is a proven technology that has found a wide range of applications. Flexible design, the compactness and the efficiency of membrane units compared to conventional gas separation methods like cryogenic distillations or absorption are particularly attractive (Staudt-Bickel and Koros, 1999). Membrane separation based on selective gas permeation competes directly with the four above-mentioned methods in many applications. Membranes offer versatility and simplicity in comparison to other methods, which must be balanced against limitations of medium purity and the need for recompression. Membrane units are also more cost effective than absorption for smaller applications and they can treat gas at the wellhead, which reduces safety problems and corrosion problems (Scott, 1998). The membrane used must be physically strong so that they can be incorporated into modules without being damaged. They must also be able to withstand the pressure differentials imposed on them during operation. A membrane module is designed to incorporate large amounts of membrane area per unit volume of pressure housing and to provide for effective contact of gas with the membrane surface. Provisions must be taken to minimize deleterious pressure drops within the module. Like the membranes, the module components need to be robust so that they can tolerate process pressures and temperatures and the materials of construction

34 10 must be chemically compatible with the process application. However, a membrane alone is not generally suitable when very high product purity and recovery is required. Membranes may be coupled with other unit operations to achieve an effective solution to such separation problems (Ho and Sirkar, 1992). In the area of CO 2 /CH 4 separation with membranes, the removal of CO 2 in landfill gas recovery processes and CO 2 removal from fractured wells as well as the removal of CO 2 in enhanced oil recovery applications (EOR) are of interest. In some of these separation applications, the membranes are exposed to high CO 2 concentrations in the feed stream. The resulting strong interaction between the CO 2 and the polymer material often affects flux and selectivity properties (Staudt-Bickel and Koros, 1999). A particularly important application of membrane technology is the use of spiral-wound modules to separate CH 4 from CO 2 /CH 4 mixtures encountered in natural gas treatment and enhanced oil recovery. Single-stage systems have low capital costs but they are appropriate only for moderate product purity and recovery requirements. Multiple separation stages and recycling are required for more demanding applications (Qi and Henson, 1998). Membrane separation processes offer many advantages, including a potential for high energy efficiency, ease of scale-up due to modular design, good weight and space efficiency and great operational flexibility for handling feed streams of varying compositions or flow rates. Membrane processes usually operate at ambient temperature, thereby avoiding the energy losses associated with heat exchange and are also environmentally safe. However, membrane processes are subject to the following limitations: (1) CH 4 losses could be higher than those in gas absorption processes (depending on the gas selectivity of the membranes), (2) presently-used membranes cannot economically reduce hydrogen sulfide (H 2 S) concentration in crude natural gas to the levels specified for pipeline quality gas ( 4 ppm H 2 S), and (3) scale-up is less economical than that of gas absorption processes (Bhide et al., 1998). The removal of CO 2 from CH 4 is very important in natural gas processing because CO 2 reduces energy content of natural gas. The world market for natural gas is estimated at approximately USD 22 billion per annual. In order to compete with traditional separation process such as amine scrubbing, it is becoming more and more

35 11 important to advance some new polymer membrane materials for the removal of CO 2 from natural gas (Ren et al., 2002). Removal of CO 2 is the only natural gas separation currently practiced on a large scale. There are two designs of typical CO 2 removal plants, which are usually used. For the single-stage plant, there are no rotating or moving parts, which leads to minimal maintenance and are preferred for small gas flows. It is also very competitive with other technologies especially if it is used for the low pressure permeate gas. In such a plant, CH 4 loss to the permeate side is often 10 to 15 percent. As the natural gas stream increases in size, the CH 4 loss from a single-stage system becomes prohibitive. Often the permeate gas is recompressed and passed through a second membrane stage. This second stage reduces the CH 4 loss to a few percent. However, the two-stage membrane processes are more expensive because a large compressor is required to compress the permeated gas and the membrane system may no longer compete with amine absorption, the conventional technology (Baker, 2000) Advantages of Membrane Technology Many polymers have been studied and polymer membrane technology appears as an attractive alternative for the separation of gases in industrial processes. However, only few polymers are used in membrane structures for commercial gas separation application. A polymer may have attractive intrinsic properties with respect to pressure-normalized flux and selectivity but may be difficult to form into defect-free asymmetric membranes (Barbari and Datwani, 1995). However, the main parameters of polymeric materials for successful gas separation membranes besides good film forming properties are the ability to form separation layers in membranes, the stability in contact with gas mixtures to be separated and reasonably low cost (Alentiev and Yampol skii, 2000). Membrane technology is attractive and preferred not only because of the compactness and its modularity but also because of the membrane separation processes which have low capital costs and easy scale-up due to modular design

36 12 when compared to the conventional separation processes (Koros, 1995, Kusakabe et al., 1996, Machado et al., 1999). According to Lim et al. (2000), separation of gases using polymeric membranes has become an important unit operation due to the discovery of economically competitive membranes with high pressure-normalized fluxes and good selectivities. This improved energy efficiency and simplicity of both operation and maintenance are the main advantages of this technique. Kaldis et al. (1998) also reported that membrane technology showed effective and reliable performance, which enable membrane technology to penetrate a wide variety of markets and applications. Bhide and Stern (1993a) reported that membrane processes offer other important advantages such as good space and weight efficiency, environmental safety and great flexibility for handling variations in flow rate, pressure and composition of feed gas streams. According to Ettouney et al. (1998), membrane technology has several advantages over conventional separation (cryogenic, solid adsorption, solvent absorption etc.). Membrane separation systems have a low capital cost, compact size and low specific power consumption, which reduce the production cost. In addition, membrane gas separation is a clean process and requires simple and inexpensive filtration. The processes are continuous and the membranes do not require regeneration, unlike adsorption or the absorption processes. Due to the enormous potential of membrane technology, continual efforts to develop and improve the economics of existing membrane processes for gas separation as well as to extend the range of applications of this technology is essential Configurations of Membrane A variety of membrane modules are available for separation processes. Table 2.1 summarizes the characteristics of different module types. The descriptions of the five most important modules are as follows.

37 13 Table 2.1: Characteristics of different module types (Franken, 1998) Module type Characteristics Surface to volume ratio [m 2 /m 3 ] Plate and frame Spiral wound Tubular Capillary Hollow fiber Flat sheet membranes Flat sheet Inner diameter > 5 mm 0.5mm < inner diameter < 5mm Inner diameter < 0.5 mm , ,000 4,000-30,000 Hollow fiber membranes consist of fibers installed as a bundle in a stainless steel tube. The shell-side hollow fiber module is shown in Figure 2.2. Two types of hollow fiber modules are used for gas separation applications. Shell-side feed modules are generally used for high-pressure applications up to 1,000 psig. Bore side feed modules are generally used for medium-pressure feed streams up to 150 psig, for which good flow control to minimize fouling and concentration polarization on the feed side of the membrane is desired (Baker, 2000). Hollow fiber permeators have approximately three times as much area per unit volume as the spiral wound units (Ho and Sirkar, 1992). Taveira et al. (2001) reported that hollow fiber membrane could pack larger quantities of membrane area into small volumes. Hollow fiber membranes have five major advantages over flat sheet membranes: (1) they have a much larger ratio of membrane area to unit volume (ratio as high as 10,000 m 2 /m 3 ) and hence higher productivity per unit volume of membrane module, (2) they are self-supporting with ease of manufacture and replacement, (3) they have good flexibility in operation (Chung et al., 2000c, Lim et al., 2000), (4) they do not require costly membrane supports, and (5) they are more damage resistant than other types modules (Stern et al., 1977).

38 14 High pressure gas outlet Fiber bundle plug Hollow fiber bundle Mixed gas feed stream Low pressure outlet Figure 2.2: Shell side feed hollow fiber module (Schendel, 1984) Spiral wound membrane modules are basically a rolled up version of the plate and frame system and is widely used in reverse osmosis. The feed flow channel, the membrane and the porous membrane support are rolled up and inserted into an outer tubular pressure vessel (Strathman, 1981). Feed passes axially down the module across the membrane envelope. A portion of the feed permeates into the membrane envelope, where it spirals toward the center and exits through the collection tube (Baker, 2000). The advantages of this system are easy and inexpensive to adjust hydrodynamics by changing the feed-spacer thickness to overcome concentration polarization and fouling. The disadvantages are rather low membrane area per volume, expensive for high pressures because extra high-pressure shells must be purchased and bypassing of feed may occur due to non-uniform wrapping of the module spiral (Koros, 1995). According to Schendel (1984), spiral wound membrane systems are to be oriented horizontally for ease of installation and replacement of membrane element. Figure 2.3 showed the spiral wound membrane module.

39 15 High pressure Gas mixture Residual gas Low pressure permeate gas Figure 2.3: Spiral wound membrane module (Schendel, 1984) Plate and frame modules were one of the earliest types of membrane systems. Membrane, feed spacers and product spacers are layered together between two end plates. The feed mixture is forced across the surface of the membrane. The filtrate permeates through the membrane, enters the permeate channel and makes its way to a central permeate collection manifold (Baker, 2000). The advantages of this system are the ease of cleaning and replacement of membranes while the disadvantage is low membrane area per volume (Koros, 1995). These systems were first used in largescale ultrafiltration and reverse osmosis units (Strathman, 1981) and this system was recently used in electrodialysis and pervaporation systems (Baker, 2000). The disadvantage of flat membranes is the low permeation rate, especially of highly selective membranes. Increasing the effective membrane area or reducing the membrane thickness can overcome this disadvantage. The increase in the packing density or reduction of the effective membrane thickness can be achieved by using a hollow fiber configuration (Tsai et al., 2002).

40 16 Capillary membrane module systems consist of a large number of membrane capillaries with a diameter in the range of 0.5 to 5 mm OD where the burst-strength of the fibers is limited to a few bars. Therefore the modules are used in applications with a low transmembrane pressure driving force, such as in (diffusion) dialysis or micro- and ultrafiltration (Franken, 1998). The feed solution is passed down the center of the capillary and the filtrates permeate its wall. The advantages of this system are good feed flow control and large membrane surface area per unit volume. The system disadvantages are limited operating pressure and the system is relatively sensitive to operating errors (Strathman, 1981). A tubular membrane module consists of a porous paper or fiberglass support with the membrane formed on the inside of the tubes (Baker, 2000). The pressurized feed solution flows down the tube bore and the product solution permeates the membrane and is collected in the outer shell. The tubes may be installed in series or parallel array. Tubular tube diameter varies from 5 to 15 mm OD. Due to the large diameter of the tube and the need to operate in cross flow, the energy requirement of tubular membranes is high compared to capillaries and hollow fibers (Franken, 1998). The advantages of this system are adjustable feed flow velocity over a wide range for control of concentration polarization effects and ease of mechanical cleaning when excessive membrane fouling occurs while the disadvantages are relatively high investment and operating costs and low ratio of membrane surface area to system volume (Strathman, 1981). Tubular modules are now limited to ultrafiltration application. Membrane separation systems have become viable alternatives to conventional gas separation technologies such as pressure swing adsorption, cryogenic distillation and amine absorption especially when high CO 2 concentrations are encountered (Scott, 1998). The demand for lower production cost and higher product purity motivates research on performance of various module configurations with combined serial and cascade arrangements. The economics of membrane separation processes depend critically on the process design. The design of a membrane system consists of two sub problems: (1) selection of an appropriate module configuration and (2) determination of the operating conditions of the

41 17 individual modules. Membrane systems currently are designed via a sequential procedure in which the module configuration is chosen a priori and the operating conditions are determined using some types of optimization procedure (Qi and Henson, 2000). An essential part in the design of membrane gas separation is the determination of the separation configuration. A single-stage arrangement with no recycle is the most common and simplest design form. However the demand for higher product purity and recovery ratio of the desired species necessitates the use of recycle streams as well as multi-stage configurations. Commonly, the multi-stage systems are designed using two, three or four-stages (Lababidi et al., 1996). Membrane has been the technology of interest in natural gas sweetening, removal of CO 2 in landfill gas recovery processes and CO 2 removal from fractured wells, and removal of CO 2 in enhanced oil recovery applications (EOR). With the progress in materials and membrane fabrication techniques, membrane system for applications in these areas became more competitive compared to conventional separation processes such as amine scrubbing and so forth. Many researchers had studied the effects of plasticization in commercial glassy polymers when exposed to high-pressure of CO 2. Glassy polymers such as polysulfones, polyimides, polycarbonates, poly(methylmethacrylate), polyurethane, polyaramide and cellulose acetate was found to plasticize when exposed to a high-pressure of CO 2. Due to the plasticizing effect, the sorption and permeation behavior of the polymer was affected. The polymer matrix was swollen due to the highly sorbed CO 2, which results in an increase in CO 2 pressure-normalized flux. Simultaneously, the CH 4 pressurenormalized flux increases and as it increases more than the CO 2 pressure-normalized flux, the selectivity decreases. This plasticizing action of CO 2 decreases the ability of the membrane to separate molecules on the basis of size, thereby causing the reduction in selectivity (Wonders and Paul, 1979, Donoue et al., 1989, Houde et al., 1992, Story and Koros, 1992, Briscoe and Kelly, 1995, Kapantaidakis et al., 1996, Bos et al., 1998, Staudt-Bickel and Koros, 1999, Chen et al., 2000, Koros and Woods, 2001, Wessling et al., 2001, Ismail and Lorna, 2002, Mikawa et al., 2002, Barsema et al., 2003, Chung et al., 2003, Kawakami et al., 2003).

42 18 Only few researchers studied the effects of membrane configurations and its effect on gas separation performance. Bhide and Stern (1991) studied the process configurations of single and two-stage for oxygen-enrichment of air. Six configurations were proposed both for series and cascades arrangements with permeate or retentate streams recycle was used in their experiment. The membranes used were silicone rubber, poly (phenylene oxide) and cellulose acetate. Later, Bhide and Stern (1993a; 1993b) studied the effect of module configuration for the removal of acid gases from natural gas. The seventh configurations, which consist of three-stage cascade configuration, were introduced. The membrane polymer used was cellulose acetate. Ettouney et al. (1998) investigated experimentally the separation characteristics of air using polysulfone hollow fiber membrane using one, two and three-modules in series. Each module has a total separation area of 2.22 m 2. As for Qi and Henson (2000), they developed spiral wound membrane network which considered two, three and four-stage of membrane configurations for the separation of CO 2 /CH 4. The total membrane area considered in their study ranging from 380 m 2 to 600 m 2. The developed system uses a random value of membrane area in order to recover the desired amount of CO 2. The cascade configuration used in their study was reported to produce 64.56% of CO 2 in the permeate stream. Ismail and Lim (2001) studied the same cascade configuration as well as series configuration through simulation works % of CO 2 in the permeate stream was achieved using three-stage cascade configuration. However, series configuration produced 82.48% of CO 2 in the permeate stream. The developed system was able to simulate several of membrane configurations in order to study the separation behavior of CO 2 /CH 4 and O 2 /N 2 mixture in polysulfone and cellulose acetate hollow fiber membranes. This configuration also provides the lowest operating cost among all other configurations studied.

43 Polymeric Membranes Polymeric membranes have been successfully used in many gas separation applications since polymeric membranes are permeable for all gases. Most of the important membrane separation processes for current industrial interest or scientific work are mainly focused on glassy type polymers because of their superior gas selectivities and excellent thermal and mechanical properties (Ismail and Lorna, 2003). Nowadays, membrane glassy polymers are used for gas separation rather than rubbery polymers because of the high selectivity achieved. Glassy polymer is defined as an amorphous polymeric material that is below its softening or glass transition temperature, T g while rubbery polymer is an amorphous polymeric material that is above its glass transition temperature, T g under the conditions used. Due to the more restricted segmental motions in glassy polymers, these materials offer enhanced mobility selectivity as compared to rubbery polymers. Since they are inherently more size and shape selective than rubbery materials, glassy polymers are more commonly used as the selective layer in gas separation membranes. Table 2.2 shows the diameters of several common gases (Ho and Sirkar, 1992). Asymmetric membrane refers to the structure in which a thin (approximately µm) dense skin layer (active layer) is integrally bonded in series with a thick (100 µm) porous substructure of the same material. The skin layer performs the separation with a high flux since it is thin and with high selectivity due to its high density. The porous substructure provides the mechanical strength for the complete membrane structure, has negligible resistance to mass transfer and prevent the mixing of local permeate fluxes of varying concentrations on the permeate side of the membrane skin surface, regardless of the flow direction of the bulk permeate stream outside the porous layer. In other words, the asymmetric membrane always gives rise to cross-flow permeation irrespective to the feed and bulk permeate flow pattern (Pan, 1983, Li, 1993, Vrentas and Vrentas, 2002). For an asymmetric hollow fiber membrane, the surface porosity and pore sizes on its selective layer play an important role to determine the mass transfer process across the membrane (Wang and Chung, 2001). Integrally skin and thin film composite are two general classes of asymmetric membranes that are universally recognized (Pandey and Chauhan, 2001).

44 20 Table 2.2: Kinetics sieving dimensions of penetrants based on zeolite sorption cutoff (Ho and Sirkar, 1992) Molecule He H 2 NO CO 2 Ar O 2 N 2 CO Kinetic diameter (Å) Molecule CH 4 C 2 H 4 C 3 H 8 n-c 4 CF 2 Cl 2 C 3 H 6 CF 4 i-c 4 Kinetic diameter (Å) Generally, asymmetric hollow fibers for gas separation have the dense skin on the outside. Therefore, in practice the high-pressure feed flows on the outside of the fiber (feed-outside mode). This approach is also followed for composite hollow fibers as in PRISM separators with silicone rubber coated asymmetric polysulfone hollow fibers (Sidhoum et al., 1988). Figure 2.4 shows a schematic illustration of the structure of integrally skinned asymmetric membranes. According to Chung et al. (2000a), for an asymmetric membrane to be useful for gas separation, its skin layer must be defect-free (surface defect is less than 10-6 ). This is to assure that permeation is exclusively controlled by a solution-diffusion mechanism to achieve the maximum selectivity and its skin layer must be as thin as possible to maximize the membrane productivity. Since the dense skin layer is formed by a phase inversion process which occurs by bringing an initially thermodynamically stable polymer solution to an unstable state during coagulation step, the complicated mass transfer and solvent exchange during the demixing generally yield defective skin layers. Thus, it is very difficult to simultaneously prepare a defect-free integral asymmetric hollow fiber membrane with an ultra-thin skin layer (thickness < 1000 Å). In the case of fluoropolyimide hollow fiber membranes, Chung et al. may be one of the pioneers applying the phase inversion technology to develop a defect-free 6FDA-durene asymmetric fiber with a skin layer of around 2000 Å to 3000 Å.

45 21 High pressure feed gas Dense nonporous active layer 1000 Å Porous sublayer ~ 0.2 mm Figure 2.4: Schematic illustration of the structure of integrally skinned asymmetric membranes (Schendel, 1984) Permeate gas Asymmetric membranes made from glassy polymers are widely used in industrial gas separations and have a very thin, dense skin and a thick, porous support. The thickness, the porosity and pore size of the dense skin control pressurenormalized flux and selectivity of asymmetric membranes at a given operating pressure and temperature. The membranes suitable for gas separation must have small enough porosity and pore size so as to increase membrane selectivity and a thin skin layer for improved gas permeation flux. In addition, the substructure of the asymmetric membrane should have good mechanical strength with negligible gas transport resistance. Asymmetric membranes may be classified to surface defects into three groups as discussed below: (1) the skin layer has a large fraction of defects. The transport behavior of gases through this group of asymmetric membranes is similar to those in porous membranes, which is controlled predominantly by pore flow (2) the skin layer is defect-free. Gas transport in this group of asymmetric membrane is similar to that in the dense polymer film, which can be described by the solution-diffusion mechanism (3) the skin layer has a small fraction of defects. The overall pressure-normalized flux for this type of membrane would be the combination of the pore flow and solution-diffusion flow.

46 22 A porous membrane is a rigid, highly voided structure with randomly distributed inter-connected pores. Figure 2.5 shows the schematic presentation of mechanism for permeation of gases through membranes. The separation of materials by porous membranes is mainly a function of the permeate character and membrane properties like the molecular size of the membrane polymer, pore size and pore size distribution. A porous membrane is very similar in its structure and function to a conventional filter. In general, only microporous membranes can separate those molecules that differ considerably in size effectively. Porous membranes for gas separation can exhibit very high levels of flux but provide for low separation or selectivity. Porous Membranes Convective flow Knudsen diffusion Molecular sieving (surface diffusion) Dense Solution-diffusion Figure 2.5: Schematic presentation of mechanism for permeation of gases through membranes (Pandey and Chauhan, 2001)

47 23 Microporous membranes are characterized by (1) the average pore diameter, d (2) the membrane, є (the fraction of the total membrane volume that is porous) and (3) the tortousity of the membrane. There are several ways to prepare porous polymeric membranes, such as solution casting, sintering, stretching, track etching and phase separation. The final morphology of the membrane obtained will vary greatly, depending on the properties of the materials and the process conditions utilized (Pandey and Chauhan, 2001). Porous polymeric membranes usually handle aqueous feed streams. Only a limited number of organic liquids can be processed with conventional polymeric membranes and this limitation has impeded the broad application of membranes across the chemical process industries (Koros, 1995). Nonporous, dense membranes consist of a dense film through which permeants are transported by diffusion under the driving force of a pressure, concentration or electrical potential gradient. There is no transport in pores or channels. Thus, nonporous, dense membranes can separate permeants of similar size if their concentration in the membrane material (i.e., their solubility) differs significantly (Murphy and de Pinho, 1995). According to Chung et al. (2000b), there are differences in gas pressure-normalized flux and selectivity between a thin (0.5 µm) and a thick (2.5 µm) dense membrane. The thin dense membrane has a higher selectivity and activation energy, but with a lower pressure-normalized flux than that of the thick membrane. Most gas separation membranes, pervaporation and reverse osmosis membranes use dense membranes to perform the separation. Usually these membranes have an asymmetric structure to improve the flux (Baker, 2000). The important feature of dense membranes that are used in separation applications is the ability to control the permeation of different species. In the solution-diffusion mechanism, the permeants dissolve in the membrane material and then diffuse through the membrane down a concentration gradient. A separation is achieved between the different permeants because of differences in the amount of material diffusing through the membranes (Pandey and Chauhan, 2001). The conventional symmetric membrane has a homogeneous structure with uniform permeation properties across its entire thickness where gas separation performances are strongly dependent on the feed and permeate flow pattern (cross-

48 24 flow, counter current or cocurrent). This type of membrane has not been widely used for gas separation mainly due to low rates of permeation imposed by the membrane thickness required for maintaining membrane integrity and strength (Pan, 1983, Pan, 1986). According to Soria (1995), glass membranes with a symmetric structure (isotropic sponge) of interconnected pores can be prepared by phase separation and leaching techniques. The main disadvantage of porous glass membranes is the instability of its surface. The pore surface is rather active mainly because of the presence of silanol groups. These groups can be modified spontaneously by feed or voluntarily by a chemical or physical action, with consequent drastic changes of the membrane characteristics. 2.2 Gas Transport Mechanism in Polymers Glassy polymers are able to discriminate effectively between extremely small differences in molecular dimensions of common gases (e.g., 0.2 Å to 0.5 Å). Solubility often dominates diffusional characteristics of transport in rubbery polymers. Meanwhile, transport in glassy polymers is most often governed by penetrant size. Gas sorption in glassy polymers is more complex and a typical sorption isotherm is presented in Figure 2.6. The physical characteristic of glassy polymers that is commonly linked to the complex sorption isotherm is the unrelaxed volume locked into these materials when they are quenched below the glass transition temperature, T g. Figure 2.7 shows the break in the volume versus temperature plot for a polymer as the temperature is lowered below T g. Below T g, the excess or the unrelaxed volume is V g V l. This excess volume is thought to be the result of trapped non-equilibrium chain conformations in quenched glasses, which result from the extraordinarily long relaxation times for segmental motions in the glassy state. The excess volume present in glassy polymers allows for the accommodation of additional penetrant above that observed in low molecular weight liquids and rubbers (Ho and Sirkar, 1992).

49 25 Henry s Law (a) Flory-Huggins BET III (b) Dual-Mode c D = + c H c H k D c, concentration in polymer (cc(stp)/cc polymer) p, partial pressure of penetrant, atm Figure 2.6: Typical gas sorption isotherm forms for polymeric media (Ho and Sirkar, 1992) GLASS RUBBER V g Actual glassy specific volume V(cc/gm) V e Equilibrium volume of densified glass T g Figure 2.7: Polymer specific volume as a function of temperature (Ho and Sirkar, 1992) Temperature

50 26 Gas permeation through polymeric membranes for nonporous membranes follows the solution-diffusion mechanism (Li, 1993, Dhingra and Marand, 1998). The movement of any species across the membrane is caused by one or more driving force (Ho and Sirkar, 1992). Dense skin membranes are formed by a delayed demixing process (Machado et al., 1999) and the transport through dense polymeric materials is modeled by considering that the diffusive flux through the polymer is Fickian and driven by the concentration gradient or chemical gradient across the polymer (Thundyil et al., 1999). The transport of a component across a membrane occurs because of a driving force, which can be one or a combination of the following: hydrostatic pressure gradient, osmotic pressure gradient, concentration gradient, electrical potential gradient and temperature gradient (Paul and Sikdar, 1998). According to Chung et al. (2000a), the selectivities for uncoated fibers indicate that gases transport through the membranes by means of a combination of Knudsen and solution-diffusions. It was reported that for asymmetric membranes, the overall transport rates have been observed to be significantly affected by the support layer. The transport of a pure component across the active layer of the asymmetric membrane can be described by the following equation, J = P( p p ) l (2.1) v i i f i / p where, J vi is the volumetric flux of i across the membrane, P is the pressurenormalized flux of i in the membrane polymer, l is the thickness of active layer, pi and f pi p are the partial pressure of i on the feed and permeate side of the membrane respectively. For commercial membranes, l is typically of the order of 1000 Å to 2000 Å, with 400 Å to 1000 Å attainable in more finely tuned membranes. If the skin layer is defect-free and the substructure resistance to gas transport is negligible, the selectivity of asymmetric membrane will be equal to intrinsic selectivity of the membrane material. A defect-free skin layer guarantees that gas

51 27 transport is determine by a solution-diffusion mechanism (Pinnau and Koros, 1991). However, if a coating material with a very high pressure-normalized flux is used, the overall transport through the membrane is hardly affected (Ho and Sirkar, 1992). For properly formed asymmetric membranes, all of the transport resistance lies in this thin selective region; so the support layer is invisible to the transport phenomena. The pressure-normalized flux P A of gas A in a membrane material simply equals the pressure-and-thickness-normalized flux. The flux is inversely proportional to the thickness of the selective layer L: [flux of A] = p A PA / L (2.2) So, thin-skinned asymmetric structures are preferred to maximize productivity. The pressure-normalized flux is a product of a thermodynamic factor, called the solubility coefficient S A and a kinetic parameter, called the diffusion coefficient D A. P = S ][ D ] (2.3) A [ A A The solubility coefficient S A equals the ratio of the concentration C A of component A able to sorb in the membrane to p A the external partial pressure of A: S = C / p (2.4) A A A The diffusion coefficient characterizes how rapidly component A can diffuse once it has been dissolved in the matrix. The temperature fixes the thermal energy of the membrane. This energy tends to reduce interactions of the sorbed penetrants with the matrix and activates the diffusion jumps; it therefore affects all of the coefficients in Equation 2.3 in a complex manner (Koros, 1995). Wang and Chung (2001) suggested a few assumptions to the gas transport mechanism to be employed in the mass transport analysis for gas diffusion through the dense skin for asymmetric membrane with a defective dense selective skin. (1)

52 28 Gas is only single dimensional. This assumption is valid for most uncoated as spun asymmetric hollow fibers. (2) The dense skin layer of the asymmetric hollow fiber membrane has a uniform thickness. (3) The pores on the skin layer are cylindrical in shape and have a length equivalent to the thickness of the dense layer. (4) The resistance of the porous substructure is minimal compared to that of the dense skin. (5) Solution-diffusion mechanisms occur at the defect-free dense skin and the gas pressure-normalized flux is independent of pressure. Porous membranes made from the basic polymer as well as from the substituted types can be applied to ultrafiltration and microfiltration (Rodemann and Staude, 1994) and is formed by an instantaneous demixing process (Machado et al., 1999). The study of gas permeation through finely microporous membranes has a long history dating back to Graham s work in the 1850s. If the pores of a microporous membrane are 0.1 µm to 10 µm, gases permeate the membrane by normal convective flows described by Poiseuille s law and no separation occurs. If the pores are smaller than 0.1 µm and the pore diameters are of the same size or smaller than the mean free path of the gas molecules, then the diffusion is govern by Knudsen diffusion. If the membrane pores are extremely small (5 Å to 20 Å), the gases are then separated by molecular sieving (Baker, 2000). For porous membranes, the mechanisms of gas separation described by Rao and Sircar hold good for inorganic membranes. Five different mechanisms may be involved in the transport of gases across a porous membrane: Knudsen diffusion, surface diffusion, capillary condensation, laminar flow and molecular sieving. The relative contributions of the different mechanisms are dependent on the properties of the membranes and the gases, as well as on the operating conditions like temperature and pressure. In a commercial ceramic membrane with pore sizes greater than 4 nm, Knudsen diffusion is likely to be the dominant mechanism of gas transport at low pressures and elevated temperatures. Capillary condensation and surface diffusion are unlikely to exist at elevated temperatures in the membranes with pore sizes in the range of 2 nm. Molecular sieving does not take place since the pore sizes are much larger than the gas molecules. The contribution of viscous flow, resulting from a pressure difference across the pores will be quite small and even if it is present, it

53 29 does not contribute to the separation process. This leaves Knudsen diffusion as the only transport mechanism contributing to the separation of various components in a gaseous mixture at elevated temperatures in a porous membrane (Pandey and Chauhan, 2001) Solution-Diffusion Model In the 1940s, the solution-diffusion model was used to explain the transport of gases across polymeric films where the permeants dissolve in the membrane material and then diffuse through the membrane down a concentration gradient. The permeants are separated because of the differences in the solubilities of the materials in the membrane and the differences in the rates at which the materials diffuse through the membrane. Diffusion, the basis of the solution-diffusion model, is the process by which matter is transported from one part of a system to another by a concentration gradient. Thus, the flux, J i, of a component i, is described by the simple equation (Wijmans and Baker, 1995, Baker, 2000) J i = Li ( dµ i / dx) (2.5) where dµ i / dx is the gradient in chemical potential of component i and L i is a coefficient of proportionality linking this chemical potential driving force with the flux. The solution-diffusion mechanism is considered to consist of three steps: (1) the absorption or adsorption at the upstream boundary, (2) activated diffusion (solubility) through the membrane, and (3) desorption or evaporation on the other side. This solution-diffusion mechanism is driven by a difference in the thermodynamic activities existing at the upstream and downstream faces of the membrane as well as the interacting force working between the molecules that constitute the membrane material and the permeate molecules. The activity

54 30 difference causes a concentration difference that leads to diffusing in the direction of decreasing activity. The solution-diffusion model assumes that the pressure within a membrane is uniform and that the chemical potential gradient across the membrane is expressed only as a concentration gradient (Pandey and Chauhan, 2001). The morphologies of most membranes based on solution-diffusion separations are asymmetric, with a thin, highly selective polymer layer on top of an open support structure. To the unaided eye, pores are not apparent in either microporous or solution diffusion membranes. Under high (10,000x-100,000x) magnification, pores become apparent in the top surfaces of filtration membranes, while the surfaces of solution-diffusion membranes still appear dense. Permeating components diffuse through this essentially dense homogeneous selective layer and then pass through the underlying support structure with minimum resistance. The transmission rates through the selective layer depend upon the molecular volume and amount of the penetrant that can sorb molecularly into the selective layer (Koros, 1995). According to Chung et al. (2000a), to assure that permeation in asymmetric membranes for gas separation is exclusively controlled by a solution-diffusion mechanism, the skin layer must be almost defect-free (surface defect is less than 10-6 ). In order to achieve the maximum selectivity, its skin layer must be as thin as possible to maximize the membrane productivity. As shown in Figure 2.8, the solution-diffusion model assumes that when a pressure is applied across a dense membrane, the pressure everywhere within the membrane is constant at the highpressure value. This assumes, in effect, that solution-diffusion membranes transmit pressure in the same way as liquids. Consequently, the pressure difference across the membrane is expressed as a concentration gradient within the membrane (Wijmans and Baker, 1995).

55 31 High-pressure solution Membrane Low-pressure solution Chemical potential, µ i Pressure, p Solvent activity, γ i c i Figure 2.8: Solution-diffusion model (Wijmans and Baker, 1995) Dual-Mode Sorption Model The dual-mode sorption model poses the most useful phenomenological description of gas transport in glassy polymers, particularly by its partial immobilization version developed by Paul and Koros and by Petropoulos. This model satisfactorily represents the dependence of solubility, diffusion and pressurenormalized flux coefficients on the penetrant gas pressures (or the concentration within the polymers) over a wide range of these variables. The validity of this model has been confirmed for a large number of gas/polymer systems by the studies of Paul and Koros and also from the contributions of many other investigators. This model is also applicable to the permeation of gas mixtures. In the dual-mode model of Paul and Koros, transport in the two types of states is postulated and different values are assigned to the four interstate diffusivities. This model satisfactorily represents the dependence of solubility, diffusion and pressure-normalized flux coefficients on the penetrant gas pressures or concentrations in the polymer. The dual-mode sorption model is strictly applicable in cases where the polymer is not significantly plasticized (swelled) by the penetrant gas. The model can be extended to the transport of plasticizing penetrants, for example, by taking into account the concentration dependence of the mutual diffusion coefficient, which

56 32 characterizes the two transport modes postulated by the model. Such modifications also require the use of additional parameters which are made necessary by the increased complexity of the transport mechanisms involved. The main limitation of phenomenological models is that they are not predictive because the model parameter is not directly related to the chemical structure of the polymers (Pandey and Chauhan, 2001). Glassy polymers typically have non-linear sorption isotherms that can be interpreted in terms of dual-mode sorption mechanisms. The penetrant molecules are visualized as sorbing into a glassy polymeric membrane following two distinct mechanisms: Henry s law dissolution and Langmuir type sorption. The two populations are believed to be connected by local equilibrium. The first refers to dissolution of the penetrant in the dense polymeric matrix and the second refers to sorption of the penetrant in the microvoids of the polymer where chain mobility and packing are highly restricted. For multicomponent sorption, competitive sorption can result in the depression of sorption of a given component due to the presence of other mixture components. Typically, glassy polymers are favored over rubbery polymers for gas separation applications such as those noted above where size and shape discrimination is a key factor in achieving good selectivity. Glassy polymers typically have an abnormally high solubility for inert gases and that the isotherms for these systems are highly non-linear (Vieth et al., 1976, Thundyil et al., 1999) but can be decomposed into a combination of a Henry s Law part and Langmuir s part. CD = k D p Henry s law isotherm C H CH ' bp = 1 + bp Langmuir s isotherm C = C D + C H = k D CH ' bp p + 1+ bp Dual Sorption Theory

57 33 where C is the solubility, cc (STP)/cc polymer, k D is the Henry s law dissolution constant, cc (STP)/cc polym atm, b is the hole affinity constant, atm -1, C H is the hole saturation constant, cc (STP)/cc polym and p is the pressure, atm. The solubility isotherm of a gas dissolved in a glassy polymer can be expressed as the sum of the Henry s law term, representing the ordinary dissolution and a Langmuir sorption term, representing the hole-filling process. According to the dual-sorption model, gas solution by the Langmuir mode involves only a filling of pre-existing microcavities or packets of free volume, whereas in the ordinary dissolution mode, work must be done in separating the polymer chains to accommodate the solute molecules (Stern and Kulkarni, 1982). The non-linear sorption isotherm can be interpreted in terms of a dual-mode sorption mechanism. Transport through glassy polymers can also be explained well in terms of a dualmode transport model that is consistent with the dual sorption model. Temperature can have a large effect on the transport rate of small penetrants in polymeric media. The lower activation energy for the glassy state is thought to be due to the abrupt change in mobility for the polymer matrix as the glass transition temperature, T g is transverse. Presumably, in the much less mobile environment of the glassy polymer, smaller segments are involved in the diffusion process. Since these smaller segments govern penetrant diffusion jumps, more subtle discrimination in penetrant sizes is obtained, which leads to higher selectivities. The solubility coefficient is much more strongly dependent on temperatures below T g. Below T g, the sorption level is thought to be strongly affected by sorption into the Langmuirlike environment. The Langmuir capacity has been found to be strongly dependent on temperature and is generally predicted to disappear at the moment of glass transition (Ho and Sirkar, 1992). According to Dhingra and Marand (1998), the gas transport and separation property studies have been especially focused on glassy polymers because of their superior gas selectivity in comparison to rubbery polymers. The presence of unrelaxed free volumes in glassy polymers plays an important role in controlling the

58 34 gas permeation rate and thereby, the separation of two or more components. The dual-mode model can been used to predict the feed pressure dependence of penetrant pressure-normalized fluxes for pure gas transport and mixed gas transport in glassy polymers (Thundyil et al., 1999). The dual-mode transport model can also be used to predict in most cases with accuracy, the multicomponent permeation across glassy polymeric membranes. From the study, the dual-mode transport model predicts that pressure-normalized flux decreases with retentate and permeate partial pressure. The dual-mode transport model also predicts that a lower pressure-normalized flux for the slower component is due to competition effects (Taveira et al., 2001) Molecular Model Molecular theories, including those of Brandt and DiBendetto, attempt to analyze the diffusion process in terms of specific postulated motions of the polymer chain relative to each other and the motion of penerant molecules. Figure 2.9 gives a schematic representation of the various models proposed for polymer microstructure for the transportation of small penetrant molecules through the matrix. Figure 2.9 (a) shows a bundle of parallel polymer chains and inclusion of gas molecules. In order to move into the polymer matrix, the gas molecules push the polymer chains and jumps into a new position. Figure 2.9 (b) shows polymer penetrant segments in a normal and activated state. In the activated state, the polymer chain accommodates a diffusing molecule and allows it to diffuse and then returns to the normal configuration after the jump. Normal and activated stages are also represented. Figure 2.9 (c) shows the model proposed by Pace and Datyner. The model proposed by Pace and Datyner accounts for the structure of the polymer contributing to gas diffusion and incorporates some of the features of previous models also. It assumes that non-crystalline polymer regions posses an appropriate semi-crystalline order with chain bundles that are parallel along distances of several nanometers and can be considered as tubules. The tubules consisting of parallel chains facilitate the movement of the penetrant and the transport occurs by leaps between these tubules. These jumps occur when the thermal motions of local segments of the polymer chain open up a sufficiently large channel to a neighboring gap. The gas particles can then

59 35 diffuse through this channel. Once the channel closes, the jump is successfully concluded. According to this model, the selectivity of a membrane material depends on the control of these leap channels. Large openings or high flexibilities cause large diffusion coefficients and low apparent energies of activation for diffusion, whereas more limited motions permit the passage of smaller species much more readily than the large particles (Pandey and Chauhan, 2001). (a) Polymer chains Penetrant molecule AFTER BRANT Bundle of paradel polymer segments Penetrant molecule NORMAL STATE ACTIVATED STATE NORMAL STATE AFTER ONE DIFFUSIONAL JUMP (b) Polymer segment Penetrant molecule NORMAL STATE ACTIVATED STATE (c) AFTER PACE AND DATYNER Figure 2.9: Models for the transport of small penetrant molecules in polymers (Pandey and Chauhan, 2001)

60 Asymmetric Membrane Formation Phase diagrams have been used to predict whether a solution of a certain polymer in a particular solvent is suitable for membrane formation or not. Binary phase diagrams show the phase boundaries as a function of temperature and composition as well as providing information on the thermally induced phase separation process. Thermal-isothermal phase diagrams are useful for the prediction of the phase transition that could occur when phase separation is induced according to the air casting of a polymer solution, preparation from vapor phase and immersion precipitation. These phase diagrams provide information on the phase transitions, a polymer solution can undergo during the formation of a porous membrane. An equilibrium phase diagram provides a map for the different phase transitions that are favored thermodynamically. The kinetics of the phase separation process determines whether the thermodynamically favored transition will occur or not (Pandey and Chauhan, 2001) Phase Inversion Process In the preparation of synthetic polymeric membranes, one of the popular methods is the phase inversion process (Han and Bhattacharyya, 1995). Phase inversion processes were used in the late 1800s. However, the physical processes involved were not directly identified (Paul and Yampol skii, 1994). The phase inversion process was developed by Loeb and Sourirajan 40 years ago and is still one of the most important means of preparing asymmetric membranes. The morphological change during the membrane formation is due to a combination of nucleation growth and spinodal decomposition (Chung et al., 2001). The phase inversion process involves a phase separation of a polymer solution in polymer rich and lean phases, which can be achieved by the immersion-precipitation technique. A polymer solution is immersed in a nonsolvent bath for the polymer, where a mass

61 37 transfer process involving interchange of solvent/nonsolvent occurs. The simplest system is formed by three components, polymer/solvent/nonsolvent, and is described by the ternary diagram shown in Figure The binodal curve delimitates the two-phase region, rich and lean polymer phases which have their compositions given by the tie lines. The spinodal curve represents the line where all possible fluctuations lead to instability. The region between binodal and spinodal correspond to metastable compositions where phase separation by nucleation and growth takes place. If the precipitation path crosses the binodal below the critical point, nucleation of a polymer rich phase may initiate the phase separation process. On the other hand, if the precipitation path crosses the binodal above the critical point, nucleation of the polymer lean phase may occur. At high polymer concentrations, phenomena such as vitrification, gelation or crystallization can occur in the polymer solution interrupting the polymer lean phase growth. These phenomena in a region of very high viscosity are not very well identified or predicted. Membranes formed by instantaneous demixing have a porous top layer and are used in microfiltration and ultrafiltration processes. Membranes formed by delayed demixing have a dense skin and are appropriate for use in gas separation, pervaporation and reverse osmosis (Machado et al., 1999). Polymer Crystallization, gelation or vitrification Binodal One phase Spinodal C Two phase Solvent Critical point Nonsolvent Figure 2.10: Schematic phase diagram for ternary consisting of polymer/solvent/ nonsolvent (Machado et al., 1999)

62 38 The resultant membranes have a skin layer, which are integrally bonded in series with a thick porous substructure. The skin and the substructure are composed of the same material. The skin layer, which contains the effective separating layer, is one of the key elements in determining the membrane pressure-normalized flux and selectivity (Chung et al., 2001). Many membranologists agree, however, that there are two dominating factors controlling the formation of phase inversion membranes: thermodynamics and kinetics correlate each other in a system during the solidification of membrane solution. The former is related to the phase equilibria between components in the system and the latter with the mutual diffusivities between them (Han and Nam, 2002). During the immersing process, the solution is cast or extruded into a quench medium, which consist of nonsolvent for the polymer. Nonsolvents diffuse into the nascent membrane, causing spinodal decomposition of the polymer solution into a polymer-rich solid phase, which forms the membrane structure or walls and a solvent-rich liquid phase, which ultimately forms the liquid-filled membrane, pores of the membrane. Temperature may also be used to induce this decomposition or to influence its rate. An evaporation step is usually provided prior to the quench, during which solvent evaporates from what ultimately will be the active layer of the membrane. The quench liquid and any solvent remaining from the membrane formation process must be removed from the membrane before it can be used for gas processing. For some membranes this step is accomplished by water washing and a simple evaporative drying step. For other membranes the quench liquid must be displaced with solvent(s), which are then evaporated. Such solvent displacement is needed in these cases in order to preserve the morphology of the membrane. Drying without the use of the intermediate solvent would result in a collapse or densification of the asymmetric structure, which can greatly reduce membrane productivity. Generally, the pores at the membrane surface where precipitation occurs first and most rapidly are much smaller than those in the interior or the bottom side of the film, and this leads to the asymmetric membrane structure (Strathman, 1981, Ho and Sirkar, 1992).

63 39 Reuvers (1987) identified two different mechanisms occurring in phase separation of ternary systems by immersion-precipitation: (1) instantaneous and (2) delayed demixing. Instantaneous demixing occurs when phase separation begins, immediately after immersion. The precipitation path crosses the binodal and two distinct phases are formed. In this case, the instant of the immersion is the onset of demixing in a barely stable casting solution. Delayed demixing can occur in the polymer solution when the precipitation path does not cross the binodal for a measurable period of time after coming in contact with the nonsolvent bath. Figure 2.11 illustrates these two mechanisms and line 1-3 depicts the binodal line. When it crosses the line 1-2, instantaneous precipitation sets in. However, if the binodal line does not cross the line 1-2, the precipitation takes some times to begin (Pandey and Chauhan, 2001). If an additive is used in the polymer solution, the phase diagram can be represented by a tetrahedron in which the binodal and spinodal are zones/surfaces intercepting each other at the initial curve (Figure 2.12). Due to the complexity of three-dimensional representations, it is usual to consider the additive along with the polymer as a single component. P P 1 1 tie - 3 line 3 tie - line t < Is t < Is S 2 NS S 2 Instantaneous Delayed NS Figure 2.11: Phase separation by instantaneous and delayed demixing of a polymer solution in a ternary system (Machado et al., 1999)

64 40 Additive C Nonsolvent Solvent Figure 2.12: Schematic phase diagram for quaternary system consisting of polymer/ solvent/nonsolvent/additive (Machado et al., 1999) Polymer Dry/Wet Phase Inversion In this study, a dry/wet phase separation process will be applied to a dry/wet spinning process in an effort to produce defect-free, ultra-thin outside-skinned hollow fiber membranes for gas separations. Dry/wet hollow fiber spinning involves the simultaneous extrusion of spinning solution and bore fluid through a spinneret die to form a nascent hollow fiber, passage of the fiber through a short air gap, coagulation in a nonsolvent precipitation bath and a subsequent collection onto a take-up drum. Dry/wet phase separation, however, describes evaporation-induced phase separation in the outermost regions of nascent flat sheet and hollow fiber membranes prior to coagulation. The dry/wet phase separation process requires a defined period for loss of a volatile solvent from a casting solution containing a carefully selected amount of a less volatile nonsolvent component. The selective loss of the volatile solvent causes destabilization in the outermost region of the nascent membrane. Interfacial dry phase separation can be observed by the almost instantaneous onset of turbidity in this outermost region.

65 41 The nascent membrane is then immersed in a nonsolvent coagulant, thus undergoing a wet phase separation where the bulk of the membrane structure is formed and extraction of the remaining solvents and nonsolvents occurs. The dry/wet technique requires multicomponent casting solutions consisting of polymer, at least one volatile solvent and one less volatile nonsolvent. It has been found, however, that the used of two solvents, a primary volatile solvent and a secondary less volatile solvent allows finer control of solvent evaporation and polymer coagulation rates. All of the solvents and nonsolvents should be miscible with the coagulation medium and the nonsolvent is added to the casting solution until the solution nears its phase boundary (Pesek and Koros, 1994). A so-called dry/wet phase inversion process can also make integrally skinned asymmetric membranes. In this case, liquid-liquid phase separation in the outermost membrane structure is induced by solvent evaporation; the bulk of the membrane structure is subsequently formed by solvent-nonsolvent exchange during a quench step. To induce evaporation-induced phase separation in the cast membrane (dry phase inversion), the casting formulations must be altered to include sufficient volatile solvent and nonvolatile nonsolvent to cause the nascent membrane to be essentially at the point of incipient phase instability. During the evaporation-induced stage of the process, a gas stream passes over a properly formulated casting solution almost at the binodal boundary. Sufficient volatile solvent is lost to drive the outside few micrometers of the nascent membrane into a spinodally-decomposed structure with an average composition. This process becomes apparent instantaneously by the formation of a fine haze on the surface of the nascent membrane. Although the occurrence of the dry phase inversion step is extremely important for achieving pore-free skin layers, the exact thickness of the dry phase separated structure appears to be of second-order importance in the generation of effective dense skin layer thicknesses. If the spinodal structure in the outermost membrane region vitrifies instantaneously during the following quench step (wet phase inversion) without undergoing further structural changes during the initial evaporation process, it

66 42 appears logical that the skin layers of the quenched membranes will be microporous. These pores would result from the interstitial spaces of the polymer-poor phase present in the outermost region of the quenched membrane at the point of vitrification. On the other hand, gas permeation experiments demonstrate that optimized membranes formed by dry/wet phase inversion could show ultra-thin and defect-free skin layers (Paul and Yampol skii, 1994) Dry Phase Separation Dry phase separation in the outer region of the nascent membrane is induced by the removal of volatile components such that the remaining solution in this thin region becomes thermodynamically unstable. This phenomenon is indicated by the formation of a cloudy region in the nascent membrane between the external gas interface and the essentially stable underlying region of the nascent membrane that has not undergone such dry phase separation. The removal of the volatile components to induce the dry phase separation can occur as a result of either free or preferably forced convection from the surface of the nascent membrane. The use of forced convection to induce the dry phase separation yields remarkably superior membrane reproducibility and defect-free selective layers. The gas or vapor used as the convective agent can include any pure or mixed stream with the ability to remove volatile components from the surface of the nascent membrane. In addition to convection by a moving gas stream across a static nascent membrane, this invention includes the embodiment of convective removal of volatile components from the surface of the nascent membrane through either a static or moving gas or vapor stream. On the other hand, free convection may require subsequent treatments to produce a defect-free selective layer. Following the dry phase separation process, the nascent membrane must be immersed in a nonsolvent precipitation liquid to cause phase inversion, thereby producing an integrally skinned asymmetric membrane with an ultra-thin separating layer (Pinnau and Koros, 1990).

67 Wet Phase Separation In the wet phase inversion process, the skin layer structure of an asymmetric membrane prepared from a binary polymer system is mainly determined by the ratio of coagulant inflow to solvent outflow. If the solvent rapidly diffuses out of the polymer solution, the concentration and viscosity on the surface layer of the nascent membrane quickly increase, limiting the mobility of polymer molecules. In this case, phase separation enters directly into the gelation region and a dense skin is formed. On the other hand, if the coagulant diffuses faster into the polymer solution, the separation path crosses the binodal curve to cause a liquid-liquid phase separation resulting in the formation of a porous surface layer (Wang et al., 1996). 2.4 Gas-Polymer Interactions CO 2 that falls into the category of acid gas such as hydrogen sulfide is commonly found in natural gas streams and hydrocarbon gases at level as high as 40% to 60% and sometimes up to 80% as hydrocarbon capacities decline and up to a feed pressure of 60 atm. These extreme operating conditions result in the swelling and plasticization of most membrane materials by the CO 2 present in the feed stream (Ho and Sirkar, 1992, Scott, 1998, Staudt-Bickel and Koros, 1999). CO 2 could be highly corrosive and rapidly destroys pipelines and equipment when combined with water unless the CO 2 is partially removed or expensive construction materials are used. In most polymers, CO 2 (critical temperature, T c = 31ºC) is more soluble than CH 4 (critical temperature, T c = 82.1ºC). Gases that are highly soluble in the membrane polymer will also affect permeation. According to Jihua (2000), gas solubility in polymers generally increases with gas condensability. Pressure-normalized flux of CO 2 in polymeric membranes is more sensitive to pressure. An increase of the CO 2 partial pressure of the feed gas will raise the CO 2 pressure-normalized flux of the membrane. This will be followed by a decrease in the selectivity for CO 2 over other gases. This is due to an increase in

68 44 the penetrant diffusion coefficient. Diffusion coefficients in polymers are sensitive to penetrant size and shape. The diffusivity of linear or oblong penetrant molecules such as CO 2 is higher than the diffusivities of spherical molecules of equivalent molecular volume. The diffusion coefficient of CO 2 is typically greater than that of CH 4. The kinetic diameter of the asymmetric molecule CO 2 (3.30Å) is smaller than that of CH 4, a symmetric molecule (3.80 Å). Based on such results, the transport of small, asymmetric molecules is understood to proceed with diffusion jumps of the penetrant through a polymer occurring principally parallel to the long axis of the penetrant. Several investigators indicated that CO 2 solubility is favored in the presence of polar groups in the main chain. Polar and flexible pendant groups such as COCH 3 or COOCH 3, being present in polymethyl-methacrylate (PMMA), polyethylmethacrylate (PEMA) and cellulose acetate (CA) attribute to the plasticization behavior of CO 2. Polar molecules (or posses a dipole moment) can be defined as a diatomic molecule consisting of different chemical species, the electron distribution is polarized and there is an accumulation of negative charge δ- on one species, the more electronegative and a corresponding depletion δ + on the other (Ladd, 1994). The polar groups of the polymer are assumed to have dipolar interactions with the polarizable CO 2 molecules. These interactions are stronger than the interactions between the chain segments. Thus CO 2 breaks these interactions providing additional diffusion pathways for the CO 2 molecules. Therefore it is suggested that the pendant group should be flexible in addition to being polar. According to Pandey and Chauhan (2001), this implies that the pressure-normalized flux will tend to be lower and selectivity higher for films of polar polymers than those of non-polar polymers. Although the polarity of a molecule or its functional group determines interchain displacement and functional free volume, they have only a moderate influence on permeation and selectivity. Gas solubility is sensitive to specific interactions between gas and polymer molecules. Gases such as CO 2, which have a quadrapole moment, are in general, more soluble in polar polymers. Many polymeric materials for CO 2 -selective membrane have been designed incorporating the use of this property. Pilato

69 45 suggested that CO 2 interacts favorably with the sulfone group in polysulfone via induced dipole-dipole interaction. In this regard, Koros found a correlation between CO 2 /CH 4 solubility selectivity and the concentration of polar sulfone and carbonyl groups in the medium in which the gases were dissolved (Ho and Sirkar, 1992, Jihua, 2000). In general, the solubility of highly polar gases such as water, ammonia, hydrogen sulfide and CO 2 is enhanced still further by the presence of polar groups on the membrane polymer. In certain cases, pressure-normalized flux is greatly increased as a result of specific interactions between condensable gases and certain polar groups which are incorporated within the membrane (Kesting and Fritzsche, 1993). Bos et al. (1999) also reported that all polymers appear to require a similar CO 2 concentration to induce plasticization, but need different pressures to reach this concentration. The reason for this is that the CO 2 solubility in each polymer is different. The average critical CO 2 concentration is 38 ± 7 cm 3 (STP)/cm 3. This result implies that the number of CO 2 molecules sorbed is important in causing plasticization and not its polar character. Figure 2.13 depicts the correlation between the solubility parameter of glassy polymers and the ideal separation factors for CO 2 /CH 4 systems. 60 Kapton P CO2 P CH Cellulose Acetate Polycarbonate Polysulfone Poly(phenylene oxide) δ poly (cal/cc) 1/2 Figure 2.13: Correlation between the solubility parameter of glassy polymers and the ideal separation factors for CO 2 /CH 4 systems (Pandey and Chauhan, 2001)

70 46 Work done by Briscoe and Kelly (1995) using polyurethane elastomer clearly describes the gas-polymer interactions since polyurethane elastomers contain a high degree of polar entities such as urethane functional groups and esters functional groups. The interactions between these polar groups are of great importance in determining the properties of polyurethanes of all types. Hydrogen bond interactions occur between N-H and C=O of the urethane groups and between the N-H of the urethane and C=O of the polyester groups. 2.5 CO 2 -Induced Plasticization on the Properties of Glassy Polymers The separation of CO 2 from CH 4 customarily takes place in the processes of natural gas sweetening, enhanced oil recovery, landfill gas and digester gas upgrading and flue gas recovery in order to reduce pipeline corrosion induced by CO 2 as well as to produce a high-purity energy product. Polymer membrane based technology is competitive for these kinds of applications in view of the following facts: (1) the high pressure of feed gas is a ready-made driving force for permeation; and (2) CO 2 is more permeable than CH 4 in most membranes. Thus, the enriched CH 4 as the residual stream still retains at a high pressure for other operations without a significant pressure loss. A successful membrane product for natural gas separation can be achieved if it possesses the following materials and performance characteristics; (1) inherently high selectivity for CO 2 and CH 4 gas pair and (2) immunity to plasticization induced by CO 2. The CO 2 -induced plasticization usually causes a severe deterioration of membrane separation performance in the natural gas application (Cao et al., 2003). Plasticization refers to the situation where the diffusivities of the penetrants are accelerated because of the swelling of the polymer matrix due to the interaction of the condensable gas, CO 2, with polymers. Although the competitive sorption decreases CH 4 solubility, the enhancement in CH 4 diffusivity is much greater than the decrease in solubility, leading to a higher pressure-normalized flux (Wang et al., 2002a). The effects of plasticization result from the plasticizer s ability to weaken

71 47 polymeric intermolecular attraction thus allowing the polymer molecules to move more readily, which increases the flexibility of the polymer. As a plasticizer, CO 2 may either swell up the interstitial place among polymer chains, which brings up a larger free volume or/and enhances segmental and side groups mobility. Though the CO 2 -induced plasticization accelerates the diffusion of penetrants, simultaneously, it severely deteriorates the gas selectivity of CO 2 /CH 4. Therefore, many efforts have been put into diminishing the effect of plasticization caused by CO 2 on the membrane separation performance (Cao et al., 2003). All polymers show the typical trend of a decreasing pressure-normalized flux with increasing pressure at low feed pressures and an increasing pressure-normalized flux with a further increase of CO 2 pressure. The decreases in pressure-normalized flux at low pressures are caused by a decreasing solubility with increasing pressure. The increase in pressure-normalized flux with increasing pressure is due to plasticization that is possible because of an increase in chain mobility. The enhanced chain mobility implies an increase in gas diffusion. Plasticization of glassy polymers can also be defined as the increase in pressure-normalized flux of CO 2 as a function of feed pressure. An increase in pressure-normalized flux, which results from an increase in chain mobility, can be observed by a depression of the glass transition temperature of the polymer-penetrant mixture. The minimum pressure necessary to induce the increasing of pressure-normalized flux is called the plasticization pressure (Bos et al., 1998, Bos et al., 1999) as shown in Figure In general, the mechanism of plasticization effects is considered to be a decrease in the cumulative intermolecular forces along the film polymer chains (Wu and McGinity, 1999). The transport of a component in a gas mixture through glassy polymeric membranes is affected by the presence of other penetrants either due to the competition among permeating species or by plasticization of the polymers if the mixture contains certain hydrocarbons and CO 2 (Wang et al., 2002a). According to Barsema et al. (2003), the commercial polyimide Matrimid 5218 exhibits a high selectivity coefficient at low CO 2 feed pressures but showed a drop in selectivity as the partial CO 2 feed pressure increased. Basically, the pressure-normalized flux of the slower gas is facilitated by the highly soluble (or faster) gas. This phenomenon is

72 48 attributed to plasticization effects caused by the high CO 2 solubility or interactions between CO 2 and the polyimide. As the membrane is plasticized the pressurenormalized flux increases significantly but the selectivity of gas mixtures decreases. According to the solution-diffusion mechanism, two aspects control the permeation of penetrants in glassy polymer membranes: (1) solution and (2) diffusion. In most glassy polymers, the diffusion coefficient contributes more to the pressure-normalized flux. Being a kinetic factor, the diffusion coefficient is correlated to the packing and motion of polymer segments and the size and shape of penetrating molecules. It is well known that high levels of CO 2 sorption can plasticize a glassy polymer and cause significant changes in its characteristics (Kawakami et al., 2003). Study by Wonders and Paul (1979) showed that glassy polycarbonate (PC) that was exposed to a high pressure of CO 2 was found to plasticize at a high pressure of CO 2. Due to the plasticizing effect, the sorption and permeation behavior of the polymer was altered. Dual-mode behavior CO 2 Pressure-normalized flux (barrer) Plasticized behavior Plasticization Pressure Feed Pressure, P (atm) Figure 2.14: CO 2 Permeation Behavior in Glassy Polymer (Ismail and Lorna, 2002)

73 49 Story and Koros (1992) investigated the sorption and permeation properties of CH 4 and CO 2 in polyphenylene oxide (PPO) and proposed that the membrane had an upturn in CO 2 pressure-normalized flux at a higher pressure due to the plasticization effect (Chen et al., 2000). Glassy polymer membranes are known to perform well in the separation of mixtures of non-condensable permanent gases (e.g. O 2 /N 2, H 2 /CO, etc.). However, their separation properties deteriorate in the presence of condensable gases and organic vapors (Barsema et al., 2003). The decrease in selectivity for many glassy polymers is related to a decrease in pressure-normalized flux. This is due to a decreasing solubility coefficient with increasing pressure. And because the pressurenormalized flux of CO 2 decreases more pronouncedly with increasing pressure than the pressure-normalized flux of CH 4, the selectivity will decrease. In the case of plasticization, it is hypothesized that the opposite occurs. The polymer matrix swells by the highly sorbed CO 2, which results in an increase in pressure-normalized flux of CO 2. Simultaneously, the pressure-normalized flux of CH 4 increases and as it increases more than the pressure-normalized flux of CO 2, the selectivity decreases (Bos et al., 1998). For the more condensable gases like CO 2, the plasticizing effect will dominate at low to medium pressures. The relatively small gas molecules, when sufficiently absorbed in a polymer network, increase the mobility of the polymer chain segments, hence plasticizing the system. However as the gas pressure reaches significant levels, the hydrostatic component of the pneumatic pressure tends to decrease the free volume, thus reducing segmental mobility. A consequence of plasticization is an increase in gas pressure-normalized flux (Briscoe and Kelly, 1995). CO 2 -induced plasticization was thoroughly reviewed by Ismail and Lorna (2002). Plasticization is the major problem faced by CO 2 -selective polymeric membranes. The increase in gas pressure-normalized flux of a membrane with increasing feed gas pressure is attributed to CO 2 -induced plasticization of the polymer matrix. This plasticizing action of CO 2 decreases the ability of the

74 50 membrane to separate molecules on the basis of size, thereby causing the reduction in selectivity. Two main factors which influence the pressure-normalized flux enhancement of the module are: (1) swelling of the glassy matrix within the nodules due to highly soluble CO 2, and (2) reduction of the effective dense skin thickness. Operating parameters such as pressure, temperature and feed composition can significantly affect pressure-normalized flux. Pressure-normalized flux of CO 2 increased dramatically with pressure that was contrasting to the normal behavior of glassy polymer. A higher pressurenormalized flux is expected with increasing operating temperature. Test results showed that the pressure-normalized flux of CO 2 increases at a faster rate with an increase in operating pressure and pressure-normalized flux of CH 4 increases at a faster rate with an increase in temperature. Thus, the separation factor increases with an increase in pressure and decreases with an increase in temperature. Hence, the effect of feed composition on selectivities showed higher concentration of CO 2 in the feed stream resulting in lower values for the same pressure. The diffusivity of a dilute penetrant in an amorphous polymer matrix is governed by the penetrant size and interactions with the polymer as well as by the shape, size, connectivity and time scales of thermal rearrangement of unoccupied space within the polymer. In a high temperature melt (T >>T g ), openings among chains that are capable of accommodating the penetrant undergo rapid redistribution in space. In other words, the penetrant is carried along by density fluctuations caused by the thermal motion of surrounding chains. At temperatures below T g, the distribution of open spaces within the configurationally arrested glassy matrix is more or less permanent. This causes the penetrant diffusion through an amorphous polymer matrix to become too slow. There is a network of preexisting cavities, the magnitude and shape of which fluctuates somewhat with thermal motion, and is modulated by the possible presence of a penetrant molecule. A dilute penetrant spends most of its time rattling within a cavity and occasionally jumps from cavity to cavity through a space that opens instantaneously via fluctuations in soft regions (e.g., regions of lower density or enhanced molecular mobility) between the cavities. The overall diffusivity in the glass would thus depend on the distribution of distances

75 51 and connectivities among the cavities as well as on the magnitude and distribution of rate constants governing the infrequent jumps of the penetrant between adjacent cavities (Neogi, 1996). Increasing the amount of plasticizer could lead to an increase in free film elongation and a decrease in tensile strength and Young s modulus (Wu and McGinity, 1999). According to Semenova et al. (1997), an increase in mobility of polymer chains can be realized not only by increasing the temperature, but also by including a plasticizer in a polymer. Particularly, penetrant specifically interacting with polymer can be a plasticizer. Theories of plasticization of polymers by low molecular compound can be summarized in the two general ideas: (1) Plasticization of a polymer with polar groups by low molecular compounds with polar groups (plasticizers) is realized by solvation. Every polar group of the polymer strongly bonds with molecules of the plasticizer. In this case, decreasing of the glass transition temperature of the polymer should be proportional to the molar fraction of added plasticizer (molar concentrations rule). Bounded molecules screen polar groups of macromolecules and increase the mobility of macromolecules and (2) In the plasticization of non-polar or low polar polymers, the main role does not belong to the energy of interaction of the polymer and plasticizer, but belongs to the change of the conformations of macromolecular chains of the polymer, which is accompanied by the change of the entropy. In this case, decreasing of the glass transition temperature should be proportional to the volume fraction of the added plasticizer (volume fractions rule) and effectivity of the plasticizer strongly depends on its specific volume Plasticization in Dense Membranes Plasticization occurs when a gas molecule sorbs into a polymer to a level sufficient enough to alter the matrix. The results of plasticization are increased pressure-normalized fluxes and reduced selectivities. One technique to avoid plasticization is to operate the membrane at or above the critical temperature (T c ) of

76 52 the suspected plasticizing molecule. This tends to cause the penetrants to be less condensable and hence minimizes the plasticization effects. If elevated pressures were involved, sufficient sorption might still occur to induce plasticization (Koros and Woods, 2001). A typical effect of plasticization is that the pressure-normalized flux versus pressure curves goes through a minimum. The pressure corresponding to the minimum pressure-normalized flux is called the plasticization pressure. For some polyimides with outstanding permeation performance a partial pressure of CO 2 of 8 to 10 bar is often sufficient to induce plasticization. These values are easily reached in, for example, natural gas upgrading where pressures up to 138 bars with CO 2 contents ranging from 5% to 40 % are reached. Landfill gas is produced at atmospheric pressure and contains approximately 40% to 50% CO 2. However, the feed stream is often compressed to higher pressures, up to 40 bars. This is to enhance process efficiency. In order to reduce the effect of plasticization, Bos et al. stabilized the polyimide Matrimid 5218 polymer film by a thermal treatment. According to Staudt-Bickel and Koros (1999), the uncross-linked 6FDA ((4,4 -hexafluoroisopropylidene) diphtalic anhydride)-mpd (m-phenylene diamine) appears to undergo plasticization at around 5 atm CO 2 feed pressure in pure gas permeation experiments at 308 K. The selectivity exhibited is around 40 at feed pressure of 10 atm and lower. At higher feed pressures, CO 2 /CH 4 selectivity decreases rapidly because of CO 2 swelling and plasticization. Meanwhile, for the chemical cross-linked 6FDA-mPD/DABA (diamino benzoic acid) 9:1 copolyimide plasticizes at a much higher CO 2 pressure of 14 atm. This is presumably due to the hydrogen bondings between the carboxylic acid groups. The CO 2 /CH 4 selectivity achieved is constant at around 48 up to a total feed pressure of 25 atm. The strongly cross-linked 6FDA-DABA did not show any signs of plasticization up to CO 2 feed pressure of 35 atm. A very high CO 2 /CH 4 selectivity of about 70 is obtained in mixed gas experiments and plasticization and swelling-induced selectivity losses were not observed up to feed pressure of 20 atm.

77 53 However, in some extreme cases CO 2 can highly plasticize the membrane, which leads to a compaction problem. In pure gas experiments with CO 2, the untreated membrane normally shows a minimum in its pressure dependence on pressure-normalized flux, whereas the treated membranes do not. Membrane performances for CO 2 /CO 4 gas mixtures showed that the plasticizing effect indeed accelerates the permeation of CH 4. The heat treatment clearly suppressed this undesired CH 4 acceleration and improved membrane performance. The favorable performance of the stabilized membrane can be attributed to less CH 4 loss and therefore a higher recovery, resulting in higher profit from gas sales (Bos et al., 1998). In the field of CO 2 -plasticization, Wessling et al. (2001) found that the plasticization behavior of polyimide (PI)/polydimethylsiloxane (PDMS) composite membrane is thickness dependent. The same material (Matrimid 5218) in the form of a relatively thick (50 µm) flat sheet dense membrane was plasticized at a partial feed pressure of 12 bars, while in the form of a thin (PI~1 µm)/pdms composite membrane, the CO 2 -plasticization was accelerated. Slightly thicker membranes (~30 µm) showed a higher resistance to CO 2 swelling (Barsema et al., 2003). Pressure-normalized flux is found to increase with increasing free volume. The free volume theory of transport postulates that the movement of molecules depends on the free volume available as well as the viability of energy sufficient to overcome polymer-polymer attractive forces. Vrentas and Duda proposed that the specific volume of the polymer and polymer-penetrant mixture is comprised of three components: (1) the occupied volume, which is the volume of the equilibrium liquid at 0ºK, (2) the interstitial free volume, which is small and distributed uniformly throughout the material, and (3) the hole free volume, which is large enough to facilitate molecular transport. Redistribution of the interstitial free volume requires a large energy input. On the other hand, redistribution of the free volume requires no additional energy, so this volume randomly migrates throughout the polymer matrix. Random mobility of the free volume may facilitate both a slow interdiffusion of polymer chains as well as penetrant transport (Ho and Sirkar, 1992). Since transport

78 54 depends on free volume, fractional free volume calculations are done. The free volume (FFV) is calculated using Equations. (2.13), (2.14) and (2.15): FFV = ( V V ) 0 / V (2.6) V = M / p (2.7) V = V (2.8) W where V is the total molar volume of the monomer unit (cm 3 /mol), M is the molar mass (g/mol) of the monomer unit and ρ is the density of the film (g/cm 3 ), which is determined experimentally. V 0 is the volume occupied by the chains (cm 3 /mol). V 0 is assumed to be impermeable for diffusing gas molecules. V w is the Van der Waals volume, which can be obtained by a group contribution method. However, according to Bos et al. (1999), a high FFV does not contribute necessarily to an increase in plasticization effect. According to Kapantaidakis et al. (1996), polyimides have been successfully commercialized by Ube and Air Product in the field of H 2 -recovery from refinery or from ammonia plant purge gases. Polyimides however are seriously affected by highly soluble penetrants, such as CO 2, which above a given partial pressure can plasticized the polymer matrix. The critical partial pressure of plasticization for polyimide is relatively low, varying between 10 to 20 bars, while the respective values for polysulfone exceed 55 bars. This is why polyimide membranes have found little application in the separation of CO 2 from light hydrocarbons in enhanced oil recovery or in landfill gas upgrading programs, where multistage or cascade operations are often needed. Chen et al. (2000) exposed polycarbonate membranes to different CO 2 densities in order to understand the CO 2 -plasticization effect on polycarbonates. CO 2 conditioning was used for the membrane treatment. The treated membrane showed a

79 55 decrease in the glass transition temperature, T g with increasing CO 2 densities. This is due to the increase of free volume of a CO 2 treated membranes that leads to an increase in gas pressure-normalized flux. Chen et al. (2000) proposed that the CO 2 sorbed into membranes changed the membrane morphology that led to an increase in excess free volume of treated membranes. It is suggested that if the excess free volume of membranes were contributed as microvoids of various sizes, then the gas sorbed into microvoids would increase the excess free volume when the microvoids were smaller than sorbed gas molecules. The gas sorbed into these microvoids would be in Langmuir form. Since the gas solubility in Langmuir-type sorption increased with increased CO 2 density, this showed that the CO 2 treatment increased the Langmuir sorption area in polycarbonate membrane. Briscoe and Kelly (1995) studied the plasticization of thick film polyurethane elastomer by CO 2. The CO 2 pressure was raised from 2 MPa to 16 MPa. The gaspolymer system was allowed to equilibrate for 1 hour at each pressure step before an infrared spectrum was recorded. At pressures up to 12 MPa, the sorption and dilation rates increased rapidly as the gas pressure increased. However, at higher pressures, the rates of sorption and dilation decreased significantly. At lower pressures, CO 2 acts as a plasticizer for the polyurethane, which causes significant swelling. At higher pressures, the hydrostatic component of the gas dominates, thus inhibiting further gas uptake. Thus, the plasticization effect became the less dominant process at pressures above 12 MPa Plasticization in Asymmetric Membrane Films Chung et al. (2003) reported that with the progress in materials and membrane fabrication, membrane systems for the removal of CO 2 in natural gas, landfill gas and enhanced oil recovery became more competitive compared to conventional separation processes such as amine scrubbing etc. However, glassy membrane materials exposed to high pressure CO 2 environments exhibit different pressure-normalized flux behavior due to plasticization induced by CO 2 sorption. As

80 56 a result, membrane permeance increases and selectivity decreases. Houde et al. (1992) investigated the plasticization mechanisms of conditioned membranes using Wide Angle X-ray Diffraction (WAXD). In the case of cellulose acetate (CA) and poly(methylmethacrylate) (PMMA) dense films, the increase in pressure-normalized flux of CO 2 with pressure was mainly due to the significant increase in the intersegmental spacing on exposure to high pressure CO 2. However, in the case of polysulfone (PSF), CO 2 -induced plasticization has enhanced segmental mobility visibly than the change in intersegmental spacing. Work done by Mikawa et al. (2002) showed that the plasticization of the asymmetric polyimide membrane (6FDA-6FAP) by CO 2 facilitates the motions of the side groups, such as bulky CF 3 groups. This phenomenon may perturb the local packing of the polyimide to an extent that it allows easy passage of small penetrant molecules and provides a kinetic opening of sufficient size to allow the penetrant to execute a diffusional jump or diffuse CO 2. The increasing pressure-normalized flux behavior observed in this study appears to be attributed to plasticization of the polyimide segments. However, the CO 2 permeation stability of the asymmetric polyimide membranes significantly depended on the molecular weight. Plasticization of the membrane prepared from the high molecular weight polyimide was not observed. It was believed that the molecular motions of the side groups are restricted in a high molecular weight polyimide with a more packed structure. In the study of Kawakami et al. (2003) using asymmetric polyimide membranes, they found that asymmetric 6FDA-APPS membranes with a skin layer thickness of 55 nm showed a pressure-normalized flux of CH 4 that obeyed a typical dual-mode transport. However, after the asymmetric 6FDA-APPS membrane was exposed to a CO 2 pressure of 760 cmhg, The CO 2 permeation with an ultra-thin skin layer increased upon exposure to the CO 2 pressure. This result indicates that the swelling in the skin layer of the asymmetric membrane occurred due to the high pressure CO 2 and caused plasticization of the polyimide segments. Wessling and Schoeman (1991) reported a structural change in a polyimide film upon exposure to CO 2. The results are interpreted as slow loosening of densely packed entanglements of polymer chains initiated by CO 2 -plasticization (White et al., 1995).

81 57 Donohue et al. (1989) had studied the permeation behavior of CO 2 /CH 4 mixtures in asymmetric cellulose acetate membranes. From the study, pressurenormalized flux of CO 2 was found to increase dramatically with pressure. They believed that plasticization had take place. However, the membrane plasticization appears to be completely reversible and is not damaged by the high partial pressure of CO 2 over the range of the conditions studied. The pressure-normalized flux of CO 2 was found to be independent of temperature compared to the data for CH 4. This is true since at high pressures the permeation is controlled by the plasticization of cellulose acetate and not by the solubility. According to Sanders (1986), CO 2 facilitates the motion of pendant acetate and ester groups in cellulose acetate and that the movement of these side groups may be of sufficient amplitude to allow easier passage of small penetrant molecules. Since it is likely that CO 2 would enable segments in straight-chain polymers to undergo similar motions, the apparent anomaly can be explained by the assumption that the diffusion of CO 2 in cellulose acetate is controlled by the motions of its pendant groups Plasticization in Asymmetric Hollow Fiber Membranes Koros and Woods (2001) carried out a study using three asymmetric hollow fiber membrane systems for applications in elevated temperature with low feed pressure systems: (1) a single component polyaramide, (2) a single component polyimide, and (3) a composite polyimide on a polyimide/polyetherimide blend support. The membranes were tested for their properties as a function of temperature by directly heating a module to the desired temperature. Its permeation properties were measured before the permeation system was allowed to equilibrate when the measured pressure-normalized flux remained stable over a period of 30 minutes. Throughout the study, hydrocarbons were shown to plasticize glassy polymers. Wang et al. (2002a) reported that when a mixture contains CO 2 such as CO 2 /CH 4 mixture at a higher pressure, the effect of plasticization influences the transport of the gas mixture and couples with the competition. This phenomenon

82 58 arises from the fact that the dense selective layer of the asymmetric hollow fiber is very thin, the inception of plasticization in the hollow fiber may occur at a very low feed pressure. According to Wang et al. (2002a), a similar observation was reported by Koros s research group where they stated that the plasticization of asymmetric hollow fibers would initiate at low feed pressure and heavier than that of thick flat dense film. Wang et al. reported an immediate increase of pressure-normalized flux of CO 2 with feed pressures in pure polyimide asymmetric hollow fiber membranes. Kapantaidakis et al. (2002) reported the same observation with ultra-thin asymmetric hollow fiber membranes based on Matrimid 5218 and polyethersulfone blends. Jordan et al. (1990) compares the effect of CO 2 conditioning on the permeation properties of asymmetric hollow fiber membranes and dense film samples of a polyimide. There is significantly different behavior between dense and asymmetric membrane samples. It is suggested that the precipitation process used to create an asymmetric membrane structure creates morphology where small nodules of polymer are tied together with regions of lower density polyimide chains. These loosely packed chains are more susceptible to movement and structural change when exposed to a gas, which can act as a plasticizer. These loosely packed chains are not present in annealed dense film samples and therefore dense film samples are less affected by plasticization effect (White et al., 1995). Permeation test carried out by Krol et al. (2001) using propane and propylene exhibit a decreasing trend with increasing feed pressure, which can be explained by the dual sorption model. Propane exhibits a constant pressure-normalized flux trend while propylene showed the opposite trend. However, the pressure-normalized flux does not decrease further with increasing feed pressure but even increases again. According to Krol et al. (2001), the pressure at which the increase in pressurenormalized flux occurs (i.e. the minimum in the pressure-normalized flux versus pressure plot) is called the plasticization pressure. At such feed pressures the gas concentration in the polymer material disrupts the chain packing. The polymer matrix swells and the segmental mobility of the polymer chains increases. This results in an increase in the gas diffusivity and therefore the pressure-normalized flux increases.

83 59 Cao et al. (2002) developed a new one-polymer and one-solvent spinning system (6FDA-2,6 DAT/N-methylpyrrolidone (NMP)) which exhibited excellent performances for CO 2 /CH 4 separation. The produced membrane showed a similar performance to the one reported in the French patent. The experimental results reveals that 6FDA-2,6-DAT asymmetric composite hollow fiber membranes have a strong tendency to be plasticized by CO 2 and suffer severe physical aging with a initial CO 2 pressure-normalized flux of 300 GPU drifting to 76 GPU at the steady state (CO 2 /CH 4 selectivity of 65). The CO 2 pressure-normalized flux of 162 GPU with CO 2 /CH 4 selectivity of 74 for pure gases was reported in the French patent. However, the 6FDA-2,6-DAT asymmetric composite hollow fiber membranes still present an impressive ultimate stabilized performance with a CO 2 /CH 4 selectivity of 40 and a CO 2 pressure-normalized flux of 59 GPU under mixed gas tests. These results showed that 6FDA-2,6-DAT polyimide is a promising membrane material candidate for CO 2 /CH 4 separation applications. 2.6 Glass Transition Temperature, T g Glass transition temperature, T g is the temperature at which the polymer changes from hard and brittle to soft and pliable. The T g of polymers can be affected by the glass molecules absorbed in the polymer matrix and the depression in T g by absorbed gas is explained by the plasticization effect. The higher the CO 2 pressure and hence the more sorbed CO 2 in the polymer, the lower the T g of the film. The glass transition measurements can also be used to estimate the free volume of the membrane since an increase in free volume leads to an increase in gas permeation of a polymer membrane (Zhong et al., 1999, Chen et al., 2000). Mi and Zheng (1998) measured the T g of polycarbonate (PC) by means of high pressure Differential Scanning Calorimetry (DSC) in helium, nitrogen and CO 2 atmosphere. According to Paul and Yampol skii (1994), T g is the most important method, which can correlate with chain mobility. As the T g is taken as a measure for chain stiffness, the lower T g of the membrane film may be attributed to a more flexible

84 60 polymer matrix as a result of less efficient packing (lower density). The exposure of glassy polymers to CO 2 that are highly sorbed could cause significant changes in the behavior of the polymer, particularly in the depression of the T g. However, sometimes an increase in glass transition is found due to CO 2 -induced crystallization (Ismail and Lorna, 2002). In some membrane applications, feed streams contain very high levels of strongly interacting gases such as CO 2 and hydrogen sulfide (H 2 S). Exposure to high levels of these gases results in a large depression of the effective T g for some polymeric materials. The glass transition depression is thought to increase the size of segmental motions that participate in the diffusion process and the increased mobility of the polymeric chains usually results in a drop in the selectivity. The CO 2 pressure necessary to induce plasticization is widely different and dependent on material properties (Ho and Sirkar, 1992). 2.7 Suppression of CO 2 -Plasticization Suppression of CO 2 -plasticization with flat sheet membrane caused a considerable reduction of gas pressure-normalized flux. However, it is known that ultra-thin dense or asymmetric hollow fiber membranes exhibit differences in physicochemical properties, like aging phenomena and plasticization behavior when compared to thicker membranes prepared from the same polymeric material (Barsema et al., 2003). Suppressing plasticization implies a suppression of polymer chain flexibility. This can be achieved by (1) cross-linking of the polymer matrix, (2) heat treatment of polymers or (3) blending the polymer with a second polymer that is less susceptible to plasticization (Krol et al., 2001, Kapantaidakis et al., 2003).

85 Cross-Linking Method Dudley et al. (2001) reported that cross-linking of a polymer matrix might influence physical properties and the ability of the material to transport and separate gases. Since cross-linking acts to resist the mobility of polymer chains by the formation of covalent bonds, it stabilizes the material against thermal and chemical degradation as well as impacting gas transport. Restriction of the polymer chains mobility can impede gas transport since the diffusion of gas molecules through a polymer involves the cooperative motion of chain segments. Gases with larger molecular diameters are more severely impacted than those with smaller diameters. Thus, cross-linking can lead to improved separation selectivity but usually only at the expense of reduced gas permeation rates. According to Liu et al. (2001), a lot of effort was made to change the chemical structure of polyimides, aiming at getting both high permeable and selective membrane materials. But most of them suffer severe ageing and performance decay due to densification or plasticization. In order to overcome these problems, cross-linking modifications were carried out. Experimental results suggested that cross-linking modifications imparted membranes with antiplasticization, good chemical resistance and even with good long-term performance. Besides, the gas pressure-normalized fluxes/selectivity relationships of some crosslinked polyimides were higher than the normal trade-off line. Liu et al. (2001) carried out an extremely simple room temperature chemical cross-linking for the modification of polyimide films for gas separations using 6FDA-durene. In this study, a chemical modification was performed by immersing the dense 6FDA-durene films in a p-xylenediamine methanol solution for a certain period of time followed by washing with fresh methanol and drying at ambient temperature. Chemical crosslinking modifications usually results in (1) tightening the interstitial spaces among the polymer chains which decreases the penetrants diffusion path (2) protecting nodule integrity from CO 2 -induced swelling, or/and (3) restricting polymer chain vibration for diffusion jumps. As a result, the diffusion of penetratnts through the membrane is likely hindered and correspondingly their pressure-normalized fluxes decrease drastically (Cao et al., 2003).

86 62 A study by Staudt-Bickel and Koros (1999) found that copolyimides containing strong polar carboxylic acid reduced plasticization possibly due to hydrogen bonding between ethylene glycol and the carboxylic acid group. However, the modification through cross-linking reactions between ethylene glycol and carboxylic acid groups in the diamino benzoic acid (DABA) based polyimide by thermal treatment at 150ºC rendered much better anti-plasticization materials. The major drawback of heat treatment at an elevated temperature is to deteriorate the subtle structures of asymmetric membranes and their gas permeation properties. Therefore, cross-linking methods performed at low temperature are necessary for the successful modification of asymmetric membranes Heat Treatment Method Another work done by Chung et al. (2003) on asymmetric 6FDA-2,6 DAT polyimide hollow fiber membranes found that CO 2 -induced plasticization occurs when the feed pressure varies from 20 to 200 psi. CO 2 -induced plasticization could be gradually suppressed when heat treatment temperature increases from room temperature to 320ºC. At temperatures greater than 250ºC, the CO 2 -induced plasticization is removed completely. The ideal CO 2 /CH 4 selectivity of hollow fiber membranes shows a slight increase with an increase in heat treatment temperature. The anti-plasticization characteristics are confirmed by Scanning Electron Microscopy (SEM), Wide Angle X-Ray Diffraction (WAXD), Differential Scanning Calorimetry (DSC) and Thermogravimetric Analysis (TGA). Heat treatment was used by Ismail and Lorna (2003) to suppress CO 2 - plasticization in asymmetric polysulfone membranes. The asymmetric membranes were given different heating temperatures and different treatment durations to investigate the suppression of undesirable CO 2 -plasticization. Pressure-normalized fluxes were reduced with the intensity of heat treatment. Experimental results showed that the membranes were stabilized against CO 2 -plasticization after heat

87 63 treatment processes, especially at a temperature of 140 C. The CO 2 permeation rate for treated membrane maintains at steady state over the applied feed pressures. Krol et al. (2001) heat-treated Matrimid hollow fiber membranes both above and below the glass transition temperature, T g in order to study the possibility of suppressing propylene plasticization. The fibers were characterized by nitrogen permeation and showed a reduced pressure-normalized flux. It was suggested that the more intense (higher temperature or longer treatment time), the larger the decrease in pressure-normalized flux. The reduction in pressure-normalized flux of flat sheet polyimide membranes is not as drastic as in the asymmetric hollow fiber membranes. Heat treatment near or above the T g of the fiber material caused a significant decrease in pressure-normalized flux. From the study, the porous substructure becomes denser when the 2 minutes treatment is compared to the untreated fiber. The structure changed even more when the heat treatment was prolonged to 5 minutes. The cell walls of the pores in the substructure became more or less molten together. For treatments at lower temperatures the changes in membrane structure were less pronounced for treatments at lower temperatures. The heat treatment of the fibers not only densifies the membrane skin, but also more seriously, causes a collapse or a densification of the porous substructure thereby causing a drastic increase in transport resistance. As for this study, it can be concluded that thermal curing at temperatures below the glass transition temperature, T g were successful in suppressing the propylene plasticization. Heat treatment of polymers may affect either the main chain linkages (often carbon-carbon bonds), or substituent atoms and side chains (Bamford and Tipper, 1975). For an overall view, the summary of the types of heat treatment used by previous researchers in order to suppress plasticization is shown in Table 2.3 and Table 2.4 summarizes the significant suppression methods for gas plasticization.

88 Blending Method Glassy polymers are seriously affected by highly sorbed gases, such as CO 2, which above a given partial pressure can plasticized the polymer matrix (Kapantaidakis et al., 2003). Polymer blending is a possibility of modifying material properties. Blending can show new properties that cannot be found in single polymers. This also holds for membrane preparation to tailor a specific separation performance. For instance, Ube prepared membranes from BPDA-based polyimide blends. Furthermore, blending is an option for reducing the price of the membranes. Considering the polyimide Matrimid in particular, several examples can be found in the patent literature where membranes are prepared from Matrimid and the much cheaper PEI. It is well known that many polymers do not mix on a molecular level. The blends contain the separate polymers as individuals domains or phrases. Blends often consist of a matrix of one polymer containing another polymer as a dispersed or cocontinuous phase. The polyimide Matrimid, however can be blended on a molecular level with polycarbonate (PC), polysulfone (PSF), polyethersulfone (PES) and polybenzimidazole (PBI). A single glass transition temperature and the optical transparency of the films, for example, indicate blending on a molecular level. Bos (1996) has extensively studied the possibilities to suppress CO 2 - plasticization of Matrimid film by blending Matrimid and Thermid FA-700. In order to induce Thermid cross-linking within the semi-interpenetrating network, heat treatment is found to suppress plasticization effectively. The dense Matrimid film was successfully suppressed by heat treatment for 15 minutes at 350ºC. However, high cross-linking density of the polymer is not always necessary to suppress plasticization. A simple thermal curing effect forming charge transfer complexes appeared to be sufficient (Krol et al., 2001). In 1999, Kapantaidakis et al. (1999) showed that the plasticization pressure of the polyimide Matrimid in pure gas CO 2 permeation experiments was shifting towards higher feed pressures by blending it with polysulfone, a polymer with a significantly higher plasticization pressure than Matrimid. Bos et al. (2001) reported that plasticization can be stabilized by blending a polymer with high plasticization tendencies with one that is hardly affected by the sorbed molecules in pure gas experiments carried out with pure CO 2.

89 Table 2.3: Summary of types of heat treatment used by previous researchers Researcher(s) Polymer T g Heat treatment Operating temperature Kawakami et al., 1996 Bos et al., 1998 Krol et al., 2001 Ismail and Lorna, 2003 Chung et al., 2003 This Study Polyimide 6FDA-m-DDS NA Below T g 150, 200, 250 C Dense Matrimid Film NA Above T g 350 C for 15 min Asymmetric Polyimide Matrimid 5218 Hollow Fiber Asymmetric Polysulfone Flat Sheet Asymmetric 6FDA-2,6 DAT Hollow Fiber Asymmetric Polysulfone Hollow Fiber 325 C Below and above T g 150, 250, 300, 350 C for 2-5 min 185 C Below T g 100, 120, 140, 160, 180 C for min 335 C Below T g C for 10 min 70 C Below T g 70 C for 2 and 5 min

90 Table 2.4: Significant suppression methods for gas-plasticization Researcher(s) Material(s) Structure/ configuration Robeson Polysulfone Makaruk et al. Maeda and Paul Maeda and Paul Bos et al. Bos et al. Staudia-Bickel and Koros Bos et al. Krol et al. Ismail and Lorna Chung et al. This study Polycarbonate Polysulfone (P-1700) Poly (phenylene oxide) Matrimid 5218 Matrimid 5218 and Thermid FA-700 6FDA-based polyimides Matrimid 5218 Matrimid 5218 Polysulfone (P-1700) 6FDA-2,6DAT polyimide Polysulfone (P-1700) Dense, flat sheet Dense, flat sheet Dense, flat sheet Dense, flat sheet Dense, flat sheet Dense, flat sheet Dense, flat sheet Dense, hollow fiber Asymmetric, flat sheet Asymmetric, hollow fiber Asymmetric, hollow fiber Plasticization suppression method Low-molecular-weight-additives: N-phenyl-2-naphtylamine, PNA and 4,4 - dichlorodiphenylsulfones, DDS Low-molecular-weight-additives: 1,1-bis(4-hydroxyphenyl)-2,2-propane diacetate (DAPP), 1,1-bis(4-hydroxy-3,5-dichlorophenyl)-2,2-propane diacetate (DACPP), 1,1-bis(4-hydroxy-3,5-dibromophenyl)-2,2-propane diacetate (DABPP), 1,1-bis(4-hydroxy-3,5-dichlorophenyl)-2,2-propane dibenzoate (DBCPP), 1,1-bis(4-hydroxy-3,5-dichlorophenyl)-2,2-propane di- 2,4-dichlorobenzoate (DCBCPP). Low-molecular-weight-diluents: tricresyl phospate, TCN, N-phenyl-2 naphtylamine, PNA and 4,4 -dichlorodiphenylsulfone, DDS Low-molecular-weight-diluents: tricresyl phospate, TCN, N-phenyl-2 naphtylamine, PNA and 4,4 -dichlorodiphenylsulfone, DDS Thermal treatm Semi-interpenetrating polymer network (s-ipn) formation Chemical crosslinking Matrimid blend with polysulfone and copolyimide P84 Mild heat treatment Heat treatment Heat treatment Mild Heat treatment Year

91 CHAPTER 3 METHODOLOGY 3.1 Research Design The widespread need for the removal of CO 2 from process gas streams associated with upgrading of natural gas, landfill gas recovery and enhanced oil recovery has led to a continuing interest in developing acid gas separation membranes. The selection of a suitable membrane material plays an important role in achieving the best performance in membrane separation processes. This chapter describes the experimental methods used in order to study the plasticization effect on the separation performance of polysulfone hollow fiber membrane when tested with various compositions of CO 2 /CH 4 gas mixture. This study consists of four main stages, namely polysulfone hollow fiber membrane fabrication, membrane characterization and membrane performance testing as depicted in Figure st Stage (Asymmetric polysulfone hollow fiber fabrication) 2 nd Stage (Membrane characterization) 3 rd Stage (Heat treatment) 4 th Stage (Membrane performance testing) Figure 3.1: Steps of the research

92 Materials Selection Polysulfone Polysulfone Udel P-1700 (Amoco Polymers, Inc., Alpharetta, GA) which is of low MW grade and has a glass transition temperature of about 185ºC (Kesting and Fritzsche, 1993) was selected as the raw material for this study. Polysulfone (PSF) is an amorphous hydrophobic polymer and also a suitable membrane polymer due to its good film-forming properties as well as its thermal and biological resistance (Rodemann and Staude, 1994, Kapantaidakis et al., 1996, Yamasaki et al., 1997). Polysulfone has satisfactory gas pressure-normalized fluxes and acceptable selectivities. It can be used with highly sorbing, plasticizing gases since this polymer has the tendency to plasticize and has the potential to demonstrate possible effects of suppressing plasticization (Ismail and Lorna, 2003). These properties and its relative low cost (e.g. $15/kg PSF Udel P-1700) established polysulfone as a standard membrane material (Kapantaidakis et al., 1996). The chemical structure of polysulfone is shown in Figure 3.2. Polysulfone is categorized into polymer groups that contain carbon-sulfur chains in its backbone structure (Grulke, 1994). Polysulfone contains aromatic hydrocarbons in the main chains and is composed of phenylene units linked by three different chemical groups-isopropyledene, ether and sulfone. Each of the three linkages imparts specific properties to the polymer, such as chemical resistance, temperature resistance and impact strength. The most important of these in imparting particular properties to the material appeared to be the diphenyl sulfone groups. Excellent thermal stability is imparted because of the strength of the chemical bonds and resistance to oxidation is ensured by the fully oxidized state of the sulfur atom in this group. The thermal stability is improved when the backbone does not contain aliphatic groups. The three types of linkage between the phenylene groups impart flexibility to the molecule, leading to excellent toughness in the polymer.

93 69 CH 3 O C O SO 2 CH 3 n Figure 3.2: The chemical structure of polysulfone (PSF) The insertion of flexible links such as O- in the aromatic ring increases the solubility but decreases the softening temperature. The C(CH 3 ) 2 - groups constitutes the weakest link (Bamford and Tipper, 1975). The combination of phenyl rings attached to the sulfone groups results in a high degree of resonance stabilization (Kesting and Fritzsche, 1993). Polysulfone has high tensile strength, moderately high modulus and impact strength and excellent resistance to creep. The resistance to high temperatures is somewhat better than that of poly(phenylene oxide) and properties remain substantially unchanged for long periods at temperatures up to 140ºC. Polysulfone is resistant to acids, alkalies and salt solutions as well as to detergents, oils and alcohols, but polar organic solvents and chlorinated and aromatic hydrocarbons attack the material severely (Bikales, 1971) Solvents The organic chemicals used in membrane solutions include N, N- dimethylacetamide (DMAc), ethanol (EtOH) and tetrahydrofuran (THF). These chemicals were obtained in reagent grade from MERCK-Schuchardt, Germany, Sigma-Aldrich Chemie, Switzerland and MERCK, Germany respectively and used as received.

94 Gases For pure gas permeation measurement, CH 4 and CO 2 were obtained from Union Carbide. The gases were specified as having a purity of 99.95%. Both gases were used without further purification. For mixed gas permeation measurement, gas mixture consists of 20% CO 2 / 80% CH 4 supplied by National Oxygen Pte Ltd was used. 3.3 Fabrication of Asymmetric Hollow Fiber Membrane Turbidimetric Titration Method In order to conduct the turbidimetric titration, a homogeneous polymer dope with specific composition was initially prepared by dissolving a polymer in a mixture of nonvolatile solvent and a volatile solvent at room temperature. After that, 100 ml of polymer solution was titrated with pure nonsolvent, titrated solution was stirred and held at constant temperature while caution had to be taken to minimize solvent loss. The solution cloud point was easily recognized by visual observation. Amount of added nonsolvent (in millimeters) was determined from the burette. Figure 3.3 shows the apparatus used for turbidimetric titration. Burette Nonsolvent Cone flask Magnetic stirrer Polymer solution Figure 3.3: Apparatus for turbidimetric titration

95 Preparation of Multi Component Polymer Dope The polymer dope should contain (1) polymer, (2) a solvent system, and (3) a nonsolvent system. The polymer can be chosen from either commercially available glassy polymers or tailored materials with desirable intrinsic gas separation properties. The solvent system chosen must be able to both fully dissolve the polymer and to tolerate the presence of the nonsolvent system so that the point of the incipient phase separation can be reached while maintaining the dope rheology that is appropriate for spinning the nascent membrane structure (Pinnau and Koros, 1990). According to Ho and Sirkar (1992), the polymer concentration of 20% to 40% by weight is commonly employed. That is why the polymer concentration of the polymer dope must be sufficiently high to produce the hollow fiber membranes with dense surface separating layer and good mechanical support layer for gas separation. The polymer dope used for spinning of hollow fiber membrane must be of sufficiently high viscosity and polymer concentration in order to produce a selfsupport extrusion prior to coagulation. However, too high viscosity of the polymer dope is undesirable as it causes difficulty in spinning (Teo et al., 2000). The apparatus as shown in Figure 3.4 below was used during the polymer dope preparation. According to Li (1993), strong mechanical stability is one of the important requirements to be fulfilled since gas separation membranes are often used under high pressure. The polysulfone pellets were vacuum-dried in a vacuum oven at 70ºC, for at least 24 hours before use in order to remove absorbed water vapor before used. N, N-dimethylacetamide (DMAc), tetrahydrofuran (THF) and ethanol (EtOH) were first poured into the flask and stir-heated until the solvent mixture reached the temperature of 55ºC. The polymers were then added part by part into the solvent mixture until all the polymers dissolved completely and were stirred well until the membrane solution became homogeneous. The membrane solution took about 8 hours to be completely homogenized. The membrane solution was then degassed in ultrasonic bath (Model Branson 3510) in order to eliminate micro bubble.

96 72 Stirrer Condenser Thermometer Flange reaction flask Heating unit Figure 3.4: Apparatus for polymer dope preparation Membrane Fabrication using Dry/Wet Spinning Process The dry/wet phase separation process was applied to dry/wet spinning process in order to prepare asymmetric polysulfone hollow fiber membranes for gas separation. Dry/wet hollow fiber spinning involves the simultaneous extrusion of spinning solution and bore fluid through a spinneret die to form a nascent hollow fiber, passage of the fiber through a short air gap, coagulation in a nonsolvent precipitation bath and subsequent collection onto a take up drum. Dry/wet phase separation, however, describes evaporation-induced phase separation in the outermost regions of nascent hollow fiber membranes prior to coagulation (Pesek and Koros, 1994). Figure 3.5 show the dry/wet spinning process with forced convection in the dry gap.

97 73 Figure 3.5: The laboratory scale for dry/wet spinning process (Ismail et al., 1999) (1) nitrogen; (2) dope reservoir; (3) gear pump; (4) filter, 7 µm; (5) syringe pump; (6) spinneret; (7) forced convection tube; (8) roller; (9) wind-up drum; (10) refrigeration/ heating unit; (11) coagulation bath; (12) washing/treatment bath; (13) wind-up bath; (14) schematic of a spinneret. Polymer dope containing of polysulfone (Udel-1700), N, N- dimethylacetamide, tetrahydrofuran and ethanol was used to fabricate the asymmetric hollow fiber membranes. In order to prepare hollow fiber membranes, polymer dope was degassed for an hour before used to eliminate micro bubbles. The dope reservoir was kept at ambient temperature (20ºC ± 2ºC) during spinning. On extrusion from the spinneret (spinneret dimensions: OD 0.6 mm/id 0.3 mm), the fiber passed through a cylindrical forced convection chamber (length 9 cm, diameter 5 cm), which was flushed with 4 l/min of nitrogen gas. The nitrogen was introduced through a ¼ in. tube, which abutted upon the chamber normal to the surface at mid height. A 2 mm clearance existed between the top of the forced convection chamber and the bottom plate of the spinneret and also between the bottom of the chamber and the water level in the first coagulation bath. Table 3.1 summarizes the spinning conditions used in this study.

98 74 Table 3.1: Summary of spinning conditions Spinning conditions Value Spinning solution PSF/DMAc/THF/EtOH Polymer concentration wt % Spinning Temperature 25ºC ± 2ºC Spinneret: (OD/ ID) 0.6 mm/0.3 mm Bore fluid 20 % Potassium Acetate, 80 % Water Bore fluid flow rate 1.0 ml/min Ratio DER:bore fluid 3:1 Air gap distance 9 cm Dope Extrusion Rate (DER) 2.5 cm 3 /min Number of revolution 1 Pump Delivery 0.3 cm 3 /rev Jet Stretch (V f /V o ) 1 Extrusion Speed, V o cm/min Wind-Up Drum, V f cm/min Time for front motor 7.20 s Time for back motor 3.00 s Spinneret Area, A sp 2.12 x 10-3 cm 2 External coagulant Water External coagulant temperature 14ºC ± 2ºC Room humidity 70 % Drying procedure 2 day in water 2 day in methanol 1 day air dry Pure water at 14ºC ± 0.5ºC was used in the external coagulation bath. The bore coagulant was 20% (w/w) solution of potassium acetate in water at ambient temperature. This equates to the water activity of 0.9. The hollow fibers were spun at dope extrusion rate (DER) of 2.5. The stretch ratio (wind up speed/extrusion speed) was fixed at 1 throughout the experiment. This is due to the fact that increases in dope extrusion rate will increase pressure-normalized flux of CO 2. This may be due to better molecular chain packing and orientation induced by shear stresses, which enhanced interaction between CO 2 and the polymer (Ismail et al., 1999, Chung et al., 2000a). The ratio of dope extrusion rate to bore fluid injection rate was kept constant at a value of 3. After spinning, the membranes were steeped in water and then dried using methanol solvent exchange technique. The nascent hollow fiber was solidified after passing through the coagulation bath and once removed from the take-up drum, the hollow fibers were washed in water at ambient temperature for 48 hours and in methanol for another 48 hours before air-dried.

99 Hollow Fiber Coating Since the formation of a few defects is inevitable during membrane preparation and module manufacture, therefore the membranes were coated with a thin layer of silicone rubber. Coating is considered as a standard technique that protects the membrane during handling. Silicone rubber is extremely permeable compared to polysulfone but has a much lower selectivity; thus the silicone rubber coating did not significantly change the selectivity or the flux through the defect-free portions of the polysulfone membrane. The silicone rubber layer also protects the membrane during handling (Li, 1993, Murphy and de Pinho, 1995, Yamasaki et al., 1999). According to Chung et al. (2000a), based on the Resistance Model, this approach can easily recover inherent membrane separation performance if the percentage of surface defects is less than 10-5 and the gas permeation through the asymmetric membrane is dominated by the solution-diffusion mechanism after surface sealing. As for the coating procedure, the hollow fibers were dip-coated in a coating solution of polydimethylsiloxane (Sylgard-184) in 97% hexane for 3 minutes. The fibers were air dried at room temperature for at least 1 day so that the silicone rubber coating was fully cured before testing Hollow Fiber Potting Twenty fine hollow fibers with a length of 30 cm were assembled tied-loose and wrapped with a plastic wrapper to hold the fibers in place before the potting process being carried out. The fibers were then tied to a holder and the fibers at the bottom were sealed with an epoxy resin (Loctite E-30CL Epoxy Adhesive). After the bottom part of the fibers had solidified, the same step was repeated for the top part of the fiber. The fibers were then put into a stainless steel tube to make a test module. Hollow fiber modules are to be installed vertically to prevent from sagging in the middle and allowing feed gas to bypass the membranes (Schendel, 1984). Figure 3.6 described the details of hollow fiber potting.

100 76 Gas inlet Hollow fibers membrane Stainless steel casing Gas outlet Figure 3.6: Hollow fiber potting 30 cm length End cap 3.4 Characterization Pure Gas Permeation Measurements Pure gas permeation measurement was conducted in order to determine whether the membrane prepared was suitable to be used in CO 2 /CH 4 separation. Permeation was conducted using pure CH 4 and CO 2 gases respectively. The hollow fiber module was coated with silicone before put to test. The gases were exposed to the hollow fiber membranes for at least 20 minutes to equilibrate. The apparatus for the gas permeation test using hollow fiber is shown in Figure 3.7. The feed and retentate pressure was measured by pressure gauges. The upstream pressure was kept at 1 bar to 10 bars while the downstream was kept open to the atmosphere. The permeate volumetric flow rate was measured by means of a bubble soap flow meter (Kapantaidakis et al., 1996, Yamasaki et al., 1997, Wang et al., 2002a) reading to 1.0 cm 3. The gas bubble flow meter was used because of the small membrane area which resulted in much lower permeate flow rates with values close to the meter accuracy (Ettouney and Majeed, 1997).

101 77 Figure 3.7: Schematic diagram of the gas permeation testing system The asymmetric hollow fibers were spun from polysulfone solution synthesized in the laboratory. Twenty hollow fibers with length of 30 cm and an outer diameter of around 0.06 cm were assembled into a bundle to make a simple module before putting into stainless steel casing for testing. All the measurements were carried out for pure CO 2 and CH 4 at room temperature using coated membrane. The flow rates and compositions in the outlet of permeate and retentate sides were measured. Pressure-normalized flux and selectivity of asymmetric membranes were determined by constant pressure-variable volume method. Properties of CO 2 and CH 4 are summarized in Table 3.2 (Uchytil et al., 1992, Lide and Frederikse, 1994, Shilton et al., 1996, Ismail et al., 1999, Lide, 2001). Intrinsic pressure-normalized flux and selectivity data of dense film are used as standard values for comparison with asymmetric membrane.

102 78 Table 3.2: Properties of carbon dioxide and methane Gas CO 2 CH 4 Kinetic diameter a, D (Å) Molecular weight b, M (g/mol) Molecular speed c, v (cm/s) Viscosity d, η (cmhg.s) 1.52 x x 10-9 Intrinsic pressure-normalized flux e, P (Barrer) Intrinsic selectivity, α (Unitless) 28 a From Ismail et al., b From Uchytil et al., c,d,e From Lide and Frederikse, 1994, Lide, 2001, data at 25 C. e From Shilton et al., The pressure-normalized flux of pure gas was measured for hollow fiber membrane samples at room temperature. Volumetric gas flow rates were determined using a bubble soap flow meters. The pressure-normalized flux is calculated using the following equation: P Qi = (3.1) l ( P )( A) i where (P/l) is the pressure-normalized flux of a membrane to gas i (GPU = 1 x 10-6 cm 3 (STP)/cm 2 -s-cmhg); Q i represents volumetric flow rate of gas i at standard temperature and pressure difference (cmhg); P is the pressure difference between the feed side and the permeation side of the membrane (cmhg); A is the membrane surface area (cm 2 ). The ideal selectivity of an asymmetric membrane, α A, B is defined as the pure gas pressure-normalized flux ratio of more permeable gas A to less permeable gas B as follows. ( P / l) A α A, B = (3.2) ( P / l) B

103 79 From the above equations, it is evident that the pressure-normalized flux of a membrane depends on the thickness of the effective separation layer while the selectivity is independent of the dense layer thickness (Chung et al., 2000a). For the simplicity of the mathematical analysis, the following assumptions were made in the evaluation of the dense skin structure parameters: (a) the pores are circular, (b) the thickness of the dense skin and the coating layer is uniform, and (c) the porous support has no resistance for gases (Wang et al., 1990, Wang et al., 1995). The mean pore size and surface porosity of the dense skin of polymeric asymmetric membranes were determined by the gas pressure-normalized flux method. For high fluxes and good separating properties, desirable substrate membranes should have an open internal structures and the resistance to gas transport should be negligible. On this basis, the total gas flux for permeant i, Q i, through the asymmetric membrane can be generally expressed by Q i P A1 P2 A2 = 1 p p 2 l + (3.3) l τ r The first term corresponds to the gas flux in the dense skin region and the second term is gas flow through pores or defects. Q i is volumetric flow rate of gas i, A 1 is the effective surface area of dense skin region, A 2 is the porous surface area, P 1 is the intrinsic pressure-normalized flux of the membrane material, P 2 is the effective pressure-normalized flux characteristic of pore, l is skin layer thickness, l r is pore length and τ is the tortousity factor. p is the pressure difference across the membrane, which is given by p = p us - p ds, where p us and p ds are the upstream and downstream pressure of the gas permeant respectively. Asymmetric membranes suitable for gas separation have a very dense skin of substantially nonporous polymer, where total surface area of membrane, A A 1. Dividing Equation (3.3) by A 1 and rearranging yields

104 80 Q P P ε K i = = + (3.4) pa i 1, i 2, i 2 1 l lrτ where K i is the total effective pressure-normalized flux and ε = A 2 /A 1 is the surface porosity. Gas flow through pores occurs by a combination of Knudsen diffusion, bulk flow and viscous diffusion. Relative contribution of each transport modes for a particular case depends on both temperature and pressure conditions in pore. Equations for flow in porous media can only be approached by analogy with flow in capillaries. A general equation for the pressure-normalized flux in capillaries with non-circular cross sections is stated as follows (Marchese and Pagliero, 1991, Ismail and Lai, 2004). P 2, i 2 m p 4mβ = + vi (3.5) k η p 3p k 0 i o o 1 The first term on the right hand side is the viscous flow term. m is the mean hydraulic radius or mean pore size, η i is the viscosity of the permeant i, p o is atmospheric pressure at which gas flux was measured, p is the mean pressure where p = p us + p ) / 2, where p us and p ds are the upstream and downstream pressure of the ( ds gas permeant respectively. The last term is the mean molecular speed of permeant i given by v i = ( 8RT / πm i ) 1/ 2 temperature and R is the gas constant. where M i is the gas molecular weight, T is the absolute Numerical factors k 1 ad k 0 in Equation (3.5) account for changes of pore shape and δ is a numerical factor for a particular system. Taking the average shape for both pore and defect as more or less rectangular in the commercial and asymmetric membranes used in the experimental work, a typical value of k 0 should be 2.5. Experiments made from various investigators showed that β/k 1 τ 2 0.4,

105 81 τ = and β/k 1 = 0.8 (Yasuda and Tsai, 1974, Marchese and Pagliero, 1991, Ismail and Lai, 2004). Equation (3.5) can now be written as P m p m = vi (3.6) p 2 2, i ηi p0 0 Introducing Equation (3.6) in Equation (3.4) leads to K 2 P m ε m ε p = 0.4 (3.7) 2 l η τ p 1, i i v 2 i + l lr p0 τ r i 0 So, a plot of K i against p will give a straight line, where the intercept K 0,i, at p = 0 and the slope B 0,i are given by P m ε = v (3.8) 1, i K 0, i 2 l lr p0 τ i 2 m ε B0, i = 0.4 (3.9) 2 lη p τ r i 0 Skin thickness is calculated using gas permeation data obtained from two different gases, A and B. Expression for skin thickness can be obtained from Equations (3.8) and (3.9), stated as P1, A - P1, BC l = with K - K C 0, A 0, B C B 0, A A A = (3.10) B 0, B η η B v v B In membrane-based gas separation process with relatively low operating pressure (or driving force), gas molecules tend to pass across porous area in skin layer (low resistance). Besides, compaction effects do not affect membrane exposed

106 82 to low pressure driving force, so pore length is assumed to be same as skin thickness. In this case, ε 1 and l r l, therefore Equation (3.7) can be rewritten as m = (3.11) K 0, i v 2 i lp0τ B 0, i 2 m = 0.4 (3.12) 2 lητ p i 0 Therefore, mean pore size can be determined by 2 B0, ilηi p0τ m = (3.13) 0.4 On the other hand, in membrane-based gas separation process with relatively high operating pressure (or driving force), gas transport is mainly contributed by gas flux through dense skin region. Furthermore, membrane exposed to high pressure driving force is affected by compaction effects, which tend to reduce pore length dramatically. Therefore, pore length in more or less rectangular shape is assumed to be same as pore radius. In this case, l r m, thus Equation (3.7) can be rewritten as K P ε m ε p = 0.4 (3.14) 2 η τ p 1, i i v 2 i + l p0τ i 0 So, a plot of K i against p will give a straight line, where the intercept K 0,i, at p = 0 and the slope B 0,i are given by P ε = v (3.15) 1, i K 0, i l p0τ i

107 83 2 m ε B0, i = 0.4 (3.16) 2 η p τ i 0 Therefore, surface porosity can be determined by ε B iηi p τ = 0, m (3.17) Membrane Configuration Use From the literature, three-stage separation system was found to provide the lowest operating cost among all other configurations. The system above can be operated in different configurations depending on which stream that need to be further purified. The design of this separation system is a combination of different configurations proposed by Bhide and Stern (1991; 1993a; 1993b), Ettouney et al. (1998), Qi and Henson (1998; 2000) etc. The typical schematic diagram of the membrane module configuration arrangement is shown in Figure 3.8. In this study, both cascade and series configuration will be investigated. The feed stream will be control by a ball valve (V1). The feed pressure and flow rate will be determined by both pressure gauge (P1) and flow meter (F1). As the feed stream enters the first membrane module (M1), separation will take place and the permeate will exit the membrane from the bore side of the membrane and the retentate will exit the membrane from the shell side. The purpose of the check valve is to assure a one-way flowing stream. The flow rates of both streams will be determined by flow meter (V2) and (V3). The product will be collected from both the permeate and retentate stream. Two ball valves are installed in the final streams. These valves act as a controller to control which stream is to be injected to the gas chromatograph (GC) to check the composition of the gas.

108 (V1) Feed ball valve (P1) Feed pressure gauge (F1) Feed flow indicator (M1) First membrane module (P2, P4, P6) Permeate pressure gauge (P3, P5, P7) Retentate pressure gauge (C1-C8) Check valves (X1-X11) On-off valves (M2) Second membrane module (M3) Third membrane module (F2) Permeate flow indicator (F3) Retentate flow indicator (V2) Permeate ball valve (V3) Retentate ball valve Figure 3.8: Typical schematic diagram of the membrane module configuration arrangement

109 85 For single-stage membrane module configuration (Figure 3.9), the gas mixture is fed to the first membrane module (M1) from the shell side of hollow fibers membrane. On-off ball valves (X2, X5 and X7) were closed in order to prevent the low pressure permeate gas from entering the second and third membrane. Meanwhile, on-off valves (X1 and X11) were also closed to prevent the high pressure retentate stream from entering the second and the third membrane module. The permeate stream will flow through pressure gauge (P2), check valves (C1, C3, C6, C8) and on-off valves (X3 and X8) before flowing through the permeate flow indicator (F2). The permeate stream will be controlled by the permeate ball valve (V2) before injected to the gas chromatograph. The retentate stream will passed along pressure gauge (P3), on-off valve (X4) and check valves (C4 and C5) before passing through retentate flow indicator (F3). The retentate stream will be controlled by retentate ball valve (F3) before been connected to the gas chromatograph. For the operation of two-stage membrane module in series configuration (Figure 3.10a), the retentate stream from the first membrane module will be used as the feed stream to the second membrane module (M2) by closing the on-off valve (X4). In this case, the retentate stream will flow through pressure gauge (P3), on-off valve (X1) and check valve (C2). On-off valve (X2) will be closed to prevent the retentate stream from mixing with the permeate stream at the upstream. The permeate stream from the first membrane module will flow through the same path as mention above and mix with the permeate stream from the second membrane module. The second stage permeate stream will flow through pressure gauge (P4) and on-off valve (X5). On-off valves (X6 and X7) will be closed to prevent the permeate stream to be recycled in the loop. From here, the combined permeate stream will proceed in the same manner as mentioned above. As for the second stage retentate stream, the stream will passed through pressure gauge (P5), on-off valve (X11) and check valve (C5) before proceed with the same process step mentioned in previous paragraph. On-off valve (X10) will be closed to prevent retentate flow from entering the third membrane module and check valves (C4 and C7) prevent the high pressure retentate stream from flowing backward into the former and later region respectively.

110 86 Figure 3.9: Single-stage module configuration Two-stage membrane module in cascade configuration (Figure 3.10 (b)) uses permeate stream from the first membrane module (M1) as the feed stream to the second membrane module (M2). However, the permeate stream pressure was very low to allow gas separation from occurring. In that case, this configuration was omitted and will not be considered in this study. Three-stage membrane module in series configuration (Figure 3.11 (a)) uses the retentate stream from the second membrane module (M2) as the feed stream to the third membrane module (M3). The retentate stream was controlled by closing the on-off valve (X4, X9 and X11). The retentate stream will then passed through the pressure gauge (P7). The rest of the retentate stream will follow the process step mentioned previously. The permeate stream will flow through pressure gauge (P6) before combined with the permeate stream from the first and the third membrane module. The permeate stream will follow the same flow path as mentioned above. (a) (b) Figure 3.10: Two-stage module configuration in (a) series and (b) cascade configuration A diagram of the three-stage membrane module in cascade configuration is presented in Figure 3.11 (b). This configuration uses retentate stream from the first

111 87 membrane as the feed stream to the second membrane module (M2) and permeate stream from the second membrane module (M2) as the feed stream to the third membrane module (M3). The permeate stream flows through pressure gauge (P4) and on-off valves (X6 and X9) while the on-off valve (X5, X7 and X10) were closed. The permeate stream then flow according to the same path as mentioned above. The high pressure retentate stream from the second membrane module passed along pressure gauge (P5), on-off valve (X11) and check valve (C5) before combined with low pressure retentate stream from the third membrane module that flow through pressure gauge (P7) and check valve (C7). The retentate stream then flow using the same path as mentioned previously. Mixed gas permeation measurements were generally conducted at 2 bars to 10 bars with 20% CO 2 /80% CH 4 gas mixture. The test gas was first stabilized for 20 minutes before the measurements were taken. The modules were kept at constant temperature with gas composition monitored by gas chromatograph and flow rate was determined with bubble flow meter. Pressure-normalized flux of mixed gas is calculated on the basis of the downstream side kept at 1 atm (14.7 psi) (White et al., 1995). According to Bos et al. (1998), pressure-normalized flux of CO 2 in the gas mixture is slightly lower compared with the pressure-normalized flux of pure CO 2 because of the presence of CH 4 reduces the pressure-normalized flux of CO 2. CO 2 /CH 4 mixed gas selectivity of the membranes is smaller than those obtained in pure gas permeation measurements at equivalent pressures. The reduction in selectivity arises from the relative net effect of three factors; namely, (1) the relative influences of the non-ideal gas behavior, (2) competitive sorption and (3) plasticization on the individual gas in the mixture. Even though CO 2 is a highly condensable and has a greater sorption tendency than CH 4, the former has a larger fugacity coefficient depression because of non-ideal gas behavior.

112 88 (a) (b) Figure 3.11: Three-stage module configuration in (a) series and (b) cascade configuration The collected data are used to calculate the pressure-normalized flux of CO 2 and CH 4. The species pressure-normalized flux is used instead of the pressurenormalized flux because of the difficulty of determining the effective membrane thickness with sufficient accuracy. The pressure-normalized flux of CO 2 is given by yq p P = (3.18) ( A( p p)) P = A(0.5( p f x yq f p p x)) yp r p (3.19) where P is the pressure-normalized flux, p is the CO 2 partial pressure on the permeate side and is given by p=yp p, A is the membrane area (cm 2 ), y the CO 2 mole fraction in the permeate stream, Q p is the permeate flow rate (cm/s) and p f and p r is the CO 2 average partial pressure in the feed and reject absolute pressures (cmhg), p p is the permeate pressure and x f and x are the CO 2 mole fractions in the feed and retentate streams. The equation for the pressure-normalized flux of CH 4 is similar to Equation (3.18), except for the use of (1-x) and (1-y) to determine the CH 4 mole fractions in the retentate and permeate streams. The calculated pressure-normalized flux gives an overall measure of the permeation rate of the CO 2 and CH 4 gases.

113 89 The module stage cut is defined as the flow rate ratio of the permeate and feed streams. The stage cut, S, is given by Q p S = (3.20) Q + Q ) ( p r The above equation applies for single-module configuration and the individual modules of the two and three-module systems (Ettouney et al., 1998). The overall stage cut for the two and three-module are given by S ( Q + p Q ) 1 p2 = (3.21) Q f S ( Q p Q p Q ) p3 = (3.22) Q f Network Model Network model developed by Ismail and Lim (2001) is used in this study in order to simulate the separation process for one, two and three-stage membrane module in series and cascade configurations. By providing input variables such as CO 2 feed composition, feed flow rate, feed pressure and CO 2 product purity, percentage of CO 2 in permeate and retentate stream together with its flow rates will be calculated by the network model. The output variables will be used to calculate the species pressure-normalized flux, selectivity and stage cut.

114 Heat Treatment Method such as cross-linking, heat treatment and blending are known to be able to suppress CO 2 -plasticization. However, heat treatment method was found to be well suited since heating even at low temperature can successfully suppress CO 2 - plasticization (Krol et al., 2001). Many researchers report heating below and above glass transition temperature, T g (Kawakami et al., 1996, Bos et al., 1998, Krol et al., 2001, Chung et al., 2003, Ismail and Lorna, 2003). In this study, heat treatment below the polymer T g is carried out since heating above the polymer T g had resulted in selectivity reduction due to extreme skin layer densification. The T g of the polysulfone hollow fiber membrane is at 182ºC and is first confirm using Differential Scanning Calorimetry (DSC). The dried hollow fiber membranes were placed in a hot air oven. Mild heat treatment at 70ºC was carried out at different duration of 3 min and 5 min, respectively. The hollow fiber membranes were then cooled down naturally to 35ºC. The cooled hollow fibers were then taken out from the oven and subsequently used for testing Density Measurements The density of untreated and treated hollow fiber was measured by weighing the well-dried hollow fiber sample in air. The volume of each sample was then calculated. For each density measurement, three samples were cut from the same hollow fiber. The final density of that particular hollow fiber was the average of the three measurements. The standard deviation varied from to g/cm 3.

115 Differential Scanning Calorimetry (DSC) Differential Scanning Calorimetry (DSC) measures the difference between reference and a sample during a controlled temperature change. In other words, DSC was used to study what happens to polymer when heated in order to determine the thermal transitions of the polymer. Changes in the heating rate of the sample relative to the reference can be converted into heat capacity and enthalpy changes. DSC responds to all thermal events, including chemical reaction, physical transition, release of strain, loss of volatiles and decomposition. In this study, DSC was used to measure the glass transition temperature, T g of the polymer as well as to measure the depression of the T g after thermal curing. The T g of the untreated hollow fiber was determined using a DSC with a heating rate of 5ºC/min. In most cases no transition was observed in the first heating run. The T g were therefore taken from the second run. The average T g determined from two samples prepared from the same hollow fiber membrane has a standard deviation of ± 1ºC Scanning Electron Microscopy (SEM) The geometrical characteristics and the morphology of the polysulfone hollow fiber membrane will be determined using Scanning Electron Microscope (SEM). Firstly, the samples (surface and cross-section) were chosen randomly. The dried fibers were then broken in liquid nitrogen before mounted on sample stubs with double-sided tape. The samples were then sputtered with a thin layer of gold using a sputtering apparatus. The cross-section and surface of the hollow fibers were examined with SEM (Philips SEMEDAX; XL40; PW6822/10) with potential of 20 kv under magnifications ranging from 250x to 500x.

116 Thermogravimetric Analysis (TGA) Thermogravimetric Analysis (TGA) was used to measure the weight change of each sample as a function of temperature. The sample was first heated at 50ºC for 5 min under nitrogen environment. The sample was then further heated from 50ºC to 1000ºC at 20ºC/min. TGA offers many advantages compared with the usual isothermal methods (Bamford and Tipper, 1975). The thermal behavior of a polymer can be characterized rapidly over a wide range of temperature in one experiment using a small amount of sample. The weight loss is probably caused by the removal of small amount of solvent still remaining in the fibers (Krol et al., 2001).

117 CHAPTER 4 RESULT AND DISSCUSSION In this chapter, the effect of CO 2 -induced plasticization and membrane configuration on asymmetric polysulfone hollow fiber membrane for CO 2 /CH 4 gas separation will be discussed. Knowing that the interaction of CO 2 with polymers can affected the membrane performance, plasticization must be suppressed. Suppressing plasticization implies a suppression of polymer chain flexibility. This can be achieved by various means. In this study, heat treatment method is used since heat treatment was able to control the increasing of the pressure-normalized flux as a function of feed pressure. Heat treatment may affect the physical structural properties of a polymer. The anti-plasticization characteristics caused by heat treatment are confirmed by density measurement, Scanning Electron Microscopy (SEM), Differential Scanning Calorimetry (DSC) and Thermogravimetric Analysis (TGA). Changes in gas transport behavior will also be discussed. In general, the permeation properties depend on intermolecular packing distance, chain stiffness, polymer-polymer interaction and polymer penetrant interaction. The effect of membrane module configuration of single, two and three-stages in series and cascade configuration will also be discussed in this chapter in order to find which configuration is best used for CO 2 /CH 4 gas separation. Results of stage cut will be discussed in each and every membrane module configuration.

118 Spinning Solution Formulation In making asymmetric membranes using the dry/wet phase inversion process, the addition of a nonsolvent, usually an organic nonsolvent for the polymer, into the membrane polymer dope is very important to obtain the optimal membrane structure and improve performance of the resulting membranes. The recent breakthrough in the preparation of gas separation membranes is essentially based on the concept of introduction of nonsolvent-additive in the membrane polymer dope. Table 4.1 summarizes the solution composition before and after titration used in this study. In polysulfone flat sheet membrane preparation, it was observed that addition of a nonsolvent-additive (water) plays an important role in controlling the membrane morphology. The addition of a suitable nonsolvent-additive into the membrane polymer dope accelerates the coagulation process from solution to gel when the polymer dope was immersed in a coagulant. As a result, membranes with thinner skin layer and more uniform structure could be obtained (Wang et al., 1995). Through visual observation, PS1 solution composition shows no changes after been titrated with 7 ml of ethanol. After titrated with ethanol in the range of 11 ml to 12 ml, PS3 and PS4 solution turned into a white solution, illustrating that the ethanol titrated into the solution had already exceeded. The PS2 solution composition was choosen to be used in the preparation of the polymer dope in this study since the solution become turbid when the last drop of the 9 ml of ethanol was titrated into the solution. Ethanol is a relative weak nonsolvent coagulant, which could be used as a nonsolvent additive to make a polymer dope approaching the incipient gelation (Ren et al., 2002). Table 4.1: Summary of the solution composition before and after titration Solution composition (wt %) Component (s) Before After PS PS1 PS2 PS3 PS4 Polysulfone (Udel 1700) N, N-dimethylacetamide (DMAc) Tetrahydrofuran (THF) Ethanol (EtOH)

119 Pure Carbon Dioxide and Methane Gas Permeation Behavior in Untreated and Treated Membranes Pure gas permeation measurement was conducted in order to determine whether the membrane prepared is suitable to be used in CO 2 /CH 4 separation. Permeation was conducted using pure CH 4 and CO 2 gases, respectively. Permeation experiment with CH 4 was performed before the permeation experiments with CO 2 to exclude any membrane conditioning effects on the inert gas permeation experiments (Wessling et al., 2001). The hollow fiber module was coated with silicone before testing. The hollow fibers were exposed to testing gases for at least 20 minutes to equilibrate. The configurations of the membrane module used in the experiments include single-stage, two-stage in series and three-stage in both series and cascade configurations. The results for the hollow fiber membrane configuration include the permeate gas pressure-normalized flux, selectivity and stage cut for both untreated and treated membranes. The results were plotted as a function of feed pressure. The hollow fiber membranes were heat treated at 70 C below the T g of polysulfone in order to suppress CO 2 -plasticization. Heat treatment will result in a reduction of chain mobility and simultaneously prevented the membrane plasticization (Ismail and Lorna, 2003). Figure 4.1 shows the pressure-normalized flux of CO 2 and CH 4 of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of feed pressure. From the figure, membrane treated for 2 and 5 min was compared with untreated membrane in terms of CO 2 permeation behavior. For the untreated membrane, the pressure-normalized flux of CO 2 exhibited an immediate increase with increasing feed pressure especially after feed pressure of about 2 bars. According to Chung et al. (2003), glassy membrane materials exposed to high-pressure CO 2 environment exhibits different pressurenormalized flux behavior due to plasticization induced by CO 2 sorption. As a result, pressure-normalized flux increases and selectivity decreases. Thus, at feed pressure of 2 bars the CO 2 gas concentration was already sufficient or high enough to disrupt the chain packing in the polymer material resulting in an increase in segmental or chain mobility which led to the increasing of gas diffusion (Bos et al., 1998, Bos et

120 96 al., 1999). As the feed pressure was further increased, CO 2 concentration will continue to build up. Increase in CO 2 concentration was the main factor that contributed to the plasticization phenomenon or membrane swelling which resulted in an increase in CO 2 pressure-normalized flux. However, the pressure-normalized flux of CH 4 for both untreated and treated membranes showed almost constant trend with increasing feed pressure. This behavior was in agreement with the dual-mode sorption transport model as studied and predicted by previous researchers (Bos et al., 1998, Krol et al., 2001, Kawakami et al., 2003). According to Krol et al. (2001), in some cases the pressure-normalized flux did not further decrease with increasing feed pressure but continue to increase. The pressure at which the increase in pressure-normalized flux occurs (i.e. the minimum in the pressure-normalized flux versus pressure plot) is called the plasticization pressure. At such feed pressures the gas concentration in the polymer material disrupts the chain packing. The polymer matrix swells and the segmental mobility of the polymer chains increases. This resulted in an increase in the gas diffusivity and therefore the pressure-normalized flux increases. In contrast, the pressure-normalized flux of treated hollow fiber membranes did not show a minimum, but level off above 2 bars. The pressure-normalized flux approached a limiting value at high feed pressure. These results indicated that the heat treatment method was successful in suppressing the CO 2 -induced plasticization. According to Bos et al. (1999), the decrease in pressure-normalized flux at low pressure was due to the reduction of solubility with increasing feed pressure. However, the disadvantage of this method is that the pressure-normalized flux of the treated polymer membrane was lower than that of the untreated one. This can be clearly seen from the figure where the pressure-normalized flux of membrane treated for 5 min showed a much lower values compared to the one treated for 2 min. Krol et al. (2001) reported that heat treatment reduces the gas pressure-normalized flux. The more intense the treatment (i.e. the higher the temperature or the longer the treatment time) the larger the decrease in the pressure-normalized flux. This is probably due to tighter packing of polymer structure (Ettouney and Majeed, 1997) or might as well indicated that a polymer matrix densification effect had taken place.

121 97 Similar results were obtained by Kawakami et al. (1996) who had thermally cured 6FDA-mDDS films and Bos et al. (1998) who had cross-linked Matrimid films. The increasing trend of the pressure-normalized flux of CO 2 in untreated membrane resulted in a reduction of selectivity with increasing of feed pressure. Figure 4.2 shows the CO 2 /CH 4 selectivity of untreated and treated asymmetric hollow fiber membrane in single-stage configuration as a function of feed pressure. This phenomenon is due to plasticization effect induced by dissolved CO 2 which taken place in the polymer as the feed pressure is further elevated (Sada et al., 1992). Similar result was observed by Wessling et al. (2001). However, the selectivity of the membrane treated for 2 and 5 min showed a more constant trend as a function of feed pressure. Selectivity of membrane treated for 5 min is slightly higher than the selectivity of membrane treated for 2 min. This is probably due to a better polymer chain packing formed which implies that the polymer segments are relatively immobile (Donohue, 1989) as a result of physical entanglements of polymer chains (Tadmor and Gogos, 1979). The increase in selectivity is mainly due to the tightened chain packing induced by shear rates while the decrease in selectivity is mainly due to relatively porous skin structures induced by low viscosity (Chung et al., 2000a). Increase in selectivity can also be attributed to the enhanced polymer molecule orientation at high shear (Shilton et al., 1997). According to Ismail et al. (1997), enhanced molecular orientation may enable membrane selectivities to be elevated beyond the recognized intrinsic value of the polymer. It is clear that sorption of CO 2 causes severe plasticization effects in single-stage hollow fiber membrane. In order to compare the plasticization effects in single-stage hollow fiber membrane module, gas permeation of a two-stage hollow fiber membrane module in series configuration was conducted. This is to determine whether membrane configuration can be manipulated in order to control plasticization effects. Figure 4.3 exhibited the pressure-normalized flux trend in two-stage hollow fiber membrane module in series configuration of untreated and treated asymmetric polysulfone hollow fiber membrane. This configuration is suitable for the production of less permeable species or retentate component. According to Callahan (1999), two-stage membrane module configuration was able to produce about 98% pure CH 4 retentate product in the separation of CO 2 from wellhead natural gas using polycarbonate membrane.

122 98 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated CH4-Untreated CO2-2 min CH4-2 min CO2-5 min CH4-5 min Figure 4.1: Pressure-normalized flux of CO 2 and CH 4 of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressures CO2/CH4 selectivity Feed pressure (GPU) Untreated 2 min 5 min Figure 4.2: CO 2 /CH 4 selectivity of untreated and treated asymmetric hollow fiber membrane in single-stage configuration as a function of feed pressures

123 99 The pressure-normalized flux of CH 4 showed a constant trend as a function of feed pressure in both untreated and treated membrane. In other words, pressurenormalized flux of CH 4 was in agreement with the dual-mode transport theory. However, the main attraction here is the pressure-normalized flux of CO 2. Since the pressure-normalized flux of CO 2 increases in single-stage membrane module configuration, the similar trend might also be exhibited in two-stage membrane module configuration. However, the pressure-normalized flux of CO 2 seems to decrease slightly as a function of feed pressure. Interestingly, treated hollow fiber membranes showed more significant reduction in CO 2 pressure-normalized flux values compared to untreated membranes. Membrane treated at 70 C for 2 min showed a drastic reduction in CO 2 pressure-normalize flux while the corresponding flux of membrane treated at 70 C for 5 min only slightly reduced before level off after feed pressure of 2 bars. This is probably due to better polymer chain packing as mentioned previously. Reduction in CO 2 pressure-normalized flux influences the selectivity by raising the selectivity values higher than the selectivity of untreated membrane. Membrane treated for 5 min at 70 C showed a better and more stable selectivity trend. This observation might be well explained by the less pronounced of CO 2 - induced plasticization effects that took place inside the membrane. This is because when the feed gas enters the membrane module, the more permeable CO 2 permeates through the first membrane module rapidly leaving behind the unrefined feed gas to be further refined in the second membrane module. The second membrane module might experiences a more efficient separation since the retentate stream contains less fast permeated gas. Therefore, the second membrane module was less plasticized which lead to better trend of pressure-normalized flux. As a result, the selectivity of two-stage membrane module revealed a more constant trend compared to the selectivity of the single-stage membrane module configuration. Figure 4.4 shows CO 2 /CH 4 selectivity of the untreated and treated asymmetric hollow fiber membrane in two-stage series configuration as a function of feed pressure. Figure 4.5 and 4.6 show the effect of feed pressure on pressure-normalized flux of CH 4 and CO 2 of untreated and treated membrane in series and cascades

124 100 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated CH4-Untreated CO2-2 min CH4-2 min CO2-5 min CH4-5 min Figure 4.3: Pressure-normalized flux of CO 2 and CH 4 of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressures CO2/CH4 selectivity Feed pressure (bar) Untreated 2 min 5 min Figure 4.4: CO 2 /CH 4 selectivity of the untreated and treated asymmetric hollow fiber membrane in two-stage series configuration as a function of feed pressures

125 101 configurations respectively. As usual, CH 4 gas showed a constant trend of pressurenormalized flux with further increase of feed pressure in both untreated and treated membranes as well as in both series and cascade configurations. The pressurenormalized flux of CH 4 of untreated and treated membrane had not only showed the same trend but the values of the pressure-normalized flux are almost the same. This behavior indicated that CH 4 does not contribute to the plasticization of the membrane. However, the pressure-normalized flux of CO 2 in untreated membrane in series and cascades configuration showed a major difference in its values. Series configuration exhibited a five time higher values of CO 2 pressure-normalized flux compared to cascade configuration. Throughout the experiment, CO 2 showed fluctuated trend of pressure-normalized flux with increasing feed pressure which was in contrast to the normal behavior of glassy polymer. Normally, the pressurenormalized flux of a glassy polymer decreases with increasing feed pressure, which has been extensively explained by the dual-mode sorption model. However, in some cases the opposite trend occur (Krol et al., 2001). The increasing trend of CO 2 as a function of feed pressure in this study indicated that CO 2 -plasticization in polysulfone membranes occurs for feed pressures greater than 1 bar for both configurations. High CO 2 concentration in the polymer film disrupts chain packing, resulting in an increased segmental mobility (Bos et al., 1998). Investigation on treated membranes revealed that the pressure-normalized flux of CO 2 in series configuration decreases drastically at first before level off as the feed pressure increases. On the other hand, pressure-normalized flux of CO 2 in cascade configuration decreases smoothly as a function of feed pressure. Pressurenormalized flux of membrane treated at 70 C for 5 min is lower than the one treated at 70 C for 2 min in both configurations. The decrease in CO 2 pressure-normalized flux with increasing CO 2 pressure in glassy polymer is consistent with the wellknown dual-mode sorption and transport model predictions (Staudt-Bickel and Koros, 1999). As mentioned in the previous section, this behavior is influenced by the tightened chain packing of polymer structure that causes the polymer structure to become more rigid and inflexible or immobile. Basically, many polymers showed typical trend of decreasing pressure-normalized flux with increasing feed pressure at low feed pressure and an increasing pressure-normalized flux with further increase of

126 102 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated CH4-Untreated CO2-2 min CH4-2 min CO2-5 min CH4-5 min Figure 4.5: Pressure-normalized flux of CO 2 and CH 4 of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure 2.5 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated CH4-Untreated CO2-2 min CH4-2 min CO2-5 min CH4-5 min Figure 4.6: Pressure-normalized flux of CO 2 and CH 4 of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascades configuration as a function of the feed pressure

127 103 CO 2 pressure. The increase of pressure-normalized flux with increasing feed pressure is due to plasticization that is possible because of an increase in chain mobility (Bos et al., 1998, Bos et al., 1999). Figure 4.7 and Figure 4.8 show CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series and cascade configuration as a function of feed pressure. From these figures, cascade configuration exhibited higher value of CO 2 /CH 4 selectivity than the series configuration. In fact, the results for the first two feed pressures exhibited selectivities exceeded the recognized polysulfone intrinsic selectivity of 28. This is because in cascade configuration, only the first membrane module undergoes separation at high-pressure while the other two modules experienced separation at low-pressure since the permeate stream was used as the feed stream. Although the permeate from the first membrane module would be mixed with the permeate from the third stage, the resultant permeate flow is still relatively low. In spite of the build up of CO 2 concentration in the first membrane module occurs as the feed pressure is further elevated, the second and the third membrane module were not severely plasticized by the CO 2 due to an exposure of low concentration of CO 2. As a result, small value of pressure-normalized flux was detected. Hence, a high selectivity was obtained. However, in series configuration, all three membrane modules experienced high feed pressures. As the feed pressure is increased, the CO 2 concentration will build up. When this is occurring, the highly sorbed CO 2 plasticized the membrane, which results in an increase in CO 2 pressure-normalized flux (Bos et al., 1998, Kawakami et al., 2003). According to Chung et al. (2003), glassy membrane materials exposed to high pressure CO 2 environment exhibited different pressurenormalized flux behavior due to plasticization induced by CO 2 sorption. As a result, pressure-normalized flux increases and selectivity decreases. Table 4.2 shows the pressure-normalized flux and CO 2 /CH 4 selectivity of polysulfone hollow fibers in both series and cascade configurations. Hollow fibers module in cascade configuration exhibited the highest selectivity (~45) compared to hollow fiber module in series configuration which only showed selectivity of about

128 CO2/CH4 selectivity Feed presssure (bar) Untreated 2 min 5 min Figure 4.7: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure CO2/CH4 selectivity Feed pressure (bar) Untreated 2 min 5 min Figure 4.8: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascades configuration as a function of the feed pressure

129 The CO 2 /CH 4 selectivity of polysulfone dense membranes is reported to be about 28. Varying the feed flow rate at a constant feed pressure and temperature may affect the stage cut. Stage cut is the ratio of permeate flow rate to feed flow rate (Coker et al., 1999). In a membrane system, the first gas that permeates is the most highly enriched in the rapidly diffusing component. A very small stage cut across the membrane yields the purest permeate product. As larger stage cuts are taken, the feed gas becomes enriched in the gases which permeate more slowly and both their driving force and concentration in the permeate are increased. The permeate becomes less pure as a larger fraction of the feed gas is permeated (Ho and Sirkar, 1992). Figure 4.9 (a) and (b) show the impact of feed flow rate on stage cut of single-stage configuration. The stage cut exhibited an increasing trend with increasing CO 2 and CH 4 feed pressure in both untreated and treated membrane. The stage cut in single-stage configuration of untreated membranes varies from 0.02 x 10-4 to 0.71 x 10-4 for CH 4 and from 0.51 x 10-4 to 15.2 x 10-4 for CO 2. Stage cut for the treated membrane showed reduction in the stage cut versus feed pressure curve. The stage cut ranges from 0.02 x 10-4 to 0.45 x 10-4 for CH 4 and from 0.41 x 10-4 to 9.06 x 10-4 for CO 2 for membrane treated at 70 C for 2 min and from 0.02 x 10-4 to 0.41 x 10-4 for CH 4 and from 0.39 x 10-4 to 8.03 x 10-4 for CO 2 for membrane treated at 70 C for 5 min. Reduction in the stage cut values of the treated membrane is probably due to the stiffening of polymer chain. As the heat treatment is prolonged, Table 4.2: Selectivity comparison of untreated asymmetric polysulfone hollow fibers in series and cascade configurations Feed pressure CO 2 /CH 4 selectivity (bar) Cascade configuration Series configuration

130 Stage cut (x10-4 ) LPM-Untreated 10 LPM-2 min 10 LPM-5 min 20 LPM-Untreated 20 LPM-2 min 20 LPM-5 min 30 LPM-Untreated 30 LPM-2 min 30 LPM-5 min Feed pressure (bar) (a) 0.8 Stage cut (x10-4 ) LPM-Untreated 10 LPM-2 min 10 LPM-5 min 20 LPM-Untreated 20 LPM-2 min 20 LPM-5 min 30 LPM-Untreated 30 LPM-2 min 30 LPM-5 min Feed pressure (bar) (b) Figure 4.9: Effect of stage cut on feed pressure in single-stage configuration of untreated and treated membrane (a) CO 2 and (b) CH 4

131 107 the rigidity of the polymer structure increases. The increasing trend of stage cut indicates better performance of CO 2 /CH 4 gas separation. More interestingly, the stage cut values increase with increasing membrane stages or membrane area. Two-stage membrane module configuration exhibited higher stage cut compared to single-stage membrane module due to larger permeation area. The stage cut trend in two-stage configuration is shown in Figure 4.10 (a) and (b). This is because the permeation driving force increases at higher feed pressure and causes the passage of larger amounts of CO 2 to diffuse through the membrane. Basically, the purity of the permeate stream, expressed in terms of CO 2 removal, decreases at higher stage cuts. Therefore, as the stage cut is increased, the CO 2 permeate purity decreases (Ettouney and Majeed, 1997). However, the stage cut value decreases as the feed flow rate is further increased. This is due to the release of high-pressure gas in the retentate stream that reduces the permeation driving force. Stage cut of two-stage series configuration ranges from 0.05 x 10-4 to 1.68 x 10-4 for CH 4 and from 0.92 x 10-4 to x 10-4 for CO 2. Stage cut of treated membrane in two-stage membrane configuration also exhibited decreasing trend as a function of feed pressure. Stage cut values for membrane treated at 70 C for 2 min lie between 0.02 x 10-4 and 0.46 x 10-4 for CH 4 and between 0.44 x 10-4 and 8.74 x 10-4 for CO 2. For membrane treated at 70 C for 5 min the stage cut values varies from 0.02 x 10-4 to 0.35 x 10-4 for CH 4 and from 0.5 x 10-4 to 7.79 x 10-4 for CO 2. The decrease in stage cut values is probably due to polymer matrix densification that restricted the polymer structure movements as discussed previously. Basically, increasing the membrane stages in series will increase the stage cut while increasing membrane stages in cascade resulted in a decrease of stage cut. Between the three-stage series and cascade configurations studied, cascade configuration exhibited lower values of stage cut due to low pressure-normalized flux in permeate stream. In other words, CO 2 concentration in the permeate stream in cascade configuration is higher than that in the series configuration. Figure 4.11 (a) and (b) and Figure 4.12 (a) and (b) show the effect of CO 2 and CH 4 stage cut on feed pressure in three-stage series and cascade configuration for both untreated and treated membrane respectively.

132 Stage cut (x10-4 ) LPM-Untreated 10 LPM-2 min 10 LPM-5 min 20 LPM-Untreated 20 LPM-2 min 20 LPM-5 min 30 LPM-Untreated 30 LPM-2 min 30 LPM-5 min Feed pressure (bar) (a) 2.0 Stage cut (x10-4 ) LPM-Untreated 10 LPM-2 min 10 LPM-5 min 20 LPM-Untreated 20 LPM-2 min 20 LPM-5 min 30 LPM-Untreated 30 LPM-2 min 30 LPM-5 min Feed pressure (bar) (b) Figure 4.10: Effect of stage cut on feed pressure in two-stage series configuration untreated and treated membrane (a) CO 2 and (b) CH 4

133 Stage cut (x10-4 ) LPM-Untreated 10 LPM-2 min 10 LPM-5 min 20 LPM-Untreated 20 LPM-2 min 20 LPM-5 min 30 LPM-Untreated 30 LPM-2 min 30 LPM-5 min Feed pressure (bar) (a) 4 Stage cut (x10-4 ) LPM-Untreated 10 LPM-2 min 10 LPM-5 min 20 LPM-Untreated 20 LPM-2 min 20 LPM-5 min 30 LPM-Untreated 30 LPM-2 min 30 LPM-5 min Feed pressure (bar) (b) Figure 4.11: Effect of stage cut on feed pressure in three-stage series configuration untreated and treated membrane (a) CO 2 and (b) CH 4

134 Stage cut (x10-4 ) LPM-Untreated 10 LPM-2 min 10 LPM-5 min 20 LPM-Untreated 20 LPM-2 min 20 LPM-5 min 30 LPM-Untreated 30 LPM-2 min 30 LPM-5 min Feed pressure (bar) (a) 0.6 Stage cut (x10-4 ) LPM-Untreated 10 LPM-2 min 10 LPM-5 min 20 LPM-Untreated 20 LPM-2 min 20 LPM-5 min 30 LPM-Untreated 30 LPM-2 min 30 LPM-5 min Feed pressure (bar) (b) Figure 4.12: Effect of stage cut on feed pressure in three-stage cascade configuration untreated and treated membrane (a) CO 2 and (b) CH 4

135 Mixed Carbon Dioxide and Methane Gas Permeation Behavior in Untreated and Treated Membranes Pure gas permeation measurements have often been used to provide an indication of possible performance of various membranes under ideal conditions. But in actual case, the transport of a component in a gas mixture through glassy polymeric membranes is affected by the presence of other penetrants either due to the competition among the permeating species or by plasticization of the polymers if the mixture contains certain hydrocarbons and CO 2. In addition, the non-ideal gas behavior of CO 2 containing mixture and the concentration polarization also cause the variation of mixed and pure gas permeation characteristics. As a result, mixed gas separation generally yields lower selectivities for membranes than those of pure gas measurements. The effect of partial feed pressure on pressure-normalized flux of untreated and treated membrane in single-stage configuration for the individual gas species at 10, 20 and 30 LPM is shown in Figure 4.13 to Figure These figures combine both results obtained from experimental (E) and simulation work (S). The results given are based on 20% CO 2 concentration in the feed stream at different values of CO 2 partial feed pressures. The pressure-normalized flux of CO 2 does not show a distinct minimum but exhibited an immediate increase in the pressure-normalized flux with increasing CO 2 partial feed pressure at each increasing feed flow rate. However, pressure-normalized flux of CH 4 increased monotonically as a function of CO 2 partial feed pressure. Simulation results however showed slightly lower results since plasticization effect was not taken into consideration. Pressure-normalized flux of CO 2 increase at faster rate compared to CH 4 since CH 4 is removed at a much lower rate. This is because CO 2 permeate quickly compared to CH 4. Basically, membranes allow selective removal of fast gases from slow gases. Therefore, pressure-normalized flux of CO 2 showed stronger dependence on feed pressure. Similar results were obtained by previous researchers (Donohue et al., 1989, Ismail, 1992, Dortmundt and Doshi, 1999).

136 112 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.13: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 10 LPM Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.14: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 20 LPM

137 113 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.15: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 30 LPM As shown in the figures, pressure-normalized flux of CO 2 for membrane treated at 70 C for 2 min and 5 min also showed an increasing trend with increasing of CO 2 partial feed pressure. However, the pressure-normalized flux showed slightly lower values compared to the untreated membrane. This is due to more restricted polymer matrixes which lead to pressure-normalized flux reduction. The increased of pressure-normalized flux in untreated membrane is due to the weakening of polymer intermolecular attraction similar to that reported by Cao et al. (2003) which takes place inside the membrane or also known as plasticization phenomena. When this happened, gas diffusion through the membrane will increase since the polymer flexibility increases. As the CO 2 partial feed pressure is further increased, the CO 2 gas concentration will build up inside the membrane which causes changes in the polymer structure (Ismail and Lorna, 2002). In this study, the pressure-normalized flux of CO 2 did not show any decreasing trend but continue to increase within the CO 2 partial feed pressure operating range. Treated membrane showed lower values of pressure-normalized flux indicating that better polymer chain packing arrangement and helps minimized plasticization effect thus resulting in a more stable selectivity as shown in Figure 4.16 to Figure 4.18.

138 CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.16: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 10 LPM 30 CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.17: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 20 LPM

139 CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.18: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in single-stage configuration as a function of the feed pressure at 30 LPM The CO 2 /CH 4 selectivity of untreated membrane showed a decreasing trend with increasing CO 2 partial feed pressure. However, selectivity for the treated membranes showed higher selectivity values. Overall, the CO 2 /CH 4 selectivity at different feed flow rate showed similar trend. The increasing or decreasing selectivity trend exhibited is influenced by the CO 2 pressure-normalized flux behavior discussed previously. However, pressure-normalized flux value of CO 2 increase with increasing of feed flow rate as a function of CO 2 partial feed pressure. This is because of larger amount of feed gas permeated through the membrane as the feed flow rate is increased. The higher the feed flow rate, the shorter the residence time will be. The difference in the transport resistance of the gas species within the membrane is also one of the contributing factors. The effect of stage cut for CO 2 /CH 4 gas mixture with CO 2 partial feed pressure is shown in Figure 4.19 to Figure The stage cut increased as a function of feed pressure but do not show much difference as the feed flow rate is further increases. The stage cut values lies in the same region as the feed flow rate is

140 Stage cut (x 10-3 ) Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.19: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in singlestage configuration of untreated and treated membrane at 10 LPM 2.0 Stage cut (x 10-3 ) Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.20: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in singlestage configuration of untreated and treated membrane at 20 LPM

141 Stage cut (x 10-3 ) Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.21: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in singlestage configuration of untreated and treated membrane at 30 LPM changed from 10 LPM to 30 LPM. However, it is obvious that the stage cut of membrane treated at 70 C for 5 min showed the lowest values experimentally as well as in simulation results. The experimental and simulation result of stage cut in single-stage configuration showed an almost similar curve. Reduction in stage cut is in agreement with the explanation mentioned above. Pressure-normalized flux of CO 2 versus CO 2 partial feed pressure of untreated and treated membrane in two-stage series configuration revealed an increasing trend as the feed flow rate is raised from 10 LPM to 30 LPM as shown in Figure 4.22 to Figure The increasing of CO 2 pressure-normalized flux trend is similar to those exhibited in single-stage configuration. However, the pressurenormalized flux value of CO 2 is higher due to an increase in membrane area. This is because addition of membrane area provides larger permeation surfaces for the gases to be separated once introduced to the membrane. As a result, more gases are permeated through the membrane at the same time.

142 118 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.22: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 10 LPM Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.23: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 20 LPM

143 119 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.24: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 30 LPM It is clear from the figures that the pressure-normalized flux of CO 2 for untreated membrane showed higher values and the pressure-normalized flux values of CO 2 for membrane treated at 70 C for 2 min and 5 min gradually decreases with increasing of CO 2 partial feed pressure. Simulation results for pressure-normalized flux of CO 2 also showed the same decreasing trend in the sequence of untreated membrane, membrane treated at 70 C for 2 min followed by membrane treated at 70 C for 5 min. Simulation results exhibited lower pressure-normalized flux values of CO 2 compared to experimental results. This is probably due to CO 2 -induced plasticization is not taken into consideration as mention above. On the other hand, pressure-normalized flux of CH 4 remains constant throughout the experiment. Figure 4.25 to Figure 4.27 shows the CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at feed flow rate of 10, 20 and 30 LPM, respectively. The selectivity trend revealed is influenced by pressurenormalized flux of CO 2 discussed above. Untreated membrane showed fluctuated

144 CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.25: CO 2 / CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 10 LPM 30 CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.26: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 20 LPM

145 CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.27: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in two-stage series configuration as a function of the feed pressure at 30 LPM selectivity trend with lower selectivity values compared to treated membranes. The instability in the selectivity trend obtained experimentally showed is due to the presence of plasticization effect. Simulation results showed a constant selectivity trend in both untreated and treated since the effect of plasticization was not taken into consideration. According to Ismail and Lorna (2002), the significant difference is most probably due to the alterations in the molecular morphology of the asymmetric hollow fiber membrane. It is believed that the decreasing selectivity behavior in two-stage membrane configuration for untreated membrane is influenced by the loosen polymer chain packing while the constant selectivity trend of treated membrane is due to better polymer chain packing and alignment. The stage cut versus feed pressure plot for two-stage series configuration revealed an increasing trend as shown in Figure 4.28 to Figure Similar trend is shown in single-stage configuration but the stage cut values in two-stage series configuration is 2.5 times higher. As the feed flow rate is further increased the stage

146 122 5 Stgae cut (x10-3 ) Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.28: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in twostage series configuration of untreated and treated membrane at 10 LPM 4 Stage cut (x10-3 ) Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.29: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in twostage series configuration of untreated and treated membrane at 20 LPM

147 123 4 Stage cut (x10-3 ) Untreated (E) 2 min (E) 5 mim (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.30: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in twostage series configuration of untreated and treated membrane at 30 LPM cut values decreased slightly. This is because of more gases were released in the retentate stream as the feed flow rate is further increased and causes the permeate flow rate to reduce due to slight pressure drop in the second membrane module. Since stage cut is the ratio of permeate flow rate to feed flow rate, the stage cut values will be reduced as the permeate flow rate decreased. It is clear that plots obtained from experimental results showed that untreated membrane revealed higher stage cut followed by membrane treated at 2 min and 5 min. Simulation results also exhibited similar trend in the same region. Untreated membrane showed higher stage cut due to the flexibility of the polymer matrix in the membrane as well as the swelling phenomena that took place during gas permeation testing. On the other hand, the polymer structure of the treated membrane became more rigid and immobile as the heating duration is further raised. The rigidity of the membrane is probably due to physical entanglements which effectively prevent sideways motion of the chain (Sperling, 1992) as well as the polymer chains are very close to each other (Staudt-Bickel and Koros, 1999). No chemical entanglements were found since both untreated and treated membrane still readily dissolved in DMAc and THF

148 124 solvents after a 5 min heat treatment at 70 C indicating that the heat treatments were not intense enough to cause any chemical entanglements. Similar result was reported by Krol et al. (2001). The alteration took place enhances the polymer structure and prevent swelling of the polymer membrane which resulted in lowering the gas diffusion through the membrane as well as lowering the stage cut. Pressure-normalized flux in three-stage series configuration showed an increasing trend as a function of feed pressure. Even though the pressure-normalized flux increase but the trend is a quite different from single and two-stage series configuration. In the first two configurations, the pressure-normalized flux increase in a linear manner. However, pressure-normalized flux in Figure 4.31 to Figure 4.33 increases linearly until feed pressure of 8 bars before increases drastically when operated at feed pressure of 10 bars. The pressure-normalized flux increases as the feed flow rate is further increased. Compared to single and two-stage series configuration, three-stage series configuration recorded the highest pressure-normalized flux values due to an increase in the number of stages or membrane areas. The pressure-normalized flux increases from 100 GPU at feed flow rate of 10 LPM to 200 GPU at feed flow rate of 30 LPM. The increase of pressure-normalized flux is due to the plasticization phenomena. According to Wessling et al.(2001), for highly sorbing gases such as CO 2, the pressure-normalized flux increases with increasing feed pressure as a result of stronger plasticization effects due to build up of high CO 2 concentration in the glassy polymer. Above the plasticization pressure, the pressure-normalized flux is no longer constant but continues to increase with time. The increase of CO 2 in gas mixtures indicates significant plasticization of the glassy polymer. As a result, the pressure-normalized flux of the product to be rejected (CH 4 ) also increases resulting in undesired product loss thereof making the separation process less economic.

149 125 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.31: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 10 LPM Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.32: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 20 LPM

150 126 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.33: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 30 LPM. The increase behavior of the pressure-normalized flux at different feed flow rates influences the outcome selectivity. The selectivity of the untreated membrane revealed a fluctuated trend throughout the operating feed pressures as the feed flow rate is further increases. This is due to the alteration done by the highly sorbing CO 2. However, selectivity for treated membranes showed a more stable or constant trend indicating that the heat treatment enhanced polymer matrix chain packing. Even though the pressure-normalized flux increases with time but the plasticization effect is less pronounced than in the untreated membrane. The selectivity versus feed pressure plot of three-stage series configuration is shown in Figure 4.34 to Figure 4.36.

151 CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.34: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 10 LPM 30 CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.35: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 20 LPM

152 CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.36: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage series configuration as a function of the feed pressure at 30 LPM The stage cut in three-stage series configuration showed similar trend as discussed previously for single and two-stage series configuration and is shown in Figure 4.37 to Figure Experimental result agrees with simulation results since the plot falls in almost the same region. The stage cut increased with increasing of feed pressure but the stage cut values is larger than the stage cut values in single and two-stage series configuration due to increase in permeation area which reduces the permeation resistance and results in the highest stage cut. Similar result is reported by Ettouney et al. (1998). However, the stage cut value decreases as the feed flow rate increases. The decreased in stage cut values is more significant than in singleand two-stage series configurations. This is probably because at higher flow rates, the gas residence time is lower, which increases the permeation resistance and as a result the stage cut decreased more severe than in the two previous configuration mentioned above. The lower stage cut value showed by treated membrane is most probably due to more restricted polymer chain packing.

153 Stage cut (x10-3 ) Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.37: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in threestage series configuration of untreated and treated membrane at 10 LPM 10 8 Stage cut (x10-3 ) Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed presure (bar) Figure 4.38: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in threestage series configuration of untreated and treated membrane at 20 LPM

154 130 8 Stage Cut (x10-3 ) Untreated (E) 2 min (E) 3 min (E) Untreated (S) 2 min (S) 5 min (S) Feed Pressure (bar) Figure 4.39: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in threestage series configuration of untreated and treated membrane at 30 LPM Similar as shown in three-stage series configuration, pressure-normalized flux of CO 2 exhibited an increasing trend as a function of CO 2 partial feed pressure in three-stage cascades configuration as shown in Figure 4.40 to Figure However, these figures showed reduction in the pressure-normalized flux values. At 10 LPM, the pressure-normalized flux of CO 2 reduced to almost 50%. The reduction in CO 2 pressure-normalized flux at 20 LPM and 30 LPM is also as extreme as the reduction at 10 LPM. The difference is resulted from the arrangement of the membrane modules. For this particular configuration, most gases were separated in the first and second membrane module. Severe plasticization occurred in the first membrane module compared to second membrane module due to slight pressure drop in the gas mixture entering the second membrane module. Since cascade arrangement is applied, retentate stream from the second membrane module is no longer used as the feed stream to the third membrane module. Instead, permeate stream is used as the feed stream. Permeate stream entering the third membrane module at atmospheric pressure with less CO 2 content lead to the assumption of the membrane to be plasticization free. Different from series configuration, the total permeate flow rate in cascade configuration is obtained from the first and second membrane module.

155 131 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.40: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascade configuration as a function of the feed pressure at 10 LPM Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.41: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascade configuration as a function of the feed pressure at 20 LPM

156 132 Pressure-normalized flux (GPU) Feed pressure (bar) CO2-Untreated (E) CO2-2 min (E) CO2-5 min (E) CO2-Untreated (S) CO2-2 min (S) CO2-5 min (S) CH4-Untreated (E) CH4-2 min (E) CH4-5 min (E) CH4-Untreated (S) CH4-2 min (S) CH4-5 min (S) Figure 4.42: Pressure-normalized flux of CO 2 /CH 4 gas mixture of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascade configuration as a function of the feed pressure at 30 LPM Since the third permeate stream is feed back into the system, the total permeate flow rate is lower compared to series configuration. As a result, the pressure-normalized flux of CO 2 in cascade configuration decreases more than in series configuration. Pressure-normalized flux of CH 4 increased monotonically as the CO 2 partial feed pressure is further increased. The reduction in the pressure-normalized flux resulted in higher selectivity values. Figure 4.43 to Figure 4.45 show the CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascades configuration as a function of the feed pressure at 10, 20 and 30 LPM. From experimental work, selectivity of membrane treated at 70 C for 5 min at 10 LPM exhibited highest selectivity value with a constant trend. Selectivity for membrane treated at 70 C for 2 min showed slightly lower selectivity value and the lowest selectivity is measured for untreated membrane revealing fluctuated trend.

157 CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.43: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascade configuration as a function of the feed pressure at 10 LPM CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (S) Untreated (E) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.44: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascade configuration as a function of the feed pressure at 20 LPM

158 CO2/CH4 selectivity Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.45: CO 2 /CH 4 selectivity of untreated and treated asymmetric polysulfone hollow fiber membrane in three-stage cascade configuration as a function of the feed pressure at 30 LPM Selectivity obtained from simulation showed lower selectivity compared to the experimental results. Constant trend showed is probably because plasticization is not taken into consideration. As mentioned by Ettouney and Majeed (1997), higher selectivity is attributed to the free volume distribution in the skin layer of the asymmetric membrane. At feed flow rate of 20 LPM and 30 LPM, the selectivity of untreated and treated membrane obtained experimentally exhibited fluctuated trend as a function of CO 2 partial feed pressure. Membrane treated at 70 C for 5 min still showed the highest selectivity but the selectivity showed decreased trend. Selectivity from simulation result remained constant throughout the experiment. According to Bos et al. (2001), one must not draw conclusions from such decreasing selectivity trends with increasing feed pressure in the plasticization behavior. Below the plasticization pressure, both CO 2 and CH 4 pressure-normalized flux show a decreasing pressure-normalized flux with increasing feed pressure. However, CO 2 generally has larger decreases than CH 4, resulting in decreasing selectivity. The stage cut versus feed pressure plot exhibited increasing trend at different feed flow rate in three-stage cascade configuration and are shown in Figure 4.46 to Figure 4.48.

159 Stage cut (x 10-3 ) Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.46: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in threestage cascade configuration of untreated and treated membrane at 10 LPM 6 5 Stage cut (x 10-3 ) Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.47: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in threestage cascade configuration of untreated and treated membrane at 20 LPM

160 136 5 Stage cut (x 10-3 ) Untreated (E) 2 min (E) 5 min (E) Untreated (S) 2 min (S) 5 min (S) Feed pressure (bar) Figure 4.48: Effect of stage cut for CO 2 /CH 4 gas mixture on feed pressure in threestage cascade configuration of untreated and treated membrane at 30 LPM Decreased in the CO 2 pressure-normalized flux influenced the stage cut behavior to decrease to about half of the stage cut value showed in three-stage series configuration. Simulation result also showed the same stage cut trend. Based on the four different membrane configuration discussed above, threestage cascade configuration using membrane treated at 70 C for 5 min posses good separation performances at feed flow rate of 20 LPM with CO 2 /CH 4 selectivity of 32 and a CO 2 pressure-normalized flux of 12 GPU when operated at feed pressure of 2 bars. However, this configuration has quite low pressure-normalized flux. At feed pressure of 2 bars to 6 bars, the selectivities obtained are slightly above the intrinsic selectivity of dense polysulfone of about 28. On the other hand, three-stage series configuration revealed the lowest selectivity compared to the other membrane configurations but with higher pressure-normalized flux values. The selectivity lies between 21 and 23. However, selectivity for two-stage series configuration reached up to 25 with almost the same value pressure-normalized flux of a three-stage cascade configuration. The selectivity of a single-stage configuration is much more stable in the range of 22 to 24.

161 Effect of Heat Treatment on Fiber Density Density measurements were performed on untreated and treated asymmetric polysulfone hollow fiber membranes. The purpose of this measurement is to examine the possible changes in the packing density of the polymer chains to be compared with the untreated membranes. Table 4.3 summarizes the fiber density of untreated and treated asymmetric polysulfone hollow fiber membrane. The density of the untreated membrane is in the range of g/cm 3 to g/cm 3 while the density of treated membrane ranges from g/cm 3 to g/cm 3. The standard deviation varies from g/cm 3 to g/cm 3. The increasing trend of fiber density as a function of heating duration is illustrated in Figure This is due to the restructuring of the polymer chains that took place during heat treatment. According to Tadmor and Gogos (1979), the rigid random coil chains of the amorphous polysulfone became flexible as the membrane was heat treated. As a result, heat treatment led to a more perfect chain packing arrangement as well as decreased the free volume in the polymer. Heat treatment caused densification of the polymer matrix. These results explained the reduction in the pressure-normalized flux of the treated membranes. The same behavior was observed by Krol et al. (2001) and Ismail and Lorna (2003). Table 4.3: Fiber density of untreated and treated asymmetric polysulfone hollow fiber membrane. Membrane type Average fiber density (g/cm 3 ) Untreated ± C for 2 min ± C for 5 min ±

162 Fiber density (g/cm 3 ) Untreated 70C for 2 min 70C for 5 min Membrane type Figure 4.49: Fiber density of the untreated and treated asymmetric polysulfone hollow fiber membrane 4.5 Effect of Heat Treatment Process on the Polymer Glass Transition Temperature Macromolecules can be tailored to control interchain displacements and segmental mobilities. Likewise, the chains of a given polymer can be packed to various extents to control density and pressure-normalized flux. Such control can be affected by thermal means. At 50 C below the T g, segmental mobility is very restricted, so that interchain displacements are fixed. Diffusive selectivity is based on the inherent ability of polymer matrices to function as size and shape-selective media. This ability is primarily determined by such factors as polymer segmental mobility and intersegmental packing. This kinetic sieve model suffices to account for the principal features of many of the important gas separations, including those involving H 2 /N 2, H 2 / hydrocarbons and O 2 /N 2 (Kesting and Fritzshe, 1993). The T g of the untreated and treated asymmetric polysulfone hollow fiber membrane used in this study is determined using DSC. The effect of heat treatment on the T g of asymmetric polysulfone hollow fiber membrane is shown in Figure 4.50.

163 Temperature (degree Celcius) Untreated 70C for 2 min 70C for 5 min Membrane type Figure 4.50: Effect of heat treatment on the glass transition temperature of asymmetric polysulfone hollow fiber membrane The midpoint T g for the untreated asymmetric polysulfone hollow fiber membrane is C. The midpoint T g of the membrane treated at 70 C for 2 min and 5 min is C and C, respectively. From the above figure, the increase in midpoint T g indicated that heat treatment had made alterations in the polymer chains structure. Heat treatment causes relaxation of the polymer chains that enable more chain segments to approach a condition of a more efficient packing as well as reduction in the free volume distribution. The polymeric chain became more rigid and immobile. This can be interpreted as a restriction or reduction of segments rotation around the main chain bond which causes the midpoint T g of the treated membrane to increase higher than the untreated membranes. Similar results were obtained by previous researchers (Mulder, 1990, Kesting and Fritzshe, 1993, Bos et al., 1998, Wong, 2002).

164 Morphology of the Develop Membrane Before and After Heat Treatment Process Scanning Electron Microscopy (SEM) pictures of cross-section were used to determine whether heat treatment caused compaction and densification in the outer skin and substructure of the hollow fiber membrane. The results will be used to compare with the untreated membranes. In order to observe the morphology of hollow fiber membranes, the dried fibers were broken in liquid nitrogen and then sputtered with a thin layer of gold. The cross section and surface of the hollow fiber were confirmed through SEM. SEM studies were performed on the asymmetric polysulfone hollow fiber membranes that were given different heat treatments. Figure 4.51 and Figure 4.52 compares SEM pictures of the partial cross section and skin surface of untreated and treated polysulfone hollow fiber membranes spun from 33 wt. % polymer dopes by dry/wet process at magnification of x250 and x 500. The structure of the produced hollow fiber membranes showed asymmetric structure; a dense top layer supported by a porous, spongy substructure. The partial cross section figures clearly reveal a thin skin layer with many teardrops or finger microvoids near the membrane surface in both untreated and treated membrane. This was due to the high dope extrusion rate used to spin the hollow fiber membrane. The same membrane morphology was reported by Sharpe et al. (1999). The presence of teardrops or finger microvoids in the treated membrane of the porous substructure region was clearly smaller when the 2 min treatment was carried out. The structure changed even more when the heat treatment was prolonged to 5 min. The reduction in the size of teardrops or finger microvoids is very obvious near the inner skin and this can be seen even at the magnification of x250. Similar result was also reported by Krol et al. (2001). From the figures, heat treatment conducted below the T g of polysulfone densified the membrane skin and this was confirmed with the calculation based on theory proposed by Ismail and Lai (2004) using the measured pressure-normalized flux and the intrinsic CO 2 and CH 4 pressure-normalized flux as determined for a dense homogeneous polysulfone membrane (Appendix F). From the calculation, it was found that as skin layer thickness increases, mean pore size and surface porosity decreases.

165 141 (a1) (a2) (b1) (b2) (c1) (c2) Figure 4.51: Scanning electron microscopy pictures of asymmetric polysulfone hollow fiber membrane before and after different heat treatments. Left: partial cross section of membrane, right: skin surface (x250). (a1 and a2: untreated membrane, b1 and b2: treated membrane at 70 C for 2 min, c1 and c2: treated membrane at 70 C for 5 min.)

166 142 (d1) (d2) (e1) (e2) (f1) (f2) Figure 4.52: Scanning electron microscopy pictures of asymmetric polysulfone hollow fiber membrane before and after different heat treatments. Left: partial cross section of membrane, right: skin surface (x500). (d1 and d2: untreated membrane, e1 and e2: treated membrane at 70 C for 2 min, f1 and f2: treated membrane at 70 C for 5 min.)

BORANG PENGESAHAN STATUS TESIS

BORANG PENGESAHAN STATUS TESIS KOLEJ UNIVERSITI KEJURUTERAAN & TEKNOLOGI MALAYSIA BORANG PENGESAHAN STATUS TESIS JUDUL: CAR LOCK INDICATOR SESI PENGAJIAN: 2005/2006 Saya HAIRULAZAM BIN HAIRULDIN (830523-08-5369) (HURUF BESAR) mengaku

More information

BIOMECHANICAL LOADING OF INSTEP KICK FOR MALAYSIAN FOOTBALLER DAYANG KHAIRUNNISA BINTI ABANG KIPRAWI

BIOMECHANICAL LOADING OF INSTEP KICK FOR MALAYSIAN FOOTBALLER DAYANG KHAIRUNNISA BINTI ABANG KIPRAWI BIOMECHANICAL LOADING OF INSTEP KICK FOR MALAYSIAN FOOTBALLER DAYANG KHAIRUNNISA BINTI ABANG KIPRAWI Report submitted in partial fulfillment of the requirements for the award of Bachelor of Mechanical

More information

INTERNAL SURFACE PIPE ROUGHNESS CLASSIFICATION USING HIGH FREQUENCY ACOUSTIC EVALUATION

INTERNAL SURFACE PIPE ROUGHNESS CLASSIFICATION USING HIGH FREQUENCY ACOUSTIC EVALUATION INTERNAL SURFACE PIPE ROUGHNESS CLASSIFICATION USING HIGH FREQUENCY ACOUSTIC EVALUATION AMAR REZA BIN MOHAMMAD FIRDAUS BACHELOR OF ENGINEERING UNIVERSITI MALAYSIA PAHANG 2010 AMAR REZA BACHELOR OF MECHANICAL

More information

UNIVERSITI TEKNIKAL MALAYSIA MELAKA

UNIVERSITI TEKNIKAL MALAYSIA MELAKA UNIVERSITI TEKNIKAL MALAYSIA MELAKA MECHANISM DEVELPOMENT OF GRIPPER FOR 5KG PAYLOAD This report submitted in accordance with requirement of the Universiti Teknikal Malaysia Melaka (UTeM) for the Bachelor

More information

Faculty of Engineering

Faculty of Engineering Faculty of Engineering A PRELIMINARY SURVEY OF COMMON MAINTENANCE AND REHABILITATION PROCEDURE: CASE STUDY FOR KUCHING-SAMARAHAN ROAD Mike Jackson Tsai Bachelor of Engineering with Honours (Civil Engineering)

More information

Faculty of Engineering

Faculty of Engineering Faculty of Engineering DETERMINATION OF SPILLWAY CAPACITY OF PROPOSED BENGOH DAM John Gary anak Jentry Bachelor of Engineering with Honours (Civil Engineering) 2009 UNIVERSITI MALAYSIA SARAWAK BORANG PENGESAHAN

More information

Recovery of helium from rejected gas streams in Natural gas processing [1, 2, 3].

Recovery of helium from rejected gas streams in Natural gas processing [1, 2, 3]. Gas Separation by Membranes Membrane technology in gas separation and purification has grown exponentially since they were first introduced about 30 years ago. Compared to other conventional gas separation

More information

Instructor: Dr. Istadi (http://tekim.undip.ac.id/staf/istadi )

Instructor: Dr. Istadi (http://tekim.undip.ac.id/staf/istadi ) Instructor: Dr. Istadi (http://tekim.undip.ac.id/staf/istadi ) Email: istadi@undip.ac.id Course Syllabus: (Part 1) 1. Definitions of Natural Gas, Gas Reservoir, Gas Drilling and Gas production (Pengertian

More information

A SYSTEM DYNAMIC MODEL FOR DECISION SUPPORT SYSTEM IN LEAN CONSTRUCTION ALI CHEGENI

A SYSTEM DYNAMIC MODEL FOR DECISION SUPPORT SYSTEM IN LEAN CONSTRUCTION ALI CHEGENI A SYSTEM DYNAMIC MODEL FOR DECISION SUPPORT SYSTEM IN LEAN CONSTRUCTION ALI CHEGENI A project report submitted in fulfilment of the Requirements for the award of the degree of Master of Engineering (Industrial

More information

UNIVERSITI TEKNIKAL MALAYSIA MELAKA

UNIVERSITI TEKNIKAL MALAYSIA MELAKA UNIVERSITI TEKNIKAL MALAYSIA MELAKA DESIGN AND DEVELOP AN ECO FLY KILLER IN UTeM s CAFETERIA This report submitted in accordance with requirement of the Universiti Teknikal Malaysia Melaka (UTeM) for the

More information

ANALYSIS OF STROKE DISTRIBUTION BETWEEN PROFESSIONAL, INTERMEDIATE AND NOVICE SQUASH PLAYERS DIYANA ZULAIKA BINTI ABDUL GHANI

ANALYSIS OF STROKE DISTRIBUTION BETWEEN PROFESSIONAL, INTERMEDIATE AND NOVICE SQUASH PLAYERS DIYANA ZULAIKA BINTI ABDUL GHANI i ANALYSIS OF STROKE DISTRIBUTION BETWEEN PROFESSIONAL, INTERMEDIATE AND NOVICE SQUASH PLAYERS DIYANA ZULAIKA BINTI ABDUL GHANI A thesis submitted in fulfilment of the requirements for the award of the

More information

THE EFFECT OF OVERSIZED LANE WIDTH AND LANE SHOULDER ON HEAVY VEHICLE PARKING ON RESIDENTIAL STREETS NURAIN BINTI MOHD SITH

THE EFFECT OF OVERSIZED LANE WIDTH AND LANE SHOULDER ON HEAVY VEHICLE PARKING ON RESIDENTIAL STREETS NURAIN BINTI MOHD SITH THE EFFECT OF OVERSIZED LANE WIDTH AND LANE SHOULDER ON HEAVY VEHICLE PARKING ON RESIDENTIAL STREETS NURAIN BINTI MOHD SITH A thesis submitted in fulfilment of the requirements for the award of the degree

More information

BLOOD PRESSURE MONITORING SYSTEM USING PIEZOELECTRIC FILM SENSOR NORSYILA BINTI BAHARUDIN

BLOOD PRESSURE MONITORING SYSTEM USING PIEZOELECTRIC FILM SENSOR NORSYILA BINTI BAHARUDIN i BLOOD PRESSURE MONITORING SYSTEM USING PIEZOELECTRIC FILM SENSOR NORSYILA BINTI BAHARUDIN UNIVERSITI TEKNIKAL MALAYSIA MELAKA ii BLOOD PRESSURE MONITORING SYSTEM USING PIEZOELECTRIC FILM SENSOR NORSYILA

More information

Preparation and Gas Separation Properties of Asymmetric Polysulfone Membranes by a Dual Bath Method

Preparation and Gas Separation Properties of Asymmetric Polysulfone Membranes by a Dual Bath Method Korean J. Chem. Eng., 17(2), 143-148 (2000) Preparation and Gas Separation Properties of Asymmetric Polysulfone Membranes by a Dual Bath Method Wan-Jin Lee, Deuk-San Kim and Jin-Hwan Kim* Faculty of Applied

More information

CTB3365x Introduction to Water Treatment

CTB3365x Introduction to Water Treatment CTB3365x Introduction to Water Treatment D4b Aeration Doris van Halem Did you know that there are not just gasses in your glass of sparkling coke, but also in the tap water you drink? Welcome to the water

More information

ASSOCIATION BETWEEN ERGONOMIC RISK FACTORS AND MUSCULOSKELETAL PAIN AMONG FIXIE BIKE CYCLISTS MOHD SHAMSHEMUN BIN MOHAMED

ASSOCIATION BETWEEN ERGONOMIC RISK FACTORS AND MUSCULOSKELETAL PAIN AMONG FIXIE BIKE CYCLISTS MOHD SHAMSHEMUN BIN MOHAMED ASSOCIATION BETWEEN ERGONOMIC RISK FACTORS AND MUSCULOSKELETAL PAIN AMONG FIXIE BIKE CYCLISTS MOHD SHAMSHEMUN BIN MOHAMED A thesis submitted in fulfilment of the requirements for the award of the degree

More information

Title: Choosing the right nitrogen rejection scheme

Title: Choosing the right nitrogen rejection scheme Title: Choosing the right nitrogen rejection scheme Authors: Nicolas Chantant, Paul Terrien, Sylvain Gérard (Air Liquide Global E&C Solutions) Abstract: Nitrogen Rejection Units (NRU) are used to extract

More information

INTEGRATED FARE PAYMENT SYSTEM IN MULTI OPERATORS SINGLE MARKET PUBLIC BUS NETWORK SAFIZAHANIN BINTI MOKHTAR UNIVERSITI TEKNOLOGI MALAYSIA

INTEGRATED FARE PAYMENT SYSTEM IN MULTI OPERATORS SINGLE MARKET PUBLIC BUS NETWORK SAFIZAHANIN BINTI MOKHTAR UNIVERSITI TEKNOLOGI MALAYSIA INTEGRATED FARE PAYMENT SYSTEM IN MULTI OPERATORS SINGLE MARKET PUBLIC BUS NETWORK SAFIZAHANIN BINTI MOKHTAR UNIVERSITI TEKNOLOGI MALAYSIA FEBRUARY 2011 INTEGRATED FARE PAYMENT SYSTEM IN MULTI OPERATORS

More information

RESIDENTIAL HOUSING DEVELOPMENT IN KURDISTAN REGION GOVERNMENT OF IRAQI FEDERAL

RESIDENTIAL HOUSING DEVELOPMENT IN KURDISTAN REGION GOVERNMENT OF IRAQI FEDERAL RESIDENTIAL HOUSING DEVELOPMENT IN KURDISTAN REGION GOVERNMENT OF IRAQI FEDERAL HAVAL MOHAMMED SALIH UNIVERSITI TEKNOLOGI MALAYSIA RESIDENTIAL HOUSING DEVELOPMENT IN KURDISTAN REGION GOVERNMENT OF IRAQI

More information

VALIDATING AND DEVELOPING A NEW AGILITY TEST FOR KARATE

VALIDATING AND DEVELOPING A NEW AGILITY TEST FOR KARATE VALIDATING AND DEVELOPING A NEW AGILITY TEST FOR KARATE By MOHAMMAD EBRAHIM MARJANI Thesis Submitted to the School of Graduate Studies, Universiti Putra Malaysia, in Fulfilment of the Requirements for

More information

You should be able to: Describe Equipment Barometer Manometer. 5.1 Pressure Read and outline 5.1 Define Barometer

You should be able to: Describe Equipment Barometer Manometer. 5.1 Pressure Read and outline 5.1 Define Barometer A P CHEMISTRY - Unit 5: Gases Unit 5: Gases Gases are distinguished from other forms of matter, not only by their power of indefinite expansion so as to fill any vessel, however large, and by the great

More information

FlashCO2, CO2 at 23 $/ton

FlashCO2, CO2 at 23 $/ton FlashCO2, CO2 at 23 $/ton A cost effective solution of capturing CO2 from Steam Methane Reforming (SMR) Hydrogen production plants by the FlashCO2 process Introduction to a cost effective solution Driven

More information

Chapter 8: Reservoir Mechanics

Chapter 8: Reservoir Mechanics PTRT 1472: Petroleum Data Management II Chapter 8: Reservoir Mechanics - Reservoir drives Types of Natural Gas Reservoir Fluids Natural gas is petroleum in a gaseous state, so it is always accompanied

More information

Figure Vapor-liquid equilibrium for a binary mixture. The dashed lines show the equilibrium compositions.

Figure Vapor-liquid equilibrium for a binary mixture. The dashed lines show the equilibrium compositions. Another way to view this problem is to say that the final volume contains V m 3 of alcohol at 5.93 kpa and 20 C V m 3 of air at 94.07 kpa and 20 C V m 3 of air plus alcohol at 100 kpa and 20 C Thus, the

More information

DISTILLATION POINTS TO REMEMBER

DISTILLATION POINTS TO REMEMBER DISTILLATION POINTS TO REMEMBER 1. Distillation columns carry out physical separation of liquid chemical components from a mixture by a. A combination of transfer of heat energy (to vaporize lighter components)

More information

KAJIAN BIOLOGI PEMBIAKAN IKAN KELAH (Tor tambra) DI SUNGAI MEUREUBO, ACEH BARAT, INDONESIA BAIHAQI

KAJIAN BIOLOGI PEMBIAKAN IKAN KELAH (Tor tambra) DI SUNGAI MEUREUBO, ACEH BARAT, INDONESIA BAIHAQI KAJIAN BIOLOGI PEMBIAKAN IKAN KELAH (Tor tambra) DI SUNGAI MEUREUBO, ACEH BARAT, INDONESIA MASTER SAINS UNIVERSITI MALAYSIA TERENGGANU 2015 KAJIAN BIOLOGI PEMBIAKAN IKAN KELAH (Tor tambra) DI SUNGAI MEUREUBO,

More information

Laser Spectrometers for Online Moisture Measurement in Natural Gas. Ken Soleyn GE M&C

Laser Spectrometers for Online Moisture Measurement in Natural Gas. Ken Soleyn GE M&C Laser Spectrometers for Online Moisture Measurement in Natural Gas Ken Soleyn GE M&C ken.soleyn@ge.com Introduction TDLAS (Tunable Diode Lase Absorption Spectroscopy) Moisture Analyzers have become the

More information

Nama Penilai. No. K/Pengenalan. Tarikh Terima :. Tarikh Siap :... No. Telefon : No. Faksimili :.. 1. Adakah manuskrip ini merupakan karya asli?

Nama Penilai. No. K/Pengenalan. Tarikh Terima :. Tarikh Siap :... No. Telefon : No. Faksimili :.. 1. Adakah manuskrip ini merupakan karya asli? OPERASI PERKHIDMATAN SOKONGAN PENERBIT UPM Kod Dokumen: OPR/PUPM/BR02/PENILAIAN MANUSKRIP BORANG PENILAIAN MANUSKRIP Tajuk Manuskrip :... Nama Penilai No. K/Pengenalan Jawatan Alamat :... :... :... :......

More information

Removing nitrogen. Nitrogen rejection applications can be divided into two categories

Removing nitrogen. Nitrogen rejection applications can be divided into two categories Removing nitrogen Doug MacKenzie, Ilie Cheta and Darryl Burns, Gas Liquids Engineering, Canada, present a comparative study of four nitrogen removal processes. Nitrogen rejection applications can be divided

More information

Gilbert Kirss Foster. Chapter 10. Properties of Gases The Air We Breathe

Gilbert Kirss Foster. Chapter 10. Properties of Gases The Air We Breathe Gilbert Kirss Foster Chapter 10 Properties of Gases The Air We Breathe Chapter Outline 10.1 The Properties of Gases 10.2 Effusion and the Kinetic Molecular Theory of Gases 10.3 Atmospheric Pressure 10.4

More information

Membrane modules for nitrogen and oxygen generator systems. Technology Overview ENGINEERING YOUR SUCCESS.

Membrane modules for nitrogen and oxygen generator systems. Technology Overview ENGINEERING YOUR SUCCESS. Membrane modules for nitrogen and oxygen generator systems Technology Overview ENGINEERING YOUR SUCCESS. Parker modules the heart of OEM tailor-made nitrogen generators OEM (Original Equipment Manufacturer)

More information

NITROGEN GENERATION FOR INDUSTRIAL APPLICATIONS

NITROGEN GENERATION FOR INDUSTRIAL APPLICATIONS TRIDENT NOTES: NUMBER 5 DECEMBER 2017 NITROGEN GENERATION FOR INDUSTRIAL APPLICATIONS Industry requires nitrogen Dozens of gases are used by industry. First among these in terms of quantity consumed is

More information

Classes at: - Topic: Gaseous State

Classes at: - Topic: Gaseous State PHYSICAL CHEMISTRY by: SHAILENDRA KR. Classes at: - SCIENCE TUTORIALS; Opp. Khuda Baksh Library, Ashok Rajpath, Patna PIN POINT STUDY CIRCLE; House No. 5A/65, Opp. Mahual Kothi, Alpana Market, Patna Topic:

More information

Chem 110 General Principles of Chemistry

Chem 110 General Principles of Chemistry CHEM110 Worksheet - Gases Chem 110 General Principles of Chemistry Chapter 9 Gases (pages 337-373) In this chapter we - first contrast gases with liquids and solids and then discuss gas pressure. - review

More information

THE DYNAMICS OF CURRENT CIRCULATION AT NEARSHORE AND VICINITY OF ISLAND IN TERENGGANU WATERS NURUL RABITAH BINTI DAUD

THE DYNAMICS OF CURRENT CIRCULATION AT NEARSHORE AND VICINITY OF ISLAND IN TERENGGANU WATERS NURUL RABITAH BINTI DAUD THE DYNAMICS OF CURRENT CIRCULATION AT NEARSHORE AND VICINITY OF ISLAND IN TERENGGANU WATERS NURUL RABITAH BINTI DAUD Thesis Submitted in Fulfillment of the Requirement for the Degree of Master of Science

More information

Optimization of Separator Train in Oil Industry

Optimization of Separator Train in Oil Industry Optimization of Separator Train in Oil Industry Pawan jain a M.E. Petroleum Engineering, Maharashtra Institute of Technology, Pune-411038 ---------------------------------------------------------------------***---------------------------------------------------------------------

More information

Chapter 5: Gases 5.1 Pressure Why study gases? An understanding of real world phenomena. An understanding of how science works.

Chapter 5: Gases 5.1 Pressure Why study gases? An understanding of real world phenomena. An understanding of how science works. Chapter 5: Gases 5.1 Pressure Why study gases? An understanding of real world phenomena. An understanding of how science works. A Gas Uniformly fills any container. Easily compressed. Mixes completely

More information

Materials Performance November Nitrogen generator inhibits corrosion within fire protection systems

Materials Performance November Nitrogen generator inhibits corrosion within fire protection systems Materials Performance November 2014 Nitrogen generator inhibits corrosion within fire protection systems When dry and pre-action fire protection systems experience internal corrosion in their piping network,

More information

The Effects of Design and Operating Parameters on The Flooding of a Gas-Liquid Mechanically Agitated, Compartmented Column

The Effects of Design and Operating Parameters on The Flooding of a Gas-Liquid Mechanically Agitated, Compartmented Column Jurnal Kejuruteraan 12(2) 99-14 The Effects of Design and Operating Parameters on The Flooding of a Gas-Liquid Mechanically Agitated, Compartmented Column Mohd. Sobri Takriff, w.r. Penney & J.B. Fasano

More information

CP Chapter 13/14 Notes The Property of Gases Kinetic Molecular Theory

CP Chapter 13/14 Notes The Property of Gases Kinetic Molecular Theory CP Chapter 13/14 Notes The Property of Gases Kinetic Molecular Theory Kinetic Molecular Theory of Gases The word kinetic refers to. Kinetic energy is the an object has because of its motion. Kinetic Molecular

More information

GASES. Unit #8. AP Chemistry

GASES. Unit #8. AP Chemistry GASES Unit #8 AP Chemistry I. Characteristics of Gases A. Gas Characteristics: 1. Fills its container a. no definite shape b. no definite vol. 2. Easily mixes w/ other gases 3. Exerts pressure on its surroundings

More information

Generating Calibration Gas Standards

Generating Calibration Gas Standards Technical Note 1001 Metronics Inc. Generating Calibration Gas Standards with Dynacal Permeation Devices Permeation devices provide an excellent method of producing known gas concentrations in the PPM and

More information

maintaining storage tank dissolved oxygen levels utilizing gas transfer membranes

maintaining storage tank dissolved oxygen levels utilizing gas transfer membranes Water Technologies & Solutions technical paper maintaining storage tank dissolved oxygen levels utilizing gas transfer membranes Authors: W.E. Haas, SUEZ, J. Helmrich, Florida Power and Light and J.E.

More information

Chemistry Chapter 10 Test

Chemistry Chapter 10 Test Chemistry Chapter 10 Test True/False Indicate whether the sentence or statement is true or false. 1. KMT stands for Kinetic Mole Theory. 2. One of the assumptions in the KMT is that the particles are spread

More information

Honors Chemistry - Problem Set Chapter 13 Classify each of these statements as always true, AT; sometimes true, ST; or never true, NT.

Honors Chemistry - Problem Set Chapter 13 Classify each of these statements as always true, AT; sometimes true, ST; or never true, NT. Honors Chemistry - Problem Set Chapter 13 Classify each of these statements as always true, AT; sometimes true, ST; or never true, NT. 1. Atmospheric pressure is 760 mm Hg. 2. The SI unit of pressure is

More information

CHAPTER 1 INTRODUCTION TO RELIABILITY

CHAPTER 1 INTRODUCTION TO RELIABILITY i CHAPTER 1 INTRODUCTION TO RELIABILITY ii CHAPTER-1 INTRODUCTION 1.1 Introduction: In the present scenario of global competition and liberalization, it is imperative that Indian industries become fully

More information

Chemistry 51 Chapter 7 PROPERTIES OF GASES. Gases are the least dense and most mobile of the three phases of matter.

Chemistry 51 Chapter 7 PROPERTIES OF GASES. Gases are the least dense and most mobile of the three phases of matter. ROERIES OF GASES Gases are the least dense and most mobile of the three phases of matter. articles of matter in the gas phase are spaced far apart from one another and move rapidly and collide with each

More information

Dr. Rogers Chapter 5 Homework Chem 111 Fall 2003

Dr. Rogers Chapter 5 Homework Chem 111 Fall 2003 Dr. Rogers Chapter 5 Homework Chem 111 Fall 2003 From textbook: 7-33 odd, 37-45 odd, 55, 59, 61 1. Which gaseous molecules (choose one species) effuse slowest? A. SO 2 (g) B. Ar(g) C. NO(g) D. Ne(g) E.

More information

Carbon Dioxide Flooding. Dr. Helmy Sayyouh Petroleum Engineering Cairo University

Carbon Dioxide Flooding. Dr. Helmy Sayyouh Petroleum Engineering Cairo University Carbon Dioxide Flooding Dr. Helmy Sayyouh Petroleum Engineering Cairo University Properties of CO 2... Properties of CO2... CO2 density increases with:.increasing pressure.decreasing temperature Viscosity

More information

Gas Pressure. Pressure is the force exerted per unit area by gas molecules as they strike the surfaces around them.

Gas Pressure. Pressure is the force exerted per unit area by gas molecules as they strike the surfaces around them. Chapter 5 Gases Gas Gases are composed of particles that are moving around very fast in their container(s). These particles moves in straight lines until they collides with either the container wall or

More information

ASAS SIMPAN KIRA DAN PERAKAUNAN Tempat: Kota Kinabalu, Sabah. Sila pilih tarikh kursus yang bersesuaian. Tandakan dalam kotak yang disediakan.

ASAS SIMPAN KIRA DAN PERAKAUNAN Tempat: Kota Kinabalu, Sabah. Sila pilih tarikh kursus yang bersesuaian. Tandakan dalam kotak yang disediakan. PENGENALAN Kursus ini memberikan peluang kepada para peserta untuk memahami kepentingan perakaunan dalam pengurusan sesuatu organisasi. Peserta akan didedahkan dengan penyediaan penyata kewangan yang betul.

More information

Chapter 13 Gases, Vapors, Liquids, and Solids

Chapter 13 Gases, Vapors, Liquids, and Solids Chapter 13 Gases, Vapors, Liquids, and Solids Property is meaning any measurable characteristic of a substance, such as pressure, volume, or temperature, or a characteristic that can be calculated or deduced,

More information

N2 Membrane+ Nitrogen Production Systems for Offshore. Where you are. Where you need to be. We are here too.

N2 Membrane+ Nitrogen Production Systems for Offshore. Where you are. Where you need to be. We are here too. N2 Membrane+ Nitrogen Production Systems for Offshore Where you are. Where you need to be. We are here too. www.airliquide.com OVERVIEW AIR LIQUIDE OIL AND GAS SERVICES is dedicated to the supply of products

More information

Kinetic-Molecular Theory

Kinetic-Molecular Theory GASES Chapter Eleven Kinetic-Molecular Theory! Recall that our only previous description of gases stated that gases completely fill and take the shape of their containers.! The Kinetic-Molecular Theory

More information

Analysis and Modeling of Vapor Recompressive Distillation Using ASPEN-HYSYS

Analysis and Modeling of Vapor Recompressive Distillation Using ASPEN-HYSYS Computer Science Journal of Moldova, vol.19, no.2(56), 2011 Analysis and Modeling of Vapor Recompressive Distillation Using ASPEN-HYSYS Cinthujaa C. Sivanantha, Gennaro J. Maffia Abstract HYSYS process

More information

PEDESTRIAN UTILIZATION; ENHANCING FROM EXISTING : A STUDY CASE OF PANTAI CHENANG, LANGKAWI AND MELAKA HISTORICAL CITY, MELAKA

PEDESTRIAN UTILIZATION; ENHANCING FROM EXISTING : A STUDY CASE OF PANTAI CHENANG, LANGKAWI AND MELAKA HISTORICAL CITY, MELAKA PEDESTRIAN UTILIZATION; ENHANCING FROM EXISTING : A STUDY CASE OF PANTAI CHENANG, LANGKAWI AND MELAKA HISTORICAL CITY, MELAKA ABD TALIP BIN ABD RAHMAN A dissertation submitted in partial fulfilment of

More information

PURE SUBSTANCE. Nitrogen and gaseous air are pure substances.

PURE SUBSTANCE. Nitrogen and gaseous air are pure substances. CLASS Third Units PURE SUBSTANCE Pure substance: A substance that has a fixed chemical composition throughout. Air is a mixture of several gases, but it is considered to be a pure substance. Nitrogen and

More information

CHM 111 Unit 5 Sample Questions

CHM 111 Unit 5 Sample Questions Name: Class: Date: As you work these problems, consider and explain: A. What type of question is it? B. How do you know what type of question it is? C. What information are you looking for? D. What information

More information

A07 Surfactant Induced Solubilization and Transfer Resistance in Gas-Water and Gas-Oil Systems

A07 Surfactant Induced Solubilization and Transfer Resistance in Gas-Water and Gas-Oil Systems A07 Surfactant Induced Solubilization and Transfer Resistance in Gas-Water and Gas-Oil Systems R Farajzadeh* (TU Delft), A. Banaei (TU Delft), J. Kinkela (TU Delft), T. deloos (TU Delft), S. Rudolph (TU

More information

CHEMISTRY - CLUTCH CH.5 - GASES.

CHEMISTRY - CLUTCH CH.5 - GASES. !! www.clutchprep.com CONCEPT: UNITS OF PRESSURE Pressure is defined as the force exerted per unit of surface area. Pressure = Force Area The SI unit for Pressure is the, which has the units of. The SI

More information

Chemistry Chapter 11 Test Review

Chemistry Chapter 11 Test Review Chemistry Chapter 11 Test Review Multiple Choice Identify the choice that best completes the statement or answers the question. 1. Pressure is the force per unit a. volume. c. length. b. surface area.

More information

Section 5.1 Pressure. Why study gases? An understanding of real world phenomena. An understanding of how science works.

Section 5.1 Pressure. Why study gases? An understanding of real world phenomena. An understanding of how science works. Chapter 5 Gases Section 5.1 Pressure Why study gases? An understanding of real world phenomena. An understanding of how science works. Copyright Cengage Learning. All rights reserved 2 Section 5.1 Pressure

More information

CP Chapter 13/14 Notes The Property of Gases Kinetic Molecular Theory

CP Chapter 13/14 Notes The Property of Gases Kinetic Molecular Theory CP Chapter 13/14 Notes The Property of Gases Kinetic Molecular Theory Kinetic Molecular Theory of Gases The word kinetic refers to. Kinetic energy is the an object has because of its motion. Kinetic Molecular

More information

how to meet today s dissolved oxygen specifications with degasification membranes

how to meet today s dissolved oxygen specifications with degasification membranes Water Technologies & Solutions technical paper how to meet today s dissolved oxygen specifications with degasification membranes Author: Fred Wiesler, Membrana Charlotte, Ionics Note: SUEZ purchased Ionics

More information

Chapter 12. The Gaseous State of Matter

Chapter 12. The Gaseous State of Matter Chapter 12 The Gaseous State of Matter The air in a hot air balloon expands When it is heated. Some of the air escapes from the top of the balloon, lowering the air density inside the balloon, making the

More information

www.ompressedairsystems.com ultra-high purity nitrogen generators nitrogen purity: 95% to 99.999% www.ompressedairsystems.com ultra-high purity nitrogen generators nitrogen purity: 95% to 99.999% Leading

More information

INSTRUCTION: This section consists of FOUR (4) questions. Answer ALL questions. ARAHAN: Bahagian ini mengandungi EMPAT (4) soalan. Jawab SEMUA soalan.

INSTRUCTION: This section consists of FOUR (4) questions. Answer ALL questions. ARAHAN: Bahagian ini mengandungi EMPAT (4) soalan. Jawab SEMUA soalan. INSTRUCTION: This section consists of FOUR (4) questions. Answer ALL questions. ARAHAN: Bahagian ini mengandungi EMPAT (4) soalan. Jawab SEMUA soalan. QUESTION 1 SOALAN 1 (a) Define the gauge pressure,

More information

THE GAS STATE. Unit 4. CHAPTER KEY TERMS HOME WORK 9.1 Kinetic Molecular Theory States of Matter Solid, Liquid, gas.

THE GAS STATE. Unit 4. CHAPTER KEY TERMS HOME WORK 9.1 Kinetic Molecular Theory States of Matter Solid, Liquid, gas. Unit 4 THE GAS STATE CHAPTER KEY TERMS HOME WORK 9. Kinetic Molecular Theory States of Matter Solid, Liquid, gas Page 4 # to 4 9. Boyles Law P α /V PV = Constant P V = P V Pressure Atmospheric Pressure

More information

Honors Chemistry Unit 7 Gas Laws Notes

Honors Chemistry Unit 7 Gas Laws Notes Honors Chemistry Unit 7 Gas Laws Notes Kinetic Molecular Theory 1. List the five assumptions: Assumption Description Extra Info 1 Basically means: the particles themselves have compared to the space between

More information

Name Chemistry Pre-AP

Name Chemistry Pre-AP Name Chemistry Pre-AP Notes: Gas Laws and Gas Stoichiometry Period Part 1: The Nature of Gases and The Gas Laws I. Nature of Gases A. Kinetic-Molecular Theory The - theory was developed to account for

More information

Ch. 11 Mass transfer principles

Ch. 11 Mass transfer principles Transport of chemical species in solid, liquid, or gas mixture Transport driven by composition gradient, similar to temperature gradients driving heat transport We will look at two mass transport mechanisms,

More information

1 PIPESYS Application

1 PIPESYS Application PIPESYS Application 1-1 1 PIPESYS Application 1.1 Gas Condensate Gathering System In this PIPESYS Application, the performance of a small gascondensate gathering system is modelled. Figure 1.1 shows the

More information

PTRT 2470: Petroleum Data Management 3 - Facilities Test 4 (Spring 2017)

PTRT 2470: Petroleum Data Management 3 - Facilities Test 4 (Spring 2017) Use scantron to answer all questions PTRT 2470: Petroleum Data Management 3 - Facilities Test 4 (Spring 2017) 1. The term dehydration of natural gas means A. addition of water vapor B. removal of water

More information

Dissolved Oxygen Guide

Dissolved Oxygen Guide Educat i onser i es Di ssol vedoxygengui de Dissolved Oxygen Guide Introduction Dissolved oxygen probes provide a convenient approach to essentially direct measurement of molecular oxygen. The membrane

More information

A NEW PROCESS FOR IMPROVED LIQUEFACTION EFFICIENCY

A NEW PROCESS FOR IMPROVED LIQUEFACTION EFFICIENCY WHITE PAPER A NEW PROCESS FOR IMPROVED LIQUEFACTION EFFICIENCY Author(s): Adam Jones and Grant Johnson, Costain Natural Resources First published: GPAE, September 2014 www.costain.com A New Process for

More information

Applied Compression s nitrogen generator packages offer users a cost-effective alternative to costly bottled nitrogen.

Applied Compression s nitrogen generator packages offer users a cost-effective alternative to costly bottled nitrogen. APPLICATIONS: Autoclaves Modified Packaging Atmospheres Coffee Packaging Oxidation Control Coil Tubing Units Pharmaceutical Manufacturing Electronic Parts Manufacturing Reflow Ovens Enhanced Oil Recovery

More information

Vibration-Free Joule-Thomson Cryocoolers for Distributed Microcooling

Vibration-Free Joule-Thomson Cryocoolers for Distributed Microcooling Vibration-Free Joule-Thomson Cryocoolers for Distributed Microcooling W. Chen, M. Zagarola Creare Inc. Hanover, NH, USA ABSTRACT This paper reports on an innovative concept for a space-borne Joule-Thomson

More information

A. What are the three states of matter chemists work with?

A. What are the three states of matter chemists work with? Chapter 10 and 12 The Behavior of Gases Chapter 10 The States of Matter A. What are the three states of matter chemists work with? Section 10.1 Pg 267 B. We will explain the behavior of gases using the

More information

THE GUIDE TO MEMBRANE NITROGEN GENERATORS

THE GUIDE TO MEMBRANE NITROGEN GENERATORS THE GUIDE TO MEMBRANE NITROGEN GENERATORS MEMBRANE NITROGEN GENERATORS The membrane nitrogen generator utilizes a hollow porous fiber to separate the N 2 molecules from other molecules, including O 2.

More information

Practice MC Test unit D (Ch 10) Gas Laws (pg 1 of 10)

Practice MC Test unit D (Ch 10) Gas Laws (pg 1 of 10) Practice MC Test unit D (Ch 10) Gas Laws (pg 1 of 10) This is practice - Do NOT cheat yourself of finding out what you are capable of doing. Be sure you follow the testing conditions outlined below. DO

More information

Fall 2004 Homework Problem Set 9 Due Wednesday, November 24, at start of class

Fall 2004 Homework Problem Set 9 Due Wednesday, November 24, at start of class 0.30 Fall 004 Homework Problem Set 9 Due Wednesday, November 4, at start of class Part A. Consider an iron surface which serves as a catalyst for the production of ammonia from nitrogen and hydrogen. The

More information

PRODUCTION OF MEMBRANE FOR CO₂/N₂SEPARATION EFFECT OF DILUENTS EXTRACTION TIME ON PP-DPE MEMBRANE

PRODUCTION OF MEMBRANE FOR CO₂/N₂SEPARATION EFFECT OF DILUENTS EXTRACTION TIME ON PP-DPE MEMBRANE PRODUCTION OF MEMBRANE FOR CO₂/N₂SEPARATION EFFECT OF DILUENTS EXTRACTION TIME ON PP-DPE MEMBRANE AHMAD FADHIL HANAFI BIN ABDULLAH Thesis submitted in partial fulfilment of the requirements for the award

More information

States of Matter. Q 7. Calculate the average of kinetic energy, in joules of the molecules in 8.0 g of methane at 27 o C. (IIT JEE Marks)

States of Matter. Q 7. Calculate the average of kinetic energy, in joules of the molecules in 8.0 g of methane at 27 o C. (IIT JEE Marks) Q 1. States of Matter Calculate density of NH 3 at 30 o C and 5 atm pressure Q 2. (IIT JEE 1978 3 Marks) 3.7 g of a gas at 25 o C occupied the same volume as 0.184g of hydrogen at 17 o C and at the same

More information

NOTES: Behavior of Gases

NOTES: Behavior of Gases NOTES: Behavior of Gases Properties of Gases Gases have weight Gases take up space Gases exert pressure Gases fill their containers Gases are mostly empty space The molecules in a gas are separate, very

More information

How To Use Getters and Getter Pumps

How To Use Getters and Getter Pumps A Journal of Practical and Useful Vacuum Technology From By Phil Danielson How To Use Getters and Getter Pumps Gettering is a vacuum pumping technology that has been with us, in many forms, for almost

More information

CONSIDERATION OF DENSITY VARIATIONS IN THE DESIGN OF A VENTILATION SYSTEM FOR ROAD TUNNELS

CONSIDERATION OF DENSITY VARIATIONS IN THE DESIGN OF A VENTILATION SYSTEM FOR ROAD TUNNELS - 56 - CONSIDERATION OF DENSITY VARIATIONS IN THE DESIGN OF A VENTILATION SYSTEM FOR ROAD TUNNELS Gloth O., Rudolf A. ILF Consulting Engineers Zürich, Switzerland ABSTRACT This article investigates the

More information

Chemistry A Molecular Approach. Fourth Edition. Chapter 5. Gases. Copyright 2017, 2014, 2011 Pearson Education, Inc. All Rights Reserved

Chemistry A Molecular Approach. Fourth Edition. Chapter 5. Gases. Copyright 2017, 2014, 2011 Pearson Education, Inc. All Rights Reserved Chemistry A Molecular Approach Fourth Edition Chapter 5 Gases Supersonic Skydiving and the Risk of Decompression Gas Gases are composed of particles that are moving around very fast in their container(s).

More information

Training Fees 4,000 US$ per participant for Public Training includes Materials/Handouts, tea/coffee breaks, refreshments & Buffet Lunch.

Training Fees 4,000 US$ per participant for Public Training includes Materials/Handouts, tea/coffee breaks, refreshments & Buffet Lunch. Training Title GAS CONDITIONING & PROCESSING Training Duration 5 days Training Venue and Dates Gas Conditioning & Processing 5 07 11 April, 2019 $4,000 Dubai, UAE Trainings will be conducted in any of

More information

Kinetic Molecular Theory imaginary Assumptions of Kinetic Molecular Theory: Problems with KMT:

Kinetic Molecular Theory imaginary Assumptions of Kinetic Molecular Theory: Problems with KMT: AP Chemistry Ms. Ye Name Date Block Kinetic Molecular Theory Explains properties of gases, liquids, and solids in terms of energy using an ideal gas, an imaginary which fits all the assumptions of kinetic

More information

Gas Law Review. Honors Chem.

Gas Law Review. Honors Chem. Gas Law Review Honors Chem. Question 1: KMT 1: What does KMT stand for? 2: Gas particles have no or. 3: Gas particles are not to or by each other. 4: measures the average kinetic energy of gas particles.

More information

Ch. 14 The Behavior of Gases

Ch. 14 The Behavior of Gases Ch. 14 The Behavior of Gases 14.1 PROPERTIES OF GASES Compressibility Compressibility: a measure of how much the volume of matter decreases under pressure Gases are easily compressed because of the spaces

More information

Introduction. S. James Zhou, Howard Meyer, Dennis Leppin and Shiguang Li Gas Technology Institute Benjamin Bikson and Yong Ding PoroGen Corporation

Introduction. S. James Zhou, Howard Meyer, Dennis Leppin and Shiguang Li Gas Technology Institute Benjamin Bikson and Yong Ding PoroGen Corporation Nano-porous PEEK Hollow Fiber-based Gas/Liquid Membrane Contactors for Sour Gas Treating International Gas Union Research Conference 2014, September 17-19, Copenhagen, Denmark S. James Zhou, Howard Meyer,

More information

AP TOPIC 6: Gases. Revised August General properties and kinetic theory

AP TOPIC 6: Gases. Revised August General properties and kinetic theory AP OPIC 6: Gases General properties and kinetic theory Gases are made up of particles that have (relatively) large amounts of energy. A gas has no definite shape or volume and will expand to fill as much

More information

Chapter 11. Recall: States of Matter. Properties of Gases. Gases

Chapter 11. Recall: States of Matter. Properties of Gases. Gases Chapter 11 Gases Recall: States of Matter Solids and Liquids: are closely related because in each case the particles are interacting with each other Gases: Properties of Gases Gases can be compressed Gases

More information

Isotopes and Gases. Isotopes as constraints to air sea gas exchange, gas exchange time scales, the various time scales of carbon (isotope) exchange

Isotopes and Gases. Isotopes as constraints to air sea gas exchange, gas exchange time scales, the various time scales of carbon (isotope) exchange Isotopes and Gases Isotopes as constraints to air sea gas exchange, gas exchange time scales, the various time scales of carbon (isotope) exchange Isotope effects in gas solution and molecular diffusion

More information

Unit 8: Gases and States of Matter

Unit 8: Gases and States of Matter Unit 8: Gases and States of Matter Gases Particles that have no definite shape or volume. They adapt to the shape and volume of their container. Ideal gases are imaginary gases that comply with all the

More information

Elements that exist as gases at 25 o C and 1 atmosphere H 2, N 2, O 2, F 2, Cl 2, He, Ne, Ar, Kr, Xe, Rn

Elements that exist as gases at 25 o C and 1 atmosphere H 2, N 2, O 2, F 2, Cl 2, He, Ne, Ar, Kr, Xe, Rn AP Chemistry Chapter 5 Sections 5. 5.9 Note Organizer Pressure, The Gas Laws of Boyle, Charles, and Avogadro, The Ideal Gas Law, Gas Stoichiometry, Dalton s Law of Partial Pressure, The Kinetic olecular

More information

GAS CYLINDERS RULES, 2004

GAS CYLINDERS RULES, 2004 GAS CYLINDERS RULES, 2004 In exercise of the powers conferred by sections 5 and 7 of the Explosives Act, 1884 (4 of 1884) and in suppression of the Gas Cylinders Rules, 1981, except in respect things done

More information

A variable-pressure constant-volume method was employed to determine the gas permeation

A variable-pressure constant-volume method was employed to determine the gas permeation Electronic Supplementary Material (ESI) for RSC Advances. This journal is The Royal Society of Chemistry 2015 S.1. Gas permeation experiments A variable-pressure constant-volume method was employed to

More information

Determination of R: The Gas-Law Constant

Determination of R: The Gas-Law Constant Determination of R: The Gas-Law Constant PURPOSE: EXPERIMENT 9 To gain a feeling for how well real gases obey the ideal-gas law and to determine the ideal-gas-law constant R. APPARATUS AND CHEMICALS: KClO

More information