© 2014 ERNESTO SILVA MOJICA ALL RIGHTS RESERVED POLYMER-SILICA HYBRIDS FOR SEPARATION OF CO2 AND CATALYSIS OF ORGANIC REACTIONS
A Dissertation Presented to The Graduate Faculty of The University of Akron
In Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy
Ernesto Silva Mojica May, 2014 POLYMER-SILICA HYBRIDS FOR SEPARATION OF CO2 AND CATALYSIS OF ORGANIC REACTIONS
Ernesto Silva Mojica
Dissertation
Approved: Accepted:
______Advisor Department Chair Dr. Steven S.C. Chuang Dr. Coleen Pugh
______Committee Member Dean of the College Dr. Matthew Becker Dr. Stephen Z.D. Cheng
______Committee Member Dean of the Graduate School Dr. Mesfin Tsige Dr. George R. Newkome
______Committee Member Date Dr. Darrell Reneker
______Committee Member Dr. Jie Zheng
ii ABSTRACT
Porous materials comprising polymeric and inorganic segments have attracted interest from the scientific community due to their unique properties and functionalities.
The physical and chemical characteristics of these materials can be effectively exploited for adsorption applications. This dissertation covers the experimental techniques for fabrication of poly(vinyl alcohol) (PVA) and silica (SiO2) porous supports, and their functionalization with polyamines for developing adsorbents with potential applications in separation of CO2 and catalysis of organic reactions.
The supports were synthesized by processes involving (i) covalent cross-linking of PVA, (ii) hydrolysis and poly-condensation of silica precursors (i.e., sol-gel synthesis), and formation of porous structures via (iii) direct templating and (iv) phase inversion techniques. Their physical structure was controlled by the proper combination of the preparation procedures, which resulted in micro-structured porous materials in the form of micro-particles, membranes, and pellets. Their adsorption characteristics were tailored by functionalization with polyethyleneimine (PEI), and their physicochemical properties were characterized by vibrational spectroscopy, microscopy, calorimetry, and adsorption techniques. Spectroscopic investigations of the interfacial cross-linking reactions of PEI and PVA with glutaraldehyde (GA) revealed that PEI catalyzes the cross-linking reactions of PVA in absence of external acid catalysts. In-situ IR spectroscopy coupled
iii with a focal plane array (FPA) image detector allowed the characterization of a gradient interface on a PEI/PVA composite membrane and the investigation of the cross-linking reactions as a function of time and position. The results served as a basis to postulate possible intermediates, and propose the reaction mechanisms.
The formulation of amine-functionalized CO2 capture sorbents was based on the spectroscopic investigation of the interactions of CO2 with amine molecules under simulated CO2 capture conditions. Industrial CO2 capture processes involve fluidization and require degradation-resistant sorbents in the form of pellets. Agglomeration of silica- based CO2 capture sorbents involved the formulation of a polymer binder solution and the design of a scalable pelletization process. The characterization of these pellets revealed the formation of a CO2-permeable polymer-silica network, which is resistant to attrition, and exhibits similar CO2 capture and degradation performance as the non- pelletized sorbents. A compartmental modeling technique was used to simulate the CO2 adsorption process and to elucidate the kinetic and thermodynamic parameters that impact the commercial viability of emerging CO2 capture technologies.
The fundamental concepts and experimental techniques developed for the preparation of CO2 capture sorbents served as a basis for fabricating amine- functionalized polymer-silica hybrids for alternative applications including enzyme immobilization and catalysis of organic reactions.
iv ACKNOWLEDGEMENTS
I express my appreciation and sincere acknowledgement to my research advisor professor Steven S.C. Chuang for his permanent guidance, support, and valuable influence in my professional life. I extend my gratitude to my committee members Dr.
Matthew Becker, Dr. Mesfin Tsige, Dr. Darrell Reneker, and Dr. Jie Zheng for their valuable suggestions on this work. Thank you to all my colleagues for their assistance and valuable discussions, especially to my friends Uma Tumuluri, Mathew Isenberg,
Chris Wilfong, Mehdi Lohrasbi, and Tritti Siengchum for creating an enjoyable work atmosphere. Thanks to Mr. Dustin Zachariasz for his support in fabrication and modification of prototypes, and to FirstEnergy, DoE and Aspen Aerogels Inc. for partially supporting the research presented in this dissertation.
Special gratitude to Alvaro Rodríguez, Maurício Echeverri, Marcela Castaño, and
Gustavo Guzmán for their friendship and for the great times I have had in Akron. My deepest and most sincere appreciation to Ana Bacco for her love, support, and understanding, for providing happiness, and motivating me every day. My greatest gratitude to my parents Víctor F. Silva and Leonor Mojica, my sister Diana Silva, and my little niece Ángela María, who have supported and encouraged every step in my life.
Those around me give me the driving force to accomplish my objectives.
v TABLE OF CONTENTS
Page
LIST OF FIGURES……………………………………………………………………..xi
LIST OF TABLES……………………………………..……………………..…….….xvi
CHAPTER
I. INTRODUCTION………………………………………………………………………1
Overview…………………………………………………………………...... 1
Porous Polymer-silica Hybrids, and Selection of the Fabrication Techniques……..3
CO2 Capture and Separation by Amine-functionalized Polymer-silica Hybrids…...6
Composite Membranes for Separation of CO2…………………………...... 6
CO2 capture by Amine-functionalized Solid Sorbents………………….……7
Catalysis of Organic Reactions…………………………………………….……….9
Approach…………………………………………………………………….…….10
Hypotheses…………………………………………………………………….…..13
Fabrication of Polymer Silica Hybrids……………………………………...13
CO2 Capture and Separation………………………………………………...14
Basic Catalysis and Enzyme Immobilization……………………………….14
Objectives…………………………………………………………………………15
Scope………………………………………………………………………………16
II. BACKGROUND……………………………………………………………………...18
Polymer-silica Hybrids…………………………………………………………….18
Sol-gel Synthesis……………………………………………………………22
vi Cross-linking of PVA and PEI……………………………………………...25
Direct Templating…………………………………………………………...33
Synthesis of mesoporous materials………………………………………….35
Phase Inversion……………………………………………………………...36
CO2 Capture by Solid Sorbents……………………………………………………42
Membranes for separation of CO2…………………………………………..48
III. EXPERIMENTAL…………………………………………………………………...50
Synthesis of SBA-15………………………………………………………………50
Construction of PVA-water-ethanol and PVA-water-acetone Ternary Phase Diagrams…………………………………………………………………………..51
Fabrication of Polymer-silica Hybrid Membranes…………………………..…….52
Fabrication of PEI/PVA membranes for cross-linking studies……………………54
Preparation of Amine-functionalized Solid Sorbents……………………………...54
Experimental System for CO2 Capture Studies……………………………………56
Step Switch Adsorption……………………………………………………………57
Diffused Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS)……...58
Attenuated Total Reflection Infrared Spectroscopy (ATR)……………………….58
Transmission Infrared Spectroscopy………………………………………………59
Focal Plane Array (FPA) image detector………………………………………….60
Mass Spectrometry………………………………………………………………...61
IV. FABRICATION AND STRUCTURAL CHARACTERIZATION OF POLYMER- SILICA HYBRIDS………………………………………………………………………62
Introduction...... ……………………………………………………….62
Experimental………………………………………………………………………64
Synthesis of SBA-15………………………………………………………...64
Fabrication of PVA/PEI/SiO2 membranes…………………………………..66
vii Results and Discussion…………………………………………………………….66
Synthesis of SBA-15………………………………………………………...66
Fabrication of Polymer-silica Hybrids………………………………………73
Conclusions………………………………………………………………………..83
V. INFRARED STUDY OF THE INTERFACIAL CROSS-LINKING REACTIONS OF PEI/PVA COMPOSITE MEMBRANES………………………………………………..84
Introduction………………………………………………………….……………85
Experimental………………………………………………………………………86
Results and Discussion…………………………………………………………….89
Membrane characterization…………………………………………………89
Conclusions………………………………………………………………………..96
V FORMULATION AND PREPARATION OF CO2 CAPTURE PELLETS………….98
Introduction……………………………………………………………………….98
Experimental………………………………………………………………………99
Sorbent and Pellet Preparation………………………………………………99
Fluidization and Attrition Tests……………………………………………101
Construction of the Adsorption Isotherms………………………………....102
Results and Discussion…………………………………………………………...103
Sorbent and Pellets Characterization………………………………………103
Conclusions………………………………………………………………………110
VII. COMPARTMENTAL MODELING OF A FIXED BED CO2 CAPTURE UNIT..111
Introduction………………………………………………………………………112
Experimental……………………………………………………………………..115
Model Description………………………………………………………………..117
Materials and Energy Balances…………………………………………………..120
viii Rate law and adsorption equilibrium…………………………………………….122
Residence time distribution (RTD) and dispersion………………………………124
Results……………………………………………………………………………126
Adsorption isotherms………………………………………………………126
Non-deal flow behavior and residence time distribution RTD…………….129
Compartmental modeling of a fixed bed CO2 adsorption unit…………….132
Parametric Studies: Heat Effects…………………………………………..134
Sensitivity analyses………………………………………………………...140
Conclusions………………………………………………………………………145
VIII. SILICA-SUPPORTED AMINE CATALYSTS FOR CARBON-CARBON ADDITION REACTIONS……………………………………………………………...146
Introduction………………………………………………………………………146
Experimental……………………………………………………………………..148
Results and Discussion…………………………………………………………...150
Amine-immobilized catalysts on silica supports…………………………..150
Catalytic activity towards the Claisen condensation of MB and MEK……155
Conclusions………………………………………………………………………162
IX. POROUS POLY(VINYL ALCOHOL) COMPOSITE MEMBRANES FOR IMMOBILIZATION OF GLUCOSE OXIDASE……………………………………...163
Introduction………………………………………………………………………163
Experimental……………………………………………………………………..165
Results and Discussion…………………………………………………………...169
Conclusions………………………………………………………………………179
BIBLIOGRAPHY………………………………………………………………………183
APPENDIX …………………………………………………………………………….207
ix LIST OF FIGURES
Figure Page
2.1 Two general reaction pathways following addition of a nucleophile to an aldehyde or ketone. The top pathway leads to an alcohol product; the bottom pathway leads to a product with a C=Nu double bond. Adapted from [108]...... 26
2.2 Chemical cross-linking of PVA with GA catalyzed by acid...... 27
2.3 Nucleophilic addition of amines; imine and enamine formation. Adapted from [108]...... 29
2.4 Chemical cross-linking of PVA with glutaraldehyde ...... 30
2.5 Preparation of porous materials by the direct templating method...... 34
2.6 Porous materials: IUPAC classification by pore size...... 35
2.7 Steps involved in the preparation of porous materials by the phase inversion technique...... 37
2.8 Ternary diagram illustration of the formation of a porous polymer by the phase inversion technique...... 41
2.9 Proposed structures of adsorbed CO2 species; carbamate and ammonium ions, and carbamic acid...... 46
3.1 Cloud point observation for a PVA-water-acetone system with initial composition of 10% PVA in water. The hollow dots represent the compositions of the system where there is only one phase. The solid dots represent the compositions of the system where there is coexistence of two phases...... 52
3.2 Schematic structure of PEI and TEPA impregnated on a silica support. The structure of PEI shown in this figure is a simple representation of a more complex, highly branched molecule...... 55
3.3 Schematics of the experimental system and DRIFTS reactor used in CO2 capture studies...... 57
x 4.1 Schematics of the synthesis process of SBA-15. The silica precursor is adsorbed on the surface of P123 micelles, where the condensation of silica occurs. The pores are generated by removal of P123 by washing or calcination...... 65
4.2 Hydrolysis and condensation of sodium silicate catalyzed by acetic acid. Sodium silicate is soluble in water where it forms basic solutions. SiO2 precipitates at neutral or acidic pH conditions...... 67
4.3 Nitrogen adsorption-desorption isotherm for SBA-15 calcined produced by condensation of sodium silicate catalyzed by acetic acid...... 68
4.4 Microscopic images of SBA-15 calcined. Top: SEM, middle and bottom:TEM...... 69
4.5 IR absorbance spectra and schematics of the proposed surface silanol groups of amorphous SiO2, SBA-15 as prepared, and SBA-15 calcined, collected in DRIFTS at 100oC...... 72
4.6 Ternary phase diagrams at 25 oC for the systems PVA-water acetone, and PVA- water-ethanol. The hollow dots represent the compositions where the system exists in one phase and the filled dots represent the compositions where the system coexists in two phases. The binodal curve was drawn by visual exploration of the experimental data...... 74
4.7 IR absorbance spectra and SEM pictures of polymer-silica membranes prepared by direct templating. The IR spectra were collected in transmission mode at 25 oC. .... 78
4.8 IR absorbance spectra and SEM pictures of polymer-silica membranes prepared by direct templating and phase inversion. The IR spectra were collected in transmission mode at 25 oC...... 80
4.9 IR absorbance spectra and SEM pictures of polymer-silica membranes prepared by direct templating, phase inversion, and cross-linking. The IR spectra were collected in transmission mode at 25 oC...... 81
4.10 Thermo gravimetric analysis of the polymer-silica hybrid membranes. Measured in TGA from 25 oC to 340 oC...... 82
5.1 (a) 3D mapping of the PVA/PEI-PVA interface generated by the IR absorbance intensity of the amine band at 1573 cm-1. (b) Picture of the IR transmission cell and SEM picture of the PVA support. (c) Schematics of the system used for evaluating the interfacial cross-linking reactions of PEI/PVA membranes. The color scale of the 3D mapping was selected arbitrarily to match the range of intensities of the IR band at 1573 cm-1...... 88
5.2 IR Single beam spectra of three regions of the PEI/PVA membrane with different concentrations of PEI, before and after cross-linking...... 90
xi 5.3: IR absorbance spectra of three regions of the PEI/PVA membrane before and after cross-linking. Abs=log (1/I) where I is the single beam of the spectrum of interest. 91
5.4 Kinetics of interfacial cross-linking reactions of a PEI/PVA membrane. Collected in transmission cell under constant flow of gas phase GA, using a focal plane detector. Absorbance =log(I/Io), where I is the single beam spectrum of the membrane at time t, and Io is the single beam spectrum of the membrane at time t=0...... 93
5.5 Mechanism of imine formation by reaction of an aldehyde with a primary amine. Adapted from [108]...... 95
6.1 Schematics of (a) wet extruder and (b) spheronizer for the fabrication of CO2 capture pellets...... 100
6.2 Experimental set up for fluidization and attrition test...... 102
6.3 IR absorbance spectra of the amine functionalized sorbent at different stages of the CO2 capture process. Absorbance=log(I/Io), where I is the single beam spectrum of the sorbent at a given stage, and Io is the single beam spectrum of the sorbent after pretreatment, prior to CO2 adsorption...... 105
6.4: Heat capacity of the amine-functionalized sorbent as a function of temperature measured by DSC...... 105
6.5 Attrition curve for amine functionalized pellets in standard ball mill for particle sizes 841 µm or greater, mesh No. 20. Adapted from ASTM E728-91 R97 ...... 106
6.6 Degradation curve for powder sorbent and rods pellets under steam at 130oC...... 108
6.7 SEM images and EDX analysis of (a) the external surface, and (b) the internal surface...... 109
7.1 (a) Schematic representation of the DRIFTS reactor and sample holder. (b) Compartmental model associated to the step-switch adsorption of CO2 in the DRIFTS reactor...... 117
7.2 (a) Experimental adsorption isotherms of CO2 on the amine-functionalized sorbent at 35, 55 and 75 oC. (b) Adsorption isotherms derived from the single site (Langmuir), dual site with no-dissociation (ND), and dual site with dissociation (DIS) models, o fitted at 55 C. (c) Surface coverage of CO2 on the sorbent predicted by each adsorption isotherm model at 55oC...... 126
7.3 Cumulative residence time distribution F(t) and external age distribution E(t) functions determined experimentally by an N2 tracer step input in the adsorption unit parallel to the adsorption of CO2...... 130
7.4 Temperature and concentration profiles obtained for the optimal values of the parameters KL0, and U, for the parametric study 1: Isothermal rate law...... 134 xii 7.5 Temperature and concentration profiles obtained for the optimal values of the parameters KL0, E, and U, for the parametric study 1: Isothermal rate law...... 139
7.6 Temperature and concentration profiles predicted by the ND adsorption isotherm model for adsorption at 35 and 75 oC...... 141
7.7 Temperature and concentration profiles predicted by the ND adsorption isotherm 3 model for a CO2 adsorption experiment with flow rates of 75 cm /min, and 225 cm3/min...... 143
7.8 Temperature and concentration profiles predicted by the ND adsorption isotherm model for a CO2 adsorption experiment using diluted (5 v%) and concentrated (25 v%) CO2...... 144
8.1 Schematics of the FTIR micro-reactor used for evaluation of liquid phase organic catalytic reactions...... 150
8.2 (a) IR absorbance spectra and corresponding band assignment for SiO2 and SBA-15 supports collected at 100 oC. (b) IR absorbance spectra of amine immobilized catalysts. Absorbance was obtained by Abs=log(1/I), where I is the normalized single beam of interest...... 152
8.3 IR absorbance spectra and corresponding band assignment for the Claisen condensation of MB and MEK on (a) SiO2, (b) SiO2-T and (c) SiO2-TE. Absorbance was obtained by Abs=-log(I/Io), where I is the normalized single beam at specific time of the reaction and Io is the normalized single beam at time t=0. The IR absorbance spectra of (d) the condensation product with SiO2-TE catalyst, (e) the condensation product without catalyst and (f) the recovered SiO2-TE catalyst washed with MEK(Abs=-log(1/I))...... 156
8.4 Absorbance intensities of 1727 and 1280 cm-1 as a function of reaction time for the immobilized amine catalysts prepared...... 159
8.5 (a) Mechanism of the Claisen condensation reaction proposed in [108]. (b) Proposed mechanism for the Claisen condensation of methyl benzoate and methyl ethyl ketone using immobilized amine catalysts...... 161
9.1 Glucose oxidation reaction catalyzed by GOx, and formation of a quinoneimine dye for indirect detection of H2O2 by UV-vis spectroscopy...... 168
9.2 IR absorbance spectra and SEM pictures of the solid particles. The spectra were collected in DRIFTS (diffuse reflectance infrared Fourier transform spectroscopy) at 100 oC, Abs=log(1/Single beam). The surface area (s.a.) was measured by the BET method...... 170
9.3 IR absorbance spectra of the PVA composite membranes before (black) and after cross-linking (red), and SEM pictures of the cross-linked membranes. The IR spectra were collected in transmission mode at 20 oC, Abs=log(1/Single beam). The surface xiii area (s.a.) was measured by the BET method, and the void fraction (ϕ) was estimated from the SEM pictures using image processing and analysis software (ImageJ). .... 172
9.4 (a) Amount of GOx adsorbed on PVA composite membranes as a function of time, (b) equilibrium amount of GOx adsorbed on the membranes as a function of pH, and (c) adsorption isotherm of GOx adsorbed on PPE-CM at pH=5.1...... 174
9.5 (a) Specific catalytic activity of free GOx, and GOx immobilized in PVA composite membranes as a function of pH. (b) Membrane activity of GOx based on the weight of enzyme plus the support (U/g-membrane)...... 177
9.6 IR absorbance spectra and intensity ratio of the amide I to amide II bands of immobilized GOx. The spectra were obtained by substraction of the absorbance spectra of the dry membranes before the adsorption experiments from the absorbance spectra of the membranes after adsorption of GOx. The intensity of amide I and amide II was measured using a baseline from the minimum points of the curves around 1700, 1600 and 1500 cm-1...... 179
xiv LIST OF TABLES
Table Page
2.1 Summary of literature reporting polymer silica hybrids fabricated by diverse methods...... 20
2.2 IR band assignments for PVA, PEI, GA and the cross-linked products...... 32
2.3 Hansen solubility parameters for PVA, water, ethanol, and acetone [169]...... 39
2.4 Proposed adsorbed species for CO2 adsorbed species on amine-functionalized solid sorbents...... 48
6.1 Solid properties and CO2 capture capacity of pellets and powder sorbent...... 104
6.2 Fluidization conditions and attrition test for pellets. Results based on fluidization of 5g of pellets...... 107
7.1 Parameters of the adsorption isotherms fitted by minimization of the SSE, and heats of adsorption calculated from equation [9]...... 129
7.2 Axial dispersion coefficients for fixed bed adsorbers simulated by the dispersion model...... 132
7.3 Optimal parameters and SSE values obtained for the parametric study 1: Isothermal rate law...... 136
7.4 Optimal parameters and SSE values obtained for the parametric study 2: Non- isothermal rate law...... 140
8.1 IR absorbance intensity ratio of Si-OH (3743 cm-1) to Si-O-Si (809, 1060, and 1180 -1 cm ) for SiO2, SBA-15 a.p. and SBA-15 cal [101, 189, 253, 272]...... 151
o 8.2 CO2 adsorption capacity and amine efficiency for the catalysts at 30 C...... 155
xv CHAPTER I
1INTRODUCTION
1.1 Overview
Polymer-silica hybrids refer to composite materials that consist of polymeric and silica domains. Their chemical and physical properties are a desirable combination of the functionality and flexibility of polymers, with the chemical, thermal, and mechanical stability of silica particles [1-3]. The reinforcement of polymers with inorganic particles is a well-known industrial practice for fabricating high performance materials with applications that extend to a wide range of science and technology fields. Advanced applications of polymer-silica hybrids include adsorption, absorption, heterogeneous catalysis, enzyme immobilization, chromatographic separation, cell culture, controlled release technologies, and environmental remediation [4-6].
The potential use of polymer-silica hybrids for separation of CO2 and catalysis of organic reactions relies very heavily on understanding and controlling their physicochemical properties and physical structure. Their physicochemical properties can be altered by cross-linking reactions, inclusion of additives, and functionalization with amine molecules. Their physical structure can be controlled by altering the preparation procedures such as tape casting, extrusion, phase inversion, and direct templating [7-14].
1 Molecular characterization techniques can be applied for elucidating the surface characteristics and molecular structure of the resulting materials. Furthermore, these techniques can be used for the mechanistic investigation of physicochemical phenomena including interfacial cross-linking reactions, heterogeneous catalysis, adsorption, desorption, and degradation [15-20].
Numerous attempts have been made to synthesize CO2 capture sorbents with high adsorption capacity and low regeneration energy [21-39]. Despite the extensive studies no sorbents have been found to serve for today’s industrial applications because the cost associated with the raw materials and synthetic methods are not justified by the overall economics of electricity generation in coal-fired power plants [40-49]. In addition, most of these sorbents exist in power forms, and are not suitable for the operation of industrial gas-solid adsorption units [50]. These units are typically large scale fixed-bed or fluidized-bed reactors filled with pellets or monoliths [47, 51, 52]. Membranes for separation of CO2 have been developed for specialized applications where the total flow and concentration of CO2 are relatively low. The use of membrane reactors for post- combustion CO2 capture is limited by the low performance of the membranes to serve in high throughput applications [53, 54]. This dissertation focuses on the fundamental investigation and development of more economically viable approaches for (i) the synthesis and characterization of amine-functionalized poly(vinyl alcohol) (PVA) and silica (SiO2) hybrids, (ii) the fabrication of cross-linked composite membranes for separation of CO2, (iii) the formulation, fabrication, and performance testing of degradation-resistant pellets for CO2 capture, and (iv) the evaluation of these materials for base-catalyzed organic condensation reactions and enzyme immobilization.
2 1.2 Porous Polymer-silica Hybrids, and Selection of the Fabrication Techniques
The motivation behind combining different approaches that are individually used for the preparation of organic and inorganic materials is based on the need of designing more convenient fabrication techniques to obtain porous materials with enhanced mechanical properties and specific functionalities. Polymer-silica hybrids offer unique properties that justify their use in engineering applications as performance materials, or in scientific applications as advanced functional materials. The development and potential applications of polymer-silica hybrids require a deep knowledge of the interactions between their functional groups [55-61]. Several of these materials require extensive manufacturing efforts because they involve intense mechanical treatments, expensive material sources, or long synthesis time. Few studies have addressed the incorporation of more than two procedures to fabricate organic-inorganic hybrids using simpler strategies
[62-65].
The most common approaches for fabrication of polymer-silica hybrids are the sol-gel synthesis, co-polymerization of pre-polymers and organic silica precursors (i.e., grafting), cross-linking of polymer solutions, and compounding/blending of polymers and fillers. Porous materials have been synthesized by co-polymerization of functional monomers (organic or inorganic) with a cross-linking agent in the presence of a template
[66, 67]. The functional monomer-template complex is formed through self assembly between these two species by covalent or non-covalent interactions. In the steps following polymerization, removal of the template is achieved either by washing, solvent extraction, or calcination [68-71]. The removal of the template originates cavities with defined size and special arrangement of functional groups.
3 Polymer membranes have been prepared by phase inversion in solvent/non- solvent systems. These membranes are usually composed of a single polymer, but recently more publications have been found presenting inorganic-polymer composites prepared by phase inversion techniques [12, 63, 72, 73]. Phase inversion consists of the precipitation of a soluble phase by the action of external factors like temperature, concentration, or additives. The most common approach involves the use of a pristine solution and the addition of a non-solvent or precipitant. The precursor can be a polymer, a pre-polymer, a polymer blend, or a monomer that undergoes polymerization by the action of the non-solvent. The pristine solution is prepared in a good solvent where the precursor is dispersed in the continuous phase; upon addition of the non-solvent the miscibility of the ternary mixture decreases significantly to produce micro-phase separation, and eventually precipitation of the solid phase. The solvent of the pristine solution acts as a template in the solid precursor, and creates pores and channels upon migration from the precipitating phase. For this reason, the phase inversion technique involves serious kinetic considerations in addition to the already complex thermodynamics that define the phase separation process [14, 74-78].
Dispersion of silica particles into a polymer matrix affects the hydrodynamic interactions and modifies the molecular dynamics of the polymer chains. The polymer chains in the vicinity of the silica surfaces experience physical interactions that limit their mobility, creating a soft shell around the hard particles [79-83]. These interactions are normally sufficient to maintain the structure of the composite, but in most cases the applications require integral materials with continuous phases and high chemical, mechanical, and thermal stability. Cross-linking is the most commonly used method for
4 improving the mechanical properties and generating polymer networks. This technique consists of linking the chains of a polymer, or the precursors of an inorganic material, across the volume occupied by its domains. The cross-linking process may consist of physical entanglement, physical interactions, ionic bonding, or covalent bonding. The molecular weight of the polymer increases significantly with the number of links, experiencing transition from a solution state to a gel, and eventually causing the formation of a solid phase. Partial cross-linking has been used to maintain solid-liquid equilibrium in soft matter applications like cell culture, protein adsorption or ionic transport (in energy storage) [84-87]. Due to the complex structures generated by cross- linking, the resulting materials are considered a network and their characterization is primarily statistical.
In this dissertation, infrared spectroscopy (IR) was employed to investigate the cross-linking reactions of polymer-silica hybrids, and to evaluate their potential for adsorption applications. Calorimetric techniques and solubility tests were used to evaluate the stability of the materials under various environments. A combination of the techniques introduced in this section was successfully applied to prepare poly(vinyl alcohol)-silica hybrids. Control over the properties of these materials involved the use of molecular characterization techniques that elucidated the effect of the phase inversion conditions, the pH of the pristine solution and precipitation bath, the nature of the template, and the extent of the cross-linking reactions.
5 1.3 CO2 Capture and Separation by Amine-functionalized Polymer-silica Hybrids
Separation of CO2 from mixed gases, containing CO, CH4, H2, N2, and O2 has been studied extensively. The results from these studies have provided the basis for the development of separation technologies in large scale applications including natural gas sweetening, purification of synthesis gas, and conditioning of enclosed spaces [88, 89].
Well known technologies for CO2 capture involve adsorption and absorption processes, membrane separation, chemical looping, and cryogenic condensation. These technologies offer specific advantages depending on the type of industry they are applied, but have multiple limitations of flexibility and large scale operation for CO2 capture from large stationary sources [90-93]. The separation of CO2 from industrial processes currently involves sophisticated technologies which costs are justified by the high value of the end products. The development of more economic technologies that offer similar efficiency for treating large amounts of CO2 are still a challenge for the research community. The CO2 capture and separation processes described in this work are primarily considered for applications in emissions control from coal-fired power plants.
1.3.1 Composite Membranes for Separation of CO2
Composite membranes for separation of CO2 consist of a porous support and an active amine layer. The porous support offers mechanical stability to the membrane and provides the structure for incorporation of an amine layer. The amine layer is incorporated on one of the surfaces of the support or in between two supports. This layer provides the functional groups for the selective permeation of CO2 and allows facilitated transport across the membrane. A viable alternative for the existing CO2 capture technologies may be accomplished by membrane separation technologies [94]. The main
6 advantages of using membrane separation technologies include: high energy efficiency, simplicity of operation, and high system flexibility [95].
Most membrane separation technologies have been accepted for treating small volumes of gas, and the motivation of investigating polymer membranes for separation of
CO2 involves their potential for obtaining (i) high permeability and selectivity, (ii) high mechanical, thermal, and chemical resistance, and (iii) more economically feasible fabrication processes in large scale manufacturing [96]. The CO2 separation using membranes has been proposed to follow the solution-diffusion mechanism and the diffusion-molecular sieving effect. Although these theories do not describe the specific mechnaisms of each membrane separation system, they have served as the basis for investigating the contribution of transport effects and physical separation throughout the membrane. In this dissertation a preparation technique is presented, that involves phase inversion and partial cross-linking of a polymer blend. The concentration gradients and cross-linking reactions across the membrane were studied by FTIR spectroscopy using a focal plane array (FPA) image detector.
1.3.2 CO2 capture by Amine-functionalized Solid Sorbents
Adsorption of CO2 by amines is being developed to reduce the emissions of CO2 generated by large stationary sources. These sources include coal-fired power plants, natural gas-fired power plants, and chemical industries, which flue gas contain between
10 and 20 % of CO2. The phenomena involved in CO2 capture from amine- functionalized solid sorbents are chemical and physical adsorption. Functionalization of amines on solid supports through grafting, impregnation, or direct synthesis changes the adsorption modes of CO2, and consequently the nature of the adsorbed species. During 7 the adsorption of CO2 on immobilized-amine solid sorbents intermediates such as ammonium ions, carbamates, and carbamic acid are formed [97-100]. These intermediates desorb as molecular CO2 by changes in temperature, pressure, or concentration; the processes are regarded as thermal swing adsorption (TSA), pressure swing adsorption (PSA), vacuum swing adsorption (VSA), or sweep stream desorption.
Water vapor is commonly used to enhance the adsorption capacity of the sorbents because it changes the stoichiometry of the CO2 adsorption reaction [32, 97, 101, 102].
Amine-functionalized solid sorbents react with CO2 via acid-base interactions, which thermodynamic and kinetic parameters can be controlled by adjusting the amine density, the functional groups of the additives, and the morphology of the supports. The formulation of these sorbents is based on the spectroscopic investigation of the interaction of CO2 with amine molecules under simulated CO2 capture conditions [103-
107]. Industrial processes involve fluidization and require degradation-resistant sorbents in the form of pellets. For this dissertation a cross-linked polymer binder solution and a scalable pelletization process were designed and evaluated using silica-based amine- functionalized solid sorbents. Pelletization creates a strong CO2-permeable network resistant to attrition, and oxidation. The performance of these pellets was tested in fixed bed and fluidized bed reactors, using in-house fabricated lab-scale and bench-scale testing units. The CO2 capture processes were designed for rapid cycling, and for production of highly concentrated CO2. A compartmental modeling technique was developed to simulate the adsorption of CO2 in fixed bed reactors, and to assess the thermodynamic and kinetic parameters used for evaluation of the commercial viability of new sorbents.
8 1.4 Catalysis of Organic Reactions
The study of materials for CO2 capture has illustrated innumerable unique properties of amine-functionalized solid sorbents and amine-functionalized polymer membranes. Some of the properties of these materials have motivated the last two chapters of this dissertation, which are dedicated to find novel applications where the related fundamental concepts and technological approaches can be applied. Amines have basic and nucleophilic properties; they can donate their lone electron pair and are likely to combine with hydrogen ions (H+). The electron pair in amines is attached to the nitrogen atom, which is highly electronegative; these electrons can attract the positive part of an ion to form polar bonds. Amines react with acids to form ammonium ions, and consequently acid-base salts [108]. These are the type of reactions that occur during the adsorption of CO2, and can be extrapolated for base-catalyzed organic reactions. In addition, the capability of amines to form hydrogen bonding with other molecules and within their own structure could be manipulated to provide the proper hydrophilic/hydrophobic balance to a hybrid material [103, 109-111]. This dissertation presents the study of silica-supported amine catalysts for organic condensation reactions, and the development of polymer composite membranes for enzyme immobilization.
9 1.5 Approach
The synthesis of micro-structured porous materials typically involves a number of steps; dissolution, templating, condensation/polymerization, and separation/purification.
Most of the synthetic routes reported in literature for porous silica, porous polymers, and porous polymer-silica hybrids are highly specialized and have little consideration of the economics and scale up limitations. The majority of these synthetic routes include costly or unavailable raw materials, excessive use of solvents, and multiple separation steps.
These processes have aroused from the initial interest to investigate the fundamental mechanisms and synthetic approaches that result in the highest surface area and mechanical stability. Low cost raw materials and simple fabrication processes for the synthesis of polymer-silica porous materials still need to be identified and evaluated.
Amine functionalization techniques must be considered to enhance the properties of these materials towards adsorption of CO2 and catalysis of organic reactions.
Poly(vinyl alcohol) (PVA) was selected for its hydrophilicity and large availability of hydroxyl functional groups. The availability of hydroxyl groups enables the cross-linking and functionalization of PVA for the fabrication of rigid, insoluble materials. The hydroxyl groups of PVA are of the same nature as those on the silica surface, which strongly interact making these two materials highly compatible. PVA and
SiO2 are commodity chemicals, produced in large scale by multiple companies, and readily available in various presentations. In addition, the production of silica materials can be achieved by the sol-gel synthesis, which includes the hydrolysis and poly- condensation of silica precursors, some of them being low cost commodity chemicals.
10 Direct templating has been used for the fabrication of porous materials for various decades. This method can generate pores in a solid structure from the molecular level to its macrostructure. The templating agents can be selected for generating pores of different sizes, and must be removed after condensation or solidification of the target material. Phase inversion is a technique driven by the thermodynamics of phase transitions in polymeric materials. The pore size and porous micro-structure are governed by the kinetics of the phase transition processes allowing the generation of multiple structures by precipitation of a polymer from a solution. This dissertation used a combination of these four techniques (i,e,. sol-gel synthesis, cross-linking, direct templating, and phase inversion) to synthesize polymer-silica hybrids in various shapes and sizes, and with a hierarchically porous micro-structures.
Amine functionalization can be achieved by introducing an amine molecule during the synthesis of polymer-silica hybrids, or by post-synthesis functionalization techniques such as grafting or impregnation. Due to their nature, amine molecules usually react with the cross-linking agents or with the solvents in the precipitation bath.
The addition of an amine molecule during the synthesis of polymer-silica hybrids requires consideration of the stability of this molecule in the pristine solutions and in the final material. Grafting consists of the covalent attachment of an amine molecule to the support. This molecule may be attached to the polymer domains, to the silica domains, or both, depending on the surface chemistry and nature of the amine molecule.
Impregnation is the simplest of the functionalization techniques, where the amine molecules are deposited on the surface of the support by physical interactions instead of covalent bonding. This technique usually requires the use of a solvent to disperse the
11 amine molecules on the support. In this work, an example of each functionalization method is presented.
The investigation of physicochemical phenomena involving adsorption, surface reaction, and desorption, requires molecular characterization approaches that help elucidate the reaction pathways and intermediates of a process. Fourier transform infrared spectroscopy (FTIR) is the most useful technique to observe the dynamic behavior of adsorbates on a surface, and to elucidate the reaction mechanisms. Mass spectrometry (MS) can be utilized to quantify the concentration of the effluents of a process involving the flow of a gas stream. This dissertation used step-switch, pulse input, and temperature programmed desorption (TPD) in combination with FTIR spectroscopy and MS spectrometry to elucidate the adsorption mechanisms and concentration of adsorbed species in CO2 capture sorbents. Similar approaches were used to observe (i) the interfacial cross-linking reactions of PVA/PEI composite membranes, where the reaction was followed in-situ by FTIR spectroscopy, using a focal plane array (FPA) detector, and (ii) to evaluate the activity of basic catalysts toward solvent-free organic condensation reactions, which was analyzed using an in-house fabricated micro-scale FTIR reactor.
The concentration of enzymes, dyes, and other substrates in aqueous solutions can be determined by UV-vis spectroscopy. This is a reliable technique in the quantification of components in liquid phase since the light absorbance is proportional to the concentration. UV-vis was used to measure the adsorption loading of enzymes on PVA composite membranes and to determine the activity of free and immobilized enzymes by indirect detection of H2O2. 12 1.6 Hypotheses
The general objective of this dissertation is to study the experimental techniques for the fabrication of amine-functionalized polymer-silica hybrids, and to evaluate their performance for adsorption applications. The general hypothesis consists of testing the feasibility of integrating silica and polymeric materials to create amine-functionalized hybrids for adsorption of CO2 and catalysis of organic reactions.
1.6.1 Fabrication of Polymer Silica Hybrids
One fundamental goal of this research is to provide the experimental techniques for the fabrication and functionalization of porous polymer-silica hybrids. Various techniques have been proposed for the fabrication of these kinds of materials including (i) sol-gel synthesis, (ii) cross-linking, (iii) direct templating, and (iv) phase inversion. The hypothesis is that the morphology of the polymer and silica segments in polymer- silica hybrids can be controlled by the proper combination of these techniques. The results presented in CHAPTER IV exhibit the feasibility of this hypothesis, and present direct evidence of the successful preparation of polymer-silica hybrids. The techniques mentioned above can be combined in a simple fabrication process to produce hierarchically structured porous materials from a single solution containing PEI, PVA, and silica precursors. The process may consist of a single step, or multiple steps depending on the desired structure. The silica precursors hydrolyze at high pH values in a pristine aqueous polymer solution containing a templating agent. Poly-condensation of
SiO2 and phase inversion of PVA happen simultaneously by contact with a non-solvent in a precipitation bath, which also serves as a template removal agent. Cross-linking and
13 functionalization may be achieved in-situ in the precipitation bath or by post fabrication techniques.
1.6.2 CO2 Capture and Separation
Powder sorbents with high adsorption capacity, fast kinetics, and slow degradation rates have been developed in lab scale. However these sorbents find no industrial applications due to their small particle size and poor scalability. The hypothesis is that a cross-linked polymer binder solution can be integrated with silica-based CO2 capture sorbents to generate degradation resistant pellets without significant loss of CO2 capture capacity. The formulation of a binder solution and the design of a scalable pelletization process are presented in CHAPTER VI. The fundamental concepts obtained in CHAPTER IV and CHAPTER V served as the basis for this study. The results that validate this hypothesis show the formation of a polymer- silica network with CO2 capture capacity similar to that of the original sorbent. The pellets were tested in lab scale, and bench scale CO2 capture units, and were resistant to attrition under fluidization and ball mill testing conditions. CHAPTER VII presents a compartmental modeling technique to evaluate the thermodynamic and kinetic parameters of the adsorption of CO2 on these sorbents. The modeling technique and the results from this study can be used in the techno-economical evaluation of emerging CO2 capture technologies.
1.6.3 Basic Catalysis and Enzyme Immobilization
The cross-linking reactions of PVA with glutaraldehyde (GA) are well known to be acid catalyzed, however in the preparation of CO2 capture pellets and polymer membranes acid catalysts were not included expecting to protect the amine sites and 14 increase the basicity of the final materials. For the applications presented in this dissertation, the cross-linking reactions have been observed to occur in the presence of
PEI, a Lewis base. The hypothesis is that the basic sites of PEI catalyze the cross- linking reactions of PVA in presence of GA and water, and are able to catalyze basic organic condensation reactions. Through the investigations presented in CHAPTER V and CHAPTER VIII, it was demonstrated that amine molecules have the ability to catalyze organic condensation reactions in the liquid phase. In the course of these studies a high throughput characterization technique was presented to evaluate the cross-linking reactions of composite polymer membranes in a single study, where all the characterization points are under the exact same environment.
1.7 Objectives
The goal of this research is to examine the hypotheses presented in section 1.6 by
(i) fabricating and characterizing micro-structured porous polymer-silica hybrids in the forms of particles, membranes, and pellets, (ii) formulating amine-functionalized solid sorbents for CO2 capture, and composite membranes for separation of CO2, and (iii) investigating physicochemical phenomena related to polymer silica hybrids including acid-base chemistry, adsorption, and heterogeneous catalysts. Other specific objectives include (iv) the fabrication and characterization of polymer-silica hybrid membranes by integration of sol-gel synthesis, cross-linking, direct templating, and phase inversion techniques, (v) the development of a polymer-binder solution, and a pelletization process for silica-based CO2 capture sorbents, and (iii) the evaluation of the potential of polymer- silica hybrids in alternative applications like heterogeneous catalysis and enzyme immobilization.
15 1.8 Scope
The scope of this dissertation is outlined by the content of its chapters.
CHAPTER I presents introductory information, and outlines the motivation behind this work, which is based on extensive research and development efforts in the areas of polymer-silica hybrids, composite membranes, CO2 capture, basic catalysts, and enzyme immobilization. CHAPTER II provides the background and pertinent literature review on the preparation techniques, inherent reactions, and physicochemical properties of polymer-silica hybrids, and their functionalization for use in separation of CO2 and catalysis of organic reactions. CHAPTER III describes the experimental techniques for fabrication and characterization of these materials, and the experimental apparatuses and procedures utilized in the spectroscopic investigations presented in this work.
CHAPTER IV reports the fabrication and characterization of hierarchically structured porous polymer-silica hybrids by processes that combine sol-gel synthesis, cross-linking, direct templating, and phase inversion. CHAPTER V presents a study of the interfacial cross-linking reactions of PVA/PEI composite membranes, observed in- situ by FTIR spectroscopy using a focal plane array (FPA) image detector. CHAPTER
VI presents the preparation and performance evaluation of CO2 capture pellets, for which a combination of the fundamental studies previously presented is applied. CHAPTER
VII shows a compartmental modeling technique to assess the kinetic and thermodynamic parameters of CO2 capture sorbents, and to evaluate the influence of the operation conditions in the adsorption of CO2. CHAPTER VIII presents a fundamental study of the potential use of amine-functionalized silica materials for basic catalysis of organic condensation reactions. CHAPTER IX presents and alternative application of composite
16 polymer membranes for immobilization of glucose oxidase. CHAPTER X summarizes the findings, and highlights the evidence for testing the feasibility of the hypotheses of this dissertation.
17 CHAPTER II
2BACKGROUND
2.1 Polymer-silica Hybrids
Polymer-inorganic hybrids have attracted considerable attention from the scientific and industrial communities for various decades. Incorporating functional macromolecules onto the surface of inorganic particles is one of the most effective routes to synthesize these types of materials. In some cases the functional macromolecules are adsorbed on the surface of the inorganic particles [56, 112-114], and in other cases they are grafted (i.e., covalently attached) to the surface functional groups of the inorganic segments [79, 106, 115, 116]. Other synthetic routes include copolymerization of silica and polymer precursors and encapsulation of organic components in sol-gel synthesis of silica particles [6]. These materials provide a good balance between the functionality of the polymer chain, and the mechanical, thermal, and chemical stability of inorganic compounds. The polymer-silica hybrids, as a particular case, have wide applications in various scientific and technological fields; this dissertation is an exploration of their applications in adsorption and catalysis.
The types of hybrids that can be prepared from polymers and silica include polymer-filled silica composites and silica-filled polymer composites, which
18 classification depends on their mass ratio. When the amount of polymer is smaller than that of silica, the material is a polymer-filled silica composite. In this case the polymer domains are completely surrounded by silica in a continuous matrix [113, 117], or located on the surface of silica particles as a continuous layer. The applications of polymer-filled silica composites involve surface phenomena like adsorption and catalysis
[2, 118-122]. When silica exists in lower proportions than the polymer, its domains are likely to be separated and completely surrounded by polymer chains; this is the case of silica-filled polymer composites [81, 123]. These kinds of materials are widely used for fabrication of engineered materials, adsorbents, electronic devices, and cell cultures [6].
Polymer-silica hybrids have been used for fabrication of highly porous materials, using the filler as a template, or porogen. This route is widely used for fabrication of hierarchically porous structures, and for the formation of highly ordered porous polymers. Table 2.1 lists a summary of some polymer-silica hybrids reported in literature, their materials sources, preparation procedures, and applications. In the scope of this work, the main interest is the fabrication procedures, reactions, and functionalization techniques of poly(vinyl alcohol) (PVA), polyethyleneimine (PEI), and silica hybrids. A background of the general concepts and specific techniques for fabrication of these materials is presented.
19 Table 2.1 Summary of literature reporting polymer silica hybrids fabricated by diverse methods. Material- Silica source Polymer source Preparation Ref. Application method Silica Silica nanoparticles Acrylamide Grafting [81] nanoparticle- (Stӧber method) monomer hybrid aerogels PEI-imprinted 3-(triethoxysilyl) Hyperbranched Silanol-PEI [124] highly porous propyl isocyanate poly(ethyleneimine) polycondensation, silica (PEI) and calcination Mesoporous 3-(methacryloxy) Methacrylic acid distillation– [55] silica propyltrimethoxy- precipitation nanospheres silane copolymerization, and calcination Silica-latex- Tetraethyl- Trimethoxysilyl- Co-hydrolysis/co- [115] fiber aerogel orthosilicate (TEOS) grafted latex condensation for insulation Mesoporous Tetraethyl- Glycidyl Sol-gel [122] magnetic silica orthosilicate (TEOS) methacrylate condensation and for protein nanoparticles deposition, and adsorption calcination Nanoparticle Silica-silane Polyetheramine Ionic grafting [106] organic hybrids colloidal suspension (Jeffamine) for CO2 capture Cationic Silica- Silicic acid Cationic Polymer-salt [125] Polyamine polyallylamine aggregate Nanoparticles hydrochloride assembly Nanosized Tetramethoxysilane Cu-Benzene Size-constrained [83] silica-filled (TMOS) tricarboxylate sol-gel synthesis porous coordination polymers Silica- Tetramethoxysilane Acrylamide and Sol-gel synthesis [82] polyacrylamide (TMOS) bis-acrylamide and photo co- aerogels polymerization Liquid-like Colloidal and Mono amine Ionic or covalent [116] nanoparticle nanosized-silica terminated bonding hybrids for suspensions polyethers and CO2 capture polyethyleneimine Cross-linked Tetramethoxysilane Phenylenediamine Sol-gel synthesis [126] sol-gel hybrid (TMOS) and 3- (three isomers) and and oxidative coatings glycidyloxypropyl polymerization trimethoxysilane epoxy/amine (GPTMS) cross-linking
20 Table 2.1 Summary of literature reporting polymer silica hybrids fabricated by diverse methods. (Continued) Material- Silica source Polymer source Preparation Ref. Application method Polymer-silica Colloidal silica, Methyl- Sol-gel [113] hybrid methacryloxypropyl- methacrylate, butyl- coprecipitation emulsions for trimethoxysilane, acrylate and, coatings aminopropyl- methacrylic acid applications triethoxysilane, and emulsions glycidoxypropyl- trimethoxysilane Hybrid SBA-15 from Methacrylic acid Impregnation and [127] nanoparticles tetramethyl monomer, and in-situ for protein orthosilicate ethylene polymerization adsorption and (TMOS) dimethacrylate controlled release Dielectric 3-aminopropyltrie- Oxydiphthalic Sol-gel synthesis [128] polyimide- thoxysilane anhydride, 2,6-bis and poly- silica thin film (3-aminophenoxy) condensation hybrids benzene PE-silica Tetraethyl- Polyethylene-block- Simultaneous [129] interpenetrating orthosilicate (TEOS) polyethyleneglycol cross-linking and network polycondensation Hierarchically Tetraethyl- Polyacrylamide Templated [4] porous silica orthosilicate (TEOS) (PAM) emulsion-Stӧber nanospheres synthesis, and calcination Biocompatible Sodium silicate Polyethyleneglycol Simultaneous [117] PEG-silica emulsion co- particles precipitation Silica-polymer Trimethoxysilyl- Poly(butyl Co-hydrolysis- [130] aerogels for modified poly(butyl methacrylate-co- deposition inslutaion methacrylate) butyl acrylate) nanoparticles Mesoporous Tetraethyl- Acrylonitrile Micelle templated [9] polymer-silica orthosilicate (TEOS) monomer polymerization- hybrids and sol-gel synthesis, carbonaceous and carbonization supercapacitors Coated silica Silica gel particles Various Solution coating- [131] for enzyme acrylonitrile deposition immobilization copolymers
21 2.1.1 Sol-gel Synthesis
Silica is an amorphous inorganic polymer with a repeating unit of siloxane groups
(Si-O-Si) in the bulk, and silanol groups (Si-OH) distributed on the surface [132].
Variations in the chemical routes of silica result in modified morphological structures, and siloxane/silanol ratios. The synthesis of organic-inorganic materials using the method of hydrolysis and condensation of organic silica precursors of the type (RO)3-Si-
R’-Si(OR)3 or Si(OR)4 (silicon alkoxides) has been commonly used in scientific and industrial research. Also popular in industrial operations and highly beneficial from the economic point of view, is the method of hydrolysis and condensation of inorganic salts like sodium silicate (Na2SiO3) [117, 133, 134], or metal alkoxides (M-OR). All these methods are achieved by the sol-gel technique, which is the most widely used technique for fabrication of silica materials and organic-silica composites, as evidenced in Table
2.1. Sol-gel synthesis offers several advantages over other techniques because the reaction conditions are mild, offers high dispersion of the guest materials, ease to control the reactants concentrations, and homogeneous dispersion of the inorganic domains after hydrolysis and condensation. In this method the inorganic and organic domains are closely dispersed in the pristine solution creating molecular level interactions between the domains before precipitation [6, 57, 69, 135].
In general the fabrication of silica materials through the sol-gel technology can be classified into a series of steps involving hydrolysis, condensation, gelation, ageing, drying, and densification. Each of these stages is sufficiently slow that allows characterization, and in some cases, manipulation of the reaction mechanisms [136].
Although all these steps are involved in the preparation of silica materials, the basis of the
22 synthesis are hydrolysis and condensation. Hydrolysis proceeds though a nucleophilic substitution of one alkoxyl group from the silicon alkoxide with one hydroxyl group provided by water, (RO)3-Si-OH. This reaction proceeds for all alkoxide groups until the entire silica precursor molecule has converted to silicon hydroxide, Si(OH)4.
Simultaneously, a reaction occurs between the hydroxyl groups of one molecule, and the alkoxyl groups of another silicon alkoxide, linking two molecules together, and starting condensation. The hydrolysis reaction produces the terminal groups of the polymer; the surface silanols. Condensation produces the intra-molecular connections that build the inorganic network; the siloxane groups.
The initial silica domains that are highly dispersed in water are called a sol. This nano-sized domains are dispersed in water forming a continuous phase that has the same properties of a liquid. When these domains agglomerate and form a network, the viscosity of the liquid increases, and a gel is formed. The network formation requires high probability for the silica domains to collide with each other and react; which is governed by the pH, the concentration of the precursors, the temperature of the solution, and the addition of a catalyst. The hydrolysis and condensation reactions of silicon alkoxides can be either acid (H+) or base (OH-) catalyzed, following slightly different mechanisms but ending in the same final product [70, 136, 137].
For the acid conditions, the hydrolysis reaction of silicon alkoxides starts by protonation of the alkoxyl by the acid, followed by water attack to the silicon atom. The attack of the hydroxyl group of water reduces the strength of the proton, and displaces the alkoxyl, forming alcohol (R-OH) as a by-product. Under basic conditions, the attack of the hydroxyl group (either from water of from the base) is the initiation step and the by- 23 product is a negatively charged alkoxy (RO-). The hydrolysis of sodium silicate happens by dissolution in water, usually at high temperature, generating sodium hydroxide
(NaOH) [134]. After hydrolysis, the liquid is a sol, and initiation of the condensation is needed to generate a gel.
The condensation reactions of the silicon hydroxides are initiated by a proton or a hydroxide ion, leading to a rapid formation of a protonated, or negatively charged intermediate. This intermediate reacts with another silicon hydroxide molecule leading to condensation, which is usually a slower step. Silica gel is generated by the neutralization of the aqueous sodium silicate with acid. The neutralization of sodium silicate initiates a polymerization reaction that generates the gel when the interaction of separate silicon domains becomes significant. The reactions that describe the hydrolysis and condensation reactions of silicon alkoxides are [130]:
(RO)3 Si OR H 2O (RO)3 Si OH ROH
(RO)3 Si OH RO Si(OR)3 (RO)3 Si O Si(OR)3 ROH
(RO)3 Si OH HO Si(OR)3 (RO)3 Si O Si(OR)3 H2O
And the reactions that describe the hydration, hydrolysis, and condensation of sodium silicate are [134, 138]:
H Na2SiO3 H2O Na2O(SiO2 )x xH 2OSi(OH)4
Si(OH)4 Si(OH)4 (OH)3 Si O Si(OH)3
24 2.1.2 Cross-linking of PVA and PEI
Cross-linking is a process that creates multiple connection points between polymer chains. In this process two or more polymer chains are linked together by a branch or a side group forming covalent bonds, ionic bonds, or physical interactions including entanglement, hydrogen bonding, or hydrophobic interactions. The consequences of cross-linking are: (i) increase of the molecular weight and viscosity of the polymer, (ii) formation of 3D networks, (iii) improvement of the mechanical strength and thermal stability, (iii) changes in the reactivity and density of functional groups of the polymer, and (iv) potential for copolymerization and incorporation of fillers [139, 140].
More specifically for the scope of this work, covalent cross-linking of poly(vinyl alcohol)
(PVA) and polyethyleneimine (PEI) and their spectroscopic identification are considered.
The cross-linking reactions of PVA and PEI with glutaraldehyde (GA) consist of nucleophilic addition reactions of aldehydes with alcohols and amines. Alcohols and amines are both nucleophiles, and aldehydes are electrophiles. In general terms, these reactions can occur in two different fashions depending on the ability of the nucleophile to protonate the carbonyl oxygen from the aldehyde. If the nucleophile is negatively charged, water or an external acid must act as a catalyst to protonate the intermediate species and produce alcohols. If the nucleophile carries a hydrogen atom that can be used for protonating the intermediate species, the nucleophilic addition gives a product with a
C=Nu double bond. Figure 2.1 shows the reaction pathways for nucleophilic additions to an aldehyde or ketone.
25
Figure 2.1 Two general reaction pathways following addition of a nucleophile to an aldehyde or ketone. The top pathway leads to an alcohol product; the bottom pathway leads to a product with a C=Nu double bond. Adapted from [108].
PVA contains alcohol groups (hydroxyl) and may contain up to 30% of acetyl functional groups depending on its degree of hydrolysis [141]. The reactions expected between PVA and GA are those that occur by the reaction of alcohols with aldehydes, the acetyl groups are not expected to react with GA, but under mild acidic conditions or under the presence of strong bases, these groups are expected to hydrolyze. An aldehyde group reacts with two OH groups in the presence of an acid to produce acetals (-O-C-O-).
Figure 2.2 shows a structure of the expected product of PVA cross-linking [141]. Most authors agree with the formation of this 6-member acetal ring structure [141-147]. The most common acids used as catalysts are HCl [1, 56, 141, 143, 148], H2SO4 [149, 150], acetic acid [151], and H3PO4 [146]. Some authors have studied this reaction in absence of a catalyst [152], and some others have used basic catalysts [142, 153]. Although basic catalysts have been reported in literature, there are very limited studies about the use basic catalysts for the cross-linking reactions of PVA.
26 The mechanism of the acetal formation by reaction of PVA with GA follows the mechanism of and acid-catalyzed nucleophilic addition reaction: (i) protonation of the carbonyl oxygen by an acid, (ii) nucleophilic attack of the alcohol to the protonated carbonyl group, (iii) formation of a hemiacetal intermediate by deprotonation of the carbonyl group, (iv) dehydration by protonation of the hydroxyl group in the hemiacetal,
(v) addition of a second alcohol to the ionic intermediate, and (vi) deprotonation to yield an acetal structure and to regenerate the acid catalyst.
Acetal bridge
PVA
GA
Figure 2.2 Chemical cross-linking of PVA with GA catalyzed by acid.
The rate limiting step has been proposed to be the transfer of protons between the acid and the carbonyl species [149]. Furthermore, the rate of formation of hemiacetal from the reaction of PVA with GA has also been investigated [146]. These works showed the equilibrium between hemiacetal which is the intermediate species, monoacetal that is produced by the reaction of an aldehyde group of GA with a single alcohol of PVA, and diacetal (or acetal bridge) that is formed by the reaction of the aldehyde groups in GA with two alcohol groups of PVA.
27 For the IR evaluation of this cross-linking reactions, the important bonds are those from GA, PVA and acetal [1, 56, 141, 143, 148, 154, 155]. The aldehyde carbonyl group
(-CH=O) shows an intense peak around 1720 cm-1, that is expected to decrease with increase in the conversion. The alcohol groups show a broad band between 3550 and
3200 cm-1 that is expected to decrease due to the consumption of alcohol groups, and by the reduction of hydrogen bonding. The crystallinity of PVA produces an absorption band at 1141 cm-1 and is also expected to be reduced after the cross-linking reaction. The acetal and hemiacetal groups are observed between 1150 and 1080 cm-1, the intensity of these IR bands is expected to increase with increase in the conversion (i.e., degree of cross-linking).
The cross-linking reactions of PEI with GA are more complex than those of PVA.
Amines react with aldehydes by nucleophilic addition reactions in which water is eliminated to create an unsaturated product. Primary amines (-NH2) react with aldehydes to produce imines (-C=N-) and secondary amines react with aldehydes to produce enamines (-C=C-N-). Tertiary amines are not expected to react with aldehydes because the non-protonated nitrogen is a weak nucleophile. PEI contains primary, secondary and tertiary amines in the ratio 1:2:1, and is expected to experience the two reactions mentioned above.
28
Figure 2.3 Nucleophilic addition of amines; imine and enamine formation. Adapted from [108].
Imines are called Shiff bases, the reaction that yields to imines is acid-catalyzed and is susceptible to hydrolysis at almost any pH value [139, 156]. The mechanism of imines formation includes (i) the nucleophilic addition of the primary amine to the carbonyl group, (ii) an intramolecular proton transfer that produces a neutral intermediate, (iii) protonation of the intermediate by an acid catalyst or water, (iv) dehydration, and (v) proton transfer to regenerate the acid and yield the unsaturated imine product. Enamines have an insaturation on the C=C bond. The mechanism for producing enamines is similar to that for producing imines; the difference is that secondary amines do not have a proton to transfer after the nucleophilic addition. Instead, the intermediate species are protonated by transfer of a proton from the α-carbon, yielding to a C=C insaturation. These enamines are stable towards hydrolysis and the reaction is almost irreversible in a large pH range [156].
29 Enamine
PEI
GA
Imine
Figure 2.4 Chemical cross-linking of PVA with glutaraldehyde
Various authors have investigated the cross-linking reactions of amines with glutaraldehyde at different pH values [139, 156, 157]. These two reactions occur at both high and low pH values, but the fastest reaction rates are achieved at weakly acid conditions (pH= 4-6). When the acidity is too high (very low pH values), the amine group protonates yielding to an ammonium ion and the nucleophilic addition does not occur. The reversibility and stoichiometry of these reactions are an indicative of the proportion of imines and enamines formed during the cross-linking [156]. Other evidence of cross-linking is the change in color; an amine turns yellow or red by reaction with glutaraldehyde [139, 158]. Although there is extensive literature about the reactions of amines and aldehyde, there are very limited studies about the cross-linking of amines at interfaces, specifically the cross-linking of PEI with vapor phase GA and in absence of a catalyst, which is the reaction studied in this dissertation [158, 159].
The IR bands observed during the cross-linking of PEI with GA are those from primary amines, secondary amines, aldehydes, imines, and enamines. Primary amines
30 present two stretching vibrations between 3500 and 3400 cm-1, and secondary amines present only one stretching vibration in this range. The N-H bending vibrations for both primary and secondary amines have an absorption band between 660 and 900 cm-1. The
NH bending vibration for secondary aliphatic amines appears around 1600 cm-1, and for aromatic amines or polymeric amines appears at a lower frequency, around 1570 cm-1
[108]. The aldehyde carbonyl group (-CH=O), as mentioned before, shows an intense absorption at 1720 cm-1. All these bands should decrease with the conversion.
The band produced by the C-N bond between 1020 and 1250 cm-1 should show a slight shift to higher frequencies due to the production of tertiary amines. The insaturations of both imines and enamines present an absorption peak around 1650 and
1950 cm-1. The intensity of these bands is expected to increase with the conversion.
Table 2.2 lists the most significant IR band assignments for PVA, PEI and their cross- linked products.
31 Table 2.2 IR band assignments for PVA, PEI, GA and the cross-linked products.
Functional group Chemical species Wavenumbers (cm-1) OH (hydrogen bonded) PVA 3550-3200 CH from alkyl groups PVA 2840-3000 CH from aldehyde GA 2830 and 2695 CH from amine PEI 3000-2810 C=O from acetate PVA 1750-1730 C=O from aldehyde GA 1730-1700 O-C-O (crystallinity) PVA 1141 C-O-C PVA/GA 1150-1085
CH2 PVA 1460-1410
CH2 PEI 1490-1420 C-O-C-O-C acetal bridge PVA-GA 1150-1080
NH2 PEI 3500-3400 (two peaks) NH PEI 3500-3400 (one peak)
NH and NH2 PEI 900-660 C-N PEI/PEI-GA 1250-1020 C=C from enamine PEI-GA 1950 and 1650 C=N from imine PEI-GA 1685-1635
32 2.1.3 Direct Templating
Direct templating is a technique used for fabrication of porous materials. In this technique a template, or porogen is blended within the solid material or its precursors to create a mixed matrix, and generate voids after removal. This is a technique for replicating the inverse structure of pre-formed templates. In general, direct templating involves three fundamental steps [68, 124, 160, 161]:
1. Template fabrication and assembling: in this step, surfactants, nanospheres,
nanotubes, or fibers are used to create a template. The template
accommodates in characteristic configurations that allows the assembling of
nano-structured domains. The morphology of these domains depends on the
nature of the template and the conditions of the mixture, which is usually in
liquid phase. Under the proper conditions of temperature, concentration, and
pH the template self-assembles in micelles, and is able to adsorb the
precursors of the final material.
2. In-situ polymerization, solidification, or cross-linking: in this step the
precursor has been deposited on the template domains and a polymerization or
condensation reaction starts to increase the molecular weight of the
precursors, creating a continuous material around the templates. The reaction
conditions must be selected in a way that the self assembled structure of the
template is not disrupted before a high polymerization degree has been
achieved.
3. Removal of the template: in this step the template domains are removed from
the polymerized or condensed solid. The removal of the template may be
33 achieved by washing, solvent extraction, dissolution, or calcination. After the
template is removed, it leaves voids that construct the porous structure of the
final material.
~2.5 nm ~10 nm
~50 nm ~50 nm
Figure 2.5 Preparation of porous materials by the direct templating method.
This technique consists essentially of the fabrication of an ordered structure by replicating the inverse of another structure. The self assembly of different templates generates different structures that may be spherical, tubular, or ordered matrices. There are several requirements for the preparation of porous materials by direct templating [7-9,
162]. The primary requirement is that the template must be able to self assemble; it must have a well a defined structure that is stable under the reaction conditions. The surface properties of the templating agent must be compatible with those of the solid precursor.
Good compatibility leads to a close replication of the self assembled structure, and to more homogeneous pores. Sometimes additives or surface modification techniques are required to increase the compatibility of these materials [163, 164]. The template is also required to be easily removed under conditions that do not disturb the structure of the final material. After removal, the surface of the porous material must have selected characteristics for post-functionalization or for direct use in specific applications. The surface chemistry of the template must allow these functionalities to remain on the
34 surface or must create new functional groups upon removal. The interactions between the template and the precursor material are normally non covalent bonds, and the adsorption properties of the template must be contemplated. When there are covalent interactions between the precursor and the template, more sophisticated technique are required for removal and functionalization [165, 166].
2.1.4 Synthesis of mesoporous materials
According to the IUPAC definition, porous materials are divided into three classes based on their pore size, as shown in Figure 2.6. Microporous materials are those with pore width smaller than 2 nm, mesoporous materials are those with pore width between 2 and 50 nm, and macroporous materials are those with pore width larger than
50 nm. Materials containing more than one type of pores are usually regarded as hierarchically porous materials* [68].
12-ring zeolite Microporous: < 2.0 nm (20 Å) 6 - 8 Å
SBA-15 Mesoporous: 2.0 nm ~ 50 nm ~5 nm=50 Å
Macroporous: > 50 nm (500 Å) Polymer membranes 0.5 – 5 µm
*Hierarchical porous structures
Figure 2.6 Porous materials: IUPAC classification by pore size.
35 One of the clearest examples of fabrication of porous materials by direct templating is the synthesis of mesoporous silica. For the purposes of this dissertation a brief summary on the synthesis of SBA-15 is provided. SBA-15 stands for Santa Barbara
Amorphous number 15, which is a mesoporous silica with pore size between 4 and 30 nm, wall thickness between 3 and 6nm, and high hydrothermal stability [71]. This material exhibits a structure with hexagonal disconnected channels (i.e., pores). SBA-15 is synthesized by using the non-ionic surfactant triblock co-polymer Pluronic 123 (P123) as a template. P123 is a poly(ethylene oxide)-b-poly(propylene oxide)-b-poly poly(ethylene oxide) copolymer with general structure EOxPOyEOx, where normally x=20 and y=70 units [163, 167]. Between 30 and 40 oC the EO chains of P123 are hydrophilic and the PO is hydrophobic, assembling in micelles that are cylindrical in shape, similar to those shown in Figure 2.5.
2.1.5 Phase Inversion
The phase inversion technique has been known for more than five decades for the preparation of porous materials and polymer membranes [74, 77]. Its applications range from separation of impurities in the chemical industry to the fabrication of advanced materials for biotechnology, pharmaceutical industry, and electronic devices [73]. The major application of phase inversion membranes is undoubtedly for separation of gas or liquid mixtures in industrial processes and for environmental remediation, which is within the scope of this work. The key factors affecting membrane preparation involve the nature of the solvents, the nature of the polymer, their concentration and temperature, the presence of additives, and the film casting conditions [72]. The separation of phases
36 results from the action of a non-solvent or by temperature changes. In general terms the process involving the action of a non-solvent consists of four steps, shown in Figure 2.7:
1. Dissolving the polymer in a good solvent
2. Casting a film or assembling specific morphologies
3. Precipitation in a non-solvent (i.e., a bad solvent for the polymer)
4. Drying
~5µm
Figure 2.7 Steps involved in the preparation of porous materials by the phase inversion technique.
Commercially available porous membranes are prepared by dissolving a polymer in an appropriate solvent and casting films of 20-200µm thick. The films are immersed in a precipitant and separation of the phases occurs. The phases obtained by immersing the homogeneous polymer film and the non-solvent are a polymer rich solid phase and a mixture of solvents in the liquid phase. The solid phase is a porous structure with pore size in the range of 0.5 to 50 µm [77]. The pore size and the pore size distribution of the resulting material can be controlled by varying the concentration of the polymer, the concentration of the non-solvent, the precipitation temperature, or by the addition of external substances that alter the solubility or viscosity of the pristine solution. The precipitation of the polymer may be alternatively induced without the effect of external substances, by reducing the temperature of the system below the solubility point of the
37 solution, causing the formation of a gel, and ultimately the precipitation of a porous polymer, this method is regarded as thermal gelation.
The thermodynamic requisite for precipitation, either by addition of a non-solvent or by thermal gelation is that the system must have a miscibility gap over a defined concentration and temperature range. The thermodynamic property that defines the phase inversion feasibility is the free energy of mixing, defined in simple terms by the Flory-
Huggins equation [168]:
F M n ln v n ln v n ln v n v n v n v kT 1 1 2 2 3 3 12 1 2 13 1 3 23 2 3
Where ΔFM is the free energy change of mixing, ni in the number of molecules of the component i in solution, k is the Boltzmann’s constant, vi is the volume fraction of the component i, and χij is the interaction parameter between the components i and j. The interaction parameter (χij) is proportional to the square of the difference of the solubility parameters of the components i and j:
ij (i dj)2
2 2 2 2 D P H
Where, the proportionality constant depends on the lattice coordination number of the system, and δi is the solubility parameter of the component I, which is composed of three contributions of forces: dispersion (δD), dipole-permanent dipole (δP), and hydrogen bonding (δH) [169]. Table 2.3 shows the solubility parameters for PVA, water, ethanol, and acetone.
38 Table 2.3 Hansen solubility parameters for PVA, water, ethanol, and acetone [169].
Substance δD δP δH δ PVA 17.2 13.6 15.0 26.6
H2O 18.1 12.9 15.5 27.1 Acetone 15.5 10.4 7.0 19.9 Ethanol 15.8 8.8 19.4 26.5
When ΔFM>0 phase separation occurs. The driving force for precipitation is the difference in chemical potential, which increases upon addition of a non-solvent inducing the polymer to flow from a region of low concentration, to a region of high concentration within the overall system [74]. The kinetic parameters that define the precipitation process are the diffusion coefficient and the viscosity of the solutions. As defined by the
Fick’s law, the diffusion coefficient is given by the diffusivity of the polymer in the solvent mixture, and by the change in chemical potential between the components of the mixture with changes in the local composition of the system:
D D i i PS x i P,T
Where µi is the chemical potential of component i, and xi is the mol fraction of the component i in the system. In the case of polymer precipitation, the diffusion coefficient is negative, which defines the driving force for the polymer to move from more diluted to more concentrated regions [74]. The formation of the porous polymer by addition of a non-solvent is illustrated in Figure 2.8. In the isothermal ternary diagram, the ternary mixture exhibits a miscibility gap in a region of high concentration of polymer and non- solvent. Upon addition of non-solvent to the initial solution A, the composition of the 39 system follows the trajectory AB, where the system exists in a single phase. When the composition of the system reaches the miscibility gap, the system separates into two phases. At this point, the polymer diffuses to a more concentrated region, because the solvent is gradually replaced by the non-solvent, and then the system follows the trajectory BC. After the solvent has been completely replaced by the non solvent, the solid reaches its final composition C leading to a porous material[73].
Due to the non-equilibrium conditions in real processes, the trajectory of the curve AC and the final point C, are governed by the kinetics of the precipitation process. The kinetic parameters determine the rate of exchange between the solvents, the size of the domains in the precipitating mixture, and ultimately the morphology of the final material. The thermodynamic considerations of a ternary system involving polymeric materials are rather complex, and since the kinetic parameters change continuously during the phase inversion, the research in this area is highly empirical
[168].
40 Non-solvent
Two phases C: Porous One phase material
Solvent Polymer B: Unstable
A: Initial solution
Figure 2.8 Ternary diagram illustration of the formation of a porous polymer by the phase inversion technique.
41 2.2 CO2 Capture by Solid Sorbents
Solid sorbents technologies for CO2 capture have been developed mostly for applications in environmental remediation. Recently, the Environmental Protection
Agency (EPA) issued a communicate expressing its interest of imposing CO2 emissions regulations for existing coal-fired power plants, and reducing the fabrication of new power plants until the regulations are settled. The motivation behind this action is driven by the impact of CO2 in environmental issues that is believed to cause climate change and adverse environment for the world’s ecosystems [90]. The main focus of CO2 capture today is on stationary sources, of which coal-fired power plants are those that produce the highest impact due to their large emissions. Carbon capture is only a part of the process to reduce these emissions. Storage and re-disposal procedures are strictly necessary for completing the carbon cycle [93]. The entire process is regarded as Carbon Capture,
Storage, and Utilization (CCUS).
Solid sorbent technologies may be applied in power plants involving pre- combustion, post-combustion, or oxyfuel combustion. Pre-combustion is achieved by gasification of the fuels in gasification combined cycles, oxyfuel combustion and post- combustion are achieved by regular combustion in fuel-based power plants. Currently, these techniques are not in a commercialization stage due to (i) their limited scalability,
(ii) their high energy consumption, (iii) their negative impact on the pressure drop and thermal efficiency of the thermodynamic cycle, (iv) and the increase in the cost of electricity (CoE) [170]. Amine based aqueous processes using aqueous solutions of monoethanolamine (MEA), diethanolamine (DEA), diglycol-amine (DGA), and other water soluble amines has been widely used in refineries and gas industries for treating gas
42 streams with higher partial pressures of CO2 and smaller volumetric flow rates than those of the flue gas of coal-fired power plants [171, 172]. The current operational conditions in industrial processes are significantly different from the operation al conditions in power plants. The gas streams are usually at high pressure which increases the driving force for adsorption of CO2. In addition, the application of aqueous amines processes is justified in these industries because the products have high value, and incorporate significant benefits for the economy of the sector. On the other hand, the motivation of
CO2 capture for environmental purposes increases the operational costs, and generates capital, maintenance and market fluctuations [42].
Techno-economical analyses present data supporting that aqueous amine processes could increase the cost of electricity in 80% [41, 93], and the solid sorbent technologies could increase this cost in only 30-50% [45]. These values are not yet suitable for commercial applications, and extensive efforts are being made to reduce the cost associated to the sorbent fabrication, the initial capital, and the operational conditions. The nature of solid sorbents allows them to have low regeneration energy, high adsorption capacity, high selectivity, reduced degradation, easy handling, and more convenient disposal than aqueous amines. In addition, the solid sorbent technologies present low corrosion to the industrial equipment, and are flexible to operate under a wide range of conditions, or under fluctuations. These technologies are driven by the thermodynamic and kinetic parameters that drive any adsorption process [29].
Depending on the nature of the sorbent and its adsorption properties, solid sorbent technologies may operate based on the driving force cause by a change of temperature
(TSA processes), on a change of total pressure, (PSA and VSA processes), or on a change
43 of concentration (i.e., by a sweep gas desorption that reduces the partial pressure of CO2)
[173].
The design of the sorbent must satisfy the economical and operational criteria for
CO2 capture from a flue gas. The adsorption characteristics must fulfill the requirements of high adsorption equilibrium and fast kinetics. High adsorption equilibrium defines the potential adsorption capacity of the sorbent and the operational temperatures and pressures, and has a significant impact on the adsorber design and the plant capacity.
Fast kinetics defines the cycling time and the potential volumetric flow rate to be treated, also playing an important role in the operational design and capital cost [42, 90]. Due to the diverse source of fuels, the composition of the flue gas may vary, and may contain other molecules that potentially degrade the sorbent. The chemical characteristics of the sorbent must fulfill the requirements of high selectivity and low degradability. High selectivity improves the degradation resistant of the sorbent and facilitates the subsequent step of compression and transport for storage, and low degradability improves the resistance to impurities and chemical degradation, specially by O2, CO2, SO2, and NOx, that are the main components of the flue gas [102]. Compression and transportation increase the cost of the CCUS processes, and require high concentration of CO2 to operate at their maximal efficiency. Other factors that must be considered for the design of solid sorbents are: low regeneration energy, fast desorption, and high mechanical stability [93].
The morphology of the sorbent must be adequate for increasing the surface area of the sorbent, which increases the availability of amine sites and consequently the CO2 adsorption capacity. However, industrial processes are intended to work in fixed bed or 44 fluidized bed reactors, for which the use of a powder sorbent is not feasible. Sorbents for industrial operation are expected to be pellets or monoliths, and must be resistant to heat, pressure, steam, and attrition. The mechanical properties of the pellets are a determining factor in the operation of CO2 capture technologies in large scale; a smooth operation relies on low attrition and facile fluidization and pneumatic transport. The size of the pellets plays an important role in the adsorption kinetics, which inside the pellets are limited by the diffusion mechanisms. The fabrication of amine-functionalized sorbents has been investigated extensively [22, 174-182], however the development of pelletized materials has been limited to a few reports [50, 183].
CO2 capture takes place at the interface between the flue gas and the immobilized-amine solid sorbent, its mechanism is similar to that of a catalytic chemical reaction. At initial conditions of temperature, pressure and concentration the CO2 (i) travels from the bulk of the flue gas to a stagnant region close to the solid sorbent, (ii) diffuses through a stagnant region inside of the pores to the solid surface and (iii) adsorbs on the active sites of the sorbent. By applying a change in temperature, pressure or concentration, the CO2 (iv) desorbs from the surface, (v) diffuses through the stagnant region and (iv) returns to the bulk of the gas. CO2 is selectively chemisorbed on amines, where it forms carbonate, bicarbonate, ammonium ions, and carbamic acid [97, 184-188].
45
Figure 2.9 Proposed structures of adsorbed CO2 species; carbamate and ammonium ions, and carbamic acid.
Carbamate and ammonium ions are formed by the reaction of two amine molecules with one molecule of CO2, but in presence of water, the stoichiometry of the reaction can increase the CO2 capture capacity of amine sorbents. At high partial pressures of CO2, or in sorbents with low amine density (i.e., isolated amine sites), the adsorption has been observed to follow a stoichiometry where two amine sites react with two CO2 molecules. The adsorption requires high CO2/amine ratios because in this structure the carbonyl ion is stabilized by adjacent adsorbed CO2 instead of adjacent amine molecules. Figure 2.9 shows the proposed structures of carbamate and ammonium ions, and that of carbamic acid. The adsorption in dry and humid conditions is believed to follow the reactions [101, 102, 189]:
46 1. Carbamate and ammonium ions:
R R R R R R
2. Carbamic acid:
3. Carbonate:
R R R R
Although the mechanisms of CO2 adsorption are still debated in the research comunity, the identification of the adsorbed species has been largely supported by spectroscopic evidence. Many studies have put efforts into differentiating the IR absorbance bands of CO2 adsorbed on amine-functionalized sorbents. The proposed structures include carbamic acid, carbamate, carbonate, ammonium ions, carbamate- ammonium ions pairs, and zwitterions. Zwitterions are intermediates that arouse from adsorption of CO2 on two amine sites of a single molecule, or on two adjacent molecules
[98, 190]. Table 2.4 Proposed adsorbed species for CO2 adsorbed species on amine- functionalized solid sorbents.shows a summary of the literature review on identification of these species.
47 Table 2.4 Proposed adsorbed species for CO2 adsorbed species on amine-functionalized solid sorbents.
Wavenumbers Species Functional groups References (cm-1) + 3400, 3010, 2470, 2170, Ammonium ion RNH3 [188, 191, 192] 1633, 1494, 1326 Carbamate RNCOO- 1680, 1560, 1409 [110, 192-194] Carbamate- - + RNCOO NH3 R’ 1408, 1318 [185, 191, 195] ammonium pairs - + Zwitterions RNCOO NH3 R 3220, 2810, 2200 [98, 196] = Carbonate CO3 1337, 1390, 1575 [97, 102] - Bicarbonate HCO3 1432, 1493, 1634 [97, 102] Carbamic acid RNCOOH 1441, 1595, 1680-1700 [97, 194]
2.2.1 Membranes for separation of CO2
Polymer membranes containing amine (NH/NH2) and hydroxyl (OH) functional groups are suitable for further functionalization to achieve better physical properties and more selective permeability. The mechanical and thermal resistance of these membranes can be improved by cross-linking reactions with aldehydes, carboxylic acids, and epoxies
[197]. Their permeability may be adjusted by changing the density of the functional groups, which is affected by non-covalent interactions like hydrogen bonding and ionic bonding. The polymer membrane functionality determines its potential applications and defines its selectivity for separation processes.
Steric hindrance in polyamines enhances the CO2 transport on cross-linked
PEI/PVA membranes because the alkyl groups of PEI prevent the formation of carbamate ions, instead forming bicarbonate (with stoichiometry 1:1). A sterically hindered amine
48 is NH2-C (tertiary) or R-NH2-C (secondary or tertiary) [196, 198]. The total flux of CO2 across the membrane is:
DA CA/ P1 CA/ P2 DAB CAB / P1m CAB / P2m N A
Where N is the flux, D is the diffusivity, C is the concentration of CO2, P is the pressure of the compartment, and δ id the thickness of the membrane. The first term of the equation represents the flux due to solution-diffusion, and the second term represents the flux due to CO2 reaction with amines, which does not exist for other gases in the mixture (i.e., facilitated transport). The transport of CO2 is given in large proportion by the facilitated transport mechanism, making the terms of this equation: 1st term << 2nd term [199]. Permeated CO2 may be removed with steam flow. The solution-diffusion model considers that CO2 dissolves in the membranes, diffuses through their thickness driven by a pressure gradient, and desorbs at the other side [200]. Fixed amine sites transport CO2 by the hopping mechanism (i.e., adsorption-desorption-readsorption), and mobile carriers do it by facilitated transport (i.e., solution-diffusion).
49 CHAPTER III
3EXPERIMENTAL
3.1 Synthesis of SBA-15
SBA-15 was synthesized by direct templating and sol gel synthesis. The template triblock copolymer (Pluronic 123, BASF) was dissolved in water, and acetic acid
(Glacial, EMD chemicals) and sodium silicate (27% sln, PQ corporation) were slowly added. The mixture was kept for 24 h at 30 oC under vigorous stirring and then was hydro-thermally treated for 24 h at 100 oC in a closed container under static conditions.
The final mixture was filtered, and the solids were washed multiple times with DI water until obtaining a washed liquid with pH=7. After washing, the solids were dried for 12 h at 100 oC in a convection oven. Oven dry SBA-15 was labeled as SBA-15 a.p. (i.e., as- prepared). To further remove the template, the as-prepared material was calcined for 15 h at 600 oC under 25 cm3/min of air, and was labeled as SBA-15 cal (i.e., calcined).
The morphology of SBA-15 was characterized by SEM (JEOL-7401 Field
Emission Scanning Electron Microscope), TEM (JEOL JSM-1230 Transmission Electron
Microscope), and N2 adsorption (BET Micromeritics ASAP 2020). The surface and bulk functional groups, and the effectiveness of the calcination were characterized by
DRIFTS-IR spectroscopy (Thermo Nicolet 6700 FTIR).
50 3.2 Construction of PVA-water-ethanol and PVA-water-acetone Ternary Phase
Diagrams
The ternary phase diagram for PVA-water-acetone and PVA-water-ethanol systems was measured for a PVA sample with average molecular weight estimated in
75000 (75k) and degree of hydrolysis higher than 99%, PVA75k (Elvanol®71-30,
DuPont). The molecular weight of this sample was estimated from viscosity measurements; the viscosity of a 4% solution of this polymer at 20 oC is 33 cps, which corresponds to a molecular weight of around 75k (by regression from the data reported in
[201]).
Eight samples with initial compositions between 2.5 % and 20 % PVA in water were prepared by dissolving the corresponding amounts at 100 oC. The solutions were equilibrated at 25 oC, and aliquots of acetone or ethanol were slowly added on each sample. After each addition the system was vigorously mixed, and stabilized at 25 oC for
1 h. The existence of precipitates was evaluated by a cloud point technique using visual inspection. The cloud point observation was performed using white light and a black background, as shown in Figure 3.1; the hollow dots on top of each sample represent the compositions of the system where only one phase was observed. The solid dots represent the compositions of the system where coexistence of two phases was observed (i.e., the formation of a cloud or a precipitate). The binodal curve was constructed using the intermediate compositions between the last sample showing a simple phase, and the first sample showing coexistence of two phases. The composition of the system was normalized from 0 to 1 after each addition.
51 Non solvent
Figure 3.1 Cloud point observation for a PVA-water-acetone system with initial composition of 10% PVA in water. The hollow dots represent the compositions of the system where there is only one phase. The solid dots represent the compositions of the system where there is coexistence of two phases.
3.3 Fabrication of Polymer-silica Hybrid Membranes
Poly(vinyl alcohol) (PVA, Elvanol®71-30, DuPont), polyethyleneimine (PEI
50% aqueous, Mn=65k, Mw=750k, Sigma-Aldrich), and sodium silicate (Na2SiO3,
26.5% of SiO2, aqueous, Sigma-Aldrich) were used to prepare polymer-silica membranes. These membranes were prepared by sol-gel synthesis, direct templating, phase inversion, and cross-linking. A series of surfactants were used for templating the polymer-silicate solutions prior to phase inversion and cross-linking. The surfactants were selected for their size, and availability: (i) tetrapropylammonium bromide (TPA-Br,
98%, Sigma-Aldrich) is a widely used template used for the fabrication of zeolites; its micelle structure has an average diameter between 0.4 and 0.8 nm, (ii) sorbitan monooleate (SPAN80, Sigma-Aldrich) is a non-ionic surfactant used for the preparation of hierarchically porous materials, and as food emulsifier; its micelle size ranges from 2-
30 nm (depending on the solvent choice and concentration), and (iii) pluronic 123 (P123,
BASF) is a triblock copolymer with micelle size range between 4 and 6 nm. The average
52 micelle sizes are estimated based on the final pore size of porous materials reported in literature [202-204].
PEI-PVA-silicate solutions were prepared and casted into membranes for phase inversion and cross-linking. First, a PEI-PVA solution was prepared by dissolving 5g of
PVA in 50 ml of water at 100 oC, and adding 2 g of PEI after the temperature was lower than 50 oC. The PEI-PVA solution was stirred vigorously for 10 min, and mixed with another solution containing 2 g of sodium silicate (aqueous solution), 5.5 g of NaOH
(1M), 4 ml of Ethanol (200% proof, Decon labs), 16 ml of water, and (i) 0.3 g of
SPAN80, (ii) 0.45 g of TPA-Br, or (iii) 0.3 g of P123. The resulting solutions were heated to 75 oC and kept under vigorous stirring for 2 h. After cooling, the solutions were maintained under mild stirring overnight, and in static conditions for 24 h. These were the pristine solutions used for preparation of polymer-silica hybrids, and were labeled based on the template used: SPAN 80, TPA-Br, or P123.
The pristine solutions were sonicated for 30 min before casting. Each solution was casted on a Mylar® sheet using a 200 µm multifilm applicator (Sheen instruments) and (i) dried without further treatment, (ii) immediately immersed in ethanol for phase inversion, or (iii) immediately immersed in an ethanol solution containing 0.15 g of
GA/ml for phase inversion and cross-linking (GA=gluraraldehyde, 25% aqueous, Alfa
Aesar). All the membranes were dried at room temperature for 24 h. The morphology of the membranes was observed by SEM (JEOL-7401 Field Emission Scanning Electron
Microscope), their composition was evaluated by transmission-IR spectroscopy (Thermo
Nicolet 6700 FTIR), and their thermal stability was analyzed by thermo-gravimetric analysis (TGA, Q500 TGA, TA instruments). 53 3.4 Fabrication of PEI/PVA membranes for cross-linking studies
PEI/PVA composite membranes were prepared by casting thin layers of PEI on
PVA membranes prepared by phase inversion. The PVA membranes (i.e., the support) were prepared by dissolving poly(vinyl alcohol) (PVA, Elvanol®71-30, DuPont) in water at 100 oC, with the correspondent amounts for obtaining solutions with composition between 5 and 15 %. The solutions were casted on a Mylar® sheet using a multifilm applicator (Sheen instruments) with blade heights of 25 to 200 µm. Immediately after casting, the Mylar® sheets with the PVA solutions were immersed in ethanol (200 proof,
Decon labs.) or acetone (Sigma-Aldrich) for 5 min and dried at room temperature for 15 h. Polyethyleneimine (PEI, branched, MW=600 or 750k, Sigma-Aldrich) was dissolved in ethanol, and casted on one exposed surface of the PVA membrane. The PEI/PVA membranes were dried at room temperature for at least 2 h.
3.5 Preparation of Amine-functionalized Solid Sorbents
Silica based sorbents were prepared by incipient wetness impregnation of polyethyleneimine (PEI) or tetraethylenepentamine (TEPA, Sigma-Aldrich) on a silica support. PEI is a branched amine-hydrocarbon polymer that contains primary, secondary, and tertiary amines in a ratio of 1:2:1. TEPA is a special case of PEI with low molecular weight (198 g/mol) and linear configuration, with three secondary amines and two primary amines. TEPA and PEI were selected for their high density of amine sites, low toxicity, and high availability. Amorphous silica (SiO2, Tixisil 68B, Rhodia) was selected as a support because of its relatively high surface area (160 m2/g), low acidity, high thermal conductivity and low cost.
54 PEI and TEPA can be impregnated on silica supports where they experience hydrogen bonding interactions with the silanol groups (Si-OH) of silica (SiO2).
Polyethyleneglycol (PEG, Mw=200, Sigma-Aldrich), a cross-linker (CL), and an antioxidant (AO) were added as additives to increase the stability of the sorbent and reduce its regeneration energy. The names of the cross-linker (CL) and the antioxidant
(AO) were not revealed to protect the intellectual property of sponsored projects related to this dissertation. Figure 3.2 shows the schematic structure of TEPA and PEI impregnated on the surface of silica.
Figure 3.2 Schematic structure of PEI and TEPA impregnated on a silica support. The structure of PEI shown in this figure is a simple representation of a more complex, highly branched molecule.
The amine-functionalized sorbents (AFS) were prepared by mixing 40 g of SiO2 with an impregnating solution. The impregnating solution was prepared by (i) dissolving
22 g of TEPA, 15 g of PEG, and 4.5 g of CL in 40 g of ethanol, and (ii) adding 80 g of an
55 aqueous solution containing 0.65 g of AO. The mixture was dried in a convection oven at 100 oC and grinded to a fine powder. The polymer binder solution (PBS) was prepared by dissolving 1.5 g of PVA in 28.5 g of water (10% solution) at 100 oC, and adding 0.3 g of GA (25% aqueous). After cooling, the PVA-GA solution was slowly mixed with a solution containing 3 g of PEI (50% aqueous) and 1.5 g of an aqueous solution containing
2% solution of sodium bicarbonate (NaHCO3, Fisher) under vigorous stirring.
The pellets were prepared by mixing various amounts of AFS sorbent and PBS solution, which in the proper ratio results in a dough-like material. The dough was extruded using a manual ram extruder, spheronized using a rotary-disk spheronizer, and dried at 130 oC in a convection oven. Details about the preparation procedures, equipment, and characteristics of these sorbents are given in CHAPTER IV.
3.6 Experimental System for CO2 Capture Studies
The experimental system for CO2 capture studies is illustrated in Figure 3.3. This system consists of (i) gas flow controller manifolds, four and six port valves, (ii) a
DRIFTS cell reactor placed in a Harrick Praying Mantis casing, with sorbent capacity of
50 mg, attached to a Thermo Nicolet 6700 FTIR (IR), and (iii) a Pfeiffer QMS 200 quadruple mass spectrometer (MS). Ar, CO2, and air and were connected to mass flow controllers (MFC) for precise control of the flow rate and composition. The four-port valve switched the gas inlet to the DRIFTS reactor between Ar and CO2/air. The six-port valve was used to inject a known volume of gases for calibration.
Inside the DRIFTS reactor, the gases were restricted to flow from top to bottom of the sorbent bed. The temperature of the sorbent bed was monitored and controlled by a
56 K-type thermocouple, located in the center of the DRIFTS sample holder. A series of IR spectra of the sorbent was collected during the CO2 capture process. Each IR spectrum was formed by 32 co-added scans with resolution of 4cm-1. The composition of the reactor effluent was monitored by MS for N2 (m/e=14 and 28), O2 (m/e=16 and 32), Ar
(m/e= 40), and CO2 (m/e=44).
(ii) DRIFTS reactor
In (i) Mass flow controllers, 4 and 6-port valves Out
Ar MFC To vent Heating element Thermocouple
Air MFC 30 cm 15 cm
CO2 MFC 85 cm By-pass 80 cm (iii) Mass spectrometer
Figure 3.3 Schematics of the experimental system and DRIFTS reactor used in CO2 capture studies.
3.7 Step Switch Adsorption
A 4-port valve (Figure 3.3), allows switching the flow between two gas streams maintaining a constant flow rate and producing nearly-ideal step changes in concentration. A steady state condition precedes the step switch, and a transient state is generated by the change in composition. The analysis of this transient state is of interest for the scope of this dissertation. Ar was flow into the reactor until reaching steady state, and then, a step-switch changed the flow from Ar to CO2/air to evaluate the adsorption of
CO2 on the amine-functionalized sorbents. This transient process allowed collecting 57 information about the nature of the CO2 adsorbed species by IR spectroscopy, and measuring the concentration of the effluent of the DRIFTS reactor via MS.
3.8 Diffused Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS)
Diffuse reflectance is a multi-modal spectroscopic technique that comprises the information from two sample properties. The reflectance is determined by (i) absorbance, which is a linear function of the absorption coefficient, and (ii) scattering, which is caused by the reflection of the light on the surface and the first crystalline layers of the solid [192]. In a DRIFTS reactor the incident light beam is shined to the sample, the scattered beams are collected and refocused by ellipsoidal mirrors, and directed to the
IR detector. Details about the DRIFTS reactor and the sample holder are presented in
CHAPTER VII.
3.9 Attenuated Total Reflection Infrared Spectroscopy (ATR)
ATR is a widely used technique for characterization of liquids and solid surfaces.
In this technique the incident IR light is refracted and transmitted through an ATR crystal or internal reflectance element (IRE) that is in contact with a sample. The spectral information is collected from the total reflection of the light, which after being reflected by the sample returns to the IRE and is directed to the detector. The detection depends on the difference in refractive index of the IRE and the sample, which defines the penetration depth of the light into the sample. The approximate penetration depth is given by the expression:
d p 2 2 1/ 2 2np (sin nsp )
58 The penetration depth for ATR measurements ranges from 0.5 to 2 µm, and is function of the wavelength of the radiation (λ), the refractive index of the crystal (np), the angle of incidence of the light beam (θ), and the ratio of the refractive indices of the sample and the crystal (nsp). The technique offers the isolated analysis of a thin layer of the sample, which is of important use in this dissertation to analyze the formation of compact layers at the surface of composite membranes.
3.10 Transmission Infrared Spectroscopy
Transmission spectroscopy is a very popular and widely applied spectroscopic technique due to its simplicity. In this technique the incident light beam is transmitted through a sample, where is absorbed at specific frequencies. In this measurement, the ratio of transmitted light (I) to the incident light (Io) is recorded. The relationship between the concentration of a species in a sample and the ratio I/Io is given by the Beer-
Lambert law:
I Abs log cl Io
Where Abs is the absorbance, ε is the molar absorption coefficient (i.e., extinction coefficient), c is the molar concentration of the species, and l is the beam pathlength across the sample. This equation shows that absorbance is proportional to the concentration of the sample; a concept that was used in this work for (i) estimating the porosity of phase inversion membranes using IR spectroscopy, and (ii) measuring the adsorption loading and activity of glucose oxidase by a colorimetric technique using UV- vis spectroscopy. The samples for transmission spectroscopy can be present in solid, liquid or gas phase, and are usually located in a closed cell. The pathlength determines 59 the degree of beam saturation, and hence the strength of the signal at the detector. Due to the low density of gas samples, the pathlength for gas samples is usually larger than that for solid and liquid samples.
3.11 Focal Plane Array (FPA) image detector
A Focal Plane Array (FPA) image detector provides a spectrometer the characteristics of a microscope where instead of visible light, IR illumination is measured. An FPA is considered a multifunctional accessory due to its flexibility to operate in transmission, diffuse reflectance, or total reflectance modes. The most important application of an IR microscope is the measurement of the chemical distribution of a compound, which allows the detection of impurities, defects, and interfaces. By simultaneously generating multiple data, an FPA originates a mapping or a 3D contour of a small area in a sample, depicting the variations in intensity of IR light at a given frequency [15, 205]. An important advantage of the FPA is that it can record measurements in multiple points under exactly the same conditions, allowing direct cross-comparison of different sections of a sample without the uncertainty of experimental variations. More details about the FPA imaging are given in CHAPTER V.
60 3.12 Mass Spectrometry
Mass spectrometry provides information on the molecular weight and molecular formula of a sample. This technique is commonly used to complement information obtained from spectroscopic methods such IR, but is fundamentally different from spectroscopy because does not involve light, absorption or emission. A mass spectrometer ionizes the sample in a high vacuum and sorts the ions according to their masses by using electric and magnetic fields. The relative number of ions of each mass is measured from the ionized particles, and the abundance of each mass-to-charge ratio is registered [206].
61 CHAPTER IV
4FABRICATION AND STRUCTURAL CHARACTERIZATION OF POLYMER-
SILICA HYBRIDS
4.1 Introduction
Hybrid materials consist of microscopic polymeric and inorganic domains forming a macroscopic continuous phase. The preparation of these materials involves a wide range of techniques, which individually provide specific characteristics like porosity, surface functionality, mechanical, chemical, or thermal stability. The combination of two or more of these techniques is expected to result in an advantageous fabrication method to obtain advanced materials with defined micro-structures. Based on the literature review and background presented in CHAPTER II, this chapter presents the fabrication methods and characterization of porous materials comprising polymeric and silica domains.
The first example is the preparation of SBA-15, a mesoporous silica. The preparation of SBA-15 involves the methods of direct templating and sol-gel synthesis.
The template is the triblock copolymer Pluronic 123, a polymeric surfactant that self assembles in water in the form of rod micelles [163]. The size of the micelles is governed by the nature of the solvent, and by the concentration, pH, and temperature of
62 the solution. The silica precursors are typically silicon alkoxides; organic silicates that undergo hydrolysis and condensation by the action of a strong acid like HCl or H2SO4, generating silica and alcohols [5, 207, 208]. Silicon alkoxides are normally used for preparation of porous materials in high end applications like microelectronics and semiconductors [68], where their cost is justified by the overall process economics.
Other applications such as CO2 capture and catalysis demand mesoporous materials of lower cost. In addition, for these applications, traces of halogens or sulfur on the final material are not desirable.
An alternative for synthesizing highly ordered mesoporous silicas is by using low cost silica precursors like sodium silicate [209, 210]. Sodium silicate (~$40/L) offers the advantages of being a commodity material of large availability and lower cost than silicon alkoxides (TEOS $65/L, TMOS $1000/kg). P123 ($90/L) is considered a fixed cost in the preparation process of SBA-15 because it is the component that defines the micro-structure of the pores. Note that these prices were obtained from laboratory suppliers catalogs, and are not expected to directly reflect industrial costs. Another advantage of sodium silicate is that its condensation reaction does not require strong acids. Silica from sodium silicate precipitates at neutral pH, achieved by neutralizing it in aqueous solutions with weak organic acids [210]. Acetic acid is strong enough to catalyze the condensation reaction and is easily removed by calcination. The first example presented in this chapter shows the feasibility of synthesizing SBA-15 using sodium silicate and acetic acid, and reveals the surface functionality of the final material before and after calcination.
63 The second example consists of a series of hybrid membranes, that in contrast with SBA-15, are intended to contain both the polymeric and inorganic segments in their final stage. The fabrication of hybrid membranes may involve one or several preparation methods, depending on the desired properties of the final material. Investigation of materials in the form of membranes is a well defined way to study the characteristics of polymer-silica hybrids. The advantages include that membranes (i) require small amounts for testing, (ii) offer large area in two of their directions and small dimensions in their thickness direction, allowing the study of multiple intensive and extensive properties, and (iii) involve simple manufacturing procedures. One of the problems with the fabrication of porous hybrids and composites is controlling the morphology of the inorganic and organic segments in the final material [58, 211]. Miscibility and phase separation are usually a controlling factor in the processing of polymer blends and composites [76]. In this chapter a series of PVA/PEI/SiO2 hybrid membranes with hierarchically porous structure is presented. The porosity of these membranes and the morphology of their polymeric and silica domains were studied by combining various preparation methods. The fundamental concepts and experimental techniques presented in this chapter served as the basis for preparing structured materials for CO2 capture and enzyme immobilization.
4.2 Experimental
4.2.1 Synthesis of SBA-15
Figure 4.1 shows the schematics of the preparation procedure of SBA-15. SBA-
15 was synthesized by dissolving 7.86 g of triblock copolymer (Pluronic 123, BASF) in
146 g of DI water, and slowly adding 6.15 g of acetic acid (Glacial, EMD chemicals) and
64 22.53 g of sodium silicate (27% sln, PQ corporation). The mixture was kept for 24 h at
30 oC under vigorous stirring and was then hydro-thermally treated for 24 h at 100 oC in a closed container under static conditions. The final mixture was filtered, and the solids were washed multiple times with DI water until obtaining a washed liquid with pH=7.
After washing, the solids were dried for 12 h at 100 oC in a convection oven. Oven dried
SBA-15 was labeled as SBA-15 a.p. (i.e., as-prepared). To further remove the template, the as-prepared material was calcined for 15 h at 600 oC under 25 cm3/min of air and was labeled as SBA-15 cal (i.e., calcined).
Template assembling
Hydrolysis and condensation
Template Removal
Figure 4.1 Schematics of the synthesis process of SBA-15. The silica precursor is adsorbed on the surface of P123 micelles, where the condensation of silica occurs. The pores are generated by removal of P123 by washing or calcination.
65 4.2.2 Fabrication of PVA/PEI/SiO2 membranes
A series of surfactants were used for templating the pristine PVA-PEI-sodium silicate solutions: (i) Tetrapropylammonium bromide (TPA-Br, 98%, Sigma-Aldrich), (ii) sorbitan monooleate (SPAN80, Sigma-Aldrich), and (iii) pluronic 123 (P123, BASF).
The templated PEI-PVA-sodium silicate solutions were prepared by dissolving 5 g of
PVA in 50 ml of water at 100 oC, cooling to 50 oC and adding 2 g of PEI (50% aqueous).
The resulting solution was stirred vigorously for 10 min, and mixed with another solution containing 2 g of sodium silicate sln, 5.5 g of NaOH (1M), 4 ml of Ethanol (200% proof,
Decon labs), and 16 ml of water. Then (i) 0.3 g of SPAN80, (ii) 0.45 g of TPA-Br, or
(iii) 0.3 g of P123 were added to the solution under vigorous stirring. The resulting solutions were heated to 75 oC and kept under vigorous stirring for 2 h. After cooling, the solutions were maintained under mild stirring overnight, and kept in static conditions for additional 24 h. The solutions were labeled according to the template used: SPAN 80,
TPA-Br, or P123. The pristine solutions were sonicated for 30 min before casting. Each solution was casted on a Mylar® sheet using a 200 µm multifilm applicator (Sheen instruments), and were either (i) dried without further treatment, (ii) immediately immersed in ethanol for phase inversion, or (iii) immediately immersed in an ethanol solution containing 0.15 g/ml of GA (GA=gluraraldehyde, 25% aqueous, Alfa Aesar).
All the membranes were dried at room temperature for 24 h.
4.3 Results and Discussion
4.3.1 Synthesis of SBA-15
Figure 4.2 shows the schematics of the hydrolysis and condensation reactions of sodium silicate catalyzed by acetic acid. The silicate solution is partially hydrolyzed
66 when added to the initial P123 solution [138]. Upon addition of acetic acid, sodium is released from the silicate, taking the place of the proton, and leading to the silanol structure. The condensation between two silanol groups releases water and creates the siloxane group which is the building block of the mesoporous silica.
Precursor: Hydrated form:
Sodium silicate (Na2SiO3 ) Na2O(SiO2)x•xH2O (NaOH)x(Na2SiO3)y•zH2O
Acetic acid + Hydrolysis
Condensation
Silica (SiO2)
Figure 4.2 Hydrolysis and condensation of sodium silicate catalyzed by acetic acid. Sodium silicate is soluble in water where it forms basic solutions. SiO2 precipitates at neutral or acidic pH conditions.
Figure 4.3 shows the BET nitrogen adsorption isotherm for SBA-15 cal. This is a typical type IV adsorption isotherm, characteristic of mesoporous materials. At low relative pressure (P/Po<0.02) the surface of the material is partially covered with nitrogen molecules, and adsorbs up to 3.5 mmol N2/g before reaching a monolayer coverage. At higher relative pressure, SBA-15 exhibits multilayer adsorption (0.02
67 The BET surface area of 659 m2/g is in good agreement with that reported for
SBA-15 synthesized from TEOS and HCl. Moreover, the pore volume of 1.19 cm3/g, and pore size of 5.2 nm fall in the range of mesoporous materials and do not differ from those reported in literature for synthesis of SBA-15 using other precursors [5, 180, 207,
208, 212]. The physical characteristics described by the adsorption isotherm are direct evidence of the synthesis of a mesoporous material.
/g) 40 2 Surface area: 659.2 m 2/g Pore volume: 1.19 cm 3/g Pore size: 52.9 Ä ~ 5.2 nm 30
20
10
0
0.0 0.2 0.4 0.6 0.8 1.0 Quantity adsorbed (mmol N Relative pressure (P/P ) 0
Figure 4.3 Nitrogen adsorption-desorption isotherm for SBA-15 calcined produced by condensation of sodium silicate catalyzed by acetic acid.
68 ~100nm
~5nm
Figure 4.4 Microscopic images of SBA-15 calcined. Top: SEM, middle and bottom:TEM.
Figure 4.4 shows the SEM and TEM images of SBA-15 cal. The hexagonal assembly of the pores is clearly observed in these images. The SEM image, at the top of the figure, shows silica flakes with an average thickness of 100 nm. These flakes are likely to be formed by the assembling of micelle rods of P123 during the static hydro- thermal treatment. The TEM images show the uniform distribution of mesopores in these
69 flakes. From the BET results the pore size was calculated as 5.2 nm. Although the TEM image with highest magnification does not have enough resolution to provide the first decimal place, the pore size was estimated as an average of 5 nm, which is in excellent agreement with the BET calculations. The hexagonal nature of the template self- assembling resulted in the generation and hexagonal arrangement of pores, and furthermore, hexagonal shape of the flakes. These pictures reveal that the sol-gel conditions were properly selected to maintain the structure of the template micelles throughout the hydrolysis and condensation of sodium silicate.
Figure 4.5 shows the IR absorbance spectra of commercial silica, SBA-15 a.p., and SBA-15 cal, collected in DRIFTS at 100 oC. SBA-15 exhibits a sharp absorption at
3745 cm-1 that is attributed to the unbounded surface silanol groups, also called isolated or free silanol groups. SBA-15 a.p. presents a broad peak between 3700 and 3200 cm-1, caused by hydrogen bonding of its surface silanol groups with residues of the template, the catalyst, and water. The IR spectrum of commercial amorphous precipitated silica
(SiO2) was plotted for comparison. The spectrum of SiO2 shows the broad IR absorption of the OH stretching vibration between 3700 and 3200 cm-1, with an evident peak at 3669 cm-1 that is attributed to geminal silanols. Geminal silanols are two hydroxyl groups that did not participate in the condensation reaction and that are connected to the same silicon atom. Either single or geminal vicinal silanol groups may experience hydrogen-bonding within themselves or with a guest molecule like water, acetic acid, or P123 [134].
The three samples show the characteristic absorption peaks of the siloxane groups
(Si-O-Si) at 817, 1058, and 1182 cm-1. The siloxane groups are the functional groups of silica in the in the bulk of the material, and can be observed by diffuse reflectance due to 70 their proximity to the surface. The availability of surface silanol groups may be estimated by correlating the aspect ratio of the surface silanol groups to the bulk siloxane groups (Si-OH/SiO-Si). For these samples, the aspect ratio increased as SBA-15 cal
(0.46) > SiO2 (0.35) > SBA-15 a.p. (0.15). SBA-15 cal owes its large surface functionality to the rigorous removal procedure of the template. Calcination is a much more effective technique for template and catalyst removal than washing and liquid extraction. The incomplete template removal of these components in SBA-15 a.p. is evidenced by the other features of its IR spectrum. These features include a strong absorption at 1733 cm-1, characteristic of the carbonyl group of carboxylic acids, an intense absorption between 300 and 2800 cm-1 that is representative of the C-H stretching vibration of organic compounds, and the broadness of the siloxane bands that is caused by the poor penetration of the IR beam into the silica bulk. Undoubtedly these features are evidence of the presence of acetic acid and P123 on the surface of SBA-15 a.p.
Although SBA-15 cal is clearly more pure than SBA-15 a.p., both materials may serve for adsorption and catalytic applications. The evidence of strongly bound organic traces on the surface of SBA-15 a.p. is an indicative of its great potential to serve as a support in impregnation and post synthesis functionalization. One clear example is revealed in CHAPTER VIII, where the acid functional groups on the surface of the SBA-
15 a.p. catalyst could contribute with a proton for the termination step of an organic condensation reaction.
71
(Si-O)
(Si-O)
unbound
1182 1058
817 817
geminal
(C=O)
(Si-OH) 0.5
H-bond SBA-15SBA-15 calcined cal
1733 1733
3745 3745
(OH)
(Si-OH)
SBA-15SBA-15 as a.p prepared
3446 3446 3669 3669
CommercialSiO2 silica
3488 IR absorbance (a.u.)
4000 3500 3000 1800 1500 1200 900 -1 Wavenumbers (cm )
Figure 4.5 IR absorbance spectra and schematics of the proposed surface silanol groups of amorphous SiO2, SBA-15 as prepared, and SBA-15 calcined, collected in DRIFTS at 100oC.
72 4.3.2 Fabrication of Polymer-silica Hybrids
1 Pristine solution design: considerations based on the ternary phase diagram. Ternary phase diagrams for aqueous solutions of PVA with non-solvents were constructed to evaluate the feasibility of phase inversion. The non- solvents selected for
PVA were ethanol and acetone [201]. These diagrams serve for estimating the maximum concentration of non-solvent that can be added to the pristine solutions, and the minimum amount of non-solvent needed in the precipitation bath for achieving solvent induced phase inversion. The precipitation bath is normally a pure solvent (i.e., dry ethanol or dry acetone) that retains the amount of water initially present in the polymer solution after phase inversion. The ternary diagrams also provide information about the depletion of the precipitant bath, and may be used to determine the number of times it can be used.
Figure 4.6 shows the ternary phase diagrams at 25 oC for the systems PVA-water- acetone, and PVA-water-ethanol, constructed by visual observation of the cloud point of the system. The solid dots represent the compositions of the system where a single phase was observed, and the hollow dots represent the compositions of the system where a cloud or a precipitate was formed (coexistence of two phases). The system PVA-water acetone exhibits a large miscibility gap with a nearly symmetric shape. The miscibility gap of this system is large due to the significant difference between the solubility parameters (δ) of PVA and water with acetone. The miscibility parameters are reported in Table 2.3 Hansen solubility parameters for PVA, water, ethanol, and acetone [169].
73 0.00 1.00
0.25 0.75 Acetone
0.50 0.50 Water 0.75 0.25
1.00 0.00 0.00 0.25 0.50 0.75 1.00
PVA 75k 0.00 1.00
0.25 0.75 Ethanol
0.50 0.50 Water
0.75 0.25
1.00 0.00 0.00 0.25 0.50 0.75 1.00
PVA75k
Figure 4.6 Ternary phase diagrams at 25 oC for the systems PVA-water acetone, and PVA-water-ethanol. The hollow dots represent the compositions where the system exists in one phase and the filled dots represent the compositions where the system coexists in two phases. The binodal curve was drawn by visual exploration of the experimental data.
74 Although the solubility parameters of PVA and ethanol are more similar than those of PVA and water, the system PVA-water-ethanol exhibits a large region of immiscibility. The insolubility of PVA in ethanol may be justified because the most significant contribution towards the solubility of PVA in water is hydrogen bonding. The solubility parameters for hydrogen bonding interactions (δH) for PVA and water only differ in 0.5, whereas those for PVA and ethanol differ in 4.4. The same observation applies for the dipole-permanent dipole solubility parameter (δD). The most significant difference between the ternary phase diagrams of PVA-water-ethanol and PVA-water- acetone is observed on the dilute polymer region. At low concentrations of PVA (i.e.,
2.5%), the system existed as a single phase with up to 50% of ethanol, whereas at the same concentration of PVA the system encountered the miscibility region at 25% acetone. By increasing the concentration of the polymer, both systems encountered the miscibility gap region at significantly lower concentrations of non-solvent.
The formulation of the pristine solution for the fabrication of polymer-silica hybrid membranes was adjusted based on the information presented in these diagrams.
Since the kinetics is the controlling factor in the precipitation of PVA, and consequently in the morphology of the precipitated membrane, the composition of the pristine solution can be modified with small amounts of non-solvent to control the kinetics and generate a more porous material. Ethanol was selected due to its non-reactive nature with PEI. The existence of ethanol in the pristine solution should accelerate the precipitation process because it moves the initial point (A) of the precipitation trajectory AC in Figure 2.8 closer to the binodal curve.
75 2 Characterization of polymer-silica hybrid membranes prepared by combination of techniques
Three sets of membranes were prepared for evaluating the effect of the preparation method on the porosity, morphology, and stability of polymer-silica hybrids.
The pristine solutions containing PEI, PVA, sodium silicate, and a template (i.e., P123,
SPAN 80, or TPA-Br) were casted in 200 µm liquid films and (i) dried without further treatment, (i) precipitated in ethanol for phase inversion, or (iii) precipitated in a
GA/ethanol solution for simultaneous phase inversion and cross-linking.
Figure 4.7 shows the IR spectra and SEM pictures of the membranes prepared directly from the pristine solution and dried without further treatment. The characteristic bands of OH stretching vibration around 3328 cm-1 and C-O stretching vibration at 1093 cm-1 correspond to the hydroxyl groups of PVA. The band at 1141 cm-1 corresponds to the crystallinity regions of PVA, which were formed by the slow assembling of PVA chains upon evaporation of water and ethanol. The formation of crystalline regions in the
PVA domains is expected because during the drying process the molecules have limited dynamic volume, and the chains migrate to a common region increasing the chance of self-organization.
The IR band at 1571 cm-1 is produced by the deformation vibration of primary and secondary amine groups of PEI, and those at 2942, 2912, 1419 (broad), and 1326 cm-
1 are produced by the C-H vibrations of PVA, PEI, and the template. The amine bands of
PEI between 3300 and 3200 cm-1 are not observable due to overlapping and hydrogen bonding with the hydroxyl groups of silica and PVA. Similarly, the silanol groups of silica are hydrogen bonded with the functional groups of the polymers and the templates.
76 The band at 836 cm-1 was attributed to the siloxane building blocks of the silica domains.
Due to its low concentration, the particular bands of the surfactant are not evident, but there is an observable difference in the absolute absorbance intensity between the spectra.
Since the membranes were casted with the same thickness, it is expected that they have the same linear concentration across the thickness direction. The IR absorbance is proportional to the concentration, and the pathlength of the IR beam. This concept may be used for estimating the relative thickness of the membranes, which is proportional to the density of the membranes, and inversely proportional to their porosity. The sample that produced the highest absorbance intensity, P123, should be the one with the highest concentration across the thickness, therefore, the less porous. Note that in these membranes the template was not removed, and it is expected that the templates are filling the pores of the membranes.
The SEM images show the morphology of the polymer and silica domains in these membranes. In this case, silica precipitated due to the increments in concentration of the pristine solution caused by solvent evaporation, and may contain large fractions of sodium silicate. Instead of undergoing a complete condensation driven by an acid catalyzed reaction, a fraction of the silanol groups precipitated as sodium silicate. The
SEM images show the formation of flower-like structures, confined in discrete segments of the polymer surface. These structures were formed by the slow condensation of the silanol groups, and the crystallization of residual sodium silicate. The slow evaporation of the solvent allowed migration of the silica components to the membrane’s surface, and caused micro-phase separation between the polymer and the silica domains. The surface of the polymer does not present macroporosity, but exhibits a nearly continuous surface
77 with uniformly distributed depressions. These depressions are most likely caused by the surface tension changes on the solution due to the change in concentration of the template. Since the templates used are surfactants, they have a significant influence in the surface properties of the pristine solution.
PVA/PEI/SiO membranes
2 20µm 3µm
cryst
def
2
OH
C-H NH C-H PVA C-O Si-O-Si
836
3328 2942 2912 1571 1419 1326 1141 1093
Template: 20µm 3µm
P-123 TPA-Br
IR Absorbance (a.u.) SPAN-80 20µm 3µm
3500 3000 1600 1200 800 Wavenumbers (cm-1)
Figure 4.7 IR absorbance spectra and SEM pictures of polymer-silica membranes prepared by direct templating. The IR spectra were collected in transmission mode at 25 oC.
Figure 4.8 shows the IR absorbance spectra and SEM pictures of the membranes prepared by direct templating and phase inversion in ethanol. Phase inversion had a significant effect on the morphology of the membranes evidenced by (i) the low absorbance intensity of TPA-Br compared to P123 and SPAN 80, (ii) the existence of macropores at the surface of the membranes, making the polymer surface discontinuous, and (iii) the formation of spherical silica domains. The IR spectra of these membranes shows the same absorption bands as those in Figure 4.7, but different features in intensity and shape. The most significant differences are: (i) the absorbance intensity of TPA-Br is
78 significantly lower than that of P123 and SPAN 80, (ii) the broad C-H band at 1419 cm-1 shifted to 1436 cm-1 and became narrow, and (iii) the intensity of the amine group at
1577 cm-1 is lower than in the case of only direct templating. From the SEM pictures it is possible to observe that the membrane with the highest porosity is TPA-Br, which is in agreement with the low intensity of its IR spectrum (i.e., the IR beam is transmitted through less amount of material). The shift of the C-H band from 1419 to 1436 cm-1, and the reduction of the intensity of the amine band at 1577 cm-1 suggest that immersion in ethanol helped washing away part of the template and removed a portion of PEI from the polymer matrix.
The macropores in these membranes are formed by phase inversion of PVA. The membranes P123 and SPAN 80 have less macropores than TPA-Br because the templates
P123 and SPAN 80 are significantly larger than TPA-Br. The presence of large templates may increase the viscosity of the solutions and produce transport limitations during the phase inversion. These transport limitations reduce the mobility of the polymer chains and the probability of pore formation. The spherical silica domains embedded on the surface of the polymer are caused by the drastic change in the solution’s surface tension upon immersion in ethanol. Not only the precipitation of silica was accelerated by lowering the pH of the solution upon contact with ethanol, but the combination of solvents (i.e., water and ethanol) with the template molecules promoted the formation of nearly uniform spherical domains. For the membranes SPAN 80 and
P123, it is possible to observe the formation of small crystals similar to those in Figure
4.7, because their precipitation process was slower than that of TPA-Br.
79
Phase inversion in ethanol
cryst def
2 15µm 1µm
OH
C-H NH C-H PVA C-O Si-O-Si
846
3328 2942 2912 1577 1436 1326 1141 1093
Template: 15µm 1µm
P-123 TPA-Br IR Absorbance (a.u.) SPAN-80 15µm 1µm 3500 3000 1600 1200 800 Wavenumbers (cm-1)
Figure 4.8 IR absorbance spectra and SEM pictures of polymer-silica membranes prepared by direct templating and phase inversion. The IR spectra were collected in transmission mode at 25 oC.
Figure 4.9 shows the IR absorbance spectra and SEM pictures of the membranes prepared by direct templating, phase inversion, and cross-linking. Significant changes are observed in the IR spectra of these membranes, where the band at 1570 cm-1 disappeared, and two bands at 1718 and 1662 cm-1 were formed. Clearly the C=O band at 1718 cm-1 is generated by unreacted aldehyde groups of GA, and the C=C-N band at
1662 cm-1, characteristic of enamine species, is caused by cross-linking of PEI. A key observation in these spectra is that PVA also cross-linked with GA, even in absence of an acid catalyst, evidenced by the increase of intensity of the C-O and O-C-O bands at 1093 and 1049 cm-1 that correspond to the acetal bridges formed by the reaction of PVA with
GA. In contrast to the majority of the literature reports about acid catalyzed cross-linking of PVA, the PVA in these membranes seemed to have cross-linked in the environment of the pristine solution/precipitation bath that is rather basic. The hypotheses and direct
80 evidence of cross-linking of PVA under neutral or basic conditions are presented in
CHAPTER V.
The SEM pictures show the formation of the spherical silica domains, similar to those in Figure 4.8, and the periodical depressions across the polymer surface similar to those in Figure 4.7. These membranes have lower macroporosity than those precipitated in pure ethanol, suggesting that cross-linking of the polymers started before the phase inversion process finished. Cross-linking increased the viscosity of the solution and limited the mobility of the polymers away the non-solvent regions to create macropores. Phase inversion in ethanol cross-linked with glutaraldehyde
15µm 5µm
cryst
OH C=O
C-H C=C-N C-H PVA C-O O-C-O Si-O-Si
852
3328 2942 2912 1436 1326 1141 1093
1718 1662 1049
Template: 15µm 0.5µm
P-123 TPA-Br
IR Absorbance (a.u.) SPAN-80 15µm 2µm 3500 3000 1600 1200 800 Wavenumbers (cm-1)
Figure 4.9 IR absorbance spectra and SEM pictures of polymer-silica membranes prepared by direct templating, phase inversion, and cross-linking. The IR spectra were collected in transmission mode at 25 oC.
81 PVA/PEI/SiO membranes 2 Non cross-linked 100 2 228 238 233
80 0 Before
60 -2 dW/dT cross-linking Weigth (%) SPAN-80 TPA-Br P-123 40 -4
Cross-linked 100 2 248 239 253
80 0 After
60 -2 dW/dT cross-linking Weigth (%) SPAN-80 TPA-Br P-123 40 -4 100 200 300 100 200 300 100 200 300 o Temperature ( C)
Figure 4.10 Thermo gravimetric analysis of the polymer-silica hybrid membranes. Measured in TGA from 25 oC to 340 oC.
Figure 4.10 shows the TGA analyses of the three membranes before and after cross-linking (Figure 4.8 and Figure 4.9). The derivative of the degradation curves show a minimum at the temperature where the membranes lost weight at a faster rate. Before cross-linking, the membranes started degrading around 220 oC, and had the fastest weight loss at around 230 oC. Cross-linking helped increasing the thermal stability of these membranes, where the temperature of maximum weight loss for SPAN 80 increased from
228 to 248 oC, for P123 increased from 233 to 253 oC, and for TPA-Br remained almost constant.
82 4.4 Conclusions
The influence of the preparation methods in the morphology, porosity, cross- linking, and condensation reactions of polymer-silica hybrids were studied by providing two comprehensive examples of the preparation and characterization of SBA-15, and
PEI/PVA/SiO2 membranes. First, it was demonstrated the feasibility of fabricating a highly ordered mesoporous silica (SBA-15) from self-assembled polymer-silicate micelles in solution. The raw materials used for the fabrication of this material are lower cost and easier to separate than most widely used catalysts and silica precursors. Second, it was shown that the morphology and physical properties of polymer silica hybrids are governed by the preparation methods. This chapter presents the conditions where different micro-structures are obtained in the polymer and silica domains, and shows an analysis of the factors that govern the morphology of polymer-silica membranes. Pristine solutions containing PEI, PVA, sodium silicate, and a template were casted in the form of membranes. Drying the membranes without removal of the template causes uniform depressions along the membrane’s surface and promotes the crystallization of silica and sodium silicate, resulting in continuous polymer matrices with flower-like silica structures. Combining the direct templating and phase inversion techniques resulted in high macroporosity and formation of uniformly sized silica micro-spheres on the membranes’ surface. Phase inversion partially removed the template, and decreased the amount of PEI on the polymer matrix. Simultaneous cross-linking and phase inversion increased the thermal stability of the membranes but reduced their macro porosity due to mass transport limitations caused by the increments in viscosity of the solutions upon formation of a polymer network.
83 CHAPTER V
5INFRARED STUDY OF THE INTERFACIAL CROSS-LINKING REACTIONS OF
PEI/PVA COMPOSITE MEMBRANES
The interfacial cross-linking reactions of poly(vinyl alcohol) (PVA) and polyethyleneimine (PEI) with glutaraldehyde (GA) were studied by FTIR spectroscopy using a focal plane array (FPA) image detector. A 20 µm PVA porous support was prepared by phase inversion technique using acetone as non-solvent. A thin layer of PEI was casted on one side of the porous support forming an asymmetric composite membrane. The composite membrane was allocated into an IR transmission cell and cross-linked with vapor phase GA for 80 min at 70oC in the presence of water vapor. The
FPA image detector allowed the simultaneous collection of spectra as a function of time and position from an area of 1.6 mm2, at a resolution of 40 µm.
The reaction of PVA with GA produces acetal groups (-O-C-O-) that are IR observable between 1100 and 1030 cm-1. The reaction of PEI with GA produces imine (-
C=N-) and enamine (-C=C-N-) groups that are IR observable around 1665 and 1580 cm-
1. Although the reaction of PVA with GA has been extensively reported to be acid- catalyzed, it was demonstrated that PVA cross-links with GA in the presence of small
84 amounts of PEI, which by nature is a base. The results presented in this chapter were used as evidence for testing the feasibility of the hypothesis 3 of this dissertation.
5.1 Introduction
Porous poly(vinylalcohol) (PVA) membranes are used as a support for permeable films in gas phase separations, and as a filtering media in ultra-filtration processes [213].
Functionalizing PVA membranes with PEI is a well know practice to increase their affinity towards specific components of a mixture and to carry out adsorption and catalytic processes [158, 214]. The mechanical properties and perm-selectivity of
PVA/PEI composite membranes can be improved by cross-linking with aldehydes, epoxies, diacids, and other molecules that provide a link between the polymer chains
[215].
The interfacial cross-linking reactions between PVA and PEI immobilize the active layer and improve the mechanical properties of the membrane [216].
Understanding these reactions is crucial to control the availability of functional groups on the active layer [156], and to tailor mechanical properties of the membrane. One of the most common cross-linking agents is glutaraldehyde (GA), a di-functional aldehyde molecule that reacts with amines and alcohols by nucleophilic addition reactions [108,
141]. The reaction of alcohols with aldehydes is catalyzed by acids, therefore the reaction of PVA with glutaraldehyde is normally carried out in acid environments. The most common approach for cross-linking PVA is by preparing aqueous solutions and adding GA and an acid like HCl or H2SO4 [217, 218]. Few studies have reported the cross-linking of PVA with gas phase GA [158, 159], or under base-catalyzed conditions
[198], but none of these have investigated the cross-linking reaction of PVA with gas- 85 phase GA in the presence of PEI and in absence of an acid catalyst. This chapter presents the characterization of PEI/PVA membranes cross-linked with gas phase GA in absence of a catalyst. The hypothesis is that PEI catalyzes the cross-linking reactions of PVA and
GA in presence of water vapor. The basis of this hypothesis relies on our observations of evidence of cross-linking in PEI/PVA systems, including membranes, pellets, and solutions, that do not contain an acid catalyst. The studies presented in this chapter were achieved by using FTIR spectroscopy involving a focal plane array (FPA) image detector.
5.2 Experimental
A PEI/PVA composite membrane was prepared by dissolving 5 g of poly(vinyl
o alcohol) (PVA, Elvanol®71-30, DuPont) in 45 ml of de-ionized water (H2O) at 100 C, cooling, and casting on a Mylar® sheet using a 200 µm multifilm applicator (Sheen instruments). Immediately after casting, the Mylar® sheet with the PVA solution was immersed in acetone (Sigma-Aldrich) for 5 min and dried at room temperature for 15 h.
Polyethyleneimine (PEI, branched, MW=600, Sigma-Aldrich) was dissolved in ethanol
(200 proof, Decon labs.) to a concentration of 10wt%, and casted on one half of the exposed surface of the PVA support creating a gradient interface between these two polymers. After drying at room temperature for 5h, the resulting membrane is composed of a pure PVA region and a PEI/PVA region, where the interface between the two regions is characterized by a gradient of less than 1.3 mm between the peak concentrations of each region.
The cross-linking reactions on the PEI/PVA membrane were studied by FTIR spectroscopy using a Focal Plane Array (FPA) image detector (Digilab Stingray 86 FastImage IR). The membrane, which is self supported after peeling off the Mylar® sheet, was placed in a transmission cell ensuring that the PVA-PEI/PVA interface is observable through the two CaF2 windows. The transmission cell configuration, shown
Figure 5.1, provides spacing of 2 mm between each window and the membrane, and allows the inlet gases to flow throughout the cell filling its entire volume. The cell was adapted with a k-type thermocouple that measures the temperature of the membrane, and was covered with a heating tape connected to a temperature controller. The upstream lines consist of a gas flow controller, a four-port valve, a water saturator, and a glutaraldehyde saturator (GA, 25% in water, Fisher). The water and GA saturators were maintained at room temperature, and the carrier gas was selected as Ar flowing at a rate of 75 cm3/min.
The FPA consists of an array of 32 x 32 MCT detectors with an approximate resolution of 40 µm per pixel, for a total area of about 1.6 mm2. The IR spectra of each pixel of an area around the PVA-PEI/PVA interface were collected every 10 min during the cross-linking reactions by co-adding 16 scans at a resolution of 4cm-1. For the cross- linking experiments the temperature of the cell was maintained at 70 oC. First, the membrane was saturated with water vapor by flowing 75 ml/min of Ar through the water saturator and into the transmission cell for at least 20 min. Then, the Ar flow was switched to pass through the GA saturator to introduce gas-phase GA into the cell. The cross-linking reaction was carried out for 80 min, and the cell was purged with Ar for 10 min before cooling the transmission cell to prevent the condensation of GA or water.
87 (a) (b)
IR absorbance at 1573 cm-1 (N-H) PEI/PVA Interface 1.2 PVA 2.5 µm (a.u)
0.0 250 µm 1.5 cm 15 µm
Ar IR beam To detector
IR transmission cell Water GA saturator saturator
(c)
Figure 5.1 (a) 3D mapping of the PVA/PEI-PVA interface generated by the IR absorbance intensity of the amine band at 1573 cm-1. (b) Picture of the IR transmission cell and SEM picture of the PVA support. (c) Schematics of the system used for evaluating the interfacial cross-linking reactions of PEI/PVA membranes. The color scale of the 3D mapping was selected arbitrarily to match the range of intensities of the IR band at 1573 cm-1.
88 5.3 Results and Discussion
5.3.1 Membrane characterization
Figure 5.1 (a) shows the 3D mapping of the PEI/PVA membrane, generated by the IR absorbance intensity of the amine band at 1573 cm-1 (N-H deformation). The color scale was selected arbitrarily to match the peak concentration of amine groups in the PVA and PEI/PVA regions. This means, the maximum intensity of the 1573 cm-1 band in the observed area was 1.2 (a.u.). The FPA window was set to analyze the center of the transmission cell, with a scanning an area of approximately 1.6 mm2 and a resolution of 40 µm. The 3D mapping shown in Figure 5.1 is a smooth surface generated by the FPA software. In this figure, the side with maximum intensity is that corresponding to the PEI/PVA region (shown in red), and the side with the lowest intensity is that corresponding to the PVA region (shown in blue). The gradient concentration of PEI is observed by the increase in intensity from the PVA region to the
PEI/PVA region, evidenced by the change in color and height of the mapping regions.
Figure 5.1 (b) shows the SEM picture of the PVA support, which has the characteristic morphology of a phase inversion membrane.
Figure 5.2 shows the IR single beam spectra of three regions of the PEI/PVA membrane, with an inset detailing the region between 1800 and 1400 cm-1, where the bands due to cross-linking of PEI are evident. These three regions were selected at equidistant points between the limits of the FPA window as labeled in the 3D map shown in Figure 5.1 (a); the PVA region (blue), the interface (green), and the PEI/PVA region
(red). Before cross-linking it is possible to observe an intense amine band at 1573 cm-1, and after cross-linking it is possible to observe the bands for imine (C=N) at 1597 cm-1,
89 and enamine (C=C-N) at 1639 cm-1. Other characteristic IR peaks of the PEI/PVA membranes are observed in these spectra including the broad OH stretching vibration of
PVA at 3356 cm-1, and the intense C-H absorption between 300 and 2800 cm-1. To observe the details of the PVA region, which is the focus of this study, the IR absorbance
spectra of the low frequency region is shown in Figure 5.3. 1597
1.0 0.9 1639
3356 1639 1597
After 0.6 cross-linking 0.5 PVA region 0.3 Interface PEI/PVA region 0.0 1800 1600 1400 0.0
1.0 0.9 1573
3356 1573
Before 0.6 cross-linking 0.5 IR Single beam (a.u.) 0.3 PVA region Interface 0.0 PEI/PVA region 1800 1600 1400 0.0 4000 3500 3000 2500 2000 1500 1000 -1 Wavenumbers (cm )
Figure 5.2 IR Single beam spectra of three regions of the PEI/PVA membrane with different concentrations of PEI, before and after cross-linking.
Figure 5.3 shows the IR absorbance spectra of the PEI/PVA membrane before and after cross-linking. In this figure it is evident that the intensity of the acetal groups at
1095 cm-1 is higher at the interface than at the PVA and PEI/PVA regions. The increase of intensity at 1095 cm-1 and the broadening of the crystallinity band at 1141 cm-1, are
90 caused by the cross-linking reaction of PVA with GA. Other spectral characteristic pertinent to this reaction is the shift of the C-H band at 1307 cm-1 to higher frequencies, which is due to the incorporation of GA into the PEI/PVA membrane [141, 192]. Before cross-linking, the interface shows an intermediate absorption intensity at most frequencies, with exception of the C-O functional groups of PVA at 1091cm-1. The unusual high intensity of this band at the interface suggests that low concentrations of
PEI increase the intra molecular hydrogen bonding interactions of PVA. The increase of
this band has been observed by the presence of water in PVA hydrogels [56].
1639 1597 1446 1431 1327 1141 1095 2.0 1342 PVA region After Interface 1.5 cross-linking PEI/PVA region 1.0
0.5
0.0
2.0
1570 1465 1431 1323 1141 1091 1307 PVA region 1.5 Interface PEI/PVA region
IR absorbance (a.u) 1.0
0.5 Before 0.0 cross-linking 1600 1400 1200 1000 -1 Wavenumbers (cm )
Figure 5.3: IR absorbance spectra of three regions of the PEI/PVA membrane before and after cross-linking. Abs=log (1/I) where I is the single beam of the spectrum of interest.
91 Analysis of the kinetics of the interfacial cross-linking reactions of PVA in presence of PEI reveals the effect of the concentration of PEI. These results could help elucidate the formation of intermediates to postulate the reaction mechanisms of PEI- catalyzed PVA cross-linking reactions. The cross-linking of PEI and PVA with gas phase GA is slow [158], and was observed by selecting time differentials of 10 min.
Figure 5.4 shows the kinetics of the cross-linking reactions for the three regions of the
PEI/PVA membrane; the PVA region, the interface, and the PEI/PVA region. For the cross-linking reactions, t=0 was selected as the moment where the flow was switched from Ar/H2O to Ar/GA/H2O, after the membrane was saturated with water vapor. The selection of this point after saturation with water is important to eliminate the contributions of water adsorption and swelling to the spectral changes of the PEI/PVA membrane.
An important observation of the spectra shown in Figure 5.4 is that the formation of enamine (C=C-N at 1639 cm-1) and the formation of imine (C=N at 1600 cm-1) do not follow the same rate. In fact, the formation of enamine is fast at the beginning of the reaction (i.e., the first 30 min) for both the PEI/PVA region and the interface, but it is slow at longer reaction times (after 40 min) only at the interface. When the rate of formation of enamine at the interface decreases, the formation of imine becomes faster, and the formation of acetal groups at 1099 cm-1 starts becoming evident. The PVA region only contains traces of amine, and the cross-linking reactions that lead to the formation of enamines and imines are slow. At long reaction times (i.e., after 40 min), the formation of imine is predominant in all three regions. The formation of acetal groups in PVA is only observed at the interface, and will be discussed in detail.
92
80
1141 1107 1103 1072
1639 1600 1141 1103 0.2 PVA region 1072 0.2 0.1 0.1
0.0 0.0 0
Time (min) Time
-0.1 -0.1 80
1099 1095
0.2 1141
1639 1593 1346 1323 1141 1095 0.2 1060 Interface 0.1 0.1 0.0 0.0 0
-0.1 -0.1 80
IR absorbance (a.u.)
1149 1141 1122
0.2 1099
1639 1600 1454 1365 1342 1203 1141 1122
0.2 1099 1570 PEI/PVA region 0.1 0.1 0.0 0.0 0 -0.1 -0.1 1650 1500 1350 1200 1050 1140 1110 1080 -1 Wavenumbers (cm )
Figure 5.4 Kinetics of interfacial cross-linking reactions of a PEI/PVA membrane. Collected in transmission cell under constant flow of gas phase GA, using a focal plane detector. Absorbance =log(I/Io), where I is the single beam spectrum of the membrane at time t, and Io is the single beam spectrum of the membrane at time t=0.
The inset of Figure 5.4 is probably the most interesting part of this kinetics. This inset shows the detail of the low frequency region of the IR spectra during the cross- linking reaction of PEI/PVA membranes with GA. It is clear that at the interface, where the concentration of PEI adopts an intermediate value between the PEI/PVA region and the PVA region, something extra-ordinary occurs. The cross-linking reaction of PVA with GA is accelerated, producing a sharp peak at 1099 cm-1. The formation of this peak is also observed in the PVA region with a much slower rate, and is not evident at the
PEI/PVA region where the concentration of PEI is high. The formation of acetal groups
93 in the PEI/PVA region is not evident because there is a large amount of PEI covering the surface of PVA. The GA molecules that approach the PEI/PVA region react fast with the amine groups of PEI, and have little chances to interact with PVA. The large availability of PEI in this region and the fast kinetics of its cross-linking reactions are evidenced by the high rate of formation of enamines and imines throughout the entire reaction.
An explanation for the fast kinetics of PVA cross-linking in the regions with low concentration of PEI (i.e., at the interface) requires analyzing the formation of intermediates. The rate of formation of enamines at the interface shows fast kinetics at initial times and slow kinetics after long times. When the rate of formation of enamines decreases, the rates of formation of imines and acetals increase. The first hypothesis is that the reaction mechanism of the cross-linking of PEI produces an intermediate that catalyzes the reaction of PVA with GA. The mechanism of the nucleophilic addition of
PEI to GA involves the nucleophilic attack on the ketone by the lone pair of the amine group to form a carbinolamine intermediate. This intermediate is a Lewis base that utilizes a proton from water to protonate its hydroxyl’s oxygen and then dehydrates. The next intermediate is an iminium ion, which necessarily loses a proton to water to finally form an imine or enamine product. Figure 5.5 illustrates the reaction mechanism of aldehydes with primary amines. In this case, the reaction of PEI is constantly producing
+ + as intermediates (i) the iminium ion (R’ = RH) and (ii) the hydronium ion (H3O ) that could protonate the oxygen in GA to activate the carbonyl group for nucleophilic attack by the hydroxyl’s oxygen of PVA. A similar mechanism follows the nucleophilic addition of secondary amines on aldehydes, where the iminium ion has no hydrogen attached to the nitrogen, and necessarily looses the proton from the second carbon in the
94 radical R. The cross-linking reactions of PVA at the interface are almost exclusively observed when the secondary amine groups stop reacting, leading to the hypothesis that when the amine groups have been partially consumed, their density decreases, and the proton from the iminium ion has a higher change to protonate the oxygen in GA for reaction with PVA.
Figure 5.5 Mechanism of imine formation by reaction of an aldehyde with a primary amine. Adapted from [108].
The second hypothesis is that in the presence of water the amine functional groups of PEI attract a proton to form a polar bond, leading to the conjugated acid
+ + (RNH2 /RNH3 ). In this case, the amines of PEI must have reduced their basicity significantly to be able to form an acid strong enough to protonate the oxygen in GA and activate the carbonyl for the nucleophilic attack of PVA to GA. The basicity of an amine can be measured by defining an analogous basicity constant (Kb) [219], for the reaction:
95
R R
R b= p b= log b R
The smaller the basicity constant and the larger the value of pKb; the stronger the conjugated acid. The primary amines of PEI have a pKb of 9-10, its secondary amines have a pKb of 7-9, and the imine and enamine groups of cross-linked PEI have a pKb of
4-6 [220], suggesting that in presence of water, the imine and enamine groups could conjugate to produce an acid capable to catalyze the cross-linking reaction of PVA with
GA. Under these circumstances, the formation of acetals is only evident after enough imine and enamine have formed close to its surface. These postulates can only be proved with stronger spectroscopic evidence; which is suggested for future work. The observations and postulates proposed in this chapter are based on spectroscopic evidence that small amounts of PEI catalyze the cross-linking reaction of PVA.
5.4 Conclusions
A spectroscopic investigation of PEI/PVA interfaces was performed using a focal plane array (FPA) detector. This instrument permitted the observation of the cross- linking reactions of PVA with GA under various concentrations of PEI. The results show that small amounts of PEI catalyze the cross-linking reactions of PVA, leading to a first hypothesis that the intermediates formed by the nucleophilic addition reactions of primary amines of PEI to the aldehyde groups of GA, are protonated species capable to catalyze the cross-linking reactions of PVA with GA. A second hypothesis is that the amine groups of PEI dissolved in water are conjugated acids capable to protonate the carbonyl oxygen of GA to activate the nucleophilic attack of the hydroxyls in PVA to the 96 carbonyl in GA. Further investigation is suggested to confirm these hypotheses, based on the spectroscopic observation that the cross-linking reactions of PVA proceed in presence of PEI.
97 CHAPTER VI
6FORMULATION AND PREPARATION OF CO2 CAPTURE PELLETS
CO2 capture units for large scale applications are intended to operate in fixed bed, fluidized bed, or moving bed reactors. The size and shape of the sorbent play an important role to define the operation conditions of these equipments. An amine- functionalized silica sorbent was pelletized using a cross-linked PEI/PVA binder solution. The pellets were prepared in the shape of rods or spheres with average diameters of 1 or 2 mm. The CO2 capture capacity of the pellets was 2.6 mmol CO2/g- sorb, only 7% lower than that of the powder sorbent. After 48 h of a fluidization attrition test, the spherical pellets did not lose any appreciable weight, and the rod pellets presented attrition of less than 1%. The design of the PEI/PVA binder solution, the fabrication methods of the pellets, and their characterization are reported in this chapter.
6.1 Introduction
Several CO2 capture sorbents have been developed in response of the environmental regulations that search to reduce the impact of greenhouse gases to the climate change. However, the commercialization of these materials for CO2 capture technologies has not yet been achieved [25, 221]. The pelletization of sorbents for large scale processes still needs significant research and development efforts [50]. The
98 fabrication of sorbents into pellets is challenging due to the diffusion limitations that CO2 may encounter to interact with the amine sites at the interior of the pellets. Furthermore, the mechanical properties of the pellets must be outstanding to comply with the industrial operation of fixed bed and fluidized bed units.
Through the fundamental research presented in CHAPTERS IV and V, an amine- functionalized silica sorbent was prepared and pelletized using a cross-linked polymer binder solution. These pellets were characterized towards adsorption of CO2, and resistance towards attrition under fluidization and milling. In situ CO2 capture studies were used to obtain the CO2 adsorption isotherms and to elucidate the nature of adsorbed species. This chapter presents the fabrication techniques and characterization of amine functionalized solid sorbents.
6.2 Experimental
6.2.1 Sorbent and Pellet Preparation
An amine functionalized sorbent was prepared by (i) dissolving 22 g of TEPA, 15 g of PEG, and 4.5 g of a cross-linker (CL) in 40 g of ethanol, (ii) dissolving 0.65 g of an inorganic antioxidant (AO) in 40 g of water, (iii) mixing these two solutions, and adding
o 40 g of amorphous precipitated silica (SiO2), and (iv) drying the mixture at 100 C in a convection oven. The names of the cross-linker (CL) and the antioxidant (AO) were not revealed to protect the intellectual property of the funded projects related to this dissertation. A polymer binder solution was prepared by (i) dissolving 1.5 g of PVA in
28.5 g of water at 100oC and adding 0.3 g of GA (25% aqueous), (ii) mixing 3 g of PEI
(50% aqueous) with 1.5 g of a 2w% solution of sodium bicarbonate (NaHCO3), and (iii) mixing these two solutions under vigorous stirring. Pellets were prepared by mixing 99 equal amounts of sorbent and binder solution. This mixture produces a wet paste similar to a clay or a dough [50]. The wet paste was extruded in a manual ram wet extruder with
1 or 2 mm holes die, where the extrudate is obtained in the shape of long rods. The extrudate was either: (i) dried in a convection oven at 130oC and cut into pieces to obtain rod pellets, or (ii) spheronized in an in-house fabricated rotary disk spheronizer to obtain spherical pellets. The spherical pellets were dried in a convection oven at 130oC.
Figure 6.1 (a) shows the schematic of the manual ram wet extruder. The wet extruder consists of a cylindrical aluminum vessel and a mobile piston. The vessel has a die with 1 or 2 mm holes to shape the wet paste into an extrudate. A manual ram system provides the pressure for extrusion. Figure 6.1 (b) shows a schematic of the rotary disk spheronizer consisting of a 19.5 cm aluminum drum equipped with a rotary disc. The clearance between the drum and the disk is less than 200 µm to avoid breaking the pellets. The disc is propelled by a variable-speed electric motor and a funnel is used to feed the extrudate. The spheronization process does not require cutting the extrudate, but requires the addition of powder sorbent or silica to avoid agglomeration.
Figure 6.1 Schematics of (a) wet extruder and (b) spheronizer for the fabrication of CO2 capture pellets. 100 The CO2 capture capacity of the sorbent was measured by a weigh method. The
o sorbent sample, typically 1 g, was heated to 100 C for 7 min to remove pre-adsorbed CO2
3 o and H2O. After this pre-treatment, the sorbent was exposed to 50cm /min of CO2 at 25 C for 10min. The CO2 capture capacity was determined by the weight difference of the sorbent before and after adsorption of CO2. This method gives an initial measurement of the CO2 capture capacity based on the assumption that the sorbent is 100 % selective towards CO2. A degradation study was performed by continuously exposing the sorbent
o to steam at 130 C and measuring the CO2 capture capacity every hour.
6.2.2 Fluidization and Attrition Tests
Figure 6.2 shows the experimental set up for fluidization and attrition tests.
Pellets were fluidized into a ” diameter glass column by flowing air at different velocities. The minimum fluidization velocity was determined experimentally for a ” height bed of pellets. The attrition was tested following a procedure adapted from the standard method ASTM E728-91 R97. A sample of 2 g of pellets was placed in a closed glass cylinder containing 6 steel balls. The dimensions of the cylinder are 1 inch diameter and 2 ¼ inches long. The diameter of each steel ball is ¼ inch, and their individual weight is 1 g. The cylinder loaded with pellets and steel balls was placed in the rotary shaft of a ball mill that provides rotation by the length axis. The rotational speed of the ball mill is 60 rpm. The sample was sieved using a No. 20 mesh (841 µm) to remove the fines produced by attrition. The weight of the remaining pellets was measured every hour until the weight changed less than 0.1 %.
101
Figure 6.2 Experimental set up for fluidization and attrition test.
6.2.3 Construction of the Adsorption Isotherms
The CO2 adsorption isotherms of the amine functionalized sorbent were measured
o o o in a DRIFTS reactor at 35 C, 55 C, and 75 C, by using various concentrations of CO2.
Before the measurements, the sorbent was pretreated by heating to 100 oC in presence of
Ar. Pretreatment is necessary before these measurements to ensure that the sorbent surface is free of pre-adsorbed water and CO2. The adsorption of CO2 was performed by a step switch experiment, where the sorbent was exposed to a fixed concentration of CO2 for 10 min. the spectra of the sorbent were recorded throughout the step switch experiment, but only the data obtained after equilibrium was considered to construct the adsorption isotherms. The residual gas in the DRIFTS reactor was purged by flowing Ar for 10 min, and a subsequent step-switch was performed to measure the concentration of adsorbed species at a different partial pressure of CO2.
102 6.3 Results and Discussion
6.3.1 Sorbent and Pellets Characterization
Table 6.1 shows the properties and average CO2 capture capacity of the pellets prepared from the amine immobilized sorbent. Small pellets are considered those with an average diameter of 1.2 mm, and big pellets are those with an average diameter of 2.5 mm. The pellets had an initial CO2 capture capacity of 2.6 mmol CO2/g-sorb, and had a degradation of 56% after being exposed to steam for18 h. Big pellets had an initial CO2 capture capacity of 2.8 mmol CO2/g-sorb and had a degradation of 58% after 18 hours exposed to steam. There was no significant difference in CO2 capture capacity between rods and spheres pellets, suggesting that for the conditions of these experiments the shape has a negligible effect on the CO2 capture capacity. The degradation of these pellets was attributed to (i) the plug of macropores by the binder solution, (ii) the generation of hot spots into the pellets due to poor heat removal by the gas, and (iii) accumulation of CO2 on the binding sites due to incomplete removal of adsorbed species.
Figure 6.3 shows the IR absorbance spectra of the sorbent after pretreatment, during adsorption of CO2, after purging by flowing Ar, and after a temperature programmed desorption (TPD) at 100 oC. The spectrum of the pretreated sorbent does not show any features because the adsorbed species were removed from its surface, this is the baseline for the spectroscopic analysis. The spectrum of the sorbent during adsorption of CO2 elucidates the CO2 adsorbed species. Carbamates, carbamic acid, and ammonium ions are identified on the surface of the amine-functionalized sorbent. The most prominent band is that at 1560, that represents the most abundant adsorbed species of CO2 on amine sorbents; the carbamate-ammonium ion pair. After Ar purge only the
103 strongly adsorbed species remained on the sorbent and after TPD at 100oC the sorbent was regenerated for a subsequent cycle. A small decrease in the amine band is observed due to thermal degradation of the sorbent during TPD. The band assignment was based on the literature review presented in CHAPTER II.
Table 6.1 Solid properties and CO2 capture capacity of pellets and powder sorbent.
ρ mmolCO /g- mmolCO /g- Dp ρ bulk Void Packing 2 2 Shape particle sorb sorb (mm) (kg/m3) fraction fraction (kg/m3) initial after deg. Rod(small) 1.2 900 565 0.37 0.63 2.6 1.1 Sphere(small) 1.2 1630 613 0.62 0.38 2.6 1.1 Rod(large) 2.5 970 602 0.38 0.62 2.8 1.2 Sphere(large) 2.5 1610 481 0.70 0.30 2.7 1.2 Powder 0.1 1650 412 0.75 0.25 2.8 1.3
104 NH +/NH +
2 3
1679 1633 1560 1494 1409 1326
3405 3012 2173 1.5 2470
After TPD
1679 1560 1409
1494 1326 1.0 Ar purge 1.5 1633
1.0 CO adsorption 0.5 2
0.5 IR absorbance (a.u)
Pretreatment 0.0 0.0 1800 1600 1400 1200 4000 3500 3000 2500 2000 1500 1000
Wavenumber (cm-1)
Figure 6.3 IR absorbance spectra of the amine functionalized sorbent at different stages of the CO2 capture process. Absorbance=log(I/Io), where I is the single beam spectrum of the sorbent at a given stage, and Io is the single beam spectrum of the sorbent after pretreatment, prior to CO2 adsorption. 8
6
4 Cp (kJ/kg.K) Cp
2 25 45 65 85 o Temperature ( C)
Figure 6.4: Heat capacity of the amine-functionalized sorbent as a function of temperature measured by DSC.
105 Figure 6.4 shows the heat capacity of the sorbent as a function of temperature.
The heat capacity was measured in DSC and was calibrated with sapphire for accuracy of the data. The heat capacity of the amine functionalized sorbent is a strong function of the temperature. A polynomial correlation was developed for using the data reported in this figure in the modeling of a CO2 adsorption unit in CHAPTER VII.
Figure 6.5 shows the attrition curve for spherical pellets measured in ball mill.
The pellets lost 2.5 % weight during the first 5 hours of attrition, and reached a plateau after 7 hours, for a total loss of 2.62 %. The pellets collected at the end of this experiment conserved their integrity in shape and size.
1.96 100.0
Resistance to attrition (%) attrition to Resistance 1.95 99.5 1.94 99.0 1.93 98.5 1.92 98.0
Weight (g) 1.91 97.5 1.90 97.0 1.89 0 1 2 3 4 5 6 7 Time (h)
Figure 6.5 Attrition curve for amine functionalized pellets in standard ball mill for particle sizes 841 µm or greater, mesh No. 20. Adapted from ASTM E728-91 R97
The selection for this attrition test method was based on previous experience with different pellets and sorbents. Other methods such as abrasion with air jets (ASTM D-
5757), dusting attrition (ASTM D-5159), radial crush strength (ASTM D-6175) and erosion by particle impingement (ASTM G-76) were considered. For the specific pellets presented in this work, their particle size, and density, the air jets methods are too mild
106 and do not cause attrition to the sample. Fluidization for 48 hours did not show any significant weight loss for the spherical pellets, and showed less than 1 % loss for the rod pellets. Dusting attrition and radial crush strength do not provide reproducible and reliable data because these techniques depend on the shape and size of one single pellet, and the results are not representative of the bulk properties of a sorbent bed.
Table 6.2 shows the flow rate, air velocity and estimated weigh loss during the attrition test on pellets. Big particles require higher velocity and flow rate for fluidization and bed expansion than small particles. Air velocity was held constant between sizes to evaluate the effect of the shape on the strength of the pellet. Spherical pellets presented no evident weight loss after 48h of fluidization. Rods suffered small attrition that represented less than 1% of weight loss. Pellets with rod shape attrite more than spheres because they have sharp edges that are weaker than the plain-like surfaces of the sphere.
Also, rods have larger surface area which increases the probability of collision .
Table 6.2 Fluidization conditions and attrition test for pellets. Results based on fluidization of 5g of pellets.
Air flow rate Air velocity Weight loss Experimental min Shape (L/min) (m/s) after 48 h fluid. velocity (m/s)
Rod(small) 36 1.2 < 50mg 0.56
Sphere(small) 36 1.2 - 0.56
Rod(large) 50 1.6 < 50mg 0.76
Sphere(large) 50 1.6 - 0.76
Figure 6.6 shows the degradation curve under steam at 130 oC for the powder sorbent, and for the small and large rod pellets. The CO2 capture capacity was tested under stagnant conditions; but under fluidization mass and heat transport limitations 107 should be added to the degradation of the pellet particles. Without transport limitations, in stagnant conditions, the capture capacity of the pellets is expected to be the same as that of the powder sorbent. These curves demonstrate that the pelletization process presented in this chapter is suitable for increasing the particle size of silica based sorbents, without affecting the CO2 capture capacity or degradation resistance. Pellets prepared with only cross-linked PVA were included for comparison, their CO2 capture capacity id significantly lower than that of the pellets prepared with the PEI/PVA binder solution.
PVA/PEI pellet Powder sorbent PVA pellet
Figure 6.6 Degradation curve for powder sorbent and rods pellets under steam at 130oC.
Figure 6.7 shows the SEM images and EDX analyzes of the external and internal surfaces of the pellets. From these images, the polymer-silica network looks like and amorphous mass of agglomerates with domains of 3-10 µm. The agglomeration was possible due to the intimate contact of the binder solution with the sorbent during mixing and extruding. The degree of mixing of the “dough” and the particle size of the silica domains are key factors in the attrition resistance of the pellets. Poor mixing causes agglomeration of powder particles that have not been impregnated by the polymer, causing rupture points in the sorbent. A large particle size of the silica domains would have the same effect where the polymer binder does not enter in contact with the internal surfaces of the sorbent, creating powder pockets that are prone to attrition. The EDX 108 analyses show that the distribution of the silica, polymer, and amine sites is almost the same both in the external and internal surface of the pellets, demonstrating a good degree of mixing, and the good distribution of active sites for adsorption of CO2.
(a) External surface
250 µm 50 µm 20 µm
(b) Internal surface
250 µm 20 µm 50 µm
Figure 6.7 SEM images and EDX analysis of (a) the external surface, and (b) the internal surface.
109 6.4 Conclusions
Cross-linked PEI/PVA solutions were used as an effective binder to agglomerate amine-functionalized silica-based powder sorbents into strong pellets. The pellets have an average CO2 capture capacity of 2.6 mmol CO2/g-sorb, only 7% lower than that of the powder sorbent. The pelletization process presented in this chapter involves wet extrusion and spheronization, which allow the preparation of pellets with different shapes and sizes. The CO2 capture capacity is not governed by the shape of the pellets when measured in stagnant conditions, demonstrating the fast kinetics of the pellets despite their large size. Large pellets yield to a faster degradation than small pellets and powder sorbent, due to the accumulation of CO2 adsorbed species on the pellet bulk, and to the generation of hot spots that potentially degrade the amine sites, as demonstrated by the spectroscopic studies of a CO2 capture cycle. Spherical pellets are completely resistant to attrition under fluidization conditions, and lose less than 3% weight when exposed to a ball milling process, which simulates the operation of a loop seal, or gate valve. The properties measured in this chapter, the fabrication techniques, and the performance tests of these sorbents and pellets was used in the model presented CHAPTER VII for evaluating the thermodynamic and kinetic properties of the sorbent in a fixed bed CO2 adsorption unit.
110 CHAPTER VII
7COMPARTMENTAL MODELING OF A FIXED BED CO2 CAPTURE UNIT
A fixed bed CO2 adsorption unit loaded with an amine-functionalized solid sorbent was simulated using a compartmental modeling technique. The model describes the bulk temperature, the concentration of CO2 in the gas phase, and the amount of CO2 adsorbed on the sorbent as a function of time and position during a step switch adsorption of CO2. The non-ideal flow behavior of the adsorption unit was described by a set of compartments that account for the dead volume and dispersion. The adsorption kinetics was assessed by the linear driving force (LDF) model with a non-isothermal mass transfer coefficient. The equilibrium concentration was modeled assuming three adsorption mechanisms: (i) single site adsorption (Langmuir), (iii) dual site adsorption with no-dissociation (ND), and (iv) dual site adsorption with dissociation (DIS). The selection of these mechanisms was based on experimental observation of CO2 adsorbed species on amine-functionalized sorbents. The thermodynamic parameters of the isotherms and the kinetic parameters of the model were fitted by minimization of the error functions against experimental data. The model was validated by in-situ CO2 adsorption experiments performed in a DRIFTS reactor. Sensitivity analyses were performed by changing three of the key variables in a CO2 capture operation; adsorption temperature, flow rate, and partial pressure of CO2. 111 7.1 Introduction
Seeking the reduction of greenhouse gases emissions to protect the environment from drastic climate change, and in an attempt to fulfill upcoming rigorous carbon pollution standards for coal-fired power plants, considerable efforts are being made in screening novel materials with favorable kinetic and thermodynamic properties for reversible adsorption of CO2 [34, 222]. These efforts include preparation of multiple samples and extensive experimentation in lab-scale adsorption units. Experimentation in small scale has multiple limitations for estimating adsorption properties because the measuring devices are usually installed at the boundaries of the sorbent bed or in the lines downstream the adsorption unit. The adsorption loading, temperature profiles, and gas concentration inside the reactor are unknown, and the heat and mass transport properties are estimated using lumped parameters. These parameters may include non-ideal flow behavior, heat effects, and diffusion limitations [223].
Various CO2 capture sorbents with fast kinetics and promising equilibrium constants have been reported in literature; amine-functionalized porous materials are among the most popular.[30, 35, 174, 224-227] These reports present intensive characterization of the adsorption properties in equilibrium state, and CO2 capture measurements under idealized conditions. Characterization of the adsorption properties heavily relies on the estimation of the kinetic and thermodynamic parameters from experimental breakthrough curves. These data are typically obtained by step-switch experiments in fixed bed reactors. Most fixed bed CO2 adsorption units’ simulations involve typical modeling approaches like tanks-in-series (TIS) and dispersion models.
These models present rigorous material and energy balances, and are supported by
112 constituting equations that serve for assessing the adsorption isotherm, the kinetic model, and the equations of state. A major drawback of the existing models is that they (i) neglect the non-ideal gas flow behavior outside the sorbent bed, (ii) use isothermal kinetic and thermodynamic approximations, (iii) do not account for heat effects or diffusion limitations, or (iv) involve complex mathematical operations that require sophisticated software for their solution.
These issues have been individually addressed by various investigations, but a more comprehensive model is still needed. A compartmental model was presented to account for the downstream extra-column volume of a small carbon-filled adsorption unit, where the adsorption column was simulated by a dispersion model, and the combination of the two models predicted the behavior of experimental CO2 breakthrough profiles.[223] The linear driving force (LDF) model has been accepted by the scientific community to describe the kinetics of adsorption processes for its simplicity and analytical consistency.[228] Modifications to this model have been made to account for fractional reaction order, multiple adsorption mechanisms, multi-component adsorption[229, 230], and mass transfer limitations[184, 190, 231]. The kinetics in particles with transport limitations has been modeled by diffusion methods, accounting for the different reaction mechanisms of CO2 on amine-functionalized sorbents [184, 190,
231].
The choice of adsorption isotherm models is inspired by the adsorption mechanism, surface properties, and degree of affinity of the sorbents.[232] The adsorption isotherms can be modeled by thermodynamic analyses, molecular simulations, or by the potential theory[233]. Various models have aroused by modification of 113 common isotherms like Langmuir, Toth, and Freundlich [62, 230, 234], by calculating the molecular interactions between the adsorbate and the adsorbent [222, 235, 236], and by predicting the chemical reactions and intermediates involved in the adsorption process.[190, 222, 237] Non-isothermal adsorption equilibrium models have been developed by incorporating expressions for heat of adsorption into the coefficients of pressure-dependant polynomial series.[233]
The energy released by adsorption of CO2 is reflected by the heat of adsorption.
This value is critical for the design of adsorption units and solid sorbents[238], and plays an important role for determining the minimum energy usage for regeneration of CO2 capture sorbents.[99, 174, 239] The heats of adsorption of CO2 on various amine- functionalized sorbents have been reported in literature. Typical values ranging from -40 to -80 kJ/mol have been commonly accepted for simulation and modeling purposes.[173,
226, 227, 230] The information derived from the thermodynamic considerations of the adsorption process has served for calculating the Gibbs free energy for the adsorption reaction, and elucidating the suitability of amine-functionalized sorbents for CO2 capture.[29] Similar considerations have been reported for predicting the behavior of alternative CO2 capture technologies including fluidized bed adsorbers[51], pressure swing adsorption[173, 240, 241], aqueous amine absorption[242], and membrane systems.[243, 244]
This chapter presents a compartmental modeling technique for simulating the adsorption of CO2. Furthermore, the compartmental model presented in this article contemplates (i) non-ideal flow behavior, (ii) heat effects, and (iii) diffusion limitations,
(iv) in a platform that involves simple mathematical calculations. This model, unique of 114 its kind, consists of a system of ordinary differential equations (ODEs) supported by several algebraic constituting equations. These equations can be solved simultaneously using simple numerical methods like Euler or Runge-Kutta, and the parameters can be accurately fitted by minimization of the error functions. The simulation was performed in a simple calculation spreadsheet, and the model was validated by the behavior of in- situ CO2 capture experiments performed in a DRIFTS reactor. Sensitivity tests showed the impact of the sorbent properties and process conditions on the temperature and concentration profiles during adsorption of CO2 on amine-functionalized solid sorbents.
The advantages and limitations of this model are discussed.
7.2 Experimental
The amine-functionalized solid sorbent was prepared by mixing 4 g of amorphous precipitated silica (SiO2, Tixosil 68B, Rhodia) with 8ml of an ethanol/water solution (50 w%) containing 2.2 g of tetraethylenepentamine (TEPA, Sigma-Aldrich), 1.5 g of polyethyleneglycol (PEG, MW=200, Sigma-Aldrich), 0.45 g of polymeric linker (PL), and 0.065 g of antioxidant (AO). The mixture was dried in a convection oven at 100 oC for 30 min. The names of the polymeric linker (PL) and antioxidant (AO) were not revealed due to the filling of a patent. The sorbent properties are reported in Table 6.1
Solid properties and CO2 capture capacity of pellets and powder sorbent..
Figure 7.1 (a) shows the schematic representation of the DRIFTS reactor and the sample holder used for the in-situ CO2 capture experiments. The system configuration has been described in detail in previous reports [245], and is shown in the experimental section of this dissertation (Figure 3.3). The upstream connections of the DRIFTS reactor consist of gas flow controller manifolds, a four-port valve, and a six-port valve. 115 The DRIFTS reactor, placed in a Harrick Praying Mantis casing, has a capacity of 0.141 cm3 (1.41x10-7 m3), and allocates approximately 58 mg of sorbent. The reactor casing is attached to a Thermo Nicolet 6700 FTIR bench (IR), and its effluent is connected to a
Pfeiffer QMS 200 quadruple mass spectrometer (MS) (i.e., downstream). The mass flow controllers regulated the flow rates of Ar, CO2, and air, the four-port valve was used to switch the inlet gas between Ar and CO2/air (15 v%), and the six-port valve was used to inject CO2 for calibration. Inside the DRIFTS reactor, the inlet gases filled the dome and were restricted to flow from top to bottom of the sorbent bed. The volume of the dome and DRIFTS reactor is approximately 15 cm3 (1.5x10-5 m3). The temperature of the sorbent bed was monitored by a K-type thermocouple, located at the center of the sample holder, and the reactor temperature was controlled using a heating element. IR spectra were collected throughout the entire experiment, and the reactor effluent was analyzed by
MS (N2: m/e=14 and 28, Ar: m/e=40, and CO2: m/e=44). The adsorption isotherms were measured in-situ in the DRIFTS reactor by adjusting the partial pressure of CO2 flowing though the sorbent bed, and measuring the concentration of adsorbed species after reaching equilibrium. The concentration of adsorbed species is assumed to be proportional to the intensity of the IR band at 1560 cm-1. Details of the step switch experiment and adsorption isotherms measurement are provided in CHAPTER VI.
116
Figure 7.1 (a) Schematic representation of the DRIFTS reactor and sample holder. (b) Compartmental model associated to the step-switch adsorption of CO2 in the DRIFTS reactor.
7.3 Model Description
Figure 7.1 (b) shows the schematic representation of the compartmental model associated with the step-switch adsorption of CO2. The adsorption unit is simulated by five sections consisting of compartments that individually behave as an unsteady-state
CSTR. The mass transfer between each compartment is considered a direct stream in which the gas flows through the void space in the dome and lines, and between the sorbent particles (i,e,. from Vgas,n-1 to Vgas,n). The first section, section 0, consists of a single compartment of volume Vgas,0 that corresponds to the dead volume inside the dome and accounts for the dispersion in the upstream lines. Sections 1, 2, and 3 represent three regions of the sorbent bed, the active zone of the adsorber unit. These sections allow the evaluation of the temperature and concentration in multiple points throughout the sorbent bed. The compartments of sections 1, 2, and 3 are (i) a flow zone defined by the volume of the gas (Vgas,n), and (ii) a stagnant zone defined by the mass of the sorbent (Wsorb,n). 117 The diffusion of CO2 into the pores of the sorbent, and its adsorption on the surface amine sites are considered as a step transferring of CO2 from the flow zone to the stagnant zone (i,e,. from Vgas,n to Wsorb,n) with a transfer rate of rCO2*, the rate of adsorption of CO2.
The concentration of CO2 in the gas phase (CCO2,n) is defined by the amount of
CO2 in each compartment’s volume, and the concentration of CO2 adsorbed on the sorbent (CCO2*,n) is defined by the amount of CO2 adsorbed on each compartment’s sorbent mass (i.e., the amount of sorbent in the compartment). The last section, section 4, consists of a single compartment of volume Vgas,4 that accounts for the dead volume and dispersion in the downstream lines and in the MS sampling port. The initial state of the
DRIFTS reactor (at T=Tin, and filled with Ar) defines the initial conditions (t=0) for all the sections. The inlet gas temperature (Tin), concentration (CCO2in), and flow rate (νin) define the boundary conditions for section 0. The boundary conditions for sections 1 to 4 are transient, and are defined by the temperature, concentration, and flow rate of the gas leaving the previous compartment.
Due to the configuration of the adsorption unit, the experimental data collected by different instruments corresponds to the properties of the gas or the sorbent in different sections (Figure 7.1 and Figure 3.3). The experimental data collected by the IR corresponds to (i) the concentration of CO2 adsorbed on the sorbent in section 1 (CCO2*,1), and (ii) the concentration of gas phase CO2 in section 0 (CCO2,0). CCO2*,1 was measured
-1 from the IR absorption intensity of the CO2 adsorbed species at 1560 cm , and CCO2,0
-1 was measured from the IR absorption intensity of gas phase CO2 at 2360 cm . This approximation is reasonable because the IR signal is enriched by the CO2 contained in 118 Vgas,0, and by the solid surface of the sorbent bed. The contribution of CO2 in Vgas,1 is sufficiently small to be ignored (compared to that of CO2 in Vgas,0), and the penetration depth of the IR beam into the sorbent bed is expected to be less than the bed height of section 1. The temperature recorded by the thermocouple corresponds to section 2, and the concentration measured by the MS corresponds to that of the reaction effluents, section 4.
The assumptions made for the development of this model are:
1. Each compartment is an ideal unsteady-state CSTR with maximum mixedness
at all times. This implies that the temperature and concentration profiles
throughout the adsorption unit are discrete, limited by the number of sections,
and negligible in the radial direction.
2. The volume of Ar filling the reactor prior to the step-switch adsorption is
rapidly evacuated, and contributes to the heat effects in the same proportion as
air. This is justified by the relatively close heat capacities of the two gases;
CPAir is 29.2 J/mol.K and CpAr is 20.7 J/mol.K.
3. The heat capacities of air and CO2 are constant within the temperature range
of these experiments. Only the heat capacity of the sorbent was a function of
temperature.
4. The gas and solid phases of each section are in thermal equilibrium at all
times.
5. The adsorption of Ar and air are negligible, only adsorption of CO2 was
considered.
119 6. The amount of CO2 adsorbed on the sorbent (CCO2*) is proportional to the
intensity of the IR band at 1560 cm-1 regardless of the adsorption mechanism
proposed for the isotherm models.
7. The mass transfer coefficient of the kinetic model is only function of
temperature, and the adsorption equilibriums are isothermal.
8. The pressure drop throughout the adsorption unit is negligible.
7.4 Materials and Energy Balances
The materials balance, equations [1] to [3], consists of three differential equations
3 that describe CCO2, CAir, (in mol/m ) and CCO2* (in mol/kg-sorb) as a function of time for each section of the adsorption unit.
dCCO2,n 1 vn1 *CCO2,n1 vn *CCO2,n Wsorb,n *rCO2*,n [1] dt Vgas,n
dCAir,n 1 vn1 *CAir ,n1 vn *CAir ,n [2] dt Vgas,n
dCCO2*,n rCO2*,n dt [3]
Where n is each section’s number, and ν is the volumetric flow rate of the gas (in
3 3 m /s), equation [4]. Vgas,n (in m ) is obtained by substracting the volume of the sorbent from the total volume of the sample holder (VT), and dividing by the number of sections
(N), equation [5]:
120 RT v W *r * n [4] n n1 sorb,n CO2*,n P
V *W V T sorb sorb [5] gas,n N
R is the ideal gas constant (8.314 J/mol.K), P is the total pressure of the system
3 ( 0 5 Pa), ρsorb (in kg/m ) is the sorbent’s particle density, and Wsorb (in kg) is the total mass of sorbent loaded in the sample holder.
The energy balance, equation [6], consists of a differential equation that describes the temperature Tn (in K) as a function of time for each section. dT UA T T v C Cp C Cp T T (H )(r *W ) n n a n n1 CO2,n1 CO2 Air,n1 Air n n1 Ads CO2*,n sorb,n [6] dt Vgas,n CCO2,nCpCO2 CAir,nCpAir,n Wsorb,n *Cpsorb1 0.044*CCO2*,n
This equation accounts for the heat exchange with the surroundings (i,e,. the
DRIFTS reactor walls) having temperature Ta (in K), global heat transfer coefficient U
2 2 (in J/s.K.m ), and contact area An (in m ). The heat released by adsorption of CO2 is quantified by the heat of adsorption Δ Ads (in J/mol). The internal energy changes are quantified by the amount and heat capacity of the participating species; CpCO2 (in
J/mol.K) is the heat capacity of CO2, CpAir (in J/mol.K) is the heat capacity of air, and
Cpsorb (J/kg.K) is the heat capacity of the sorbent (Figure 6.4). The factor 0.044 is the molecular weight of CO2 (in kg/mol), which multiplied by CCO2*, n accounts for the weight increments of the sorbent.
121 Initial conditions for all compartments, at t=0:
Tn=T(t=0), CCO2,n=0, CCO2*,n=0, CAir,n=P/RT
Boundary conditions; at reactor inlet (section 0):
Tin=Tin CCO2,in=0.15*P/RTin, CCO2*,in=0, CAir,in=0.85*P/RT0
7.5 Rate law and adsorption equilibrium
The adsorption kinetics was assessed by the linear driving force (LDF) model, equation [7].
r K C C CO2*,n L,n CO2*eq,n CO2*,n [7]
The driving force is generated by the difference between CCO2*eq and CCO2*.
Where CCO2*eq, is the amount of adsorbed CO2 that would be in equilibrium with the momentaneous partial pressure of CO2, and CCO2* is the actual concentration of CO2 adsorbed on the sorbent at that moment. CCO2*eq is obtained from the adsorption isotherms, equations [9] to [11]. KL (in 1/s) is the effective mass transfer coefficient, a lumped parameter that accounts for the internal and external mass transfer coefficients.
In this work KL is assumed to be only a function of the temperature, described by the
Arhenius equation with parameters KL0 (in 1/s) and E (activation energy in J/mol), equation [8].
E K K exp L,n L0 RT n [8]
122 The adsorption isotherms were derived from the elementary reactions of three possible adsorption mechanisms of CO2 on amine-functionalized solid sorbents. The first mechanism describes the simplest case where molecular CO2 adsorbs on a single amine site. This mechanism yields to a Langmuir isotherm, equation [9]. Molecular adsorption of CO2 has been observed at high partial pressures of CO2 (PCO2>10KPa), and has been attributed to physisorption on isolated sites.[99, 188, 230] The second mechanism describes the adsorption of CO2 on two amine sites, where the adsorbed species stay as a molecule. This mechanism yields to a dual site adsorption isotherm with no-dissociation of the adsorbed species (ND), equation [10]. Dual site adsorption with no-dissociation has been observed on densely loaded amine sorbents, and the adsorbed species have been attributed to carbamate-ammonium ion pairs and zwitterions.[35, 98, 191, 245, 246] The third mechanism describes the adsorption of CO2 on two amine sites, where the carbamate-ammonium ion pairs and zwitterions dissociate generating two individual adsorbed species. This mechanism yields to a dual site adsorption isotherm with dissociation of the adsorbed species (DIS), equation [11]. Dual site adsorption with dissociation has been observed on sorbents containing mixed primary and secondary amine groups[35, 187, 245], and under humid conditions.[190, 230] The adsorbed species has been attributed to carbamate ions, ammonium ions, carbamic acid, and zwitterions.[187, 188, 195] The elemental reactions and proposed mechanisms are:
1. Single site adsorption, Langmuir isotherm (R2NH+CO2=CO2*):
CT KC PCO2 CCO2*eq [9] 1 KC PCO2
123 2. Dual site adsorption with no-dissociation (ND) (2R2NH+CO2=CO2*):
1 1 8K C P 4K C P C C T CO2 C T CO2 CO2*eq 8K P C CO2 [10]
3. Dual site adsorption with dissociation (ND) (DIS) (2R2NH+CO2=2CO2*):
C K P C T C CO2 CO2*eq 1 2 K P C CO2 [11]
PCO2 is the partial pressure of CO2 in the gas phase: PCO2=RT/P*CCO2. The parameters of these adsorption isotherms are the total concentration of active sites (CT, in molN/kg-sorb), and the equilibrium constant (KC). Note that for the dual site adsorption with dissociation (DIS), the stoichiometry of the adsorption changes; one CO2 molecule produces two adsorbed species. The net change of moles must be accounted for in the
1 rate law and in the materials and energy balances (-rCO2= /2 rCO2*).
The heat of adsorption can be calculated by the van’t off’s equation using the equilibrium constants of adsorption isotherms at two adjacent temperatures, equation
[12].
H 1 1 K , K , exp Ads [12] C T2 C T1 R T2 T1
7.6 Residence time distribution (RTD) and dispersion
During the step-switch adsorption of CO2, nitrogen (N2) served as a tracer to evaluate the residence time distribution of the gas in the adsorption unit. Due to its inertness and non-adsorption nature on amine sorbents, the concentration of N2 in the 124 effluent stream (analyzed by MS, section 4) can be recorded as a function of time, and used to calculate the cumulative residence time distribution F(t), the external age
2 distribution E(t), the mean residence time (tm), and the variance σ of the species flowing through the adsorption unit, equations [13] to [16].
C (t) C (0) F(t) N 2 N 2 C () C (0) N 2 N 2 [13]
d E(t) F(t) dt [14]
t tE(t)dt m 0 [15]
2 (t t )2 E(t)dt m 0 [16]
Where CN2(t) is the concentration of N2 as a function of time and CN2(∞) is the equilibrium concentration of N2 after the step switch.
125 7.7 Results
7.7.1 Adsorption isotherms
(a) (b) (c) 2.0 2.0 1 0.8 1.5 2 1.5 2 1.0 0.6
Langmuir Langmuir (mol/kg) (mol/kg) 1.0 1.0 1 35 C 1 ND 0.4 0.5 ND 55 C DIS DIS
0.5 75 C 0.5 55 C 0.2 CO2*eq
CO2*eq 0 0 0.0
C C
0 0.05 0.1 0 0.05 0.1 coverage Surface 0 0.05 0.1 0.0 0.0 0 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 PCO2 PCO2 PCO2
Figure 7.2 (a) Experimental adsorption isotherms of CO2 on the amine-functionalized sorbent at 35, 55 and 75 oC. (b) Adsorption isotherms derived from the single site (Langmuir), dual site with no-dissociation (ND), and dual site with dissociation (DIS) o models, fitted at 55 C. (c) Surface coverage of CO2 on the sorbent predicted by each adsorption isotherm model at 55oC.
Figure 7.2 shows (a) the experimental adsorption isotherms of CO2 on the amine- functionalized sorbent at 35, 55 and 75 oC, (b) the adsorption isotherms derived from the single site (Langmuir), dual site with no-dissociation (ND), and dual site with
o dissociation (DIS) models, fitted at 55 C, and (c) the surface coverage of CO2 on the sorbent predicted by each adsorption isotherm model at 55oC. The insets in these figures show detail of the low partial pressure region, which is of interest in CO2 capture experiments. Table 7.1 shows the parameters and heats of adsorption calculated at 35oC,
o o 55 C, and 75 C. KC and CT were fitted by minimization of the sum of square errors
o (SSE), and Δ Ads was calculated using equation [12]. Δ Ads at 55 C was interpolated by
o o a logarithmic regression of the values of Δ Ads calculated between 35 C and 55 C, and between 55oC and 75oC.
126 The parameter CT represents the total concentration of active sites on the sorbent.
The value of CT for the Langmuir isotherm is expected to be half of that for the dual site adsorption isotherms (ND and DIS) because of the stoichiometry proposed in each mechanism. The theoretical concentration of amine sites of our sorbent is 6.76 molN/kg.
The values of CT obtained from the isotherms, are 1.97 molN/kg for the Langmuir isotherm, and 4.07 molN/kg for ND and DIS isotherms. The calculated values are lower than the theoretical amine loading of the sorbent, indicating that the maximum amine
o efficiency of the sorbent (molCO2/molN) at 55 C is 29% for the Langmuir model, and
30.1% for the ND and DIS models. This observation suggests that some of the amine sites in our sorbent are inaccessible to CO2 at the conditions of these experiments. We hypothesize that the inaccessible amine sites are “buried” under the surface amine layers, which after adsorbing CO2 form a dense network that limits the transport of CO2 into the internal layers[245]. In addition, a fraction of the amine sites may be hindered by other additives in the sorbent, hydrogen-bonded to the silica support[185], or may have been oxidated during the preparation process[103].
The steep slopes of these isotherms at low partial pressures of CO2, evidenced by the high values of the equilibrium constant KC, show the great potential of our amine-
o functionalized sorbent to uptake CO2 from diluted systems, especially below 55 C. KC changes drastically with temperature, a highly desired feature for CO2 capture sorbents used in temperature swing adsorption processes (TSA). The characteristics of the equilibrium constant allow the sorbent to have fast kinetics because it generates high values of CCO2*eq even at low partial pressures of CO2. The high values of CCO2*eq help increase the driving force for adsorption of CO2 (i.e., CCO2*eq-CCO2*) at short times after
127 the step-switch, when CO2 is entering the adsorption unit in a dilute state. The values of
Δ Ads obtained for each mechanism are significantly different. Δ Ads for the Langmuir isotherm is around -36kJ/mol and is nearly constant within the temperature range evaluated in this work. The value of Δ Ads for the ND isotherm is around -51kJ/mol,
41% larger than that for the Langmuir isotherm, and it variation with temperature is less
o o than 13% between 35 C and 75 C. For the DIS isotherm the value of Δ Ads changes drastically with temperature from -28kJ/mol at 35 oC to -69 k/mol at 75 oC. According to the DIS model, the heat of adsorption our sorbent is sensitive to temperature, which is proportional to the availability of amine sites (CT), and to the equilibrium constant KC.
This observation suggests that the sorbent would bind weakly large amounts of CO2 at low temperature, and strongly smaller amounts at high temperature[185]; ideal for isothermal concentration swing adsorption processes.[173, 247]
The low values of SSE indicate the good fit of the adsorption isotherms, and suggest that the three proposed mechanisms are reasonable. According to these values, the mechanism that better describes the adsorption of CO2 on our sorbent is the dual site adsorption with no-dissociation (ND) (lowest SSE). The excellent fit of the isotherms shows that simple isotherm models can predict the amount of CO2 adsorbed on amine- functionalized sorbents.
128 Table 7.1 Parameters of the adsorption isotherms fitted by minimization of the SSE, and heats of adsorption calculated from equation [9].
T (K) 308 328 348
CT (mol/kg) 1.99 1.97 1.69
KC 1967.75 819.85 382.00 Langmuir ΔHAds (J/mol) -36768.31 -36519.04 -36237.41 SSE 2.55E-05 2.99E-06 1.04E-05
CT (mol/kg) 4.07 4.07 3.58
KC 1431.73 453.78 143.03 ND ΔHAds (J/mol) -48253.97 -51318.97 -54782.03 SSE 6.86E-07 9.56E-07 7.81E-05
CT (mol/kg) 4.14 4.08 3.61
KC 1415.96 732.91 169.62 DIS ΔHAds (J/mol) -27655.44 -47273.96 -69440.34 SSE 2.30E-05 2.48E-06 1.77E-04
7.7.2 Non-deal flow behavior and residence time distribution RTD
Figure 7.3 shows the cumulative residence time (F(t)) and external age distribution (E(t)) functions for the tracer (N2) in the step switch experiment. E(t) was also plotted for the two ideal conditions of plug flow (no dispersion) and mixed flow
(complete dispersion) for comparison. From the RTD curves we calculated tm=13.7 s,
2 2 and σ =602.5 s . In absence of dispersion, tm should be equivalent to the space-time, τ, of the adsorption unit (tm=τ=VT/ν0, Vt=ΣVgas,n). The space-time of the adsorption involves the effects of non-ideal flow outside the sorbent bed, and the effects of flow throw the sorbent bed (τbed=Vbed/ν0). The calculated value of τbed is 0.06 s, which is much lower than that of tm, elucidating that the shape of the RTD curve is produced mostly by the contribution of the external dead volume and dispersion inside the lines. All the 129 deviations from ideal flow may be accounted for by using the compartments Vgas,0 and
Vgas,4 connected in series with the sorbent bed. These compartments provide the additional space-time to the adsorption unit necessary to match the experimental RTD curve. Vgas,0 was calculated by minimizing the error function SSE between the
-1 experimental CO2 curve inside the dome (from IR, 2360 cm ) and the CCO2 curve calculated for the compartment in section 0. Vgas,4 was calculated by minimizing the difference between tm and the space-time of the reactor.
1.0
0.8 0.10 D/νL=∞ Mixed flow Plug flow Exp. data 0.6 D/νL=0
0.05
E(t) F(t) 0.4 D/νL=0. 6
0.2 0.00 0 5 10 15 20 25 30 τ=tm=13.1s Time (s) 0.0 0 30 60 90 120 150 180
Time (s)
Figure 7.3 Cumulative residence time distribution F(t) and external age distribution E(t) functions determined experimentally by an N2 tracer step input in the adsorption unit parallel to the adsorption of CO2.
3 The results show that Vgas,0 is 16.6 cm ; since the volume of the dome is approximately 15 cm3, it can be concluded that the dispersion in the upstream lines is
130 3 small, adding only 1.6 cm to this compartment. Similarly, Vgas,4 was calculated as 16.0 cm3; in this compartment the death volume may be neglected because the downstream lines are connected only to the sample port of the MS. The calculated volume of Vgas,4 is attributed primarily to dispersion caused by the capillary flow inside the MS sampling port. The variance calculated from the RTD curves (σ2) serves to calculate the dispersion module (D/νL), which gives a clear idea of the degree of dispersion in the adsorption unit, equation [17]. D/νL0 represents negligible dispersion (plug flow), and D/νL∞ represents large dispersion (mixed flow). To provide an idea of the orders of magnitude, a reactor with dispersion module of 0.002 is considered to have small dispersion and one with dispersion module of 0.2 is considered to have large dispersion (Levenspiel and
Smith, 1957).
2 2 D D 2 2 8 L L [17]
Where D (in m2/s) is the axial dispersion coefficient and L (in m) is the reactor length. The dispersion module in our adsorption unit is D/νL=0.26, and the axial dispersion coefficient is D=1.4x10-6 m2/s (using L=2.1 m, Figure 3.3). If we assume that
-9 2 the dispersion happens only in the sorbent bed Dbed would be 3.3x10 m /s. The shape of the E(t) curve in Figure 7.3, and the large dispersion module further validate our assumption that the compartments of each section can be modeled as a perfectly mixed unsteady-state CSTRs. For comparison, Table 2.1 shows the axial dispersion coefficient of other tubular reactors reported in literature and calculated using the dispersion model.
The value of D in our model is small due to the large flow rate.
131 Table 7.2 Axial dispersion coefficients for fixed bed adsorbers simulated by the dispersion model.
Adsorption unit D (m2/s) Ref. 15 mm tubular reactor 1x10-5 [35, 234] 61 cm tubular reactor 1.2x10-4 [248] 1.2 m tubular reactor 1x10-3 [241] 12 cm tubular reactor 1.7x10-5 [226] 5 mm DRIFTS reactor 3.3x10-9 This work 2.1 m adsorption unit 1.4x10-6 This work
7.7.3 Compartmental modeling of a fixed bed CO2 adsorption unit
Although it was previously demonstrated that the adsorption unit as a whole operates close to a mixed flow reactor, the sorbent bed experiences temperature and concentration gradients in the axial direction because to its geometry and activity towards adsorption of CO2. Since the gases are restricted to flow from top to bottom of the bed, emulating a tubular reactor operation, the sorbent bed was modeled by three vertical sections. The compartments in sections 1, 2, and 3 account for (i) the dispersion throughout the sorbent bed, which is lower than that of the entire adsorption unit, and (ii) the concentration and temperature variations caused by the transfer of CO2 from the gas phase to the sorbent. The system of differential equations was solved simultaneously using a simple finite-differences numerical method; the Runge-Kutta method of second order (or trapezoidal rule). The adsorption process was simulated for 180s with time differentials of 0.006s, and the parameters U, KL0, and E were fitted by minimizing the sum of the individual error functions (ΣSSE), between the experimental data and the profiles generated for the specific compartments where the data was collected (T2, CCO2,4,
132 and CCO2*,1.as labeled in Figure 7.1 (b)). First we present two parametric studies, where the parameters were fitted assuming (i) isothermal rate law, and (ii) non-isothermal rate law, and then we present three sensitivity analyses evaluating the effect of the adsorption temperature, flow rate, and partial pressure of CO2. For these simulations we used an
Excel calculation spreadsheet in a computer with a CORE i5 processor. Fitting the parameters took between 4and 9 minutes depending on the number of parameters and their initial value, and the simulations for the sensitivity analyses took less than 10 seconds. The initial values were selected based on the literature review, and a visual inspection of the profiles.
133 7.7.4 Parametric Studies: Heat Effects
Parametric study 1: Isothermal rate law
Figure 7.4 Temperature and concentration profiles obtained for the optimal values of the parameters KL0, and U, for the parametric study 1: Isothermal rate law. 134 In this study we consider the adsorption of CO2 with isothermal rate law, where the effective mass transfer coefficient is independent of the temperature (KL=KL0, E=0).
Table 7.3 shows the optimal values of KL0 and U, and Figure 7.4 shows the temperature and concentration profiles obtained after fitting the parameters. The experimental temperature profile shows oscillating behavior after the temperature spike due to the action of the temperature controller. Due to the on-off nature of this controller, the
o heating element is only activated when Tn The experimental data at the reactor outlet (MS) shows that the concentration of CO2 started to rise immediately after the step-switch, suggesting that there was channeling throughout the adsorption unit. Channeling is a characteristic of the non-ideal flow behavior that was not included in the compartmental model, and all the predicted breakthrough curves are overestimated. Note that CCO2 rises very rapidly, and the profile curves are very close to each other. To make the details more evident, the CCO2 profile was plotted only from 0 to 90 s. The existence of channeling, and the concentration gradients in the sorbent bed are caused by the high flow rate of the gases compared to the volume of the sorbent bed. 135 Table 7.3 Optimal parameters and SSE values obtained for the parametric study 1: Isothermal rate law. Parameters Sum of square errors (SSE) Isotherm 2 U (J/s.K.m ) KL0 (1/s) E (J/mol) T2 CCO2,4 CCO2*,1 Langmuir 0.00 0.0638 0a 585.91 49.25 0.13 Isothermal ND 49.61 0.0488 0a 165.23 34.96 0.28 rate law DIS 0.00 0.0917 0a 507.45 25.75 0.16 Adiabatic ND 0a 0.0155 0a 1670.71 10.18 5.91 a: Arbitrarily selected to fulfill the conditions of the parametric study. By visual inspection and supported by the SSE values, the ND isotherm model gives the best fit for the temperature, the DIS model gives the best fit for the concentration of CO2, and the Langmuir model gives the best fit for the concentration of adsorbed CO2. Overall, the model that has the best fit is ND, whit the least error function ΣSSE=200.47. The values of KL0 increase as ND transfer rate of CO2 onto the sorbent is faster if the adsorbed species dissociate, as proposed for the DIS model. Interestingly, the Langmuir and DIS models show that the optimal value of U is 0. J/s.K.m2, which corresponds to an adiabatic operation. The value 2 of U for ND is 49 J/s.K.m , suggesting that the adsorption of CO2 based on the ND adsorption mechanism generates more heat than the adsorption based on the mechanisms of the Langmuir and DIS models. The heat generated by adsorption is an indicative of the amount of CO2 adsorbed by the sorbent, and should be proportional to the transfer rate of CO2. The excess heat generated cannot be removed solely by the inlet gases, and must be removed through the DRIFTS reactor wall. Note that at the conditions of the experiment this phenomenon is feasible because the inlet gases contain CO2, which gets adsorbed on the sorbent and produces heat. It is unlikely that the heat that the carrier gas 136 (i.e., air) is able to remove is exactly equivalent as the heat generated by the adsorption of CO2 minus the heat used to increase the temperature of the sorbent, resulting in a perfectly adiabatic operation. For comparison, we fitted KL0 for the ND model under arbitrary adiabatic conditions, obtaining an optimal value of KL0 68% smaller than that of the original case (i.e., non-adiabatic). The adiabatic analysis for the model ND shows poor agreement with the experimental data, shown at the bottom of Figure 7.4. In this special case the heat generated by the adsorption reaction accumulates in the reactor, heating the sorbent bed and causing a causing a broad temperature rise. The prediction for the peak temperature has a delay of more than 60 s with respect to the experimental data and a long cooling trail. Parametric study 2: Non-isothermal Rate Law Since the objective of these parametric studies is to obtain the best fit for the experimental data, in this study we test our hypothesis that contemplating the effect of temperature on the effective mass transfer coefficient improves the correlation between the profiles predicted by the model and the experimental data. For this study we fitted all three parameters U, KL0, and E. The expression that we assumed for KL is the Arrhenius equation, were KL changes exponentially with temperature. The diffusivity of CO2 into the amine layers of the sorbent increases with temperature, reducing the mass transport limitations and increasing the rate of transfer of CO2 between the gas phase and the sorbent. It is obvious that the effect of temperature on KL, a kinetic parameter, is hindered by the thermodynamic nature of the adsorption process. In the adsorption 137 isotherms analysis, Figure 7.2 and Table 7.1, we showed that the equilibrium constant of our amine-functionalized sorbent is a strong function of temperature. Small temperature increments reduce the adsorption equilibrium, and consequently reduce the driving force for adsorption of CO2. Table 7.4 shows the optimal parameters and SSE values obtained for the parametric analysis with non-isothermal rate law, and the profiles calculated using these parameters are shown in Figure 7.5. The temperature and concentration profiles are very similar to those of the parametric study 1. The values of KL0 for the Langmuir and DIS models increased in 8%, and the value of KL0 for ND did not change. The overall heat transfer coefficient (U) for the Langmuir model is very close to zero, and could be neglected to conclude that the best fit for this model is the adiabatic operation. The value of U for the DIS model is -11.39 J/s.K.m2, which is not feasible for the experimental conditions of this work. A negative value of U would imply heat transfer from the DRIFTS reactor to the sorbent, which is impossible when Tn>Ta. The activation energy (E) is probably the most significant evidence of the small dependence of KL with temperature. The values of E obtained in this study are about three orders of magnitude smaller than the value of RT (2727 J/mol) in the denominator of the exponential in equation [8]. To have an increase of 10% in KL, the value for E should be at least 262 J/mol. Furthermore, for the ND model in this study, the value of the overall error function is ΣSSE= 00.47, the same as that in the isothermal rate law study. ΣSSE slightly increased for the Langmuir isotherm model, and although it decreased for the DIS model, the values of the parameters U and E are not feasible. From these observations it is possible to conclude that the temperature dependence of KL is either 138 small enough that can be neglected, or hindered by the temperature effects on the thermodynamic properties. Figure 7.5 Temperature and concentration profiles obtained for the optimal values of the parameters KL0, E, and U, for the parametric study 1: Isothermal rate law. 139 Table 7.4 Optimal parameters and SSE values obtained for the parametric study 2: Non- isothermal rate law. Parameters Sum of square errors (SSE) Isotherm 2 U (J/s.K.m ) KL0 (1/s) E (J/mol) T2 CCO2,4 CCO2*,1 Non- Langmuir -0.0055 0.0689 -0.0068 612.23 56.72 0.12 isothermal ND 40.61 0.0488 -1.1391 165.20 34.99 0.28 rate law DIS -11.39 0.0993 -18.437 393.81 27.64 0.19 7.7.5 Sensitivity analyses The optimal parameters obtained for the ND model gave the best fit of the simulation to the experimental data. This model was used to simulate the adsorption unit under variations of three key variables; adsorption temperature, flow rate and partial pressure of CO2. The adsorption temperature is a key variable because it defines the thermodynamic equilibrium of CO2 adsorbed on the sorbent, and consequently the CO2 capture capacity of the sorbent. We demonstrated that mass transfer coefficient of the adsorption kinetics are nearly insensitive to temperature, however the adsorption kinetics could be largely affected by the adsorption temperature due to the thermodynamic contributions to the driving force. Adsorption at low temperature is ideal to favor the adsorption of CO2, but the costs associated with the processes of reducing the temperature of a flue gas are probably not justified by the increments in CO2 capture capacity. Figure 7.6 shows the temperature and concentration profiles predicted for the o CO2 adsorption experiment at 35 and 75 C. The most evident difference between these simulated experiments is the temperature spikes. Although the rate of heat removal of the DRIFTS reactor is large (40.61 J/s.K.m2), the difference in temperature spike of the two experiments is significant. For adsorption at 35 oC the temperature spike is 15 oC, and for adsorption at 75 oC the temperature spike is 9 oC. Note that the temperature spike 140 for the original experiment at 55 oC was 13 oC. As we mentioned before, the heat generated by adsorption is an indicative of the amount of CO2 adsorbed by the sorbent, in agreement with the observations in Figure 7.6, were the concentration of adsorbed CO2 is o lower for the experiment at 75 C. The concentration of gas phase CO2 is also lower at 75 oC than at 35 oC because of the thermal expansion of the gases at higher temperature. Figure 7.6 Temperature and concentration profiles predicted by the ND adsorption isotherm model for adsorption at 35 and 75 oC. Studying the effect of the flow rate on the adsorption experiments is of vital importance to calculate the dimensions of the adsorption units and to estimate the potential outputs of fluctuations in the CO2 capture operation. Figure 7.7 shows the predicted temperature and concentration profiles for CO2 adsorption experiments using 141 flow rates of 75 cm3/min and 225 cm3/min. The heat removal at higher flow rate is evidenced by the narrow temperature spike, compared to that of the experiment at low flow rate. The heat is removed by the change in internal energy due to the flow of the inlet gases at T0; the higher the flow rate, the higher the amount of heat removed by the inert gas. The temperature spike of the two experiments is very similar, because the sorbent has fast kinetics and adsorbs almost the same amount of CO2 from the two experiments. The flow rate changes the saturation time of the sorbent and the heat removal rate, but has little effect on the rate of heat generation. Another significant difference between these experiments is the time and shape of the breakthrough curves. As expected the experiment with lower flow rate has slow breakthrough (around 18 s), and the curve is not as steep as that for the experiment with higher flow rate. A flow rate of 225 cm3/min is excessively high for this adsorption unit and the breakthrough of the experiment is not evident. 142 Figure 7.7 Temperature and concentration profiles predicted by the ND adsorption 3 isotherm model for a CO2 adsorption experiment with flow rates of 75 cm /min, and 225 cm3/min. Evaluating the effect of the concentration of CO2 in the adsorption experiments is of interest in the CO2 capture community because different industrial sectors have emissions with more diluted, or more concentrated CO2 than those used in this experiment. The concentration of CO2 in the flue gas of coal-fired power plants is between 10 and 15 v%, in natural gas-fired power plants is around 5%, and in the blast furnace of steel industries can reach up to 25 v%. Figure 7.8 shows the predicted temperature and concentration profiles for the adsorption experiment using 5v% and 25v%. The characteristics of the concentration profiles in these experiments are similar to those evaluated at different flow rates. The broad dispersion in the CO2 concentration 143 profile is caused by the slow introduction of CO2 into the reactor. Regardless of the flow rate or concentration of CO2, the breakthrough curves are characterized by the total amount entering the reactor, a balance between flow rate, and concentration. The partial pressure determines the amount of CO2 present in the reactor, but this amount has little effect on the temperature spike because the adsorption equilibrium of CO2 on the sorbent is almost the same for any partial pressure above 1 v%, as predicted by the adsorption isotherms in Figure 7.2. Figure 7.8 Temperature and concentration profiles predicted by the ND adsorption isotherm model for a CO2 adsorption experiment using diluted (5 v%) and concentrated (25 v%) CO2. 144 7.8 Conclusions In this chapter was developed a compartmental model for simulating a fixed bed CO2 adsorption unit. The deviations from ideal flow were characterized by a residence time distribution RTD experiment, and modeled by two compartments connected in series with the sorbent bed. The RTD showed that the DRIFTS reactor operates close to a mixed flow reactor, and the dispersion and dead volume can be accounted for by increasing the space-time of the reactor. Three adsorption isotherms models, (i) single site adsorption (Langmuir), (iii) dual site adsorption with no dissociation (ND), and (iv) dual site adsorption with dissociation (DIS), were used to model the adsorption kinetics of CO2 on the sorbent. The adsorption isotherm that better describes the equilibrium concentration of CO2 on the sorbent is the dual site adsorption isotherm with no-dissociation, which is in agreement with the mechanism of the adsorption of CO2 on densely loaded amine sorbents. The kinetics was modeled by the linear driving force (LDF) model with a temperature- dependant mass transfer coefficient. And it was demonstrated that temperature has little effect on the effective mass transfer coefficient of the kinetics model. The optimal model parameters were used for evaluating the impact of key variables like adsorption temperature, flow rate, and concentration of CO2. 145 CHAPTER VIII 8SILICA-SUPPORTED AMINE CATALYSTS FOR CARBON-CARBON ADDITION REACTIONS Basic catalysts for carbon-carbon addition reactions were synthesized by immobilization of amine species on silica supports. Tetraethylenepentamine (TEPA) was impregnated and immobilized onto amorphous silica (SiO2) and SBA-15 using an epoxy resin. The basicity of the catalysts was determined by adsorption-desorption of CO2 and the degree of immobilization was evaluated by FTIR. The catalytic activity towards the Claisen condensation reaction of methyl benzoate (MB) and methyl ethyl ketone (MEK) was evaluated by an in-situ FTIR micro-scale reactor. A mechanism is proposed to show that the catalysts promote the formation of β-diketone and methanol; the effects of the support and amine immobilization degree are discussed. 8.1 Introduction Functionalized silica supports are beneficial for applications that require substrates with high surface area such as adsorption, absorption and catalysis. SBA-15 has been prepared by various methods [97, 164, 249, 250]; the porosity and structure have been modified to improve the sorption and catalytic properties [101, 164, 251]. Surface functionalization has been achieved by both co-condensation [5, 120, 252] and 146 impregnation methods [97, 250, 253, 254]. Amine functionalized mesoporous materials have been used for applications in CO2 capture [97, 101, 189, 255], removal of ions in water waste treatment [253, 254, 256], purification of bio-oil and pyrolysis by-products [257-259], cell growth on selective substrates [260], and heterogeneous catalysis [5, 261- 265]. The catalytic activity of amine functionalized substrates is obtained from the basicity of amine functional groups. The basicity of an amine depends on the number of substituents (primary, secondary and tertiary amines), the character of neighboring functional groups (electron donor and/or electron withdrawing groups), and the molecular weight [101, 266]. In general, basic catalysts containing amine functional groups have been demonstrated to be useful for organic chemical synthesis via Aldol condensation [261, 264, 267], Michael addition [267-269], Knovenagel reaction [5, 251, 263, 265] and Claisen condensation reaction [5, 251]. These kinds of catalysts have also been used in catalytic esterification and transesterification reactions [120, 270, 271]. The efficient immobilization of amine molecules on silica supports is essential for catalyzed liquid phase reactions. Incomplete immobilization yields to migration of amine molecules, decreasing the activity of the catalysts and reducing their life cycle. In addition, side reactions may occur due to the release of amines into the reacting liquid phase, decreasing the selectivity and product yield, and increasing the separation steps. This article presents a method to immobilize amine molecules on silica supports by using an epoxy resin as an immobilization agent. Immobilized amines have higher molecular weight and are more likely to remain on the supports due to their elevated boiling point [110]. 147 A novel method is presented to evaluate solvent-free organic synthesis reactions in batch mode using a micro-scale reactor and FTIR spectroscopy. The micro-scale reactor allows the fast screening of organic reactions in micro liter scale, reducing the amount of reactants and limiting the production of chemical wastes. A ZnSe window incorporated to the reactor allows the transmission of light in the mid-IR region and permits to follow the reaction by FTIR spectroscopy. The liquid phase Claisen condensation of methyl benzoate and methyl ethyl ketone is shown as an example for the use of our micro-scale reactor to evaluate the activity of the catalysts towards organic catalytic reactions. 8.2 Experimental The synthesis of SBA-15 consisted of (i) dissolving 7.86 g of triblock copolymer (Pluronic 123, BASF) in 146 g of DI water, (ii) adding 6.15 g of acetic acid (Glacial, EMD chemicals) and 22.53 g of sodium silicate (27% sln, PQ corporation) under vigorous stirring for 24 h at 30 oC, (ii) hydro treating the mixture in a closed container for 24 h at 100 oC under static conditions, (iv) filtering and washing the solids with DI water until obtaining a washed liquid with pH=7 and (v) drying in a convection oven for 12 h at 100 oC. The resulting material was divided into two batches: the first batch containing the as-prepared material was labeled as SBA-15 a.p.. The second batch was calcined for 15 h at 600 oC under 25 cm3/min of air and was labeled as SBA-15 cal. Commercial precipitated amorphous silica (Rhodia) was used as received and labeled as SiO2. The amine used for this study was tetraethylenepentamine (TEPA, Sigma- Aldrich) and the immobilization agent was a commercial epoxy resin (bisphenol-A diglycidyl ether, 180 g/epoxide). The immobilized amine catalysts were prepared by 148 impregnating 1 g of SBA-15 a.p., SBA-15 cal or SiO2 with 40 g of an ethanol solution containing 0.8 g of TEPA and 0.4 g of epoxy resin. The solutions and the supports were o mixed for 15 min at 80 C for immobilizing amines on the SBA-15 or SiO2 surface. The catalysts were dried in an oven at 100 oC and were further washed in a 50 wt % solution of water-ethanol for 30 min at 45 oC to remove the incomplete immobilized species. One sample was prepared by impregnating TEPA on SiO2 without epoxy for comparison. This sample was not immobilized or further washed. The catalysts were labeled with the name of the support used and the initials of the impregnated species. For example the catalyst prepared by impregnating TEPA/epoxy on SBA-15 cal was labeled as SBA15cal-TE. The single beam spectra of silica supports and catalysts was collected at 100oC using a DRIFTS cup attached to a Thermo Nicolet 6700 FTIR. The basicity of the o catalysts was estimated by the CO2 adsorption capacity at 30 C. Figure 8.1 shows the schematics of the in-situ FTIR micro-reactor used to evaluate the liquid phase of the Claisen condensation reaction of methyl benzoate (MB, 99%, Avocado research chemicals) and methyl ethyl ketone (MEK, JT Baker). The reactor consists of (i) a ZnSe window, (ii) a stainless steel plate with a 100 µl cavity, (iii) a k-type thermocouple, and (iv) a heating element connected to a temperature controller. The diffuse reflectance reactor is placed in a Harrick Praying Mantis casing collector attached to a Thermo Nicolet 6700 FTIR (IR). For each reaction the reactor was filled with 1 mg of catalyst, 60 µl of MB and 40 µl of MEK. The reactor was heated from 30 to 65 oC at 5 oC/min, and held for 60 min. Air was not allowed inside the reactor and the system remained closed throughout the whole time of reaction. 149 The single beam spectra of the reaction mixture were collected by 32 co-adding scans, resolution of 4 cm-1, at a rate of 6 scans/min. The absorbance spectrum was obtained by the equation Abs = -log (I/I0) where I is the single beam spectrum of interest o and I0 is the single beam spectrum when the temperature of the reaction reached 65 C, selected as time t=0. Figure 8.1 Schematics of the FTIR micro-reactor used for evaluation of liquid phase organic catalytic reactions. 8.3 Results and Discussion 8.3.1 Amine-immobilized catalysts on silica supports The major difference between SiO2 and SBA-15 is the morphology. SiO2 is an inexpensive material, amorphous, with surface area of 160 m2/g and a wide pore size distribution. SBA-15 has an ordered structure with mesoporous channels with surface area of 470 m2/g and high concentration of unbounded silanol groups on its surface. Figure 8.2 (a) shows the IR absorbance spectra of SBA-15 a.p., SBA-15 cal and SiO2. The availability of unbounded silanol groups (Si-OH) is estimated by the absorbance -1 intensity of the band at 3743 cm . Other vibration modes of SiO2 are observed at 809, 1060 and 1180 cm-1 (Si-O and Si-O-Si) and are present in all three supports. 150 Table 8.1 shows that the IR absorbance intensity ratio of Si-OH (3743 cm-1) to Si- -1 O-Si (809, 1060 and 1180 cm ) increases as SBA-15 a.p. << SiO2 < SBA-15 cal for the supports used in this study. Si-OH represents the Si-containing surface functional groups and Si-O-Si represents the Si-containing functional groups in the bulk structure of silica. The IR absorbance intensity ratio (Si-OH/Si-O-Si) gives an estimate of the surface/bulk group contribution towards the functionalization of the supports. The significantly low amount of Si-OH groups for SBA-15 a.p. is due to the presence of hydrocarbon residues and water on its surface. For completely clean surfaces, the IR absorbance intensity ratio (Si-OH/Si-O-Si) should be proportional to the surface area of the supports. Table 8.1 IR absorbance intensity ratio of Si-OH (3743 cm-1) to Si-O-Si (809, 1060, and -1 1180 cm ) for SiO2, SBA-15 a.p. and SBA-15 cal [101, 189, 253, 272]. Aspect ratio of (Si-OH/Si-O-Si) Support 3743/809 3743/1060 3743/1180 SiO2 0.347 0.466 0.466 SBA-15 a.p. 0.153 0.168 0.146 SBA-15 cal 0.465 0.537 0.487 151 (a) asymmetric) -asymmetric) symmetric) 3 2 3 (Si-O-Siasymmetric) 3743(Si-OH unbounded) 3450(H bonding, water) 2980(-CH 2935(-CH 2884(-CH 1735(-C=O from RCOOH) 1180 1060 809(Si-O-Si symmetric) 0.5 SBA-15 cal SBA-15 a.p. SiO 2 IR absorbance (a.u.) 4000 3500 3000 1500 1000 Wavenumbers (cm-1) (b) ) 2 asymmetric) -asymmetric) symmetric) -asymmetric) -deformation) deformation) 3 2 3 2 2 2 3362(N-H asymmetric) 2292(N-H symmetric) 2964(-CH 2930(-CH 2873(-CH 2825(-CH 1676(-C=O NH 1605(NH 1509(Aromatic ring) 1365(C-N-C stretching) 1263(-CH 829(Si-O-C stretching) 0.5 SBA15cal-TE SBA15ap-TE SiO -TE 2 SiO -T 2 IR absorbance (a.u.) 3500 3000 1500 1000 -1 Wavenumbers (cm ) Figure 8.2 (a) IR absorbance spectra and corresponding band assignment for SiO2 and SBA-15 supports collected at 100 oC. (b) IR absorbance spectra of amine immobilized catalysts. Absorbance was obtained by Abs=log(1/I), where I is the normalized single beam of interest. 152 -1 SiO2 and SBA-15 a.p. show a shoulder at 3743 cm due to the unbounded silanol groups and a broad absorption centered at 3450 cm-1 due to the hydrogen bonding interactions between SiO2 and water. SBA-15 a.p. shows strong absorption peaks at 2884, 2935 and 2980 cm-1 (C-H stretching modes) due to the organic residues from the template and other materials used during the preparation. The absorption at 1735 cm-1 for SBA-15 a.p. is produced by the C=O bond of remaining acetic acid on the support. These acidic sites may be localized inside the pores, where the mass transport is limited due to diffusion, hindering the effectiveness of the washing step. The spectrum of SBA- 15 cal exhibits a strong absorption peak for unbounded silanol groups (3743 cm-1) and does not present significant hydrogen bonding or C-H stretching. SBA-15 cal has a cleaner and more hydrophobic surface than SBA-15 a.p. and SiO2 because the calcination step removes organic residues and further oxidizes the support. Impregnation (i.e., physical grafting) of the amine on silica supports is possible due to the hydrogen bond interactions between Si-OH groups from silica and NH/NH2 functional groups from TEPA. Hydrogen bonding is stronger than van der Waals interactions but weaker than a covalent bond, as a consequence is susceptible to dissociate at high temperatures or under the presence of solvents. Immobilization increases the molecular weight of the amine and creates covalent bonds with the support. Figure 8.2 (b) shows the IR absorbance spectra of the four catalysts prepared. The characteristic bands for amines at 3362, 2292 cm-1 (N-H stretching) and 1605 cm-1 (H-N- H bending) confirm the presence of amines in all the catalysts. Those catalysts containing epoxy show a well defined peak at 1509 cm-1 due to the C-C stretching on the aromatic ring of the epoxy chain. The peak at 829 cm-1 is produced by the reaction of 153 epoxy with unbounded silanol groups, and the peak at 1365 cm-1 is attributed to the C-N- C bond formed by the reaction of epoxy with TEPA. The 1676 cm-1 band, a characteristics of an amide species on SiO2-T and SBA15ap-TE, could be produced from (i) the reaction of amine with acetic acid present in the SBA15 a.p. structure and (ii) oxidation of amine on the support surface [107, 108]. Table 8.2 shows the CO2 adsorption capacity for the four catalysts. CO2 adsorption capacity is an indicative of the number of basic sites because CO2 is an acid gas (pka=3.6) and adsorbs on the amine catalysts via acid-base interactions [101, 102, 189]. SiO2-TE exhibits the highest basicity; however, its CO2 adsorption capacity decreases by 0.69 mmol CO2/g-cat (43%) after the washing step. The decrease in CO2 adsorption capacity is due to the removal of incomplete immobilized amine species. SBA15ap-TE and SBA15cal-TE adsorb less CO2 and their adsorption capacity increase after the washing step, suggesting the high stability of these catalysts. The increase in adsorption capacity for SBA-15 catalysts may be a result of redistribution (spacing) of adsorption sites, which increases the availability of occluded amines for interaction with CO2. The high stability may be attributed to a higher degree of immobilization due to the high concentration of silanol groups on SBA-15 surface. Although the Si-OH groups from SBA-15 a.p. were initially occluded by hydrogen bonding with water, the immobilization of TEPA with epoxy was successful. The hydrogen bonding dissociates by the presence of ethanol from the impregnation solution and allows the Si-OH groups to react with the epoxy and amine groups. 154 o Table 8.2 CO2 adsorption capacity and amine efficiency for the catalysts at 30 C. a CO2 adsorption capacity (mmol CO2/g-cat) Amine efficiency Catalyst Before washing After washing (mol CO2/mol NH2) SiO2-T 1.48 - 15.6% SiO2-TE 1.59 0.90 16.8% SBA15ap-TE 0.69 0.71 7.3% SBA15cal-TE 0.59 0.70 6.2% a Amine efficiency is calculated based on the total amount of TEPA loaded, and the CO2 adsorption capacity before washing. 8.3.2 Catalytic activity towards the Claisen condensation of MB and MEK A qualitative IR study of the immobilized amine catalysts towards the Claisen condensation reaction of MB and MEK is described in this section. All the catalysts were active for the formation of β-diketone species, which is the product of this reaction. The IR spectra recorded throughout the reaction provide an insight on the formation of β- diketone and intermediate products for all the catalysts. Figure 8.3 shows the IR absorbance spectra with band assignment for the reaction using (a) SiO2, (b) SiO2-T, and (c) SiO2-TE as catalysts. The absorbance spectra of a control reaction with SiO2 showed flat profiles throughout the time, suggesting that no reaction occurred. The absorbance spectra for SiO2-T and SiO2-TE showed an increase in intensity of some peaks located in the fingerprint region. The most prominent growth in absorption was observed at 1727 cm-1, which corresponds to the C=O stretching of ketone functional group. The relative position of the C=O band depends on the electronic and mass effects from neighboring substituents and intermolecular interactions. For a saturated diketone in equilibrium with enolate species, the predominant vibrations are those observed at 1727 cm-1 for C=O of diketone and 1280 cm-1 for [O=C-C-C=O]- of enolate [192]. 155 ) + 3 ) 3 ) , -NH , 3 asymmetric) asymmetric) + -deformation) amide) 3 2 2 CO) 2 C 2 3 iso (-NH (-CH (R- 3597(R-OH) 3482(R-OH) 3423(NH 2977(-CH 2951(-CH 1727(C=O ketone) 1599 1579 1493(Ar-CO-N) 1452 1434 1365(-CO-CH 1314(CH 1280(-OCCCO-) 1193 1174 1111(-C-C-) (f) (e) (d) 0.2 (c) SiO -TE Time 2 (min) 60 30 0 (b)SiO -T 2 60 30 0 IR Absorbance (a.u) (a) SiO 2 60 30 0 3600 3300 3000 1600 1400 1200 Wavenumbers (cm-1) Figure 8.3 IR absorbance spectra and corresponding band assignment for the Claisen condensation of MB and MEK on (a) SiO2, (b) SiO2-T and (c) SiO2-TE. Absorbance was obtained by Abs=-log(I/Io), where I is the normalized single beam at specific time of the reaction and Io is the normalized single beam at time t=0. The IR absorbance spectra of (d) the condensation product with SiO2-TE catalyst, (e) the condensation product without catalyst and (f) the recovered SiO2-TE catalyst washed with MEK(Abs=-log(1/I)). 156 The band at 1111 cm-1 shows the formation of carbon-carbon bond and those at 1193 and 1174 cm-1 show the isomer resonance structure for a tertiary carbon [192, 273]. -1 + The absorption bands at 1599 and 1579 cm correspond to ammonia ions (-NH2 and - + NH3 ). Ammonia ions are the intermediate species of immobilized amine molecules after the abstraction of a proton to activate the reaction. The stretching modes of CH2 and CH3 for aliphatic portions of the molecule are observed at 2951 and 2977 cm-1 and the C-H -1 stretching on the aromatic ring is observed at 3050 cm . For the reaction with SiO2-T the characteristic peaks of diketone and enolate species (1727 cm-1 and 1280 cm-1) are broader and less intense than those observed over the reaction with SiO2-TE. In addition, -1 absorption peaks at 1493 and 3423 cm are present for the reaction with SiO2-T and -1 absent for the reaction with SiO2-TE. The 1493 cm peak is located in the absorption region of O=C-N bonds and may be attributed to the formation of aromatic amides. The 3423 cm-1 peak represents the N-H stretching absorption of aromatic amides, note that the N-H stretching for amides occur at higher frequencies compared to N-H stretching from amines. These two bands (1493 and 3423 cm-1) are only present for the reaction with SiO2-T, suggesting the formation of amides as a secondary product. These amides were produced from the reaction between reactants or intermediate species (i.e., ester or enolate ion) with incomplete immobilized TEPA molecules that migrated from the catalyst surface. Figure 8.3 also shows the IR absorbance spectra of (d) the condensation product with SiO2-TE catalyst after evaporation of reactants and by-products, (e) the condensation product after removal of the catalyst by filtration, and (f) the recovered catalyst washed with MEK. The bands at 1727 and 1280 cm-1 in spectra (d) and (e) 157 confirm the presence of the C=O and [O=C-C-C=O]- functional groups. The spectrum of the recovered catalyst (f) does not show traces of the reaction product and resembles the characteristic peaks of the fresh catalyst (1509, 1605, 2292 and 3362 cm-1), shown in Figure 8.2, indicating that the product can be completely removed from the catalyst surface. Comparison of the IR spectrum of the recovered catalyst in Figure 8.3 (f) with that of fresh SiO2-TE catalyst in Figure 8.2 (b) shows that less than 50% of the immobilized amine was leached from the catalyst surface. Figure 8.4 shows the absorbance intensities of diketone and enolate species (1727 and 1280 cm-1) as a function of time for the reaction with the four catalysts and pure SiO2. For SiO2 there is no observable change in the intensity for any of these bands, indicating that SiO2 is not active for the catalytic condensation of MB and MEK. The -1 band at 1727 cm (C=O) gradually grows at similar relative rates for SiO2-TE and for -1 - SiO2-T, with a delay of 10 min, but the band at 1280 cm [O=C-C-C=O] grows much faster for SiO2-TE than for SiO2-T. This behavior indicates that in the reaction with SiO2-T some of the = bonds formed belong to species other than β-diketone. These species are probably amides or diesters, which are not produced by carbon-carbon addition reactions. 158 0.50 0.25 SBA15ap-TE SBA15ap-TE 0.40 0.20 SBA15cal-TE SBA15cal-TE 0.30 0.15 SiO -TE SiO -TE 2 2 intensity (a.u.) intensity (a.u.) -1 -1 SIO -T 0.20 2 0.10 0.10 0.05 SIO -T 2 1280 cm 1727 cm 0.00 SiO 0.00 SiO 2 2 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Time (min) Time (min) Figure 8.4 Absorbance intensities of 1727 and 1280 cm-1 as a function of reaction time for the immobilized amine catalysts prepared. For the reaction with SBA15cal-TE, both bands 1727 cm-1 (C=O) and 1280 cm-1 [O=C-C-C=O]- grow at the same relative rate. The growth rate for these bands in the case of SBA15ap-TE is much greater than that for SBA15cal-TE. The shape of the curves for the catalysts prepared on SBA-15 is different than the shape for those prepared on SiO2. The faster rates of production of β-diketone on SBA-15 catalysts may be attributed to the well defined pore size and spacing of amine sites on the surface. The amine packing density on the catalysts surface may change the adsorption modes of the reactants, changing the mechanism of the reaction. The remaining acetic acid on SBA15ap-TE neutralizes the enolate ion by increasing the acidity of ammonia ions (- + + NH2 and -NH3 ). This neutralization produces the irreversible formation of the condensation products. This observation is in concordance with the proposed mechanism for the reaction, but further investigation is required to confirm both the mechanism and the effect of the presence of acetic acid in SBA-15 a.p. Figure 8.5 (b) shows the proposed mechanism for the Claisen condensation reaction of MB and MEK using amine immobilized catalysts. The proposed mechanism is an adaptation from the mechanism of the Claisen condensation reaction for two 159 enolizable species containing α-carbons proposed elsewhere (shown in Figure 8.5 (a)) [108]. In the first step, the basic amine site (pka= ) extracts an α-hydrogen from MEK - (pka=20), yielding a ketone enolate ion in the liquid phase (-C=O-C ) and an ammonia ion + on the catalyst surface (-NH3 ). The enolate ion is a nucleophile and attacks the MB producing the carbon-carbon addition and yielding an alkoxide intermediate (-C-O-). - This alkoxide rearranges the carbonyl group releasing methoxide ion (CH3O , pka=16) and producing the diketone product (pka=9). The strength of the methoxide ion is sufficient to deprotonate the diketone formed, yielding to equilibrium between diketone + methoxide and enolate ion + methanol. In the final step the ammonia ion formed on the catalyst, acts as a Lewis acid, accepting the free electron pair from the enolate ion and irreversibly protonating the intermediate products to condensate β-diketone and methanol. 160 Figure 8.5 (a) Mechanism of the Claisen condensation reaction proposed in [108]. (b) Proposed mechanism for the Claisen condensation of methyl benzoate and methyl ethyl ketone using immobilized amine catalysts. 161 8.4 Conclusions Basic catalysts were prepared by immobilization of tetraethylenepentamine (TEPA) on amorphous silica and SBA-15 supports. Immobilization was achieved by the use of an epoxy resin as an immobilization agent between the silica supports and the amine molecules. During immobilization, covalent bonds are formed between unbounded silanol groups, amine functional groups and epoxy. The basicity of the catalysts before and after washing show that better immobilization of amines is achieved on SBA-15, which has a higher concentration of unbounded silanol groups. Basic catalysts containing immobilized amine molecules are active for the solvent-free Claisen condensation reaction of methyl benzoate and methyl ethyl ketone. A qualitative analysis of formed species was achieved by following the reaction using an in-situ FTIR micro reactor. The reaction was carried out using four different catalysts, and the results show that the formation of β-diketone by a carbon-carbon addition is faster on the catalysts prepared with SBA-15 supports than on those prepared with SiO2. A reaction mechanism was proposed for this reaction. Incomplete immobilization of amine molecules on the supports yields to side reactions and slower formation of the Claisen condensation products. “The final publication is available at link.springer.com: DOI: 10.1007/s11244- 012-9828-9” 162 CHAPTER IX 9POROUS POLY(VINYL ALCOHOL) COMPOSITE MEMBRANES FOR IMMOBILIZATION OF GLUCOSE OXIDASE Particle loaded porous poly(vinyl alcohol) (PVA) composite membranes were selected for immobilization of glucose oxidase (GOx) for their hydrophilicity and unique interactions with amino functional groups. GOx was immobilized on the membranes by adsorption at pH values between 3.5 and 7.1. The highest adsorption loading was observed at pH=7.1 and the highest catalytic activity was observed at pH=5.1. Infrared studies showed that the highest ratio of amide I to amide II at pH=5.1 is obtained for GOx immobilized on membranes loaded with amine-functionalized micro-particles, suggesting that the conformational changes of GOx on these membranes yield to higher catalytic activity than in other supports. 9.1 Introduction Enzymes are catalysts used in biological conversions, synthesis of ultra pure compounds, and food processing. Industrial processes using enzymes have been commercialized, but relatively few use immobilized enzymes because the costs associated with immobilization are not justified in the overall process economics [274]. Advantages of immobilized enzymes include reusability, high chemical and thermal 163 stability, and simple product separation [275, 276]. The motivation for the development of immobilized enzyme processes is that immobilization could lead to continuous operations with high production capacity [275, 277]. The major disadvantage of enzyme immobilization is the loss of catalytic activity [274, 278]. The key issues to be addressed in the development of immobilized enzyme processes with commercial viability are (i) selecting an adequate immobilization technique, and (ii) finding support materials with physicochemical properties that help retain the catalytic activity of the enzyme in industrial operation conditions. Immobilization techniques include enzyme adsorption, covalent attachment, affinity immobilization and encapsulation. We have selected enzyme adsorption because it immobilizes the enzyme without changing its chemical structure; adsorption is achieved by non-covalent interactions between the functional groups of the enzyme and those of the support. These interactions include electrostatic forces [279, 280], hydrophobic interactions [278, 281-283], and hydrophilic interactions [281, 284]. Immobilization may affect the conformation of the enzyme, changing the optimal operation conditions and reducing its catalytic activity [278, 280, 281]. Support materials must be stable, inexpensive, and able to retain the catalytic activity of the enzyme [275, 285]. Enzyme immobilization has been reported on particle-loaded polymeric supports, revealing the effect of the particle properties like conductivity [286-288], surface potential [279, 280], particle size [278], hydrophobicity [281-283] and hydrophilicity [213, 281, 282] on the adsorption and catalytic activity of immobilized enzymes. Poly(vinyl alcohol) (PVA), an inexpensive and non toxic polymer, has been used as a 164 support for enzyme immobilization in the form of hydrogels [289, 290], fibers [291, 292], beads [279, 293], and membranes [213, 294]. This article presents the results of our study on particle-loaded porous PVA composite membranes for adsorption of glucose oxidase (GOx). PVA was selected for its hydrophilicity and unique interactions with amino functional groups. Solid particles were used to enhance the physical properties and adsorption capacity of the membranes. GOx was selected as a model enzyme for its availability, and significance in the medical, pharmaceutical, and food industries [277]. We discuss the effect of particle size and phase inversion on the morphology of the membranes [78, 118, 295], and the effect of cross-linking on the solubility and surface functionality of these materials. The experimental techniques and fundamental understanding obtained from this study could (i) guide the development of novel materials for glucose sensors and biological fuel cells [296], and (ii) be extended for immobilization of other industrial enzymes like carbonic anhydrase, glucose isomerase, and lipase [213, 274, 281, 297]. 9.2 Experimental Membranes casting solution: The preparation of the casting solution consisted of dissolving 5 g of PVA (Elvanol®71-30, DuPont) in 45 mL of de-ionized water (H2O) at 100 oC, cooling, and mixing with another solution containing 2 g of PEI (50 % (w/v) in H2O, Sigma-Aldrich), 14 mL of H2O, 6 mL of ethanol (200 proof, Decon Labs), and 0.5 g of surfactant (SA). Synthesis of solid particles: (i) PVA/PEI micro-particles (PPE) were prepared by dissolving 2 g of cross-linker (CL) in 25 mL of ethanol, and adding to 22 g of the casting 165 solution, the mixture was cross-linked at 75 oC for 5 h. The solids were recuperated by sedimentation, repeatedly washed with ethanol, and dried at 100 oC. (ii) PVA microspheres (PVAsp) were prepared by dissolving 1 g of polyethyleneglycol MW=10,000 (PEG10k, Sigma-Aldrich) and 1 g of polyethyleneglycol MW=200 (PEG200, Sigma-Aldrich) in 80 mL of H2O, and adding 5 g of a 10 wt% PVA solution, 1.8 g of surfactant (SA), 0.85 g of HCl (37 % sln., EMD chemicals), and 1 g of glutaraldehyde o (GA, 25 % sln. in H2O, Alfa Aesar). The solution was cross-linked for 2 h at 90 C, the solids were recuperated by sedimentation, washed with water, and dried at 100 oC. (iii) SBA-15 was prepared as described in a previous publication [298]; 7.86 g of triblock copolymer Pluronic 123 (P-123, BASF) were dissolved in 146 g of H2O, then 6.15 g of acetic acid (Glacial, EMD chemicals) and 22.53 g of sodium silicate (Na2SiO3, Sigma- Aldrich) were slowly added to the solution and kept under vigorous stirring at 30 oC for 24 h. The solution was subjected to hydrothermal treatment under static conditions at 100 oC for 24 h. The solids were recuperated by filtration, repeatedly washed with water to remove the organic residues (P-123 and acetic acid), and dried at 100 oC. Note: In this article we provide detailed characterization of the materials prepared with the surfactant (SA) and the cross-linker (CL), but their names have not been disclosed due to our intentions to fill a patent. The use of SA and CL does not have an effect on how GOx interacts with PVA, which is the focus of this work. Membranes fabrication: The membranes were prepared by mixing 5 g of the casting solution with (i) 0.165 g of PPE, (ii) 0.035 g of PVAsp, or (iii) 0.035 g of SBA- 15, and casting on a Mylar® sheet using a 200 µm casting blade. Immediately after casting, the films were immersed in acetone (Sigma-Aldrich) for 5 min and dried at room 166 temperature for 15 h. The membranes were cross-linked in an aqueous solution containing 20 wt% of Na2SO4 (Fisher), 5 wt% of H2SO4 (95-98 % sln., J.T. Baker), and 0.5 wt% of glutaraldehyde (GA, 25 % sln.) at 70 oC for 1 h. The cross-linked membranes were washed with hot water and dried at 75 oC for 2 h. The membranes were named based on the particles loaded, followed by the letters “ M” to denote “composite membrane”: (i) PPE-CM, (ii) PVAsp-CM, and (iii) SBA-15-CM. Immobilization of glucose oxidase (GOx): Immobilization was achieved by adsorption of GOx on the PVA composite membranes. Four GOx solutions with a concentration of 0.067 mgGOx/mL were prepared by dissolving GOx (Type VII, >100,000 U/g, Sigma-Aldrich) in water (GOx/H2O, 1.4 mg/mL), and diluting it in buffers with pH values of 3.5, 4.1, 5.1, and 7.1. The buffers were prepared using phosphate buffer powder (Sigma-Aldrich) and acetic acid. For the adsorption experiments 51 mg of each membrane were immersed in 21 mL of GOx solution at each pH value. The amount of GOx adsorbed on the membranes was calculated by measuring the concentration of GOx remaining in the solution at different times using UV-vis spectroscopy; GOx absorbs light at 280 nm (GOx product information sheet, Sigma). After adsorption the membranes were repeatedly washed with fresh buffer, and dried at room temperature for 12 h. Enzyme activity tests: The activity of free and immobilized GOx was evaluated by indirect detection of hydrogen peroxide (H2O2), a by-product of the glucose oxidation reaction [299]. A substrate solution containing 0.80 mg of horseradish peroxidase (HRPx, Type I, 50-150 U/mg, Sigma-Aldrich), 6.30 mg of 4-aminoantipyrine (4-AAP, Sigma-Aldrich), 1.95 g of phenol (Sigma-Aldrich), and 0.766 g of glucose (D-glucose, 167 o Sigma-Aldrich) in 500 mL of H2O was prepared and incubated at 4 C for 24 h. An aliquot of 250 µl of the substrate solution was added to vials containing 10 mL of buffer and (i) 21 mg of GOx-immobilized membrane or (ii) 10 µL of GOx/H2O solution (1.4 mgGOx/mL) for the activity tests with free enzyme (i,e,. non-immobilized). The UV-vis spectra of the solutions were recorded as a function of time. GOx + H O + O + H O 2 2 Glucose oxidase 2 2 (160 kDa) UV-visabs=280 nm Hydrogen Glucose Gluconic acid peroxide (180g/mol) (196g/mol) (34g/mol) Quinoneimine dye (294g/mol) UV-visabs= + H O 2 2 400, 505 nm HRPx Horseradish peroxidase (44 kDa) 4-aminoantipyrine Phenol (203g/mol) (34g/mol) + 2 H2O Figure 9.1 Glucose oxidation reaction catalyzed by GOx, and formation of a quinoneimine dye for indirect detection of H2O2 by UV-vis spectroscopy. Figure 9.1 shows the oxidation reaction of glucose to gluconic acid and hydrogen peroxide catalyzed by GOx in presence of water and oxygen. H2O2 reacts with 4-AAP and phenol in presence of HRPx to produce a quimoneimine dye. The H2O2 detection by UV-vis spectroscopy is based on the formation of this dye which absorbs light at 505 nm, and the formation of other species of the co-oxidation of phenol and 4-AAP that absorb light at 400 nm [300]. The physical structure of the membranes was characterized by SEM (Hitachi, tabletop TM 3000) and BET (Micromeritics ASAP 2020). IR spectroscopy (Thermo Nicolet 6700 FTIR) was used to analyze the cross-linking 168 reactions and the conformation of immobilized GOx. UV-vis spectroscopy (Hitachi U- 3900) was used to quantify the adsorption loading and catalytic activity of GOx. 9.3 Results and Discussion Preparation of PVA composite membranes: PVA composite membranes were prepared by phase inversion of the casting solution loaded with solid particles followed by cross-linking with GA. Phase inversion in acetone precipitates PVA and removes water from the casted films creating macropores across the membranes. The presence of ethanol in the casting solution reduces the driving force for polymer precipitation, inducing the formation of a homogeneous porous structure [78]. Solid micro-particles adsorb the polymer chains of the casting solution limiting their mobility and promoting the precipitation of the polymer. The presence of solid particles enhances the mechanical properties of the membranes [295], but reduces the probability of macropores formation during the phase inversion. This effect is more drastic for particles smaller than 1 µm, and for those with significantly large surface area [118]. Figure 9.2 shows the IR absorbance spectra and SEM pictures of the solid particles. The spectrum of PPE shows the O-H and C-O stretching vibrations of PVA at 3405 and 1038 cm-1, and the N-H deformation and C-N stretching vibrations of secondary amines at 1608 and 1250 cm-1. The secondary amines in PPE are generated from the reaction of the cross-linker (CL) with the primary amine groups of PEI. We calculated the ratio of the IR intensity of N-H to O-H as 0.77, to demonstrate the large availability of amine functional groups in PPE. The spectrum of PVAsp shows the O-H stretching vibrations of cross-linked PVA at 3490 cm-1, the C-O and C-O-C stretching vibrations of acetal groups at 1141 and 1008 cm-1, and the C=O stretching vibration of 169 un-reacted aldehyde groups at 1725 cm-1. The spectrum of SBA-15 shows the stretching vibration of hydrogen bonded Si-OH at 3440 cm-1 and those of Si-O at 1018, 1060, and 809 cm-1. The band at 1725 cm-1 is produced by traces of acetic acid remaining from the synthesis of SBA-15. Most of these bands overlap with those of the pristine membrane, and are not evident in the spectra presented in Figure 9.3. (Si-O) SBA-15 (C=O) 2 s.a. = 695 m /g (O-H) 809 1118 1060 dp ~ 0.45 m 0.5 1725 1 µm 3440 1141 1008 PVAsp (C-O) 2 3490 s.a. = 475 m /g dp ~ 0.42 m 1 µm 1250 1038 (C-N) 3405 1608 PPE (N-H) s Absorbance (a.u.) dp ~ 4.5 m s.a. = 326 m2/g 3400 1500 1000 -1 5 µm NH/OH=0.77 Wavenumbers (cm ) Figure 9.2 IR absorbance spectra and SEM pictures of the solid particles. The spectra were collected in DRIFTS (diffuse reflectance infrared Fourier transform spectroscopy) at 100 oC, Abs=log(1/Single beam). The surface area (s.a.) was measured by the BET method. Figure 9.3 shows the IR spectra of the composite membranes before and after cross-linking, and the SEM pictures of the cross-linked membranes. The presence of PEI is confirmed by the N-H deformation of amine groups at 1571 cm-1. The N-H stretching vibrations of PEI at 3350 and 3280 cm-1 are not visible due to overlapping with the broad absorption of the O-H stretching vibrations of PVA at 3340 cm-1. Other characteristic bands of PVA include the C-O stretching vibrations at 968 and 1093 cm-1, and that of its crystalline domains at 1141 cm-1. The stretching vibrations, deformation, and wagging of -1 CH2 appear at 2939, 2848, 1446 and 1328 cm . After cross-linking, the amine band at 170 1571 cm-1 disappears and a new band at 1629 cm-1 is formed. This band represents the C=C- and = stretching vibrations of enamines and imines (Schiff’s bases) generated by the reaction of PEI with GA. Cross-linking also produces an absorption band at 968 cm-1 and an evident increase of that at 1093 cm-1, which represent the C-O and C-O-C stretching vibrations of acetal groups formed by the cross-linking reaction of PVA with GA. The cross-linked membranes are insoluble in water and stable at the pH values used in this study. The spectra of the composite membranes have the same characteristic bands as that of the pristine membrane (not shown), and their total absorption intensity increases as PPE-CM < PVAsp-CM < SBA-15-CM. This observation reflects the difference in void fraction (ϕ) between the membranes; in transmission mode the IR absorbance intensity is proportional to the sample concentration (c) and to the pathlength of the light being transmitted through the sample (l). The samples with higher void fraction have lower density and consequently lower concentration in the light pathlenght. This phenomenon is explained by the Beer-Lambert law where absorbance=εcl, and the molar absorptivity (ε) is approximately the same for all the membranes. The SEM pictures and BET analyses confirm that PPE-CM has larger void fraction and surface area than PVAsp-CM and SBA-15-CM. These structural differences may be explained by examining the size of the particles in each membrane. PVAsp and SBA-15 are small particles with high surface area; these particles adsorb the PEI and PVA chains more effectively than PPE, stretching the polymer matrix, and limiting the formation of macropores. 171 /g /g /g 2 2 2 ~ 0.19 ~ 0.32 ~ 0.53 SBA-15-CM 141 m = s.a. PVAsp-CM 191 m = s.a. PPE-CM 317 m = s.a. 10 µm 10 µm 10 µm 10 Top view Top 50 µm 50 µm 50 µm 50 50 µm 50 µm 50 µm 50 Side view Side s (C-O-C) 968 (C-O) 1093 1000 cryst (C-O) 1141 ) -1 1200 2 ) (CH 1328 2 ) (CH 1446 1400 s (N-H) 1571 (C=C-N) 1629 (C=N) 1600 2 s ) (CH 2848 After cross-linking After cross-linking Before 2910 2 as ) (CH 2939 Wavenumbers (cm Wavenumbers 3000 (O-H) 3340 3500 Absorbance (a.u.) Absorbance 0.5 Figure 9.3 IR absorbance spectra of the PVA composite membranes before (black) and after cross-linking (red), and SEM pictures of the cross-linked membranes. The IR spectra were collected in transmission mode at 20 oC, Abs=log(1/Single beam). The surface area (s.a.) was measured by the BET method, and the void fraction (ϕ) was estimated from the SEM pictures using image processing and analysis software (ImageJ). 172 Immobilization of glucose oxidase (GOx): Glucose oxidase was immobilized on the PVA composite membranes by adsorption from dilute solutions at various pH values. Figure 9.4 shows (a) the amount of GOx adsorbed on the membranes as a function of time, (b) the amount of GOx adsorbed at the end of the experiment (i,e,. after 69 h) as a function of pH, which is assumed to be in equilibrium with the concentration of GOx in the solutions, and (c) the adsorption isotherm of GOx adsorbed on PPE-CM at pH=5.1. At pH=3.5, below the isoelectric point of GOx (pI=4.2), the membranes show the lowest adsorption values because the enzyme adopts its protonated form and may experience repulsive electrostatic interactions with the membrane. In acid environments the hydroxyl groups of PVA may carry a weak positive charge (p a≈ -4), and the imine and enamine groups of cross-linked PEI are highly protonated (p a≈8-10 [220]). These repulsive electrostatic interactions could work against other attractive interactions for immobilization, especially in PPE- M whose particles have amine groups (p a≈9-10) that are highly protonated at low pH values and partially protonated at high pH values [301, 302]. At pH>4.2 the enzyme carries a negative charge and the protonation degree of the membrane decreases significantly, the electrostatic interactions are much weaker than those at pH=3.5 and become less significant in the adsorption mechanism. At higher pH values other attractive forces are more important for the adsorption of GOx; the primarily adsorption mechanism has been attributed to hydrophobic interactions between the non- polar aminoacids of the enzyme and the hydrophobic segments of the membrane [278, 280-282]. These segments include the backbone of PVA, the ethylene groups of PEI, and the cross-linking acetal groups. Hydrophilic interactions like hydrogen bonding between 173 the polar aminoacids of GOx and the hydroxyl groups of PVA could also participate in the adsorption of GOx in the entire range of pH studied. The adsorption values of PPE- CM are lower than those of PVAsp-CM and SBA-15-CM because the amine groups in PPE create more hydrogen bonding interactions with GOx and PVA, reducing the probability of hydrophobic interactions between GOx and the membrane. in equilibrium (69 h) 30 (a) (b) pH=3.5 20 GOx/PPE-CM GOx/PVAsp-CM 30 10 GOx/SBA-15-CM 0 30 20 pH=4.1 20 10 10 0 0 30 GOx ads. (mg GOx/g) 3 4 5 6 7 pH=5.1 pH 20 (c) 10 400 GOx/PPE-CM GOx ads. (mg GOx/g) 0 300 pH=5.1 30 pH=7.1 200 20 10 100 0 0 0 10 20 30 40 50 60 70 GOx ads. (mg GOx/g) 0.0 0.5 1.0 1.5 Time (h) GOx in sln. (mgGOx/ml) Figure 9.4 (a) Amount of GOx adsorbed on PVA composite membranes as a function of time, (b) equilibrium amount of GOx adsorbed on the membranes as a function of pH, and (c) adsorption isotherm of GOx adsorbed on PPE-CM at pH=5.1. Figure 9.4 (b) shows that the adsorption values of GOx on the membranes increase with the pH of the impregnation solution. PVAsp-CM and SBA-15-CM reached 174 their highest adsorption value of 23.6 and 23.2 mgGOx/g at pH=7, and PPE-CM reached its highest adsorption value of 8.2 mgGOx/g at pH=5.1. Note that these adsorption experiments were performed in dilute concentrated solutions; up to 331 mgGOx/g from GOx solutions with concentration of 1.4 mg GOx/mL. Adsorption loadings up to 270 mgGOx/g have been reported on other supports for adsorption from significantly more concentrated GOx solutions (>1.5 mgGOx/mL) and at pH values different from 5.1, the optimal pH for GOx activity [278, 280, 303]. The adsorption performance of the membranes presented in this work is better than that reported for other supports, especially in the diluted range. Adsorption in the diluted range is of interest because adsorption from highly concentrated solutions could result in agglomeration and multi- layer deposition, leading to inefficient immobilization (i,e., enzyme leaching) and rapid loss of enzymatic activity. GOx solutions (0.067 mgGOx/mL) and the maximum adsorption value possible to achieve is 27.6 mgGOx/g. Use of GOx solutions with concentrations up to 1.4 mg GOx/mL produced the adsorption isotherm of GOx on PPE- CM at pH=5.1, shown in Figure 9.4 (c). The slope of the curve indicates that this membrane has the potential to adsorb significantly larger amounts of GOx from more Enzyme activity tests: Figure 9.5 shows the catalytic activity of free and immobilized GOx towards oxidation of glucose to gluconic acid and hydrogen peroxide at various pH values. The catalytic activity of the enzyme is expressed in units per mass; one unit (U) will oxidize one µmole of glucose per minute at pH=5.1 and 35 oC (GOx product information sheet, Sigma). The specific activity of GOx is based on the net weight of the enzyme (U/mg-GOx), and the membrane activity is based on the weight of the enzyme plus the support (U/g-membrane). The highest activity of free and 175 immobilized GOx was observed at pH=5.1, suggesting that immobilization does not change the optimal pH conditions for the catalytic activity of GOx. The highest specific activity of free GOx was 149.8 U/mg-GOx, and the highest specific activity of immobilized GOx was 15.5 U/mgGOx, observed for PPE-CM at pH=5.1. Similar activity reductions between free and immobilized GOx have been reported for immobilization of GOx on other supports; from 141 to 12 U/mgGOx for adsorption on organo-functionalized gold nanoparticles [304], from 159 to 11 U/mgGOx for adsorption on ion-exchange mixed matrix hollow fibers [280], and from 162 to 13 U/mgGOx for GOx adsorbed on graphene oxide [278]. These activity reductions have been attributed to conformational changes caused by interactions with the supports. The membrane activity, shown in Figure 9.5 (b), is about at the same level for all the membranes. Although the specific activity of GOx on PPE-CM is high, its adsorption values is lower than that on PVAsp-CM and SBA-15-CM. The low adsorption value of GOx on PPE-CM compensates for its high specific activity, resulting in similar membrane activity for all the membranes. The higher activity retention of GOx on PPE-CM suggests that the physical and chemical environment of PPE-CM affects the conformation of GOx less than PVAsp-CM and SBA-15-CM. Two significant differences of the micro-environment of PPE-CM compared to that of PVAsp-CM and SBA-15-CM are (i) the presence of secondary amine groups (NH/OH ratio of 0.77), and (ii) the size of the loaded particles. It has been reported that the presence of amines could help increase the activity of GOx in sensors [305], increase its stability on amine functionalized zeolites [306], and yield to high adsorption loadings on silica supports and magnetic particles [279, 307]. The results 176 presented in this article suggest that GOx has better activity retention in presence of amine functional groups. The size of the support also affects the conformation and activity of GOx. Adsorption on nano-particles has been found to produce conformational changes in GOx resulting from extensive protein folding and deformation due to excessive hydrophobicity and high curvature of particular domains in the support with sizes in the same order of magnitude as GOx [278, 284, 304]. GOx is an elipse of ~10 nm long and ~5 nm wide [280, 282]. Our results show that despite of the low adsorption value, the microenvironment of PPE-CM could help retaining the activity of GOx. (a) (b) GOx/PPE-CM GOx/PVAsp-CM 160 160 GOx/SBA-15-CM 120 80 120 Free GOx 40 80 15 10 40 5 GOx Activity (U/mg-GOx) 0 0 3 4 5 6 7 GOx Activity (U/g-membrane) 3 4 5 6 7 pH pH Figure 9.5 (a) Specific catalytic activity of free GOx, and GOx immobilized in PVA composite membranes as a function of pH. (b) Membrane activity of GOx based on the weight of enzyme plus the support (U/g-membrane). Infrared spectroscopy studies: We investigated the conformation of immobilized GOx by analyzing the IR spectra of the dry membranes before and after the adsorption experiments. Figure 9.6 shows the IR absorbance spectra of immobilized GOx, and the 177 intensity ratio of the amide I and amide II bands. Amide I appears between 1700 and 1600 cm-1, and amide II appears between 1600 and 1500 cm-1. The absorbance spectra of GOx were obtained by substracting the absorbance spectra of the dry membranes before the adsorption experiments from the absorbance spectra of the membranes after adsorption of GOx. Variations in the intensity ratio of amide I to amide II have been reported to represent conformational changes in immobilized GOx [278, 282, 308]. This ratio is proportional to the amount of protein present on the membrane and is useful to analyze structural changes of the enzyme; in general higher ratios reflect small conformational changes. The most relevant observation in Figure 9.6 is the behavior of the amide I to amide II ratio with respect to the pH. PVAsp-CM and SBA-15-CM show high ratios at pH=3.5 and pH=7.1, and the lowest ratio at pH=5.1. PPE-CM shows low ratios at pH=3.5 and 7.1, and the highest ratio at pH=5.1. These observations suggest that the conformational changes of GOx immobilized at the optimal pH value of 5.1 are less significant on PPE-CM, whose micro-particles contain amine functional groups. It is important to note that the activity of immobilized GOx does not only depend on the conformation of the enzyme, governed by the chemical and physical environment of the support, but also on the pH, temperature, and other parameters of the reaction media. This proposition is supported by the results of GOx immobilized on PVAsp-CM and SBA-15-CM at pH=3.5 and pH=7.1. Although the ratio of amide I to amide II is high, suggesting that the conformational of changes of GOx are low, the pH of the reaction media altered the enzyme resulting in low catalytic activity. The ratio of amide I to amide II may be used to study the conformational changes of the enzyme immobilized 178 in different environments but it is not a direct measurement of the activity of immobilized GOx. Future studies will focus on the in-situ spectroscopic investigation of the conformation of the immobilized on supports with various NH/OH ratios. GOx/PPE-CM GOx/PVAsp-CM GOx/SBA-15-CM Amide I Amide II Amide I Amide II Amide I Amide II pH 0.05 7.1 5.1 4.1 3.5 Absorbance (a.u.) 1700 1600 1500 1700 1600 1500 1700 1600 1500 Wavenumbers (cm-1) 1.6 1.4 1.2 1.0 0.8 3 4 5 6 7 Ratio (AmideI/AmideII) pH Figure 9.6 IR absorbance spectra and intensity ratio of the amide I to amide II bands of immobilized GOx. The spectra were obtained by substraction of the absorbance spectra of the dry membranes before the adsorption experiments from the absorbance spectra of the membranes after adsorption of GOx. The intensity of amide I and amide II was measured using a baseline from the minimum points of the curves around 1700, 1600 and 1500 cm-1. 9.4 Conclusions This article presents the detailed characterization of particle-loaded PVA composite membranes for adsorption of GOx from dilute solutions (0.067 mgGOx/mL). The predominant adsorption mechanism of GOx on the membranes was attributed to 179 hydrophobic interactions between the non-polar aminoacids of GOx and the hydrophobic segments of the membranes. The highest adsorption value of 23.6 mgGOx/g was obtained for PVA membranes at pH=7.1, and represents 83% of the maximum possible loading of 27.6 mgGOx/g. Our results show that porous PVA composite membranes loaded with solid micro-particles containing amine functional groups serve as a support for GOx and could help retaining its catalytic activity. The results from this study may serve as a guide for designing and fabricating support materials with the proper chemical and physical environments for retaining the catalytic activity of industrial enzymes. 180 CHAPTER X 10CONCLUSIONS Porous polymer-silica hybrids have unique properties that are of interest for adsorption and catalysis applications. The physical and chemical properties of these materials were modified by combining a set of preparation methods including (i) sol-gel synthesis, (ii) direct templating, (iii) cross-linking, and (iv) phase inversion. This dissertation demonstrated through its hypothesis 1, that economic alternative routes for the synthesis of poly(vinyl alcohol) (PVA) and silica (SiO2) porous supports are feasible, and that the combination of the mentioned techniques can be used for tailoring the morphology and physicochemical properties of polymer-silica hybrid materials. These approaches were demonstrated by the synthesis of low cost SBA-15 and polymer-silica membranes with multiple morphologies. Supporting the hypothesis 2 of this dissertation, the formulation of amine- functionalized CO2 capture sorbents was presented. The development of silica based pellets using a cross-linked polymer binder solution was studied with aims of demonstrating the applicability of the fundamental concepts and experimental techniques developed throughout the course of this dissertation. A compartmental modeling technique was developed for simulating the adsorption of CO2 on amine-functionalized 181 silica sorbents. The heat effects on the CO2 adsorption kinetics were investigated by the analysis of two parametric studies, and sensitivity tests were applied to investigate the outcomes of key variables in the operation of a CO2 capture unit; adsorption temperature, flow rate, and partial pressure of CO2. Using spectroscopic techniques, it was shown that the cross-linking reactions of PVA with GA are catalyzed by the presence of the amine groups of PEI. The results of this study supported the hypothesis 3 of this dissertation, and generated another sets of hypotheses regarding the mechanism and intermediates involved in the cross-linking reactions of PVA. Furthermore, the polymer-silica hybrids developed in this dissertation found applications in catalysis of organic reactions and enzyme immobilization. 182 BIBLIOGRAPHY 1. Mansur, H.S., R.L. Orefice, and A.A.P. Mansur, Characterization of poly(vinyl alcohol)/poly(ethylene glycol) hydrogels and PVA-derived hybrids by small-angle X-ray scattering and FTIR spectroscopy. Polymer, 2004. 45(21): p. 7193-7202. 2. Moriya, Y., et al., Preparation of polymer hybrids of the silica-PVA or silica-PEG system and porous materials made from the hybrid gels. J. Ceram. Soc. Jpn., 1993. 101(May): p. 518-21. 3. Ogoshi, T. and Y. Chujo, Synthesis of anionic polymer-silica hybrids by controlling pH in an aqueous solution. J. Mater. Chem., 2005. 15(2): p. 315-322. 4. Ahmed, A., et al., Porous silica spheres in macroporous structures and on nanofibers. Philos. Trans. R. Soc., A, 2010. 368(1927): p. 4351-4370. 5. Cheng, S., X. Wang, and S.-Y. Chen, Applications of Amine-functionalized Mesoporous Silica in Fine Chemical Synthesis. Top. Catal., 2009. 52(6-7): p. 681-687. 6. Sanchez, C., et al., Applications of hybrid organic-inorganic nanocomposites. Journal of Materials Chemistry, 2005. 15(35-36): p. 3559-3592. 7. Wu, S.-H., C.-Y. Mou, and H.-P. Lin, Synthesis of mesoporous silica nanoparticles. Chem Soc Rev, 2013. 42(9): p. 3862-75. 8. Studart, A.R., et al., Hierarchical Porous Materials Made by Drying Complex Suspensions. Langmuir, 2011. 27(3): p. 955-964. 9. Jang, J. and J. Bae, Fabrication of mesoporous polymer/silica hybrid using surfactant-mediated sol-gel method. J. Non-Cryst. Solids, 2006. 352(38-39): p. 3979- 3984. 10. Schmitz, P. and S. Janocha, Films, in Ullmann's Encyclopedia of Industrial Chemistry. 2000, Wiley-VCH Verlag GmbH & Co. KGaA. 11. Johansson, B. and G. Alderborn, Degree of pellet deformation during compaction and its relationship to the tensile strength of tablets formed of microcrystalline cellulose pellets. International Journal of Pharmaceutics, 1996. 132(1–2): p. 207-220. 183 12. Shih, C.K., Fundamentals of polymer compounding: the phase-inversion mechanism during mixing of polymer blends. Adv. Polym. Technol., 1992. 11(3): p. 223- 6. 13. Li, N., et al., Preparation and properties of PVDF/PVA hollow fiber membranes. Desalination, 2010. 250(2): p. 530-537. 14. Kim, S.-G. and K.-H. Lee, Poly(vinyl alcohol) membranes having an integrally skinned asymmetric structure. Mol. Cryst. Liq. Cryst., 2009. 512: p. 1878-1885. 15. Bhargava, R., B.G. Wall, and J.L. Koenig, Comparison of the FT-IR Mapping and Imaging Techniques Applied to Polymeric Systems. Appl. Spectrosc., 2000. 54(4): p. 470- 479. 16. Oh, S.J. and J.L. Koenig, Phase and Curing Behavior of Polybutadiene/Diallyl Phthalate Blends Monitored by FT-IR Imaging Using Focal-Plane Array Detection. Analytical Chemistry, 1998. 70(9): p. 1768-1772. 17. Lin, S. and P. Theato, CO2-Responsive Polymers. Macromolecular Rapid Communications, 2013. 34(14): p. 1118-1133. 18. Madejová, J., FTIR techniques in clay mineral studies. Vibrational Spectroscopy, 2003. 31(1): p. 1-10. 19. Ping, Z.H., et al., States of water in different hydrophilic polymers — DSC and FTIR studies. Polymer, 2001. 42(20): p. 8461-8467. 20. Daniliuc, L., C. De Kesel, and C. David, Intermolecular interactions in blends of poly(vinyl alcohol) with poly(acrylic acid)—1. FTIR and DSC studies. European Polymer Journal, 1992. 28(11): p. 1365-1371. 21. Sculley, J.P. and H.-C. Zhou, Enhancing Amine-Supported Materials for Ambient Air Capture. Angew. Chem., Int. Ed., 2012. 51(51): p. 12660-12661. 22. Wang, J., et al., Surfactant promoted solid amine sorbents for CO2 capture. Energy Environ. Sci., 2012. 5(2): p. 5742-5749. 23. Wang, L., et al., Experimental and modeling investigation on post-combustion carbon dioxide capture using zeolite 13X-APG by hybrid VTSA process. Chem. Eng. J. (Amsterdam, Neth.), 2012. 197: p. 151-161. 24. Liu, Y., et al., Carbon Dioxide Capture by Functionalized Solid Amine Sorbents with Simulated Flue Gas Conditions. Environ. Sci. Technol., 2011: p. ACS ASAP. 25. Duan, Y., D.C. Sorescu, and D. Luebke, Efficient theoretical screening of solid sorbents for CO2 capture applications. Proc. - Annu. Int. Pittsburgh Coal Conf., 2011. 28th: p. a19/1-a19/17. 184 26. Chaikittisilp, W., H.-J. Kim, and C.W. Jones, Mesoporous Alumina-Supported Amines as Potential Steam-Stable Adsorbents for Capturing CO2 from Simulated Flue Gas and Ambient Air. Energy Fuels, 2011. 25(11): p. 5528-5537. 27. Qi, G., et al., High efficiency nanocomposite sorbents for CO2 capture based on amine-functionalized mesoporous capsules. Energy Environ. Sci., 2011. 4(2): p. 444-452. 28. Su, F., et al., Adsorption of CO2 on Amine-Functionalized Y-Type Zeolites. Energy Fuels, 2010. 24(2): p. 1441-1448. 29. Alesi, W.R., M. Gray, and J.R. Kitchin, CO2 Adsorption on Supported Molecular Amidine Systems on Activated Carbon. ChemSusChem, 2010. 3(8): p. 948-956. 30. Heydari-Gorji, A. and A. Sayari, CO2 capture on polyethylenimine-impregnated hydrophobic mesoporous silica: Experimental and kinetic modeling. Chemical Engineering Journal, 2011. 173(1): p. 72-79. 31. Heydari-Gorji, A., Y. Belmabkhout, and A. Sayari, Polyethylenimine- Impregnated Mesoporous Silica: Effect of Amine Loading and Surface Alkyl Chains on CO2 Adsorption. Langmuir, 2011. 27(20): p. 12411-12416. 32. Harlick, P.J.E. and A. Sayari, Applications of Pore-Expanded Mesoporous Silicas. 3. Triamine Silane Grafting for Enhanced CO2 Adsorption. Ind. Eng. Chem. Res., 2006. 45(9): p. 3248-3255. 33. Sayari, A. and S. Hamoudi, Periodic Mesoporous Silica-Based Organic−Inorganic Nanocomposite Materials. Chemistry of Materials, 2001. 13(10): p. 3151-3168. 34. Rezaei, F., et al., Aminosilane-Grafted Polymer/Silica Hollow Fiber Adsorbents for CO2 Capture from Flue Gas. ACS Applied Materials & Interfaces, 2013. 35. Bollini, P., S.A. Didas, and C.W. Jones, Amine-oxide hybrid materials for acid gas separations. J. Mater. Chem., 2011. 21(39): p. 15100-15120. 36. Choi, S., et al., Application of Amine-Tethered Solid Sorbents for Direct CO2 Capture from the Ambient Air. Environ. Sci. Technol., 2011. 45(6): p. 2420-2427. 37. Chaikittisilp, W., et al., Poly(allylamine)-mesoporous silica composite materials for CO2 capture from simulated flue gas or ambient air. Ind. Eng. Chem. Res., 2011. 50(24): p. 14203-14210. 38. Chaikittisilp, W., et al., Poly(L-lysine) brush-mesoporous silica hybrid material as a biomolecule-based adsorbent for CO2 capture from simulated flue gas and air. Chemistry, 2011. 17(38): p. 10556-61. 185 39. Hicks, J.C., et al., Designing Adsorbents for CO2 Capture from Flue Gas - Hyperbranched Aminosilicas Capable of Capturing CO2 Reversibly. J. Am. Chem. Soc., 2008. 130(10): p. 2902-2903. 40. Heyne, S. and S. Harvey, Impact of choice of CO2 separation technology on thermo-economic performance of Bio-SNG production processes. International Journal of Energy Research, 2014. 38(3): p. 299-318. 41. Li, S., et al., Post-combustion CO2 capture using super-hydrophobic, polyether ether ketone, hollow fiber membrane contactors. J. Membr. Sci., 2013. 430: p. 79-86. 42. Zhao, M., A.I. Minett, and A.T. Harris, A review of techno-economic models for the retrofitting of conventional pulverised-coal power plants for post-combustion capture (PCC) of CO2. Energy & Environmental Science, 2013. 6(1): p. 25-40. 43. Skorek-Osikowska, A., J. Kotowicz, and K. Janusz-Szymanska, Comparison of the Energy Intensity of the Selected CO2-Capture Methods Applied in the Ultra- supercritical Coal Power Plants. Energy Fuels, 2012. 26(11): p. 6509-6517. 44. Kulkarni, A.R. and D.S. Sholl, Analysis of Equilibrium-Based TSA Processes for Direct Capture of CO2 from Air. Ind. Eng. Chem. Res., 2012. 51(25): p. 8631-8645. 45. Cormos, C.-C., Integrated assessment of IGCC power generation technology with carbon capture and storage (CCS). Energy (Oxford, U. K.), 2012. 42(1): p. 434-445. 46. Meerman, J.C., et al., Techno-economic assessment of CO2 capture at steam methane reforming facilities using commercially available technology. International Journal of Greenhouse Gas Control, 2012. 9: p. 160-171. 47. Rodríguez, N., et al., Comparison of experimental results from three dual fluidized bed test facilities capturing CO2 with CaO. Energy Procedia, 2011. 4(0): p. 393-401. 48. Damen, K., et al., A comparison of electricity and hydrogen production systems with CO2 capture and storage - Part B: Chain analysis of promising CCS options. Prog. Energy Combust. Sci., 2007. 33(6): p. 580-609. 49. Damen, K., et al., A comparison of electricity and hydrogen production systems with CO2 capture and storage. Part A: Review and selection of promising conversion and capture technologies. Prog. Energy Combust. Sci., 2006. 32(2): p. 215-246. 50. Sharma, P., et al., Amine modified and pelletized mesoporous materials- Synthesis, textural-mechanical characterization and application in adsorptive separation of carbon dioxide. Powder Technol., 2012. 219: p. 86-98. 51. Modekurti, S., D. Bhattacharyya, and S.E. Zitney, Dynamic Modeling and Control Studies of a Two-Stage Bubbling Fluidized Bed Adsorber-Reactor for Solid– 186 Sorbent CO2 Capture. Industrial & Engineering Chemistry Research, 2013. 52(30): p. 10250-10260. 52. Veneman, R., et al., Continuous CO2 capture in a circulating fluidized bed using supported amine sorbents. Chem. Eng. J. (Amsterdam, Neth.), 2012. 207-208: p. 18-26. 53. Nguyen, P.T., et al., A dense membrane contactor for intensified CO2 gas/liquid absorption in post-combustion capture. J. Membr. Sci., 2011. 377(1-2): p. 261-272. 54. Reijerkerk, S.R., et al., Highly hydrophilic, rubbery membranes for CO2 capture and dehydration of flue gas. Int. J. Greenhouse Gas Control, 2011. 5(1): p. 26-36. 55. Liu, B., et al., Synthesis of organic-inorganic hybrid microspheres and the corresponding mesoporous silica nanoparticles. J. Colloid Interface Sci., 2013. 411: p. 98-104. 56. Fonseca dos Reis, E., et al., Synthesis and characterization of poly(vinyl alcohol) hydrogels and hybrids for rMPB70 protein adsorption. Mater. Res. (Sao Carlos, Braz.), 2006. 9(2): p. 185-191. 57. Innocenzi, P., T. Kidchob, and T. Yoko, Hybrid Organic-Inorganic Sol-Gel Materials Based on Epoxy-Amine Systems. Journal of Sol-Gel Science and Technology, 2005. 35(3): p. 225-235. 58. Airoldi, C. and O.A.C. Monteiro, Chitosan–organosilane hybrids—Syntheses, characterization, copper adsorption, and enzyme immobilization. Journal of Applied Polymer Science, 2000. 77(4): p. 797-804. 59. Pereira, A.P.V., W.L. Vasconcelos, and R.L. Oréfice, Novel multicomponent silicate–poly(vinyl alcohol) hybrids with controlled reactivity. Journal of Non-Crystalline Solids, 2000. 273(1–3): p. 180-185. 60. Zok, F.W. and C.G. Levi, Mechanical Properties of Porous-Matrix Ceramic Composites. Advanced Engineering Materials, 2001. 3(1-2): p. 15-23. 61. Pham, H.Q. and M.J. Marks, Epoxy Resins, in Ullmann's Encyclopedia of Industrial Chemistry. 2000, Wiley-VCH Verlag GmbH & Co. KGaA. 62. Drisko, G.L., et al., One-Pot Preparation and CO2 Adsorption Modeling of Porous Carbon, Metal Oxide, and Hybrid Beads. ACS Applied Materials & Interfaces, 2013. 5(11): p. 5009-5014. 63. Keshtkar, A.R., M. Irani, and M.A. Moosavian, Comparative study on PVA/silica membrane functionalized with mercapto and amine groups for adsorption of Cu(II) from aqueous solutions. Journal of the Taiwan Institute of Chemical Engineers, 2013. 44(2): p. 279-286. 187 64. Zhang, S., et al., Fabrication of Ordered Porous Polymer Film via a One-Step Strategy and Its Formation Mechanism. Macromolecules, 2009. 42(10): p. 3591-3597. 65. Qian, Z., et al., Facile preparation of monodisperse hollow cross-linked chitosan microspheres. Journal of Polymer Science Part A: Polymer Chemistry, 2008. 46(1): p. 228-237. 66. Chen, K.-P., et al., Synthesis of CO2-philic Xanthate-Oligo(vinyl acetate)-Based Hydrocarbon Surfactants by RAFT Polymerization and Their Applications on Preparation of Emulsion-Templated Materials. Macromolecules (Washington, DC, U. S.), 2010. 43(22): p. 9355-9364. 67. Thomas, A., et al., Porous Polymers: Enabling Solutions for Energy Applications. Macromolecular Rapid Communications, 2009. 30(4-5): p. 221-236. 68. Walcarius, A., Mesoporous materials and electrochemistry. Chemical Society Reviews, 2013. 69. Hoffmann, F., et al., Silica-Based Mesoporous Organic–Inorganic Hybrid Materials. Angewandte Chemie International Edition, 2006. 45(20): p. 3216-3251. 70. Bai, J., et al., Synthesis of hybrid mesoporous polystyrene-silica materials with non-surfactant citric acid as template via sol-gel process. Chin. J. Polym. Sci., 2002. 20(6): p. 565-572. 71. Galarneau, A., et al., Thermal and mechanical stability of micelle-templated silica supports for catalysis. Catalysis Today, 2001. 68(1–3): p. 191-200. 72. Guillen, G.R., et al., Preparation and Characterization of Membranes Formed by Nonsolvent Induced Phase Separation: A Review. Industrial & Engineering Chemistry Research, 2011. 50(7): p. 3798-3817. 73. Strathmann, H., Membrane Separation Processes, 3. Membrane Preparation and Membrane Module Constructions, in Ullmann's Encyclopedia of Industrial Chemistry. 2000, Wiley-VCH Verlag GmbH & Co. KGaA. 74. Strathmann, H. and K. Kock, The formation mechanism of phase inversion membranes. Desalination, 1977. 21(3): p. 241-255. 75. Wang, D., K. Li, and W.K. Teo, Relationship between mass ratio of nonsolvent- additive to solvent in membrane casting solution and its coagulation value. J. Membr. Sci., 1995. 98(3): p. 233-40. 76. Sundararaj, U., C.W. Macosko, and C.-K. Shih, Evidence for inversion of phase continuity during morphology development in polymer blending. Polymer Engineering & Science, 1996. 36(13): p. 1769-1781. 188 77. Smolders, C.A., et al., Microstructures in phase-inversion membranes. Part 1. Formation of macrovoids. J. Membr. Sci., 1992. 73(2–3): p. 259-275. 78. Kim, S.-G., et al., Preparation of asymmetric PVA membranes using ternary system composed of polymer and cosolvent. J. Appl. Polym. Sci., 2003. 88(13): p. 2884- 2890. 79. Zhang, Y. and B. Gao, Preparation of Functional Grafted Particles PVA/SiO2 with High Grafting Degree and Preliminary Research of Their Adsorption Character. Journal of Macromolecular Science, Part A, 2013. 50(2): p. 238-247. 80. Labreche, Y., et al., Post-spinning infusion of poly(ethyleneimine) into polymer/silica hollow fiber sorbents for carbon dioxide capture. Chem. Eng. J. (Amsterdam, Neth.), 2013. 221: p. 166-175. 81. Yang, J. and C.-R. Han, Dynamics of Silica-Nanoparticle-Filled Hybrid Hydrogels: Nonlinear Viscoelastic Behavior and Chain Entanglement Network. J. Phys. Chem. C, 2013. 117(39): p. 20236-20243. 82. Ramadan, H., et al., Synthesis and characterization of mesoporous hybrid silica- polyacrylamide aerogels and xerogels. Silicon, 2011. 3(2): p. 63-75. 83. Uemura, T., et al., Incarceration of Nanosized Silica into Porous Coordination Polymers: Preparation, Characterization, and Adsorption Property. Chem. Mater., 2011. 23(7): p. 1736-1741. 84. Nagai, D., A. Suzuki, and T. Kuribayashi, Synthesis of Hydrogels from Polyallylamine with Carbon Dioxide as Gellant: Development of Reversible CO2 Absorbent. Macromolecular Rapid Communications, 2011. 32(4): p. 404-410. 85. Rokita, B., J.M. Rosiak, and P. Ulanski, Ultrasound-Induced Cross-Linking and Formation of Macroscopic Covalent Hydrogels in Aqueous Polymer and Monomer Solutions. Macromolecules, 2009. 42(9): p. 3269-3274. 86. Xiao, C. and M. Yang, Controlled preparation of physical cross-linked starch-g- PVA hydrogel. Carbohydrate Polymers, 2006. 64(1): p. 37-40. 87. Ossipov, D.A. and J. Hilborn, Poly(vinyl alcohol)-Based Hydrogels Formed by “Click Chemistry”. Macromolecules, 2006. 39(5): p. 1709-1718. 88. Zou, J. and W.S.W. Ho, CO2-selective polymeric membranes containing amines in crosslinked poly(vinyl alcohol). J. Membr. Sci., 2006. 286(Copyright (C) 2012 American Chemical Society (ACS). All Rights Reserved.): p. 310-321. 89. Xomeritakis, G., C.-Y. Tsai, and C.J. Brinker, Microporous sol–gel derived aminosilicate membrane for enhanced carbon dioxide separation. Separation and Purification Technology, 2005. 42(3): p. 249-257. 189 90. Hedin, N., et al., Adsorbents for the post-combustion capture of CO2 using rapid temperature swing or vacuum swing adsorption. Appl. Energy, 2013. 104: p. 418-433. 91. Ho, W.S.W. and K. Li, Recent advances in separations. Curr. Opin. Chem. Eng., 2012. 1(2): p. 145-147. 92. Ramdin, M., L.T.W. de, and T.J.H. Vlugt, State-of-the-Art of CO2 Capture with Ionic Liquids. Ind. Eng. Chem. Res., 2012. 51(24): p. 8149-8177. 93. Samanta, A., et al., Post-Combustion CO2 Capture Using Solid Sorbents: A Review. Ind. Eng. Chem. Res., 2012. 51(4): p. 1438-1463. 94. Patel, N.P., A.C. Miller, and R.J. Spontak, Highly CO2-Permeable and Selective Polymer Nanocomposite Membranes. Advanced Materials, 2003. 15(9): p. 729-733. 95. Liu, S.L., et al., Recent progress in the design of advanced PEO-containing membranes for CO2 removal. Progress in Polymer Science, 2013(0). 96. DORTMUNDT, et al., CO[2] Removal membrane technology: Recent development. Vol. 38. 2003, Mumbai, INDE: Jasubhai. 12. 97. Chang, A.C.C., et al., In-Situ Infrared Study of CO2 Adsorption on SBA-15 Grafted with gamma -(Aminopropyl)triethoxysilane. Energy Fuels, 2003. 17(2): p. 468- 473. 98. Chang, F.-Y., et al., Adsorption of CO2 onto amine-grafted mesoporous silicas. Separation and Purification Technology, 2009. 70(1): p. 87-95. 99. Berger, A.H. and A.S. Bhown, Comparing physisorption and chemisorption solid sorbents for use separating CO2 from flue gas using temperature swing adsorption. Energy Procedia, 2011. 4(0): p. 562-567. 100. Shimizu, S., C. Song, and M. Strano, Role of Specific Amine Surface Configurations for Grafted Surfaces: Implications for Nanostructured CO2 Adsorbents. Langmuir, 2011. 27(6): p. 2861-2872. 101. Zelenak, V., et al., Amine-modified SBA-12 mesoporous silica for carbon dioxide capture: Effect of amine basicity on sorption properties. Microporous Mesoporous Mater., 2008. 116(1-3): p. 358-364. 102. Khatri, R.A., et al., Thermal and Chemical Stability of Regenerable Solid Amine Sorbent for CO2 Capture. Energy Fuels, 2006. 20(4): p. 1514-1520. 103. Srikanth, C.S. and S.S.C. Chuang, Spectroscopic Investigation into Oxidative Degradation of Silica-Supported Amine Sorbents for CO2 Capture. ChemSusChem, 2012. 5(8): p. 1435-1442. 190 104. Sharma, S., et al., The simultaneous measurement of support, metal, and gas phase temperatures using an in situ IR spectroscopic technique. Journal of Catalysis, 1988. 110(1): p. 103-116. 105. Richner, G. and G. Puxty, Assessing the Chemical Speciation during CO2 Absorption by Aqueous Amines Using in Situ FTIR. Industrial & Engineering Chemistry Research, 2012. 51(44): p. 14317-14324. 106. Petit, C., et al., Spectroscopic Investigation of the Canopy Configurations in Nanoparticle Organic Hybrid Materials of Various Grafting Densities during CO2 Capture. J. Phys. Chem. C, 2012. 116(1): p. 516-525. 107. Chakravartula S. Srikanth, S.S.C.C., Spectroscopic investigation on oxidative degradation of silica-supported amine sorbents for CO2 capture. ChemSusChem, 2012, DOI:10.1002/cssc.201100662. 108. McMurry, J., Organic Chemistry. 8th ed. 2010, Belmont, CA: Brooks/Cole CENGAGE learning. 109. Teo, L.-S., C.-Y. Chen, and J.-F. Kuo, Fourier Transform Infrared Spectroscopy Study on Effects of Temperature on Hydrogen Bonding in Amine-Containing Polyurethanes and Poly(urethane−urea)s. Macromolecules, 1997. 30(6): p. 1793-1799. 110. Tanthana, J. and S.S.C. Chuang, In Situ Infrared Study of the Role of PEG in Stabilizing Silica-Supported Amines for CO2 Capture. ChemSusChem, 2010. 3(8): p. 957-964. 111. Rocher, N. and R. Frech, Hydrogen Bonding and Cation Coordination Effects in Primary and Secondary Amines Dissolved in Carbon Tetrachloride. J. Phys. Chem. A, 2007. 111(14): p. 2662-2669. 112. Chrissopoulou, K., et al., Crystallinity and Chain Conformation in PEO/Layered Silicate Nanocomposites. Macromolecules, 2011. 44(24): p. 9710-9722. 113. Liao, W., et al., Preparation of Organic/Inorganic Hybrid Polymer Emulsions with High Silicon Content and Sol-gel-derived Thin Films. Chinese Journal of Chemical Engineering, 2010. 18(1): p. 156-163. 114. Voigt, I., et al., Controlled synthesis of stable poly(vinyl formamide- <i>co</i> -vinyl amine)/silica hybrid particles by interfacial post-cross- linking reactions. Colloid & Polymer Science, 2000. 278(1): p. 48-56. 115. Fidalgo, A., et al., Flexible hybrid aerogels prepared under subcritical conditions. J. Mater. Chem. A, 2013. 1(39): p. 12044-12052. 116. Lin, K.-Y.A. and A.-H.A. Park, Effects of Bonding Types and Functional Groups on CO2 Capture using Novel Multiphase Systems of Liquid-like Nanoparticle Organic Hybrid Materials. Environ. Sci. Technol., 2011. 45(15): p. 6633-6639. 191 117. Elimelech, H. and D. Avnir, Sodium-Silicate Route to Submicrometer Hybrid PEG@Silica Particles. Chemistry of Materials, 2008. 20(6): p. 2224-2227. 118. Harton, S.E., et al., Immobilized Polymer Layers on Spherical Nanoparticles. Macromolecules, 2010. 43(7): p. 3415-3421. 119. Zhao, B., A Combinatorial Approach to Study Solvent-Induced Self-Assembly of Mixed Poly(methyl methacrylate)/Polystyrene Brushes on Planar Silica Substrates: Effect of Relative Grafting Density. Langmuir, 2004. 20(26): p. 11748-11755. 120. Saravanamurugan, S., et al., Transesterification reactions over morphology controlled amino-functionalized SBA-15 catalysts. Catal. Commun., 2007. 9(1): p. 158- 163. 121. Choi, M., et al., Controlled Polymerization in Mesoporous Silica toward the Design of Organic−Inorganic Composite Nanoporous Materials. Journal of the American Chemical Society, 2005. 127(6): p. 1924-1932. 122. Wang, Y., et al., Synthesis of monodisperse, hierarchically mesoporous, silica microspheres embedded with magnetic nanoparticles. ACS Appl. Mater. Interfaces, 2012. 4(5): p. 2735-2742. 123. Shen, C., et al., Physicochemical properties of poly(ethylene oxide)-based composite polymer electrolytes with a silane-modified mesoporous silica SBA-15. Electrochimica Acta, 2009. 54(12): p. 3490-3494. 124. Tsiourvas, D., et al., A novel hybrid sol-gel method for the synthesis of highly porous silica employing hyperbranched poly(ethyleneimine) as a reactive template. Microporous Mesoporous Mater., 2013. 175: p. 59-66. 125. Nguyen, Q.X., et al., Water-Phase Synthesis of Cationic Silica/Polyamine Nanoparticles. Chemistry of Materials, 2012. 24(8): p. 1426-1433. 126. Croes, K.J., et al., An electrochemical study of corrosion protection by in situ oxidative polymerization in phenylenediamine crosslinked sol–gel hybrid coatings. Electrochimica Acta, 2011. 56(23): p. 7796-7804. 127. Kim, T.-W., et al., Ordered Mesoporous Polymer−Silica Hybrid Nanoparticles as Vehicles for the Intracellular Controlled Release of Macromolecules. ACS Nano, 2010. 5(1): p. 360-366. 128. Hamciuc, E., C. Hamciuc, and M. Olariu, Thermal and electrical behavior of polyimide/silica hybrid thin films. Polym. Eng. Sci., 2010. 50(3): p. 520-529. 129. Elimelech, H., et al., Cross-linked polyethylene and silica: the first full interpenetrating network hybrid particles. J. Mater. Chem., 2010. 20(42): p. 9515-9522. 192 130. Ilharco, L.M., et al., Nanostructured silica/polymer subcritical aerogels. Journal of Materials Chemistry, 2007. 17(21): p. 2195-2198. 131. Godjevargova, T., R. Nenkova, and V. Konsulov, Immobilization of glucose oxidase by acrylonitrile copolymer coated silica supports. J. Mol. Catal. B: Enzym., 2006. 38(2): p. 59-64. 132. Cestari, A.R. and C. Airoldi, Diamine Immobilization on Silica Gel through the Sol-Gel Process and Increase in the Organic Chain by Using Glutaraldehyde followed by Ethylenediamine. Langmuir, 1997. 13(10): p. 2681-2686. 133. Weichold, O., et al., A comparative study on the dispersion stability of aminofunctionalised silica nanoparticles made from sodium silicate. Journal of Colloid And Interface Science, 2008. 324(1-2): p. 105-109. 134. Flörke, O.W., et al., Silica, in Ullmann's Encyclopedia of Industrial Chemistry. 2000, Wiley-VCH Verlag GmbH & Co. KGaA. 135. Cong, H., et al., Polymer–inorganic nanocomposite membranes for gas separation. Separation and Purification Technology, 2007. 55(3): p. 281-291. 136. Wright, J.D.S., Nico A.J.M., Sol-Gel Materials: Chemistry and Applications. Advanced Chemistry Texts, ed. D. Phillips. 2001, Australia: Gordon and Breach Science Publishers. 125. 137. Livage, J. and C. Sanchez, Sol-gel chemistry. Journal of Non-Crystalline Solids, 1992. 145(0): p. 11-19. 138. Lagaly, G., et al., Silicates, in Ullmann's Encyclopedia of Industrial Chemistry. 2000, Wiley-VCH Verlag GmbH & Co. KGaA. 139. Peters, K. and F.M. Richards, Chemical cross-linking: reagents and problems in studies of membrane structure. Annual Review of Biochemistry, 1977. 46: p. 523-551. 140. Peterson, A.M., R.E. Jensen, and G.R. Palmese, Reversibly Cross-Linked Polymer Gels as Healing Agents for Epoxy−Amine Thermosets. ACS Applied Materials & Interfaces, 2009. 1(5): p. 992-995. 141. Mansur, H.S., et al., FTIR spectroscopy characterization of poly(vinyl alcohol) hydrogel with different hydrolysis degree and chemically crosslinked with glutaraldehyde. Mater. Sci. Eng., C, 2008. 28(4): p. 539-548. 142. Bai, H. and W.S.W. Ho, Carbon Dioxide-Selective Membranes for High-Pressure Synthesis Gas Purification. Ind. Eng. Chem. Res., 2011. 50(21): p. 12152-12161. 143. Mansur, H.S., et al., Cytocompatibility evaluation in cell-culture systems of chemically crosslinked chitosan/PVA hydrogels. Mater. Sci. Eng., C, 2009. 29(5): p. 1574-1583. 193 144. Xing, R. and W.S.W. Ho, Synthesis and characterization of crosslinked polyvinylalcohol/polyethyleneglycol blend membranes for CO2/CH4 separation. J. Taiwan Inst. Chem. Eng., 2009. 40(6): p. 654-662. 145. Bolto, B., et al., Crosslinked poly(vinyl alcohol) membranes. Progress in Polymer Science, 2009. 34(9): p. 969-981. 146. Hansen, E.W., et al., Reaction of poly(vinyl alcohol) and dialdehydes during gel formation probed by 1H n.m.r.—a kinetic study. Polymer, 1997. 38(19): p. 4863-4871. 147. Hallensleben, M.L., Polyvinyl Compounds, Others, in Ullmann's Encyclopedia of Industrial Chemistry. 2000, Wiley-VCH Verlag GmbH & Co. KGaA. 148. Srinivasa Rao, P., et al., Preparation and performance of poly(vinyl alcohol)/polyethyleneimine blend membranes for the dehydration of 1,4-dioxane by pervaporation: Comparison with glutaraldehyde cross-linked membranes. Separation and Purification Technology, 2006. 48(3): p. 244-254. 149. Kim, K.-J., S.-B. Lee, and N.-W. Han, Kinetics of crosslinking reaction of PVA membrane with glutaraldehyde. Korean Journal of Chemical Engineering, 1994. 11(1): p. 41-47. 150. Wu, L. and C.S. Brazel, Modifying the release of proxyphylline from PVA hydrogels using surface crosslinking. International Journal of Pharmaceutics, 2008. 349(1–2): p. 144-151. 151. Kim, J.H., et al., Properties and swelling characteristics of cross-linked poly(vinyl alcohol)/chitosan blend membrane. J. Appl. Polym. Sci., 1992. 45(10): p. 1711-1717. 152. Figueiredo, K.C.S., T.L.M. Alves, and C.P. Borges, Poly(vinyl alcohol) films crosslinked by glutaraldehyde under mild conditions. J. Appl. Polym. Sci., 2009. 111(6): p. 3074-3080. 153. Zou, J. and W.S.W. Ho, CO2-selective polymeric membranes containing amines in crosslinked poly(vinyl alcohol). J. Membr. Sci., 2006. 286(1+2): p. 310-321. 154. Kudo, S., E. Otsuka, and A. Suzuki, Swelling behavior of chemically crosslinked PVA gels in mixed solvents. Journal of Polymer Science Part B: Polymer Physics, 2010. 48(18): p. 1978-1986. 155. Gohil, J.M. and P. Ray, Polyvinyl alcohol as the barrier layer in thin film composite nanofiltration membranes: Preparation, characterization, and performance evaluation. J. Colloid Interface Sci., 2009. 338(1): p. 121-127. 156. Okuda, K., et al., Reaction of glutaraldehyde with amino and thiol compounds. Journal of Fermentation and Bioengineering, 1991. 71(2): p. 100-105. 194 157. Gölander, C.-G. and J.C. Eriksson, ESCA studies of the adsorption of polyethyleneimine and glutaraldehyde-reacted polyethyleneimine on polyethylene and mica surfaces. J. Colloid Interface Sci., 1987. 119(1): p. 38-48. 158. Fang, X., et al., Facile immobilization of gold nanoparticles into electrospun polyethyleneimine/polyvinyl alcohol nanofibers for catalytic applications. Journal of Materials Chemistry, 2011. 21(12): p. 4493-4501. 159. Fang, X., et al., Fabrication and characterization of water-stable electrospun polyethyleneimine/polyvinyl alcohol nanofibers with super dye sorption capability. New Journal of Chemistry, 2011. 35(2): p. 360-368. 160. Wu, D., et al., Design and Preparation of Porous Polymers. Chemical Reviews, 2012. 112(7): p. 3959-4015. 161. Coenen, A., T.L. Church, and A.T. Harris, Biological versus Synthetic Polymers as Templates for Calcium Oxide for CO2 Capture. Energy Fuels, 2012. 26(1): p. 162- 168. 162. Abramson, S., et al., Highly porous and monodisperse magnetic silica beads prepared by a green templating method. Journal of Materials Chemistry, 2010. 20(23): p. 4916-4924. 163. Chaibundit, C., et al., Effect of Ethanol on the Micellization and Gelation of Pluronic P123. Langmuir, 2008. 24(21): p. 12260-12266. 164. Zhu, J., et al., High-Surface-Area SBA-15 with Enhanced Mesopore Connectivity by the Addition of Poly(vinyl alcohol). Chem. Mater., 2011. 23(8): p. 2062-2067. 165. Lee, M.-H., et al., Hydrolysis of Magnetic Amylase-Imprinted Poly(ethylene-co- vinyl alcohol) Composite Nanoparticles. ACS Appl. Mater. Interfaces, 2012. 4(2): p. 916-921. 166. Piletsky, S.A., et al., Surface Functionalization of Porous Polypropylene Membranes with Molecularly Imprinted Polymers by Photograft Copolymerization in Water. Macromolecules, 2000. 33(8): p. 3092-3098. 167. Lettow, J.S., et al., Hexagonal to Mesocellular Foam Phase Transition in Polymer-Templated Mesoporous Silicas. Langmuir, 2000. 16(22): p. 8291-8295. 168. Yilmaz, L. and A.J. McHugh, Analysis of nonsolvent–solvent–polymer phase diagrams and their relevance to membrane formation modeling. J. Appl. Polym. Sci., 1986. 31(4): p. 997-1018. 169. Hansen, C.M., Hansen Solubility Parameters. Taylor & Francis Group, 2007. Second edition. 195 170. Zhai, H. and E.S. Rubin, Techno-Economic Assessment of Polymer Membrane Systems for Postcombustion Carbon Capture at Coal-Fired Power Plants. Environ. Sci. Technol., 2013. 47(6): p. 3006-3014. 171. Yang, Z.-Z., et al., Carbon dioxide utilization with C-N bond formation: carbon dioxide capture and subsequent conversion. Energy Environ. Sci., 2012. 5(5): p. 6602- 6639. 172. Wang, M., et al., Post-combustion CO2 capture with chemical absorption: A state-of-the-art review. Chemical Engineering Research and Design, 2011. 89(9): p. 1609-1624. 173. Pirngruber, G.D. and D. Leinekugel-le-Cocq, Design of a PSA process for post- combustion CO2 capture. Ind. Eng. Chem. Res., 2013. 174. Zhao, Y., et al., Carbon dioxide adsorption on polyacrylamide-impregnated silica gel and breakthrough modeling. Appl. Surf. Sci., 2012. 261: p. 708-716. 175. Yan, X., et al., Amine-Modified SBA-15: Effect of Pore Structure on the Performance for CO2 Capture. Ind. Eng. Chem. Res., 2011. 50(6): p. 3220-3226. 176. Yan, X., et al., Amine-modified mesocellular silica foams for CO2 capture. Chem. Eng. J. (Amsterdam, Neth.), 2011. 168(2): p. 918-924. 177. Witoon, T., Polyethyleneimine-loaded bimodal porous silica as low-cost and high-capacity sorbent for CO2 capture. Materials Chemistry and Physics, 2012. 137(1): p. 235-245. 178. Wang, X., et al., Molecular basket sorbents polyethylenimine–SBA-15 for CO2 capture from flue gas: Characterization and sorption properties. Microporous and Mesoporous Materials, 2013. 169(0): p. 103-111. 179. Wang, H.-B., P.G. Jessop, and G. Liu, Support-Free Porous Polyamine Particles for CO2 Capture. ACS Macro Letters, 2012. 1(8): p. 944-948. 180. Subagyono, D.J.N., et al., Amine modified mesocellular siliceous foam as a sorbent for CO2. Chem. Eng. Res. Des., 2011. 89(9): p. 1647-1657. 181. Roth, E.A., S. Agarwal, and R.K. Gupta, Nanoclay-Based Solid Sorbents for CO2 Capture. Energy & Fuels, 2013. 182. Qi, G., et al., Efficient CO2 sorbents based on silica foam with ultra-large mesopores. Energy & Environmental Science, 2012. 5(6): p. 7368-7375. 183. Stuckert, N.R. and R.T. Yang, CO2 Capture from the Atmosphere and Simultaneous Concentration Using Zeolites and Amine-Grafted SBA-15. Environ. Sci. Technol., 2011. 45(23): p. 10257-10264. 196 184. Suh, D.-M. and X. Sun, Particle-scale CO2 adsorption kinetics modeling considering three reaction mechanisms. Int. J. Greenhouse Gas Control, 2013. 17: p. 388- 396. 185. Srikanth, C.S. and S.S.C. Chuang, Infrared Study of Strongly and Weakly Adsorbed CO2 on Fresh and Oxidatively Degraded Amine Sorbents. The Journal of Physical Chemistry C, 2013. 186. Isenberg, M. and S.S.C. Chuang, The Nature of Adsorbed CO2 and Amine Sites on the Immobilized Amine Sorbents Regenerated by Industrial Boiler Steam. Industrial & Engineering Chemistry Research, 2013. 52(35): p. 12530-12539. 187. Danon, A., P.C. Stair, and E. Weitz, FTIR Study of CO2 Adsorption on Amine- Grafted SBA-15: Elucidation of Adsorbed Species. The Journal of Physical Chemistry C, 2011. 115(23): p. 11540-11549. 188. Bacsik, Z.n., et al., Mechanisms and Kinetics for Sorption of CO2 on Bicontinuous Mesoporous Silica Modified with n-Propylamine. Langmuir, 2011. 27(17): p. 11118-11128. 189. Khatri, R.A., et al., Carbon Dioxide Capture by Diamine-Grafted SBA-15: A Combined Fourier Transform Infrared and Mass Spectrometry Study. Ind. Eng. Chem. Res., 2005. 44(10): p. 3702-3708. 190. Mebane, D.S., et al., Transport, Zwitterions and the Role of Water for \ce{CO2} Adsorption in Mesoporous Silica-Supported Amine Sorbents. The Journal of Physical Chemistry C, 2013. 191. Bacsik, Z., et al., Temperature-Induced Uptake of CO2 and Formation of Carbamates in Mesocaged Silica Modified with n-Propylamines. Langmuir, 2010. 26(12): p. 10013-10024. 192. Norman B. Colthup, L.H.D., Stephen E. Wiberley, Introduction to Infrared and Raman Spectroscopy. 3rd ed. 1990: Elsevier Science. 193. Fisher, J.C., II, J. Tanthana, and S.S.C. Chuang, Oxide-supported tetraethylenepentamine for CO2 capture. Environ. Prog. Sustainable Energy, 2009. 28(4): p. 589-598. 194. Tanthana, J., Study of Amine Impregnated on Silica Support for CO2 Capture, in Department of Chemical and Biomolecular Engineering. 2008, The University of Akron: Akron, OH. 195. Aziz, B., G. Zhao, and N. Hedin, Carbon Dioxide Sorbents with Propylamine Groups-Silica Functionalized with a Fractional Factorial Design Approach. Langmuir, 2011. 27(7): p. 3822-3834. 197 196. Vinoba, M., et al., CO2 Absorption and Sequestration as Various Polymorphs of CaCO3 Using Sterically Hindered Amine. Langmuir, 2013. 29(50): p. 15655-15663. 197. Wu, X.-Y., et al., Preparation and characterization of novel physically cross- linked hydrogels composed of poly(vinyl alcohol) and amine-terminated polyamidoamine dendrimer. Macromol Biosci, 2004. 4(2): p. 71-5. 198. Zhao, Y. and W.S. Winston Ho, Steric hindrance effect on amine demonstrated in solid polymer membranes for CO2 transport. J. Membr. Sci., 2012(0). 199. Kim, T.-J., et al., Separation performance of PVAm composite membrane for CO2 capture at various pH levels. J. Membr. Sci., 2013. 428: p. 218-224. 200. Ahmad, F., et al., Process simulation and optimal design of membrane separation system for CO2 capture from natural gas. Computers & Chemical Engineering, 2012. 36(0): p. 119-128. 201. Mark, J.E., Polymer Data Handbook. Oxford University Press, 1998. V3. electronic version. 202. STEVEN S.C. C HUANG , B.E.C.H., Y AWU C HI , A BDELHAMID S AYARI, NANOSTRUCTURED MATERIALS: Synthesis of Zeolites. Chemical Engineering Education, 2004. 38(1, Winter 2004): p. 34-47. 203. Kato, K., et al., Temperature-Sensitive Nonionic Vesicles Prepared from Span 80 (Sorbitan Monooleate). Langmuir, 2008. 24(19): p. 10762-10770. 204. Santhanalakshmi, J. and S.I. Maya, Solvent effects on reverse micellisation of Tween 80 and Span 80 in pure and mixed organic solvents. Proceedings of the Indian Academy of Sciences - Chemical Sciences, 1997. 109(1): p. 27-38. 205. Ribar, T., J.L. Koenig, and R. Bhargava, FTIR Imaging of Polymer Dissolution. 2. Solvent/Nonsolvent Mixtures. Macromolecules, 2001. 34(23): p. 8340-8346. 206. Wade, J.L.G., Organic Chemistry. Sixth edition ed. 2006, Upper Saddle River, NJ: Prentice Hall. 207. Zhao, S., et al., Amino-functionalized SBA-15 immobilized NiBr2(PPh3)2 as a highly effective catalyst for ATRP of MMA. Microporous Mesoporous Mater., 2010. 137(1-3): p. 36-42. 208. Dirk, J., S. Carsten, and H. Martin, Covalent Anchoring of Chloroperoxidase and Glucose Oxidase on the Mesoporous Molecular Sieve SBA-15. International Journal of Molecular Sciences, 2010. 11(2). 209. Kim, S.-S., T.R. Pauly, and T.J. Pinnavaia, Non-ionic surfactant assembly of ordered, very large pore molecular sieve silicas from water soluble silicates. Chemical Communications, 2000. 0(17): p. 1661-1662. 198 210. Kim, S.-S. and T.J. Pinnavaia, A low cost route to hexagonal mesostructured carbon molecular sieves. Chemical Communications, 2001. 0(23): p. 2418-2419. 211. Pirzada, T., et al., Hybrid Silica–PVA Nanofibers via Sol–Gel Electrospinning. Langmuir, 2012. 28(13): p. 5834-5844. 212. Thielemann, J.P., et al., Pore structure and surface area of silica SBA-15: influence of washing and scale-up. Beilstein Journal of Nanotechnology, 2011. 2: p. 110- 118. 213. Djennad, M.h., et al., Poly(vinyl alcohol) ultrafiltration membranes: Synthesis, characterization, the use for enzyme immobilization. Eng. Life Sci., 2003. 3(11): p. 446- 452. 214. Huang, Y., et al., Efficient Catalytic Reduction of Hexavalent Chromium Using Palladium Nanoparticle-Immobilized Electrospun Polymer Nanofibers. ACS Appl. Mater. Interfaces, 2012. 4(6): p. 3054-3061. 215. Peng, F., Z. Jiang, and E.M.V. Hoek, Tuning the molecular structure, separation performance and interfacial properties of poly(vinyl alcohol)–polysulfone interfacial composite membranes. J. Membr. Sci., 2011. 368(1–2): p. 26-33. 216. Yang, Y.-F., L.-S. Wan, and Z.-K. Xu, Surface hydrophilization of microporous polypropylene membrane by the interfacial crosslinking of polyethylenimine. J. Membr. Sci., 2009. 337(1–2): p. 70-80. 217. Hirai, T., et al., Effect of chemical cross-linking under elongation on shape restoring of poly(vinyl alcohol) hydrogel. Journal of Applied Polymer Science, 1992. 46(8): p. 1449-1451. 218. Lang, K., et al., A study on the preparation of polyvinyl alcohol thin-film composite membranes and reverse osmosis testing. Desalination, 1996. 104(3): p. 185- 196. 219. McMurry, J., Organic Chemistry. 2000, Pacific Groove, CA: Brooks/Cole. 220. Zhu, S., M.F. Brown, and S.E. Feller, Retinal Conformation Governs pKa of Protonated Schiff Base in Rhodopsin Activation. Journal of the American Chemical Society, 2013. 135(25): p. 9391-9398. 221. Bandyopadhyay, A., Amine versus ammonia absorption of CO2 as a measure of reducing GHG emission: a critical analysis. Clean Technol. Environ. Policy, 2011. 13(2): p. 269-294. 222. Mebane, D.S., et al., Bayesian calibration of thermodynamic models for the uptake of CO2 in supported amine sorbents using ab initio priors. Phys. Chem. Chem. Phys., 2013. 15(12): p. 4355-4366. 199 223. Joss, L. and M. Mazzotti, Modeling the extra-column volume in a small column setup for bulk gas adsorption. Adsorption, 2012. 18(5-6): p. 381-393. 224. Yan, W., et al., Carbon Dioxide Capture by Amine-Impregnated Mesocellular- Foam-Containing Template. Ind. Eng. Chem. Res., 2012. 51(9): p. 3653-3662. 225. Begag, R., et al., Superhydrophobic amine functionalized aerogels as sorbents for CO2 capture. Greenhouse Gases: Sci. Technol., 2013. 3(Copyright (C) 2013 American Chemical Society (ACS). All Rights Reserved.): p. 30-39. 226. Serna-Guerrero, R. and A. Sayari, Modeling adsorption of CO2 on amine- functionalized mesoporous silica. 2: Kinetics and breakthrough curves. Chemical Engineering Journal, 2010. 161(1–2): p. 182-190. 227. Knöfel, C., et al., Study of Carbon Dioxide Adsorption on Mesoporous Aminopropylsilane-Functionalized Silica and Titania Combining Microcalorimetry and in Situ Infrared Spectroscopy. The Journal of Physical Chemistry C, 2009. 113(52): p. 21726-21734. 228. Sircar, S. and J.R. Hufton, Why Does the Linear Driving Force Model for Adsorption Kinetics Work? Adsorption, 2000. 6(2): p. 137-147. 229. Garcia, S., et al., Predicting Mixed-Gas Adsorption Equilibria on Activated Carbon for Precombustion CO2 Capture. Langmuir, 2013. 29(20): p. 6042-6052. 230. Gebald, C., et al., Single-Component and Binary CO2 and H2O Adsorption of Amine-Functionalized Cellulose. Environ. Sci. Technol., 2014. 48(4): p. 2497-2504. 231. Bollini, P., et al., Dynamics of CO2 Adsorption on Amine Adsorbents. 2. Insights Into Adsorbent Design. Ind. Eng. Chem. Res., 2012. 51(46): p. 15153-15162. 232. Foo, K.Y. and B.H. Hameed, Insights into the modeling of adsorption isotherm systems. Chem. Eng. J. (Amsterdam, Neth.), 2010. 156(1): p. 2-10. 233. Bahadori, A. and H.B. Vuthaluru, New method accurately predicts carbon dioxide equilibrium adsorption isotherms. Int. J. Greenhouse Gas Control, 2009. 3(6): p. 768- 772. 234. Bollini, P., et al., Dynamics of CO2 Adsorption on Amine Adsorbents. 1. Impact of Heat Effects. Ind. Eng. Chem. Res., 2012. 51(46): p. 15145-15152. 235. García, E.J., et al., Modeling Adsorption Properties on the Basis of Microscopic, Molecular, and Structural Descriptors for Nonpolar Adsorbents. Langmuir, 2013. 29(30): p. 9398-9409. 236. Builes, S. and L.F. Vega, Effect of Immobilized Amines on the Sorption Properties of Solid Materials: Impregnation versus Grafting. Langmuir, 2013. 29(Copyright (C) 2013 American Chemical Society (ACS). All Rights Reserved.): p. 199-206. 200 237. Jackson, P., et al., In situ Fourier Transform-Infrared (FT-IR) analysis of carbon dioxide absorption and desorption in amine solutions. Energy Procedia, 2009. 1(1): p. 985-994. 238. Sircar, S., R. Mohr, and M.B. Rao, Isosteric Heat of Adsorption: Theory and Experiment. The Journal of Physical Chemistry B, 1999. 103(31): p. 6539-6546. 239. Watabe, T. and K. Yogo, Isotherms and isosteric heats of adsorption for CO2 in amine-functionalized mesoporous silicas. Separation and Purification Technology, 2013. 120(0): p. 20-23. 240. Casas, N., et al., Fixed bed adsorption of CO2/H2 mixtures on activated carbon: experiments and modeling. Adsorption, 2012. 18(2): p. 143-161. 241. Schell, J., et al., Precombustion CO2 Capture by Pressure Swing Adsorption (PSA): Comparison of Laboratory PSA Experiments and Simulations. Ind. Eng. Chem. Res., 2013. 52(24): p. 8311-8322. 242. Posch, S. and M. Haider, Dynamic modeling of CO2 absorption from coal-fired power plants into an aqueous monoethanolamine solution. Chem. Eng. Res. Des., 2013. 91(6): p. 977-987. 243. Zhang, Y.-T., et al., Modeling of CO2 mass transport across a hollow fiber membrane reactor filled with immobilized enzyme. AIChE J., 2012. 58(7): p. 2069-2077. 244. Choi, J.H., et al., Modelling and analysis of pre-combustion CO2 capture with membranes. Korean J. Chem. Eng., 2013. 30(6): p. 1187-1194. 245. Wilfong, W.C. and S.S.C. Chuang, Probing the Adsorption/Desorption of CO2 on Amine Sorbents by Transient Infrared Studies of Adsorbed CO2 and C6H6. Industrial & Engineering Chemistry Research, 2014. 246. Khanna, R.K. and M.H. Moore, Carbamic acid: molecular structure and IR spectra. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 1999. 55(5): p. 961-967. 247. Wang, T., K.S. Lackner, and A. Wright, Moisture Swing Sorbent for Carbon Dioxide Capture from Ambient Air. Environ. Sci. Technol., 2011. 45(15): p. 6670-6675. 248. Mulgundmath, V.P., et al., Fixed bed adsorption for the removal of carbon dioxide from nitrogen: Breakthrough behaviour and modelling for heat and mass transfer. Separation and Purification Technology, 2012. 85(0): p. 17-27. 249. Lu, Q., et al., Catalytic Upgrading of Biomass Fast Pyrolysis Vapors with Pd/SBA-15 Catalysts. Ind. Eng. Chem. Res., 2010. 49(6): p. 2573-2580. 250. Chary, K. and C. Srikanth, Selective Hydrogenation of Nitrobenzene to Aniline over Ru/SBA-15 Catalysts. Catal. Lett., 2009. 128(1): p. 164-170. 201 251. Sujandi, E.A. Prasetyanto, and S.-E. Park, Synthesis of short-channeled amino- functionalized SBA-15 and its beneficial applications in base-catalyzed reactions. Appl. Catal., A, 2008. 350(2): p. 244-251. 252. Lu, Q., et al., Analytical pyrolysis-gas chromatography/mass spectrometry (Py- GC/MS) of sawdust with Al/SBA-15 catalysts. J. Anal. Appl. Pyrolysis, 2009. 84(2): p. 131-138. 253. Shahbazi, A., H. Younesi, and A. Badiei, Functionalized SBA-15 mesoporous silica by melamine-based dendrimer amines for adsorptive characteristics of Pb(II), Cu(II) and Cd(II) heavy metal ions in batch and fixed bed column. Chem. Eng. J. (Amsterdam, Neth.), 2011. 168(2): p. 505-518. 254. Zhao, Y., et al., Effective NH2-grafting on mesoporous SBA-15 surface for adsorption of heavy metal ions. Mater. Lett., 2011. 65(6): p. 1045-1047. 255. Wang, X., et al., Infrared Study of CO2 Sorption over "Molecular Basket" Sorbent Consisting of Polyethylenimine-Modified Mesoporous Molecular Sieve. J. Phys. Chem. C, 2009. 113(17): p. 7260-7268. 256. Ebrahimzadeh, H., et al., Extraction of trace amounts of silver on various amino- functionalized nanoporous silicas in real samples. Microchim. Acta, 2010. 170(1-2): p. 171-178. 257. Bridgwater, A.V., D. Meier, and D. Radlein, An overview of fast pyrolysis of biomass. Org. Geochem., 1999. 30(12): p. 1479-1493. 258. Mohan, D., C.U. Pittman, Jr., and P.H. Steele, Pyrolysis of Wood/Biomass for Bio-oil: A Critical Review. Energy Fuels, 2006. 20(3): p. 848-889. 259. Wildschut, J., I. Melian-Cabrera, and H.J. Heeres, Catalyst studies on the hydrotreatment of fast pyrolysis oil. Appl. Catal., B, 2010. 99(1-2): p. 298-306. 260. Hu, H., et al., Polyethyleneimine Functionalized Single-Walled Carbon Nanotubes as a Substrate for Neuronal Growth. J. Phys. Chem. B, 2005. 109(10): p. 4285-4289. 261. Lopez, J., et al., Hydrogen Transfer Reduction of 4-tert-Butylcyclohexanone and Aldol Condensation of Benzaldehyde with Acetophenone on Basic Solids. J. Catal., 2002. 208(1): p. 30-37. 262. Motokura, K., M. Tada, and Y. Iwasawa, Layered Materials with Coexisting Acidic and Basic Sites for Catalytic One-Pot Reaction Sequences. J. Am. Chem. Soc., 2009. 131(23): p. 7944-7945. 263. Parida, K.M. and D. Rath, Amine functionalized MCM-41: An active and reusable catalyst for Knoevenagel condensation reaction. J. Mol. Catal. A Chem., 2009. 310(1-2): p. 93-100. 202 264. Yu, X., et al., The effect of the distance between acidic site and basic site immobilized on mesoporous solid on the activity in catalyzing aldol condensation. J. Solid State Chem., 2011. 184(2): p. 289-295. 265. Inaki, Y., et al., New basic mesoporous silica catalyst obtained by ammonia grafting. Chemical Communications, 2001(22): p. 2358-2359. 266. Cherkasov, A.R., M. Jonsson, and V. Galkin, A novel approach to the analysis of substituent effects: quantitative description of ionization energies and gas basicity of amines. Journal of Molecular Graphics and Modelling, 1999. 17(1): p. 28-42. 267. Srivastava, R., An efficient, eco-friendly process for aldol and Michael reactions of trimethylsilyl enolate over organic base-functionalized SBA-15 catalysts. J. Mol. Catal. A Chem., 2007. 264(1-2): p. 146-152. 268. McKittrick, M.W. and C.W. Jones, Toward Single-Site Functional Materials- Preparation of Amine-Functionalized Surfaces Exhibiting Site-Isolated Behavior. Chem. Mater., 2003. 15(5): p. 1132-1139. 269. Nee, M., et al., A nitrogen-15 nuclear magnetic resonance study of the base- catalyzed NH2 exchange reactions of acetamide and thioacetamide. Org. Magn. Reson., 1982. 18(3): p. 125-7. 270. Fang, J., et al., One-Step Synthesis of Bifunctional TiO2 Catalysts and Their Photocatalytic Activity. J. Phys. Chem. C, 2010. 114(17): p. 7940-7948. 271. Xu, J., et al., Production of hydrocarbon fuels from pyrolysis of soybean oils using a basic catalyst. Bioresour. Technol., 2010. 101(24): p. 9803-9806. 272. Ahmed, A., et al., Porous silica spheres in macroporous structures and on nanofibres. Philos. Trans. R. Soc., A, 2010. 368(1927): p. 4351-4370. 273. Robert M. Silverstein, F.X.W., David J. Kiemle, Spectrometric Identification of Organic Compounds. Seventh Edition ed. 2005: John Wiley & Sons. 274. DiCosimo, R., et al., Industrial use of immobilized enzymes. Chemical Society Reviews, 2013. 42(15): p. 6437-6474. 275. Datta, S., L.R. Christena, and Y. Rajaram, Enzyme immobilization: an overview on techniques and support materials. 3 Biotech, 2013. 3(1): p. 1-9. 276. Liu, W., L. Wang, and R. Jiang, Specific Enzyme Immobilization Approaches and Their Application with Nanomaterials. Topics in Catalysis, 2012. 55(16-18): p. 1146- 1156. 277. Bankar, S.B., et al., Glucose oxidase — An overview. Biotechnology Advances, 2009. 27(4): p. 489-501. 203 278. Shao, Q., et al., Graphene oxide-induced conformation changes of glucose oxidase studied by infrared spectroscopy. Colloids and Surfaces B: Biointerfaces, 2013. 109(0): p. 115-120. 279. Schultz, N., et al., Zeta potential measurement as a diagnostic tool in enzyme immobilisation. Colloids Surf., B, 2008. 66(1): p. 39-44. 280. André, J., Z. Borneman, and M. Wessling, Enzymatic Conversion in Ion- Exchange Mixed Matrix Hollow Fiber Membranes. Industrial & Engineering Chemistry Research, 2013. 52(26): p. 8635-8644. 281. Bellusci, M., et al., Lipase Immobilization on Differently Functionalized Vinyl- Based Amphiphilic Polymers: Influence of Phase Segregation on the Enzyme Hydrolytic Activity. Biomacromolecules, 2012. 13(3): p. 805-813. 282. Guiomar, A.J., J.T. Guthrie, and S.D. Evans, Use of Mixed Self-Assembled Monolayers in a Study of the Effect of the Microenvironment on Immobilized Glucose Oxidase. Langmuir, 1999. 15(4): p. 1198-1207. 283. Carbone, K., M. Casarci, and M. Varrone, Crosslinked poly(vinyl alcohol) supports for the immobilization of a lipolytic enzyme. J. Appl. Polym. Sci., 1999. 74(8): p. 1881-1889. 284. Gao, F. and G. Ma, Effects of Microenvironment on Supported Enzymes. Topics in Catalysis, 2012. 55(16-18): p. 1114-1123. 285. Gaffney, D., J. Cooney, and E. Magner, Modification of Mesoporous Silicates for Immobilization of Enzymes. Topics in Catalysis, 2012. 55(16-18): p. 1101-1106. 286. Xiao, C., P. Ma, and N. Geng, Multi-responsive methylcellulose/Fe-alginate-g- PVA/PVA/Fe3O4 microgels for immobilizing enzyme. Polym. Adv. Technol., 2011. 22(12): p. 2649-2652. 287. Wang, J., et al., Facile Fabrication of Gold Nanoparticles-Poly(vinyl alcohol) Electrospun Water-Stable Nanofibrous Mats: Efficient Substrate Materials for Biosensors. ACS Appl. Mater. Interfaces, 2012. 4(4): p. 1963-1971. 288. Zhang, N., J. Xie, and V.K. Varadan, Soluble functionalized carbon nanotube/poly(vinyl alcohol) nanocomposite as the electrode for glucose sensing. Smart Mater. Struct., 2006. 15(1): p. 123-128. 289. Lozinsky, V.I., A.L. Zubov, and E.F. Titova, Poly(vinyl alcohol) cryogels employed as matrices for cell immobilization. 2. Entrapped cells resemble porous fillers in their effects on the properties of PVA-cryogel carrier. Enzyme and Microbial Technology, 1997. 20(3): p. 182-190. 204 290. Fejerskov, B., et al., Engineering Surface Adhered Poly(vinyl alcohol) Physical Hydrogels as Enzymatic Microreactors. ACS Appl. Mater. Interfaces, 2012. 4(9): p. 4981-4990. 291. Ren, G., et al., Electrospun poly(vinyl alcohol)/glucose oxidase biocomposite membranes for biosensor applications. React. Funct. Polym., 2006. 66(12): p. 1559- 1564. 292. Tran, D. and K. Balkus, Jr., Enzyme Immobilization via Electrospinning. Topics in Catalysis, 2012. 55(16-18): p. 1057-1069. 293. De, Q.A.A.A., et al., Alginate-poly(vinyl alcohol) core-shell microspheres for lipase immobilization. J. Appl. Polym. Sci., 2006. 102(2): p. 1553-1560. 294. Wong, F.-L. and A. Abdul-Aziz, Comparative study of poly(vinyl alcohol)-based support materials for the immobilization of glucose oxidase. J. Chem. Technol. Biotechnol., 2008. 83(1): p. 41-46. 295. Bhattacharya, M. and S. Chaudhry, High-performance silica nanoparticle reinforced poly (vinyl alcohol) as templates for bioactive nanocomposites. Materials Science and Engineering: C, 2013(0). 296. Ha, S., Y. Wee, and J. Kim, Nanobiocatalysis for Enzymatic Biofuel Cells. Topics in Catalysis, 2012. 55(16-18): p. 1181-1200. 297. Rayalu, S., et al., Nanobiocatalysts for Carbon Capture, Sequestration and Valorisation. Topics in Catalysis, 2012. 55(16-18): p. 1217-1230. 298. Silva M, E., S. Chakravartula S, and S.C. Chuang, Silica-Supported Amine Catalysts for Carbon–Carbon Addition Reactions. Topics in Catalysis, 2012. 55(7-10): p. 580-586. 299. Lott, J.A. and K. Turner, Evaluation of Trinder's Glucose Oxidase Method for Measuring Glucose in Serum and Urine. Clinical Chemistry, 1975. 21(12): p. 1754-1760. 300. Ghosh Datta, S., et al., DNA template-assisted modulation of horseradish peroxidase activity. International Journal of Biological Macromolecules, 2012. 50(3): p. 552-557. 301. Sharma, K.P., et al., Assembly of Polyethyleneimine in the Hexagonal Mesophase of Nonionic Surfactant: Effect of pH and Temperature. The Journal of Physical Chemistry B, 2011. 115(29): p. 9059-9069. 302. Bessbousse, H., et al., Removal of heavy metal ions from aqueous solutions by filtration with a novel complexing membrane containing poly(ethyleneimine) in a poly(vinyl alcohol) matrix. Journal of Membrane Science, 2008. 307(2): p. 249-259. 205 303. Ge, L., et al., Immobilization of glucose oxidase in electrospun nanofibrous membranes for food preservation. Food Control, 2012. 26(1): p. 188-193. 304. Seehuber, A. and R. Dahint, Conformation and Activity of Glucose Oxidase on Homogeneously Coated and Nanostructured Surfaces. J. Phys. Chem. B, 2013. 117(23): p. 6980-6989. 305. Khodadadei, F., et al., Rapid and clean amine functionalization of carbon nanotubes in a dielectric barrier discharge reactor for biosensor development. Electrochimica Acta, 2014. 115(0): p. 378-385. 306. Mukhopadhyay, K., et al., Gold Nanoparticles Assembled on Amine- Functionalized Na−Y Zeolite: A Biocompatible Surface for Enzyme Immobilization. Langmuir, 2003. 19(9): p. 3858-3863. 307. Oh, C., et al., New approach to the immobilization of glucose oxidase on non- porous silica microspheres functionalized by (3-aminopropyl)trimethoxysilane (APTMS). Colloids and Surfaces B: Biointerfaces, 2006. 53(2): p. 225-232. 308. Haouz, A., et al., Dynamic and structural properties of glucose oxidase enzyme. European Biophysics Journal, 1998. 27(1): p. 19-25. 206 APPENDIX 11GAS-SOLID FLUIDIZATION Abstract Gas-solid fluidization is a unit operation widely used in industry and industrial research due to its advantages of high heat and mass transfer coefficients, and great flexibility. This appendix shows a review on the basic principles of the gas-solid fluidization, and presents a general algorithm to determine some of the important parameters in fluidized bed columns sizing, such as minimum fluidization velocity, terminal velocity, bed height and column diameter. Introduction As one of the most important industrial applications of the transport phenomena, the gas-solid fluidization is an operation that takes place between two extremes; the solids packed bed and the pneumatic transport of solids. Under fluidization a bed of solid particles is changed to a fluid-like state by the passage of a gas at certain velocity through the bed. Typical applications occur in adsorption/desorption technologies; in the drying of solids particles (like grains, sand, agglomerates and food); in heat treatment of solids (for example catalyst regeneration, calcinations, etc); and in both catalytic and non- catalytic reactions [1]. 207 Five regimes of fluidization may be identified according to the velocity of the gas [1, 6]; (1) particulate fluidization, at the minimum fluidization velocity; (2) aggregative or bubbling fluidization; (3) slugging regime; (4) turbulent regime and (5) fast fluidization at the terminal velocity. Particle diameter, sphericity, and particle density are important properties that must be known for the solid particles; while void fraction and bulk density must be known for the solids bulk and may be predicted after fluidization [2, 5]. Heat and mass transfer are also present in the gas-solid fluidization; they are very high in most of the cases because of the very high surface area exposed by the solid and the flow regime of the gas. Heat transfer may occur primarily by convection (between the flowing gas and the solid particle) followed by conduction and radiation (between particles and/or particles and tube wall); mass transfer occurs by diffusion (inside of the particles and/or in the lean and dense regions) and by convection between the gas and the solid surfaces [2, 4, 5]. Several correlations have been proposed in order to determine many dimensionless numbers, useful to predict coefficients of fluidization, heat and mass transfer [1]. Hydrodynamics: Particles Fluidization [1, 2, 3, 6] Particulate fluidization: As long as the gas velocity increases and reaches Umf, the velocity of minimum fluidization, the solid particles start moving individually and randomly along the bed in a very softly manner allowing the bed to be homogeneous and behave as a fluid with well defined top surface. At this point no bubbles or aggregates are formed and a smooth expansion of the bed is observed. Particulate fluidization 208 occurs typically in a very narrow range of velocities and the bed expansion is often very low and for some systems is not observed Aggregate or Bubbling Fluidization: When gas velocity increases up to the minimum bubbling velocity, Umb, some aggregates or bubbles are formed in the bed, inside of these aggregates the solid concentration is nearly zero. The bubbles appear at the bottom of the bed and rise rapidly to the homogeneous top surface breaking trough like in a boiling liquid. The movement of the aggregates produces a more vigorous movement of the solid particles and some pressure fluctuations. Slugging regime: This regime is produced when the aggregate bubbles fill most of the column cross section, at this point the top surface of the bed rises and collapses periodically while pressure experiments large and frequently fluctuations. Most of the column designs avoid this regime, but when desired, the column must be tall and relatively small in diameter. Turbulent regime; The gas velocity reaches a transition point in which turbulence starts, many aggregates are present along the whole bed and some of them break down, yielding to a wide range of bubble sizes. The movement of the bed is vigorous and the top surface is not homogeneous any more, pressure fluctuations present small amplitude. Fast Fluidization: When gas velocity is equal or above Ut, the terminal velocity, the particles from the bed are transported all along the column and only return to the bottom very close to the walls or by the use of a cyclone. Fast fluidization yields to a very homogeneous solids concentration in the columns and often operates above 80 % of void volume. 209 Figure 1 shows a schematic of the fixed bed and the five regimes described. Figure 1: Regimes of fluidization identified with increasing gas superficial velocity. (a) represents the fixed bed and (b) to (f) represent the five fluidization regimes described above. From ref [6]. Fluidization also depends on the sizes, shapes and densities of the particles. Small particles (particle diameter; dp=30 µm - 150 µm) with low particle densities ( < .5g/cc) show an “abnormal bed behavior” (figure ) in which the bed expands appreciably before bubbling sets in, for them the minimum bubbling velocity is always greater than the minimum fluidization velocity. For bigger and more dense particles (dp>150 µm an > 1.5 g/cc) the minimum fluidization velocity and the minimum bubbling velocity are nearly the same and they show a “normal bed behavior”. Bulk characteristics are also important in fluidized bed systems, porosity or void fraction and bulk density are important variables that must be known for the fixed bed and may be calculated after fluidization as well as bed expansion and bed height. Figure 3 shows how to determine the minimum fluidized porosity ( ) for a wide variety of 210 solid particles, and equation 1 is a compliment of this chart when another solid is used. Equation 2 relates density and bed height with porosity between the minimum fluidization point and any other condition of the bed. Figure 3: Bed voidage at minimum fluidization. (Leva, 1959). Agarwal and Storrow: (a) soft brick; (b) absorption carbon; (c) broken Raschig rings; (d) coal and glass powder; (e) Figure 2: Progress of carborundum; (f) sand. U.S. Bureau of Mines: (g) pressure drop and bed height with round sand, ϕs=0:86; (h) sharp sand, ϕs= 0:67; (i) increasing velocity [6]. Fischer–Tropsch catalyst, ϕs=0:58; (j) anthracite coal, ϕs= 0:63; (k) mixed round sand, ϕs= 0:86. Van Heerden et al.: (l) coke; (m) carborundum [6]. Leva (1959) proposed one of the first applicable correlations with relatively high grade of confidence to be applied to predict the fluidized void volume, Grace, J. R presented equations 4 to 6 to calculate the minimum fluidization velocity [1] and Gledart and Abrhamsen (1978) presented a relation for the minimum bubbling velocity [6]. (1) 211 (2) represents the height of the bed at the minimum fluidization point. is the height, the void fraction and the density of the bed at any point. Equation 3 applies for . (3) (4) (5) Where (varies for different authors) and 0.0408, and gas properties must be used at working pressure and temperature. is the Arquimedes number, is the Reynolds number at minimum fluidization point, g is the acceleration of the gravity and is the viscosity of the gas. (6) (7) (8) Where is the terminal velocity (maximum allowed before pneumatic transport) and Cd is the drag coefficient and is calculated from the following correlations: 212 is the Reynolds number for the solid particles. The minimum pressure drop (Δ ) that the gas experiments should equal the value predicted by Bernoulli equation and can be calculated from [1]: Δ (9) However this equation may be calculated even further the point of minimum fluidization. Heat Transfer: The Convective Heat Transfer Coefficient [1, 2, 5] Heat transfer occurs by the three mechanisms; (1) conduction between the solid particles when they are in contact and between particles and wall, (2) convection between the solids surface and the gas and between the gas and cylinder walls, and (3) radiation between all the bodies present in the system when high temperatures are reached. Numerous correlations and algorithms have been proposed to calculate the heat transfer coefficients but only one simple approach is presented in this work, with the purpose of illustration. From Kunii and Levenspiel (1969) [1]: 213 (10) (11) (12) The convective coefficient (hc) is obtained from Nusselt number ( ). is Prandtl number, is the therma1 conductivity and is the specific heat capacity of the gas. The heat transfer in fluidized beds is relatively high due to the high surface area of the particles exposed and the high flow regime (turbulence) of the gas [2]. Mass Transfer: Sizing Of A Fluidized Bed Column [1, 2, 5] Most of the common applications of fluidized beds are drying, adsorption, agglomeration, and heterogeneous reactions; they require knowing the mass transfer coefficients. The mass transfer mechanisms diffusion and convection are observed during fluidization. There are regions in which aggregates (large void fraction) are formed and species present in the bubble must travel to a denser region or vice-versa; however in the solid-dense regions the mass transfer occurs between the bulk gas and the solid. Mass transfer coefficients have been correlated using the Sherwood number, , Wen and Fane presented correlations 13 to 15 for the prediction of Kfb (the mass transfer coefficient) [1]. 214 (13) (14) (15) is a modification of Reynolds number, in Smith number, D is the column diameter and z represents the bed height. is the fraction of active mass transport particles, and in the case there are no inert particles its value is 1. In the design of fluidized beds, the bed height is calculated according to the desired change of concentration in the gas (or in the solid, by applying a mass balance): The flux of species between the solid particles and the gas (NA) is given by: (16) is the concentration of species ‘A’ and is the saturation (maximum) concentration of ‘A’ in the gas. From a balance over the column A may be replaced by an expression that relates the gas concentration along the height of the bed [1]: (17) Where the volumetric surface factor is; , and ϕ is the sphericity of the particles. This is a separable ordinary differential equation that must be integrated under the following boundary conditions: 215 (18) From a direct integration, the concentration profile becomes: (19) And re arranging: (20) Where Note that if the ratio of Lfb to bed diameter is less than 10, axial dispersion should be considered. Using the minimum fluidization velocity and the fluidized bed height ( ; the total height and diameter of the column (D), may be calculated, as well as the gas flow needed for the operation. (21) Conclusions Gas-solid fluidization is an important unit operation and is very used in chemical and process engineering. Due to the high contact between the solids and the gas, the main applications include adsorption/desorption; drying; heterogeneous (catalytic) reactions and heat treatment. The most important parameters in gas-solid fluidization are the minimum fluidization velocity, terminal velocity, pressure drop, heat and mass 216 transfer coefficients. In the subjects presented in this dissertation related to CO2 capture, the main concepts and some calculations were shown to describe the operations in which transport phenomena concepts are applied. References [1] Robert S. Brodkey, arry . ershey, TRA SP RT P E ME A “A U IFIED APPR A ”, The Ohio State University, McGraw-Hill Book Company. [2] Wen-Ching Yang (edited by), FLUIDIZATION, SOLIDS HANDLING, AND PROCESSING, INDUSTRIAL APPLICATIONS, Siemens Westinghouse Power Corporation Pittsburgh, Pennsylvania. [3] L. G. Gibilaro, FLUIDIZATION-DY AMI S, University of L’Aquira, L’Aquira Italy. [4] Liang-Shin Fan and Chao Zhu, PRICIPLES OF GAS-SOLID FLOWS, Ohio State University, USA. [5] Wen-Ching Yang (edited by), HANDBOOK OF FLUIDIZATION AND FLUID- PARTICLE SYSTEMS, Siemens Westinghouse Power Corporation Pittsburgh, Pennsylvania, USA. [6] James R. Couper, W. Roy Penney, James R. Fair and Stanley M. Walas, CHEMICAL PROCESS EQUIPMENT, SELECTION AND DESIGN, Gulf Professional Publishing, Elsevier. 217