My Project Final

Hydrogen Recovery of Refinery purge/flare Gases Using Membrane Processes BSc. Thesis

Author: Ehsan Ahmadpour Samani

Supervisor: Dr. Reza Mosayebi Behbahani

Ahwaz - Iran June 2012

I

© Copyright by EHSAN AHMADPOUR SAMANI 2012 All Rights Reserved.

II

III

Acknowledgements I wish to express my gratitude to those who helped me in the past four years during my study in Petroleum University of Technology, and to those who contributed to the completion of this thesis. At this moment of accomplishment, first of All, I would like to take the opportunity to thank my supervisor Dr. Reza Mosayebi Behbahani, who motivate and supervise me to knock out difficulties and make my thesis work progressing smoothly. I am sincerely and heartily grateful to Mr. Abolhamid Salahi, PHD student at Iran University of Science and Technology, who offers me valuable advice on the direction of my thesis work. Moreover, he shared his experiences and knowledge unreservedly, which will be great treasure for me in my personal development and future career. I am sure it would have not been possible without his help. Moreover, I would like to thank Dr. Ghayyem and Dr. Mehrabani, who helped me understand many practical problems, shared their experience and guided me on how to become a matured engineer. Sincere thanks are given to my close friend Mr. Vahid Sarfaraz who always offers me valuable advice when I face dilemma, and encourage me to confront and tackle challenges. He comforts me when I was depressed, and share my every piece of happiness and sadness. I wish to thank my excellent friends Mohammad Ahmadvand, Esmaeail Hamidpour, Kamran Janghorban, Farshad Riahi, Alireza Bolandghamatpour, Reza Shokrani, masoud rahnemai and Masih Adyani. I would like to dedicate this piece of work to my beloved grandfather who passed away. His love always inspired and encouraged me to fulfill the desire. Finally, from deep within my heart, I am greatly grateful to my parents for their endless love and support, and I am sincerely grateful to my brothers for their constant encouragement.

IV

Abstract Membrane processes are considered to be viable and effective technologies for the separation of purge/flare gaseous mixtures at the industrial-scale due to their high efficiency, simple operation and low (capital and operating) cost. Membrane technology is becoming more useful and effective for separation of gas mixture and offers great advantages in its industrial operations. Ceramic, metallic and polymeric membranes had been successfully used for this purpose. However, their permeability-selectivity combination is still not up to the industries satisfaction and their application is limited especially related to severe environment such as higher temperature and corrosive operation. Hydrogen separation by membrane is an attractive alternative compared to traditional technologies such as pressure swing adsorption and cryogenic distillation. This research focuses on the use of different types of membrane processes for H2 recovery from refinery purge/flare gases. The literature also discusses the advantages and disadvantages of different hydrogen separation membranes in compare with traditional H2 separation methods and also reports the performance of selected membranes in terms of hydrogen selectivity and permeability.

V

Contents Introduction ................................................................................................................................................... 1 Chapter 1 ....................................................................................................................................................... 1 1.1Overview of Hydrogen Production and Uses .......................................................................................... 1 1.2 Usage of off-gases in hydrogen production plant ............................................................................... 2 1.3 H2 Separation ...................................................................................................................................... 2 1.3.1 Cryogenic separation.................................................................................................................... 3 1.3.2 Absorption Methods..................................................................................................................... 4 1.3.3 Pressure Swiping Adsorption ....................................................................................................... 5 1.3.4 Membrane technology.................................................................................................................. 6 Literature review ........................................................................................................................................... 7 Chapter 2 ....................................................................................................................................................... 7 2. Gas Separation Membrane ........................................................................................................................ 7 2.1 Historical Development of Membranes .............................................................................................. 7 2.2 membrane nomenclature and classification ...................................................................................... 10 2.3 Gas separation mechanism ................................................................................................................ 11 2.3.1 Transport of gas through non-porous membrane ....................................................................... 12 2.3.2 Transport of gas through porous membranes ............................................................................. 13 2.4 Membrane Construction Techniques ................................................................................................ 14 Chapter 3 ..................................................................................................................................................... 19 3. Hydrogen Selective Membranes ............................................................................................................. 19 3.1. Introduction ...................................................................................................................................... 19 3.2 Polymeric membranes ....................................................................................................................... 20 3.3 Metallic membrane ........................................................................................................................... 22 3.3 Carbon membrane ............................................................................................................................. 26 3.4 Ceramic and Zeolite Membranes ...................................................................................................... 27 3.5 Glass membranes .............................................................................................................................. 28 3.6 Mixed-Matrix membrane .................................................................................................................. 29 3.7 Motivation and guidelines for development of advanced or novel functional membranes .............. 31

VI

Chapter 4 ..................................................................................................................................................... 33 4. MEMBRANE SYSTEM DESIGN ......................................................................................................... 33 4.1 Introduction ....................................................................................................................................... 33 4.2 Current membrane module designs................................................................................................... 34 4.2.1

The cartridge membrane module ........................................................................................ 35

4.2.2

The plate-and-frame membrane module ............................................................................. 36

4.2.3

The plate-and-frame membrane module ............................................................................. 38

4.2.4

The tubular membrane module ........................................................................................... 40

4.2.5 The capillary membrane module................................................................................................ 42 4.2.6 The hollow fiber membrane module .......................................................................................... 43 4.2.7 Other membrane modules .......................................................................................................... 44 4.3 Calculating membrane permeate fluxes for gas separation............................................................... 46 4.3.1 Pressure Ratio ............................................................................................................................ 47 4.3.2 Stage-cut .................................................................................................................................... 49 4.4 Single stage membrane processes ..................................................................................................... 51 4.5 Multiple and Multistage membrane process ..................................................................................... 56 4.6 Recycle Designs ................................................................................................................................ 58 4.7 Application........................................................................................................................................ 59 4.7.1 Hydrogen Separation.................................................................................................................. 59 4.8 Membrane reactors............................................................................................................................ 63 4.9 Some practical issues ........................................................................................................................ 64 Result and Discussion ................................................................................................................................. 66 Chapter 5 ..................................................................................................................................................... 66 5.1 Comparison on different hydrogen separation methods ................................................................... 66 5.2 Comparison/overview of the different membranes ........................................................................... 72 Conclusion .................................................................................................................................................. 79 Reference .................................................................................................................................................... 80

VII

List of Tables Table 4. 1 Commercially available membrane modules, there costs and major applications [12] ............. 45 Table 4. 2 Hydrogen separation membrane [13]......................................................................................... 60

Table 5. 1 Comparison of hydrogen purification techniques [64] .............................................................. 68 Table 5. 2 comparison of H2 capture cost [65]............................................................................................ 70 Table 5. 3 Properties of the relevant hydrogen selective membranes [27] ................................................. 73 Table 5. 4 Selected Hydrogen Separation Metallic Membranes and Their Performance [69] ................... 76 Table 5. 5 Performance of H2 Selective Ceramic Membranes [69] ............................................................ 77

VIII

List of Figures Figure 2. 1 Milestones in the development of gas separation [13] ............................................................... 9 Figure 2. 2 Membrane unit nomenclatures ................................................................................................. 10 Figure 2. 3 Mechanism of permeation of hydrogen through metal membranes[13] .................................. 12 Figure 2. 4 Transport mechanisms in porous membranes: (a) Knudsen diffusion, (b) surface diffusion, (c) capillary condensation, (d) molecular sieving [27] .................................................................................... 13

Figure 3. 1 Mechanism of permeation of hydrogen through metal membranes[13] .................................. 23 Figure 3. 2 Hydrogen permeability as a function of temperature for the selected metals.[69] .................. 26 Figure 3. 3 Gas permeation through mixed-matrix membranes containing different amounts of dispersed zeolite particles [13] ................................................................................................................................... 29

Figure 4. 1 Schematic drawing showing a cartridge filter unit [20] .......................................................... 35 Figure 4. 2 Schematic drawing illustrating the concept of a plate-and-frame membrane module[20] ...... 36 Figure 4. 3 Circular plate-and-frame filter device with one baffle to extend the feed flow path length [13] .................................................................................................................................................................... 37 Figure 4. 4 Schematic drawing of a spiral-wound membrane module [13] ............................................... 38 Figure 4. 5 Schematic drawing illustrating the construction of a multi-leaf spiral wound module [20] .... 39 Figure 4. 6 Schematic drawing illustrating the tubular membrane module [26] ....................................... 41 Figure 4. 7 Tubular module with seven individual tubes bundled in a shell tube [26] ............................... 41 Figure 4. 8 Schematic diagram showing a capillary membrane module [20] ............................................ 42 Figure 4. 9 SEM of a capillary membrane with the selective “skin” on the inside of capillary [20] ......... 42 Figure 4. 10 Schematic drawing illustrating the construction of a hollow fiber module [26] .................... 43 Figure 4. 11 Parameters affecting the performance of membrane gas separation systems[13] ................. 46 Figure 4. 12 Calculated permeate vapor concentration for a vapor-permeable membrane with a vapor/nitrogen selectivity of 30 as a function of pressure ratio.[13] ......................................................... 49 Figure 4. 13 The effect of stage-cut on the separation of a 50/50 feed gas mixture (pressure ratio, 20; membrane selectivity, 20).[13] ................................................................................................................... 50 Figure 4. 14 Simple membrane set-up......................................................................................................... 51 Figure 4. 15 Comparison co-current, counter-current, perfect permeate mixing and perfect mixing set up [27] ............................................................................................................................................................. 53 Figure 4. 16 Single stage membrane processes with feed flow compression .............................................. 54 Figure 4. 17 Single stage membrane process with permeate vacuum......................................................... 54 Figure 4. 18 Single stage process with permeate dilution by means of sweep flow .................................... 55 Figure 4. 19 Single stage membrane processes with recycle ...................................................................... 55 Figure 4. 20 Two-stage membrane process as simple split up of single stage process............................... 56 Figure 4. 21 Two-stage membrane process with permeate recycle ............................................................ 56 Figure 4. 22 two-stage system to produce a highly concentrated permeate stream [13] ........................... 57 IX

Figure 4. 23 Two-and-one-half-stage systems: by forming a recycle loop around the second stage, a small, very concentrated product stream is created [13] ........................................................................... 58 Figure 4. 24 Simplified flow schematic of the PRISM® membrane system to recover hydrogen from an ammonia reactor purge stream [13] ........................................................................................................... 60 Figure 4. 25 Hydrogen recovery from a hydrotreater used to lower the molecular weight of a refinery oil stream. Permea polysulfone membranes (PRISM®) are used[13] .............................................................. 62

Figure 5. 1 Cash position versus time for different process [65] ................................................................ 70

X

Introduction Chapter 1 1.1Overview of Hydrogen Production and Uses Worldwide, industrial hydrogen is currently produced at over 41 MM tons/yr with 80% of production coming from the steam reforming of natural gas. Globally, hydrogen produced “on-purpose”, i.e., not as part of a petrochemical processing, is about 16 trillion scf/year; and refinery by-product hydrogen is about 14 trillion scf/yr, or about half the global total. The growing demand for hydrogen in chemical manufacturing, petroleum refining, and the new emerging clean energy concepts will place greater demands on supply and will most certainly impact pricing [1]. Hydrogen is used commercially in petroleum and chemical processing for hydrodesulfurization, and the production of syngas, ammonia, methanol, higher alcohols, urea and hydrochloric acid [2-5]. It is also used in Fischer Tropsch reactions, as a reducing agent (metallurgy), and to upgrade petroleum products and oils (hydrogenation, hydrocracking) [2-5]. Due to increased demand, H2 is increasingly being produced from natural gas by steam reforming, partial oxidation and autothermal reforming. In some refining processes, hydrogen is both used as a raw material and obtained in some processes as a by-product. Generally, in refineries or petrochemical complexes, the off-gas streams contain considerable amount of hydrogen, which are mostly incinerated in refinery flares as a waste gas. At the same time valuable hydrocarbons are converted into hydrogen in reformers in order to produce the hydrogen needed for hydrogen consuming processes. The hydrogen demands in refineries are constantly growing due to more sever environmental regulations. This leads to higher capital cost needed to build new hydrogen plants or increase the capacity of the existing ones. Therefore, the hydrogen recovery from the off-gases could be considered as a promising approach to make effective use of the existing facilities. The novel approach may considerably decrease the hydrogen production cost in the refineries. The recovered hydrogen can either be used as a raw material in a refinery or can be sold as a new product to downstream industries. 1

The reforming of natural gas to produce H 2 consumes about 31,800 Btu/lb of H 2 produced at 331 psig based on 35.5 MM tons/yr production [6]. It is estimated that 450 trillion Btu/yr could be saved with a 20% improvement in just the H 2 separation and purification train after the hydrogen reformer [6]. Clearly, improved separation technology can offer substantial dividends. The four most widely used separations methods, i.e., absorption, adsorption, membrane and partial separation, are summarized below.

1.2 Usage of off-gases in hydrogen production plant Steam reforming is the most ordinary process to produce hydrogen from natural gas. This process is based on the reforming of natural gas in a catalytic reformer at high temperature and pressure. The off-gas streams could be consumed inside the hydrogen production plant as a raw material and two options may be taken into account accordingly: 

Hydrogen production plant’s retrofit, in this case off- gas stream is used as a raw material inside the hydrogen production plant with considering constant plant capacity.



Hydrogen production plant’s revamp.

In order to have an off-gas stream with a high flow rate, revamping of hydrogen production plant is not recommended because; the simulation results show that heat exchangers are not adequate. Also, when the capacity is increased, the reformer duty and the consumption rate will be accordingly increased leading to an enlargement in reformer size. However, usage of off-gas stream with considering constant plant capacity could be considered. Due to check usage of the off-gas in existing hydrogen production plant, existing plant could be simulated using the sequential modular approach.

1.3 H2 Separation The separation technology used in H2 production depends on the application, the desired H2 purity and the downstream impact of CO or N2. Four different H2 purification technologies are widely practiced in industry; these include 1) cryogenic separation, 2) absorption, both chemical and physical, 3) pressure swiping adsorption and 4) membranes. These studies have the potential for both near term and longer term impact on the adsorptive applications for H2 production. 2

Adsorbent/membrane hybrid technologies are also being explored for sequential operation. A brief summary of each of these emerging areas is provided below.

1.3.1 Cryogenic separation Cryogenic technology is based on the difference in the relative component volatility at low temperatures. Since hydrogen has a higher volatility than the other components presented in the off-gas stream, it remains as a gas, while the other components become condensed easily when the temperature is lowered extensively. This would result in condensating the impurities of the off-gas, which in turn, leads to separate the gaseous hydrogen from these impurities. The disadvantages of this method, which result in higher capital cost of such purification unit, are as followings [7]: 

Off-gases are usually available at low pressure and have to be compressed up to at least 20 bars in order to be able to return to the hydrogen main header. Therefore, additional equipment, i.e., two centrifugal compressors and two after coolers have to be employed.



Off-gases usually contain about 8% C0 2 (mole) as an impurity that may be freeze during low temperature process. Consequently, a CO 2 separation plant, as a pretreatment stage, has to be considered before passing the gas into the cryogenic process. Therefore, the auxiliary unit should be included using this technique for CO2 separation, however in this way, the compressed CO 2-free off-gas enters the cryogenic section.

There are three options to be considered in the cryogenic process: 

Using a Joule-Thomson valve



Using a turbo-expander



Using plate-fin heat exchangers

In order to have a low pressure off- gas stream inherently, Using a Joule-Thomson valve or a turbo-expander process is not recommended because hydrogen content in the final product stream would be low (less than 90% mole). However, plate-fin heat exchangers would produce the desired hydrogen purity up to 97% (mole). Such high purity could be 3

achieved using two plate-fin heat exchangers. In addition to the use of cold process streams, which are produced in the separators and then used as the refrigerant in the heat exchangers, employing an auxiliary refrigerant such as liquefied nitrogen is necessary to provide enough cooling. This process is simulated using the sequential modular approach. 1.3.2 Absorption Methods Absorption methods are also gas–liquid separation ones, which use absorbents to remove soluble components from crude hydrogen gases. The equipment consists of an absorption column in which soluble components are trapped in an absorbent at a higher pressure or a lower temperature, followed by a regeneration column in which the absorbed components are released at a lower pressure or a higher temperature. The absorbent is circulated between the two columns. Such operations as washing out of soluble impurities into a liquid absorbent are often called "washing". Water washing is the oldest absorption process and was used mainly to remove carbon dioxide contained in crude hydrogen gases. The absorption processes are roughly classified into physical ones that utilize the differences in solubility between hydrogen and other components, and chemical ones based on chemical reactions between impurity components and the absorbent [9]. Physical absorption processes, in which solubility of gaseous components generally obeys Henry’s law, are mostly applied to separation of hydrogen from feed gases at considerably high pressures and high impurity concentrations for removal of residual impurities, and have the advantages that the product hydrogen leaving the absorption column is near the feed pressure and the regeneration of absorbents is relatively easy. The representative applications are removal of carbon dioxide by the use of organic solvents or solutions, removal of methane, nitrogen and argon by scrubbing pre-purified hydrogen gas at a cryogenic temperature, and removal of heavy hydrocarbons and separation of light hydrocarbons from gases rich in hydrogen by the use of heavy hydrocarbons such as gasoline, oil, toluene, or methyl cyclohexane as solvent [9]. Chemical absorption is suitable for complete removal of specific impurities from crude hydrogen and recovery of hydrogen from that at its relatively low pressure or low concentration in comparison with physical absorption. This chemical absorption method is used mainly for separation of acidic substances such as carbon dioxide, hydrogen sulfide, carbonyl sulfides, and 4

hydrogen cyanide from raw hydrogen gases produced from hydrocarbons and coals. Hot alkaline or amine solutions are used as the absorbing agents to remove the acidic impurities [9]. 1.3.3 Pressure Swiping Adsorption Adsorption separation methods utilize the preferential adsorption of some constituent species other than hydrogen mainly due to the difference in adsorption equilibrium. Representative adsorbents are silica gel, activated carbon, activated aluminum and molecular sieves. Most of the conventional separation units are operated in a dual mode of adsorption by passing the feed hydrogen gas and regeneration by means of heating, pressure reduction, gas purge, or a combination of these methods. They are widely applied to removal of impurities such as water vapor, dusts, tars, heavy hydrocarbons, mists and acidic components, and upgrading of moderately high purity hydrogen. The adsorption processes at cryogenic temperatures are indispensable for the final removal of residual impurities in the production of the ultra-highpurity hydrogen with an overall impurity level of less than 0.1 parts per million (ppm), which is required in hydrogen liquefaction and semiconductor industries [10]. Pressure swing adsorption is known to be one of the most economic and widespread commercial processes for hydrogen purification, air separation and small scale air driers. PSA was introduced in 1960s and today it has numerous other actual and potential uses such as the recovery of methane from landfill gas, the production of carbon dioxide and other large scale applications. Pressure Swing Adsorption processes rely on the fact that under pressure gases tend to be attracted to solid surfaces, or adsorbed(Typical adsorbents are activated carbon, silica gel, alumina and zeolite). The higher the pressure, the more gas is adsorbed, when the pressure is reduced, the gas is released, or desorbed. PSA processes can be used to separate gases in a mixture because different gases tend to be attracted to different solid surfaces more or less strongly. A typical PSA system involves a cyclic process where a number of connected vessels containing adsorbent material undergo successive pressurization and depressurization steps in order to produce a continuous stream of purified product gas [11].

5

1.3.4 Membrane technology Hydrogen enrichment using membrane technology is one of the economical methods to recover hydrogen from off-gases due to low costs (i.e., operating and capital) and ease of operation. In the last 50 years there have been strong developments in membrane technology for gas separation. Currently the number of hydrogen selective membrane applications is still limited to two, and both applications are not in critical parts of the process. These two are hydrogen recovery from off-gases in the ammonia industry and production of pure hydrogen in the electronics industry. Research is going on, intended to apply hydrogen selective membranes in more critical parts of processes. Different types of hydrogen selective membranes are presented for hydrogen separation from refinery off-gases and separation of hydrogen from syngas. Each membrane has its own operating ranges, in terms of temperatures and flow compositions. The properties of the flow to be separated are therefore a starting point to select a suitable membrane. This literature summarizes the state of the art in hydrogen gas separation membrane technology. Chapter 2 gives a general introduction to membranes, where nomenclature, definitions, transport mechanisms and manufacturing technology are presented. The most important membranes for hydrogen separation from refinery off-gases and separation of hydrogen from syngas, a mixture of hydrogen and carbon monoxide, are presented in chapter 3. Chapter 4 provides an introduction to the design of membrane systems. In chapter 5 different hydrogen separation methods are compared and in addition some types of membranes are discussed. Finally in chapter 6 some conclusions are drawn.

6

Literature review Chapter 2 2. Gas Separation Membrane A membrane is a barrier that permits selective mass transport between two phases. It is selective because some components can pass the membrane more easily than others. This makes membranes a suitable means to separate a mix of components. The phases on either side of the membrane can be liquid or gaseous. Although we may not be aware of their presence, membranes play an important role in life. Probably the best-known example of a membrane is the human skin. The skin admits selective transport of both gases and liquids (e.g., water can't flow in but it can flow out when sweating) [12].

2.1 Historical Development of Membranes Systematic studies of membrane phenomena can be traced to the eighteenth century philosopher scientists. For example, Abbé Nolet coined the word ‘osmosis’ to describe permeation of water through a diaphragm in 1748. Through the nineteenth and early twentieth centuries, membranes had no industrial or commercial uses, but were used as laboratory tools to develop physical/chemical theories. For example, the measurements of solution osmotic pressure made with membranes by Traube and Pfeffer were used by van’t Hoff in 1887 to develop his limit law, which explains the behavior of ideal dilute solutions; this work led directly to the van’t Hoff equation. At about the same time, the concept of a perfectly selective semipermeable membrane was used by Maxwell and others in developing the kinetic theory of gases [13]. Early membrane investigators experimented with every type of diaphragm available to them, such as bladders of pigs, cattle or fish and sausage casings made of animal gut. Later, collodion (nitrocellulose) membranes were preferred, because they could be made reproducibly. In 1907, Bechhold devised a technique to prepare nitrocellulose membranes of graded pore size, which he determined by a bubble test [14]. Other early workers, particularly Elford [15], Zsigmondy and Bachmann [16] and Ferry [17] improved on Bechhold’s technique, and by the early 1930s microporous collodion membranes were commercially available. Development of gas separation membranes started later. The first large-scale gas separation membrane process was used in the mid-1940s by the United States government to separate UF6 7

isotopes for nuclear fuel enrichment [18]. The first commercially significant gas separation membranes were introduced only in late 1979. However, within 10 years a range of gas separation membranes has been developed [18]. The development of high-flux membranes and large-surface-area membrane modules for reverse osmosis applications in the late 1960s and early 1970s provided the basis for modern membrane gas separation technology. The first company to establish a commercial presence was Monsanto, which launched its hydrogen-separating Prism® membrane in 1980 [19]. By the mid-1980s, Cynara, Separex and Grace Membrane Systems were producing membrane plants to remove carbon dioxide from methane in natural gas. This application, although hindered by low natural gas prices in the 1990s, has grown significantly over the years. At about the same time, Dow launched Generon®, the first commercial membrane system for nitrogen separation from air. Initially, membrane-produced nitrogen was cost-competitive in only a few niche areas, but the development by Dow, Ube and Du Pont/Air Liquide of materials with improved selectivities has since made membrane separation much more competitive. This application of membranes has expanded very rapidly and is expected to capture more than onehalf of the market for nitrogen separation systems within the next few years. To date, approximately 10000 nitrogen systems have been installed worldwide. Gas separation membranes are also being used for a wide variety of other, smaller applications ranging from dehydration of air and natural gas to organic vapor removal from air and nitrogen streams. Application of the technology is expanding rapidly and further growth is likely to continue for the next 10 years or so. Figure 2.1 provides a summary of the development of gas separation technology [13].

8

Figure 2. 1 Milestones in the development of gas separation [13]

In the last 50 years there have been strong developments in membrane technology, making application of manufactured membranes a viable option. Today's membrane applications are very diverse, ranging from reverse osmosis (to produce clean water) to micro filtration (to filter bacteria). Membranes owe their popularity largely to the following advantages [20]: 

(generally) low energy consumption;



possibility to carry out separation continuously;



mild process conditions;



easy scaling up;



absence of additives;



Possibility to combine with other separation technologies.

Important disadvantages are, depending on the specific membrane type: 

fouling tendency;



low membrane lifetime;



low selectivity or flux;

9



More or less linear up-scaling factor (whereas competing processes exhibit economies of scale).

In this literature only gas separation membranes are studied, i.e. membranes with gaseous phases on both sides. This is because membrane applications in refineries are limited to hydrogen selective membranes. Even though conditions in refineries may be somewhat extreme (in terms of pressures and temperatures), hydrogen will not be present in liquid form and it will not be (in large quantities) dissolved in liquids from which it should be separated.

2.2 membrane nomenclature and classification Figure 2.2 illustrates the nomenclature used for membranes. The two sides of the membrane are called feed side (or upstream side) and permeate side (or downstream side). In practice, permeation can take place in both directions. Generally speaking, feed side and permeate side are chosen consistent with the rule that the permeation of the (most) relevant species takes place from feed to permeate side. The feed side flow is initially called the feed flow. The flow resulting after permeation is called retentate (or residue) flow. On the permeate side the inlet flow is called sweep flow and the exit flow permeate flow.

Figure 2. 2 Membrane unit nomenclatures

Performance and efficiency of membranes are usually measured in terms of flow (or flux) through the membrane and membrane selectivity towards mixtures. The flow can be measured in volume or mass per time unit (if measured per unit of surface it is called flux). The selectivity is a measure for the difference in permeabilities (the relative ease with which species can permeate) of different components. In other words, it is a measure for the membrane separation effectiveness. The selectivity factor

of two components A and B in a mixture is defined as: eq. 2.1

Where

and

are the fractions of components A and B in the permeate and

and

are the

fractions of the components A and B in the feed. A and B are usually chosen in such a way that the selectivity factor is greater than unity. If the selectivity factor is equal to one, there is no 10

separation. The higher the selectivity factor, the more selective the membrane is to certain species (which is usually a desirable membrane property).

Two other important ratios to describe membrane performance are the recovery and volume reduction. The recovery or yield (S) is the fraction of the feed flow passing through the membrane: eq. 2.2 Where

is the permeated flow and

is the feed flow. The volume reduction (VR) is the ratio

of the initial feed flow rate and the retentate flow rate:

eq. 2.3 To subdivide membranes, a number of membrane properties can be used. One is the material the membrane is made of. So-called organic membranes are made of polymers. The group of inorganic membranes comprises membranes made of glass, metal (including carbon), and ceramics (including zeolites). Another property of membranes used for subdivision, is the membrane structure and connected to this the way in which transport through the membrane takes place. Porous membranes enable transport through their pores, whereas dense membranes allow transport through the bulk of the material. Finally, the morphology or structural make-up of membranes can also be used for classification. Symmetrical membranes have a homogeneous structure. Asymmetric membranes consist of several layers with different characteristics. There can also be a gradual transition from a dense membrane to a porous support. Membranes consisting of different layers of different materials are called composite membranes.

2.3 Gas separation mechanism As noted before, there are two main membrane permeation mechanisms: through the bulk of the material (dense membranes) and through pores (porous membranes). In the following sections, a number of membrane transport mechanisms are described. The mechanisms described are not limitative. Dense membranes usually have high selectivities, yet low fluxes. This principle also 11

applies to small-pore membranes. Larger pores will increase the fluxes, but decrease selectivities.

2.3.1 Transport of gas through non-porous membrane The solution/diffusion mechanism is the most commonly used physical model to describe gas transport through dense membranes. A gas molecule is adsorbed on one side of the membrane, dissolves in the membrane material, diffuses through the membrane and desorbs on the other side of the membrane. If diffusion through the membrane takes place in the form of ions and electrons (= proton exchange transport) or as atoms (e.g. for hydrogen transport through dense metal), the molecule needs to split up after adsorption and recombine after diffusing through the membrane. Hydrogen permeation through a metal membrane is believed to follow the multistep process illustrated in Figure 2.3 [21]. Hydrogen molecules from the feed gas are sorbed on the membrane surface, where they dissociate into hydrogen atoms. Each individual hydrogen atom loses its electron to the metal lattice and diffuses through the lattice as an ion. Hydrogen atoms emerging at the permeate side of the membrane reassociate to form hydrogen molecules, then desorb, completing the permeation process. Only hydrogen is transported through the membrane by this mechanism; all other gases are excluded.

Figure 2. 3 Mechanism of permeation of hydrogen through metal membranes[13]

12

2.3.2 Transport of gas through porous membranes

Four types of diffusion mechanisms can be utilized to effect separation in porous membranes (see Figure 3.4) [18]. In some cases, molecules can move through the membrane by more than one mechanism. These mechanisms are described below. Knudsen diffusion gives relatively low separation selectivities compared to surface diffusion and capillary condensation. Shape selective separation or molecular sieving can yield high selectivities. The separation factor for these mechanisms depends strongly on pore size distribution, temperature, pressure and interactions between gases being separated and the membrane surfaces.

Figure 2. 4 Transport mechanisms in porous membranes: (a) Knudsen diffusion, (b) surface diffusion, (c) capillary condensation, (d) molecular sieving [27]

Knudsen (or free-molecule) diffusion (see Figure 3.4a) is said to appear for large Knudsen numbers. The Knudsen number

is defined as the ratio of the mean free path of the gas

molecules (average distance between collisions) and a representative physical length scale (e.g., the pore radius). eq. 2.4 13

If the pore radius is used as representative physical length scale the mean free path lengths are substantially higher than pore radiuses at Knudsen numbers larger than 10. The result is that mainly the lighter molecules permeate through the pores. Selectivity, however, is limited and can be calculated with the square root of the ratio of the molar masses of the gasses involved [23]. The smaller the Knudsen number, the larger the pores become (relative to the mean free path of the gas molecules), the lower selectivity becomes. For Knudsen numbers < 1 the dominant transport mechanism is viscous flow, which is non-selective. Surface diffusion (see Figure 3.4b) can occur in parallel with Knudsen diffusion. Gas molecules are adsorbed on the pore walls of the membrane and migrate along the surface. Surface diffusion increases the permeability of the components adsorbing more strongly to the membrane pores. At the same time, the effective pore diameter is reduced. Consequently, transport of non-adsorbing components is reduced and selectivity is increased. This positive contribution of surface diffusion only works for certain temperature ranges and pore diameters. Capillary condensation (see Figure 3.4c) occurs if a condensed phase (partially) fills the membrane pores. If the pores are completely filled with condensed phase, only the species soluble in the condensed phase can permeate through the membrane. Fluxes and selectivities are generally high for capillary condensation. The appearance of capillary condensation, however, strongly depends on gas composition, pore size, and uniformity of pore sizes. If pore sizes become sufficiently small (3.0-5.2 A), molecular sieving (see Figure 3.4d) can be used to separate molecules that differ in kinetic diameter: the pore size becomes so small, that only the smaller gas molecules can permeate through the membrane.

2.4 Membrane Construction Techniques All kinds of materials such as ceramic, glass, metal or polymers can be used for preparing membranes. The aim is to modify the material by means of an appropriate technique to obtain a membrane structure with morphology suitable for a specific separation. The material limits the preparation techniques employed, the membrane morphology obtained and the separation principle applied. In this section the most important membrane manufacturing techniques for gas separation membranes are presented. A more extensive review of manufacturing techniques is presented in [13, 20].

14

Several factors contribute to the successful fabrication of a high-performance membrane module. First, membrane materials with the appropriate chemical, mechanical and permeation properties must be selected; this choice is very process-specific. However, once the membrane material has been selected, the technology required to fabricate this material into a robust, thin, defect-free membrane and then to package the membrane into an efficient, economical, high surface- area module is similar for all membrane processes. Therefore, this chapter focuses on methods of forming membranes and membrane modules [13]. Membrane preparation techniques are organized by membrane structure: isotropic membranes, anisotropic membranes, ceramic and metal membranes, and liquid membranes. Isotropic membranes have a uniform composition and structure throughout; such membranes can be porous or dense. Anisotropic (or asymmetric) membranes, on the other hand, consist of a number of layers each with different structures and permeabilities. A typical anisotropic membrane has a relatively dense, thin surface layer supported on an open, much thicker microporous substrate. The surface layer performs the separation and is the principal barrier to flow through the membrane. The open support layer provides mechanical strength. Ceramic and metal membranes can be either isotropic or anisotropic. However, these membranes are grouped separately from polymeric membranes because their preparation methods are so different. Sintering involves compression of powders and subsequent formation of solids from the particles by sintering at elevated temperatures. As the particles are not cubically shaped, pores will result. It is a very simple technique to prepare porous organic as well as inorganic materials with pore sizes ranging from 0.1 to 10 μm, the lower limit being determined by the minimum particle size. It is primarily suitable for materials with very high chemical, thermal and mechanical stability. A drawback is, however, that only certain types of membranes can be prepared by means of sintering. Besides, the porosity (pore area as fraction of total area) of porous polymeric membranes is generally low, in the order of 10 to 20%. Higher porosities (up to 90%) can be achieved using stretching. This technique involves extension of an extruded film or foil made from a partially crystalline polymeric material. The extension is performed perpendicular to the direction of the extrusion. The result is that crystalline regions are located parallel to the extrusion direction. Small ruptures occur if mechanical stresses are applied, resulting in a porous structure. Pore sizes are in the range

15

between 0.1 and 3 μm. This technique can only be used to manufacture membranes from (semi) crystalline polymeric materials. Track-etching comprises tracking (the subjection of a film or foil to perpendicular high energy particle radiation) and subsequently etching away polymeric material along the tracks obtained by immersing the film in an acid or alkaline bath. The results are practically cylindrical pores with a narrow pore size distribution. Pore size can be obtained in the range from 0.02 to 10 μm. Surface porosity, however, is low (maximum 10%). These membranes are almost a perfect screen filter; therefore, they are widely used to measure the number and type of suspended particles in air or water. This method is mainly used for organic materials, although it can be used for a limited number of inorganic materials too. Template leaching is another method of producing isotropic microporous membranes from insoluble polymers such as polyethylene, polypropylene and poly(tetrafluoroethylene). In this process a homogeneous melt is prepared from a mixture of the polymeric membrane matrix material and a leachable component. To finely disperse the leachable component in the polymer matrix, the mixture is often homogenized, extruded, and pelletized several times before final extrusion as a thin film. After formation of the film, the leachable component is removed with a suitable solvent, and a microporous membrane is formed. The leachable component can be a soluble, low-molecular-weight solid, a liquid such as liquid paraffin, or even a polymeric material such as polystyrene. Wide range of pore diameters can be obtained, with a minimum of 5 nm (1 nm = 10-9 m) by this method. Templates are also used to prepare zeolite membranes. A support is immersed in a water-based sol containing silica, sodium hydroxide and the organic template. During heating, the zeolite crystals grow. Finally, the template is burned out. Phase inversion is the most frequently used method for the current range of commercially available polymer membranes. In this method a polymer in a liquid state is transformed into a solid state, creating a membrane. Usually, the liquid state is first transformed into two separated liquids. During this so-called demixing stage, one of the liquids solidifies into a solid matrix. The term phase separation describes the process most clearly, namely, changing a one-phase casting solution into two separate phases. In all phase separation processes, a liquid polymer solution is precipitated into two phases: a solid, polymer-rich phase that forms the matrix of the membrane and a liquid, polymer-poor phase that forms the membrane pore. This method can be used to prepare both porous and non-porous membranes. The concept of phase inversion includes 16

techniques such as solvent evaporation, precipitation by controlled evaporation, thermal precipitation, precipitation from the vapor phase and immersion precipitation. The majority of the phase inversion membranes are prepared by immersion precipitation [24]. Coating methods are especially important for dense polymeric membranes and inorganic composite membranes. Because the fluxes are generally lower for dense membranes than for porous membranes, membrane thickness needs to be minimized. To attain structural strength for reduced membrane thicknesses, composite membranes are used (the membrane is supported by a porous sub layer). A number of coating procedures in use are dip coating, plasma polymerization, interfacial polymerization, and in-situ polymerization. The sol-gel process was the first process to enable pore sizes in the nanometer range for ceramic membranes. In this process, an alkoxide precursor is formed into a gel via either the colloidal suspension route, or the polymeric gel route. In the colloidal suspension route, the precursor is hydrolysed to form a sol (a colloidal dispersion of particles in a liquid). By changing the surface charge of the particles in the sol or by increasing the concentration the particles tend to agglomerate and a three-dimensional network structure, a so-called gel is obtained. In the polymeric gel route a precursor is hydrolysed slowly to form an inorganic polymer and finally a polymer network. The most difficult step is drying the gel, which can result in crack formation. The final step in the process is sintering. Even smaller pore sizes can be obtained by densification of the mesoporous structure. Some surface modification techniques for ceramic membranes are internal deposition of pores by monolayer or multi-layer, pore plugging of nanoparticles, coating an inorganic polymeric gel layer on top of a support and constrictions at sites in the top layer by Chemical vapor deposition (CVD). With CVD a metal organic component is vaporized in the carrier gas, which enters from one side of the membrane and the other reactant enters from the other side of the membrane. Reactants diffuse into the pores and react there. The product is deposited on the pores. Depending on the structure of the material and the process parameters the deposited layer will be dense or (micro) porous. Metals can be also deposite on a substrate layer by means of physical vapour deposition (PVD). In this process, the solid material to be deposited is first evaporated in a vacuum system using physical techniques. The thin to medium thickness film is subsequently condensed and deposited on the cooler substrate [25]. 17

Somewhat thicker metallic membranes can also be prepared by means of alloy casting and rolling. This is the standard practice for large scale manufacturing of metallic plates and sheets and together with electroless plating (see below) it is the method most commonly used for preparation of metallic membranes. The process involves melting raw materials at very high temperature, ingot casting, high-temperature homogenization, hot and cold forging or pressing, followed by repeated sequences of alternate cold rolling and anneals, down to the required thickness. If the cooling speed of the melts is chosen fast enough, this method can also be used to obtain amorphous metals called metallic glasses. This method may be suitable for thin membranes if the metal purity is high enough, because contamination becomes a problem as metal foils become thinner [25].

18

Chapter 3 3. Hydrogen Selective Membranes 3.1. Introduction For separation of hydrogen from gaseous streams, membranes can provide an attractive alternative to PSA and cryogenic separation, depending on the scale and purity of the product streams required. There are currently two applications in which hydrogen selective membranes are used. Since 1979 polymeric membranes are used in the ammonia industry for hydrogen recovery from off-gases and in the second half of the 1990s a new application appeared: the production of pure hydrogen in the electronics industry using palladium membranes. Pilot projects are currently under way to demonstrate the technical feasibility of hydrogen selective membranes in other applications, more at the heart of processes. Both refineries and IGCCs may be good candidates for hydrogen selective membrane applications, as they possess gaseous streams relatively rich in hydrogen. Hydrogen selective membranes can be broadly separated into four categories: polymeric (organic), metallic, carbon and ceramic (the latter three jointly called inorganic). For a long time, considerably more effort has been put in development of polymeric membranes than in inorganic membranes. Consequently polymeric membranes have wide ranging applications and can be bought at relatively low cost. However, interest in inorganic membranes has started to grow in the last decade. Inorganic materials can operate under higher temperatures than polymeric materials and generally possess superior chemical stability relative to polymeric materials. Ceramics form the main class of inorganic membranes [20]. The following sections provide an overview of the current status of the various hydrogen selective membranes. The information is mainly taken from [13, 20, and 26]. The main properties such as the transport mechanisms, the range of possible working conditions, development status, and strong and weak points of the various types of hydrogen permeating membranes are presented. To characterize the membranes the material of which they are made is used.

19

3.2 Polymeric membranes Gas separation from polymer membranes is already an existing technology. Polymeric membranes are a dense-type of membranes, transporting species through the bulk of the material. Depending on their state, polymeric membranes can be subdivided into glassy (prepared at temperatures below the glass transition temperature) and rubbery (prepared at temperatures above the glass transition temperature) polymeric membranes. Glassy membranes have relatively high selectivity and low flux, whereas rubbery membranes have increased flux but lower selectivity. In absolute terms both types have moderate fluxes and selectivity. They are usually produced using the phase inversion method. According to Kluiters [27] operating temperatures for polymer membranes are ~373 K. Good ability to cope with high pressure drops and low cost are key advantages of polymer membranes. However, limited mechanical strength, relatively high sensitivity to swelling and compaction, and susceptibility to certain chemicals such as hydrochloric acid (HCl), sulfur oxides (SOx), and CO2 make polymeric membranes less attractive. Polymer membranes used for separation processes operate according to the solution-diffusion mechanism [28]. Polymeric membranes are in an advanced stage of development. An overview of the state-of-theart polymeric materials, used for the manufacturing of commercial membranes, is given in [29]. A closer inspection reveals that most of the membranes currently on the market are based on relatively few polymers which had originally been developed for other engineering applications. Polymeric membranes for hydrogen separation are commercially available from gas producing companies like Air Products, Linde, BOC and Air Liquide. A classification will be made between the open porous membranes, which are applied in microfiltration and ultrafiltration and the dense nonporous membranes, applied in gas separation and pervaporation. The reason for this classification is the different requirements when the polymeric materials are used as membranes. For the porous microfiltration/ultrafiltration membranes the choice of the meterial is mainly determined by the process requirements (membrane manufacture), fouling tendency and chemical and thermal stability of the membrane. For the second class of polymers which are used for gas separation/pervaporation, the choice of the material directle determines the membrane performance (selectivity and flux). The choice of material is determined by the type of application and the polymer type can range from an

20

elastomer to a glassy material. For these processes either composite or asymmetric membranes are used [20]. In gas separation frequently glassy polymers are used with a high Tg showing high selectivities. An example is the class of polyoxadiazoles and polytriazoles. These polymers shoe an extremely high thermal stability, e.g. polyoxadiazoles has glass transition temperatures above the degradation temperature [20]. Most gas separation processes require that the selective membrane layer be extremely thin to achieve economical fluxes. Typical membrane thicknesses are less than 0.5 μm and often less than 0.1 μm. Early gas separation membranes [30] were adapted from the cellulose acetate membranes produced for reverse osmosis by the Loeb–Sourirajan phase separation process. These membranes are produced by precipitation in water; the water must be removed before the membranes can be used to separate gases. However, the capillary forces generated as the liquid evaporates cause collapse of the finely microporous substrate of the cellulose acetate membrane, destroying its usefulness. This problem has been overcome by a solvent exchange process in which the water is first exchanged for an alcohol, then for hexane. The surface tension forces generated as liquid hexane is evaporated are much reduced, and a dry membrane is produced [13]. Experience has shown that gas separation membranes are far more sensitive to minor defects, such as pinholes in the selective membrane layer, than membranes used in reverse osmosis or ultrafiltration. Even a single small membrane defect can dramatically decrease the selectivity of gas separation membranes, especially with relatively selective membranes such as those used to separate hydrogen from nitrogen. For example, a good polymeric hydrogen/nitrogen separating membrane has a selectivity of more than 100. A small defect that allows as little as 1% of the permeating gas to pass unseparated doubles the nitrogen flux and halves the membrane selectivity. The sensitivity of gas separation membranes to defects posed a serious problem to early developers. Generation of a few defects is very difficult to avoid during membrane preparation and module formation [13]. For common (inert) gases, the permeability coefficient, P is the product of the diffusion coefficient, D solubility constant, S with the common units noted:

21

P=DS cc (STP) cm/ cm2 s cm Hg 1 Barrer = cc (STP) cm/ cm2 s cm Hg × 10-10 The separation factor, αij , is defined as the ratio of Pi /Pj , where i , j=gases being separated. The diffusion of gases through polymers is an activated energy process, thus, the relationships for D, S, and P as a function of temperature can be expressed as: D=D0 exp(-Ed /RT )

eq. 3.1

S=S0 exp(-ΔHs /RT )

eq. 3.2

P=P0 exp(-Ep /RT )

eq. 3.3

Where E is the activation energy of diffusion, is ΔHs the heat of sorption and Ep = Ed + ΔHs is the activation energy of permeation. The permeability of gases through polymers is dependent upon the gas–polymer combination. For specific gases, the range of permeability through known organic polymers can range from 6 to 10 orders of magnitude. As an example, the O permeability ranges from 10000 Barrers for poly(trimethyl silyl propyne) (PTMSP) to 0.0001 Barrers for poly(vinyl alcohol). A survey of permeability/permselectivity values ([31, 32]) noted an upper bound relationship for polymeric materials whereby virtually no data resided above the ‘upper bound’. Some gas pairs of interest also have been shown to exhibit upper bound behavior (e.g. H2 /N2; H2 /CH4; He /N2; He/CH4; CO2 /CH4; H2 /O2). The empirical upper bound relationship has been shown to be predicted by theory as noted by Freeman [33].

3.3 Metallic membrane If very pure hydrogen is required, dense metallic membranes may be a good option. Especially palladium and palladium alloys (practically the only types of hydrogen selective metallic membranes used) are extremely selective as only hydrogen can permeate through them. The study of gas permeation through metals began with Graham’s observation of hydrogen permeation through palladium. Pure palladium absorbs 600 times its volume of hydrogen at room temperature and is measurably permeable to the gas. Hydrogen permeates a number of other metals including tantalum, niobium, vanadium, nickel, iron, copper, cobalt and platinum [34]. In most cases, the metal membrane must be operated at high temperatures (>300 ) to obtain useful permeation rates and to prevent embrittlement and cracking of the metal by sorbed hydrogen. Poisoning of the membrane surface by oxidation or sulfur deposition from trace amounts of hydrogen sulfide also occurs. A breakthrough in metal permeation studies occurred 22

in the 1960s when Hunter at Johnson Matthey discovered that palladium/silver alloy membranes showed no hydrogen embrittlement even when used to permeate hydrogen at room temperature [35]. Although most work on gas permeation through membranes has focused on hydrogen, oxygen-permeable metal membranes are also known; however, the permeabilities are low. Hydrogen-permeable metal membranes are extraordinarily selective, being extremely permeable to hydrogen but essentially impermeable to all other gases. The gas transport mechanism is the key to this high selectivity. Hydrogen permeation through a metal membrane is believed to follow the multistep process illustrated in Figure 3.1[36]. Hydrogen molecules from the feed gas are sorbed on the membrane surface, where they dissociate into hydrogen atoms. Each individual hydrogen atom loses its electron to the metal lattice and diffuses through the lattice as an ion. Hydrogen atoms emerging at the permeate side of the membrane reassociate to form hydrogen molecules, then desorb, completing the permeation process. Only hydrogen is transported through the membrane by this mechanism; all other gases are excluded.

Figure 3. 1 Mechanism of permeation of hydrogen through metal membranes[13]

If the sorption and dissociation of hydrogen molecules is a rapid process, then the hydrogen atoms on the membrane surface are in equilibrium with the gas phase. The concentration, c, of hydrogen atoms on the metal surface is given by Sievert’s law: eq. 3.4

23

Where K is Sievert’s constant and p is the hydrogen pressure in the gas phase. At high temperatures (>300 ◦C), the surface sorption and dissociation processes are fast, and the ratecontrolling step is diffusion of atomic hydrogen through the metal lattice. This is supported by the data of Holleck and others, who have observed that the hydrogen flux through the metal membrane is proportional to the difference of the square roots of the hydrogen pressures on either side of the membrane. At lower temperatures, however, the sorption and dissociation of hydrogen on the membrane surface become the rate-controlling steps, and the permeation characteristics of the membrane deviate from Sievert’s law predictions. The permeation of hydrogen through a metallic film is a complex process. The permeability can be considered as product of solubility and diffusivity. The permeation rate of hydrogen can be given by: [(

)

( (

where is the

) ] )

is the rate of more permeable species A (mol m−2 s−1), feed-side pressure and

eq. 3.5 eq. 3.6 is the membrane thickness (m),

permeate-side pressure (kPa),

is the membrane

permeability of more permeable component A (mol m m−2 s−1 kPa−n),

is the diffusivity of

hydrogen (m2 s−1), S is the hydrogen solubility in metal film (molm−3),

is the activation energy

for permeation (equal to the sum of the diffusion energy and the heat of dissolution) (kJ mol−1), is the mole fraction in feed side (a being more permeable) and

mole fraction in permeate

side. Palladium-alloy membranes were studied extensively during the 1950s and 1960s, and this work led to the installation by Union Carbide of a full-scale demonstration plant to separate hydrogen from a refinery off-gas stream containing methane, ethane, carbon monoxide and hydrogen sulfide [37]. The plant could produce 99.9% or better pure hydrogen in a single pass through the membrane. The plant operated with 25-μm-thick membranes, at a temperature of 370

and a

feed pressure of 450 psi. The high cost of the membranes and the need to operate at high temperatures to obtain useful fluxes made the process uncompetitive with other hydrogen recovery technologies. In the 1970s and early 1980s, Johnson Matthey built a number of systems to produce on-site hydrogen by separation of hydrogen/carbon dioxide mixtures made by reforming methanol [38]. This was not a commercial success, but the company still produces 24

small systems using palladium–silver alloy membranes to generate ultrapure hydrogen from 99.9% hydrogen for the electronics industry. If palladium membranes are exposed to hydrogen at lower temperatures, they can be seriously damaged, because hydrogen can become locked inside the palladium lattice. This will cause stresses in the membrane, increasing the likelihood of membrane failure. A solution to this problem is to dope the palladium with other elements such as silver or copper. Operating temperatures of the today's palladium alloy membranes are in the range 300-600 °C. A major technical disadvantage of palladium membranes in most applications is their high sensitivity to chemicals such as sulphur, chlorine and even CO. These chemicals can poison the membrane surface reducing the effective hydrogen fluxes by 20 to even 100%. Although much attention is focused on development of palladium membranes, their commercial availability is still limited. Johnson Matthey produces palladium-silver alloy membranes up to 60 cm in size commercially for the production of ultra-pure hydrogen in the electronics industry [26]. Metallic membranes for hydrogen separation could be of many types, such as (i) pure metals: Pd, V, Ta, Nb, and Ti; (ii) binary alloys of Pd: Pd-Cu, Pd-Ag, Pd-Y, and also Pd alloyed with Ni, Au, Ce, and Fe; (iii) complex alloys: Pd alloyed with 3-5 other metals; (iv) amorphous alloys: typically Group IV and Group V metals; and (v) coated metals: Pd over Ta, V, etc. [39]. Figure 3.2 presents hydrogen permeability for selected metals based on data presented elsewhere [40]. Body centered cubic (BCC) metals such as Nb and V have higher permeability than face centered cubic (FCC) metals such as Pd and Ni [40]. Hydrogen permeability decreases with increasing temperature in the case of Nb, V, and Ta. This phenomenon is due to the decrease of hydrogen solubility more rapidly than the increase of the diffusion coefficient. Although Nb, Ta, and V have higher permeability on the order of 10- 15× greater than that of Pd, these metals form oxide layers and are difficult to use as hydrogen separation membranes [40, 41]. However, this problem could be overcome by coating a thin layer of Pd on the surface of the aforementioned metals [40]. Moreover, Nb, Ta, and V metals are also cheaper than Pd and they could meet the cost target of the system.

25

Figure 3. 2 Hydrogen permeability as a function of temperature for the selected metals.[69]

3.3 Carbon membrane There are two types of carbon membranes, applying different transport mechanisms: molecular sieving and surface diffusion membranes. Molecular sieving membranes are identified as promising, both in terms of separation properties (including achievable fluxes) and stabilities, but are not yet commercially available at a sufficiently large scale. The pore sizes are in the order of the size of H2-molecules. Reported selectivities are in the range of 4-20. Adsorption selective carbon membranes separate non- (or weakly) absorbable from absorbable gases (such as H2S, NH3 and CFCs). The performance of these membranes will deteriorate severely if feed streams contain organic traces or other strongly adsorbing vapors. The use of molecular sieve carbon membranes (MSCM) for gas separations was first reported by Koresh and Soffer (1983), who based their research on molecular sieve carbon adsorbents, and developed MSCM by controlled pyrolysis of thermosetting polymer membranes [42]. They demonstrated that the permeation characteristics of the membrane could be controlled by mild stepwise thermochemical treatments. The permeability for both hydrogen and methane reached a maximum as the heat treatment temperature varied from 400 to 800 26

[43]. For light gases, such

as helium, hydrogen, argon, and oxygen, at relatively low pressures, the permeability is essentially independent of pressure, a consequence of free molecules in the pores or a liner adsorption isotherm. They further demonstrated that the separation factor for H2/CH4 appeared to be independent of the mixture composition and decreased from 57 at 200

to 35 at 500

[44].

Although these values of separation factors appear to be reasonable, the suitability of such membranes for high pressure applications requires careful study because of the possible loss of separation efficiency due to the possible change from free molecules in the pores to Knudsen or even molecular diffusion at high pressures. Surface-selective flow membranes made of nanoporous carbon, which is a variation of molecular sieving membranes, were developed by Rao et al. (1992) and Rao and Sircar (1993) [45]. The membrane can be produced by coating poly(vinylidene chloride) on the inside of a macroporous alumina tube followed by carbonization to form a thin membrane layer. The mechanism of separation is by adsorption–surface-diffusion–desorption. Certain gas components in the feed are selectively adsorbed, permeated through the membrane by surface diffusion, and desorbed at the low-pressure side of the membrane. This type of membrane was used to separate H2 from a mixture of H2 and CO2 [46] and their main advantage is that the product hydrogen is at the highpressure side eliminating the need for recompression. The membrane, however, is not industrially viable because of its low overall separation selectivity. In addition, since the separation mechanism involves physical adsorption, operation at low temperatures is required. 3.4 Ceramic and Zeolite Membranes During the last few years, ceramic- and zeolite-based membranes have begun to be used for a few commercial separations. These membranes are all multilayer composite structures formed by coating a thin selective ceramic or zeolite layer onto a microporous ceramic support. Ceramic membranes are prepared by the sol–gel technique described before; zeolite membranes are prepared by direct crystallization, in which the thin zeolite layer is crystallized at high pressure and temperature directly onto the microporous support [47]. Ceramic membranes are constructed by combination of a metal with a non-metal in the form of an oxide, nitride or carbide. Ceramic membranes can be both, porous or dense. Porous ceramic membranes generally have a two-layer structure: the separation membrane itself and a thicker, more porous ceramic supporting layer. The separation membrane is usually made of alumina, 27

zirconia, titania, or silica. Depending on the components to be separated the selectivity can reach values up to 140. Hydrogen fluxes through the membrane are promising. Operating temperatures for the porous ceramic membranes are within the range 200-600 °C. The development of porous membranes is at an early stage and most gas separation measurements have been made on samples of only a few square centimeters in surface area. Tubular samples of 20-90 cm have been fabricated but measurements under simulated operating conditions indicate that there are many issues to be resolved before this technology can be applied. One of these issues is limited stability in atmospheres containing steam. This stability may be increased by modifying silica membranes with methyl-groups [26]. In dense ceramic membranes, so-called proton conducting membranes (or proton exchange membranes), hydrogen is transported in the solid phase as ions (protons). Preferred materials are SrCeO3-δ and BaCeO3-δ. The selectivity of this type of membranes is very high as basically only the hydrogen ions can migrate through the membrane material. Reported operating temperatures are in the range 600-900 °C. Some sources go as high as 1000 °C [48] but no experimental results at temperatures above 900 °C have been published. High temperatures (around 900°C) are required to achieve acceptable fluxes. Chemical stability in the presence of certain species (e.g., CO2, H2S) is a major concern. These membranes are still in an early stage of development. Quite a separate category of (porous) ceramic membranes is formed by zeolites. These materials have channels inherent in their crystal structure and act as natural molecular sieves. However, achievable separation of small molecules is limited and fabrication of large thin zeolite plates still poses problems. They are presently at a very early stage of development. Generally speaking, ceramic membranes still require considerable development and testing is mainly confined to bench scale level. Demonstration of feasibility of technology scale-up is necessary [26].

3.5 Glass membranes Glass membranes are actually not very important for hydrogen separation. One of the reasons is their low selectivity. Glass membranes are porous. Depending on the pore size, they can be subdivided into micro porous (pores below 2 nm) and mesoporous (pores 2-5 nm). Micro porous membranes have higher selectivity yet lower fluxes. Both membrane types are usually produced

28

from silica using the leaching manufacturing process. The temperature range where they can be used has an upper limit of 400-500 °C. Vycor glass membranes are commercially available. 3.6 Mixed-Matrix membrane The ceramic and zeolite membranes described above have been shown to have exceptional selectivities for a number of important separations. However, the membranes are not easy to make and consequently are prohibitively expensive for many separations. One solution to this problem is to prepare membranes from materials consisting of zeolite particles dispersed in a polymer matrix. These membranes are expected to combine the selectivity of zeolite membranes with the low cost and ease of manufacture of polymer membranes. Such membranes are called mixed-matrix membranes. Mixed-matrix membranes have been a subject of research interest for more than 15 years [4951]. The concept is illustrated in Figure 3.3. At relatively low loadings of zeolite particles, permeation occurs by a combination of diffusion through the polymer phase and diffusion through the permeable zeolite particles. The relative permeation rates through the two phases are determined by their permeabilities. At low loadings of zeolite, the effect of the permeable zeolite particles on permeation can be expressed mathematically by the expression shown below, first developed by Maxwell in the 1870s [52].

Figure 3. 3 Gas permeation through mixed-matrix membranes containing different amounts of dispersed zeolite particles [13]

29

[

]

Where P is the overall permeability of the mixed-matrix material dispersed zeolite phase

eq. 3.7 is the volume fraction of the

is the permeability of the continuous polymer phase, and

is the

permeability of the dispersed zeolite phase. At low loadings of dispersed zeolite, individual particles can be considered to be well separated. At higher loadings, some small islands of interconnected particles form; at even higher loadings, these islands grow and connect to form extended pathways. At loadings above a certain critical value, continuous channels form within the membrane, and almost all the zeolite particles are connected to the channels. This is called the percolation threshold. At this particle loading, the Maxwell equation is no longer used to calculate the membrane permeability. The percolation threshold is believed to be achieved at particle loadings of about 30 vol %. Figure 3.4, adapted from a plot by Robeson et al. [53], shows a calculated plot of permeation of a model gas through zeolite-filled polymer membranes in which the zeolite phase is 1000 times more permeable than the polymer phase. At low zeolite particle loadings, the average particle is only in contact with one or two other particles, and a modest increase in average permeability occurs following the Maxwell model. At particle loadings of 25–30 vol% the situation is different—most particles touch two or more particles and most of the permeating gas can diffuse through interconnected zeolite channels. The percolation threshold has been reached, and the Maxwell model no longer applies. Gas permeation is then best described as permeation through two interpenetrating, continuous phases. At very high zeolite loadings, the mixed-matrix membrane may be best described as a continuous zeolite phase containing dispersed particles of polymer. The Maxwell model may then again apply, with the continuous and the dispersed phases in above Equation reversed. The figure also shows that the highly permeable zeolite only has a large effect on polymer permeability when the percolation threshold is reached. That is, useful membranes must contain more than 30 vol% zeolite. This observation is borne out by the limited experimental data available.

30

3.7 Motivation and guidelines for development of advanced or novel functional membranes In the last two decades, membrane technology had been established in the market, in particular for tasks where no technically and/or economically feasible alternatives exist. The successful implementation had been due to the unique separation principle based on using a membrane. By far the most processes in liquid separation are dealing with aqueous solutions, mostly at ambient or relatively low temperatures. Technically mature membrane separations with a large growth potential in the next few years include especially Ultrafiltration and Nanofiltration (with large membrane area modules) for concentration, fractionation and purification in the food, pharma and other industries [13]. Here, the selectivity of separation is still often limited, especially due to an uneven pore size distribution of the membranes. Gas Separation with membranes is also industrially established for selected applications, some in large scale. Nevertheless, many more processes could be realized if membranes with high selectivities, competitive flux and sufficient long-term stability would be available. Here, main limitations are due to insufficient membrane selectivity and/or stability. In addition, membranes suited for all kinds of applications in organic media, including higher temperatures, are still rare. Progress in all these latter areas will open the doors into large scale membrane applications in the chemical industry [54]. Furthermore, the presumably largest potential for membrane technology is in process intensification, e.g. via implementation of reaction/separation hybrid processes (membrane reactors). Therefore, membrane processes will largely contribute to the development of sustainable technologies. Finally, using specialized support and/or separation membranes in cell and tissue culture will pave the road towards biohybrid and artificial organs for medical and other applications [55]. Many scientifically interesting, technically challenging and commercially attractive separation problems cannot be solved with membranes according to the state-of-the-art. Novel membranes with a high selectivity, e.g. for isomers, enantiomers or special biomolecules are required. Consequently, particular attention should be paid to truly molecule-selective separations, i.e. advanced membranes for Nanofiltration and Ultrafiltration. Especially the development of Nanofiltration membranes for separations in organic solvents will require a much better understanding of the underlying transport mechanisms and, hence, the requirements to the polymeric materials. In addition, a membrane selectivity which can be switched by an external 31

stimulus or which can adapt to the environment/process conditions is an important vision. Such advanced or novel selective membranes, first developed for separations, would immediately find applications also in other fields such as analytics, screening, membrane reactors or bio-artificial membrane systems [29]. Specialized (tailor-made) membranes should not only have a significantly improved selectivity but also a high flux along with a sufficient stability of membrane performance. Of similar relevance is a minimized fouling tendency, i.e. the reduction or prevention of undesired interactions with the membrane. Furthermore, it should be possible to envision membrane manufacturing using or adapting existing technologies or using novel technologies at a competitive cost. The following general strategies will lead to a higher separation’s performance: 

non-porous membranes—composed of a selective transport and a stable matrix phase at an optimal volume ratio along with a minimal tortuosity of the transport pathways, thus combining high selectivity and permeability with high stability;



porous membranes—with narrow pore size distribution, high porosity and minimal tortuosity (ideally: straight aligned pores though the barrier);



additional functionalities for selective interactions (based on charge, molecular recognition or catalysis) combined with non-porous or porous membrane barriers;



membrane surfaces (external, internal or both) which are ‘inert’ towards uncontrolled adsorption and adhesion processes.

In addition, minimizing the thickness of the membrane barrier layer will be essential. For certain completely novel membrane processes, e.g. in micro-fluidic systems, it should be possible to fulfill special processing requirements. This can be envisioned considering the large flexibility with respect to the processing of polymeric materials. All these above outlined requirements can efficiently be addressed by various approaches within the field of nanotechnology.

32

Chapter 4 4. MEMBRANE SYSTEM DESIGN 4.1 Introduction Chapter 3 presented the membranes relevant to hydrogen separation and their most important properties. This chapter takes membranes a step further by providing an introduction into the design of actual membrane systems. A strong point of membranes is their versatility. A great number of variables can be chosen to arrive at an optimal design. At the same time, this versatility means it's very difficult to provide an in-depth membrane system design manual. Therefore, this chapter attempts not to provide a complete membrane system design overview, but merely some basic insights into the membrane design process. It is based on the design considerations presented in [18]. The first step in membrane system design is selecting a feed flow to be separated. The properties of this flow (temperature, pressure, composition), together with cost considerations, determine the membrane to be used, using the information provided in chapter 3. The next choice is what module shape is to be used. A module is the smallest practical unit containing a set membrane area and any supporting structures. For each membrane certain module shapes are commercially available. If another shape is desired, this will need to be designed and produced specifically for this application, leading to high cost. Apart from cost, other factors influencing the choice of a module shape are the operating pressure and fouling (mainly concentration polarization). Once the module shape is selected, the overall membrane system can be designed. In between the module and system level is the stage level. A stage is formed by one or more membrane modules assembled into an operating unit that provides a specific function different from any other membrane stages that may be utilized in the same process. System design is a matter of combining a number of stages in the optimal way [27]. This chapter starts with the membrane module designs currently used. Section 4.3 presents the transport equation used to calculate membrane fluxes. Section 4.4 shows the basics for a single stage membrane system. Some limitations of single stage systems can be overcome by using a multistage system. Multiple and Multistage membrane process and recycle design come after in section 4.5 and 4.6. Subsequently the application in hydrogen separation comes in section 4.7. If (substantial) chemical reactions occur inside a membrane system it is called a membrane reactor. 33

Some design considerations for membrane reactors are described in section 4.8. Finally, some practical issues, such as membrane deterioration are treated in section 4.9 at last.

4.2 Current membrane module designs The building block of a membrane system is called the module. All module types applied so far are based on two types of membrane configurations: flat and tubular. To use a membrane in a given application it has to be installed in a proper device which is generally referred to as membrane module. Membrane modules must meet certain requirements as far as their production costs, their packing density, energy consumption, and especially the control of concentration polarization and membrane fouling is concerned. A large number of different module types are described in the literature. In laboratory application often a stirred cell is used in batch or feed and bleed operation mode. However, on a large industrial scale mainly six basic types are used today. These modules are shown schematically in Figures 4.1-10. They are quite different in their design, their mode of operation, their production costs, and the energy requirements for pumping the feed solution through the module. A very important criterion is the control of concentration polarization and membrane fouling. Module types based on flat membranes are the plate-and-frame and spiralwound modules. Tubular-type membrane modules are subdivided into tubular, capillary and hollow fiber modules. The difference between the latter three is mainly based on their dimensions: tubular membranes are defined as having diameters larger than 10 mm, capillary membranes having diameters between 10 mm and 0.5 mm and hollow fiber membranes having diameters below 0.5 mm. A more extensive review of membrane modules is presented in [13, 20, 26].In some modules, such as the tubular, the plate-and-frame and the capillary type module, concentration polarization and membrane fouling can effectively be controlled by the proper feed flow. Other modules, such as the spiral-wound and the hollow fiber module are more sensitive for membrane fouling. There is not one module type that serves all membrane processes and applications. The different modules which are commercially available today are designed for a certain application and process in which they provide the technical and commercially best solution. The same is true for the process design and mode of operation. In certain applications batch and feed-and-bleed operation is used while in other applications continuous operation is more efficient.

34

4.2.1 The cartridge membrane module The pleated cartridge filter membrane module, which is shown in Figure 4.1, is used mainly in dead-end microfiltration. It consists of a pleated membrane cartridge installed in a pressurized housing. The feed solution enters the filter from the housing side and the product is collected in a center tube which is sealed against the housing by an O-ring. Cartridge type filters are operated at relatively low hydrostatic pressures of 1 to 2 bars. Their useful life is limited due to plugging of the membrane pores by the retained solutes.

membrane env elope f eed f low

f eed f low

membrane

knif e-edge seal permeate Figure 4. 1 Schematic drawing showing a cartridge filter unit [20]

The useful life of a cartridge filter depends very much on the feed solution constituents and their concentration. It can be a couple of days up to several months. Cartridge filters are disposable items. If the membrane which has the typical characteristics of a depth filter is blocked by the retained particles it can generally not be cleaned and the original flux restored. The main application of cartridge filters is in sterile filtration of water, beverages, such as wine, beer or 35

fruit juices and pharmaceutical solutions. Furthermore, they are used for the filtration of surface water for industrial or household use to remove particles and suspended matters and tohe serve as pre-filters in reverse osmosis water desalination plants. An important application of cartridge filters is in the production of ultra-pure water for the electronic industry where they are used as point-of-use filters to remove all traces of particles from the ultra-pure water used to rinse electronic components. The actual cartridge is made by pleating a membrane sheet and potting the ends by an appropriate resin or hot-melt-glue. Since the process of pleating the membrane puts a considerable mechanical stress on the membrane only materials that are rather flexible and have a certain mechanical strength are suited for the fabrication of cartridge filters. The cartridge filter module provides a relatively high surface area per unit volume and the production costs are also relatively low. 4.2.2 The plate-and-frame membrane module Another module type used on an industrial scale for various membrane separation processes including ultrafiltration, reverse osmosis, and gas separation is the plate-and-frame module. Its design has its origin in the conventional filter press-concept. The membranes, porous membrane support plates, and spacers forming the feed flow channel are clamped together and stacked between two endplates and placed in housing as indicated in the schematic diagram of Figures 4.2 and 4.3.

permeate channel spacer f eed channelspacer permeate retentate

f eed solution

permeate Figure 4. 2 Schematic drawing illustrating the concept of a plate-and-frame membrane module[20]

The feed solution is pressurized in the housing and forced across the surface of the membrane. The permeate is leaving the module through the permeate channel to a permeate collection

36

manifold which in circular devices is central tube as indicated in the Figure 4.3. Often the device contains one or more baffles to extend the path-length of the feed solution in the device.

f eed solution

retentate permeate Figure 4. 3 Circular plate-and-frame filter device with one baffle to extend the feed flow path length [13]

There are various types of plate-and-frame modules on the market which offer, however, only slight variations in their basic configuration. In many planet-and-frame membrane modules the membranes can easily be exchanged. This makes the module suitable for batch-type operations and multi-purpose applications using different membranes for different separation tasks. Plate-and-frame units are mainly 37

used in small scale applications such as in the production of certain pharmaceuticals, bio-products, or fine chemicals. The housings and other components of plate-and–frame modules to be used in the food and pharma industry are made from stainless steel so that they can easily be steam sterilized. These units, however, are quite expensive and the exchange of the membranes is labor intensive. Therefore, the plateand-frame module is quite expensive.

4.2.3 The plate-and-frame membrane module A variation of the basic plate-and-frame concept is the spiral-wound module, which is widely used today in reverse osmosis, ultrafiltration, and gas separation. Its basic design is illustrated in Figure 4.4.

collection pipe retentate f low f eed f low f eed spacer

permeate

membrane

f eed f low

permeate f low

membrane f eed spacer Figure 4. 4 Schematic drawing of a spiral-wound membrane module [13]

The feed flow channel spacer, the membrane, and the porous membrane support form an envelope which is rolled around a perforated central collection tube and inserted into an outer tubular pressure shell. The feed solution passes in axial direction through the feed channel across the membrane surface. The filtrate is moves along the permeate channel and is collected in a perforated tube in the center of the roll. Small spiral wound units consist of just one envelope which limits the total membrane area that can be installed in one unit to about 1 to 2 m 2. The main reason for the limitation of the surface area which can be installed in a module containing one single envelope is the pressure drop encountered by the permeate moving down the permeate channel to the central collection tube. Because the channel in a practical unit is very narrow its 38

length is limited to 2 to 5 m. A significantly longer path would resultr in an unacceptable pressure drop in the permeate channel. To install larger membrane surfaces in a spiral wound module a multi-leaf arrangement in used as indicated in the Figure 4.5. concentrate

permeate outlet feed flow membrane permeate flow

feed solution spacer

membrane porous centre tube permeate spacer Figure 4. 5 Schematic drawing illustrating the construction of a multi-leaf spiral wound module [20]

Commercial spiral wound modules are about 1 meter long and have a diameter of 10 to 60 cm. The membrane area in a spiral-wound element is 3 and 60 m2. Generally, 2 to 6 elements are placed in series in a pressure vessel. The spiral-wound module provides a relatively large membrane area per unit volume. The large scale production is quite cost effective and module costs per membrane area quite low. The major application of the spiral-wound module is in reverse osmosis sea and brackish water desalination. But it is also extensively used in ultrafiltration and gas separation. However, the spiral-wound module is quite sensitive to fouling, and the feed channels can easily be blocked and particles or fibers should be removed from the feed solution by a proper pretreatment procedure. 39

4.2.4 The tubular membrane module While the previously described three membrane module types required flat sheet membrane material for their preparation, special membrane configurations are needed for the preparation of the tubular, capillary, and hollow fiber modules. The tubular membrane module consists of membrane tubes placed into porous stainless steel of fiber glass reinforced plastic pipes. The pressurized feed solution flows down the tube bore and the permeate is collected on the outer side of the porous support pipe, as indicated in Figure 4.6. The diameters of tubular membranes are typically between 1-2.5 cm. In some modules, the membranes are cast directly on the porous pipes and in others they are prepared separately as tubes and then installed into the support pipes. Today, tubular modules are used in ultrafiltration at low hydrostatic pressures. This allows the membrane tubes to be made by a welding or gluing procedure of flat sheet membranes that are cast on a relatively thick and mechanically strong porous polyester support material. These tubes which have a diameter of 0.5 to 1 cm do not need additional support when operated at hydrostatic pressures of less than 2 to 4 bars. Usually, 10 to 30 individual tubes are installed in a larger tube and potted at the end of the tube. The feed solution is fed in parallel through the tubular bundle while the permeate of the individual tubes is collected in the outer shell tube as indicated in the schematic drawing of Figure 4.7. The main advantage of the tubular module is that concentration polarization effects and membrane fouling can be easily controlled, and plugging of the membrane module is avoided even with feed solutions that have very high concentration of solid matter and thus high viscosity. The disadvantage of the tubular module design is the low surface area, that can be installed in a given unit volume, and the very high costs. Therefore, tubular membrane modules are generally only applied in applications where feed solutions with high solid content and high viscosity have to be treated and other module concepts fail due to membrane fouling and module plugging. This is the case in certain applications in the food and pharma industry and in the treatment of certain industrial effluents.

40

feed permeate

concentrate

permeate

feed solution

membrane porous tube Figure 4. 6 Schematic drawing illustrating the tubular membrane module [26]

retentate

permeate f eed Figure 4. 7 Tubular module with seven individual tubes bundled in a shell tube [26]

41

4.2.5 The capillary membrane module The capillary membrane module, which is shown schematically in Figure 4.8, consists of a large number of membrane capillaries with an inner diameter of 0.2 to 3 mm arranged in parallel as a bundle in a shell tube. The feed solution is passed down the center of the membrane capillary and the filtrate, which permeates the capillary wall, is collected in the shell tube. capillary membrane

permeate

f eed retentate

Figure 4. 8 Schematic diagram showing a capillary membrane module [20]

The capillary membrane module requires membranes in a self-supporting capillary configuration, which when asymmetrically structured, carry the selective barrier on the inner side of the capillary as indicated in the scanning electron micrograph of Figure 4.9, which shows a typical capillary, ultrafiltration membrane prepared in a wet-spinning process.

Figure 4. 9 SEM of a capillary membrane with the selective “skin” on the inside of capillary [20]

42

The capillary membrane module provides a high membrane area per module volume. The production costs are very low and concentration polarization and membrane fouling can effectively be controlled by the proper feed flow and back-flushing of the permeate in certain time intervals. The main disadvantage of the capillary membrane module is the required low operating pressure. Because of the limited stability of the capillary membranes operating pressures generally can not exceed 4 to 6 bars. Therefore, the capillary membrane is used in applications where low transmembrane pressures are applied, i.e. in dialysis, microfiltration, and low pressure ultrafiltration. The most significant application of the capillary membrane module is as artificial kidney. 4.2.6 The hollow fiber membrane module The same basic spinning process is used for the preparation of hollow fiber membranes, which have an outer diamter of 50 to 100 µm. In hollow fiber membranes, the selective layer is on the outside of the fibers, which are installed as a bundle of several thousand fibers in a half loop with the free ends potted with an epoxy resin in a pressure tube as indicated in Figure 4.10. The filtrate passes through the fiber walls and flows up the bore to the open end of the fibers at the epoxy head.

feed solution

shell tube c oncentrate

permeate

hollow fiber

epoxy resin

Figure 4. 10 Schematic drawing illustrating the construction of a hollow fiber module [26]

43

The hollow fiber membrane module has the highest packing density of all module types available on the market today. Its production is very cost effective and hollow fiber membrane modules can be operated at pressures in excess of 100 bars. The main disadvantage of the hollow fiber membrane module is the difficult control of concentration polarization and membrane fouling. When operated with liquid solutions the modules do not tolerate any particals, macromolecules or other materials that may easily precipitated at the membrane surface. Therefore, an extensive pretreatment is required when hollow fiber membranes are used for the treatment of liquid mixtures. The main application of the hollow fiber module is today in reverse osmosis desalination of sea water and in gas separation. Both application require high operating pressures and low cost membranes which have a long useful life. In reverse osmosis, of sea water an extensive pretreatment of the sea water is required.

4.2.7 Other membrane modules There are several more membrane module concepts, such as the rotating disk module or the transversal flow module. In a rotating disk module the membranes are arranged as in the circular plate-and-frame module. However, the feed solution is not pumped through the module but the membrane plates are rotating and thus providing some turbulence in the feed solution at the membrane surface. This mudule is rather costly per installed membrane area and is used only in very specific applications in the pharmaceutical industry when solutions with very shear sensitive constituents have to be processed. In the following table 4.1 the different the more significant membrane modules, there costs per membrane area, and main application are sumarized.

44

Table 4. 1 Commercially available membrane modules, there costs and major applications [12]

Membrane

Membrane area

Membrane

Control of

module

per unit volume

costs

concentration polarization

low

Very poor

Dead-end MF

medium

good

MF, UF, RO, D,

low

good

UF, RO, GS

Tubular module 20 - 100

very high

very good

MF, UF, RO

Capillary

low

very good

UF,

(m2 m-3 ) Filter cartridge

800 -1000

Application

module Plate-and-frame 400 - 800 module Spiral-wound

800 - 1200

module

600 - 1200

module Hollow fiber

MF,

D,

SLM 2000 - 5000

very low

very poor

RO, GS

module MF = microfiltration UF = ultrafiltration RO = reverse osmosis D = dialysis GS = gas separation

The selection of a module shape depends on a number of factors, among which cost are very important. For each membrane certain module shapes are commercially available. If another shape is desired, it will need to be designed a new and produced, leading to high cost. For gas separation the module price depends to a large extent on the specific application (pressure, temperature, etc.). As the pressure vessel in which the membranes are located is the main determinant in total membrane unit cost for high temperature and pressure applications, under 45

these circumstances a high packing density is desirable. This favors application of hollow fiber and to a lesser extent capillary and spiral wound modules. A disadvantage of thinner tubes is that the number of seals is relatively high. This will lead to extra cost, as cheaper sealing solutions, mainly resins, cannot cope with high temperatures.

4.3 Calculating membrane permeate fluxes for gas separation The three factors that determine the performance of a membrane gas separation system are illustrated in Figure 4.11. The role of membrane selectivity is obvious; not so obvious are the importance of the ratio of feed pressure ( usually called the pressure ratio,

) to permeate pressure (

) across the membrane,

and defined as eq. 4.1

and of the membrane stage-cut,

, which is the fraction of the feed gas that permeates the

membrane, defined as eq. 4.2

Figure 4. 11 Parameters affecting the performance of membrane gas separation systems[13]

46

4.3.1 Pressure Ratio The importance of pressure ratio in the separation of gas mixtures can be illustrated by considering the separation of a gas mixture with component concentrations of feed pressure

and

at a

. A flow of component i across the membrane can only occur if the partial

pressure of i on the feed side of the membrane ( the permeate side of the membrane (

) is greater than the partial pressure of i on

) that is:

eq. 4.3 It follows that the maximum separation achieved by the membrane can be expressed as:

eq. 4.4 That is, the separation achieved can never exceed the pressure ratio ϕ, no matter how selective the membrane: eq. 4.5 The relationship between pressure ratio and membrane selectivity can be derived from the Fick’s law expression for the fluxes of components i and j,

eq. 4.6 And (

)

eq. 4.7

The total gas pressures on the feed and permeate side are the sum of the partial pressures and the volume fractions of components i and j on the feed and permeate side are also related to partial pressures. From mass balance consideration:

eq. 4.8

47

By combining above equations yields an expression linking the concentration of component i on the feed and permeate sides of the membrane:

[

√(

)

]

eq. 4.9

This somewhat complex expression breaks down into two limiting cases depending on the relative magnitudes of the pressure ratio and the membrane selectivity. First, if the membrane selectivity (α) is very much larger than the pressure ratio ( ), that is: eq. 4.10 The equation becomes: eq. 4.11 This is called the pressure-ratio-limited region, in which the performance is determined only by the pressure ratio across the membrane and is independent of the membrane selectivity. If the membrane selectivity (α) is very much smaller than the pressure ratio ( ), that is: eq. 4.12 The equation becomes: eq. 4.13 This is called the membrane-selectivity-limited region, in which the membrane performance is determined only by the membrane selectivity and is independent of the pressure ratio. There is, of course, an intermediate region between these two limiting cases, in which both the pressure ratio and the membrane selectivity affect the membrane system performance. These three regions are illustrated in Figure 4.12, in which the calculated permeate concentration (

) is plotted

versus pressure ratio ( ) for a membrane with a selectivity of 30 [56]. At a pressure ratio of 1, feed pressure equal to the permeate pressure, no separation is achieved by the membrane. As the difference between the feed and permeate pressure increases, the concentration of the more permeable component in the permeate gas begins to increase, first according to Eq. 4.11 and then, when the pressure ratio and membrane selectivity are comparable, according to Eq. 4.9. At very high pressure ratios, that is, when the pressure ratio is four to five times higher than the membrane selectivity, the membrane enters the membrane-selectivity controlled region. In this region the permeate concentration reaches the limiting value given by Eq. 4.13. 48

Figure 4. 12 Calculated permeate vapor concentration for a vapor-permeable membrane with a vapor/nitrogen selectivity of 30 as a function of pressure ratio.[13]

4.3.2 Stage-cut Another factor that affects membrane system design is the degree of separation required. The usual target of a gas separation system is to produce a residue stream essentially stripped of the permeable component and a small, highly concentrated permeate stream. These two requirements cannot be met simultaneously; a tradeoff must be made between removal from the feed gas and enrichment in the permeate [13]. The system attribute that characterizes this tradeoff is called the stage-cut. The effect of stage-cut on system performance is illustrated in Figure 4.13. In the example calculation shown in Figure 4.13, the feed gas contains 50% of a permeable gas (i) and 50% of a relatively impermeable gas (j). Under the assumed operating conditions of this system (pressure ratio 20, membrane selectivity 20), it is possible at zero stage-cut to produce a permeate stream containing 94.8% of component i. But the permeate stream is tiny and the residue stream is still very close to the feed gas concentration of 50 %. As the fraction of the feed gas permeating the membrane is increased by increasing the membrane area, the concentration of

49

the permeable component in the residue and permeates streams falls. At a stage-cut of 25 %, the permeate gas concentration has fallen from 94.8% (its maximum value) to 93.1 %. The residue stream concentration of permeable gas is then 35.5 %. Increasing the fraction of the feed gas that permeates the membrane to 50% by adding more membrane area produces a residue stream containing 11.8% of the permeable gas. However, the gas permeating the added membrane area only contains 83.0% of the permeable component, so the average concentration of permeable component in the permeate stream is reduced from 93.1 to 88.1 %. If the fraction of the feed gas that permeates the membrane is increased to 75% by adding even more membrane area, the concentration of the permeable component in the residue stream is reduced to only 0.04 %. However, the gas permeating the added membrane area only contains 23.8% of the permeable component, less than the original feed gas. The average concentration of the permeable component in the feed gas is, therefore, reduced to 66.7 %. This means that one-half of the less permeable component has been lost to the permeate stream.

Figure 4. 13 The effect of stage-cut on the separation of a 50/50 feed gas mixture (pressure ratio, 20; membrane selectivity, 20).[13]

The calculations shown in Figure 4.13 illustrate the trade-off between recovery and purity. A single-stage membrane process can be designed for either maximum recovery or maximum 50

purity, but not both. The calculations also show that membranes can produce very pure residue gas streams enriched in the less permeable component, although at low recoveries. However, the enrichment of the more permeable component in the permeate can never be more than the membrane selectivity, so a membrane with low selectivity produces an only slightly enriched permeate. This is why membranes with an oxygen/nitrogen selectivity of 4–6 can produce very pure nitrogen (>99.5 %) from air on the residue side of the membrane, but the same membranes cannot produce better than 50–60% oxygen on the permeate side. If the more permeable component must be pure, very selective membranes are required or multistage or recycle membrane systems must be used. Finally, the calculations in Figure 4.13 show that increasing the stage-cut to produce a pure residue stream requires a disproportionate increase in membrane area. As the feed gas is stripped of the more permeable component, the average permeation rate through the membrane falls. In the example shown, this means that permeating the first 25% of the feed gas requires a relative membrane area of 1, permeating the next 25% requires a membrane area increment of 1.8, and permeating the next 25% requires an increment of 6.7.

4.4 Single stage membrane processes The simplest membrane processes are single stage membrane processes. A stage is formed by one or more membrane modules assembled into an operating unit that provides a specific function different from any other membrane stages that may be utilized in the same process.

Figure 4. 14 Simple membrane set-up

The basic membrane stage set-up is shown in Figure 4.14. Generally, membrane operations can be subdivided into dead-end and cross-flow operations. Dead-end operations occur if there is no retentate stream (the only exit of the feed side is the membrane). Although the recovery is high

51

for this kind of operation, it is usually not preferred as non-permeating species in time become more abundant on the feed side, leading to so-called concentration polarization. This means that the feed side concentration of the permeating specie decreases (in time), reducing the driving force and therewith the transport through the membrane. Instead, cross-flow operations are preferred, i.e. configurations in which (some of) the flows run alongside the membrane [27]. In this set up, deterioration of membrane flux in time is limited. In general, 4 cases of cross-flow operations are distinguished: 

co-current;



counter-current;



cross-flow with perfect permeate mixing;



perfect mixing.

Co-current operation means that feed and the permeate (sweep) flows run in the same direction, whereas counter-current flow means that feed and permeate (sweep) flows run in opposite direction. In the perfect permeate mixing set up, the permeate is mixed, resulting in one permeate composition along the membrane length coordinate (see Figure 4.15). The perfect mixing set up results in one feed side composition and one permeate side composition. In the co-current set up, the initial driving force is large, but it decreases as permeation takes place (see Figure 4.15). If the membrane length is chosen large enough, the partial pressures on both sides will become (almost) equal. Therefore, the working efficiency of a membrane decreases with increasing membrane surface area. In the counter-current set up the permeate partial pressure decreases towards the (feed flow) exit of the module, resulting in a substantial driving force (partial pressure difference) even there. In the perfect permeate mixing set up, the permeate partial pressure is the same over the whole membrane length. The result is a retentate partial pressure approaching the permeate partial pressure, if the membrane surface area is chosen large enough. As the partial pressures on feed and permeate side are constant for each side of the membrane for the perfect mixing set up, this set up results in a constant driving force. Generally, the best membranes results are obtained using a counter-current flow set up, followed by cross-flow with permeate mixing and by co-current flow set up. Perfect mixing usually delivers the worst results [20].

52

Figure 4. 15 Comparison co-current, counter-current, perfect permeate mixing and perfect mixing set up [27]

53

Apart from the flow operation, an endless number of variables can be adjusted to arrive at the optimal membrane set up. A few options will be presented here. It should be kept in mind that both the retentate and the permeate can be the desired flows. Figure 4.16 shows a membrane stage with feed compression. Feed compression will increase the pressure differential, and thus the driving force over the membrane. Consequently product recovery will increase. Compression in itself adds to the total plant cost, but reduced product loss results in smaller membrane plant size and lower cost. The net effect on cost will depend on the specific situation.

Figure 4. 16 Single stage membrane processes with feed flow compression

Another way to increase the pressure differential over the membrane is to decrease the permeate pressure, possible down to vacuum. This set up is shown in Figure 4.17. An advantage of this set up over feed flow compression is that the flow to be compressed by the vacuum pump is generally smaller than the feed flow.

Figure 4. 17 Single stage membrane process with permeate vacuum

As shown before, the partial pressure differential over the membrane can be influenced by increasing the absolute pressure differential over the membrane (e.g., by feed compression or permeate vacuum), but also by changing the compositions of feed and permeate flow. Figure 4.18 shows one way of achieving just this, by diluting the permeate with a sweep flow. Again, the improved membrane transport and thus a smaller membrane unit may offset the extra cost of providing a sweep flow (e.g., steam production). 54

Figure 4. 18 Single stage process with permeate dilution by means of sweep flow

Furthermore, the membrane retentate or permeate flows can also be partly recycled (see Figure 4.19). This provides a more flexible set up and can enhance product recovery. Especially with polluted flows and concentration polarization this can be a good option. Noble and Stern come to the following generalizations for single-stage membrane processes: 

Membrane area requirements are reduced and product recoveries are increased as the pressure differential over the membrane increases;



The absolute permeate pressure (or pressure ratio) is also important. Higher pressure ratios lead to improved separation performance. Pressure differential alone does not define membrane performance;



Increasing membrane area leads to a purer residue but a less pure permeate;



Product recoveries tend to drop rapidly as the purity requirement is increased.

Figure 4. 19 Single stage membrane processes with recycle

55

4.5 Multiple and Multistage membrane process Because the membrane selectivity and pressure ratio achievable in a commercial membrane system are limited, a one-stage membrane system may not provide the separation desired. To improve membrane system performance, multistage membrane systems can be built. These usually require additional equipment but the costs are often small relative to the gains from the process improvements [18]. Even a simple split up of a single membrane stage into two stages may prove advantageous under certain circumstances. Although the final permeate flow in Figure 4.20 will not be any different from the permeate flow of the single stage system before the split up, the separate permeate flows of both membrane units will be different (and can be quite different). This difference can be used if the streams are used for different purposes.

Figure 4. 20 Two-stage membrane process as simple split up of single stage process

Most multistage membrane systems incorporate some sort of recycle to enhance product separation and recovery. Such designs are easy to implement from a membrane standpoint, but always require compression for the recycle stream. Gas compression is expensive, but recycling generally improves overall process efficiency [18].

Figure 4. 21 Two-stage membrane process with permeate recycle

56

The process shown in Figure 4.21 can be used to enhance product recovery. The first membrane unit produces a permeate with the desired product purity, yet the retentate still contains a relatively large fraction of the desired product. The second membrane unit is used to reclaim the main part of this desired product from the retentate flow. Since the purity of the permeate of this second membrane unit is considered too low, it is recycled to the feed flow. A multistep design of this type can achieve almost complete removal of the permeable component from the feed stream to the membrane unit. However, greater removal of the permeable component is achieved at the expense of increases in membrane area and power consumption by the compressor. As a rule of thumb, the membrane area required to remove the last 9% of a component from the feed equals the membrane area required to remove the first 90%. Sometimes, 90% removal of the permeable component from the feed stream is acceptable for the discharge stream from the membrane unit, but a higher concentration is needed to make the permeate gas usable.

Figure 4. 22 two-stage system to produce a highly concentrated permeate stream [13]

In this situation, a two-stage system of the type shown in Figure 4.22 is used. In a two-stage design, the permeate from the first membrane unit is recompressed and sent to a second membrane unit, where a further separation is performed. The final permeate is then twice enriched. In the most efficient two-stage design, the residue stream from the second stage is reduced to about the same concentration as the original feed gas, with which it is mixed. In the example shown in Figure 4.22, the permeate stream, concentrated a further five-fold, leaves the system at a concentration of 21 %. Because the volume of gas treated by the second-stage membrane unit is much smaller than in the first stage, the membrane area of the second stage is 57

relatively small. Thus, incorporation of a second stage only increases the overall membrane area and power requirements by approximately 15–20%. Multistage/multistep combinations of two-step and two-stage processes can be designed but are seldom used in commercial systems—their complexity makes them uncompetitive with alternative separation technologies. More commonly some form of recycle design is used.

4.6 Recycle Designs A simple recycle design, sometimes called a two-and-one-half-stage system, proposed by Wijmans [57] is shown in Figure 4.23. In this design, the permeate from the first membrane stage is recompressed and sent to a two-step second stage, where a portion of the gas permeates and is removed as enriched product. The remaining gas passes to another membrane stage, which brings the gas concentration close to the original feed value. The permeate from this stage is mixed with the first-stage permeate, forming a recycle loop. By controlling the relative size of the two second stages any desired concentration of the more permeable component can be achieved in the product. Recycle designs of this type are limited to applications in which the components of the gas mixture, if sufficiently concentrated, can be separated from the gas by some other technique.

Figure 4. 23 Two-and-one-half-stage systems: by forming a recycle loop around the second stage, a small, very concentrated product stream is created [13]

58

4.7 Application The membrane gas separation industry is still growing and changing. Most of the large industrial gas companies now have membrane affiliates: Air Products (Permea), MG (Generon), Air Liquide (Medal) and Praxair (IMS). The affiliates focus mainly on producing membrane systems to separate nitrogen from air, but also produce some hydrogen separation systems. Another group of companies, UOP (Separex), Natco (Cynara), Kvaerner (GMS) and ABB Lummus Global (MTR), produces membrane systems for natural gas separations. A third group of smaller independents are focusing on the new applications, including vapor separation, air dehydration and oxygen enrichment. The following section covers the major current application in Hydrogen separation. 4.7.1 Hydrogen Separation The first large-scale commercial application of membrane gas separation was the separation of hydrogen from nitrogen in ammonia purge gas streams. The process, launched in 1980 by Monsanto, was followed by a number of similar applications, such as hydrogen/methane separation in refinery off-gases and hydrogen/carbon monoxide adjustment in oxo-chemical synthesis plants [58]. Hydrogen is a small, noncondensable gas, which is highly permeable compared to all other gases. This is particularly true with the glassy polymers primarily used to make hydrogen selectivity membranes; fluxes and selectivities of hydrogen through some of these materials are shown in Table 4.2. With fluxes and selectivities as high as these, it is easy to understand why hydrogen separation was the first gas separation process developed. Early hydrogen membrane gas separation plants used polysulfone or cellulose acetate membranes, but now a variety of specifically synthesized materials, such as polyimides (Ube, Praxair), polyaramide (Medal) or brominated polysulfone (Permea), are used. A typical membrane system flow scheme for recovery of hydrogen from an ammonia plant purge gas stream is shown in Figure 4.24. During the production of ammonia from nitrogen and hydrogen, argon enters the high-pressure ammonia reactor as an impurity with the nitrogen stream and methane enters the reactor as an impurity with the hydrogen. Ammonia produced in the reactor is removed by condensation, so the argon and methane impurities accumulate until they represent as much as 15% of the gas in the reactor. To control the concentration of these components, the reactor must be continuously purged. The hydrogen lost with this purge gas can

59

represent 2–4% of the total hydrogen consumed. These plants are very large, so recovery of the hydrogen for recycle to the ammonia reactor is economically worthwhile. Table 4. 2 Hydrogen separation membrane [13]

Membrane (developer)

selectivity

Hydrogen Pressure-normalized flux [10-6 2 cm3(STP)/cm .s.cmHg]

H2 /CO

H2 /CH4

H2 /N2

Polyaramide (Medal) Polysulfone (Permea)

100

>200

>200

—

40

80

80

100

Cellulose acetate (Separex)

30-40

60-80

60-80

200

50

100-200

100-200

80-200

Polyimide (Ube)

Figure 4. 24 Simplified flow schematic of the PRISM® membrane system to recover hydrogen from an ammonia reactor purge stream [13]

60

In the process shown in Figure 4.24, a two-step membrane design is used to reduce the cost of recompressing the hydrogen permeate stream to the very high pressures of ammonia reactors. In the first step, the feed gas is maintained at the reactor pressure of 135 atm, and the permeate is maintained at 70 atm, giving a pressure ratio of 1.9. The hydrogen concentration in the feed to this first step is about 45 %, high enough that even at this low pressure ratio the permeate contains about 90% hydrogen. However, by the time the feed gas hydrogen concentration has fallen to 30 %, the hydrogen concentration in the permeate is no longer high enough for recycle to the reactor. This remaining hydrogen is recovered in a second membrane step operated at a lower permeate pressure of 28 atm and a pressure ratio of 4.7. The increased pressure ratio increases the hydrogen concentration in the permeate significantly. By dividing the process into two steps operating at different pressure ratios, maximum hydrogen recovery is achieved at minimum recompression costs. A second major application of hydrogen-selective membranes is recovery of hydrogen from waste gases produced in various refinery operations [58-60]. A typical separation—treatment of the high-pressure purge gas from a hydrotreater—is shown in Figure 4.25. The hydrogen separation process is designed to recycle the hydrogen to the hydrotreater. As in the case of the ammonia plant, there is a trade-off between the concentration of hydrogen in the permeate and the permeate pressure and subsequent cost of recompression. In the example shown, a permeate of 96.5% hydrogen is considered adequate at a pressure ratio of 3.9.

61

Figure 4. 25 Hydrogen recovery from a hydrotreater used to lower the molecular weight of a refinery oil stream. Permea polysulfone membranes (PRISM®) are used[13]

62

4.8 Membrane reactors Following the work of Gryaznov on membrane reactors using palladium alloys, the use of membranes to increase the conversion of reversible reactions by separating gaseous products has been largely studied by a number of researchers. Some significant advantageous of membrane reactors, compared to conventional reactors are [61]: 

Yield-enhancement of equilibrium limited reactions compared to conventional reactors.



The reactor and the membrane can be divided into two individual compartments. For some reactions (e.g., oxidative dehydrogenation reaction) this aspect may be very important: by separating the stream and the oxidant, the extent of the side-reactions can be significantly decreased.



By using a membrane, it is possible to control the interaction between two reactants.



The stoichiometry of the reaction can be easily maintained.



The combination of the two processes (catalytic reactor and down-stream separation units) into one unit will reduce capital costs.

An important advantage of membranes derives from the ability to selectively permeate species taking place in equilibrium reactions. If no membranes are applied, temperatures are adjusted and/or reactant species are added to shift these reactions to the product side. Both options are often not favorable from a system efficiency point of view. Membranes offer the opportunity to selectively take away reaction products, thereby shifting the equilibrium to the product side. If chemical reactions are carried out (on a considerable scale) in a membrane module, the system is called a membrane reactor. The fact that reactions occur in membrane modules complicates the module design. The reactions may result in substantial composition changes, influencing membrane operation through changes in partial pressures or formation/depletion of contaminants. For reactions to occur in a membrane module, catalysts may be needed. These catalysts will need to be accommodated inside the membrane reactor. Three types of arrangements are found to accomplish this [20]: 

catalyst placed inside the feed stream;



catalyst placed in a membrane top layer;



catalyst placed inside the membrane itself.

63

The most simple catalyst arrangement is the first option: placed inside the feed stream. This arrangement is easy to prepare and operate. If needed, the catalyst can also easily be replaced. If the catalyst is placed in a membrane top layer or inside the membrane itself, replacing the catalyst involved usually means replacing the complete membrane. The most important issue of a membrane reactor is likely to be heat balancing. Reactions usually have temperature effects, either endothermic or exothermic. If several reactions occur at the same time they may roughly balance each other thermally. However, it is more probable that there will be a need for heat exchange. This can place large constraints on membrane reactor design. For example, in case of methane reforming membrane reactor, a large heat influx is required, since the reforming reaction is strongly endothermic. To introduce this heat at the places where the reaction has to occur (near the membrane where the hydrogen needs to be taken away) is a complicated task. The potential applications of membrane reactors are numerous, but commercial applications are hindered by practical limitations such as low separation factors, leakage at higher temperatures, poisoning of catalysts and mass transfer limitations [20].

4.9 Some practical issues Membrane performance generally decreases with time. This phenomenon can be caused by concentration polarization and fouling. Concentration polarization occurs because of limited permeation of certain species. These species will become higher in concentration directly adjacent to the membrane reducing permeate transport. The size of this effect depends on the type of species used and the flow set up. Concentration polarization is usually not a very severe problem for gas separation membranes. If concentration polarization does occur, turbulent flow can be promoted in the gas flow directly adjacent to the membrane to minimize the negative effect. By doing so, a better mix of the concentrated layer with the feed flow will be obtained. Decrease in flux due to concentration polarization is generally constant over time Fouling is said to occur when species adsorb to the membrane surface (also inside the pores), limiting or even blocking permeation. Notable examples of species fouling membranes are sulphur containing species such as H2S and SO2. To mitigate the effects of fouling, membranes can be cleaned by heating and purging with non-adsorbing gases. Small particles can best be removed from the feed flow using a filter.

64

Membrane deterioration may also be caused by compaction, i.e. a reduction of pore size, because of pressurization. This phenomenon occurs with polymer membranes and is usually irreversible; most often the pore size doesn't return to its original value when pressure is decreased. In making choices to obtain the optimal design, other practical considerations come into play. One of these is the effect of thermal stresses on the structural integrity. If temperature variations occur, several parts of the system may experience different degrees of expansion. If there is no room to accommodate these differences in expansion, the system can be seriously damaged. Furthermore, the pressure drop over a membrane unit (not the membrane itself) is directly proportional to the module length. To reduce the pressure drop, it would be advantageous to apply shorter modules. However, shorter modules will lead to more seals.

65

Result and Discussion Chapter 5 5.1 Comparison on different hydrogen separation methods Hydrogen is manufactured by removing other components or recovering it from gaseous mixtures generated in various chemical processes. The crude hydrogen-based gases produced from hydrocarbons by steam reforming or partial oxidation contain various co-products, byproducts and residual reactants such as carbon dioxide, water vapor, carbon monoxide, and methane. Moisture and traces of oxygen exist in hydrogen generated by water electrolysis. Unreacted or by-product hydrogen is obtained in a mixed or fairly pure state in hydrocracking, hydrotreating, and brine electrolysis processes. Hydrogen with a moderate or high purity is prepared from these hydrogen-rich gases through preliminary Separation and purification operations. The product hydrogen can be further purified to a specified level according to its intended application. A variety of methods to separate and purify hydrogen exist. The processes in the removal of impurities from crude hydrogen to obtain a pure product can be roughly divided into three steps. The first step is pretreatment of the crude hydrogen for the removal of specific contaminants that are detrimental to subsequent separation processes and for their conversion to easily separable species. Three methods of conventional adsorption, physical absorption and chemical reactions are effective for these purposes. The second step is the removal of both major and minor impurities to yield an acceptably pure hydrogen level. The prime separation technology here is the pressure swing adsorption (PSA) unit, which has several advantages over the other methods and is widely used in various fields of hydrogen separation. Physical absorption and polymer membrane processes are also applicable to hydrogen recovery from crude hydrogen mixed with hydrocarbons. In addition, hot alkaline absorption and conventional adsorption processes are available for removal of carbon dioxides and water vapor. Technical progress in the separation technology based on metal hydrides, especially in avoiding or minimizing deactivation and degradation, is still needed, but would be very useful for recovering hydrogen selectively from streams at low partial pressures and low concentrations. The third step is the final purification to a specified level. This is typically a cryogenic adsorption method at a liquid nitrogen temperature 66

or the use of a palladium membrane. Both are capable of reducing impurities to below 1 ppm [62]. In order to choose a feasible technique among the separation techniques for hydrogen recovery from the off-gas stream several factors have to be considered: 

Off-gas stream properties (composition, pressure, flow rate)



Recovery gas properties (purity, pressure)

The choice of a suitable separation process depends on the specifications and operating conditions of the feed and product gases. In the use of hydrogen as a combustion fuel, the separation processes at the first and second steps mostly suffice [63]. In terms of the scale of operation most, if not all, of the available techniques are capable of being operated over a wide range from small laboratory requirements through to large scale industrial production. However, practical and economic considerations impose restraints so that only two techniques are actually utilized over this whole range, namely Catalytic Purification and Polymer Membrane Diffusion. The two physical techniques, Cryogenic Separation and Pressure Swing Adsorption are best suited to large scale applications and the remaining techniques, including Palladium Alloy Membrane Diffusion, are utilized for small to medium outputs. There are two other aspects of importance to be considered in selecting a purification technique besides the scale of operation. The first is its ability to cope with a range of gas feedstock in terms of hydrogen content (that is whether rich or lean) and the second concerns the limitations imposed by the technique in terms of the chemical composition of the feedstock. These include both the selectivity of the technique against gases other than hydrogen and its resistance to “poisoning” by constituents present as impurities in the feed gas [64]. All techniques can operate well with hydrogen-rich gas feedstock. Catalytic Purification removes oxygen by reaction with hydrogen to form water, and carbon monoxide by oxidation or methanation; this technique is used to upgrade relatively pure hydrogen produced, for example, by electrolysis. Pressure Swing Adsorption also requires hydrogen-rich gas streams since it functions by selective adsorption of impurities. Cryogenic Separation can tolerate a wider range of hydrogen content in the feed gas, typically 30 to 80 per cent, but is limited in gas composition to those constituents that will selectively condense at cryogenic temperatures. Metal Hydride 67

Separation and the two diffusion techniques, Palladium Membrane and Polymer Membrane, have the ability to deal with feed gases lean in hydrogen. The Hydride Separation and Palladium Diffusion techniques are based on the very selective adsorption and diffusion of hydrogen, respectively, and the purity of the output hydrogen is not affected by the leanness of the feed gas. However, Polymer Membrane Diffusion is based on a differential diffusion rate principle and purity of the output hydrogen will be affected by the hydrogen concentration as well as by the nature of the other constituents that are present in the input gas stream. Many of the techniques are sensitive to “poisoning” by certain impurities in the gas feedstock, particularly those which rely on selective reactions (adsorption, diffusion); sulphur compounds and, in some cases, carbon dioxide are the chief poisons. The major poisons for each technique are listed in the Table 5.1[64]. Table 5. 1 Comparison of hydrogen purification techniques [64] Comparison of Hydrogen Purification Techniques Hydrogen output Technique

Cryogenic Separation

Pressure Swing Adsorption

Principle

Typical feed gas

90-98

95

Large scale

necessary to remove

Petrochemical and refinery

low temperatures

off-gases

CO 2 ,H 2 S and water

Any hydrogen

relatively low as

Selective adsorption of impurities from gas stream

rich gas

gases and

Membrane

through a permeable

ammonia

membrane

purge gas

hydrogen through a palladium alloy membrane

99.999

70-85

large

hydrogen is lost in the purging step

Refinery off-

Selective diffusion of

Prepurification step

The recovery is

diffusion of gases

Diffusion

Recovery

Comments

of gas mixtures at

Differential rate of

Membrane

Purity

Scale of use

Partial condensation

Polymer

Palladium

percent

He, CO 2 and water may 92-98

>85

medium

also permeate trough membrane

Sulphur-containing Any hydrogen containing

compounds and ≥99.9999

gas stream

Up to 99

medium

unsaturated hydrocarbon impair permeability

68

The level of recovery of hydrogen from feed gases varies considerably from one technique to another. Metal Hydride Separation, Pressure Swing Adsorption and Polymer Membrane Diffusion have relatively poor recovery levels, typically in the range 70 to approximately 85 per cent, while Cryogenic Separation and Solid Polymer Electrolyte techniques can attain recovery levels of about 95 per cent. Only Palladium Membrane Diffusion and Catalytic Purification techniques can achieve high recovery levels of up to 99 per cent from hydrogen-rich gases. Where impure or lean hydrogen feed gases are used the recovery levels are somewhat lower. As can be seen from the Table 5.1, many techniques have limitations with regard to the purity of hydrogen produced. At the lowest levels, typically low to mid 90 per cent, are the Cryogenic Separation and Polymer Membrane Diffusion methods. Metal Hydride Separation can achieve a 99 per cent purity level with the Solid Polymer Electrolyte technique attaining almost a factor of 10 better purity. Two methods are capable of producing moderately high purity levels of 99.999, namely Pressure Swing Adsorption and Catalytic Purification, but where very high purities of 99.9999 or better are required, only Palladium Membrane Diffusion comes into consideration. Economical aspects : The performance of the H 2 separation techniques were evaluated by a techno-economic analysis. Cost estimation results have been reported in Table 5.2 for these separation technologies. As expected, membrane technique has a low capital and operating costs. Cash position diagram has been shown in Figure 5.1 to calculate payback period. It can be concluded from this figure that membrane technology has the lowest payback period among the separation techniques and its usage is recommended as an economical way to separate hydrogen from the refinery off-gases [65].

69

Table 5. 2 comparison of H2 capture cost [65] process

Initial capital cost (million $)

Electrical cost ($/hr)

compression

8.2

546

PSA

3.1

---

Joule-Thomson

2.5

13

turbo expander

8.0

13

Plate fin heat exchanger

4.1

13

2.4

---

Cryogenic

Membrane

Figure 5. 1 Cash position versus time for different process [65]

The selection of the optimum purification technique for specific industrial applications must be based on both technical and economic considerations. As the Table 5.1 shows, the degree of purification of hydrogen obtained from the different methods varies from around 90 percent for the Cryogenic and Polymer Membrane techniques to 99.9999 per cent for Palladium Alloy Membrane diffusion. The amount of hydrogen recovered also varies considerably and can have a major impact on process economics, particularly for large scale applications. Furthermore table 70

5.2 indicates that membrane technology has the lowest payback period among the separation techniques and its usage is recommended to separate hydrogen from the refinery off -gases.

71

5.2 Comparison/overview of the different membranes As it has been proved that membrane technology is the best for hydrogen separation it is better to compare different membrane that are used in hydrogen separation. Table 5.3 presents the properties of the most important membranes that can be used for hydrogen separation processes. A practical classification can be based on their operating temperatures: 

For temperatures up to 100 °C only dense polymer membranes can be used.



For temperatures between 200 °C (or 300 °C) and 600 °C dense metallic or micro porous ceramic membranes could be used.



For temperatures between (500 °C or) 600 °C and 900 °C porous carbon and dense ceramic membranes are suitable.

Hydrogen permselective polymeric membranes are widely used for H2 recovery from refinery streams at low temperatures. Diffusion selectivity and solubility selectivity, along with permeance (the absolute magnitude of permeability), are the key factors governing performance of a polymeric membrane for H2 separation [66]. Diffusion selectivity favors smaller molecules and solubility selectivity favors larger molecules. In general, the selectivity of glassy polymers (i.e., polymers with glass transition temperatures above the operating temperature) is dominated by diffusion selectivity, which is governed primarily by the size difference between the gas molecules and the size sieving ability of the polymer material. Membranes made of glassy polymers are used for removing lighter gases like H2. The selectivity of rubbery polymers (i.e., polymers with glass transition temperature below the operating temperature) is dominated by solubility selectivity. These membranes are used for removing heavier gases from a mixture. Temperature affects these selectivities in different ways. For a given polymer, within its glassy or rubbery range, diffusion selectivity generally becomes more important as temperature increases, while the opposite is generally true for solubility selectivity, particularly for temperatures below room temperature. Freeman [66] showed in a typical H2/N2 vs. H2 permeability plot why glassy polymers are preferred over rubbery membrane materials for most H2 separation applications; where high selectivity is needed to meet permeate purity specifications. Important exceptions are the rubbery membranes (e.g., MTR VaporSep) that are used to recover H2 from refinery streams. Here the high permeability of the rubbery polymer membrane is more important than H2 purity in driving the system economics. 72

Table 5. 3 Properties of the relevant hydrogen selective membranes [27] Dense polymer Temperature range H2 selectivity H2 flux (10-3 mol/m2s) at dP=1 bar

Micro porous ceramic

Dense metallic

Porous carbon

<100 °C

200-600 °C

300-600 °C

500-900 °C

600-900 °C

low

5-139

>1000

4-20

>1000

low

60-300

60-300

10-200

6-80

Stability issues

Swelling, compaction, mechanical strength

Poisoning issues

HCl, SOx, (CO2)

Stability in H2O

Phase transition

Brittle, oxidising

Stability in CO2

H2S, HCl, CO

Strong adsorbing vapors, organics

H2S

Materials

Polymers

Silica, alumina, zirconia, titania, zeolites

Transport mechanism

Solution/ diffusion

Molecular sieving

Solution/ diffusion

Surface diffusion; molecular sieving

Commercial by Air Products, Linde, BOC, Air Liquid

Prototype tubular silica membranes available up to 90 cm. Other materials only small samples (cm2)

Commercial by Johnson Matthey; prototype membrane tubes available up to 60 cm

Small membrane modules commercial, mostly small samples (cm2) available for testing

Development status

Dense ceramic

Palladium alloy

Carbon

Proton conducting ceramics (mainly SrCeO3-δ, BaCeO3-δ) Solution/ diffusion (proton conduction)

Small samples available for testing

Information on polymeric membranes that selectively permeate H2 over CO2 is limited. Due to the high permeability of CO2, the selectivity of H2 in the presence of CO2 is typically low for organic polymers. Orme et al. showed that for a wide range of polymers the H2/CO2 selectivities varied between 0.5-2.5. The H2 permeabilities and selectivities for other polymeric membranes are shown in Tables 1.15 and 1.16 in [67]. For these polymers the H2/CO2 selectivity varied between 2 and 15. Of particular interest are the results provided by Hradil et al. [68] with 73

alumina supported styrene-divinylbenzene membranes. High permeabilities (i.e., 500-4000 Barrers) were reported for H2, but H2/CO2 selectivity data is lacking. Achieving high H2 permeability with high H2/CO2 selectivity remains an important technical challenge. However, a polymer membrane that selectively permeates both H2 and CO2 relative to CO could still be used to drive the unfavorable equilibrium of the water gas shift reaction, though downstream scrubbing of the CO2 might be required. Hydrogen selective membranes such as metallic and dense ceramics, as well as less selective porous inorganic and organic membranes, have been evaluated for commercial hydrogen separation. Because hydrogen is transported in dissociated form, both metallic and dense ceramic membranes can be 100% selective towards hydrogen. This particular ability allows for ultra-pure hydrogen, containing little (< 1 ppm) or no carbon oxides. However, no porous (Knudsen based) membrane has been able to meet hydrogen separation purity and economic requirements. Despite this, porous membranes can still be useful to drive the reaction. More work is needed to explore practical opportunities in this area. Among hydrogen selective membranes, Pd membranes remain the most promising. These Pd based membranes have limitations that have restricted commercial use. Key limitations include embrittlement, thin films that are free of cracks or pinholes (hillocks), delamination, and sulfur poisoning. Because of the presence of a two-phase region below the critical temperature, Pd membranes are not suitable for use at low temperatures (<573 K). Use of Pd membranes at low temperature causes hydrogen embrittlement [70]. To combat this problem, Pd is alloyed with other materials such as Cu, Ag, Ru, Y, Ni, Au, Ce, and Ta [70, 71]. Alloying not only reduces the critical temperature but also increases the operability of Pd alloy membranes. Thermal and chemical stability of Pd can also be improved by alloying. More importantly, alloying Pd with some metals increases the permeation rate of hydrogen, [72] such as Pd77Ag23 (alloy compositions are expressed by the metal symbol followed by weight percentage as a subscript), Pd90Y10, and Pd60Cu40. Pd77Ag23, for example, exhibits 1.7×more hydrogen flux than pure Pd membranes [73]. This is attributed to the higher diffusivity of atomic hydrogen in the alloys mentioned before. Hydrogen flux at different compositions of Ag, Y, and Cu with Pd can be found elsewhere [73]. Furthermore, hydrogen sulfide (H2S) and other feedstock have a poisoning effect on the

74

hydrogen transport mechanism in Pd. This can be reduced by forming binary alloys with Cu and Au [74]. Cost is the major barrier for the preparation of Pd membranes. Recent studies have been focused on thin metallic membranes. Thin membranes would reduce the cost of materials as well as increase the hydrogen flux. The thin membranes are prepared on supports. Two of the supports that were widely used are porous glass, such as porous Vycor (silica glass with symmetric structure and a mean average pore size of 4 nm), and porous ceramic R-alumina with asymmetric multilayer structure [75]. Porous ceramic and glass supports can be used because of their smooth surface. However, fitting ceramics to metal lacks the mechanical stability. In this regard, stainless steel could be used as a support material because of its mechanical durability, its thermal expansion coefficient close to that of Pd, and its ease of gas sealing [76]. As mentioned earlier, three methods (electroless-plating, CVD, and physical sputtering) have been developed to coat thin metal films on porous metallic or ceramic supports, and all of them produce good quality thin membranes with the high hydrogen-to-nitrogen selectivity [75]. Electrolessplating and CVD can be easily scaled up and prepared on supports of different geometry. However, with these methods, it is difficult to control the composition of the deposited metal alloy. Electrodeposition and microfabrication of Pd alloys show significantly higher hydrogen permeance and selectivity than CVD and electroless-plating (Table 5.4) [69]. According to Tong et al., [77] this higher permeance may be due to the thin Pd-Ag membrane with high composition control and the nanostructures. Table 5.4 presents the selected metallic membranes, construction techniques, supporting substrates, and performance data in terms of hydrogen permeance and selectivity. Membranes produced from the microfabrication technique seem to have the best permeance, whereas electrodeposition shows higher H2 selectivity over N2. Metals that were used to prepare the membranes are given in the parenthesis. Permeances are given for the comparison purpose; however, the values of the permeation rates (or permeances) of membranes should be used with caution. While comparing permeance values, the operating pressure should be noted. There are significant discrepancies in the permeation data, and this could not be simply explained by the membrane thickness, pressure differences between two sides of the membranes, and construction techniques. Perhaps crystallite size could be important, and some discussion on hydrogen permeation due to the effect of grain size can be found elsewhere [78].

75

Table 5. 4 Selected Hydrogen Separation Metallic Membranes and Their Performance [69]

method

permeance (10-6 mol m-2 s-1 Pa-1)

support

selectivity

CVD (Rh)

R- alumina

1.58 at 773 K

H2/N2 = 80

CVD (Ir)

R- alumina

1.81at 773 K

H2/N2 = 93

CVD (Pd)

γ- alumina

0.1-0.2 at 573 K

H2/He = 200-300

electroless (Pd)

titania/ceramic

6.3 at 773 K

H2/N2 = 1140

electroless (Pd)

alumina and stainless steel

0.22 at 623 K

H2/N2 = 110

electroless (Pd-Cu)

R-alumina/zirconia

0.023 at 723 K

H2/N2 = 1 150

electrodeposition (Pd-Ni)

s. steel

6.7 at 723 K

H2/N2 = 3 000

electrodeposition (Pd-Cu)

Ni-porous stainless steel

8.4 at 723 K

H2/N2 > 10 000

microfabrication (Pd-Ag)

silicon wafer

≈ 45 at 723 K

H2/N2 = 4 000

Metal alloys or composite metal membranes have been used for hydrogen purification, but they are sensitive to some gases such as CO or H2S. Therefore, a ceramic membrane, inert to poisonous gas, might be desirable [79,80]. Although dense inorganic membranes are currently being commercialized, inorganic microporous hydrogen membranes are yet to be commercialized for large industrial use. Inorganic microporous membranes (pore sizes <2 nm) offer many advantages over thin-film Pd membranes for the separation of hydrogen from a mixed-gas stream [69]. More importantly, in microporous membranes, the flux is directly proportional to the pressure, whereas in palladium membranes, it is proportional to the square root of the pressure. Therefore, microporous membranes become the more attractive option for systems that operate at increased pressures. An added feature of the microporous membranes is that their permeance increases dramatically with temperature. Consequently, they can operate at higher pressures and temperatures. Furthermore, microporous membranes cost less than thin-film Pd membranes because the fabrication does not require precious metals. However, microporous inorganic membranes are porous, and they can never produce 100% pure gas streams as can thinfilm palladium or ion-transport membranes. The membranes can be fabricated from a variety of materials such as ceramics, metals, etc. Ceramic membranes are chemically inert and can be used at high temperatures [81]. Ceramic membranes can be made by combining metals with nonmetals in the form of nitride, carbide, or oxide. They can be either dense or porous. Porous ceramic membranes generally have a two-layer structure: the separation membrane and a porous 76

ceramic supporting layer. Supporting material could be Vycor, alumina, or glass. Ceramic membranes can be prepared either by CVD, sol-gel, or electroless techniques. Table 5.5 presents the selected hydrogen selective ceramic membranes, their hydrogen permeability, and methods of preparation, as well as support materials. A considerable amount of work has been done with ceramic membranes, and especially sol-gel processing and CVD have been the methods of choice in most of the studies. It can be seen from the table 5.5 that sol-gel provides good hydrogen permeability and selectivity. However, it lacks reproducibility [82]. A membrane (called Nanosil) prepared by Prabhu and Oyama [82] exhibited significantly higher hydrogen selectivity with respect to CH4. Table 5. 5 Performance of H2 Selective Ceramic Membranes [69]

method

H2 permeance (10-8 mol m-2 s-1 Pa-1)

support

H2 selectivity

CVD (SiCl4 + H2O)

porous vycor

2.2 at 873 K

H2/N2 = 500-1000

CVD (SiH4 + O2)

porous vycor

1.5 at 873 K

H2/N2 = 2000-3000

CVD (TEOS)

alumina

4.0 at 873 K

H2/N2 = 100-1000

CVD (SiO2, TiO2, Al2O3, B2O3)

porous glass

0.71-2.2 at 723 K

H2/N2 = 1000-5000

CVD (TEOS, SiCl4 + O2)

porous glass

2.9 at 973 K

H2/N2 = 500-3000

high-temperature atmospheric CVD (TEOS)

modified Vycor (Nanosil)

1.8 at 873 K

H2/CH4 = 23000-27000

sol-gel and CVD (SiO2)

alumina

5.4 at 500 K

H2/N2 > 300

sol-gel (SiO2)

alumina

50 at 473 K

H2/CH4 = 5000

There are two types of carbon membranes based on the transport mechanisms: molecular sieving and surface diffusion membranes. Molecular sieving membranes are seen as promising, both in terms of separation properties as well as reasonable flux and stabilities, but are not yet commercially available at a sufficiently large scale [27]. Carbon molecular sieving (CMS) membranes can be prepared in two ways: (i) unsupported CMS membranes such as flat membranes, capillary tubes, or hollow fibers, and (ii) supported membranes on a macroporous material [83]. According to Wang and Hong, [83] the former might suffer from the problem of brittleness, and the latter is relatively difficult to prepare. Carbon membranes can be used in nonoxidizing environments with temperatures in the range of 773-1173 K. Carbon membranes are brittle and can be difficult to package if the membranes surfaces become large [27]. Furthermore, carbon membranes are still expensive. The performance of the membranes will 77

deteriorate severely if feed streams contain organic traces or other strongly adsorbing vapors such as H2S, NH3, or chlorofluorocarbons (CFCs). Table 5.3 illustrates the selected properties of the hydrogen separation membranes. The major properties listed are operating temperature range, hydrogen selectivity, hydrogen flux, stability, and poisoning issues. Details can be found elsewhere [27]. Although there might be some variations in the data for each type of membrane as seen table 5.3, table 5.4 and table 5.5 presents the general characteristics of different types of membranes. On the basis of the membrane’s operating temperature, polymer membranes are mainly suitable for temperatures <373 K, whereas porous carbon and dense ceramic membranes are suitable for a higher temperature range (773-1173 K). Metallic membranes would be suitable somewhere around 573873 K. It is reported that the selectivity of Pd for H2 separation was lost at temperatures >823 K because of the tin present at the interface of Pd and the support [75]. The dense-metallic and ceramic membranes are highly selective for H2 permeation. Different types of hydrogen separation membranes have been discussed. Currently, inorganic membranes have attracted wider interest than organic membranes. Inorganic membranes can tolerate harsher conditions than organic membranes and they offer advantages such as high flux and high temperature operation, and can be further divided into metallic (dense phase) and ceramic (porous and non-porous). Metallic Pd types theoretically offer infinite selectivity, but have issues such as precious metals costs and poor thermomechanical stability of the selective film. Since Pd is an expensive material, ceramic membranes are becoming more attractive. Membranes produced from microfabrication techniques performed best for permeance, whereas electrodeposition showed higher H2 selectivity over N2 in metallic membranes. Similarly, sol-gel provides good hydrogen permeability and selectivity in the case of ceramic membranes.

78

Conclusion Although there is certainly competition from other separation technologies, such as pressure swing adsorption (PSA) and cryogenic separation, membrane systems for gas separation have enjoyed growing interest in the last half century. Due to the low capital and operating costs and ease of operation, high quality H2 can be obtained by membrane technology to recover the hydrogen from the refinery off-gases with the typical throughput. There are several hydrogen selective membranes, each with their own operating ranges, in terms of temperatures and flow compositions. The properties of the feed flow to be separated are therefore a starting point to select a suitable membrane. For temperatures below 100 °C, only dense polymer membranes can be used; in the temperature range between 200 and 600 °C, dense metallic or micro porous ceramic membranes can be used; and in the temperature range between 600 °C and 900 °C, porous carbon and dense ceramic membranes are most suitable. Applicability of these membranes is also limited by sensitivity towards certain species and cost. Moreover, the development status of these membranes varies. Once the membrane type is selected, the membrane module can be chosen or designed taking into account considerations such as manufacturability, maintainability, operability, efficiency, membrane deterioration, and costs. Currently five module types exist, plate-and-frame and spiral-wound modules, based on flat membranes, and tubular, capillary and hollow-fiber modules, based on tubular membrane geometries. Generally, for gas separation the module price depends to a large extent on the specific application (feed pressure, temperature, etc.). Especially for high temperature and pressure applications a high packing density is desirable. This favors application of hollowfiber and to a lesser extent capillary and spiral wound modules. Modules are combined into stages. The options for membrane system layout are virtually endless, as stages can be combined in various ways, and processes can be changed by adding compressors and recycles. A design also needs to accommodate occasional cleaning/replacement and possibly carrying out of chemical reactions. If chemical reactions are carried out on a large scale, heat management issues may complicate membrane operations. Moreover catalysts may need to be incorporated into the membrane systems.

79

Reference [1] M.P. Ramage, et al., "The hydrogen economy: opportunities, costs, barriers and R&D needs," The National Academy Press, Washington, DC (2005). [2] Kirk and Othmer, “Concise Encyclopedia of Chemical Technology”, 4th ed., John Wiley & Sons (1999). [3] Editorial Staff, "Gas Processing 2002", Hydrocarbon Processing, May (2002). [4] Editorial Staff, "Petrochemical Processes 2003", Hydrocarbon Processing, March (2003). [5] R.A. Meyers, "Handbook of petroleum refining processes, 3rd ed.," McGraw-Hill, NY (2004). [6] J.L. Humphrey, G.E. Keller II, “Separation process technology”, McGraw-Hill, NY (1997). [7] Gas processors suppliers association, Engineering data book, Tulsa, oklahama (1994). [8] B.A. Hands, “Cryogenic Engineering”, Academic Press, (1986). [9] I. Uehara, “Hydrogen Separation and Handling”, in Energy Carriers and Conversion Systems, [Ed. Tokio Ohta], in Encyclopedia of Life Support Systems (EOLSS), Eolss Publishers, Oxford, UK, (2004). [10] M.S.A. Baksh, C.E. Terbot, "Pressure-swing adsorption process for the production of hydrogen", US Patent 6,503,299 B2 (2003). [11] G. Kolb, “Fuel Processing”, John Wiley & Sons, Germany (2008). [12] S.C.A. Kluiters, “Status review on membrane systems for hydrogen separation”, Intermediate report EU project MIGREYD NNE5, (2001). [13] R.W. Baker, “Membrane technology and applications”, 2nd Ed., John Wiley & Sons, England (2004). [14] H. Bechhold, “Kolloidstudien mit der Filtrationsmethode”, Z. Physik Chem. 60, (1907), 257. [15] W.J. Elford, “Principles Governing the Preparation of Membranes Having Graded Porosities. The Properties of ‘Gradocol’ Membranes as Ultrafilters”, Trans. Faraday Soc. 33, (1937), 1094. [16] R. Zsigmondy and W. Bachmann, “Űber Neue Filter”, Z. Anorg. Chem. 103, (1918), 119. [17] J.D. Ferry, “Ultrafilter Membranes and Ultrafiltration”, Chem. Rev. 18, (1936), 373. [18] R.D. Noble, S.A. Stern, “Membrane Separations Technology - Principles and Applications”, Elsevier Science B.V., Amsterdam (1995). 80

[19] J.M.S. Henis and M.K. Tripodi, “A Novel Approach to Gas Separations Using Composite Hollow Fiber Membranes”, Sep. Sci. Technol. 15, (1980), 1059. [20] M. Mulder, “Basic Principles of Membrane Technology”, Kluwer Academic Publishers, Dordrecht (2000). [21] J.B. Hunter, Ultrapure Hydrogen by Diffusion through Palladium Alloys, Disv. Pet. Chem. Prepr. 8, (1963), 4. [22] D.R. Paul, Y.P. Yampol'skii, ”Polymeric Gas Separation Membranes”, CRC Press Inc., Boca Raton (1994). [24] J.S. Ribberink, “Status Review on Membranes”, ECN-C--93-063, ECN, Petten, December (1993). [25] J. Shu, B.P.A Grandjean, A.V. Neste, S. Kaliaguine, “Catalytic Palladium-based Membrane Reactors: A review”, The Canadian Journal of Chemical Engineering, Canadian Society for Chemical Engineering, vol. 69, Ottawa (1991). [26] Y.C.V. Delft, J.A. Hugill, E.M.V Dorst, H.M.V Veen, “Hydrogen Preparation in Process Industry with Inorganic Membranes”, ECN-CX-03-110, ECN, Petten, (2003). [27] S. C. A. Kluiters, “Status review on membrane systems for hydrogen separation”, Energy Center of the Netherlands: NNE5-2001-670 Petten, The Netherlands (2004). [28] B. D. Freeman and I. Pinnau, “ Gas and liquid separations using membranes: An overview”, In Advanced materials for membrane separations; ACS Symposium Series 876, American Chemical Society, Washington, DC, (2004), 1-21,. [29] M. Ulbricht, “Advanced functional polymer membranes”, Polymer 47, (2006), 2217–2262 [30] U. Merten and P.K. Gantzel, “Method and Apparatus for Gas Separation by Diffusion”, US Patent 3,415,038, (1968). [31] L.M. Robeson, “Correlation of separation factor versus permeability for polymeric membranes”, J. Membrane Science, 62, (1991), 165–85. [32]L.M. Robeson, W.F. Burgoyne, M. Langsam, A.C. Savoca, C.F. Tien, “High performance polymers for membrane separation”, Polymer, 35, (1994), 4970–8. [33]B.D. Freeman, “Basis of permeability / selectivity tradeoff relations in polymeric gas separation membranes”, Macromolecules, 32, (1999), 375–80. [34] G. Alefeld and J. Völkl, “Hydrogen in Metals I: Basic Properties”, Springer- Verlag, Germany, (1978). 81

[35] J.B. Hunter, “Silver-Palladium Film for Separation and Purification of Hydrogen”, US Patent 2,773,561, (1956). [36] J.B. Hunter, “Ultrapure Hydrogen by Diffusion through Palladium Alloys”, Disv. Pet. Chem. Prepr. 8, (1963), 4. [37] R.B. McBride and D.L. McKinley,” A New Hydrogen Recovery Route”, Chem. Eng. Prog. 61, (1965), 81. [38] J.E. Philpott, Hydrogen Diffusion Technology, Commercial Applications of Palladium Membrane, Platinum Metals Rev. 29, (1985), 12. [39] D. J. Edlund, J. M. McCarthy, “The relationship between intermetallic diffusion and flux decline in composite-metal membranes: implications for achieving long membrane lifetime”, J. Membrane Science, 107, (1995), 147-153. [40] S. Uemiya, “Brief review of steam reforming using a metal membrane reactor”, Top. Catal, 29, (2004), 79-84. [41] R. Hughes, “Composite palladium membranes for catalytic membrane reactors”, Membr. Technol., 131, (2001), 9-13. [42] J. E. Koresh and A. Soffer, “Molecular sieve carbon membranes for gas separation”, Sep. Sci. Technol. 18, (1983), 723. [43] J. E. Koresh and A. Soffer, “Mechanism of permeation through molecular-sieve carbon membrane”, J. Chem. Soc. Faraday Trans. I82, (1986), 2057. [44] J. Koresh, “Carbon molecular sieve membranes: General properties and the permeability of CH4/H2 mixtures”, Sep. Sci. Technol. 22, (1987), 973. [45] Y.H. MA, “Hydrogen Separation Membranes”, Advanced Membrane Technology and Applications, Wiley, (2008), 671. [46] S. Sircar and M. B. Rao, “Nano-porous carbon membranes for gas separation”, In N. Kanellopoulos (Ed.), Recent Advances on Gas Separation by Microporous Membranes, Elsevier, Amsterdam, The Netherlands, (2000), 473. [47] J. Caro, M. Noack, P. Kolsch and R. Schäfer, “Zeolite Membranes: State of Their Development and Perspective”, Microporous Mesoporous Mater. 38, (2000), 3. [48] M. M. Shah, R. F. Drnevich and U. Balachandran, “Integrated Ceramic Membrane System for Hydrogen Production”, Proceedings of the 2000 Hydrogen Program Review, NREL/CP-57028890, (2000). 82

[49] H.J.C. te Hennepe, D. Bargeman, M.H.V. Mulder and C.A. Smolders, “Zeolite-filled Silicone Rubber Membranes Part I: Membrane Preparation and Pervaporation Results”, J. Membrane Science, 35, (1987), 39. [50] J.-M. Duval, B. Folkers, M.H.V. Mulder, G. Desgrandchamps and C.A. Smolders, “Adsorbent Filled Membranes for Gas Separation”, J. Membrane Science, 80, (1993), 183. [51] M.-D. Jia, K.V. Peinemann and R.-D. Behling, Preparation and Characterization of Thinfilm Zeolite-PDMS Composite Membranes, J. Membrane Science, 73, (1992), 119. [52] J.C. Maxwell, “Treatise on Electricity and Magnetism”, Vol. I, Oxford University Press, London (1873). [53] L.M. Robeson, A. Noshay, M. Matzner and C.N. Merian, “Physical Property Characteristics of Polysulfone/poly(dimethyl siloxane) Block Copolymers”, Angew. Makromol. Chem. 29, (1973), 47. [54] S. Nunes, K.V. Peinemann, “Membrane technology in the chemical industry”, Weinheim, Wiley–VCH (2001). [55] R.P. Lanza, R. Langer, J. Vacanti, “Principles of tissue engineering”, 2nd ed. San Diego, CA, Academic Press; (2000). [56] R.W. Baker and J.G. Wijmans, “Membrane Separation of Organic Vapors from Gas Streams”, in Polymeric Gas Separation Membranes, CRC Press, Boca Raton, FL, (1994), 353– 398. [57] R.W. Baker and J.G. Wijmans, “Two-stage Membrane Process and Apparatus”, US Patents 5,256,295 and 5,256,296, (1993). [58] D.L. MacLean, W.A. Bollinger, D.E. King and R.S. Narayan, “Gas Separation Design with Membranes, in Recent Developments in Separation Science”, CRC Press, Boca Raton, FL, (1986), 9. [59] W.A. Bollinger, D.L. MacLean and R.S. Narayan, “Separation Systems for Oil Refining and Production”, Chem. Eng. Prog. 78, (1982), 27. [60] J.M.S. Henis, “Commercial and Practical Aspects of Gas Separation Membranes, in Polymeric Gas Separation Membranes”, CRC Press, Boca Raton, FL, (1994), 441–530. [61] 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”, Journal of Colloid and Interface Science, 314, (2007), 589–603. 83

[62] K. Lui, C. Song and V. Subramani, “Hydrogen and Syngas Production and Purification Technologies”, John Wiley & Sons, USA (2010). [63] E.A. Maciula, “Membrane processing favors hydrogen recovery”, Hydro. Proc., (1980), 115-118. [64] G. J. Grashoff, C. E. Pilkington and C. W. Corti, “The Purification of Hydrogen. a review of the technology emphasizing the current status of palladium membrane diffusion”, Platinum Metals Rev., 27, (1983), 157-169. [65] S. Faraji, R. Sotudeh-Gharebagh and N. Mostoufi, “Hydrogen Recovery from Refinery Off-gases”, Journal of Applied Sciences, 5, (2005), 459-464. [66] B.D. Freeman, "Basis of Permeability/Selectivity Tradeoff Relations in Polymeric Gas Separation Membranes", Macromolecules, 32, (1999), 375-380. [67] “Separation Technology R&D Needs for Hydrogen Production in the Chemical and Petrochemical Industries”, Prepared for U.S. Department of Energy, Chemical Industry Vision2020 Technology Partnership, December, (2005). [68] J. Hradil, V. Krystl, P. Hrabanek, B. Bernauer and M. Kocirik, "Heterogeneous Membranes Based On Polymeric Adsorbents for Separation of Small Molecules", Reactive & Functional Polymers, 61, (2004), 303-313. [69] S. Adhikari and S. Fernando, “Hydrogen Membrane Separation Techniques”, Ind. Eng. Chem. Res. 45, (2006), 875-881. [70] F. Roa, M.J. Block and J.D. Way, “The influence of alloy composition on the H2 flux of composite Pd-Cu mmbranes”, Desalination, 147, (2002), 411-416. [71] R. Hughes, “Composite palladium membranes for catalytic membrane reactors”, Membr. Technol., 131, (2001), 9-13. [72] T. Ishihara, A. Kawahara, A. Fukunaga, H. Nishiguchi, H. Shinkai, M. Miyaki and Y. Takita, “CH4 decomposition with a Pd-Ag hydrogenpermeating membrane reactor for hydrogen production at decreased temperature”, Ind. Eng. Chem. Res., 41, (2002), 3365-3339. [73] F. Roa and J.D. Way, “Influence of alloy composition and membrane fabrication on the pressure dependence of the hydrogen flux of palladiumcopper membranes”, Ind. Eng. Chem. Res., 42, (2003), 5827-5835. [74] F.Roa, D.J. Way and S.N. Paglieri, “Process for preparing palladium alloy composite membranes for use in hydrogen separation”, U.S. Patent 2,003,190,486, (2003). 84

[75] S. Uemiya, “State-of-the-art of supported metal membranes for gas separation”, Sep. Purif. Methods, 28, (1999), 51-85. [76] P. F. Aparicio and M.J. Benito, “High performance membrane reactor system for hydrogen production from methane. Ind. Eng. Chem. Res., 44, (2005), 742-748. [77] H. D. Tong, F.C. Gielens, J.G.E. Gardeniers, H.V. Jansen, C.J.M.V. Rijn, M.C. Elwenspoek, W. Nijdam,

“Microfabricated palladium-silver alloy membranes and their

application in hydrogen separation”, Ind. Eng. Chem. Res., 43, (2004), 4182-4187. [78] Y. S. Lin, “Microporous and dense inorganic membranes: Current status and prospective”, Sep. Purif. Technol., 25, (1995), 39-55. [79] M.W. Lee, “Separation membrane development for hydrogen”, Westinghouse Savannah River Company, Aiken, SC (2000). [80] D.H. Edlund, “ Hydrogen separation and purification using dense metallic membranes”, DOE Hydrogen Separation and Purification Technologies: Arlington, VA (2004). [81] R. W. Baker, E.L. Cussler, W. Eykamp, W.J. Koros, R.L. Riley, H. Strathmann, “Membrane separation systems: Recent developments and future directions”, Noyes Data Corporation, Park Ridge, NJ (1991). [82] A.K. Prabhu, S.T. Oyama, “Highly hydrogen selective ceramic membranes: Application to the transformation of greenhouse gases”, J. Membr. Sci. 176, (2000), 233-248. [83] L.J. Wang, F.C.N. Hong, “Effects of surface treatments and annealing on carbon-based molecular sieve membranes for gas separation”, Appl. Surf. Sci. 240, (2005), 161-174.

85