1 CO2-Selective Membranes for Fuel Cell H2 Purification

Home , Membrane gas separation

CO2-selective Membranes for Fuel Cell H2 Purification and Flue Gas CO2 Capture:

From Lab Scale to Field Testing

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

Witopo Salim

Graduate Program in Chemical Engineering

The Ohio State University

2018

Dissertation Committee

Dr. W.S. Winston Ho, Advisor

Dr. Bhavik R. Bakshi

Dr. Nicholas A. Brunelli

Copyrighted by

Witopo Salim

2018

Abstract

Membrane processes are attractive for gas separation applications, including H2 purification and CO2 capture. Advanced nanostructured polymer membranes based on facilitated transport mechanism showed potential in H2 purification for fuel cell applications due to its oxidative stability and ability to simultaneously remove CO2 and

H2S from a feed gas. In addition, the membranes could achieve high CO2 permeance and hence also suitable for CO2 capture from flue gas. In this work, CO2-selective membranes research development from lab-scale to field-test was conducted.

CO2-selective membranes comprised of a quaternaryammonium hydroxide mobile carrier, a quaternary ammonium fluoride fixed-site carrier, borate-based additives, and crosslinked polyvinylalcohol was developed. The optimized membrane composition containing tetrafluoroboric acid demonstrated oxidative stability with a CO2 permeance of

100 GPU and CO2/H2 selectivity > 100 for at least 144 hours at 120°C with humid air as the sweep gas. The membrane was scaled-up to fabricate 14-in wide flat sheet membranes with > 1400-ft long in total length. The scale-up membrane performed similarly compared to the lab-scale membranes and demonstrated the potential for membrane processes that use air as the sweep gas including the H2 purification in fuel cell applications.

Fabrication of CO2-selective membranes was scaled-up to prepare membranes with 14- in wide and > 150-ft long with a uniform selective-layer thickness of around 15 microns.

The membrane contained aminoisobutyric potassium salt and polyvinylamine as the carriers for facilitated transport of acid gases and crosslinked polyvinylalcohol as the

ii membrane matrix. The scale-up membranes demonstrated similar performances as the lab- scale membranes with a CO2 permeance > 200 GPU, CO2/H2 selectivity > 200, and H2S/H2 selectivity > 600 and were used for a field test of prototype spiral-wound membrane modules with autothermal reformate gas as the feed gas. The modeling and optimization of the H2 purification process were performed and showed that a retentate stream containing less than 10 ppm of H2S and greater than 99% of H2 recovery was obtained.

The field test results showed that the retentate H2S concentrations were in the range of 4 –

10 ppm on a wet basis, which agrees with the modeling results and showed the potential use for removal of CO2 and H2S from reformate gas for H2 purification for fuel cells.

A new design of spiral-wound membrane module comprising face compression “O” rings was developed. An inorganic/polymer composite membrane, consisting of a selective amine-containing polymer cover layer / a zeolite nanoparticle layer / a polymer support was used to prepare spiral-wound membrane elements. The concentration polarization phenomenon was observed at a dry feed gas flow rate < 1000 sccm. The spiral-wound membrane modules showed similar performance results with a CO2 permeance of greater than 800 GPU and a CO2/N2 selectivity of more than 140. The pressure drops of the spiral- wound membrane modules were less than 1.5 psi/m. Similar results were obtained during the field test and some unexpected issues encountered during the field test were resolved by improving the spiral-wound membrane module fabrication using a smoother feed spacer and a longer glue curing time. The field test results of the spiral-wound membrane modules at the National Carbon Capture Center have shown the potential for the post-combustion

CO2 capture from coal-fired power plants.

iii

Dedication

Dedicated to my parents, brother, sisters, and Indah

Acknowledgments

Firstly, I would like to thank my advisor Dr. W.S. Winston Ho for his guidance, encouragement, support, and life lessons throughout my PhD study. His dedication, enthusiasm, and focus on continuous improvement is a role model for inspiration. It was truly an honor to be his student.

I am thankful to Dr. Andre Palmer, Dr. Nicholas A. Brunelli, and Dr. Bhavik R.

Bakshi, who served as my qualifying exam, candidacy exam, or final defense committee members and provided their valuable insights and advices of my research work.

I would also like to acknowledge my colleagues in Dr. Ho’s research group, including

Dr. Kartik Ramasubramanian, Dr. Yanan Zhao, Dr. Lin Zhao, Dr. Yuanxin Chen, Dr.

Varun Vakharia, Dr. Zi Tong, Mr. Dongzhu Wu, Mr. Yang Han, Mr. Kai Chen, and Ms.

Tingyu Chen. It was my pleasure to work together with all of you to accomplished challenging research project objectives. I am fortunate to meet visiting scholars and professors including Dr. Norman Loney, Dr. Yue Wang, Dr. Hang Dong, Dr. Ruizhi Pang,

Dr. Luca Ansaloni, and others.

I acknowledge the faculty, staff, graduate students, and friends at William G. Lowrie

Department of Chemical and Biomolecular Engineering for their assistance during my study. The support from Mr. Michael A. Wilson, Mr. Leigh Evrard, and Mr. David Cade in my research work are very well appreciated.

I acknowledge the financial support of my research work by the United States

Department of Energy, Ohio Development and Services Agency, Bloom Energy

Corporation, Office of Naval Research, and The Ohio State University.

Finally, I would like to thank my parents, brother, and sisters for their endless love, support, and encouragement.

Vita

1983 ...…………………...... Born in Jakarta, Indonesia

2001 ……...……………...... B. Chem. Eng., Institute Teknologi Bandung,

Bandung, Indonesia

2005 ...…………………...... M. Sci. Chem. Eng., National Taiwan University

of Science and Technology, Taipei, Taiwan R.O.C.

2007 ...…………………...... Project Engineer, CTCI Corporation, Taipei,

Taiwan R.O.C.

2012 to present ...………………... Graduate Research Associate, Department of

Chemical and Biomolecular Engineering, The

Ohio State University, Columbus, Ohio, USA

vii

Publications

1. W. Salim, W.S.W. Ho, Recent developments on nanostructured polymer-based

membranes, Curr. Opinion Chem. Eng. 8 (2015) 76–82.

2. D. Wu, L. Zhao, V. K. Vakharia, W. Salim, W.S.W. Ho, Synthesis and characterization

of nanoporous polyethersulfone membrane as support for composite membrane in CO2

separation: From lab to pilot scale, J. Membr. Sci. 510 (2016) 58–71.

3. W.S.W. Ho, V. Vakharia, W. Salim, Borate-containing membranes for gas separations,

U.S. Provisional Patent Application No. 62416434 (filed Nov. 2, 2016).

4. W.S.W. Ho, W. Salim, V. Vakharia, Membranes for gas separation, U.S. Utility Patent

No. 20170056839 A1 (March 2, 2017).

5. W.S.W. Ho, W. Salim, V. Vakharia, Spiral-wound membrane module for gas

separations, U.S. Provisional Patent Application No. 62557477 (filed Sep. 12, 2017).

6. V. Vakharia, W. Salim, M. Gasda, W.S.W. Ho, Oxidatively stable membranes for CO2

separation and H2 purification, J. Membr. Sci. 533 (2017) 220–228.

Fields of Study

Major Field: Chemical Engineering

viii

Table of Contents

Abstract ...... ii Dedication ...... iv Acknowledgments...... v Vita ...... vii List of Tables ...... xii List of Figures ...... xiii

Chapter 1. Introduction and Literature Review ...... 1

1.1. Membrane processes for gas separation ...... 1 1.2. Polymer membranes for gas separation ...... 5 1.3. Facilitated transport polymer membranes ...... 11 1.4. Scope and outline ...... 13

Chapter 2. Oxidatively Stable Borate-containing Membranes for H2 Purification for Fuel Cell ...... 22

2.1. Summary ...... 22 2.2. Introduction ...... 23 2.3. Experimental ...... 26 2.3.1. Materials ...... 26 2.3.2. Membrane preparation ...... 27 2.3.3. Gas transport measurement ...... 29 2.3.4. Scale-up membrane fabrication ...... 31 2.4. Results and discussion ...... 32 2.4.1. Effect of borate-based catalyst on membrane performance ...... 32 2.4.2. Effect of tetrafluoroborate-based catalyst on membrane performance ...... 36 2.4.3. Effect of membrane thickness ...... 39 2.4.4. Membrane stability with air as sweep gas ...... 41 2.4.5. Membrane composition optimization with TFBA as catalyst ...... 42 2.4.6. Effect of substrates on water permeation of borate-containing membranes ... 43 2.4.7. Scale-up membrane fabrication ...... 44 ix

2.5. Future directions ...... 46 2.6. Conclusions ...... 47

Chapter 3. Scale-up of Amine-containing Membranes for H2 Purification for Fuel Cells ...... 54

3.1. Summary ...... 54 3.2. Introduction ...... 55 3.3. Experimental ...... 59 3.3.1. Materials ...... 59 3.3.2. Pilot-scale membrane fabrication ...... 60 3.3.3. Membrane gas transport measurement ...... 65 3.4. Results and discussion ...... 68 3.4.1. Scale-up membrane fabrication parameters optimization...... 68 3.4.1.1. Effect of coating solution on membrane thickness ...... 69 3.4.1.2. Effect of coating speed on membrane thickness ...... 72 3.4.1.3. Effect of coating knife gap setting on membrane thickness ...... 73 3.4.2. Scale-up membrane fabrication and test at OSU ...... 74 3.4.3. Membane module fabrication ...... 77 3.4.4. Modeling and process optimization ...... 79 3.4.4.1. Effect of sweep-to-feed flow rate ratio and sweep water content ...... 80 3.4.4.2. Effect of feed water content ...... 84 3.4.4.3. Design operating conditions for membrane module testing ...... 86 3.4.4.4. Comparison between modeling results and membrane module field test results ...... 88 3.5. Future directions ...... 91 3.6. Conclusions ...... 92

Chapter 4. Fabrication and Field Testing of Spiral-wound Membrane Modules for CO2 Capture from Flue Gas ...... 100

4.1. Summary ...... 100 4.2. Introduction ...... 101 4.3. Experimental ...... 105

4.3.1. Materials ...... 105 4.3.2. Spiral-wound membrane element fabrication ...... 106 4.3.3. Spiral-wound membrane module: new design ...... 107 4.3.4. Spiral-wound membrane module test ...... 110 4.4. Results and discussion ...... 117 4.4.1. Effect of feed gas flow rate on membrane module performance ...... 117 4.4.2. Module performance and pressure drop results at > 1000 sccm feed gas flowrates ...... 119 4.4.3. Module performance and pressure drop results at 1000 sccm feed gas flowrates and performance comparison with flat-sheet membranes...... 120 4.4.4. Module performance and pressure drop results at 1000 sccm feed gas with SO2 and O2 ...... 123 4.4.5. Membrane module field test results with actual flue gas at NCCC ...... 126 4.5. Future directions ...... 132 4.6. Conclusions ...... 133

Chapter 5. Final Remarks and Recommendations ...... 142

5.1. Final remarks ...... 142 5.2. Recommendations ...... 144

Bibliography ...... 147 Appendix A. Design of experiment for membrane composition optimization with TFBA as catalyst ...... 158 Appendix B. Modeling results of membrane module performance for a feed pressure of 5 psig ...... 163 Appendix C. Modeling for spiral-wound membrane module ...... 166

List of Tables

Table 1.1. Comparison of polymeric and inorganic membranes for H2 purification ...... 5

Table 2.1. Gas transport performances of the membranes with BA as the catalyst ...... 33

Table 2.2. Gas transport performances of the membranes with TFBA as the catalyst ..... 36

Table 2.3. Performances of membranes containing different borate- and tetrafluoroborate- based catalysts ...... 38

Table 2.4. The optimized membrane compositions and transport performances of TFBA- containing membranes ...... 42

Table 2.5. Effects of additional substrates on water permeation in TFBA-containing membranes ...... 44

Table 2.6. The performance comparison of the lab-scale and scale-up quaternaryammonium hydroxide- and fluoride-containing membranes with TFBA as the catalyst ...... 45

Table 3.1. Five runs of the prototype membrane (14 inches) fabricated by using the continuous membrane fabrication machine at OSU ...... 75

Table 3.2. Cases considered in the modeling study of effect of temperature on membrane performance ...... 89

Table 4.1. Gas transport performances and pressure drops of spiral-wound membrane modules at different feed gas flow rates ...... 120

Table 4.2. Gas transport performances and pressure drops of spiral-wound membrane modules at the optimized feed gas flow rate ...... 122

Table 4.3. Gas transport performances of spiral-wound membrane elements at 57oC tested at OSU ...... 124

Table 4.4. Gas transport performances of spiral-wound membrane elements at 57oC tested at NCCC ...... 127

Table A.1. The experimental design membrane compositions and transport performances of TFBA-containing membranes ...... 159

xii

List of Figures

Figure 1.1. Simplified diagram of the three approaches of carbon capture ...... 1

Figure 1.2. Schematic of membrane process configuration: (a) vacuum permeate (b) sweep gas ...... 4

Figure 1.3. Schematics of nanostructured membranes: (a) dense membrane incorporated with nanomaterials, i.e., nanoparticles (b) porous membrane with nanoscale pores, (c) dense membrane on a porous membrane with nanoscale pores (d) dense membrane incorporated with nanomaterials on a porous membrane with nanoscale pores (e) inorganic-organic multiple layer composite membrane...... 7

Figure 1.4. Schematic of solution-diffusion and facilitated transport mechanism ...... 12

Figure 2.1. Schematic of the gas permeation unit for gas transport measurements ...... 31

Figure 2.2. Effect of different borate-based catalysts and concentrations on the membrane gas transport performances: (a) CO2 permeance (b) CO2/H2 selectivity ...... 34

Figure 2.3. Schematic representations of the molecular interactions for (a) borate-based compound and (b) tetrafluoroborate-based compound with a H2O molecule for the H2O- CO2 reaction ...... 35

Figure 2.4. Effect of selective-layer membrane thickness on transport performances: (a) CO2 permeance (b) CO2/H2 selectivity ...... 40

Figure 2.5. The stability plot of the membrane containing TFBA as catalyst with air as the sweep gas ...... 41

Figure 2.6. The scale-up membrane fabricated by using the roll-to-roll fabrication machine: (a) wet film membrane before curing (b) dry membrane after curing ...... 46

Figure 3.1. The schematic of the structure of scale-up membranes for H2 purification ... 61

Figure 3.2. Schematic representation of pilot-scale thin-film-coating (TFC) assembly of the coating machine for the scale-up fabrication of membranes for H2 purification ...... 63

Figure 3.3. The pilot-scale coating machine for the scale-up fabrication of membranes: (a) machine setup at OSU and (b) the schematic of coating knife assembly ...... 64

Figure 3.4. Schematic representation of the gas permeation unit for the membrane gas transport measurements ...... 68

xiii

Figure 3.5. Effect of small amine molecules (TETA or EDA) on the increasing rate coating solution viscosity ...... 70

Figure 3.6. Effect of coating solution concentration on selective-layer membrane thickness ...... 71

Figure 3.7. Effect of substrate web coating speed on selective-layer membrane thickness ...... 73

Figure 3.8. Effect of coating knife gap setting on selective-layer membrane thickness. . 74

Figure 3.9. The images of the scale-up membrane before (a) and after (b) the curing in oven ...... 75

Figure 3.10. The stability plot of membrane sample taken from scale-up run No. 5 ...... 77

Figure 3.11. The image of the spiral-wound membrane module that contained the CO2- selective membrane developed in this work ...... 78

Figure 3.12. Modeling results of module performance for a feed pressure of 10 psig: (a) CO2 removal (b) H2S removal (c) H2 recovery (d) retentate H2S concentration on wet basis (e) retentate water content ...... 81

Figure 3.13. Modeling results of module performance with 40% water contents in both the feed and sweep gases: (a) 5 psig feed pressure (b) 10 psig feed pressure ...... 85

Figure 3.14. Modeling results of module performance with 12.7% and 40% water contents in the feed and sweep gases: (a) 5 psig feed pressure (b) 10 psig feed pressure ...... 87

Figure 3.15. The H2S concentration in the retentate side obtained for Cases 1 – 5 in Table 3.2...... 91

Figure 4.1. The step-by-step procedure of the spiral-wound membrane element fabrication ...... 107

Figure 4.2. Spiral-wound membrane element fabricated ...... 108

Figure 4.3. The spiral-wound membrane element placed inside and glued to the Plexiglas FRP ...... 109

Figure 4.4. The face-compression “O” ring design of the spiral-wound membrane element with the Plexiglas FRP ...... 109

xiv

Figure 4.5. Spiral-wound membrane module consisting of the membrane element inside a housing ...... 110

Figure 4.6. The leak test assembly for the spiral-wound membrane module ...... 112

Figure 4.7. Schematic of the gas permeation unit for membrane module gas transport measurements ...... 114

Figure 4.8. The setup of the gas permeation unit installed inside the analytical lab of NCCC ...... 116

Figure 4.9. A spiral-wound membrane module placed inside the oven of the gas permeation unit for gas transport measurements ...... 116

Figure 4.10. The effect of feed gas flow rate on the CO2 permeance of the spiral-wound membrane module (concentration polarization phenomenon) ...... 118

Figure 4.11. Transport performances of scale-up flat-sheet membranes and spiral-wound membrane modules at 57oC ...... 123

Figure 4.12. The stability plots of the spiral-wound membrane modules SW-16 tested at OSU and SW-19 tested at NCCC ...... 125

Figure 4.13. The stability plots of the spiral-wound membrane modules SW-17 tested at OSU and SW-20 tested at NCCC (both models with a longer glue curing time of 48 hours) ...... 126

Figure 4.14. The stability plot of the spiral-wound membrane module SW-18 tested at NCCC ...... 128

Figure 4.15. The images of the membrane before element rolling showing a smooth membrane surface and after the module testing with the indentations caused by the feed spacer ...... 129

Figure 4.16. The color change of epoxy glue over testing time at NCCC (a) membrane element prior to testing (b) SW-18 after 96 hours of test (c) SW-20 after 208 hours of test ...... 130

Figure 4.17. The stability plot of the improved spiral-wound membrane module tested with 3% O2 and 3 ppm SO2...... 131

Figure A.1. The JMP analysis results (analysis of variance, effect tests, prediction expression, and profiler) for response variables: (a) CO2 permeance (b) CO2/H2 selectivity ...... 161 xv

Figure B.1. Modeling results of module performance for a feed pressure of 5 psig: (a) CO2 removal (b) H2S removal (c) H2 removal (d) retentate H2S concentration on wet basis (e) retentate water content ...... 163

Figure C.1. Module configuration and flow directions in the crossflow module ...... 166

xvi

Chapter 1. Introduction and Literature Review

1.1. Membrane processes for gas separation

Industrial applications of gas separation include natural gas treatment, air separation, hydrogen purification, carbon capture, etc. Natural gas treatment is a process to clean raw natural gas by separating impurities (CO2, H2S) and various non-methane hydrocarbons (ethane, propane, butane, O2, N2), to produce pipeline quality dry natural gas. Air separation is a process to remove certain components in the air to produce either

N2-enriched air, O2-enriched air, or dehydrated air. Hydrogen purification is a process to produce high purity H2 gas stream from the products of the water gas shift reaction that mainly contains CO2 and H2. Carbon capture is a process to remove CO2 from a CO2 emission source and can be followed by sequestration of the collected CO2. There are three approaches of carbon capture including pre-combustion, oxy-combustion, and post- combustion. A simplified diagram of the carbon capture approaches is shown in Figure 1.1.

Pre -combustion CO2 to Fuel (CO2, H2) storage Combustion CO2 Capture Air/O2

Oxy -combustion CO2 to storage Fuel Combustion O2

Post -combustion Flue gas CO2 to Fuel storage Combustion CO2 Capture Air (CO2, N2)

Figure 1.1. Simplified diagram of the three approaches of carbon capture. 1

For hydrogen purification from the products of the water gas shift reaction or syngas, separation of CO2 from H2 is necessary. For carbon capture, CO2 is separated from either

H2 (in pre-combustion) or N2 (in post-combustion). There are three major technologies for

CO2 separation including absorption, adsorption, and membrane. Other technologies also include a hybrid process such as membrane reactors for water gas shift reaction to achieve higher H2 recovery with high purity.

In absorption process by using amines such as monoethanolamine (MEA), the main issues encountered are the solvent loss and the required energy for the solvent regeneration.

However, so far it is the most well-established process and a CO2 recovery of 98% with a purity of 99% can be achieved [1]. Physical absorption processes such as Rectisol® and

® Selexol have also been used for H2 purification from syngas.

In adsorption process (either pressure swing adsorption – PSA or temperature swing adsorption – TSA), the main issues include the limited capacity and the low CO2 selectivity of the existing adsorbents such as zeolites and activated carbon. The main advantage of physical adsorption over chemical or physical absorption is its simple and energy efficient operation and regeneration, which can be achieved with a pressure swing or temperature swing cycle.

Compared to these technologies, the membrane process has the following advantages:

• It does not need any additional energy penalty for the regeneration step.

• Membrane system is compact, lightweight, and can be positioned either horizontally or

vertically which makes it suitable for retrofitting applications.

• Modular design of membrane system allows the process optimization by using multi-

stage and recycling operation.

• No moving parts in the membrane unit and hence low maintenance requirements.

In a membrane process, the driving force for the separation of gases is generated by either using vacuum in permeate side or applying sweep gas. Figure 1.2 shows an example for acid gases removal from a steam reformate gas. In the vacuum permeate case, the CO2 and H2S is separated from the steam reformate gas and permeate through the membrane due to the vacuum at the permeate side. In the sweep gas case, the CO2 and H2S is separated from the steam reformate gas and permeate through the membrane due to the partial pressure difference between feed and sweep side created by the sweep gas.

The performances of a membrane (permeability Pi and selectivity αij) are defined as:

Ji yi / y j Pi  ij  pi /  xi / x j

( p  p )  ( p  p ) i, feed in i,sweep out i, feed out i,sweep in pi  ln( pi, feed in  pi,sweep out )  ln( pi, feed out  pi,sweep in) where i denotes the gas component CO2 or H2S and j denotes another gas component (H2,

CH4, CO, or N2); y and x are the mole fractions of each gas component in the sweep and feed sides of the membrane, respectively; Ji is the steady-state CO2 molar flux across the selective-layer;  is the selective-layer thickness; and Δpi is the partial pressure difference between the feed and sweep sides (driving force) for the permeation process, which is determined using the logarithmic mean method. The ratio of permeability to the selective- layer thickness (Pi /  ) is referred as the permeance and its common unit is the gas permeation unit (GPU), which is equal to 10-6 cm3 (STP) / (cm2 s cmHg). 3

Steam reformed gas CO , H S CO , H S (CO , H , H S) 2 2 2 2 High purity H2 2 2 2 Feed/retentate

Membrane Membrane thickness

Permeate

CO2 + H2S

Vacuum

(a)

Steam reformed gas CO2, H2S CO2, H2S High purity H2 (CO2, H2, H2S) Feed/retentate

Membrane Membrane thickness

Sweep/permeate Sweep Gas + CO + H S Sweep Gas 2 2

(b)

Figure 1.2. Schematic of membrane process configuration: (a) vacuum permeate (b) sweep gas.

Membrane technology is considered as a promising one due to its advantages; however, the membrane itself still needs to be tailor made to suit the H2 purification or CO2 capture application. The H2 purification membrane for CO2 separation from fuel-cell fuel is required to have the oxidative stability, withstand the high temperature operation (> 100oC) and is also required to have a high CO2/H2 selectivity to minimize the loss of H2, which is

4 the energy source in the fuel-cell. The membrane for CO2 capture from flue gas of a power plant needs to be designed to have very high permeability in order to be able to handle the huge volume of flue gas. In addition, the relatively high scale-up factor (close to 1) for membrane processes can also be a disadvantage for large scale H2 purification and CO2 capture. Therefore, development of advanced membrane with excellent gas separation performance is necessary to answer the challenges in these applications.

1.2. Polymer membranes for gas separation

Two categories of membranes are available for gas separation including polymer membranes and inorganic membranes. There classes of inorganic membranes used for gas separation include dense phase metal, metal alloys and ceramics (perovskites) and porous ceramic membranes [2]. Table 1.1 summarizes the comparison of the inorganic and polymer membranes. In summary, inorganic membranes is favored for harsh temperature and chemical conditions, while polymer membrane is more economical.

Table 1.1. Comparison of polymeric and inorganic membranes for H2 purification. Membrane Advantages Disadvantages Status Inorganic Long term durability Brittle (Pd) Small scale Good thermal Expensive application stability Chemical stability Polymeric Cheap price Short life durability Wide application Mass production Inferior thermal stability in gas separation ability Prone to chemical Good quality control contamination

The membranes that are developed in this work are nanostructured polymer membranes, which can be mass produced more straightforwardly than inorganic membranes and the adsorbents in the adsorption processes. According to the International

Organization for Standardization (ISO) [3], nanostructured material is defined as “material having internal nanostructure or surface nanostructure”, where nanostructure is defined as

“composition of inter-related constituent parts, in which one or more of those parts is a nanoscale region” and nanoscale is defined as “size range from approximately 1 nm to

100 nm”. Based on the definitions, nanostructured membranes can be defined as membranes having internal or surface nanostructure [4]. The nanostructured membrane can be a dense membrane incorporated with nanomaterials, a porous membrane with nanoscale pores, or a combination of both, as illustrated in Figure 1.3.

In some approaches, as in Figure 1.3 (a) and (d), various nanomaterials have been incorporated into polymer matrixes to prepare nanostructured membranes for gas separations, including titanium dioxide (TiO2), silica (SiO2), carbon nanotubes, zeolite, and metal organic frameworks (MOF). The selection of the nanomaterial and polymer matrix has been the subject of extensive research recently [5–15]. Titanium dioxide was favored due to its hydrophilicity, chemical and thermal stability, and gas separation properties [5–

7]. Up to 10 wt.% of fumed TiO2 nanoparticles (size of 21 nm) was homogeneously dispersed in polyvinylacetate membranes [8]. Addition of 1 – 10 wt.% of TiO2 improved both the permeability and selectivity of the nanocomposite membranes.

(a) (b) )

(e)

Nanoparticles Polymer chains Nanoscale pores

Silica can enhance the thermal and mechanical stabilities as well as the gas separation property of membranes [9–11]. Xing and Ho [12] incorporated fumed silica nanoparticles

(FS) with a size of 7 nm into crosslinked poly(vinyl alcohol)-poly(siloxane) membrane matrixes for high-pressure CO2/H2 separation. They obtained the best performance with a

-10 3 CO2/H2 selectivity of 87 and a CO2 permeability of 1296 Barrers (1 Barrer = 10 cm

(STP) cm/ (cm2 s cmHg) = 3.35 x 10-16 mol m/ (m2 s Pa)) at 107oC and 220 psia with 22.3 wt.% FS loading in the membrane.

Carbon nanotubes have excellent mechanical and thermal stabilities and potential to minimize the compaction effect in high-pressure gas separation [13,14]. Deng and Hägg

[15] reinforced polyvinylamine/polyvinylalcohol blend membrane with 1 wt.% of carbon nanotubes (CNTs, 1 – 2 µm in length and 20 – 80 nm in diameter) for CO2/CH4 separation.

The CNTs were compatible with the blend polymer membrane and enhanced the water swelling, performance, and durability against the compaction effect at elevated pressures up to 15 bar (1.5 MPa). Zhao et al. [16] utilized both untreated multi-walled carbon nanotubes (MWNTs, 10 – 15 nm in diameter and 0.1 – 10 µm in length) and acid-treated multi-walled carbon nanotubes (AT-MWNTs) as mechanical reinforcing fillers in the crosslinked poly(vinyl alcohol)-poly(siloxane) membrane matrix for high-pressure CO2/H2 separation. During the first 18.5 days test at 1.52 MPa and 380.15 K, the membrane had no change of performance with a CO2 permeability of 836 Barrers and CO2/H2 selectivity of 43. The improvement was attributed to the ability of MWNTs to improve the mechanical strength and anti-compaction property of the mixed matrix membranes. The optimum untreated MWNTs loading was 2 wt.%, and as the MWNTs increased further, the membrane performance reduced with time due to the poor dispersion of the MWNTs.

However, the dispersion improved with AT-MWNTs. The membrane with 4 wt.% of AT-

MWNTs had a stable membrane performance with a CO2 permeability of 896 Barrers and a CO2/H2 selectivity of 50.9 for at least 11 days.

Graphene oxide has the potential in gas separation application once an ultra-thin (1 –

10 nm) of few layers of graphene is used [17–18]. Shen et al. [19] incorporated laminar structures of graphene oxide with a thickness of 6 – 15 nm into the polyether block amide

(PEBA) matrix and reported a CO2 permeability of 100 Barrers and a CO2/N2 selectivity of 91.

Metal organic frameworks (MOFs), which consist of metal ions and organic molecules in three-dimensional structures with high surface area and porosity that can be fine-tuned, have attracted interests from many researchers [20–22]. A dual layer (organic- and inorganic-based) was formed when the zeolitic imidazolate framework (ZIF)-8 with a pore size of 0.34 nm was used as an inorganic filler in the PEBAX-2533 polymer matrix [23].

As the loading of ZIF-8 increased, the inorganic-based layer thickness increased and hence the permeability was increased whereas the selectivity was constant or slightly reduced.

Polymer of intrinsic microporosity (PIMs) has been considered for gas separation application primarily due to its microporosity, which resulted in a high free volume [24–

26]. One type of polymers of intrinsic microporosity (PIMs), PIM-1, has recently been blended with other components such as mesoporous chromium (III) terephthalate MIL-101

[24], polyetherimide [25], and porous aromatic framework PAF-1 [26] in order to improve the specific surface area, gas transport properties, and reduce the porosity loss (physical aging).

Many efforts were made recently to reduce the thickness of the selective-layer membrane in order to improve the permeance of gas separation, and some researchers were getting close to the nanoscale range of membrane thickness [27–31]. Nanoporous ultrafiltration membranes were utilized as the supports for very thin selective-layers to maintain a good mechanical property. Qiao et al. [27] coated a thin layer of polyvinylamine/piperazine with a thickness as thin as 130 nm on a polysulfone

9 ultrafiltration membrane. The membrane with 220 nm of thickness had a CO2 permeance

2 o of 2.18 µmol/(m s Pa) and a CO2/N2 selectivity of 277 at 22 C and 0.11 MPa.

Ho and coworkers [28] developed the continuous fabrication of 14-inch wide nanostructured polyethersulfone (PES) supports with various pore sizes ranging from 40 nm to 90 nm, as illustrated in Figure 1.3 (b). The membrane with a pore size of 69.5 nm and a porosity of 16.9% was very similar to the commercial PES ultrafiltration membrane

(pore size = 72.3 nm, porosity = 15.8%) used in protein purification. Alternatively, the membrane can also be used as a nanoporous support for a very thin selective-layer for gas separation application since the nanoporous support will provide a negligible resistance for gas transport. The nanostructured PES support was then used to develop a novel inorganic/polymer composite membrane, consisting of a selective amine-containing polymer cover layer / a zeolite nanoparticle layer / a polymer support, for CO2 capture.

Zeolite-Y nanoparticles with around 40 nm of diameter were synthesized with a rapid (1 hour) zeolite growth method [29] and then packed into a layer with a thickness of 250 nm.

A selective amine-containing polymer cover layer with a thickness of 200 nm was coated on the zeolite layer. They continuously fabricated a 14” wide of the inorganic/polymer composite membrane and rolled it into a spiral-wound module with a countercurrent configuration. The membrane showed great potential for the post-combustion CO2 capture from flue gas in coal-fired power plants [30].

Merkel and coworkers [31] developed a CO2-selective membrane for H2 production

2 and CO2 capture in three stages: laboratory tests (membrane area = 30 cm ), pilot-scale test with real syngas (membrane area = 1 – 4 m2), and membrane demonstration system

(membrane area = 20 m2) at the National Carbon Capture Center (NCCC) in Wilsonville,

AL. The thin-film composite PolarisTM membranes consisted of around 50 – 200 nm of a

CO2-selective-layer on top of around 50 – 200 nm of a gutter layer, which were prepared by dip-coating on a nanoporous ultrafiltration support. The membrane demonstration system fed with 227 kg/h of syngas with around 9% of CO2 produced a liquid CO2 product with > 95% of CO2.

1.3. Facilitated transport polymer membranes

There are two mechanism of gas transport in polymer membranes including solution- diffusion mechanism and facilitated transport mechanism. Figure 1.4 shows the schematic of these mechanisms in case of a CO2/H2 separation. The membrane contains CO2 carriers, mobile carrier and/or fixed-site carrier, which can react reversibly with CO2 to form a CO2- carrier complex. In the case of mobile carriers, the CO2-carrier complex diffuses along the membrane thickness. In the case of fixed-site carriers, CO2 moves or hops among the functional groups of the fixed-site carriers along the membrane thickness. The CO2 is then released through the reverse reaction at the permeate side. The inert gases such as H2, N2,

CO, and CH4 cannot react with the CO2 carriers and can only be transported through the membrane by the solution-diffusion mechanism. This mechanism relies on the physical properties including the solubility and diffusivity of the gases in the polymer matrix.

Meanwhile, not only solution-diffusion mechanism naturally occurs for CO2, but also reaction between CO2 and CO2 carriers takes place. The reaction-diffusion of the facilitated

11 transport mechanism has a faster rate compared to the solution-diffusion mechanism and resulted in a significantly improved permeability and selectivity of the membrane.

Feed Gas Retentate

Membrane

Permeate Sweep Gas

CO Fixed-site Carrier 2 H2 Mobile Carrier

Figure 1.4. Schematic of solution-diffusion and facilitated transport mechanism.

The CO2 carriers that can be used in the facilitate transport mechanism include amines, hydroxides, and carbonates according to the following reactions:

+  CO2 + 2 R-NH2 ⇌ R-NH3 + R-NH-COO (1)

+  CO2 + R-NH2 + H2O ⇌ R-NH3 + HCO3 (2)

2-  CO2 + CO3 + H2O ⇌ 2 HCO3 (3)

− − CO2 + OH ⇌ HCO3 (4)

Among these CO2 carriers, amines are the most effective one and therefore are more extensively studied.

In addition, amines can also be used to facilitated transport H2S according to the following reaction:

+  H2S + R-NH2 ⇌ R-NH3 + HS (5)

Therefore, simultaneous removal of acid gases (CO2 and H2S) can be performed by using amines. Huang and Ho demonstrated that H2S permeability was 3 times higher than CO2 permeability and H2S concentration < 10 ppb in the retentate stream could be obtained from a feed gas containing 50 ppm of H2S [32].

Ho and coworkers developed various facilitated transport flat-sheet nanostructured membranes for CO2 separation and H2 purification [32-39]. The lab-scale membranes demonstrated excellent performances and showed potential for various gas separation applications. In order to move towards commercialization of the membrane processes that use the membranes, subsequent steps are necessary including the scale-up fabrication of the flat-sheet membranes, reconfiguration of the flat-sheet membranes into membrane modules, membrane modules field-test with industrial feed gas. Upon completion of these activities, the membrane technology is ready to be transferred to industrial membrane company for commercialization.

1.4. Scope and outline

This dissertation covers the whole range of membrane research from lab-scale to field- test. The research topics includes the oxidatively stable CO2-selective membranes for CO2 removal and H2 purification for fuel cells, CO2-selective membranes for simultaneous CO2 and H2S removal and H2 purification for fuel cells, and CO2-selective membranes for CO2

13 capture from flue gas. Chapter 1 covers introduction to membrane process and polymer membranes for gas separation as well as literature review on facilitates transport polymer membranes.

Chapter 2 relates to the development of an oxidatively stable borate-containing membranes H2 purification for fuel cells. Lab-scale membrane formulation, synthesis, and test of CO2-selective membranes containing borate- and tetrafluoroborate-based additives was performed. The membrane composition was optimized by using the design of experiment method to optimize the CO2 permeance and CO2/H2 selectivity. Scale-up membranes with 14 inches wide and around 1000 feet long were fabricated based on the optimized membrane composition.

Chapter 3 discusses the scale-up of amine-containing membranes for H2 purification for fuel cells. Roll-to-roll membrane fabrication machine was described and effect of the process parameters on the selective-layer thickness of the membrane product was studied.

Scale-up membranes with 14 inches wide and around 150 feet long were fabricated and used to roll into spiral-wound membrane modules. Modeling and process optimization was performed to support the field-test of the membrane modules by determining the design operating condition and analyzing the comparison between the field-test results and the modeling results.

Chapter 4 discloses the design, fabrication, and field-test of spiral-wound membrane modules for CO2 capture from flue gas of coal-fired power plant. A new design of spiral- wound membrane module was developed to enable the transition from flat-sheet membranes into membrane modules. The effect of feed and sweep gases flow rates and

14 contaminants such as SO2 and O2 in the feed gas on the spiral-wound membrane module performance was studied. The test results obtained by using simulated feed gas at OSU were compared to the field-test results. Based on the lessons learned during the field test, modifications on the spiral-wound membrane module fabrication procedures were completed. Chapter 5 covers the author’s final remarks and recommendations for future work.

Nomenclatures

Pi is the permeability of species i

αij is the selectivity of species i over species j i is the gas component CO2

j is another gas component (H2, CO, or H2S); x is the mole fraction of gas components in the feed side of the membrane y is the mole fraction of gas components in the sweep side of the membrane

Ji is the steady-state CO2 molar flux through the selective-layer membrane

is the selective-layer membrane thickness

Δpi is the partial pressure difference between the feed and sweep sides and is defined by using the logarithmic mean method.

References

[1] K. Liu, C. Song, V. Subramani, Hydrogen and syngas production and purification,

John Wiley & Sons, 2010.

[2] G.Q. Lu, J.C. Diniz da Costa, M. Duke, S. Giessler, R. Socolow, R.H. Williams, T.

Kreutz, Inorganic membranes for hydrogen production and purification: A critical

review and perspective, J. Colloid Inter. Sci. 314 (2007) 589–603.

[3] ISO/TS 80004-1:2010 ISO Nanotechnologies, Vocabulary, Core Terms.

International Organization for Standardization (ISO), Geneva; 2010.

[4] N. C. Mueller, V. Van der Bruggen, V. Keuter, P. Luis, T. Melin, W. Pronk, R.

Reisewitz, D. Rickerby, G.M. Rios, W. Wennekes, B. Nowack, Nanofiltration and

nanostructured membranes - Should they be considered nanotechnology or not?, J.

Haz. Mater. 211-212 (2012) 275–280.

[5] S.S. Madaeni, M.M.S. Badieh, V. Vatanpour, N. Ghaemi, Effect of titanium dioxide

nanoparticles on polydimethylsiloxane/polyethersulfone composite membranes for

gas separation, Polym. Eng. Sci. 52 (2012) 2664–2674.

[6] C.Y. Liang, P. Uchytil, R. Petrychkovych, Y.C. Lai, K. Friess, M. Sipek, M.M.

Reddy, S. Y. Suen, A comparison on gas separation between PES

(polyethersulfone)/MMT(Na-montmorillonite) and PES/TiO2 mixed matrix

membranes, Sep. Purif. Technol. 92 (2012) 57–63.

[7] H. Sun, C. Ma, B. Yuan, T. Wang, Y. Xu, Q. Xue, P. Li, Y. Kong, Cardo

polyimides/TiO2 mixed matrix membranes: Synthesis, characterization, and gas

separation property improvement, Sep. Purif. Technol. 122 (2014) 367–375.

[8] J. Ahmad, M.B. Hägg, Polyvinyl acetate/titanium dioxide nanocomposite

membranes for gas separation, J. Membr. Sci. 445 (2013) 200–210.

[9] A.C. Lua, Y. Shen, Preparation and characterization of polyimide–silica composite

membranes and their derived carbon–silica composite membranes for gas separation,

Chem. Eng. J. 220 (2013) 441–451.

[10] B. Zornoza, C. Tellez, J. Coronas, Mixed matrix membranes comprising glassy

polymers and dispersed mesoporous silica spheres for gas separation, J. Membr. Sci.

368 (2011) 100–109.

[11] V. Nafisi, M.B. Hägg, Development of nanocomposite membranes containing

modified Si nanoparticles in: PEBAX-2533 as a block copolymer and 6FDA-Durene

Diamine as a glassy polymer. ACS Appl. Mater. Interfaces 6 (2014) 15643–15652.

[12] R. Xing, W.S.W. Ho, Crosslinked polyvinylalcohol–polysiloxane/fumed silica

mixed matrix membranes containing amines for CO2/H2 separation, J. Membr. Sci.

367 (2011) 91–102.

[13] A.F. Ismail, N.H. Rahim, A. Mustafa, T. Matsuura, B.C. Ng, S. Abdullah, S.A.

Hashemifard, Gas separation performance of polyethersulfone/multi-walled carbon

nanotubes mixed matrix membranes. Sep. Purif. Tech. 80 (2011) 20–31.

[14] B.L. Ge, Z. Zhu, V. Rudolph, Enhanced gas permeability by fabricating

functionalized multi-walled carbon nanotubes and polyethersulfone nanocomposite

membrane, Sep. Purif. Tech. 78 (2011) 76–82.

[15] L. Deng, M.B. Hägg, Carbon nanotube reinforced PVAm/PVA blend FSC

nanocomposite membrane for CO2/CH4 separation, Int. J. Greenhouse Gas Control

26 (2014) 127–134.

[16] Y. Zhao, B.T. Jung, L. Ansaloni, W.S.W. Ho, Multiwalled carbon nanotube mixed

matrix membranes containing amines for high pressure CO2/H2 separation, J. Membr.

Sci. 459 (2014) 233–243.

[17] H. Li, Z. Song, X. Zhang, Y. Huang, S. Li, Y. Mao, H.J. Ploehn, Y. Bao, M. Yu,

Ultrathin, molecular-sieving graphene oxide membranes for selective hydrogen

separation, Science 342 (2013) 95–98.

[18] H.W. Kim, H.W. Yoon, S.M. Yoon, B.M. Yoo, B.K. Ahn, Y.H. Cho, H.J. Shin, H.

Yang, U. Paik, S. Kwon, J.Y. Choi, H.B. Park, Selective gas transport through few-

layered graphene and graphene oxide membranes, Science 342 (2013) 91–94.

[19] J. Shen, G. Liu, K. Huang, W. Jin, K.R. Lee, N. Xu, Membranes with fast and

selective gas-transport channels laminar graphene oxide for efficient CO2 capture,

Angew. 127 (2015) 588–592.

[20] S. Japip, H. Wang, Y. Xiao, T.S. Chung, Highly permeable zeolitic imidazolate

framework (ZIF)-71 nano-particles enhanced polyimide membranes for gas

separation, J. Membr. Sci. 467 (2014) 162–174.

[21] S. Sorribas, B. Zornoza, C. Téllez, J. Coronas, Mixed matrix membranes comprising

silica-(ZIF-8) core–shell spheres with ordered meso–microporosity for natural- and

bio-gas upgrading, J. Membr. Sci. 452 (2014) 184–192.

[22] T. Li, Y. Pan, K.V. Peinemann, Z. Lai, Carbon dioxide selective mixed matrix

composite membrane containing ZIF-7 nano-fillers, J. Membr. Sci. 425-426 (2013)

235–242.

[23] V. Nafisi, M.B. Hägg, Development of dual layer of ZIF-8/PEBAX-2533 mixed

matrix membrane for CO2 capture, J. Membr. Sci. 459 (2014) 244–255.

[24] A.Y. Alentiev, G.N. Bondarenko, Y.V. Kostina, V.P. Shantarovich, S.N. Klyamkin,

V.P. Fedin, K.A. Kovalenko, Y.P. Yampolskii, PIM-1/MIL-101 hybrid composite

membrane material: transport properties and free volume, Petroleum Chem. 54

(2014) 477–481.

[25] L. Hao, J. Zuo, T.S. Chung, Formation of defect-free polyetherimide/PIM-1 hollow

fiber membranes for gas separation, AIChE J. 60 (2014) 3848–3858.

[26] C.H. Lau, K. Konstas, A.W. Thornton, A.C.Y Liu, S. Mudie, D.F. Kennedy, S.C.

Howard, A.J. Hill, M.R. Hill, Gas-separation membranes loaded with porous

aromatic frameworks that improve with age, Angew. Chem. Int. Ed. 54 (2015) 1–6.

[27] Z. Qiao, Z. Wang, C. Zhang, S. Yuan, Y. Zhu, J. Wang, PVAm–PIP/PS composite

membrane with high performance for CO2/N2 separation, AIChE J. 59 (2013) 215–

228.

[28] W.S.W. Ho, P.K. Dutta, S. Schmit, Novel inorganic/polymer composite membranes

for CO2 capture. 2014 NETL CO2 Capture Technology Meeting, (Available online:

http://www.netl.doe.gov/File%20Library/Events/2014/2014%20NETL%20CO2%2

0Capture/W-Ho-OSU-Inorganic-Polymer-Composite-Membranes.pdf).

[29] M. Severance, B. Wang, K. Ramasubramanian, L. Zhao, W.S.W. Ho, P.K. Dutta,

Rapid crystallization of faujasitic zeolites: mechanism and application to zeolite

membrane growth on polymer supports, Langmuir 30 (2014) 6929–6937.

[30] K. Ramasubramanian, W.S.W. Ho, Recent developments on membranes for post-

combustion carbon capture, Curr. Opinion Chem. Eng. 1 (2011) 47–54.

[31] H. Lin, Z. He, Z. Sun, J. Vu, A. Ng, M. Mohammed, J. Kniep, T.C. Merkel, T. Wu,

R.C. Lambrecht, CO2-selective membranes for hydrogen production and CO2 capture

– Part I: Membrane development, J. Membr. Sci. 457 (2014) 149–161.

[32] J. Huang, CO2 (H2S) membrane separations and WGS membrane reactor modeling

for fuel cells (Ph.D. dissertation), The Ohio State University (2007).

[33] J. Zou, Carbon dioxide-selective membranes and their applications in hydrogen

processing (Ph.D. dissertation), The Ohio State University (2007).

[34] H. Bai, High temperature proton-exchange and fuel processing membranes for fuel

cells and other applications (Ph.D. dissertation), The Ohio State University (2008).

[35] K. Ramasubramanian, CO2(H2S)-selective membranes for fuel cell hydrogen

purification and flue gas carbon capture: An experimental and process modeling

study (Ph.D. dissertation), The Ohio State University (2013).

[36] Y. Zhao, Carbon dioxide-selective membranes containing sterically hindered amines

(Ph.D. dissertation), The Ohio State University (2013).

[37] Y. Chen, Polymer membranes for flue gas carbon capture and fuel cell application

(Ph.D. dissertation), The Ohio State University (2015).

[38] V. Vakharia, Development of membrane technology for CO2 removal and H2

purification: A techno-economic, lab-scale, and pilot-scale study (Ph.D. dissertation),

The Ohio State University (2016).

[39] Z. Tong, CO2 facilitated transport membranes for hydrogen purification and flue gas

carbon capture (Ph.D. dissertation), The Ohio State University (2017).

Chapter 2. Oxidatively Stable Borate-containing Membranes for H2 Purification for Fuel

Cell

2.1. Summary

An oxidatively stable borate-containing membrane for H2 purification in fuel cell application was developed. The membrane comprised a quaternaryammonium hydroxide as the mobile carrier for the facilitated transport of CO2, a quaternaryammonium fluoride- containing polymer as the fixed-site catalyst for the reaction between CO2 and water molecules, a borate-based additive as another catalyst for the same reaction, and a crosslinked polyvinylalcohol-polysiloxane as the membrane matrix. The effect of borate- based additive on the membrane performance was investigated. The optimized membrane containing tetrafluoroboric acid achieved a CO2 permeance of 100 GPU and CO2/H2 selectivity > 100. The oxidative stability of the membrane was demonstrated for at least

144 hours at 120°C with humid air as the sweep gas. As the membrane thickness was reduced from 15 µm to 10 µm, a significant drop in CO2/H2 selectivity was observed due to the increase of the H2 permeance while the CO2 permeance was not significantly improved. The water permeation through the membrane was reduced by adding the number of the substrate layers underneath the membrane. The water permeance was reduced almost proportionally as the number of the substrate was increased up to 3 pieces.

The membrane was scaled up by using a continuous roll-to-roll machine to fabricate 14- inch wide flat-sheet membranes with > 1400 feet in total length. The scale-up membrane performed similarly compared to the lab-scale membranes. The membrane developed in

22 this work has the potential for membrane processes that use air as the sweep gas including the H2 purification for fuel cell applications.

2.2. Introduction

Natural gas has been used to produce synthesis gas (syngas), which comprises hydrogen, carbon monoxide, and carbon dioxide. The hydrogen content in the syngas can be increased by utilizing the water gas shift (WGS) reaction, in which steam is added to react with carbon monoxide and produces hydrogen and carbon dioxide. Removal of carbon dioxide in the gas mixture is needed to produce a high-purity hydrogen for fuel cells. As the usage of hydrogen and natural gas is increasing, so is the need for their purification [1–4].

Compared to the existing commercial separation processes for the removal of carbon dioxide, such as absorption, adsorption and cryogenic distillation, a membrane process has the advantages of energy efficiency, system compactness, operation and maintenance simplicity, and low capital cost. The driving force of a membrane process typically arises by using either a vacuum or sweep gas on the permeate side. The sweep gases that can be used include air, nitrogen, steam, etc. In certain application, it is desirable to use air as the sweep gas in the membrane process [5]. Hence, the development of an oxidatively stable membrane for such applications is required.

Polymer membranes have advantages compared to inorganic membranes including ease in scale-up manufacturing and lower cost [6–7]. Among the polymer membranes, facilitated transport membranes are more suitable for fuel cell application since they offer a high selectivity [8–10]. A high selectivity is important to limit the H2 loss in the

23 membrane process since H2 is precious as the fuel. Amines are considered effective CO2 carriers for the facilitated transport of CO2 through the membrane. Ho and coworkers have developed amine-based membranes with excellent performance for H2 purification [11–

20]. However, the oxidative degradation of amines by oxygen in the air at high temperatures (> 100oC) has prevented the use of air as the sweep gas in the membrane process [21–25].

Hydroxide- and fluoride-containing quaternaryammonium compounds have been considered for the facilitated transport of CO2 [26–33]. Facilitated transport membranes based on quaternaryammonium polyelectrolytes have demonstrated attractive CO2 transport performance and stability at room temperature [27–29]. The membranes may be used in the presence of air as the sweep gas.

Like amines, hydroxide ions are considered as CO2 carriers for the facilitated transport

− membranes. The hydroxide ion reacts with CO2 to form the bicarbonate ion (HCO3 ) as shown in Reaction (1). Due to the concentration gradient, the reaction product diffuses from the feed side to the permeate side and eventually releases the CO2 molecule on the permeate side with a lower CO2 partial pressure than the feed side. Upon the release of

CO2, the hydroxide ion diffuses back to the feed side due to the concentration gradient. In other words, the CO2 transport is continuously facilitated by the reaction-diffusion mechanism.

− − CO2 + OH ⇌ HCO3 (1)

− + CO2 + H2O ⇌ HCO3 + H (2)

As shown in Reaction (2), H2O molecules can also react with CO2 in the membrane.

However, the reaction rate is significantly slower than Reaction (1) [34]. The CO2 transport of the selectively permeable membranes can be improved by incorporating a catalyst for

Reaction (2), such as a fluoride-containing compound. The fluoride ions in such compounds can make the oxygen atom of the water molecule more basic via hydrogen bonding and increase the rate of reaction with CO2 molecules [35].

Researchers have explored the use of different catalysts for the CO2-H2O reaction

[34,36–41]. Sharma and Danckwerts [34] concluded that catalysts with weak acids as the anions such as selenite, arsenite, and hypobromite could catalyze the CO2-H2O reaction.

However, the carcinogenic nature of arsenite and the stability of hypobromite prevented their use for industrial applications. Eickmeyer [40] and Field [41] demonstrated that the reaction rate of the CO2-H2O reaction was enhanced by using environmentally friendly sodium or potassium borate as catalysts. Field [41] disclosed the effect of sodium or potassium metaborate and tetraborate salts as the catalysts for the CO2-H2O reaction.

Recently, Ho and coworkers developed oxidatively stable membranes that comprised oxidatively stable quaternaryammonium hydroxide small molecules and fluoride- containing macromolecules as the mobile carrier for the facilitated transport of CO2 and the fixed-site catalyst for the reaction between CO2 and water molecules, respectively, in a crosslinked polyvinylalcohol-polysiloxane (XLPVA-POS) matrix [42–44]. They used

o borate additives to enhance the performance of the membranes at 120 C for H2 purification applications. In this work, the borate-containing membranes with desirable CO2 permeance and excellent CO2/H2 selectivity were synthesized and investigated.

Optimization of the membrane composition was conducted on the lab scale. Later, the scale-up of the optimized membrane was performed, and the performance of the scale-up membrane compared favorably to that of the lab-scale membrane.

2.3. Experimental

2.3.1. Materials

Tetramethylquaternaryammonium hydroxide (TMQOH, 25%), tetraethoxysilane

(TEOS, 99%), glutaraldehyde (GA, 50%), poly(diallyldimethylquaternaryammonium chloride) (PDADMQ-Cl, 20%), potassium hydroxide (KOH, >85%), potassium fluoride

(KF, 99%), boric acid (BA, 99.5%), sodium tetraborate (STB, 99%), potassium tetraborate

(KTB, 99%), tetrafluoroboric acid (TFBA, 48%), sodium tetrafluoroborate (STFB, 98%), potassium tetrafluoroborate (KTFB, 96%), and tetramethylquaternaryammonium tetrafluoroborate (TMQ-TFB, 97%) were purchased from Sigma-Aldrich (Milwaukee, WI,

USA). Hydrochloric acid (HCl, 36.5%) was bought from Fisher Scientific Inc. (Pittsburgh,

PA, USA). Polyvinylalcohol (PVA, POVAL S-2217 MW 150000, 92%) was kindly donated by Kuraray America, Inc. (Houston, TX, USA).

Porous polysulfone substrates (average pore size = 9 nm, porosity = 7 %), polyethersulfone UF10 (PES UF10), polyvinylidene fluoride (PVDF), and nonwoven fabric (NWF) were purchased from TriSep Corporation (Goleta, CA, USA). The Ohio State

University polyethersulfone (OSU PES) and hydrophilic polyethersulfone (OSU PES-H) were developed by Ho and coworkers [45]. The feed gas mixture (certified composition of

26.5% H2, 14% CO, 59.5% CO2), sweep gas (air), and argon (99.998%) gas cylinders were purchased from Praxair, Inc. (Danbury, CT, USA) and used for the gas transport performance measurements.

2.3.2. Membrane preparation

The poly(diallyldimethylammonium fluoride) (PDADMQ-F) polymer solution was prepared from the poly(diallyldimethylammonium chloride) (PDADMQ-Cl) by using the following procedure. First, 20 g of 20 wt.% PDADMQ-Cl aqueous solution was dried by air-purging at room temperature and then dried in a vacuum oven at 102oC to obtain 4.62 g of dried PDADMQ-Cl. The dried PDADMQ-Cl was dissolved in 30 g of methanol and followed by addition of 1.86 g of potassium fluoride (KF) under vigorous stirring. The ion-exchange reaction between PDADMQ-Cl and KF to obtain the PDADMQ-F was performed for 24 hours at room temperature. The byproduct salt, KCl, was insoluble in methanol and therefore precipitated. The salt was then removed from the mixture by using the centrifugation at 8000 rpm for 5 minutes. Before being used as one of the components of the coating solution for membrane preparation, the PDADMQ-F solution was air-purged to evaporate the methanol, and was subsequently re-dissolved in water to obtain an aqueous solution of PDADMQ-F at 14.61 wt.%.

The crosslinked polyvinylalcohol-polysiloxane (XLPVA-POS) solution was prepared by using the following procedure. First, PVA was dissolved in water to obtain an aqueous

PVA solution with 13 wt.% of total solid concentration. The PVA was crosslinked by tetraethylorthosilicate (TEOS) and glutaraldehyde (GA) (mole ratio = 40:60), with a

27 targeted degree of crosslinking of 100%, based on the previous study [42]. The aqueous

PVA solution temperature was increased to 80oC under a continuous stirring before the

TEOS and HCl solution were added. The crosslinking reaction with TEOS was conducted at 80oC for 80 minutes. Subsequently, the KOH solution (40 wt.% in water) was added dropwise to the PVA-TEOS solution under gentle stirring. The solution was mixed for 30 minutes before the GA solution (50 wt.% in water) was added dropwise under vigorous stirring. The crosslinking reaction with GA was conducted at 80oC for 2 ½ hours. The total solid content in the crosslinked PVA solution was targeted at 15 wt.%, consisting of 12 wt.% of crosslinked PVA and 3 wt.% of KOH.

An appropriate amount of water was added to the XLPVA-POS solution to avoid gel formation of the coating solution and adjust the final concentration of the total solid in the coating solution. Afterwards, the TMQOH solution and the PDADMQ-F polymer solution were added dropwise to the XLPVA-POS solution under vigorous stirring based on the membrane composition. Subsequently, the calculated amount of a borate solution was added dropwise to the coating solution. The coating solution was mixed continuously after the addition of all the components and then air purged until a desired viscosity was achieved. A high viscosity of the coating solution was required to minimize the penetration of the coating solution into the pores of the substrate. The coating solution was centrifuged at 8000 rpm for 3 minutes to remove air bubbles before it was coated on a flat-sheet polysulfone support at a controlled gap setting by using a GARDCO adjustable micrometer film applicator (Paul N. Gardner Company, Pompano Beach, FL, USA). The membrane was cured at 120 °C in a convection oven for 6 hours to complete the removal of water and

28 crosslinking reaction of PVA. The total thickness of the membrane was measured by a

Mitutoyo electronic indicator (Model 543-252B, Mitutoyo America Corp, Aurora, IL,

USA) with an accuracy of ± 0.5 μm. The thickness of the selective-layer was calculated by subtracting the polysulfone support thickness from the total thickness of the composite membrane.

2.3.3. Gas transport measurement

Figure 2.1 shows the schematic representation of the gas permeation unit that was used to measure the gas transport performance of the membranes. The setup consisted of mass flow controllers, water pumps, humidifiers, a permeation cell, water knock-out vessels, and a gas chromatography. The membrane was placed in a rectangular gas permeation cell between the upper and lower parts. A membrane active area of 3.4 cm2 was used for the gas transport performance measurement at 120oC. During the gas transport permeation measurement, a dry feed gas of 60 cc/min at 1.5 psig and a dry sweep gas of 30 cc/min at

1 psig were humidified with water vapor in the feed and sweep humidifiers, respectively.

Controlled water amounts of 0.054 sccm and 0.01 sccm were injected into the feed and sweep sides, respectively. The humidified feed and sweep gases entered the cell in a countercurrent configuration at the top and bottom compartments of the permeation cell, respectively. The retentate and permeate streams flowed out of the permeation cell and through the water knock-out vessels so that most of the water in the gas streams could be removed. Afterwards, the retentate and permeate gas streams were directed to the Agilent

6890N gas chromatography (GC) which was equipped with a Supelco stainless steel

29 micropacked column Carboxen 1004 with 80/100 mesh (Bellefonte, CA, USA) and a thermal conductivity detector (TCD) for gas composition analysis. The gas transport performances of the membranes, including the permeability Pi and selectivity αij, were evaluated by using the gas compositions obtained from the GC in the flux equation as follows:

퐽푖 푦푖⁄푦푗 푃푖 = 훼푖푗 = 훥푝푖/ℓ 푥푖⁄푥푗

(푝푖,푓푒푒푑 푖푛 − 푝푖,푠푤푒푒푝 표푢푡) − (푝푖,푓푒푒푑 표푢푡 − 푝푖,푠푤푒푒푝 푖푛) 훥푝푖 = ln(푝푖,푓푒푒푑 푖푛 − 푝푖,푠푤푒푒푝 표푢푡) − ln(푝푖,푓푒푒푑 표푢푡 − 푝푖,푠푤푒푒푝 푖푛) where i denotes the gas component CO2 and j denotes another gas component (H2 or CO),

Ji is the steady-state CO2 molar flux through the selective-layer membrane, is the selective-layer membrane thickness, and Δpi is the log-mean partial pressure difference between the feed and sweep sides. y and x are the mole fractions of gas components in the sweep and feed sides of the membrane, respectively.

−10 3 2 The common unit of the permeability (Pi) is Barrer (10 cm (STP)·cm /cm ·s ·cmHg

= 3.35 × 10−16 mol·m / m2·s·Pa). The permeance of a gas component i is defined as the ratio of the permeability of gas i to the thickness of the selective-layer membrane (Pi / ) and its common unit is the gas permeation unit (GPU) (10−6 cm3(STP) / cm2·s·cmHg).

Sweep Feed Gas Gas

Figure 2.1. Schematic of the gas permeation unit for gas transport measurements.

2.3.4. Scale-up membrane fabrication

The scale-up fabrication of the oxidatively stable borate-containing membranes was demonstrated by using the pilot-scale thin-film-coating (TFC) assembly of a continuous roll-to-roll membrane machine at The Ohio State University. The schematic representation of the thin-film coating assembly that was used for the scale-up membrane fabrication was described previously [37]. A roll of 14-in wide polysulfone substrate at the unwind roll

31 was rotated by the unwind motor and was led to the coating assembly. The coating solution was stored in the chamber between the coating knife and the back plate. The thickness of the membrane was controlled by setting the gap between the coating knife and the base of the coating assembly as well as the web speed of the substrate. After passing the coating knife, the polysulfone substrate was covered with a uniform layer of a wet film of the coating solution. This composite membrane was passed into the curing oven, which provided a convective hot air countercurrently to the substrate motion. After exiting the curing oven, the dried composite membrane was collected at the rewind roller.

2.4. Results and discussion

2.4.1. Effects of different borate-based catalysts and concentrations on membrane performances

Boric acid (BA), sodium tetraborate (STB), and potassium dihydrogenborate (KDHB) were investigated as the catalysts for the CO2-H2O reaction that facilitates the CO2 transport through quaternaryammonium hydroxide- and fluoride-containing membranes.

Table 2.1 shows the compositions and gas transport performances of the membranes with

BA as the catalyst. The membranes that comprised STB and KDHB were also prepared with the same composition as shown in Table 2.1. The effects of BA, STB, and KDHB at different catalyst concentrations on the membrane gas transport performances are shown in Figure 2.2. As the catalyst concentration increased up to 2 wt.%, the CO2 permeances increased significantly, whereas the H2 permeances increased slightly. Overall, there was

32 an increase in the CO2/H2 selectivity as the catalyst concentration increased. These results suggested that borate-based compounds, including BA, STB, and KDHB could be used as the catalyst for enhancing CO2 transport. Since there was no significant improvement of the membrane performance after increasing the catalyst content to 2 wt.%, the catalyst content was kept up to this maximum amount for subsequent experiments.

Table 2.1. Gas transport performances of the membranes with BA as the catalyst. Sample TMQOH PDADMQ-F XLPVA- BA CO2 CO2/H2 No. (wt.%) (wt.%) POS (wt.%) (wt.%) permeance selectivity (GPU) M-1 11 54 35 0 52 161 M-2 10.9 53.8 34.9 0.4 59 163 M-3 10.9 53.6 34.7 0.8 64 169 M-4 10.8 52.9 34.3 2 68 164

Figure 2.3 shows the schematic representations of the interactions for borate-based and tetrafluoroborate-based compounds with a water molecule during the CO2-H2O reaction.

Vakharia et al. [42] proposed that two fluoride ions interacted with one water molecule to cause an increase of the basicity of the water molecule (Figure 2.3(b)) and hence an increase of reactivity for the CO2-H2O reaction. Similarly, the oxygen atoms of the borate- based compound interact with the hydrogen atoms of the water molecules via hydrogen bonding, resulting in an increased rate of CO2-H2O reaction, as seen in Figure 2.3(a).

Moreover, the incorporation of borate ions increases the ionic character of the membrane, reducing the H2 solubility in the membrane. Consequently, these membranes can exhibit an improved CO2/H2 selectivity. However, it is expected that a tetrafluoroborate-based 33 compound can increase the basicity of the water molecule more than a borate-based compound, which is discussed in the following section.

120

100

Permeance (GPU) Permeance STB 2 KDHB

CO 20 BA 0 0 0.5 1 1.5 2 2.5 Catalyst concentration (wt%)

(a)

220 200 180 160 140 120

Selectivity 100 2

/H 80 2

60 STB CO 40 KDHB 20 BA 0 0 0.5 1 1.5 2 2.5 Catalyst concentration (wt%)

(b)

Figure 2.2. Effect of different borate-based catalysts and concentrations on the membrane gas transport performances: (a) CO2 permeance (b) CO2/H2 selectivity.

O - δ O O δ+ B H

- C δ- δ H O δ+ O H H O

(a)

- + - δ O H F B F δ + H C δ- F + δ O H δ- O

(b)

Figure 2.3. Schematic representations of the molecular interactions for (a) borate-based compound and (b) tetrafluoroborate-based compound with a H2O molecule for the H2O- CO2 reaction.

2.4.2. Effect of different tetrafluoroborate-based catalysts and concentrations on membrane performances

Tetrafluoroboric acid (TFBA) was investigated as a catalyst for the CO2-H2O reaction in quaternaryammonium hydroxide- and fluoride-containing membranes. Different coating solutions with total solid concentrations of 12 wt.% were prepared by using different contents of TFBA of 0.4 – 2 wt.% of total solid. The membrane compositions and gas transport performances are shown in Table 2.2. The gas transport performance of the membranes with TFBA as a catalyst (M-2 to M-5), exhibited significantly higher CO2 permeance along with a CO2/H2 selectivity higher than or comparable to the membrane without any catalyst (M-1). As the catalyst content increased to 2 wt.% for M-5, the CO2 permeance increased up to 91 GPU while the CO2/H2 selectivity was 193. This table demonstrated the significant effect of tetrafluoroboric acid (TFBA) on the CO2 permeance improvement by catalyzing the CO2-H2O reaction in the membrane.

Table 2.2. Gas transport performances of the membranes with TFBA as the catalyst. Sample TMQOH PDADMQ-F XLPVA- TFBA CO2 CO2/H2 No. (wt.%) (wt.%) POS (wt.%) (wt.%) permeance selectivity (GPU) M-1 11 54 35 0 52 161 M-2 10.9 53.8 34.9 0.4 73 174 M-3 10.9 53.6 34.7 0.8 77 205 M-4 10.8 53.1 34.5 1.6 87 163 M-5 10.8 52.9 34.3 2 91 193

The above results showed that replacing boric acid or its salts with tetrafluoroboric acid

(TFBA) or its salt further enhanced the membrane performance. Figure 2.3(b) shows a schematic representation of the interaction of TFBA with a water molecule for the CO2-

H2O reaction. The hydrogen bonding strength between the -F and -H atoms is stronger than that between the -O and -H atoms shown in Figure 2.3(a). Moreover, a TFBA molecule comprises four fluoride atoms as hydrogen bonding donors in comparison with three in a boric acid molecule. Both factors are believed to enhance the basicity of the water molecule further (via hydrogen bonding), thus increasing the rate of CO2-H2O reaction in the presence of TFBA.

The membranes, which comprised 2 wt.% of other tetrafluoroborate-based catalysts, including sodium tetrafluoroborate (STFB), potassium tetrafluoroborate (KTFB), and tetramethylquaternaryammonium tetrafluoroborate (TMQ-TFB), were also prepared with the same composition as membrane sample M-5 in Table 2.2. The transport performances of these membranes were compared to those of the membranes containing the borate-based catalysts in Table 2.3. The highest CO2 permeance was observed in the membrane that comprised a tetrafluoroborate salt with a quaternaryammonium cation (TMQ-TFB).

However, its CO2/H2 selectivity was significantly lower as compared to the membranes with borate-based catalysts or other salts of tetrafluoroborate as the catalysts. The membranes containing the sodium and potassium salts of tetrafluoroborate demonstrated

CO2 permeances higher than the membranes containing sodium and potassium tetraborate, but lower than the membrane containing TFBA as the catalyst. On the other hand, the

CO2/H2 selectivity of these membranes seemed similar. Therefore, the

quaternaryammonium hydroxide- and fluoride-containing membranes with TFBA as the

catalyst achieved a desirable CO2 permeance and CO2/H2 selectivity.

Table 2.3. Performances of membranes containing different borate- and tetrafluoroborate- based catalysts. Sample TMQOH PDADMQ- XLPVA- Catalyst (wt.%) CO2 CO2/H2 No. (wt.%) F (wt.%) POS permeance selectivity (wt.%) (GPU) M-1 10.8 52.9 34.3 BA (2) 65 171 M-2 10.8 52.9 34.3 STB (2) 79 185 M-3 10.8 52.9 34.3 KDHB (2) 65 189 M-4 10.8 52.9 34.3 TFBA (2) 91 193 M-5 10.8 52.9 34.3 STFB (2) 85 190 M-6 10.8 52.9 34.3 KTFB (2) 83 186 M-7 10.8 52.9 34.3 TMQ-TFB (2) 96 163 M-8 10.6 51.8 33.6 TFBA (2) + 90 160 TMQ-TFB (2)

The combination of tetrafluoroboric acid (TFBA) and

tetramethylquaternaryammonium tetrafluoroborate (TMQ-TFB) as the catalyst for the

CO2-H2O reaction in quaternaryammonium hydroxide- and fluoride-containing

membranes was investigated. The equal weight amounts of TMQ-TFB solution and TFBA

solution were added dropwise to the coating solution for membrane preparation. As shown

in Table 2.3, the transport performances showed the improved CO2 permeances at the

expense of the CO2/H2 selectivity for the total catalyst content of 2 wt.%. But, for the total

catalyst content of 4 wt.%, the reduced performance was observed, presumably due to the

reduction of the CO2 carrier amounts. 38

2.4.3. Effect of membrane thickness

The effect of selective-layer thickness on gas transport performance for the borate- and tetrafluoroborate-containing membranes was investigated. Membranes with a composition of 10.8 wt.% TMQOH, 52.9 wt.% PDADMQ-F, 34.3 wt.% XLPVA-POS, and 2 wt.% of catalyst were used for the investigation. The catalysts that yielded better membrane performance, including STB, KDHB, TFBA, and TMQ-TFB, were used in this study. As shown in Figure 2.4, when the selective-layer membrane thickness reduced, the CO2 permeance increased whereas the CO2/H2 selectivity decreases pronouncedly. However, the tetrafluoroborate-catalyzed membranes with different thicknesses showed very good performances with a CO2 permeance up to 107 GPU and a CO2/H2 selectivity of > 90.

Also shown in Figure 2.4, as the selective-layer membrane thickness reduced, the CO2 permeance did not increase as prominently as the H2 permeance. This phenomenon was caused by the different gas transport mechanisms of the H2 and CO2 molecules. The H2 molecules followed the solution-diffusion mechanism to transport through the membrane from the feed side to the sweep side. For this mechanism, the diffusional resistance for the

H2 transport decreased as the selective-layer membrane thickness reduced, resulting in a higher H2 permeance. On the other hand, besides the solution-diffusion mechanism, the

CO2 molecules mainly followed the reaction-diffusion transport mechanism, i.e., facilitated transport, to permeate through the membrane. In this situation, the reaction of

CO2 with the CO2 carrier molecules along with the diffusion of the reaction product was

39 the controlling step. Therefore, the thinner membranes did not result in a significant increase in CO2 permeance as compared to the H2 permeance.

120

100

Permeance (GPU) Permeance 40 TFBA 2

CO TMQ-TFB 20 STB KDHB 0 8 10 12 14 16 Selective-layer thickness (µm)

(a)

220 200 180 160 140 120

Selectivity 100 2

/H 80 2 TFBA

CO 60 TMQ-TFB 40 STB 20 KDHB 0 8 10 12 14 16 Selective-layer thickness (µm)

(b)

Figure 2.4. Effect of selective-layer membrane thickness on transport performances: (a) CO2 permeance (b) CO2/H2 selectivity.

2.4.4. Membrane stability with air as sweep gas

The oxidative stability of quaternaryammonium hydroxide- and fluoride-containing membrane with TFBA as a catalyst was investigated at by using air as the sweep gas. The membrane with a composition of 7.8 wt.% TMQOH, 58.9 wt.% PDADMQ-F, 31.9 wt.%

XLPVA-POS, and 2 wt.% of TFBA was used for the measurement of gas transport performance stability. This composition was chosen as it resulted in the optimum performance of both CO2 permeance > 100 GPU and CO2/H2 selectivity > 100. The stability plot of the membrane is shown in Figure 2.5. The membrane reached a steady- state performance with CO2 permeance of 101 GPU and CO2/H2 selectivity of 105 after around 24 hours of the test. Overall, the CO2 permeance and CO2/H2 selectivity were very stable without any changes during the 144 hours of the test.

120 200 CO Permeance 2 180 100 160 140 80 H2 Permeance (x 100) 120

60 100 Selectivity

80 2 /H

40 2

60 Permeance (GPU) Permeance 40 CO 20 20 0 0 0 20 40 60 80 100 120 140 Time (hours)

Figure 2.5. The stability plot of the membrane containing TFBA as catalyst with air as the sweep gas.

2.4.5. Membrane composition optimization with TFBA as catalyst

The effect of tetrafluoroboric acid (TFBA) as the catalyst was investigated further by optimizing the total solid component compositions of the quaternaryammonium hydroxide- and fluoride-containing membranes. The membrane compositions were optimized by using the design of experiment method, which is described in Appendix A

(the membrane compositions generated by JMP software [51,52] are shown in Table A.1).

The experimental data were analyzed by using JMP software [51,52], and the results are included in Appendix A. The compositions and transport performances of the TFBA- containing membranes with the best performances, M-16 and M-22 in Appendix A, are shown in Table 2.4. As seen in this table, these two membranes were able to achieve CO2 permeances of 100 GPU and CO2/H2 selectivities > 100. The prediction equation obtained from JMP pointed to an optimum membrane composition of 7.5 wt.% TMQOH, 51.5 wt.%

PDADMQ-F, 36 wt.% XLPVA-POS, and 5 wt.% of TFBA. This optimum composition was used in the membrane scale-up fabrication, which is described later.

Table 2.4. The optimized membrane compositions and transport performances of TFBA- containing membranes. TMQOH PDADMQ-F XLPVA- TFBA CO2 permeance CO2/H2 (wt.%) (wt.%) POS (wt.%) (wt.%) (GPU) selectivity 7.5 51.5 35 6 100 114 8 56.5 32.5 3 100 118

2.4.6. Effects of additional substrates on water permeation in TFBA-containing membranes

The effects of additional substrates underneath the quaternaryammonium hydroxide- and fluoride-containing membranes with TFBA as the catalyst on the water permeance through the membranes were investigated. The membranes (15 µm thickness) with total solid compositions as shown in Table 2.5 were prepared for the measurement of gas transport performances. The gas transport measurements were performed in the same way as described in the previous examples. Inside the membrane cell, additional different substrates were stacked underneath the typical composite membrane as shown in Table 2.5, including polysulfone (PSf), polyethersulfone UF10 (PES UF10), polyvinylidene fluoride

(PVDF), Ohio State University polyethersulfone (OSU PES), Ohio State University hydrophilic polyethersulfone (OSU PES-H), and nonwoven fabric (NWF). As seen in

Table 2.5, the water permeance was almost linearly correlated to the number of substrates below the selective-layer (except for nonwoven fabric) while the CO2 permeances and

CO2/H2 selectivities were not significantly affected by the number of substrates except for

PES UF10. The PES UF10 with a MW cutoff of 10000 Da had smaller pores than the other substrates, and it presumably provided significant mass transfer resistance for the CO2 molecules. Hence, the permeation of water through the membrane can be controlled by adding layers of substrates under the membrane.

Table 2.5. Effects of additional substrates on water permeation in TFBA-containing membranes. Sample Composition – wt.% Additional CO2 H2O CO2/H2 No. (TMQOH : PDADMQ-F Substrate Permeance Permeance Selectivity : XLPVA-POS: TFBA) (GPU) (GPU) M-1 7 : 57.5 : 32.5 : 3 0 99 2423 120 M-2 7 : 57.5 : 32.5 : 3 1 PSf 98 1578 118 M-3 7 : 57.5 : 32.5 : 3 2 PSf 97 830 116 M-4 7 : 56.5 : 32.5 : 4 0 100 2440 115 M-5 7 : 56.5 : 32.5 : 4 PES UF10 42 1050 81 M-6 7 : 56.5 : 32.5 : 4 PVDF 99 1345 114 M-7 7 : 56.5 : 32.5 : 4 2 PVDF 97 930 114 M-8 7 : 56.5 : 32.5 : 4 OSU PES 99 1336 116 M-9 7 : 56.5 : 32.5 : 4 2 OSU PES 98 761 115 M-10 7 : 56.5 : 32.5 : 4 OSU PES-H 100 1471 116 M-11 7 : 56.5 : 32.5 : 4 2 OSU PES-H 99 810 116 M-12 7 : 56.5 : 32.5 : 4 NWF 100 1728 114 M-13 7 : 56.5 : 32.5 : 4 2 NWF 100 1710 115 2.4.7. Membrane scale-up fabrication

The fabrication of the oxidatively stable quaternaryammonium hydroxide- and

fluoride-containing membrane with TFBA as the catalyst was successfully scaled-up by

using a pilot-scale thin-film coating (TFC) assembly of the continuous roll-to-roll

fabrication machine at The Ohio State University. The total solid compositions as shown

in Table 2.6 were used for the scale-up fabrication for the targeted selective-layer thickness

of 15 µm. Figure 2.6 shows the wet film membrane prior to entering the curing oven as

well as the dry membrane coming out of the curing oven. A total of more than 1400-feet

44 long and 14-inch wide scale-up membranes was successfully fabricated with the targeted selective-layer thickness.

A small representative sample with a membrane testing area of 3.4 cm2 was taken from each of the scale-up membrane and was tested in the gas permeation unit. Table 2.6 also shows the gas transport performances of the scale-up membranes from each of scale-up runs and the performance comparison with the lab-scale synthesized membrane. As shown in this table, the gas transport performances of the membranes fabricated by using the pilot- scale roll-to-roll machine were reasonably similar to those of the lab-scale synthesized membranes. The selective-layer thickness variation of ±1 µm along the 14-inch width and the ≥110-ft length for the scale-up membranes could explain the small differences in the

CO2 permeance and CO2/H2 selectivity results.

Table 2.6. The performance comparison of the lab-scale and scale-up quaternaryammonium hydroxide- and fluoride-containing membranes with TFBA as the catalyst. Sample Composition (wt.%) Fabrication CO2 CO2/H2 No. TMQOH : PDADMQ-F : Scale permeance selectivity XLPVA-POS : TFBA (GPU) Lab-scale M-1 6.8 : 59.3 : 31.9 : 2 97 108 (2” × 3”) Lab-scale M-2 7 : 56.5 : 32.5 : 4 100 115 (2” × 3”) Lab-scale M-3 7.5 : 51.5 : 36 : 5 100 118 (2” × 3”) Scale-up SM-1 6.8 : 59.3 : 31.9 : 2 96 109 (14” × 330’) Scale-up SM-2 7 : 56.5 : 32.5 : 4 99 114 (14” × 110’) Scale-up SM-3 7.5 : 51.5 : 36 : 5 100 117 (14” × 990’)

Figure 2.6. The scale-up membrane fabricated by using the roll-to-roll fabrication machine: (a) wet film membrane before curing (b) dry membrane after curing.

2.5. Future directions

This work has demonstrated that borate-containing membranes can achieve a desirable

CO2 permeance and an excellent CO2/H2 selectivity with air as the sweep gas. Other borate-containing compounds, e.g., tetrafluoroborate compounds, or the combinations thereof can be incorporated into the membrane to improve the membrane performance.

Other hydroxide- and fluoride-containing salts and polymers instead of TMQOH and

PDADMQ-F can also be incorporated in the membrane. Hydrophilic polymers with higher molecular weights can also be used as the membrane matrix instead of the crosslinked polyvinylalcohol-polysiloxane.

Since the membrane synthesis has been scaled-up to produce > 1400-ft of flat-sheet membranes, they can be used to prepare membrane modules for a field test with actual syngas used as the fuel for fuel cells. Afterwards, integration of the membrane modules with the fuel cell stack can be considered [5]. The long-term stability test of the membrane can be performed to ensure the robustness of the membrane performance over fluctuations

46 of operating conditions. Finally, the membrane fabrication can be scaled up to wider than

14 inches for mass production for commercial applications.

2.6. Conclusions

New oxidatively stable membranes containing quaternaryammonium hydroxide and fluoride, and borate (or tetrafluoroborate) in the crosslinked polyvinylalcohol-polysiloxane matrix were developed in this study. The optimized membrane showed a CO2 permeance of 100 GPU and a CO2/H2 selectivity of > 100. This membrane showed a stable performance for at least 144 hours at 120oC with air as the sweep gas. A significant drop in CO2/H2 selectivity was observed as the selective-layer membrane thickness was reduced from 15 µm to 10 µm due to the increase in the H2 permeance. The water permeation through the membrane was controllable by adding more layers of substrate underneath the membrane. The membrane was scaled up to 14 inches in width for more than 1400 feet.

The scale-up membrane showed similar performance compared to the lab-scale membrane.

Acknowledgments

I acknowledge the contributions of Dr. Varun Vakharia, Mr. Kai Chen, and Dr. Michael

Gasda to this work. I would like to thank Kuraray America Inc. in Houston, TX for donating free samples of polyvinylalcohol. I would also like to gratefully acknowledge

Bloom Energy for the financial support of this work.

Nomenclatures

Pi is the permeability of species i

αij is the selectivity of species i over species j i is the gas component CO2

j is another gas component (H2, CO, or H2S); x is the mole fraction of gas components in the feed side of the membrane y is the mole fraction of gas components in the sweep side of the membrane

Ji is the steady-state CO2 molar flux through the selective-layer membrane

is the selective-layer membrane thickness

Δpi is the partial pressure difference between the feed and sweep sides and is defined by using the logarithmic mean method.

References

[1] J.A. Turner, A realizable renewable energy future, Science 285 (1999) 687–689.

[2] C. Song, Global challenges and strategies for control, conversion and utilization of

CO2 for sustainable development involving energy, catalysis, adsorption and

chemical processing, Catalysis Today 115 (2006) 2–32.

[3] K. Ramasubramanian, Y. Zhao, W.S.W. Ho, CO2 capture and H2 purification:

Prospects for CO2-selective membrane processes, AIChE J. 59 (2013) 1033–1045.

[4] United States Department of Energy Fuel Cell Technologies Office, State of the

states: Fuel cells in America 2016, 7th ed., (2016)

[5] M. Gasda, A. Ballantine, C. Karuppaiah, W.S.W. Ho, Carbon dioxide separator, fuel

cell system including same, and method of operating the fuel cell system, U.S. Patent

2016/0248111 A1, 2016.

[6] W.J. Koros, R. Mahajan, Pushing the limits on possibilities for large scale gas

separation: which strategies?, J. Membr. Sci. 175 (2000) 181–196.

[7] R. Baker, Future directions of membrane gas-separation technology, Membr. Tech.

138 (2001) 5–10.

[8] V. Vakharia, W.S.W. Ho, Separation and Purification of Hydrogen Using CO2-

Selective Facilitated Transport Membranes, Hydrogen Production from Renewable

Resources, Springer Book Series - Biofuels and Biorefineries, Chap. 11, (2015) 315–

338.

[9] K. Ramasubramanian, CO2 (H2S)-selective membranes for fuel cell hydrogen

purification and flue gas carbon capture: an experimental and process modeling study

(Ph.D. dissertation), The Ohio State University (2013).

[10] P. Li, Z. Wang, Z. Qiao, Y. Liu, X. Cao, W. Li, J. Wang, S. Wang, Recent

development in membranes for efficient hydrogen purification, J. Membr. Sci. 495

(2015) 130–168.

[11] J. Zou, W.S.W. Ho, CO2-selective polymeric membranes containing amines in

crosslinked poly (vinyl alcohol), J. Membr. Sci. 286 (2006) 310–321.

[12] J. Huang, J. Zou, W.S.W. Ho, Carbon dioxide capture using a CO2-selective

facilitated transport membrane, Ind. Eng. Chem. Res. 47 (2008) 1261–1267.

[13] R. Xing, W.S.W. Ho, Crosslinked polyvinylalcohol-polysiloxane/fumed silica mixed

matrix membranes containing amines for CO2/H2 separation, J. Membr. Sci. 367

(2011) 91–102.

[14] H. Bai, W.S.W. Ho, Carbon dioxide-selective membranes for high-pressure synthesis

gas purification, Ind. Eng. Chem. Res. 50 (2011) 12152–12161.

[15] Y. Zhao, W.S.W. Ho, Steric hindrance effect on amine demonstrated in solid polymer

membranes for CO2 transport, J. Membr. Sci. 415 (2012) 132–138.

[16] Y. Zhao, B.T. Jung, L. Ansaloni, W.S.W. Ho, Multiwalled carbon nanotube mixed

matrix membranes containing amines for high pressure CO2/H2 separation, J. Membr.

Sci. 459 (2014) 233–243.

[17] L. Ansaloni, Y. Zhao, B.T. Jung, K. Ramasubramanian, M.G. Baschetti, W.S.W. Ho,

Facilitated transport membranes containing amino-functionalized multi-walled

carbon nanotubes for high-pressure CO2 separations, J. Membr. Sci. 490 (2015) 18–

28.

[18] V. Vakharia, K. Ramasubramanian, W.S.W. Ho, An experimental and modeling

study of CO2-selective membranes for IGCC syngas purification, J. Membr. Sci. 488

(2015) 56–66.

[19] W. Salim, W.S.W. Ho, Recent developments on nanostructured polymer-based

membranes, Current opinion in Chem. Eng. 8 (2015) 76–82.

[20] Z. Tong, V. Vakharia, M. Gasda, W.S.W. Ho, Water vapor and CO2 transport

through amine-containing facilitated transport membranes, React. Funct. Polym. 86

(2015) 111–116.

[21] H. Lepaumier, D. Picq, P.L. Carrette, New amines for CO2 capture. II. oxidative

degradation mechanisms, Ind. Eng. Chem. Res. 48 (2009) 9068–9075.

[22] A. Bello, R.O. Idem, Comprehensive study of the kinetics of the oxidative

degradation of CO2 loaded and concentrated aqueous monoethanolamine (MEA)

with and without sodium metavanadate during CO2 absorption from flue gas, Ind.

Eng. Chem. Res. 45 (2006) 2569–2579.

[23] S.A. Didas, R. Zhu, N.A. Brunelli, D.S. Sholl, C.W. Jones, Thermal, oxidative and

CO2 induced degradation of primary amines used for CO2 capture: effect of alkyl

linker on stability, J. Phys. Chem. C 118 (2014) 12302–12311.

[24] A. Heydari-Gorji, A. Sayari, Thermal, oxidative, and CO2-induced degradation of

supported polyethylenimine adsorbents, Ind. Eng. Chem. Res. 51 (2012) 6887–6894.

[25] Z. Tong, CO2 facilitated transport membranes for hydrogen purification and flue gas

carbon capture (Ph.D. dissertation), The Ohio State University (2017).

[26] R. Quinn, G.P. Pez, J.B. Appleby, Molten salt hydrate membranes for the separation

of gases, U.S. Patent 4,780,114, 1988.

[27] R. Quinn, D.V. Laciak, J.B. Appleby, G.P. Pez, Polyelectrolyte membranes for

separation of acid gases, U.S. Patent 5,336,298, 1994.

[28] R. Quinn, D.V. Laciak, G.P. Pez, Polyelectrolyte-salt blend membranes for acid gas

separation, J. Membr. Sci. 131 (1997) 61–69.

[29] R. Quinn, D.V. Laciak, G.P. Pez, Process for separating acid gases from gaseous

mixtures utilizing composite membranes formed from salt-polymer blends, U.S.

Patent 6,315, 968, 2001.

[30] L. Xiong, S. Gu, K.O. Jensen, Y.S. Yan, Facilitated transport in hydroxide-exchange

membranes for post-combustion CO2 separation, Chem. Sus. Chem. Com. 7 (2014)

114–116.

[31] S. Gu, J. Wang, B. Zhang, R.B. Kaspar, Y. Yan, Hydroxide exchange membranes

and ionomers, in: B. Ladewig, S.P. Jiang, Y. Yan, M. Lu (Eds.), Materials for Low

Temperature Fuel Cells, Wiley-VCH, New York, 2015, pp. 130–134.

[32] J.F. Brennecke, E.J. Maginn, Ionic liquids: innovative fluids for chemical processing,

AIChE J. 47 (2001) 2384–2389.

[33] D. Chinn, D. Vu, M. Driver, L. Boudreau, CO2 removal from gas using ionic liquid

absorbents, U.S. Patent 0,129,598, 2005.

[34] M.M. Sharma, P. V. Danckwerts, Catalysis by Brönsted bases of the reaction between

CO2 and water, Transactions of the Faraday Society 59 (1963) 386–395.

[35] P.A. Giguere, S. Turrell, The nature of hydrofluoric acid. A spectroscopic study of

+ - the proton-transfer complex H3O .F , J. Am. Chem. Soc. 102 (1980) 5473–5477.

[36] R.J. Lander, J.A. Quinn, The use of membranes in studies of reaction kinetics:

Arsenite catalysis of CO2 hydration, J. Membr. Sci. 3 (1978) 47–56.

[37] M.M. Sharma, P. V. Danckwerts, Fast reactions of CO2 in alkaline solutions – (a)

Carbonate buffers with arsenite, formaldehyde and hypochlorite as catalysts (b)

Aqueous monoisopropanolamine (1-amino-2-propanol) solutions, Chem. Eng. Sci.

18 (1963) 729–735.

[38] D.T. Phan, M. Maeder, R.C. Burns, G. Puxty, Catalysis of CO2 absorption in

aqueous solution by inorganic oxoanions and their application to post combustion

capture.

[39] U. Ghosh, S. Kentish, G. Stevens, “Absorption of carbon dioxide into aqueous

potassium carbonate promoted by boric acid”, Energy Pro. 1 (2009) 1075–1081.

[40] A. G. Eickmeyer, Method and compositions for removing acid gases from gaseous

mixtures, U.S. Patent 4,271,132, 1981.

[41] J. Field, Separation of CO2 from gas mixtures, U.S. Patent 3,907,969, 1975.

[42] V. Vakharia, W. Salim, M. Gasda, W.S.W. Ho, Oxidatively stable membranes for

CO2 separation and H2 purification, J. Membr. Sci. 533 (2017) 220–228.

[43] W.S.W. Ho, W. Salim, V. Vakharia, Membranes for gas separation, U.S. Patent No.

20170056839 A1, 2017.

[44] W.S.W. Ho, V. Vakharia, and W. Salim, Borate-containing membranes for gas

separations, U.S. Patent No. 62416434, 2016.

[45] D. Wu, L. Zhao, V.K. Vakharia, W. Salim, W.S.W. Ho, Synthesis and

characterization of nanoporous polyethersulfone membrane as support for composite

membrane in CO2 separation: From lab to pilot scale, J. Membr. Sci. 510 (2016) 58–

71.

Chapter 3. Scale-up of Amine-containing Membranes for H2 Purification for Fuel Cells

3.1. Summary

Fabrication of amine-containing membranes for removal of CO2 and H2S from reformate for hydrogen purification for fuel cells was scaled-up by using a continuous roll- to-roll membrane fabrication machine. The membrane contained aminoisobutyric acid- potassium salt and polyvinylamine as the carriers for facilitated transport of acid gases and crosslinked polyvinylalcohol as the membrane matrix. Three key variables that controlled the selective-layer membrane thickness were identified and studied, including the coating solution, substrate web coating speed, and coating knife gap setting. Membranes with 14- in wide and > 150-ft long were fabricated with a uniform selective-layer thickness of around 15 microns. The scale-up membranes demonstrated similar performances as the lab-scale membranes with a CO2 permeance of more than 200 GPU, CO2/H2 selectivity of greater than 200, and H2S/H2 selectivity of higher than 600. The scale-up membranes were used for the fabrication of prototype spiral-wound membrane modules for a field test with autothermal reformate gas as the feed gas. The modeling and optimization of the hydrogen purification process were performed to identify the optimum test condition for the spiral- wound membrane modules. The modeling results showed that the membrane removed H2S to less than 10 ppm and CO2 to <1% in the hydrogen product with more than 99% of H2 recovery for the case of using a wet sweep gas with 40% moisture. The field test results showed that the H2S and CO2 concentrations were about 4 – 10 ppm and <3%, respectively,

54 in the hydrogen product on a wet basis and that the agreement between the modeling results and the test results was reasonably well.

3.2. Introduction

Fuel cell is an attractive technology that can provide electricity with high fuel efficiency in various scales due to its inherent modularity [1]. Therefore, fuel cells can be used as the power source in places with limited space such as a ship with ease of installation and maintenance, reduced heat, noise, and air requirements as compared to traditional power generation modules [2]. Hydrogen is generally considered as the preferred fuel for the current high-efficiency fuel cells, such as proton-exchange membrane fuel cells (PEMFC), because of its high reactivity and clean emission. The hydrogen fuel can be produced by reforming hydrocarbon fuels and then followed by the water gas shift (WGS) reaction.

Hydrocarbon fuels are reformed to produce reformate gas that contains H2, CO, CO2, and

H2S. The water gas shift reaction further converts CO to H2 as shown in reaction (1).

CO + H2O ⇌ CO2 + H2 (1)

The impurities that remain in the hydrogen fuel stream produced by the reforming step, such as CO and H2S, are proven to be harmful to the fuel cells, including the loss of operation due to contamination and the shortened fuel cell life due to deterioration of the platinum electrode. It is necessary to minimize all impurities in the hydrogen reformate stream to the required levels, notably H2S to < 10 ppm and CO to < 100 ppm, which are sufficient for the subsequent high temperature PEMFC [3]. Therefore, a robust purifier that is capable of continuously treating the hydrogen reformate stream and producing the high purity hydrogen fuel for PEMFC needs to be developed. In addition, special requirements

55 in niche applications, such as shipboard fuel cell applications, also needs to be fulfilled, including a minimum system size, modular packaging, operational and maintenance simplicity, and minimum cost.

Membrane systems have the potential for the continuous removal of CO2, CO, and H2S in a shipboard environment that has space constraints, shock requirements, and fluctuating hydrogen demands. Both CO2-selective and H2-selective membranes can be used for hydrogen purification; however, highly CO2-selective membranes have the advantage of ability to separate the CO2 and H2S while retaining hydrogen and CO for the use in the

WGS reaction. In other words, the H2 loss is minimized and the H2 pressure is maintained during the separation process, which can also aid in reaching almost 100% CO conversion in an otherwise equilibrium-dominated low-temperature WGS reaction.

Polymeric dense membranes generally follow the solution-diffusion mechanism for the gas transport through the membrane. This mechanism relies on the physical and chemical properties of the polymers, including the solubility of a specific gas and its diffusivity through the polymer matrix of the membrane. The separation of gases depends on the molecular sizes of the gases and the chemical interactions between the gases and the polymers. Hao et al. performed a process design and economics of membrane stages without recycle streams by using both CO2-selective (fluorine-containing polyimide

6FDA-HAB) and H2S-selective membrane (poly (ether urethane urea)) [4]. Merkel and

Toy compared the H2S transport properties in fluorinated and nonfluorinated polymers and concluded that fluorinated polymers showed significantly greater resistance to H2S transport than an analogous nonfluorinated polymers [5]. Recently, Lin et al. developed

o Polaris™ membranes and demonstrated H2S/CO2 selectivity around 3 at 30 C from a syngas containing 320 ppm of H2S [6,7].

Facilitated transport membranes offer enhanced selectivity as well as increased permeance due to the existence of carriers, which can effectively transport acid gases by reacting reversibly with the acid gases. One of the promising carriers for facilitated transport is amine, which can react reversibly with both CO2 and H2S according to reactions

(2) and (3) and reaction (4), respectively.

+  CO2 + 2 R-NH2 ⇌ R-NH3 + R-NH-COO (2)

+  CO2 + R-NH2 + H2O ⇌ R-NH3 + HCO3 (3)

+  H2S + R-NH2 ⇌ R-NH3 + HS (4)

George et al. [8] recently summarized the results of various polymer membranes used for acid gas removal from natural gas to meet the United States pipeline specifications of

4 ppm H2S and 2% CO2 [9,10]. They concluded that current efforts should focus on the improvement of the membrane performance along with physical and chemical stability of the membrane. Quinn et al. developed polyelectrolyte-salt membranes for acid gas separations and utilized fluoride ions to facilitate CO2 transport [11]. Uddin and Hägg used the fixed-site carrier facilitated transport mechanism in polyvinylamine/polyvinylalcohol blend membrane and reported that no permanent damage was observed during the exposure to aggressive environment including 1 mole% H2S and condensable hydrocarbons including n-hexane and propane [12].

Ho and coworkers developed CO2-selective facilitated transport membranes that comprised amines as the CO2 carriers for its facilitated transport [13–22]. A water-gas-shift

57 membrane reactor with a CO2-selective membrane was studied for fuel cell hydrogen processing [13]. A CO2-selective membrane comprising 2-aminoisobutyric acid-potassium salt as the mobile carrier and polyallylamine as the fixed-site carrier was developed for

CO2 capture [14]. Comparison between membranes based on polyvinylalcohol and sulfonated polybenzimidazole as the membrane matrixes for high pressure synthesis gas purification at 220 psia was performed [15]. The polyvinylalcohol-based membranes demonstrated better performance at a higher temperature of 106oC compared to the sulfonated polybenzimidazole-based membrane. Multi-walled carbon nanotubes were incorporated to form mixed-matrix membranes to resolve the membrane compaction issue during high pressure CO2/H2 separation at more than 1.52 MPa [16]. In addition, experimental and modeling study of CO2-selective membranes for IGCC syngas purification was conducted [17]. Ramasubramanian and Ho synthesized a CO2-selective membrane that showed high CO2 and H2S permeances and CO2 and H2S selectivities vs.

H2, N2, and CO [18]. The membrane utilized 2-aminoisobutyric acid potassium and polyvinylamine as the mobile and fixed-site CO2 carriers, respectively.

Despite the recent significant advances in membrane scale-up and field testing [23–

30], the details of the scale-up membrane fabrication have not been well documented. In this study, transition from lab-scale membrane synthesis to a continuous roll-to-roll fabrication of membrane scale-up was performed for the CO2-selective membrane proposed by Ramasubramanian and Ho [18]. The key variables that affected the quality of the scale-up membrane was identified and optimized. For testing with the spiral-wound membrane modules fabricated from the scale-up membrane, modeling for process

58 optimization was conducted to identify optimal process conditions, including feed gas composition, sweep-to-feed gas flow rate ratio, and feed and sweep water contents.

Comparison between the modeling results and the results from the membrane module testing was also made.

The scale-up membranes fabricated in this work are suitable for the removal of hydrogen sulfide and carbon dioxide from synthesis gas using steam or nitrogen as the sweep gas or pulling vacuum on the permeate side, allowing the subsequent water-gas-shift reaction to convert any remaining carbon monoxide in the synthesis gas to generate more hydrogen. The membrane technology is critical since it allows an on-demand, low-cost, and robust source of high-purity hydrogen fuel for fuel cell applications. In addition, this technology can be extended for purifying hydrogen in chemical industry facilities, capturing CO2 emissions from power plants and other manufacturing facilities, and assisting the industries in meeting future stringent control of CO2 greenhouse gas.

3.3. Experimental

3.3.1. Materials

Lupamin® 9095, a polyvinylamine (PVAm)-based commercial product, was kindly donated by BASF AG (Wyandotte, MI, USA). Kuraray polyvinylalcohol 217SB (PVA, 94 wt.%, 87 – 89% hydrolysis degree) was kindly donated by Kuraray America Inc. (Houston,

TX, USA). 2-aminoisobutyric acid (AIBA, 99.46%) was purchased from Chem-Impex

International (Wood Dale, IL, USA). Glutaraldehyde (GA, 50 wt.% aqueous solution),

59 potassium hydroxide (KOH, 85%), and ethylene diamine (EDA, 99%) were bought from

Sigma Aldrich (Milwaukee, WI, USA). Triethylenetetraamine (TETA, 60%) was acquired from Fisher Scientiﬁc (Bridgewater, NJ, USA). All the chemicals were used without further purification.

The nanoporous polysulfone support used in the membrane fabrication was purchased from TriSep Corporation (Goleta, CA, USA). The specialty feed gas mixture with a certified composition of 20% CO2, 40% H2, 40% N2, and 50 ppm H2S on a dry basis and the pre-purified argon used as sweep gas, carrier gas, and reference gas for gas chromatography were bought from Praxair Inc. (Danbury, CT, USA) for gas permeation measurements. Pre-purified helium and oxygen used for the sulfur chemiluminescent detector (SCD) for H2S determination were also bought from Praxair Inc.

3.3.2. Pilot-scale membrane fabrication

The membrane developed in this work consisted of a dense polymeric selective-layer on a nanoporous substrate as shown in Figure 3.1. A polysulfone layer with a pore size of about 9 nm and a porosity of about 7% on a nonwoven fabric (from TriSep Corporation) was used as the nanoporous substrate. For the scale-up membrane fabrication work, a polysulfone substrate roll with around 14-inch in width and 500 feet in length was used.

The dense polymeric selective layer was obtained after a uniform wet film of coating solution was dried and cured at 130oC for about 16 min (for a coating speed of 0.5 ft/min).

Selective-layer (15 μm) Nanoporous polysulfone (50 μm)

Nonwoven fabric (100 μm)

Figure 3.1. The schematic of the structure of scale-up membranes for H2 purification.

The coating solution used in the scale-up fabrication work was prepared by adding a carrier solution to a crosslinked PVA solution. The crosslinked PVA solution was prepared by using the following procedure. Firstly, the PVA 217SB was dissolved in water and heated to 80oC for 2 h with continuous stirring to obtain a homogeneous solution with a concentration of 10 wt.%. To the PVA solution at 80oC under stirring was added the KOH aqueous solution with a concentration of 35 wt.% dropwise, and the solution was mixed continuously for 30 min. The KOH served as the catalyst for the crosslinking reaction of polyvinylalcohol by glutaraldehyde. The crosslinking reaction was started by adding the glutaraldehyde (GA) solution dropwise at 80oC under vigorous stirring. The crosslinking reaction was carried out for about 150 min and aimed at a molar crosslinking degree of

15%. The total solid content in the crosslinked PVA solution was aimed at 14%, consisting of 11% crosslinked PVA and 3% KOH.

The carrier solution consisted of a mixture of AIBA-K and commercial PVAm solution

(Lupamin® 9095). The AIBA-K solution was prepared by neutralizing AIBA solution with

KOH at room temperature. The carrier solution was prepared by mixing Lupamin® 9095 with the AIBA-K solution. The composition of the membrane consisted of 31 wt.% crosslinked PVA, 20 wt.% Lupamin® 9095, 23 wt.% AIBA-K, 17 wt.% KOH, and 9 wt.%

EDA.

The coating solution for the scale-up membrane fabrication was prepared by adding the carrier solution dropwise to the calculated amount of the crosslinked PVA solution under stirring. The total solid concentration in the coating solution was required to be low enough to avoid instantaneous gelling of the coating solution. However, a coating solution with a high viscosity was desirable to minimize the penetration of the coating solution into the pores of the substrate. Thus, the coating solution used contained 17 – 20 wt.% total solid content. The coating solution was centrifuged at 8000 rpm for 3 min to remove air bubbles before using it in scale-up membrane fabrication.

The scale-up fabrication of the membrane was demonstrated by using the pilot-scale thin- film-coating (TFC) assembly of the continuous roll-to-roll machine at The Ohio State

University. The schematic representation of the thin-film coating assembly used for the pilot-scale membrane fabrication is shown in Figure 3.2. The coating machine was equipped with a control panel, which included: 1) Human Machine Interface (HMI), a touch-screen panel that user could use to input the parameters for machine control, 2) Drive

“Start” button, 3) Drive “Stop” button, 4) Drive “Jog” button, and 5) “Emergency Stop” button. The parameters that could be set with the HMI were the substrate web tension and

62 substrate web coating speed with the controllable ranges of 0 – 20 lbf and 0 – 5 ft/min, respectively. The length of the convection oven was 8-ft; hence, the residence time of the membrane inside the convection oven was 16 minutes if a coating speed of 0.5 ft/min was used.

Hot Air Hot Air In Out

Convection Oven Coating Unwind Knife Roller

Rewind Roller

Figure 3.2. Schematic representation of pilot-scale thin-film-coating (TFC) assembly of the coating machine for the scale-up fabrication of membranes for H2 purification.

The setup of the thin-film-coating assembly of the coating machine for the scale-up fabrication of membranes is shown in Figure 3.3(a). The polysulfone substrate roll at the unwind roll was rotated by the unwind motor to roll the 14-inch wide polysulfone substrate web onto the coating knife assembly. The coating solution was stored in the coating trough chamber between the coating knife and the back plate. The thickness of the membrane was

63 controlled by setting the gap between the coating knife and the base of the coating knife assembly (gap setting, as shown in Figure 3.3(b)) as well as the coating speed of the polysulfone substrate roll. The membrane was dried and cured at 130oC for 16 minutes inside the convection oven by flowing hot air into the oven for a complete removal of water and a complete crosslinking reaction of PVA with glutaraldehyde. The tension used to flatten the polysulfone substrate rolls was 18 lbf.

Oven Coating Trough

Unwind Roll

Coating Knife Wet Film Coating TFC Assembly Rewind Gap Setting Roll Substrate Motion

(a) (b) Figure 3.3. The pilot-scale coating machine for the scale-up fabrication of membranes: (a) machine setup at OSU and (b) the schematic of coating knife assembly.

The empirical equation (Eq. 5) was used to determine the gap setting of the coating knife in order to control the final thickness of the dry membrane.

 ρdry = 0.5 c ρ gap (5)

3 where  (nm) is the dry membrane thickness, ρdry (g/cm ) is the density of the dry membrane, ρ (g/cm3) is the density of the coating solution, c (by weight) is the total solids concentration of the coating solution, and gap (nm) is the gap setting of the coating knife.

The thickness of the selective layer of the scale-up membrane was measured by a Mitutoyo electronic indicator Model 543-252B (Mitutoyo America Corp, Aurora, IL, USA) with an accuracy of ± 0.5μm. The selective layer thickness was calculated by subtracting the polysulfone/nonwoven fabric support thickness from the total thickness of the composite membrane. The polysulfone/nonwoven fabric support as well as the total composite membrane thickness were measured at 20 different points on each of the membrane samples. The thickness of the selective layer was controlled at around 15 μm, unless otherwise noted.

3.3.3. Membrane gas transport measurement

The schematic of the gas permeation unit used to measure the transport performance of the scale-up membrane is shown in Figure 3.4. The unit consisted of mass flow controllers, water pumps, and humidifiers to simulate the actual reformate gas composition in the feed gas. The feed and sweep gases entered the membrane cell in a countercurrent configuration. The membrane was placed in a rectangular membrane cell between the upper and lower parts. An active membrane area of 3.4 cm2 was used for the gas transport performance measurements.

During the test, the dry feed gas flow rate was typically 60 sccm at 2 atm, and the dry sweep gas flow rate was typically 30 sccm at around 1 atm. Both the feed and sweep gases were humidified with water vapor by injecting controlled amounts of water into the two respective humidifiers. At the test temperature of 106oC, the water contents were maintained at 40% and 57% for the feed and sweep sides, respectively, by pumping 0.03 cc/min of water into each of the feed and sweep sides. The retentate and permeate streams flowed out of the permeation cell to the water knock-out vessels to remove most of the water in the gas streams. Afterwards, the retentate and permeate gas streams were directed to the Agilent 6890N gas chromatography (GC) which was equipped with a stainless steel micropacked column Carboxen 1004 with 80/100 mesh size (Supelco, Bellefonte, CA,

USA) and a thermal conductivity detector (TCD) for gas composition analysis. The gas transport performance of the membrane, in terms of permeability Pi and selectivity αij, was evaluated by using the gas compositions obtained from the GC in the flux equation as follows:

yi / y j Ji P  ij  i p /  x / x i i j

( p  p )  ( p  p ) p  i, feed in i,sweep out i, feed out i,sweep in i ln( p  p )  ln( p  p ) i, feed in i,sweep out i, feed out i,sweep in

where i denotes the gas component CO2 and j denotes another gas component (H2, CO, or

H2S); y and x are the mole fractions of gas components in the sweep and feed sides of the

66 membrane, respectively; Ji is the steady-state CO2 molar flux through the membrane; is the selective-layer membrane thickness; Δpi is the partial pressure difference between the feed and sweep sides and defined by using the logarithmic mean method.

-10 The common unit of the permeability (Pi) is Barrer, which is equal to 10 cm3(STP)·cm / (cm2·s ·cmHg) = 3.34 x 10-16 mol·m / (m2·s·Pa). The permeance of a gas component i is defined as the ratio of the permeability of gas i to the selective-layer membrane thickness, i.e., Pi/, and its common unit is the gas permeation unit (GPU), which is equal to 10-6 cm3(STP) / (cm2·s·cmHg).

Sweep Feed Gas Gas

Back Pressure Regulator

Figure 3.4. Schematic representation of the gas permeation unit for the membrane gas transport measurements.

3.4. Results and discussion

3.4.1. Optimization of scale-up membrane fabrication parameters

In the previous lab-scale work, Ramasubramanian demonstrated that a membrane with a selective-layer thickness of 15 µm resulted in a CO2 permeance of > 200 GPU, CO2/H2

68 selectivity of > 200, and H2S/CO2 selectivity of 3 [18]. Therefore, this work aimed at the continuous membrane fabrication with around 15 µm of selective-layer thickness. Three parameters that were considered as the key variables to affect the selective-layer membrane thickness in the continuous roll-to-roll fabrication were the coating solution, substrate web coating speed, and coating knife gap setting [31–35].

3.4.1.1. Effect of coating solution on membrane thickness

After incorporating all the components of the membrane, the viscosity of the coating solution started to increase due to the interaction between the amino groups in Lupamin® in the carrier solution and the aldehyde groups in the crosslinked PVA solution, which led to gel formation after 10 minutes, as shown in Figure 3.5. Small amine molecules, such as

EDA or TETA (9 wt.% of the total solids), were added to the crosslinked PVA solution prior to the mixing with the carrier solution in order to slow down the kinetics of gelling.

This figure shows a pronounced effect of these two additives on the coating solution viscosities and the comparison of viscosity increase over time for the coating solutions with

TETA and EDA. TETA and EDA are much smaller molecules compared to the PVAm macromolecules in Lupamin®, and they were believed to preferentially interact with the aldehyde groups prior to the addition of the carrier solution. The lesser amount of free aldehyde groups, due to aldehyde and EDA/TETA interaction, was believed to reduce the interaction of aldehyde groups with the PVAm and therefore reduced the increasing rate of the coating solution viscosity after the addition of the carrier solution. The coating solutions containing TETA or EDA were still able to be used for the coating of scale-up

69 membrane at least 1 hour after the coating solution was prepared. The increasing rate of the coating solution viscosity with EDA was slower compared to the one with TETA due to the more molecules of EDA could be incorporated into the solution for the same basis of 9 wt.% of the total solids in the membrane. Since the coating solution containing EDA showed a slower increasing rate of the coating solution viscosity compared to the one with

TETA, the subsequent work focused on using the coating solution with EDA.

1200

1000

No additive TETA as )

800 (gel formation additive cP in 10 minutes) 600

400

Shear Viscosity ( Viscosity Shear 200 EDA as additive

0 0 10 20 30 40 50 60 70 Time (minutes)

Figure 3.5. Effect of small amine molecules (TETA or EDA) on the increasing rate coating solution viscosity.

The effect of coating solution on the selective-layer membrane thickness is shown in

Figure 3.6. The coating knife gap setting was maintained at 8 mils as used in the lab-scale membrane preparation. The substrate web coating speed was maintained at 0.5 ft/min for these experiments. The selective-layer membrane thickness increased with increasing 70 coating solution concentration, indicating that the coating solution concentration was one of the key variables to be optimized.

A higher coating solution concentration resulted in a higher coating solution viscosity and helped in reducing the penetration of the coating solution into the pores of the polysulfone substrate. However, too high of the coating solution concentration caused the gelling of the coating solution and nonuniform coating as marked in Figure 3.6 with the insert showing a non-uniform gelled coating. In that case, it was not feasible to use the coating solution to form a uniform coating on the polysulfone substrate. In order to reduce coating solution penetration and avoid gel formation, a coating solution concentration of

20 wt.% was selected and used in the subsequent experiments.

24 Gelled Coating 22 Solution

layer thickness (microns) thickness layer - 14

12 Selective

10 15 16 17 18 19 20 21 22 Coating solution concentration (wt.%)

Figure 3.6. Effect of coating solution concentration on selective-layer membrane thickness.

3.4.1.2. Effect of coating speed on membrane thickness

The effect of substrate web coating speed on the selective-layer membrane thickness is shown in Figure 3.7. The coating knife gap setting was maintained at 8 mils as used in lab- scale membrane preparation. The coating solution at 20 wt.% was used for these experiments. The selective-layer membrane thickness reduced with increasing substrate web coating speed, indicating that the substrate web coating speed is one of the key variables that should be optimized. The thickness reduction with the coating speed was presumably due to the case that the solution delivery rate could not keep up with the coating speed.

The coating machine at OSU could be operated up to 5 ft/min of substrate web coating speed. A higher coating speed resulted in higher efficiency of the scale-up membrane fabrication as more membranes can be fabricated at a certain amount of time. However, a higher coating speed resulted in reduced residence time of the membrane in the curing oven and might lead to insufficient drying and curing of the membrane. On the other hand, a substrate web coating speed less than 0.5 ft/min resulted in the chattering of the substrate web and non-uniform coating along the membrane length as shown in the insert in Figure

3.7. In order to eliminate any chattering of the substrate web, a substrate web coating speed of 0.5 ft/min was selected and used in the subsequent experiments. The maximum path length in the curing oven of the coating machine was 8-ft, resulted in a maximum residence time of 16 min for a substrate web coating speed of 0.5 ft/min.

layer thickness (microns) thickness layer - 14 Chattering Substrate

12 Web Selective 10 0 1 2 3 4 5 6 Substrate web coating speed (ft/min)

Figure 3.7. Effect of substrate web coating speed on selective-layer membrane thickness.

3.4.1.3. Effect of coating knife gap setting on membrane thickness

The effect of coating knife gap setting on the selective-layer membrane thickness is shown in Figure 3.8. The substrate web coating speed was varied at 0.5, 1, 2, 3, 4, and 5 ft/min and the coating solution was at a concentration of 20 wt.% for these experiments.

The selective-layer membrane thickness reduced with decreasing coating knife gap setting.

A coating knife gap setting of 6 mils was required to obtain a selective-layer membrane thickness of around 15 µm when the 0.5 ft/min of substrate web coating speed was used.

24 Gap setting 6 mils 22 6.5 mils 20 7 mils

16 layer thickness (microns) thickness layer - 14

12 Selective 10 0 1 2 3 4 5 6 Substrate web coating speed (ft/min)

Figure 3.8. Effect of coating knife gap setting on selective-layer membrane thickness.

3.4.2. Scale-up membrane fabrication and test at OSU

In order to fulfill one of the tasks of the project (N00014-11-C-0062) supported by the

Office of Naval Research, five scale-up runs were performed and a total of 155-ft long and

14-in wide of scale-up membranes were fabricated. Figure 3.9 shows the images of the scale-up membrane before (immediately after the coating knife station) and after the curing in the oven. The scale-up membranes appeared to have a uniform coverage of the selective- layer on the polysulfone substrate. For these five runs, Table 3.1 lists the lengths of the scale-up membrane and the selective-layer membrane thicknesses. As seen from this table, longer lengths of the membranes were fabricated per batch in the latter runs by synthesizing multiple batches of the coating solution for each run. The average membrane thickness

74 was 15.48 µm, which was close to the targeted selective-layer membrane thickness. The scale-up membrane with a total length of 155 ft was handed to DJW Technology (Dublin,

OH, USA) for the subsequent work on spiral-wound membrane module fabrication and testing.

(a) (b) Figure 3.9. The images of the scale-up membrane before (a) and after (b) the curing in oven.

Table 3.1. Five runs of the prototype membrane (14 inches) fabricated by using the continuous membrane fabrication machine at OSUa.

Run number 1 2 3 4 5

Length (ft) 20 25 30 35 45 Membrane thickness (µm) 15.7 15.3 15.8 15.4 15.2

CO2 permeance (GPU) - - 203 - 205

CO2/H2 selectivity - - 248 - 220

H2S/CO2 selectivity - - 3.3 - 2.8 a Coating speed = 0.5 ft/min, gap setting = 6 mils, coating solution = 20 wt.%.

The gas transport properties of the scale-up membrane were measured at 106oC and 2 atm feed pressure using a simulated synthesis gas consisting of 20% CO2, 40% H2, 40%

N2, and 50 ppm H2S on dry basis. Both the feed and sweep gases were humidified to have about 40% steam and 57% steam, respectively. Table 3.1 also lists the gas transport results measured from membrane samples taken from Runs 3 and 5, which had the highest and lowest average of selective-layer membrane thickness, respectively.

As shown in Table 3.1, the gas transport properties of the scale-up membranes were good with a high CO2 permeance of 203 – 205 GPU and a very high CO2/H2 selectivity of

220 – 248. These results were consistent with those obtained previously from similar membranes that were fabricated in lab-scale [18]. Figure 3.10 shows the stability of the membrane sample taken from scale-up run No. 5. For 30 hours of test, the membrane exhibited a reasonably stable membrane performance at 205 GPU of CO2 permeance and

CO2/H2 selectivity of 220.

250 350

CO2 Permeance 300 200 250

150

200 Selectivity

150 2

100

2 Permeance (GPU) Permeance 100 CO 50 50 H2 Permeance (x100)

0 0 0 5 10 15 20 25

Time (hours)

Figure 3.10. The stability plot of membrane sample taken from scale-up run No. 5.

3.4.3. Membrane module fabrication

The scale-up membrane fabricated in this work was delivered to DJW Technology and was rolled into spiral-wound membrane modules for a field test at Precision Combustion

Inc. (PCI, North Haven, CT, USA). Figure 3.11 shows the image of the spiral-wound membrane modules that contained the CO2-selective membrane developed in this work.

The purpose of the spiral-wound membrane module skid test was to demonstrate the membrane performance to achieve less than 10 ppm H2S in the purified H2 stream

(retentate) by using an industrial reformate gas. An autothermal reformer at PCI provided a reformate gas stream of 53 slpm with a composition of 6.2% CO2, 43.2% H2, 32.1% N2,

18.7% CO, and 57 ppm H2S on dry basis as the feed gas into the spiral-wound membrane modules. The pressure of the reformate gas stream was expected to be around 1 barg.

Nitrogen was used as the sweep gas with a sweep-to-feed flow rate ratio of more than 1:1.

The sweep gas was humidified to obtain a water content of up to 40% on the sweep side.

Based on the conditions available at PCI, a modeling and process optimization work was performed to support the membrane module skid test at PCI as described next.

Figure 3.11. The image of the spiral-wound membrane module that contained the CO2- selective membrane developed in this work.

3.4.4. Modeling and process optimization

The modeling of the membrane module performance for the CO2 and H2S removal was conducted to study the effects of sweep water content, sweep-to-feed flow rate ratio, and feed water content. This modeling determined the design operating conditions for the spiral-wound membrane module testing. Finally, the modeling results were compared to the experimental results obtained from the field test at PCI.

Gas transport results of the scale-up membrane in this work were used. Based on the experimental results of the scale-up membranes, the performances of the membrane used in the calculations were a CO2 permeance of 200 GPU, a CO2/H2 selectivity of 200, and a

H2S/CO2 selectivity of 3. Two different feed gas pressures of 5 and 10 psig were considered for the modeling while the sweep pressure of 1 atm was used. The feed gas inlet flow rate used in the modeling was 53 slpm and the sweep gas inlet flow rate was up to 223 slpm; both flow rates were on wet basis. The modules were arranged in a parallel configuration and 2 modules were considered with a 32 ft2 of membrane area for each 4-inch diameter module. Therefore, 50% of the design flow rates for the feed and sweep sides were used for each module. The dry feed gas composition considered for the modeling consisted of

6.2% CO2, 43.2% H2, 32.1% N2, 18.7% CO, and 57 ppm H2S, which was based on the composition of the feed gas on dry basis for the ATR (autothermal reforming) reformate from PCI. The moisture content of 12.7% on the feed side was considered, unless otherwise noted.

3.4.4.1. Effect of sweep-to-feed flow rate ratio and sweep water content

Various cases were evaluated by varying the sweep-to-feed flow rate ratio and the water content on the sweep side. The ratios of the sweep-to-feed flow rate were set as 4.2:1

(based on 100% design flow), 3:1, 2:1, and 1:1; both flow rates were on wet basis. The water content on the sweep side were set at 0, 10, 20, 22, 30, and 40%.

The CO2 removal, H2S removal, H2 recovery, retentate H2S concentration, and retentate water content for a feed pressure of 10 psig with different sweep-to-feed flow rate ratios (1

– 4.2) and sweep water contents (0 – 40%) are shown in Figure 3.12(a) – (e), respectively.

The corresponding plots for a feed pressure of 5 psig are provided as a comparison in

Appendix B. Figure 3.12(a) and Figure 3.12(b) showed that both CO2 removal and H2S removal increased with increasing sweep-to-feed flow rate ratio, decreasing sweep water content, and increasing feed gas pressure. Figure 3.12(c) showed that the H2 recovery increased with increasing sweep water content and decreasing feed gas pressure; while no effect was observed when the sweep-to-feed flow rate ratio was varied. Figure 3.12(d) showed that the retentate H2S concentration reduced with increasing sweep-to-feed flow rate ratio, decreasing sweep water content, and increasing feed gas pressure. In addition, some observations from the calculation results were as follows: (1) A sweep-to-feed flow rate ratio higher than 2:1 was sufficient to achieve a H2S concentration lower than 10 ppm in the treated gas stream; (2) A CO2 removal higher than 90% could only be reached at a feed pressure of 10 psig and a relatively low sweep water content; (3) At all operating conditions the H2 recovery was higher than 98.5%; (4) Water content in the feed and sweep sides might affect membrane performance. Figure 3.12(e) showed that in order to maintain

80 a relatively constant water content in the feed side, the sweep water content should be at least 17% for 5 psig feed gas and 22% for 10 psig feed gas. A lower sweep water content caused a significant decrease of water content in the feed side; a higher sweep water content resulted in water permeation from the sweep to the feed side.

100%

90%

80%

70%

Removal 2

10 psig feed pressure CO 60% Sweep water content 0% 10% 20% 22% 30% 40% 50%

40% 1 2 3 4 Sweep:Feed Flow Rate Ratio

(a) continued

Figure 3.12 continued

100%

95%

90% S Removal S

2 10 psig feed pressure H Sweep water content 0% 10% 20% 85% 22% 30% 40%

80% 1 2 3 4 Sweep:Feed Flow Rate Ratio

(b)

100%

99%

98% Recovery

2 97% 10 psig feed pressure H Sweep water content 0% 10% 20% 96% 22% 30% 40%

95% 1 2 3 4 Sweep:Feed Flow Rate Ratio

Figure 3.12 continued

10 psig feed pressure 10 Sweep water content 0% 10% 20% 22% 30% 40% 8

6 S Conc., Wet Basis Basis (ppm) WetConc., S 2 4

2 Retentate H Retentate

0 1 2 3 4 Sweep:Feed Flow Rate Ratio

(d)

50%

45% 10 psig feed pressure Sweep water content 40% 0% 10% 20%

35% 22% 30% 40%

30%

25%

20%

15% RetentateWaterContent 10% Close to initial feed water content 5%

0% 1 2 3 4 Sweep:Feed Flow Rate Ratio

(e) 83

3.4.4.2. Effect of feed water content

Figure 3.12(e) indicated that with 40% water in the sweep gas, the water content from feed inlet to retentate outlet increased from 12.7% to about 25% for the sweep to feed flow rate ratio ranging from 1 to 4.2. However, the scale-up membrane performance was obtained by using 40% of water content in the feed gas. In order to investigate the possible reduction of spiral-wound membrane module performance caused by a lower water content on the feed side, modeling cases with 40% water content on both feed and sweep sides were also calculated. When water was injected into the feed gas, the sweep-to-feed flow rate ratio would reduce concomitantly, which might also affect the module performance.

Consequently, two different scenarios were considered for each case: (1) The sweep-to- feed flow rate ratio was kept as the one before injecting water into the feed gas, denoted as

“unfixed” and (2) The sweep flow rate was increased accordingly to keep the sweep-to- feed flow rate ratio as a constant, denoted as “fixed”.

As shown in Figure 3.13, in both cases the H2S removal increased with increasing both sweep-to-feed flow rate ratio and feed gas pressure; while the retentate H2S concentration reduced with increasing both sweep-to-feed flow rate ratio and feed gas pressure. In addition, the calculation results exhibited that pumping water in the feed gas will not deteriorate the membrane module performance significantly, especially for the cases with accordingly increased sweep flow rate (“fixed” case).

2 2.5 3 3.5 4 100% 12

11 95% (2.9) (2.1) (1.4) 10 90% (1.4) 9

85% 8

80% (2.1) 7 S Removal S

2 75% 6

S Conc., Wet Basis Basis (ppm) WetConc., S H

(2.9) 2 5 70% 4 5 psig feed pressure 65% 40% feed moisture 3 40% sweep moisture H Retentate 60% 2 2 2.5 3 3.5 4 Sweep:Feed Flow Rate Ratio

H2SH2S Removal, Removal, unfixed unfixed H2SH2S Removal, Removal, fixed fixed

H2SH2S Conc., Conc., unfixed unfixed H2SH2S Conc., Conc., fixed fixed

(a)

2 2.5 3 3.5 4 100% 12 (2.9) (1.4) (2.1) 95% 11 10 90% 9 85% 8 80% 7

S Removal S 10 psig feed pressure

2 75% 6 S Conc., Basis Conc., Wet S (ppm)

H 40% feed moisture 2 (1.4) 40% sweep moisture 5 70% (2.1) (2.9) 4

65% 3 Retentate H Retentate 60% 2 2 2.5 3 3.5 4 Sweep:Feed Flow Rate Ratio

H2SH2S Removal, Removal, unfixed unfixed H2SH2S Removal, Removal, fixed fixed H2SH S Conc., Conc., unfixed unfixed H2SH S Conc., Conc., fixed fixed 2 2 (b)

Figure 3.13. Modeling results of module performance with 40% water contents in both the feed and sweep gases: (a) 5 psig feed pressure (b) 10 psig feed pressure. The numbers in the parentheses indicate the actual sweep to feed flow rate ratios in the unfixed cases. 85

3.4.4.3. Design operating conditions for membrane module testing

Figure 3.14 shows the calculated module performances with 12.7% and 40% water contents in the feed and sweep gases, respectively, at 106oC. It is important to have a water vapor content of about 40% in the sweep gas stream (a mixture of H2O and N2) to have a good membrane performance for H2S and CO2 removal while maintaining a H2 recovery

> 99%. A sweep-to-feed flow rate ratio of 2:1 was sufficient to achieve a H2S concentration to be lower than the target of 10 ppm in the treated H2 gas stream. Therefore, the design operating conditions for the membrane module test at PCI were determined as follows:

(1) The raw feed gas (6.2% CO2, 43.2% H2, 32.1% N2, 18.7% CO, and 57 ppm H2S on dry basis) at a flow rate of 53 slpm and 10 psig pressure;

(2) The sweep-to-feed flow rate ratio of 2:1 or higher;

(3) The sweep water content of 40%.

1 1.5 2 2.5 3 3.5 4 100% 10

8 90% 7

6 80%

5 S conc., wetconc., S (ppm) basis 4 2 70% 3

40% sweep moisture 2 H Retentate 5 psig feed pressure 60% 1 1 2 3 4 Sweep:Feed Flow Rate Ratio

CO2CO 2removalremoval H2SH2S removal removal H2H recoveryrecovery H2S conc. 2 H2S conc. (a)

1 1.5 2 2.5 3 3.5 4 100% 7

6 90% 5

80% 4 S conc., wet(ppm) conc., basis S 3 2 70% 2

40% sweep moisture H Retentate 10 psig feed pressure 60% 1 1 2 3 4 Sweep:Feed Flow Rate Ratio

CO2CO 2removalremoval H2SH2S removal removal H2H recoveryrecovery H2S conc. 2 H2S conc. (b)

Figure 3.14. Modeling results of module performance with 12.7% and 40% water contents in the feed and sweep gases: (a) 5 psig feed pressure (b) 10 psig feed pressure.

3.4.4.4. Comparison between modeling results and membrane module field test results

DJW Technology, LLC conducted the membrane module testing at PCI and obtained

7 – 8 ppm H2S in the retentate fuel gas from the feed fuel gas containing 32 ppm H2S, meeting the target of less than 10 ppm H2S. However, it was significantly higher than the results obtained from modeling, which should have been around 2 – 5 ppm H2S in the retentate fuel gas from the feed fuel gas containing 57 ppm H2S.

The major differences noticed between the actual test conditions and the design operating conditions were as follows:

(1) The reformate gas H2S concentration was 32 ppm, lower than the 57 ppm of H2S

concentration used in the modeling. The composition of the feed gas from the ATR

(autothermal reforming) reformate at PCI were 1.6% CO2, 17.5% H2, 41.2% N2, 17%

CO, 22.8% H2O, and 32 ppm H2S on wet basis.

(2) The feed and sweep flow rates were different from those used in the modeling case

study.

(3) The sweep nitrogen temperature was lower than the design temperature of 120oC (99oC

and 76oC at 34 slpm and 53 slpm sweep gas, respectively, compared to the feed

temperature of 120oC). The temperature difference between the feed and sweep gases

brought about a temperature variation inside of the module, which would affect the

membrane transport properties. Presumably, the CO2 permeance would drop from 200

to 100 GPU if the operating temperature decreased from 120oC to 76oC.

Therefore, the model calculations were revised to accommodate the changes of feed and sweep gases and to take the temperature effect into account. Table 3.2 lists the cases

88 which were studied to consider the reduced membrane performance at lower temperature.

The governing equations for heat transfer in the module was given by Ramasubramanian et al. [18] and the heat transfer coefficient was set as 4 W/(m2·s) empirically [36]. The permeance-temperature dependence was expressed by an Arrhenius form equation:

2162[K ] P  CO2  48952[GPU ] e T (1)

where P is the CO2 permeability; is the membrane selective layer thickness; T is the CO2 absolute temperature [K]; and P / is the CO2 permeance in GPU. CO2

Table 3.2. Cases considered in the modeling study of effect of temperature on membrane performance. CO permeance Sweep flow rate Sweep inlet temperature Case 2 (GPU) (slpm) (oC) 1 100 34 76 2 100 53 76 3 Temperature- 34 99 dependent 4 200 34 120 5 Temperature- 53 76 dependent

The calculated H2S concentrations in the retentate fuel gas for different cases considered in Table 3.2 for a feed pressure of 10 psig are shown in Figure 3.15. Cases 1 and 2 were for two different feed gas flow rates of 34 and 53 slpm, respectively, and based on an assumption that the modules have a uniform temperature of 76oC. Case 4 was for a feed gas flow rate of 34 slpm and based on an assumption that the modules have a uniform 89 temperature of 120oC. The Cases 3 and 5 were the two representative ones for temperature- dependent permeance and different permeances were projected along the membrane module length by using equation (9) due to the temperature differences between feed gas

(120oC) and sweep gases (99 and 76oC for 34 slpm and 53 slpm sweep gases, respectively).

In Case 3, the fuel gas was cooled down from 120oC to 105°C and the sweep gas was heated up from 99oC to 117°C. In Case 5, the fuel gas was cooled down 120oC to 78°C while the sweep gas was heated up from 76oC to 115°C. The lower average temperature of Case 5 was manifested by a reduced permeance; however, the much larger sweep flow rate increased the permeation driving force and eventually resulted in a lower retentate H2S concentration.

From the comparison between the modeling results and the test results at PCI (7 – 8 ppm H2S), the following points were observed: (1) For either the modeling results or the test results at PCI, the retentate H2S concentrations were in the range of 4 – 10 ppm on wet basis. In other words, the agreement between the modeling results and the test results at

PCI was reasonably good; (2) The calculation gave a slightly lower retentate H2S concentration compared to the test data. This might stem from the 20% leakage from the fuel gas side to the sweep gas side for the field test at PCI due to the poor sealing in the vendor-fabricated membrane modules. This leakage resulted in a higher H2S concentration in the sweep gas, leading to a lower permeation driving force and consequently, a slightly lower H2S removal for the field test.

10 Representative case I 8 Representative case II 6

1100 GPU, sweep 34 slpm S conc., wet S wet (ppm) conc., basis 2 4 2100 GPU, sweep 53 slpm 3 Temp. dependent permeance, sweep 34 slpm 2 4200 GPU, sweep 34 slpm

Retentate Retentate H 5 Temp. dependent permeance, sweep 53 slpm 0 1 2 3 4 5 Case

Figure 3.15. The H2S concentration in the retentate side obtained for Cases 1 – 5 in Table 3.2.

3.5. Future directions

This work described the fabrication of scale-up membranes with 14 inches in width and a selective-layer thickness of around 15 microns. The roll-to-roll coating machine at OSU can be used to fabricate up to 21-inch wide membranes, which is half of the width of commercial membranes. The efficiency of the membrane fabrication can be enhanced by using a higher coating speed up to 5 ft/min. In order to accommodate the higher speed, the web length in the curing oven can be increased so that a sufficient drying and curing of the membrane can still be obtained.

The technique developed in this work can also be utilized in the fabrication of thin membranes with a membrane thickness of less than 1 µm. In that case, the variation of the membrane thickness along the width and length needs to be reduced and maintained at less than 10% in order to have a good quality membrane. Modifications that can be done to 91 achieve that goal include the substitution of the coating knife base from the granite slab to a stainless-steel rod to improve the flatness of the substrate below the coating knife. The coating knife trough can also be modified in order to control the delivery rate of the coating solution onto the membrane substrate. Moreover, other additives to control the increasing rate of coating solution viscosity can be investigated and utilized.

Another field test with industrial reformate gas using the design operating conditions can be performed to demonstrate the optimum membrane performance, preferably in specific application such as shipboard. The long-term stability as well as robustness of the spiral-wound membrane modules can be investigated. The field test results will be a major step forward toward the commercialization of the CO2-selective membrane for H2S removal in fuel cell applications.

3.6. Conclusions

A CO2-selective amine-containing membrane for H2S and CO2 removal was successfully scaled up to 14-inch in width for 150 feet in length with an average selective- layer membrane thickness of around 15 µm through continuous roll-to-roll fabrication. In order to obtain this desirable membrane thickness, the optimized fabrication parameters identified were a coating solution concentration of 20 wt.%, a substrate web coating speed of 0.5 ft/min, and a coating knife gap setting of 6 mils. The scale-up membrane exhibited more than 200 GPU of CO2 permeance, greater than 200 of CO2/H2 selectivity, and around

3 of H2S/CO2 selectivity, which were similar to those obtained from the lab-scale membranes.

The scale-up membranes were used to fabricate spiral-wound membrane modules for field testing with the feed gas from autothermal reforming (ATR) reformate at PCI. For the test, the design operating conditions were determined based on modeling calculations.

The results obtained from the test were in reasonably good agreement with the modeling results, and the H2S concentration in the treated hydrogen product of less than 10 ppm was obtained. This work has demonstrated the potential of the CO2-selective membrane for the

H2S removal in fuel cell applications, particularly in shipboard environment.

Acknowledgments

I specially thank Dr. Kartik Ramasubramanian for his contributions in developing the membrane composition utilized in this work and installation of the pilot-scale thin-film- coating assembly of the coating machine at OSU. I also acknowledge the contributions of

Dr. Varun Vakharia, Mr. Dongzhu Wu, and Mr. Yang Han to this work. I would like to thank Douglas J. Wheeler of DJW Technology for his helpful discussion and contribution.

I appreciate the donations of Lupamin® 9095 and PVA 217SB from BASF and Kuraray, respectively. I gratefully acknowledge Office of Naval Research for their financial support of this work. This work was supported by the United States Office of Naval Research under Award Number N00014-14-C-098.

Nomenclatures

Pi is the permeability of species i

αij is the selectivity of species i over species j

93 i is the gas component CO2

j is another gas component (H2, CO, or H2S); x is the mole fraction of gas components in the feed side of the membrane y is the mole fraction of gas components in the sweep side of the membrane

Ji is the steady-state CO2 molar flux through the selective-layer membrane

is the selective-layer membrane thickness

Δpi is the partial pressure difference between the feed and sweep sides and is defined by using the logarithmic mean method.

ρdry is the density of the dry membrane

ρ is the density of the coating solution c is the total solids concentration of the coating solution (by weight)

gap is the gap setting of the coating knife

References

[1] United States Department of Energy Fuel Cell Technologies Office, State of the

states: Fuel cells in America 2016, 7th ed., (2016)

[2] H. John, M. Cervi, A. Nickens, D. Hoffman, J. Kuseian, Fuel cell system models for

U.S. Navy shipboard applications, Fuel Cell Seminar Palm Springs (2005).

[3] Advanced hydrogen reformate stream purifier for fuel cell applications (available

online: http://www.navysbir.com/n09_s/navst09-014.htm)

[4] J. Hao, P.A. Rice, S.A. Stern, Upgrading low-quality natural gas with H2S- and CO2-

selective polymer membranes Part I. Process design and economics of membrane

stages without recycle streams, J. Membr. Sci. 209 (2002) 177–206.

[5] T.C. Merkel and L.G. Toy, Comparison of Hydrogen Sulfide Transport Properties in

Fluorinated and Nonfluorinated Polymers, Macromolecules 39 (2006) 7591–7600.

[6] H. Lin, Z. He, Z. Sun, J. Vu, A. Ng, M. Mohammed, J. Kniep, T.C. Merkel, T. Wu,

R.C. Lambrecht, CO2-selective membranes for hydrogen production and CO2 capture

– Part I: membrane development, J. Membr. Sci. 457 (2014) 149–161.

[7] H. Lin, Z. He, Z. Sun, J. Kniep, A. Ng, R.W. Baker, T.C. Merkel, CO2-selective

membranes for hydrogen production and CO2 capture – Part II: techno-economic

analysis, J. Membr. Sci. 493 (2015) 794–806.

[8] G. George, N. Bhoria, A. AlHallaq, A. Abdala, V. Mittal, Polymer membranes for

acid gas removal from natural gas, Sep. Purif. Tech. 158 (2016) 333–356.

[9] R. Baker, Future directions of membrane gas-separation technology, Membr. Tech.

138 (2001) 5–10.

[10] W.J. Koros, R. Mahajan, Pushing the limits on possibilities for large scale gas

separation: which strategies?, J. Membr. Sci. 175 (2000) 181–196.

[11] R. Quinn, D.V. Laciak, G.P. Pez, Polyelectrolyte-salt blend membranes for acid gas

separation, J. Membr. Sci. 131 (1997) 61–69.

[12] M.W. Uddin, M.B. Hägg, Natural gas sweetening – the effect on CO2-CH4 separation

after exposing a facilitated transport membrane to hydrogen sulfide and higher

hydrocarbons, J. Membr. Sci. 423-424 (2012) 143–149.

[13] J. Zou, W.S.W. Ho, CO2-selective polymeric membranes containing amines in

crosslinked poly(vinyl alcohol), J. Membr. Sci. 286 (2006) 310–321.

[14] J. Huang, J. Zou, W.S.W. Ho, Carbon dioxide capture using a CO2-selective

facilitated transport membrane, Ind. Eng. Chem. Res. 47 (2008) 1261–1267.

[15] H. Bai, W.S.W. Ho, Carbon dioxide-selective membranes for high-pressure synthesis

gas purification, Ind. Eng. Chem. Res. 50 (2011) 12152–12161.

[16] Y. Zhao, B.T. Jung, L. Ansaloni, W.S.W. Ho, Multiwalled carbon nanotube mixed

matrix membranes containing amines for high pressure CO2/H2 separation, J. Membr.

Sci. 459 (2014) 233–243.

[17] V. Vakharia, K. Ramasubramanian, W.S.W. Ho, An experimental and modeling

study of CO2-selective membranes for IGCC syngas purification, J. Membr. Sci. 488

(2015) 56–66.

[18] K. Ramasubramanian, CO2 (H2S)-selective membranes for fuel cell hydrogen

purification and flue gas carbon capture: an experimental and process modeling study

(Ph.D. dissertation), The Ohio State University (2013).

[19] Z. Tong, V. Vakharia, M. Gasda, W.S.W. Ho, Water vapor and CO2 transport through

amine-containing facilitated transport membranes, React. Funct. Polym. 86 (2015)

111–116.

[20] W. Salim, W.S.W. Ho, Recent developments on nanostructured polymer-based

membranes, Current Opinion in Chem. Eng. 8 (2015) 76–82.

[21] D. Wu, L. Zhao, V.K. Vakharia, W. Salim, W.S.W. Ho, Synthesis and

characterization of nanoporous polyethersulfone membrane as support for composite

membrane in CO2 separation: from lab to pilot scale, J. Membr. Sci. 510 (2016) 58–

71.

[22] D. Wu, C. Sun, P.K. Dutta, W.S.W. Ho. "SO2 interference on separation performance

of amine-containing facilitated transport membranes for CO2 capture from flue gas."

J. Membr. Sci. 534 (2017) 33–45.

[23] K.A. Lokhandwala, I. Pinnau, Z. He, K.D. Amo, A.R. DaCosta, J.G. Wijmans, R.W.

Baker, Membrane separation of nitrogen from natural gas: a case study from

membrane synthesis to commercial deployment, J. Membr. Sci. 346 (2010) 270–279.

[24] H. Lin, S.M. Thompson, A.S. Martin, J.G. Wijmans, K.D. Amo, K.A. Lokhandwala,

T.C. Merkel, Dehydration of natural gas using membranes. Part I: composite

membranes, J. Membr. Sci. 413-414 (2012) 70–81.

[25] H. Lin, S.M. Thompson, A.S. Martin, J.G. Wijmans, K.D. Amo, K.A. Lokhandwala,

B.T. Low, T.C. Merkel, Dehydration of natural gas using membranes. Part II:

sweep/countercurrent design and field test, J. Membr. Sci. 432 (2013) 106–114.

[26] T. Brinkmann, C. Naderipour, J. Pohlmann, J. Wind, T. Wolff, E. Esche, D. Muller,

G. Wozny, B. Hoting, Pilot scale investigations of the removal of carbon dioxide

from hydrocarbon gas streams using poly(ethylene oxide) – poly(butylene

terephthalate) PolyactiveTM thin film composite membranes, J. Membr. Sci. 489

(2015) 237–247.

[27] C.A. Scholes, A. Qader, G.W. Stevens, S.E. Kentish, Membrane pilot plant trials of

CO2 separation from flue gas, Greenhouse Gas Sci. Technol. 5 (2015) 229–237.

[28] M. Sandru, T.J. Kim, W. Capala, M. Huijbers, M.B. Hägg, Pilot scale testing of

polymeric membranes for CO2 capture from coal fired power plants, Energy Procedia

37 (2013) 6473–6480.

[29] L.S. White, X. Wei, S. Pande, T. Wu, T.C. Merkel. Extended flue gas trials with a

membrane-based pilot plant at a one-ton-per-day carbon capture rate, J. Membr. Sci.

496 (2015) 48–57.

[30] W. Yave, A. Car, J. Wind, K.V Peinemann, Nanometric thin film membranes

manufactured on square meter scale: ultra-thin films for CO2 capture, Nanotechno.

21 (2010) 395301.

[31] J. Morse, Nanofabrication technologies for roll-to-roll processing, NIST-NNN

Workshop Report, September 2011.

[32] S. Bae, H. Kim, Y. Lee, X. Xu, J.S. Park, Y. Zheng, J. Balakrishnan, T. Lei, H.R.

Kim, Y.I. Song, Y.J. Kim, K.S. Kim, B. Ozyilmaz, J.H. Ahn, B.H. Hong, S. Iijima,

Roll-to-roll production of 30-inch graphene films for transparent electrodes, Nat.

Nanotechnol. 5 (2010) 574–578.

[33] L. Wengeler, M. Schmitt, K. Peters, P. Scharfer, W. Schabel, Comparison of large

scale coating techniques for organic and hybrid films in polymer based solar cells,

Chem. Eng. Proc. 68 (2013) 38–44.

[34] F.C. Krebs, Fabrication and processuing of polymer solar cells: A review of printing

and coating techniques, Solar Energy Mat. Solar Cells. 93 (2009) 394–412.

[35] U. Siemann, Solvent cast technology - A versatile tool for thin film production, Prog.

Colloid Polym. Sci. 130 (2005) 1–14.

[36] V. Vakharia, W. Salim, M. Gasda, W.S.W. Ho, Oxidatively stable membranes for

CO2 separation and H2 purification, J. Membr. Sci. 533 (2017) 220-228.

[37] J. Huang, L. El-Azzami, W.S.W. Ho, Modeling of CO2-selective water gas shift

membrane reactor for fuel cell, J. Membr. Sci. 261 (2005) 67–75.

Chapter 4. Fabrication and Field Testing of Spiral-wound Membrane Modules for CO2

Capture from Flue Gas

4.1. Summary

A spiral-wound membrane module comprising face compression “O” rings for effective seal of gases was developed, including a spiral-wound membrane element,

Plexiglas fiber-reinforced plastic tube, and membrane housing. The spiral-wound membrane modules demonstrated essentially no leakage. The concentration polarization phenomenon was observed at a dry feed gas flow rate of less than 1000 sccm before humidification. Hence, the performances of the spiral-wound membrane modules were evaluated using both dry feed and sweep gas flow rates each at 1000 sccm before humidification at the typical flue gas temperature of 57oC. The feed and sweep gas pressures were at 1.5 and 1 psig, respectively. By using either a simulated flue gas or an actual flue gas, the spiral-wound membrane modules showed similar performance results with a CO2 permeance of greater than 800 GPU and a CO2/N2 selectivity of more than 140, which were essentially identical to the results obtained from the flat-sheet membrane used for the module fabrication and tested in laboratory with the simulated flue gas. The pressure drops of the modules were satisfactory, i.e., less than 1.5 psi/m. During the field test using the actual flue gas at the National Carbon Capture Center in Wilsonville,

Alabama, the performances of the modules revealed some unexpected issues including the feed spacer indentations on the selective-layer membrane and the leakage at glue lines.

These issues were resolved by improving the spiral-wound membrane module fabrication

100 using a smoother feed spacer and a longer glue curing time. The field test results of the modules have shown their good potential for the post-combustion CO2 capture from coal- fired power plants.

4.2. Introduction

One of the main sources of CO2 emission, which is generally accepted as the cause of the global warming, is the flue gases from coal-fired power plants [1–3]. The CO2 capture processes that can be used in such power plants are pre-combustion, oxy-combustion, and post-combustion processes. The advantage of the post-combustion CO2 capture processes is mainly the direct retrofitting to the thousands of existing pulverized coal combustion power plants worldwide [4]. A membrane process for post-combustion CO2 capture has advantages over absorption and adsorption processes due to the following reasons [1,2]:

• Compactness, lightweight, and options to be positioned horizontally or vertically,

which make great ease and flexibility for retrofitting applications,

• Modular design of membrane system allows the process optimization by using

multi-module and recycling operation,

• No moving parts in the membrane unit and hence low maintenance requirements,

• No separate regeneration step and hence no additional energy penalty for

regeneration,

• Kinetic ability to overcome thermodynamic solubility limitation.

The typical types of membrane modules for large-scale gas separation application include spiral-wound, hollow-fiber, and plate-and-frame configurations. Commercial

101 membrane processes for gas separation have been successfully developed by using such membrane modules [5–7]. Lokhandwala et al. reported the installation of twelve membrane-based systems, named as NitroSepTM, for separation of 4 – 30 mol.% nitrogen from natural gas streams [5]. They concluded that the CH4-selective membranes based on polydimethylsiloxane (PDMS) were preferable, and a simple two-step bank of spiral- wound membrane modules was suitable for treating natural gas streams containing 4 – 8 mol.% nitrogen. Lin et al. designed a sweep/countercurrent spiral-wound membrane modules that showed a low methane loss and minimum membrane area requirement for the dehydration of natural gas [6,7]. After field testing of the spiral-wound membrane module

(4 in (0.1016 m) diameter, 40 in (1.016 m) long, and 3 m2 membrane area) in a natural gas processing plant, they demonstrated that the use of a dry sweep gas on the permeate side

® increased the water vapor flux across the H2O-selective Pebax -based membranes.

Remarkable advances towards large-scale membrane processes for CO2 separation and capture have been made recently [8–10]. Brinkmann et al. investigated the performance of an envelope-type membrane module in pilot scale by using PolyActiveTM, a poly(ethylene oxide)-poly(butylene terephthalate) composite CO2-selective membrane, for

CO2 removal from methane and CO2 separation from hydrocarbon streams [8]. They reported that the 310 mm envelopes could house up to 75 m2 of membrane area, which resulted in up to a membrane packing density of 950 m2/m3, depending on the envelope and spacer thickness. Sandru et al. tested a plate-and-frame membrane module (24 membranes of 0.25 m × 0.25 m and membrane area of 1.5 m2) in a coal-fired power plant and showed that the poly(vinylamine) membranes kept constant performance after

102 exposure to high levels of NOx and SO2 as well as the power plant outages [9]. The membranes reached a maximum of 75% CO2 content in permeate and a CO2 permeance of

0.2 – 0.6 m3(STP)/(m2 bar h), which is equal to 74 – 222 GPU (1 GPU = 10-6

3 2 cm (STP)/(cm s cmHg)), and CO2/N2 selectivity of 80 – 300. Scholes et al. compared the performances of two commercial membrane modules: a hollow-fiber Air Products PRISM

PA1020 polysulfone membrane module and a spiral-wound Filmtec® NF3838/30FF polyamide membrane module [10]. They observed that the hollow-fiber membrane module performance decreased upon exposure to the flue gas due to the competitive sorption from increasing concentration of water, concentration polarization, and possibly membrane fouling. Meanwhile, the spiral-wound membrane module performance increased in the presence of the flue gas as the saturated water allowed the facilitate transport mechanism to occur. Moreover, the pressure-drop observations indicated that only the spiral-wound membrane module was suitable for post-combustion CO2 capture.

However, the performance of the spiral-wound membrane module was not of the order required to be competitive since both the CO2 permeance of 29 GPU and CO2/N2 selectivity of 7 were too low.

Field testing and demonstration of membrane processes for CO2 capture have shown significant progress recently [11–13]. For pre-combustion CO2 capture, Lin et al. fabricated spiral-wound membrane modules, each with a diameter of 0.2 m and membrane

2 TM area of 20 m , for a CO2-selective Polaris membrane process to capture CO2 from a syngas stream equivalent to 0.15 MWe in the IGCC power plant and produce liquid CO2 at the National Carbon Capture Center (NCCC), Wilsonville, Alabama, USA [11,12]. For

103 the post-combustion CO2 capture, White et al. performed extended flue gas trials at this

TM NCCC by installing a CO2-selective Polaris -based membrane unit capable of treating 1 tonne/day of CO2 gas [13]. Their two-step process including cross-flow and countercurrent sweep spiral-wound membrane modules (0.2 m diameter, 1 m long) for each step, respectively, was used to achieve a stable performance and obtain the carbon capture over

90%.

Ho and coworkers [14–26] have developed a two-stage membrane process for the post- combustion CO2 capture from flue gas that exhibited an excellent performance by using a novel polymer/inorganic composite membrane, consisting of a selective amine-containing polymer cover layer, a zeolite nanoparticle layer, and a polymer support. The 14” wide polymer/inorganic composite membrane was fabricated by using a continuous roll-to-roll coating machine at The Ohio State University (OSU). The scale-up membrane showed a great potential for the post-combustion CO2 capture from flue gas in coal-fired power plants [27]. However, transformation from the flat-sheet membrane test results to membrane module performances is a critical step towards the large-scale application of the membrane process. This work is a continuation of the previous lab-scale work and focuses on the fabrication and field testing of spiral-wound membrane modules with a countercurrent configuration. The details regarding the design, fabrication, and test results of the spiral-wound membrane modules are valuable to demonstrate the effective CO2 capture from flue gas.

104

4.3. Experimental

4.3.1. Materials

The novel polymer/inorganic composite membrane used in this work comprised a polymer support, a zeolite nanoparticle layer, and a selective amine-containing polymer cover layer. The pilot-scale fabrication of the polymer support was described in detail in the previous publication by Wu et al. [20]. Polyester non-woven fabric was purchased from TriSep Corporation (Goleta, CA, USA). BASF polyethersulfone (PES) Ultrason®

E7020P with a MW of 92000 was donated by BASF Corporation (Wyandotte, MI, USA).

The deposition of the Zeolite-Y nanoparticles onto the polymer support was described in detail in the previous publications [21–24]. The synthesis method of the high molecular weight polyvinylamine (PVAm) used in the amine-containing cover layer was described in the previously published papers by Chen et al. [25,26]. N-methyl-2-pyrrolidone (NMP), piperazine (99%), and glycine (98.5%) were purchased from Sigma-Aldrich (Milwaukee,

WI, USA). 2-Methoxyethanol (2-ME) was purchased from Fisher Scientific (Bridgewater,

NJ, USA). The NMP, 2-ME, and PES were used for polymer support fabrication [20].

Piperazine, glycine, and PVAm were used for amine-containing polymer cover layer preparation [25,26]. The scale-up fabrication of the amine-containing polymer cover layer for CO2 separation is being described as a separate publication by Vakharia et al., and the

105 fabricated scale-up membrane was used in this work to prepare the spiral-wound membrane modules. Unless otherwise noted, all chemicals were used without any further purification.

Epoxy glue (GSP 1394-1) was purchased from GS Polymers (Mira Loma, CA, USA).

Teflon feed spacer (ET8120), polypropylene feed spacer (XN1673), and polypropylene sweep spacer (XN3234) were purchased from Industrial Netting (Minneapolis, MN, USA).

Crystalline polyethyleneterephthalate (CPET) feed spacer was purchased from Dexmet

Corporation (Wallingford, CT, USA). Aluminum tape (VEN-3520CW) was acquired from

Progress Supply (Columbus, OH, USA) while acrylic tubes and sheets, “O” rings, bolts and nuts, were purchased from McMaster-Carr (Cleveland, OH, USA).

4.3.2. Spiral-wound membrane element fabrication

As illustrated in Figure 4.1, the membrane element was prepared according to the following procedure that was developed based on the available literature [28,29]. Firstly, a 30-inch long scale-up membrane with a 12-inch width was folded once to have a membrane leaf length of 15 inches, and the permeate spacer was inserted in between. An epoxy glue was used to seal the membrane leaf at three sides and to prepare the glue lines required to enable the crosscurrent flow inside the permeate channel. Afterwards, the membrane leaf was glued and taped to the central tube, and the feed spacer was placed on the membrane leaf. Finally, the membrane leaf was rolled by using the rolling machine that was developed at OSU. The membrane element was wrapped with an outer-wrap aluminum tape.

106

The spiral-wound membrane element was rolled with about 1.8” in diameter (using a

1.5” OD central tube) by 14” in length using a single membrane leaf of 15” in length with an effective width of 12”. The membrane leaf consisted of one piece of the once folded polymer/zeolite composite membrane, a permeate spacer in between the folded composite membrane, and a feed spacer. The length of the membrane element was about 14 inches, and the total membrane area was about 300 cm2.

(1) Glue application (2) Attachment to central tube Sweep Inlet

Sweep Outlet

Feed Outlet Sweep Inlet Feed Inlet Sweep (4) Outer-wrapping Outlet (3) Membrane element rolling

Figure 4.1. The step-by-step procedure of the spiral-wound membrane element fabrication.

4.3.3. Spiral-wound membrane module: new design

A major drawback commonly encountered in conventional spiral-wound membrane modules is the feed gas bypass. The ineffective sealing of the end-cap flanges associated with the conventional spiral-wound membrane module has led to significant “bypass” of the feed gas to the feed outlet (retentate) without flowing between the wrapped membrane layers. This has resulted in low CO2 permeance (lower than the flat-sheet membrane 107 performance). In principle, if the issue of the feed gas bypass is resolved, the performance and quality of the spiral-wound membrane element/module can be improved significantly.

The design of the new spiral-wound membrane element and housing in the present work is divided into 3 sections, including the spiral-wound membrane element, a Plexiglas FRP

(fiber reinforced plastic) tube, and a membrane housing. The picture of one of the fabricated spiral-wound membrane elements is shown in Figure 4.2.

Figure 4.2. Spiral-wound membrane element fabricated.

The fabricated spiral-wound membrane element was consequently inserted inside and glued to the Plexiglas FRP tube as shown in Figure 4.3. The Plexiglas FRP was a Plexiglas tube that was analogous to and serves as the FRP outer-wrap in the commercial spiral- wound membrane elements. The Plexiglas FRP was 1/4” thick. The presence of 1/4” thick

Plexiglas FRP aided in sealing by face-compression of the “O” rings on the flanges as shown in Figure 4.4. The face compression “O” rings successfully helped in ensuring that the element had no leakage and hence enhancing the CO2 transport in the element.

108

Plexiglas FRP

Sweep Sweep inlet outlet

Figure 4.3. The spiral-wound membrane element placed inside and glued to the Plexiglas FRP.

Face Face compression compression “O” ring “O” ring on FRP on central tube

Face compression “O” ring Face on FRP compression “O” ring on central tube

Figure 4.4. The face-compression “O” ring design of the spiral-wound membrane element with the Plexiglas FRP. 109

The assembled spiral-wound membrane module consisting of the membrane element

inside a housing is shown in Figure 4.5. This face-compression of the “O” rings is unlike

the conventional spiral-wound design for reverse osmosis application that relies on sliding

the “O” ring installed at the anti-telescoping device for sealing [30].

Feed inlet

Sweep Sweep inlet outlet

Feed outlet

Figure 4.5. Spiral-wound membrane module consisting of the membrane element inside a housing.

4.3.4. Spiral-wound membrane module testing

The leak test was a part of the procedural protocol to check if there was any leak of the

overall element and module prior to the testing for the gas transport performance. The

step-by-step procedure for the leak test is as follows:

1) Load the spiral-wound membrane element to the membrane housing (Figure 4.5).

2) Connect the feed inlet of the housing to a dry air tubing, which is connected to a

rotameter to control the air flow rate. Close the valve to make sure that there is no air

flow prior to the installation of the spiral-wound membrane module. 110

3) Connect the feed outlet and the permeate inlet (countercurrent configuration) to the

pressure gauges as shown in Figure 4.6. The needle valve for each of the pressure

gauges should be kept open to the atmosphere.

4) Plug the outlet end of the permeate side as shown in Figure 4.6. This ensures that any

increase in the pressure on the permeate side is being detected by the pressure gauge

connected to the inlet end of the permeate side.

5) Start the air flow to the feed inlet at a flow rate of 1000 sccm as controlled by the

rotameter.

6) Gradually rotate and close the needle valve for the pressure gauge connected to the feed

outlet/retentate side in order to increase the pressure on the feed side up to 1.5 – 2 psig.

7) Afterwards, completely close the needle valve for the pressure gauge connected to the

inlet end of the permeate side. Any leak from the feed side to the permeate side will

be immediately indicated by the pressure gauge on the permeate side since the pressure

on the permeate side will immediately increase and eventually will be equal to the

pressure on the feed side (1.5 psig in this case). This indicates that there is a leak in

the spiral-wound membrane module.

8) If there is no increase in the pressure on the permeate pressure gauge, then check all

the connections for any possible source of leak from the housing to the atmosphere.

One of the ways to check the connections for leak is the use of the soap bubble test for

all the connections.

9) If no leak is observed and the pressure gauge on the permeate side still shows 0 psig

for a period of 5 minutes, then it is confirmed that there is no convectional flow leak

111

from the feed side to the permeate side. Only leak-free spiral-wound membrane

modules will be installed in the gas permeation unit for gas transport measurement.

Permeate pressure gauge

Retentate pressure Rotameter gauge

Plug

Figure 4.6. The leak test assembly for the spiral-wound membrane module.

The gas permeation unit that was used to measure the transport performance of the spiral-wound membrane module is shown in Figure 4.7. The unit consisted of mass flow controllers, water pumps, and humidifiers to simulate the actual gas composition in the flue gas. After passing the leak test, the membrane module was placed schematically inside the oven of the unit as shown in Figure 4.7. The feed and sweep gases entered the module in a countercurrent configuration. During the test, the dry feed gas flow rate used was

112 typically 1000 sccm (a lower or higher rate could also be employed) at 1.5 psig, and the dry sweep gas flow rate used for the permeate side was also typically 1000 sccm (a lower or higher rate could also be employed) at 1 psig. Both the feed gas and the sweep gas were humidified with water vapor by injecting controlled amounts of water into them to obtain the controlled concentrations of water vapor in them. At the typical flue gas temperature of 57oC, the saturation water vapor content is about 17.2% for both the feed and sweep gases at about 1 atm.

113

Water reservoir

Figure 4.7. Schematic of the gas permeation unit for membrane module gas transport measurements.

As described in the previous work [18], the pressure drop of the spiral-wound membrane module was measured by installing 4 pressure gauges at the following positions of the spiral-wound membrane module: (1) feed inlet, (2) retentate outlet, (3) sweep inlet, and (4) permeate outlet. The pressures at (2) and (4) were set to be 1.5 and 1.0 psig, respectively.

The pressure difference between (1) and (2) divided by the module length was considered as the feed side pressure drop of the spiral-wound membrane module. Similarly, the 114 pressure difference between (3) and (4) divided by the module length was taken as the sweep side pressure drop of the spiral-wound membrane module.

The field test conducted at the NCCC in Wilsonville, AL was based on the testing collaboration agreement between OSU and NCCC and was mainly related to the tasks of the DOE-NETL (Department of Energy-National Energy Technology Laboratory) project of DE-FE007632. The main objective of the field test was to evaluate the spiral-wound membrane module performance with the real flue gas produced at NCCC as the feed gas.

Additionally, the preliminary stability of the spiral-wound membrane module could also be investigated. Figure 4.8 shows the setup of the gas permeation unit inside the analytical lab at NCCC. Prior to the start of testing, the membrane module was placed inside the oven of the unit as shown in Figure 4.9. The details of the model used for the determination of the gas transport performance for the spiral-wound membrane module is described in detail as Appendix C.

115

Gas flow controller Oven

Gas chromatography Water pumps Oven control Water panel reservoir

Figure 4.8. The setup of the gas permeation unit installed inside the analytical lab of NCCC.

Humidifier (2 L)

Humidifier (0.5 L) Sweep inlet Feed outlet

Feed inlet

Sweep outlet

Figure 4.9. A spiral-wound membrane module placed inside the oven of the gas permeation unit for gas transport measurements.

116

4.4. Results and discussion

4.4.1. Effect of feed gas flow rate on membrane module performance

The testing conditions for the spiral-wound membrane module were similar to the flat- sheet membrane test. However, higher feed and sweep flow rates (500 – 1000 sccm) were used for the initial testing of the gas transport performance of the spiral-wound membrane modules. Figure 4.10 shows the effect of feed gas flow rate on the gas separation performance of the spiral-wound membrane module. As shown in this figure, higher CO2 permeance values were observed as the feed gas flow was increased from 500 to 1000 sccm for the same spiral-wound membrane module. This observation could be explained by the concentration polarization effect, which appeared to be significant at lower flow rates of less than 1000 sccm. The concentration polarization effect is a very common phenomenon in membrane processes that involves liquid systems such as reverse osmosis, nanofiltration, ultrafiltration, and microfiltration [31-39]. Since the diffusion coefficient in the gas phase is much larger than that in the liquid phase, the concentration polarization effect in general is assumed to be negligible for gas separation membranes that have low fluxes. However, this assumption is not valid for membranes with high gas permeances, and both theoretical and experimental investigations support that the concentration polarization effect indeed exists for gas separation membranes [40-49].

117

900

800

700

600

500

400

300 Permeance (GPU) Permeance

2 200

CO 100

0 0 250 500 750 1000 1250

Feed Gas Flow Rate (sccm)

Figure 4.10. The effect of feed gas flow rate on the CO2 permeance of the spiral-wound membrane module (concentration polarization phenomenon).

The concentration polarization phenomenon occurred on the feed side of the spiral- wound membrane module. In other words, the concentration of CO2 in the bulk gas phase was higher than that at the membrane surface whereas the concentration of N2 in the bulk gas phase was lower than that at the membrane surface. This was made possible for high flux membranes because CO2 permeated through the membrane much faster than N2 and resulted in most of N2 being retained by the membrane on its surface in the feed side. The concentration polarization phenomenon reduced the CO2 permeance, particularly at lower flow rates where the stagnant boundary layers were thicker. The effects of this concentration polarization became less pronounced when the feed gas flow rate was increased to around 1000 sccm in the spiral-wound membrane module testing.

118

4.4.2. Module performance and pressure drop results at > 1000 sccm feed gas flowrates

The transport performances and pressure drops of representative spiral-wound membrane modules at different feed gas flow rates above 1000 sccm are shown in Table

4.1. As the flow rate was increased further to more than 1000 sccm, sometimes there was a reduction of CO2/N2 selectivity of the spiral-wound membrane modules, as seen in SW-

3 and SW-4 in Table 4.1, which could be caused by spiral-wound membrane module glue failure at the high flow rates. Moreover, as shown in Table 4.1, the pressure drop of the spiral-wound membrane module was usually higher as a higher flow rate was used. For instance, the pressure drop of SW-2 on the feed side increased from 1.48 to 2.3 psi/m as the feed flow rate increased from 1000 to 2000 sccm. The spiral-wound membrane modules showed an average pressure drop of less than 1.5 psi/m for both the feed and sweep sides when 1000 sccm of feed gas flow rate was used. These results have met the requirement of ≤ 1.5 psi/m for the pressure drop of the spiral-wound membrane module.

This pressure drop criterion is critical for post-combustion CO2 capture to reduce the energy that is required to increase the pressure of the flue gas to provide the driving force of the gas separation. Furthermore, an excessively high feed-gas flow rate corresponds to a low membrane stage cut, which leads to a deviation compared to the actual membrane process. In order to maintain the pressure drop at less than 1.5 psi/m and to prevent the spiral-wound membrane module glue failure issue, the flow rates for both feed and sweep sides were maintained at 1000 sccm in the subsequent tests of the spiral-wound membrane modules.

119

Table 4.1. Gas transport performances and pressure drops of spiral-wound membrane modules at different feed gas flow ratesa. Module Feed/sweep CO2 CO2/N2 Feed/sweep No. flow rate (sccm) permeance selectivity pressure drop (GPU) (psi/m) SW-1 1000/1000 350 220 1.48/1.31 2000/1000 350 220 1.64/1.48 SW-2 1000/1000 430 350 1.48/1.64 1500/1000 480 280 1.97/1.97 2000/1000 480 250 2.30/1.97 SW-3 1000/1000 570 250 1.48/1.64 1500/1000 600 200 1.64/1.97 2000/1000 600 40 1.97/2.30 SW-4 1000/1000 600 200 1.48/1.64 1500/1000 600 150 1.64/1.97 2000/1000 590 30 1.97/2.30 SW-5 1000/1000 715 360 1.48/1.48 1500/1000 720 250 1.64/1.97 a Other test conditions: feed gas = 20% CO2 and 80% N2 (on dry basis), sweep gas = argon, feed/sweep pressures = 1.5 psig/1 psig, and temperature = 57oC.

4.4.3. Module performance and pressure drop results at 1000 sccm feed gas flowrates and performance comparison with flat-sheet membranes

Table 4.2 shows the transport results of the spiral-wound membrane modules which were tested with 20% CO2 and 80% of N2 (on dry basis) at the optimized feed gas flow rates of 1000 sccm. The comparison of the gas transport performances obtained from flat- sheet membranes and spiral-wound membrane modules is shown in Figure 4.11. This figure, along with Table 4.2, shows that the spiral-wound membrane modules achieved a

120 high CO2 permeance of up to 820 GPU with a very high CO2/N2 selectivity of greater than

200, which were similar to the results obtained from the flat-sheet membranes. This high transport performance was achieved after we resolved the glue line failure, module leakage, and membrane indentation issues that were encountered initially. The issue of the glue line failure was successfully resolved by the proper selection of a suitable glue (GSP 1394-1 epoxy glue) and the glue curing at room temperature for 16 hours. The quality of the spiral- wound membrane element fabrication was improved by optimizing the glue-line application procedure, including the location of the glue line as shown in Figure 4.1 and the width of the glue line of around 0.5 inch. The new procedure aided in adequate sealing of the nonwoven fabric in the membrane leaf and sealing between the membrane leaf and the central tube. The quality of fabrication was successfully improved as indicated by the new elements that demonstrated essentially no leakage during the leak testing. The physical impact of the feed spacer on the selective-layer during the initial element preparation, resulting in indentations, was believed to be the main reason for the poor selectivity. Thus, a layer of a fine and smooth polymer spacer (Teflon feed spacer ET8120) was incorporated between the feed spacer and the selective polymer layer to minimize/eliminate the indentations. The indentations were eventually successfully eliminated in the elements prepared by OSU, resulting in the improved CO2 transport performance with good CO2/N2 selectivity.

121

Table 4.2. Gas transport performances and pressure drops of spiral-wound membrane modules at the optimized feed gas flow ratea. Module Feed/sweep CO2 CO2/N2 Feed/sweep No. flow rate (sccm) permeance selectivity pressure drop (GPU) (psi/m) SW-6 1000/1000 780 180 1.31/1.31 SW-7 1000/1000 800 220 1.31/1.31 SW-8 1000/1000 750 200 1.48/1.31 SW-9 1000/1000 770 200 1.48/1.31 SW-10 1000/1000 820 270 1.48/1.48 SW-11 1000/1000 800 200 1.48/1.48 SW-12 1000/1000 820 200 1.48/1.48 SW-13 1000/1000 750 140 1.48/1.48 SW-14 1000/1000 800 200 1.48/1.48 SW-15 1000/1000 800 160 1.48/1.48 a Other test conditions: feed gas = 20% CO2 and 80% N2 (on dry basis), sweep gas = argon, feed/sweep pressures = 1.5 psig/1 psig, and temperature = 57oC.

Table 4.2 also shows the results of the pressure drop measurements of the spiral-wound membrane modules. As shown in this table, low pressure drops of less than 1.5 psi/m

(meter of the membrane length) were obtained for the spiral-wound membrane modules with good transport performance after we solved the glue line failure, module leakage and membrane indentation issues encountered initially. This low pressure drop value is highly desirable as less energy is needed to increase the flue gas pressure for providing the sufficient driving force for CO2 separation [50].

122

400 Flat-sheet membranes 350 Spiral-wound membrane modules 300

250

200 electivity

S 150

2 /N

2 100

CO 50

0 500 600 700 800 900 1000 CO Permeance (GPU) 2 Figure 4.11. Transport performances of scale-up flat-sheet membranes and spiral-wound membrane modules at 57oC.

4.4.4. Module performance and pressure drop results at 1000 sccm feed gas with SO2 and O2

For comparison with the NCCC field test results, spiral-wound membrane modules

SW-16 and SW-17 were tested at The Ohio State University with a simulated flue gas containing SO2 and O2. The gas transport performances of the spiral-wound membrane modules are summarized in Table 4.3. The SW-17 module had a longer glue curing time

(of 48 hours) at 57oC than the SW-16 module (of 16 hours). This longer glue curing time gave a better selectivity stability result with the SW-17 module than with the SW-16 module as shown in this table.

123

Table 4.3. Gas transport performances of spiral-wound membrane elements at 57oC tested at OSUa. Module Feed/sweep CO2 CO2/N2 Feed/sweep No. flow rate (sccm) permeance selectivity pressure drop (GPU) (psi/m) SW-16 1000/1000 700 210  160 1.48/1.48

SW-17 1000/1000 620  580 205 1.97/2.30

a Other test conditions: feed gas = 20% CO2, 7% O2, 3 ppm SO2, and balance of N2 (on dry basis) for SW-16 and 20% CO2, 3% O2, 1 – 3 ppm SO2, and balance of N2 (on dry basis) for SW-17, sweep gas = argon, feed/sweep pressures = 1.5 psig/1 psig, and temperature = 57oC.

The stability test results of modules SW-16 and SW-17 are included in Figures 12 and

13, respectively, for comparison with the spiral-wound membrane modules that were tested

at NCCC. The module SW-16 stability was tested with 20% CO2, 7% O2, 1 – 3 ppm SO2,

and balance of N2 (on dry basis). As can be seen from Figure 4.12, the subject module

showed a CO2 permeance of around 700 GPU throughout the test. The CO2/N2 selectivity

fluctuated around 210 until the 150th hour of test and then dropped to 160, which could

possibly be caused insufficient glue curing and membrane indentations by the rough

surface of the feed spacer.

124

1000 350 900 NCCC Test CO Permeance 300 2 800 700 250 OSU Test CO Permeance 2 600 200 OSU Test CO2/N2 Selectivity

500 Selectivity

150 2

400

/N 2

Permeance (GPU) Permeance 300 100 2 200 CO NCCC Test CO /N Selectivity CO 2 2 50 100 60-h NCCC Flue Gas Shutdown 0 0 0 20 40 60 80 100 120 140 160 180 200

Run Time (h)

Figure 4.12. The stability plots of the spiral-wound membrane modules SW-16 tested at OSU and SW-19 tested at NCCC.

The module SW-17 stability was tested with 20% CO2, 3% O2, 1-3 ppm SO2, and balance of N2. As can be seen from Figure 4.13, the subject module showed a CO2 permeance of around 620 GPU initially. After 3% O2 was incorporated in the feed gas, the

CO2 permeance fluctuated around 600 GPU. It did not show a dropping trend during the

115 hours of test, which could be considered relatively stable. Also from this figure, after

th the incorporation of 1 ppm SO2 and 3 ppm SO2 at around the 115 hour of the test, the CO2 permeance dropped from 600 GPU to around 580 GPU, and then the CO2 permeance remained stable at 580 GPU. The overall test took around 204 hours and showed a relatively stable result, which was a result of the longer glue curing time.

125

1000 300 NCCC Test CO /N Selectivity 2 2 900 NCCC Test CO Permeance 250 800 2 OSU Test CO /N Selectivity 2 2 700

(GPU) 200 600

500 150 OSU Test CO Permeance Selectivity

2 2

400 /N 2

Permeance 100

300

2 CO

CO 200 50

100 48-h NCCC Flue Gas Shutdown 0 0 0 20 40 60 80 100 120 140 160 180 200

Run Time (h)

Figure 4.13. The stability plots of the spiral-wound membrane modules SW-17 tested at OSU and SW-20 tested at NCCC (both models with a longer glue curing time of 48 hours).

4.4.5. Membrane module field test results with actual flue gas at NCCC

The spiral-wound membrane module testing at NCCC lasted from May 28 to June 22,

2015. During this period, three spiral-wound membrane modules were tested after passing their leak tests. The flue gas composition provided by NCCC contained approximately

12% (± 1%) CO2, 7% (± 1%) O2, 81% (± 1%) N2, 0.5 – 5 ppm SO2, and 1.5 – 4 ppm NO2.

Among those components, the oxygen concentration was much higher than the typical average of 2.3% in coal-fired power plants. Our gas chromatograph (GC) settings, including the GC column, were incapable of separating the oxygen concentration from the nitrogen concentration in the permeate stream sample. However, we solved this problem

126

with the great help from the analytical lab team of NCCC by using a non-dispersive infrared

(NDIR) oxygen analyzer to measure the oxygen concentration in the retentate stream. By

using this concentration, the permeate oxygen concentration was determined from the mass

balance, and the accurate CO2/N2 selectivities for our membrane modules were obtained.

As shown in Table 4, the three spiral-wound membrane modules tested at NCCC

showed repeatable results with > 800 GPU of CO2 permeance and > 150 of CO2/N2

selectivity. Those results were well in line with the spiral-wound membrane modules

tested in our OSU laboratory. The first spiral-wound membrane module (SW-18) was

tested for 96 hours and showed a stable result at 820 GPU of CO2 permeance and a CO2/N2

selectivity of 150. As shown in Figure 4.14, this element showed a promising stable

performance of 96 hours. However, the test was stopped since the other elements could

give a higher CO2/N2 selectivity.

Table 4.4. Gas transport performances of spiral-wound membrane elements at 57oC tested at NCCCa. Module Feed/sweep CO2 CO2/N2 Feed/sweep No. flow rate (sccm) permeance selectivity pressure drop (GPU) (psi/m) SW-18 1000/1000 820 150 0.98/1.31 SW-19 1000/1000 800 170  60 0.98/1.48 SW-20 1000/1000 800  630 270  180 1.31/1.48

a Other test conditions: feed gas = 12% (± 1%) CO2, 7% (± 1%) O2, 81% (± 1%) N2, 0.5 – 5 ppm SO2, and 1.5 – 4 ppm NO2 (on dry basis), sweep gas = argon, feed/sweep pressures = 1.5 psig/1 psig, and temperature = 57oC.

127

1000 350 NCCC Test CO Permeance 900 2 300 800 700 250 600 200

500 Selectivity

150 2 400

NCCC Test CO2/N2 Selectivity /N 2

Permeance (GPU) Permeance 300 100 2 200 CO CO 50 100 0 0 0 20 40 60 80 Run Time (h) Figure 4.14. The stability plot of the spiral-wound membrane module SW-18 tested at NCCC.

The second spiral-wound membrane module (SW-19) was tested for 208 hours, and there was a flue gas shutdown for of 60 hours in the middle of the test. The SW-19 module showed an initial CO2 permeance of around 800 GPU and a CO2/N2 selectivity of 170 before the flue gas shutdown; however, the selectivity dropped to around 60 after the restart of the test following the flue gas return. Figure 4.12 shows the stability plot of the SW-19 module and its comparison to that of the SW-16 module that was tested at OSU. Both modules showed a reasonably stable CO2 permeance throughout the test and a drop of

CO2/N2 selectivity. The selectivity drop was presumably due to the insufficient curing of the glue used and the membrane indentations caused by the rough surface of the feed spacer. The indentations on the membrane sample cut from the SW-19 module after the

128 test are shown in Figure 4.15. These indentations might have introduced the leakage of the spiral-wound membrane element and resulted in the CO2/N2 selectivity drop.

1 inch (2.54 cm) Nonwoven fabric

Scale-up membrane (before rolled to SW-19)

SW-19 sample cut (unrolled after test)

Figure 4.15. The images of the membrane before element rolling showing a smooth membrane surface and after the module testing with the indentations caused by the feed spacer.

The third spiral-wound membrane module (SW-20) was tested for 200 hours, and there was a flue gas shutdown at the 48th hour in the middle of the test. The SW-20 module with a longer glue curing time of 48 hours showed reasonably stable selectivity of more than

200 initially. Figure 4.13 shows the stability plot of the SW-20 module and its comparison to the SW-17 module that was tested at OSU. Both modules showed a reasonably stable

CO2/N2 selectivity throughout the test, except for the drop of CO2/N2 selectivity after the restart of the test following the flue gas return. The SW-20 module showed an initial CO2 permeance of around 800 GPU; however, the CO2 permeance dropped to ~ 630 GPU after 129 the restart of the test following the flue gas return. The CO2 permeance drop in the spiral- wound membrane module SW-20 was presumably due to the feed gas bypass which was caused by the glue failure. This failure was indicated by the change of epoxy glue color to light yellowish to brownish after 208 hours of testing as shown in Figure 4.16.

(a) (b) (c)

Figure 4.16. The color change of epoxy glue over testing time at NCCC (a) membrane element prior to testing (b) SW-18 after 96 hours of test (c) SW-20 after 208 hours of test.

Following the completion of the field test, , the glue curing was further improved and membrane indentations were reduced in order to resolve the issues that were identified from the field test. The glue curing was improved by increasing the curing temperature as well as the curing time from room temperature and 16 hours to 57oC and 48 hours, respectively. The membrane indentations were eliminated by using a nanoporous layer, such as polyvinylidene fluoride (PVDF) or polyethersulfone (PES), between the feed spacer and the selective-layer membrane or by using a feed spacer with a smoother surface

(Dexmet CPET feed spacer). Figure 4.17 shows the improved stability plot of the spiral- wound membrane module with the smooth Dexmet CPET feed spacer. As shown, this

130 module had stable CO2 permeance and CO2/N2 selectivity over a testing course of 205 hours. No obvious indentations were observed on the selective-layer after test completion, indicating the improvement in the spiral-wound membrane module fabrication. This improvement will most likely enable the spiral-wound membrane module to overcome the previously encountered issues during future field tests.

1000 350

900 300 800

700 250

600 200

500 Selectivity

150 2

400 /N

16% CO2 + 64% N2 + 17%H2O 2 Permeance (GPU) Permeance 300 3% O2 and 3 ppm SO2 100

2 16.6% CO2, CO 200 66.4% N2,

CO 17% H O 2 50 100

0 0 0 20 40 60 80 100 120 140 160 180 200 Time (hour) Figure 4.17. The stability plot of the improved spiral-wound membrane module tested with 3% O2 and 3 ppm SO2.

The field test results clearly showed the potential of the membrane which exhibited high gas transport performance, including high CO2/N2 selectivity along with the high CO2 permeance. Based on the U.S. Department of Energy NETL report [50], Gradient

Technology, in cooperation with Ho and coworkers, carried out a techno-economic analysis [51]. This analysis showed that the CO2 permeance of > 800 GPU obtained by 131 the spiral-wound membrane modules in this work corresponded to a CO2 capture cost of <

$40 (in 2007 dollar)/tonne, which is desirable [51].

4.5. Future directions

This work has demonstrated that an effective gas separation with the spiral-wound membrane module can be achieved. The spiral-wound membrane module developed in this work was able to achieve a similar result compared to the flat-sheet membrane used for the module fabrication. Further improvement on the spiral-wound membrane module may include the identification of a more appropriate glue and the optimization of the glue line and the feed and permeate spacers. Although the existing glue obtained from GS

Polymers can ensure a leak-free module after curing for 48 hours at 57oC, a glue that requires less curing time while provides the same performance is desirable for our future work. The glue lines inside the permeate channel can be adjusted to manage the flow path of the sweep gas in order to enable the full countercurrent configuration in the membrane leaf. The thickness of the feed and sweep spacers can be adjusted to accommodate a suitable pressure drop and packing density of the spiral-wound membrane module. The mesh configuration of the spacers (e.g., square, diamond, etc.) can possibly be engineered to improve the mixing of the gas inside each of the feed and sweep channels and eliminate the concentration polarization.

After the spiral-wound membrane module is optimized, another field test by using a larger active membrane area and a longer test time can be accomplished. A membrane skid that includes the spiral-wound membrane modules can be built and installed either at

132

NCCC or other coal-fired power plants for flue gas slipstream testing. The long-term stability of the performance of the spiral-wound membrane module as well as the thorough study of the effect of the contaminants in the actual flue gas can be conducted. In case the pressure drop of the spiral-wound membrane module is too high, the plate-and-frame configuration can be considered as an alternative since it typically yields a low pressure drop. The field test results will be a valuable information and a major step forward toward the commercialization of the post-combustion CO2 capture from the flue gas.

4.6. Conclusions

A new design of a spiral-wound membrane module by using face compression “O” rings as the sealing method for the gases was developed. The spiral-wound membrane modules fabricated were leak free and tested with both simulated flue gas at The Ohio State

University and the actual flue gas at the National Carbon Capture Center in Wilsonville,

Alabama. The concentration polarization phenomenon in the spiral-wound membrane module testing was observed at feed gas flow rates less than 1000 sccm. By using a feed gas flow rate of 1000 sccm, the modules demonstrated excellent gas transport performances with an average CO2 permeance of > 800 GPU and an average CO2/N2 selectivity of > 140, which were similar to the results obtained from the flat-sheet membrane used for the module fabrication. The pressure drops of the spiral-wound membrane modules acceptable, i.e., less than 1.5 psi/m. The fabrication procedure of the spiral-wound membrane module was improved in order to resolve the issues that were encountered during the field test. The gas transport performances obtained from the field test have

133 indicated that the proposed technology has the potential as the cost-effective capture of

CO2 from flue gas in coal-fired power plants.

Acknowledgments

I would like to thank José D. Figueroa of the U.S. Department of Energy/National

Energy Technology Laboratory for his helpful discussion. I also acknowledge the contributions of Dr. Varun Vakharia, Dr. Yuanxin Chen, Mr. Yang Han, and Mr. Dongzhu

Wu to this work. I thank Michael A. Wilson of the OSU Chemical and Biomolecular

Engineering and Peter Knappe and Jeff Flowers at TriSep Corporation for their suggestions and shared experiences in the spiral-wound membrane module fabrication. The smooth membrane module testing at NCCC would not be possible without the great efforts of

NCCC team members, particularly Tony Wu and Bob Lambrecht. I gratefully acknowledge the U.S. Department of Energy/National Energy Technology Laboratory

(DE-FE007632) and the Ohio Development Services Agency (OOE-CDO-D-13-05) for their financial support of this work. This work was partly supported by the United States

Department of Energy under Award Number DE-FE0007632 with substantial involvement of the National Energy Technology Laboratory, Pittsburgh, PA, USA.

Nomenclatures

푓 푛푖 molar flow rate of species i on the feed side

푝 푛푖 molar flow rate of species i on the sweep side

푊 length of membrane leaf

134

퐿 axial length of the module element

푥 axial direction of module element as shown in Figure C.1.

푦 longitude direction of the membrane leaf in the material coordinate as shown in

Figure C.1.

푃푖 permeance of species i

푝푓 absolute feed pressure

푝푠 absolute sweep pressure

푓 푛푖,0 molar flow rate of species i on the feed inlet

푝 푛푖,0 molar flow rate of species i on the sweep inlet

References

[1] J. Wilcox, Carbon capture, Springer, New York (2012).

[2] K. Ramasubramanian, Y. Zhao, W.S.W. Ho, CO2 capture and H2 purification:

prospects for CO2-selective membrane processes, AIChE J. 59 (2013) 1033–1045.

[3] M.E. Boot-Handford, J. C. Abanades, E.J. Anthony, M.J. Blunt, S. Brandani, N. Mac

Dowell, J.R. Fernandez, M.C. Ferrari, R. Gross, J.P. Harlett, R.S. Haszeldine, P.

Heptonstall, A. Lyngfelt, Z. Makuch, E. Mangano, R.T.J. Porter, M. Pourkashanian,

G.T. Rochelle, N. Shah, J.G. Yao, P.S. Fennell, Carbon capture and storage update,

Energy Environ. Sci. 7 (2014) 130–189.

[4] T.C. Merkel, H. Lin, X. Wei, R. Baker, Power plant post-combustion carbon dioxide

capture: an opportunity for membranes, J. Membr. Sci. 359 (2010) 126–139.

135

[5] K.A. Lokhandwala, I. Pinnau, Z. He, K.D. Amo, A.R. DaCosta, J.G. Wijmans, R.W.

Baker, Membrane separation of nitrogen from natural gas: a case study from

membrane synthesis to commercial deployment, J. Membr. Sci. 346 (2010) 270–279.

[6] H. Lin, S.M. Thompson, A.S. Martin, J.G. Wijmans, K.D. Amo, K.A. Lokhandwala,

T.C. Merkel, Dehydration of natural gas using membranes. Part I: composite

membranes, J. Membr. Sci. 413-414 (2012) 70–81.

[7] H. Lin, S.M. Thompson, A.S. Martin, J.G. Wijmans, K.D. Amo, K.A. Lokhandwala,

B.T. Low, T.C. Merkel, Dehydration of natural gas using membranes. Part II:

sweep/countercurrent design and field test, J. Membr. Sci. 432 (2013) 106–114.

[8] T. Brinkmann, C. Naderipour, J. Pohlmann, J. Wind, T. Wolff, E. Esche, D. Muller,

G. Wozny, B. Hoting, Pilot scale investigations of the removal of carbon dioxide

from hydrocarbon gas streams using poly(ethylene oxide) – poly(butylene

terephthalate) PolyactiveTM thin film composite membranes, J. Membr. Sci. 489

(2015) 237–247.

[9] M. Sandru, T.J. Kim, W. Capala, M. Huijbers, M.B. Hägg, Pilot scale testing of

polymeric membranes for CO2 capture from coal fired power plants, Energy Procedia

37 (2013) 6473–6480.

[10] C.A. Scholes, A. Qader, G.W. Stevens, S.E. Kentish, Membrane pilot plant trials of

CO2 separation from flue gas, Greenhouse Gas Sci. Technol. 5 (2015) 229–237.

[11] H. Lin, Z. He, Z. Sun, J. Vu, A. Ng, M. Mohammed, J. Kniep, T.C. Merkel, T. Wu,

R.C. Lambrecht, CO2-selective membranes for hydrogen production and CO2 capture

– Part I: membrane development, J. Membr. Sci. 457 (2014) 149–161.

136

[12] H. Lin, Z. He, Z. Sun, J. Kniep, A. Ng, R.W. Baker, T.C. Merkel, CO2-selective

membranes for hydrogen production and CO2 capture – Part II: techno-economic

analysis, J. Membr. Sci. 493 (2015) 794–806.

[13] L.S. White, X. Wei, S. Pande, T. Wu, T.C. Merkel. Extended flue gas trials with a

membrane-based pilot plant at a one-ton-per-day carbon capture rate, J. Membr. Sci.

496 (2015) 48–57.

[14] J. Zou, W.S.W. Ho, CO2-selective polymeric membranes containing amines in

crosslinked poly(vinyl alcohol), J. Membr. Sci. 286 (2006) 310–321.

[15] J. Huang, J. Zou, W.S.W. Ho, Carbon dioxide capture using a CO2-selective

facilitated transport membrane, Ind. Eng. Chem. Res. 47 (2008) 1261–1267.

[16] H. Bai, W.S.W. Ho, Carbon dioxide-selective membranes for high-pressure synthesis

gas purification, Ind. Eng. Chem. Res. 50 (2011) 12152–12161.

[17] Y. Zhao, W.S.W. Ho, Steric hindrance effect on amine demonstrated in solid polymer

membranes for CO2 transport, J. Membr. Sci. 415 (2012) 132–138.

[18] Y. Chen, Polymer membranes for flue gas carbon capture and fuel cell application

(Ph.D. dissertation), The Ohio State University (2015).

[19] W. Salim, W.S.W. Ho, Recent developments on nanostructured polymer-based

membranes, Current Opinion in Chem. Eng. 8 (2015) 76–82.

[20] D. Wu, L. Zhao, V.K. Vakharia, W. Salim, W.S.W. Ho, Synthesis and

characterization of nanoporous polyethersulfone membrane as support for composite

membrane in CO2 separation: from lab to pilot scale, J. Membr. Sci. 510 (2016) 58–

71.

137

[21] K. Ramasubramanian, CO2 (H2S)-selective membranes for fuel cell hydrogen

purification and flue gas carbon capture: an experimental and process modeling study

(Ph.D. dissertation), The Ohio State University (2013).

[22] K. Ramasubramanian, M.A. Severance, P.K. Dutta, W.S.W. Ho, Fabrication of

zeolite/polymer multilayer composite membranes for CO2 capture: Deposition of

zeolite particles on polymer supports, J. Colloid Interf. Sci. 452 (2015) 203–214.

[23] Y. Chen, B. Wang, L. Zhao, P. Dutta, W.S.W. Ho, New Pebax®/zeolite Y composite

membranes for CO2 capture from flue gas, J. Membr. Sci. 495 (2015) 415–423.

[24] L. Zhao, Y. Chen, B. Wang, C. Sun, S. Chakraborty, K. Ramasubramanian, P.K.

Dutta, W.S.W. Ho, Multilayer polymer/zeolite y composite membrane structure for

CO2 capture from flue gas, J. Membr. Sci. 498 (2016) 1–13.

[25] Y. Chen, L. Zhao, B. Wang, P. Dutta, W.S.W. Ho, Amine-containing

polymer/zeolite-Y composite membranes for CO2/N2 separation, J. Membr. Sci. 497

(2016) 21–28.

[26] Y. Chen, W.S.W. Ho, High-molecular-weight polyvinylamine/piperazine glycinate

membranes for CO2 capture from flue gas, J. Membr. Sci. 514 (2016) 376–384.

[27] W.S.W. Ho, FE0007632 Final project review meeting presentation: Novel

inorganic/polymer composite membranes for CO2 capture, Feb 2016 (available

online

https://www.netl.doe.gov/File%20Library/Research/Coal/carbon%20capture/post-

combustion/FE0007632-Project-Meeting-Final-public-release-2-26-16.pdf).

138

[28] D. Reddy, T.Y. Moon, C.E. Reineke, Counter current dual-flow spiral wound dual-

pipe membrane separation, U.S. Patent 5,034,126 (1991).

[29] D. Reddy, K. Denslow, Spiral-wound membrane separation device with feed and

permeate/sweep fluid flow control, U.S. Patent 5,096,584 (1992).

[30] P. Goebel, D. Olson, D.M. Polizzitti, A.R. Khwaja, Spiral wound membrane element

and treatment of SAGD produced water or other temperature alkaline fluids, U.S.

Patent 2012/0228219 (2012).

[31] S.S. Sablani, M.F.A. Goosen, R. Al-Belushi, M. Wilf, Concentration polarization in

ultrafiltration and reverse osmosis: a critical review, Desalination 141 (2001) 269–

289.

[32] G.T. Gray, J.R. McCutcheon, M. Elimelech, Internal concentration polarization in

forward osmosis: role of membrane orientation, Desalination 197 (2006) 1–8.

[33] J.G. Wijmans, S. Nakao, J.W.A. Van Den Berg, F.R. Troelstra, C.A. Smolders,

Hydrodynamic resistance of concentration polarization boundary layers in

ultrafiltration, J. Membr. Sci. 22 (1985) 117–135.

[34] S. Kim, E.M.V. Hoek, Modeling concentration polarization in reverse osmosis

processes, Desalination 186 (2005) 111–128.

[35] I. Sutzkover, D. Hasson, R. Semiat, Simple technique for measuring the

concentration polarization level in a reverse osmosis system, Desalination 131 (2000)

117–127.

[36] S. Lee, Y. Aurelle, H. Roques, Concentration polarization, membrane fouling and

cleaning in ultrafiltration of soluble oil, J. Membr. Sci. 19 (1984) 23–38.

139

[37] C.R. Bouchard, P.J. Carreau, T. Matsuura, S. Sourirajan, Modeling of ultrafiltration:

predictions of concentration polarization effects, J. Membr. Sci. 97 (1994) 215–229.

[38] S. Bhattacharjee, J.C. Chen, M. Elimelech, Coupled model of concentration

polarization and pore transport in crossflow nanofiltration, AIChE J. 47 (2001) 2733–

2745.

[39] G. Schulz, S. Ripperger, Concentration polarization in crossflow microfiltration, J.

Membr. Sci. 40 (1989) 173–187.

[40] O. Lüdtke, R.-D. Behling, K. Ohlrogge, Concentration polarization in gas

permeation, J. Membr. Sci. 146 (1998) 145–157.

[41] H. Takaba, S. Nakao, Computational fluid dynamics study on concentration

polarization in H2/CO separation membranes, J. Membr. Sci. 249 (2005) 83–88.

[42] S. Bhattacharya, S.T. Hwang, Concentration polarization, separation factor, and

Peclet number in membrane processes, J. Membr. Sci. 132 (1997) 73–90.

[43] K. Haraya, T. Hakuta, H. Yoshitome, A study of concentration polarization

phenomenon on the surface of a gas separation membrane, Sep. Sci. Tech. 22 (1987)

1425–1438.

[44] G. He, Y. Mi, P.L. Yue, G. Chen, Theoretical study on concentration polarization in

gas separation membrane processes, J. Membr. Sci. 153 (1999) 243–258.

[45] J. Zhang, D. Liu, M. He, He. Xu, W. Li, Experimental and simulation studies on

concentration polarization in H2 enrichment by highly permeable and selective Pd

membranes, J. Membr. Sci. 274 (2006) 83–91.

140

[46] J. Catalano, M.G. Baschetti, G.C. Sarti, Influence of the gas phase resistance on

hydrogen flux though thin palladium-silver membranes, J. Membr. Sci. 339 (2009)

57–67.

[47] A.M. Avila, H.H. Funke, Y. Zhang, J.L. Falconer, R.D. Noble, Concentration

polarization in SAPO-34 membranes at high pressures, J. Membr. Sci. 335 (2009)

32–36.

[48] A. Mourgues, J. Sanchez, Theoretical analysis of concentration polarization in

membrane modules for gas separation with feed inside the hollow-fibers, J. Membr.

Sci. 252 (2005) 133–144.

[49] A. Caravella, G. Barbieri, E. Drioli, Concentration polarization analysis in self

supported Pd-based membranes, Sep. Puri. Tech. 66 (2009) 613–624.

[50] J. Black, Cost and performance baseline for fossil energy plants, U.S. Department of

Energy, DOE/NETL-2010/1397, 2010.

[51] S.J. Schmit and J.R. Hitchingham, Technical and economic feasibility study for the

use of novel inorganic/polymer composite membrane for carbon dioxide capture,

Gradient Technology, GT/CR-2014-002-V1, 2014.

141

Chapter 5. Final Remarks and Recommendations

5.1. Final remarks

This dissertation covers the development of three different membranes for gas separation applications including H2 purification and CO2 capture. Two different membrane compositions were developed for H2 purification including oxidatively stable borate-containing membrane aimed for membrane process with air as sweep gas as well as amine-containing membrane aimed for simultaneous removal of acid gases (CO2 and H2S).

A novel inorganic/polymer composite membrane was developed for CO2 capture from flue gas of a coal-fired power plant.

The oxidatively stable membrane containing borate- and tetrafluoroborate-compounds showed improved and stable performances at least for 144 hours at 120oC with air as the sweep gas. The optimized membrane composition achieved a CO2 permeance > 100 GPU and CO2/H2 selectivity > 100. As the selective-layer thickness was reduced from 15 µm to

10 µm, a significant drop in CO2/H2 selectivity was observed due to the increase of the H2 permeance. Addition of more layers of substrate underneath the membrane helped in controlling the water permeation through the borate-containing membranes. The borate- containing membranes was scaled-up to prepare > 1400-ft of flat sheet membranes with a width of 14-in. The scale-up membranes showed similar performances compared to the lab-scale membranes.

The CO2-selective amine-containing membrane for H2S removal was scaled-up to fabricate an average selective-layer membrane thickness of around 15 µm with 14-in wide

142 and 150-ft long in total. In order to obtain the desired thickness, a coating solution concentration of 20 wt.%, a substrate web coating speed of 0.5 ft/min, and a coating knife gap setting of 6 mils were used. The scale-up membrane exhibited a CO2 permeance > 200

GPU, a CO2/H2 selectivity > 200, and a H2S/CO2 selectivity around 3, which is similar to the performance of the lab-scale membranes. The modeling results indicated that < 10 ppm of H2S concentration in the purified hydrogen gas stream was obtained, meeting the requirements set for shipboard fuel cell applications. The scale-up membranes were used to prepare spiral-wound membrane modules and the design operating conditions for the spiral-wound membrane modules test was determined based on the modeling calculations.

The results obtained from the field-test was < 10 ppm seemed of H2S concentration in the purified hydrogen gas stream, in agreement with the modeling results. This work demonstrated the potential of the CO2-selective membrane for the H2S removal in fuel cell applications, particularly in shipboard environment.

The novel inorganic/polymer composite membrane was used for fabrication of spiral- wound membrane modules and field-test at the National Carbon Capture Center in

Wilsonville, Alabama. A new design of a spiral-wound membrane module by using face compression “O” rings as an effective sealing method for the gases was developed. The concentration polarization phenomenon was observed at feed gas flow rates < 1000 sccm.

By using a feed gas flow rate of 1000 sccm, the modules demonstrated CO2 permeances of

> 800 GPU of and CO2/N2 selectivity > 140, similar to the results obtained from the flat- sheet membranes. In addition, the spiral-wound membrane modules showed a desirable pressure drops of < 1.5 psi/m. The results obtained from the field test have indicated that

143 the membrane process has the potential as the cost-effective technology for CO2 capture from flue gas in coal-fired power plants.

The scope of membrane research in this dissertation starts from lab-scale work until field test with actual industrial feed gas. Therefore, a unique perspective on the development of a membrane process was accomplished. While the research conducted in this dissertation has significantly move the membrane development towards commercialization, there remains future work and recommendations that can be planned as described in the next section.

5.2. Recommendations

The borate-containing membranes developed in this work could obtain a desirable CO2 permeance and excellent CO2/H2 selectivity with air sweep. Other borate-containing compounds, tetrafluoroborate-containing compounds, or the combinations thereof can be incorporated into the membrane to improve the membrane performance. Other CO2 carriers based on the hydroxide- and fluoride-containing salts and polymers, possibly with quaternaryphosphonium as the cation, can also be incorporated instead of the TMQOH and

PDADMQ-F that was used in this work. Other polymers can also be used as the membrane matrix instead of the crosslinked polyvinylalcohol, e.g., polymers that have higher molecular weights such as crosslinked polyvinylamine. The scale-up membrane can be used to prepare membrane modules, either spiral-wound or plate-and-frame, for a field test with industrial syngas used for solid oxide fuel cell. Integration of the membrane system with the solid oxide fuel cell stack can be performed subsequently. The robustness test of

144 the membrane can be performed to ensure the long-term stability and sensitivity of the membrane performance over fluctuations of operating conditions. The membrane synthesis can be mass-produced to allow the commercialization of the membrane process.

The scale-up amine-containing membranes in this work with 14-in wide and membrane thickness around 15 microns can be increased further to fabricate up to 21-in wide membranes, which is half of the width of commercial membranes. A higher substrate web coating speed up to 5 ft/min can be used to increase the efficiency of the membrane fabrication. The path length in the curing oven can be increased so that a sufficient curing of the membrane can still be obtained when the higher speed of 5 ft/min is used. Other additives can be used to slow down the increasing rate of viscosity of the coating solution and increase the shelf life of the coating solution. Another field test with industrial reformate gas using the design operating conditions can be performed to demonstrate the optimum membrane performance, preferably in specific application such as shipboard. The long-term stability as well as robustness test of the spiral-wound membrane modules can be conducted. The field test results will be a major step forward toward the commercialization of the CO2-selective membranes for simultaneous CO2 and H2S removal in fuel cell applications.

This work has demonstrated that an effective gas separation with the spiral-wound membrane module containing the novel inorganic/polymer composite membrane can be accomplished. Improvement on the spiral-wound membrane module fabrication procedure may include the identification of an improved glue, the optimization of the glue line, and the optimization of feed and permeate spacers. While the GS Polymers glue used in this

145 work can ensure a leak-free module after curing for 48 hours at 57oC, an improved glue that requires less curing time while provides the same performance is desirable for our future work. The glue lines inside on the permeate channel can be adjusted to manage the flow path of the sweep gas in order to enable the full countercurrent configuration in the membrane leaf. The thickness of the feed and sweep spacers can be adjusted to accommodate a suitable pressure drop and packing density of the spiral-wound membrane module. The mesh configuration of the spacers (e.g., square, diamond, etc.) can possibly be engineered to improve the mixing of the gas inside each of the feed and sweep channels and eliminate the concentration polarization. The configuration of the spiral-wound membrane module in this work is aimed for air sweep membrane process. Some modifications are necessary for vacuum permeate membrane process, including the removal of center plug in the central tube, increase of central tube holes size, change of membrane module material from acrylic to stainless steel, reduction of permeate spacer mesh size. Once the membrane module configuration is optimized, technology transfer to a membrane production company can be conducted to ensure the path forward to commercialization of the membrane technology.

146

Bibliography

A. Bello, R.O. Idem, Comprehensive study of the kinetics of the oxidative degradation of CO2 loaded and concentrated aqueous monoethanolamine (MEA) with and without sodium metavanadate during CO2 absorption from flue gas, Ind. Eng. Chem. Res. 45 (2006) 2569– 2579.

A. Caravella, G. Barbieri, E. Drioli, Concentration polarization analysis in self supported Pd-based membranes, Sep. Puri. Tech. 66 (2009) 613–624.

A. G. Eickmeyer, Method and compositions for removing acid gases from gaseous mixtures, U.S. Patent 4,271,132, 1981.

A. Heydari-Gorji, A. Sayari, Thermal, oxidative, and CO2-induced degradation of supported polyethylenimine adsorbents, Ind. Eng. Chem. Res. 51 (2012) 6887–6894.

A. Mourgues, J. Sanchez, Theoretical analysis of concentration polarization in membrane modules for gas separation with feed inside the hollow-fibers, J. Membr. Sci. 252 (2005) 133–144.

A.C. Lua, Y. Shen, Preparation and characterization of polyimide–silica composite membranes and their derived carbon–silica composite membranes for gas separation, Chem. Eng. J. 220 (2013) 441–451.

A.F. Ismail, N.H. Rahim, A. Mustafa, T. Matsuura, B.C. Ng, S. Abdullah, S.A. Hashemifard, Gas separation performance of polyethersulfone/multi-walled carbon nanotubes mixed matrix membranes. Sep. Purif. Tech. 80 (2011) 20–31.

A.M. Avila, H.H. Funke, Y. Zhang, J.L. Falconer, R.D. Noble, Concentration polarization in SAPO-34 membranes at high pressures, J. Membr. Sci. 335 (2009) 32–36.

A.Y. Alentiev, G.N. Bondarenko, Y.V. Kostina, V.P. Shantarovich, S.N. Klyamkin, V.P. Fedin, K.A. Kovalenko, Y.P. Yampolskii, PIM-1/MIL-101 hybrid composite membrane material: transport properties and free volume, Petroleum Chem. 54 (2014) 477–481.

Advanced hydrogen reformate stream purifier for fuel cell applications (available online: http://www.navysbir.com/n09_s/navst09-014.htm)

B. Zornoza, C. Tellez, J. Coronas, Mixed matrix membranes comprising glassy polymers and dispersed mesoporous silica spheres for gas separation, J. Membr. Sci. 368 (2011) 100–109.

147

B.L. Ge, Z. Zhu, V. Rudolph, Enhanced gas permeability by fabricating functionalized multi-walled carbon nanotubes and polyethersulfone nanocomposite membrane, Sep. Purif. Tech. 78 (2011) 76–82.

C. Song, Global challenges and strategies for control, conversion and utilization of CO2 for sustainable development involving energy, catalysis, adsorption and chemical processing, Catalysis Today 115 (2006) 2–32.

C.A. Scholes, A. Qader, G.W. Stevens, S.E. Kentish, Membrane pilot plant trials of CO2 separation from flue gas, Greenhouse Gas Sci. Technol. 5 (2015) 229–237.

C.H. Lau, K. Konstas, A.W. Thornton, A.C.Y Liu, S. Mudie, D.F. Kennedy, S.C. Howard, A.J. Hill, M.R. Hill, Gas-separation membranes loaded with porous aromatic frameworks that improve with age, Angew. Chem. Int. Ed. 54 (2015) 1–6.

C.R. Bouchard, P.J. Carreau, T. Matsuura, S. Sourirajan, Modeling of ultrafiltration: predictions of concentration polarization effects, J. Membr. Sci. 97 (1994) 215–229.

C.Y. Liang, P. Uchytil, R. Petrychkovych, Y.C. Lai, K. Friess, M. Sipek, M.M. Reddy, S. Y. Suen, A comparison on gas separation between PES (polyethersulfone)/MMT(Na- montmorillonite) and PES/TiO2 mixed matrix membranes, Sep. Purif. Technol. 92 (2012) 57–63.

D. Chinn, D. Vu, M. Driver, L. Boudreau, CO2 removal from gas using ionic liquid absorbents, U.S. Patent 0,129,598, 2005.

D. Reddy, K. Denslow, Spiral-wound membrane separation device with feed and permeate/sweep fluid flow control, U.S. Patent 5,096,584 (1992).

D. Reddy, T.Y. Moon, C.E. Reineke, Counter current dual-flow spiral wound dual-pipe membrane separation, U.S. Patent 5,034,126 (1991).

D. Wu, C. Sun, P.K. Dutta, W.S.W. Ho. "SO2 interference on separation performance of amine-containing facilitated transport membranes for CO2 capture from flue gas." J. Membr. Sci. 534 (2017) 33–45.

D. Wu, L. Zhao, V.K. Vakharia, W. Salim, W.S.W. Ho, Synthesis and characterization of nanoporous polyethersulfone membrane as support for composite membrane in CO2 separation: From lab to pilot scale, J. Membr. Sci. 510 (2016) 58–71.

D.T. Phan, M. Maeder, R.C. Burns, G. Puxty, Catalysis of CO2 absorption in aqueous solution by inorganic oxoanions and their application to post combustion capture, Environ. Sci. and Tech. 48 (2014) 4623–4629.

148

F.C. Krebs, Fabrication and processuing of polymer solar cells: A review of printing and coating techniques, Solar Energy Mat. Solar Cells. 93 (2009) 394–412.

G. George, N. Bhoria, A. AlHallaq, A. Abdala, V. Mittal, Polymer membranes for acid gas removal from natural gas, Sep. Purif. Tech. 158 (2016) 333–356.

G. He, Y. Mi, P.L. Yue, G. Chen, Theoretical study on concentration polarization in gas separation membrane processes, J. Membr. Sci. 153 (1999) 243–258.

G. Schulz, S. Ripperger, Concentration polarization in crossflow microfiltration, J. Membr. Sci. 40 (1989) 173–187.

G.Q. Lu, J.C. Diniz da Costa, M. Duke, S. Giessler, R. Socolow, R.H. Williams, T. Kreutz, Inorganic membranes for hydrogen production and purification: A critical review and perspective, J. Colloid Inter. Sci. 314 (2007) 589–603.

G.T. Gray, J.R. McCutcheon, M. Elimelech, Internal concentration polarization in forward osmosis: role of membrane orientation, Desalination 197 (2006) 1–8.

H. Bai, High temperature proton-exchange and fuel processing membranes for fuel cells and other applications (Ph.D. dissertation), The Ohio State University (2008).

H. Bai, W.S.W. Ho, Carbon dioxide-selective membranes for high-pressure synthesis gas purification, Ind. Eng. Chem. Res. 50 (2011) 12152–12161.

H. John, M. Cervi, A. Nickens, D. Hoffman, J. Kuseian, Fuel cell system models for U.S. Navy shipboard applications, Fuel Cell Seminar Palm Springs (2005).

H. Lepaumier, D. Picq, P.L. Carrette, New amines for CO2 capture. II. oxidative degradation mechanisms, Ind. Eng. Chem. Res. 48 (2009) 9068–9075.

H. Li, Z. Song, X. Zhang, Y. Huang, S. Li, Y. Mao, H.J. Ploehn, Y. Bao, M. Yu, Ultrathin, molecular-sieving graphene oxide membranes for selective hydrogen separation, Science, 342 (2013) 95–98.

H. Lin, S.M. Thompson, A.S. Martin, J.G. Wijmans, K.D. Amo, K.A. Lokhandwala, T.C. Merkel, Dehydration of natural gas using membranes. Part I: composite membranes, J. Membr. Sci. 413–414 (2012) 70–81.

H. Lin, S.M. Thompson, A.S. Martin, J.G. Wijmans, K.D. Amo, K.A. Lokhandwala, B.T. Low, T.C. Merkel, Dehydration of natural gas using membranes. Part II: sweep/countercurrent design and field test, J. Membr. Sci. 432 (2013) 106–114.

149

H. Lin, Z. He, Z. Sun, J. Kniep, A. Ng, R.W. Baker, T.C. Merkel, CO2-selective membranes for hydrogen production and CO2 capture – Part II: techno-economic analysis, J. Membr. Sci. 493 (2015) 794–806.

H. Lin, Z. He, Z. Sun, J. Vu, A. Ng, M. Mohammed, J. Kniep, T.C. Merkel, T. Wu, R.C. Lambrecht, CO2-selective membranes for hydrogen production and CO2 capture – Part I: Membrane development, J. Membr. Sci. 457 (2014) 149–161.

H. Sun, C. Ma, B. Yuan, T. Wang, Y. Xu, Q. Xue, P. Li, Y. Kong, Cardo polyimides/TiO2 mixed matrix membranes: Synthesis, characterization, and gas separation property improvement, Sep. Purif. Technol. 122 (2014) 367–375.

H. Takaba, S. Nakao, Computational fluid dynamics study on concentration polarization in H2/CO separation membranes, J. Membr. Sci. 249 (2005) 83–88.

H.W. Kim, H.W. Yoon, S.M. Yoon, B.M. Yoo, B.K. Ahn, Y.H. Cho, H.J. Shin, H. Yang, U. Paik, S. Kwon, J.Y. Choi, H.B. Park, Selective gas transport through few-layered graphene and graphene oxide membranes, Science, 342 (2013) 91–94.

I. Sutzkover, D. Hasson, R. Semiat, Simple technique for measuring the concentration polarization level in a reverse osmosis system, Desalination 131 (2000) 117–127.

ISO/TS 80004-1:2010 ISO Nanotechnologies, Vocabulary, Core Terms. International Organization for Standardization (ISO), Geneva; 2010.

J. Ahmad, M.B. Hägg, Polyvinyl acetate/titanium dioxide nanocomposite membranes for gas separation, J. Membr. Sci. 445 (2013) 200–210.

J. Black, Cost and performance baseline for fossil energy plants, U.S. Department of Energy, DOE/NETL-2010/1397, 2010.

J. Catalano, M.G. Baschetti, G.C. Sarti, Influence of the gas phase resistance on hydrogen flux though thin palladium-silver membranes, J. Membr. Sci. 339 (2009) 57–67.

J. Field, Separation of CO2 from gas mixtures, U.S. Patent 3,907,969, 1975.

J. Hao, P.A. Rice, S.A. Stern, Upgrading low-quality natural gas with H2S- and CO2- selective polymer membranes Part I. Process design and economics of membrane stages without recycle streams, J. Membr. Sci. 209 (2002) 177–206.

J. Huang, CO2 (H2S) membrane separations and WGS membrane reactor modeling for fuel cells (Ph.D. dissertation), The Ohio State University (2007).

150

J. Huang, J. Zou, W.S.W. Ho, Carbon dioxide capture using a CO2-selective facilitated transport membrane, Ind. Eng. Chem. Res. 47 (2008) 1261–1267.

J. Huang, L. El-Azzami, W.S.W. Ho, Modeling of CO2-selective water gas shift membrane reactor for fuel cell, J. Membr. Sci. 261 (2005) 67–75.

J. Morse, Nanofabrication technologies for roll-to-roll processing, NIST-NNN Workshop Report, September 2011.

J. Shen, G. Liu, K. Huang, W. Jin, K.R. Lee, N. Xu, Membranes with fast and selective gas-transport channels laminar graphene oxide for efficient CO2 capture, Angew. 127 (2015) 588–592.

J. Wilcox, Carbon capture, Springer, New York (2012).

J. Zhang, D. Liu, M. He, He. Xu, W. Li, Experimental and simulation studies on concentration polarization in H2 enrichment by highly permeable and selective Pd membranes, J. Membr. Sci. 274 (2006) 83–91.

J. Zou, Carbon dioxide-selective membranes and their applications in hydrogen processing (Ph.D. dissertation), The Ohio State University (2007).

J. Zou, W.S.W. Ho, CO2-selective polymeric membranes containing amines in crosslinked poly (vinyl alcohol), J. Membr. Sci. 286 (2006) 310–321.

J.A. Turner, A realizable renewable energy future, Science 285 (1999) 687–689.

J.F. Brennecke, E.J. Maginn, Ionic liquids: innovative fluids for chemical processing, AIChE J. 47 (2001) 2384–2389.

J.G. Wijmans, S. Nakao, J.W.A. Van Den Berg, F.R. Troelstra, C.A. Smolders, Hydrodynamic resistance of concentration polarization boundary layers in ultrafiltration, J. Membr. Sci. 22 (1985) 117–135.

K. Haraya, T. Hakuta, H. Yoshitome, A study of concentration polarization phenomenon on the surface of a gas separation membrane, Sep. Sci. Tech. 22 (1987) 1425–1438.

K. Liu, C. Song, V. Subramani, Hydrogen and syngas production and purification, John Wiley & Sons, 2010.

K. Ramasubramanian, CO2 (H2S)-selective membranes for fuel cell hydrogen purification and flue gas carbon capture: an experimental and process modeling study (Ph.D. dissertation), The Ohio State University (2013).

151

K. Ramasubramanian, M.A. Severance, P.K. Dutta, W.S.W. Ho, Fabrication of zeolite/polymer multilayer composite membranes for CO2 capture: Deposition of zeolite particles on polymer supports, J. Colloid Interf. Sci. 452 (2015) 203–214.

K. Ramasubramanian, W.S.W. Ho, Recent developments on membranes for post- combustion carbon capture, Curr. Opinion Chem. Eng. 1 (2011) 47–54.

K. Ramasubramanian, Y. Zhao, W.S.W. Ho, CO2 capture and H2 purification: Prospects for CO2-selective membrane processes, AIChE J. 59 (2013) 1033–1045.

K.A. Lokhandwala, I. Pinnau, Z. He, K.D. Amo, A.R. DaCosta, J.G. Wijmans, R.W. Baker, Membrane separation of nitrogen from natural gas: a case study from membrane synthesis to commercial deployment, J. Membr. Sci. 346 (2010) 270–279.

L. Ansaloni, Y. Zhao, B.T. Jung, K. Ramasubramanian, M.G. Baschetti, W.S.W. Ho, Facilitated transport membranes containing amino-functionalized multi-walled carbon nanotubes for high-pressure CO2 separations, J. Membr. Sci. 490 (2015) 18–28.

L. Deng, M.B. Hägg, Carbon nanotube reinforced PVAm/PVA blend FSC nanocomposite membrane for CO2/CH4 separation, Int. J. Greenhouse Gas Control, 26 (2014) 127–134.

L. Hao, J. Zuo, T.S. Chung, Formation of defect-free polyetherimide/PIM-1 hollow fiber membranes for gas separation, AIChE J. 60 (2014) 3848–3858.

L. Wengeler, M. Schmitt, K. Peters, P. Scharfer, W. Schabel, Comparison of large scale coating techniques for organic and hybrid films in polymer based solar cells, Chem. Eng. Proc. 68 (2013) 38–44.

L. Xiong, S. Gu, K.O. Jensen, Y.S. Yan, Facilitated transport in hydroxide-exchange membranes for post-combustion CO2 separation, Chem. Sus. Chem. Com. 7 (2014) 114– 116.

L. Zhao, Y. Chen, B. Wang, C. Sun, S. Chakraborty, K. Ramasubramanian, P.K. Dutta, W.S.W. Ho, Multilayer polymer/zeolite y composite membrane structure for CO2 capture from flue gas, J. Membr. Sci. 498 (2016) 1–13.

L.S. White, X. Wei, S. Pande, T. Wu, T.C. Merkel. Extended flue gas trials with a membrane-based pilot plant at a one-ton-per-day carbon capture rate, J. Membr. Sci. 496 (2015) 48–57.

M. Gasda, A. Ballantine, C. Karuppaiah, W.S.W. Ho, Carbon dioxide separator, fuel cell system including same, and method of operating the fuel cell system, U.S. Patent 2016/0248111 A1, 2016.

152

M. Sandru, T.J. Kim, W. Capala, M. Huijbers, M.B. Hägg, Pilot scale testing of polymeric membranes for CO2 capture from coal fired power plants, Energy Procedia 37 (2013) 6473–6480.

M. Severance, B. Wang, K. Ramasubramanian, L. Zhao, W.S.W. Ho, P.K. Dutta, Rapid crystallization of faujasitic zeolites: mechanism and application to zeolite membrane growth on polymer supports, Langmuir, 30 (2014) 6929–6937.

M.E. Boot-Handford, J. C. Abanades, E.J. Anthony, M.J. Blunt, S. Brandani, N. Mac Dowell, J.R. Fernandez, M.C. Ferrari, R. Gross, J.P. Harlett, R.S. Haszeldine, P. Heptonstall, A. Lyngfelt, Z. Makuch, E. Mangano, R.T.J. Porter, M. Pourkashanian, G.T. Rochelle, N. Shah, J.G. Yao, P.S. Fennell, Carbon capture and storage update, Energy Environ. Sci. 7 (2014) 130–189.

M.M. Sharma, P. V. Danckwerts, Catalysis by Brönsted bases of the reaction between CO2 and water, Transactions of the Faraday Society 59 (1963) 386–395.

M.M. Sharma, P. V. Danckwerts, Fast reactions of CO2 in alkaline solutions – (a) Carbonate buffers with arsenite, formaldehyde and hypochlorite as catalysts (b) Aqueous monoisopropanolamine (1-amino-2-propanol) solutions, Chem. Eng. Sci. 18 (1963) 729– 735.

M.W. Uddin, M.B. Hägg, Natural gas sweetening – the effect on CO2-CH4 separation after exposing a facilitated transport membrane to hydrogen sulfide and higher hydrocarbons, J. Membr. Sci. 423–424 (2012) 143–149.

N. C. Mueller, V. Van der Bruggen, V. Keuter, P. Luis, T. Melin, W. Pronk, R. Reisewitz, D. Rickerby, G.M. Rios, W. Wennekes, B. Nowack, Nanofiltration and nanostructured membranes - Should they be considered nanotechnology or not?, J. Haz. Mater. 211-212 (2012) 275–280.

O. Lüdtke, R.-D. Behling, K. Ohlrogge, Concentration polarization in gas permeation, J. Membr. Sci. 146 (1998) 145–157.

P. Goebel, D. Olson, D.M. Polizzitti, A.R. Khwaja, Spiral wound membrane element and treatment of SAGD produced water or other temperature alkaline fluids, U.S. Patent 2012/0228219 (2012).

P. Li, Z. Wang, Z. Qiao, Y. Liu, X. Cao, W. Li, J. Wang, S. Wang, Recent development in membranes for efficient hydrogen purification, J. Membr. Sci. 495 (2015) 130–168.

P.A. Giguere, S. Turrell, The nature of hydrofluoric acid. A spectroscopic study of the + - proton-transfer complex H3O .F , J. Am. Chem. Soc. 102 (1980) 5473–5477.

153

R. Baker, Future directions of membrane gas-separation technology, Membr. Tech. 138 (2001) 5–10.

R. Quinn, D.V. Laciak, G.P. Pez, Polyelectrolyte-salt blend membranes for acid gas separation, J. Membr. Sci. 131 (1997) 61–69.

R. Quinn, D.V. Laciak, G.P. Pez, Process for separating acid gases from gaseous mixtures utilizing composite membranes formed from salt-polymer blends, U.S. Patent 6,315, 968, 2001.

R. Quinn, D.V. Laciak, J.B. Appleby, G.P. Pez, Polyelectrolyte membranes for separation of acid gases, U.S. Patent 5,336,298, 1994.

R. Quinn, G.P. Pez, J.B. Appleby, Molten salt hydrate membranes for the separation of gases, U.S. Patent 4,780,114, 1988.

R. Xing, W.S.W. Ho, Crosslinked polyvinylalcohol-polysiloxane/fumed silica mixed matrix membranes containing amines for CO2/H2 separation, J. Membr. Sci. 367 (2011) 91–102.

R.J. Lander, J.A. Quinn, The use of membranes in studies of reaction kinetics: Arsenite catalysis of CO2 hydration, J. Membr. Sci. 3 (1978) 47–56.

S. Bae, H. Kim, Y. Lee, X. Xu, J.S. Park, Y. Zheng, J. Balakrishnan, T. Lei, H.R. Kim, Y.I. Song, Y.J. Kim, K.S. Kim, B. Ozyilmaz, J.H. Ahn, B.H. Hong, S. Iijima, Roll-to-roll production of 30-inch graphene films for transparent electrodes, Nat. Nanotechnol. 5 (2010) 574–578.

S. Bhattacharjee, J.C. Chen, M. Elimelech, Coupled model of concentration polarization and pore transport in crossflow nanofiltration, AIChE J. 47 (2001) 2733–2745.

S. Bhattacharya, S.T. Hwang, Concentration polarization, separation factor, and Peclet number in membrane processes, J. Membr. Sci. 132 (1997) 73–90.

S. Gu, J. Wang, B. Zhang, R.B. Kaspar, Y. Yan, Hydroxide exchange membranes and ionomers, in: B. Ladewig, S.P. Jiang, Y. Yan, M. Lu (Eds.), Materials for Low Temperature Fuel Cells, Wiley-VCH, New York, 2015, pp. 130–134.

S. Japip, H. Wang, Y. Xiao, T.S. Chung, Highly permeable zeolitic imidazolate framework (ZIF)-71 nano-particles enhanced polyimide membranes for gas separation, J. Membr. Sci. 467 (2014) 162–174.

S. Kim, E.M.V. Hoek, Modeling concentration polarization in reverse osmosis processes, Desalination 186 (2005) 111–128. 154

S. Lee, Y. Aurelle, H. Roques, Concentration polarization, membrane fouling and cleaning in ultrafiltration of soluble oil, J. Membr. Sci. 19 (1984) 23–38.

S. Sorribas, B. Zornoza, C. Téllez, J. Coronas, Mixed matrix membranes comprising silica- (ZIF-8) core–shell spheres with ordered meso–microporosity for natural- and bio-gas upgrading, J. Membr. Sci. 452 (2014) 184–192.

S.A. Didas, R. Zhu, N.A. Brunelli, D.S. Sholl, C.W. Jones, Thermal, oxidative and CO2 induced degradation of primary amines used for CO2 capture: effect of alkyl linker on stability, J. Phys. Chem. C 118 (2014) 12302–12311.

S.J. Schmit and J.R. Hitchingham, Technical and economic feasibility study for the use of novel inorganic/polymer composite membrane for carbon dioxide capture, Gradient Technology, GT/CR-2014-002-V1, 2014.

S.S. Madaeni, M.M.S. Badieh, V. Vatanpour, N. Ghaemi, Effect of titanium dioxide nanoparticles on polydimethylsiloxane/polyethersulfone composite membranes for gas separation, Polym. Eng. Sci. 52 (2012) 2664–2674.

S.S. Sablani, M.F.A. Goosen, R. Al-Belushi, M. Wilf, Concentration polarization in ultrafiltration and reverse osmosis: a critical review, Desalination 141 (2001) 269–289.

T. Brinkmann, C. Naderipour, J. Pohlmann, J. Wind, T. Wolff, E. Esche, D. Muller, G. Wozny, B. Hoting, Pilot scale investigations of the removal of carbon dioxide from hydrocarbon gas streams using poly(ethylene oxide) – poly(butylene terephthalate) PolyactiveTM thin film composite membranes, J. Membr. Sci. 489 (2015) 237–247.

T. Li, Y. Pan, K.V. Peinemann, Z. Lai, Carbon dioxide selective mixed matrix composite membrane containing ZIF-7 nano-fillers, J. Membr. Sci. 425-426 (2013) 235–242.

T.C. Merkel and L.G. Toy, Comparison of Hydrogen Sulfide Transport Properties in Fluorinated and Nonfluorinated Polymers, Macromolecules 39 (2006) 7591–7600.

T.C. Merkel, H. Lin, X. Wei, R. Baker, Power plant post-combustion carbon dioxide capture: an opportunity for membranes, J. Membr. Sci. 359 (2010) 126–139.

U. Ghosh, S. Kentish, G. Stevens, “Absorption of carbon dioxide into aqueous potassium carbonate promoted by boric acid”, Energy Pro. 1 (2009) 1075–1081.

155

U. Siemann, Solvent cast technology - A versatile tool for thin film production, Prog. Colloid Polym. Sci. 130 (2005) 1–14.

United States Department of Energy Fuel Cell Technologies Office, State of the states: Fuel cells in America 2016, 7th ed., (2016).

V. Nafisi, M.B. Hägg, Development of dual layer of ZIF-8/PEBAX-2533 mixed matrix membrane for CO2 capture, J. Membr. Sci. 459 (2014) 244–255.

V. Nafisi, M.B. Hägg, Development of nanocomposite membranes containing modified Si nanoparticles in: PEBAX-2533 as a block copolymer and 6FDA-Durene Diamine as a glassy polymer, ACS Appl. Mater. Interfaces 6 (2014) 15643–15652.

V. Vakharia, Development of membrane technology for CO2 removal and H2 purification: A techno-economic, lab-scale, and pilot-scale study (Ph.D. dissertation), The Ohio State University (2016).

V. Vakharia, K. Ramasubramanian, W.S.W. Ho, An experimental and modeling study of CO2-selective membranes for IGCC syngas purification, J. Membr. Sci. 488 (2015) 56–66.

V. Vakharia, W. Salim, M. Gasda, W.S.W. Ho, Oxidatively stable membranes for CO2 separation and H2 purification, J. Membr. Sci. 533 (2017) 220–228.

V. Vakharia, W.S.W. Ho, Separation and Purification of Hydrogen Using CO2-Selective Facilitated Transport Membranes, Hydrogen Production from Renewable Resources, Springer Book Series - Biofuels and Biorefineries, Chap. 11, (2015) 315–338.

W. Salim, W.S.W. Ho, Recent developments on nanostructured polymer-based membranes, Current opinion in Chem. Eng. 8 (2015) 76–82.

W. Yave, A. Car, J. Wind, K.V Peinemann, Nanometric thin film membranes manufactured on square meter scale: ultra-thin films for CO2 capture, Nanotechno. 21 (2010) 395301.

W.J. Koros, R. Mahajan, Pushing the limits on possibilities for large scale gas separation: which strategies?, J. Membr. Sci. 175 (2000) 181–196.

W.S.W. Ho, FE0007632 Final project review meeting presentation: Novel inorganic/polymer composite membranes for CO2 capture, Feb 2016 (available online https://www.netl.doe.gov/File%20Library/Research/Coal/carbon%20capture/post- combustion/FE0007632-Project-Meeting-Final-public-release-2-26-16.pdf).

W.S.W. Ho, P.K. Dutta, S. Schmit, Novel inorganic/polymer composite membranes for CO2 capture. 2014 NETL CO2 Capture Technology Meeting, (Available online: 156 http://www.netl.doe.gov/File%20Library/Events/2014/2014%20NETL%20CO2%20Capt ure/W-Ho-OSU-Inorganic-Polymer-Composite-Membranes.pdf).

W.S.W. Ho, V. Vakharia, and W. Salim, Borate-containing membranes for gas separations, U.S. Patent No. 62416434, 2016.

W.S.W. Ho, W. Salim, V. Vakharia, Membranes for gas separation, U.S. Patent No. 20170056839 A1, 2017.

Y. Chen, B. Wang, L. Zhao, P. Dutta, W.S.W. Ho, New Pebax®/zeolite Y composite membranes for CO2 capture from flue gas, J. Membr. Sci. 495 (2015) 415–423.

Y. Chen, L. Zhao, B. Wang, P. Dutta, W.S.W. Ho, Amine-containing polymer/zeolite-Y composite membranes for CO2/N2 separation, J. Membr. Sci. 497 (2016) 21–28.

Y. Chen, Polymer membranes for flue gas carbon capture and fuel cell application (Ph.D. dissertation), The Ohio State University (2015).

Y. Chen, W.S.W. Ho, High-molecular-weight polyvinylamine/piperazine glycinate membranes for CO2 capture from flue gas, J. Membr. Sci. 514 (2016) 376–384.

Y. Zhao, B.T. Jung, L. Ansaloni, W.S.W. Ho, Multiwalled carbon nanotube mixed matrix membranes containing amines for high pressure CO2/H2 separation, J. Membr. Sci. 459 (2014) 233–243.

Y. Zhao, Carbon dioxide-selective membranes containing sterically hindered amines (Ph.D. dissertation), The Ohio State University (2013).

Y. Zhao, W.S.W. Ho, Steric hindrance effect on amine demonstrated in solid polymer membranes for CO2 transport, J. Membr. Sci. 415 (2012) 132–138.

Z. Qiao, Z. Wang, C. Zhang, S. Yuan, Y. Zhu, J. Wang, PVAm–PIP/PS composite membrane with high performance for CO2/N2 separation, AIChE J. 59 (2013) 215–228.

Z. Tong, CO2 facilitated transport membranes for hydrogen purification and flue gas carbon capture (Ph.D. dissertation), The Ohio State University (2017).

Z. Tong, V. Vakharia, M. Gasda, W.S.W. Ho, Water vapor and CO2 transport through amine-containing facilitated transport membranes, React. Funct. Polym. 86 (2015) 111– 116.

157

Appendix A. Design of experiment for membrane composition optimization with TFBA

as the catalyst

The membrane compositions generated by JMP are shown in Table A.1. Five levels of

TMQOH contents were explored including 6, 7, 7.5, 8, and 9 wt%. Four levels of XLPVA-

POS content were explored including 32.5, 35, 36.5, and 38 wt%. Four levels of TFBA content were explored including 2, 3, 5, and 6 wt%. In total, 28 membranes were prepared in randomized order with compositions as in Table A.1. The gas transport properties of each compositions were collected and shown in Table A.1.

The experimental data were analyzed by using JMP software. The analysis of variance, effect tests, prediction expression, and profiler for both response variables (CO2 permeance and CO2/H2 selectivity) are shown in Figure A.1. The analysis of variances indicated that the model used is adequate. The effect tests showed that the CO2 permeance and CO2/H2 selectivity are significantly affected by all the terms proposed in the model. The prediction expressions generated by JMP should be valid for the range of compositions listed in Table

A.1. The profiler showed the correlation between the CO2 permeance and CO2/H2 selectivity and each components of the membrane compositions.

158

Table A.1. The experimental design membrane compositions and transport performances of TFBA-containing membranes. Sample TMQOH XLPVA- TFBA PDADMQ- CO2 CO2/H2 (wt.%) POS (wt.%) F Permeance Selectivity (wt.%) (wt.%) (GPU) M-1 6 32.5 2 59.5 81 99 M-2 6 32.5 3 58.5 85 96 M-3 6 35 2 57 72 109 M-4 6 35 3 56 75 107 M-5 7 32.5 2 58.5 96 120 M-6 7 32.5 3 57.5 99 119 M-7 7 35 2 56 90 130 M-8 7 35 3 55 91 127 M-9 7 35 5 53 96 116 M-10 7 35 6 52 96 114 M-11 7 36.5 5 51.5 93 127 M-12 7 36.5 6 50.5 93 128 M-13 7 38 5 50 89 142 M-14 7 38 6 49 89 141 M-15 7.5 35 5 52.5 99 117 M-16 7.5 35 6 51.5 100 114 M-17 7.5 36.5 5 51 98 127 M-18 7.5 36.5 6 50 98 122 M-19 7.5 38 5 49.5 92 137 M-20 7.5 38 6 48.5 92 134 M-21 8 32.5 2 57.5 99 115 continued

159

Table A.1 continued

M-22 8 32.5 3 56.5 100 118 M-23 8 35 2 55 93 127 M-24 8 35 3 54 96 128 M-25 9 32.5 2 56.5 94 109 M-26 9 32.5 3 55.5 96 107 M-27 9 35 2 54 85 123 M-28 9 35 3 53 87 118

160

Analysis of Variance

Source DF Sum of Mean Square F Ratio Squares Model 5 237972.09 47594.4 26118.90 Error 23 41.91 1.8 Prob > F C. Total 28 238014.00 <.0001*

Effect Tests

Source Nparm DF Sum of F Ratio Prob > F Squares TMAOH 1 1 617.18374 338.6985 <.0001* XLPVA-POS 1 1 267.17579 146.6209 <.0001* TFBA 1 1 128.33565 70.4281 <.0001* TMAOH*TMAOH 1 1 564.11042 309.5729 <.0001* XLPVA-POS*XLPVA-POS 1 1 152.14128 83.4922 <.0001*

Prediction Expression

Profiler

(a) continued

Figure A.1. The JMP analysis results (analysis of variance, effect tests, prediction expression, and profiler) for response variables: (a) CO2 permeance (b) CO2/H2 selectivity.

161

Figure A.1 continued

Analysis of Variance

Source DF Sum of Mean Square F Ratio Squares Model 5 409129.85 81826.0 6599.960 Error 23 285.15 12.4 Prob > F C. Total 28 409415.00 <.0001*

Effect Tests

Source Nparm DF Sum of F Ratio Prob > F Squares TMAOH 1 1 876.8129 70.7224 <.0001* XLPVA-POS 1 1 789.0558 63.6441 <.0001* TFBA 1 1 396.9913 32.0207 <.0001* TMAOH*TMAOH 1 1 831.3085 67.0521 <.0001* XLPVA-POS*XLPVA-POS 1 1 1182.1297 95.3488 <.0001*

Prediction Expression

Profiler

(b)

162

Appendix B. Modeling results of membrane module performance for a feed pressure of 5

psig

In comparison to the modeling results for a feed pressure of 10 psig as shown in section

3.4.1, the modeling results for a feed pressure of 5 psig is presented here. The CO2 removal,

H2S removal, H2 recovery, retentate H2S concentration, and retentate water content for a feed pressure of 5 psig with different sweep-to-feed flow rate ratios (1 – 4.2) and sweep water contents (0 – 40%) are shown in Figure B.1(a) – (e).

100%

90%

80%

70% Removal

2 60%

CO 5 psig feed pressure Sweep water content 50% 0% 10% 17% 20% 30% 40% 40%

30% 1 2 3 4 Sweep:Feed Flow Rate Ratio

(a) continued Figure B.1. Modeling results of membrane module performance for a feed pressure of 5 psig: (a) CO2 removal (b) H2S removal (c) H2 recovery (d) retentate H2S concentration on wet basis (e) retentate water content.

163

Figure B.1 continued

100%

95%

90%

S Removal S 2

H 5 psig feed pressure Sweep water content 85% 0% 10% 17% 20% 30% 40%

80% 1 2 3 4 Sweep:Feed Flow Rate Ratio

(b)

100%

99%

98%

ecovery R

2 97% H 5 psig feed pressure Sweep water content 0% 10% 17% 96% 20% 30% 40%

95% 1 2 3 4 Sweep:Feed Flow Rate Ratio

Figure B.1 continued

12 5 psig feed pressure Sweep water content 10 0% 10% 17% 20% 30% 40% 8

6 S Conc., Conc., S BasisWet (ppm) 2 4

etentate H etentate R

0 1 2 3 4 Sweep:Feed Flow Rate Ratio

(d)

50% 5 psig feed pressure 45% Sweep water content 40% 0% 10% 17% 20% 30% 40% 35%

30%

25%

20% Close to initial feed water content

15%

etentate WaterContent etentate R 10%

0% 1 2 3 4 Sweep:Feed Flow Rate Ratio

(e) 165

Appendix C. Modeling for spiral-wound membrane module

The gas permeation in a membrane module is simulated by a two-dimensional crossflow model. The module configuration, along with the flow directions, is illustrated in Figure C.1. This model assumes:

• The sweep flow channel is formed by two pieces of membrane;

• The membrane is only permeable for CO2, N2, and H2O;

• The permeance of each species is independent of pressure and gas composition;

• The feed and permeate streams are in plug flow;

• There is negligible pressure drop along the flow directions.

푊 푦

푥 퐿

Figure C.1. Module configuration and flow directions in the crossflow module. Blue and red arrows denote the feed and sweep flows, respectively.

The mass balance of each species on the feed and permeate sides can be expressed as

휕푛푓 푛푓 푛푝 푖 = −2푊푃 (푝 푖 − 푝 푖 ) , 0 ≤ 푥 ≤ 퐿 (A. 1) 휕푥 푖 푓 푓 푝 ∑ 푛푝 ∑ 푛푖 푖

휕푛푝 푛푓 푛푝 퐿 푖 = −퐿푃 (푝 푖 − 푝 푖 ) , 0 ≤ 푥 < (A. 2) 휕푦 푖 푓 푓 푝 ∑ 푛푝 2 ∑ 푛푖 푖

166

휕푛푝 푛푓 푛푝 퐿 푖 = 퐿푃 (푝 푖 − 푝 푖 ) , ≤ 푥 ≤ 퐿 (A. 3) 휕푦 푖 푓 푓 푝 ∑ 푛푝 2 ∑ 푛푖 푖

푓 푝 where 푛푖 and 푛푖 are the molar flow rates of species 푖 in the feed and permeate sides, respectively; 푝푓 and 푝푝 are the feed and permeate pressures; 푃푖 is the permeance of species

푖. The coordinate and module dimensions are shown in Figure C.1. The permeances of gas components are related by the selectivity 훼푖/푗 = 푃푖⁄푃푗. The boundary conditions for the crossflow module are

푛푓| = 푛푓 (A. 4) 푖 푥=0 푖,0

푝 푝 푛 | 퐿 = 푛 (A. 5) 푖 푦=0, ≤푥≤퐿 푖,0 2

퐿 퐿 2 ∫ (푛푝| ) 푑푥 = ∫ (푛푝| ) 푑푥 (A. 6) 푖 푦=푊 퐿 푖 푦=푊 0 2

푓 푝 where 푛푖,0 and 푛푖,0 are the molar flow rates of species 푖 at the feed and sweep inlets, respectively. Equations (A.1) – (A.6) are fed into the partial differential equation solver in

® Comsol Multiphysics and solved by the collocation method. To extract the CO2 permeance and CO2/N2 selectivity from the module permeation test data, the molar fractions of CO2 and N2 on the retentate and permeate sides, respectively, are calculated based on the gas chromatograph analysis results. Then, an optimization node is incorporated in the partial differential solver to calculate the required CO2 permeance and

CO2/N2 selectivity that fulfill the retentate and permeate stream compositions.

167