Printable Version

Maximizing a Potentially Significant Energy Source

SwRI researchers develop ultra-thin metal membranes for hydrogen gas separation

By James Arps, Ph.D., Bruce Lanning, Ph.D. and Geoffrey Dearnaley, Ph.D.

Dr. James Arps (center) is manager of the Surface Engineering Section in SwRI’s Mechanical and Materials Engineering Division. Arps specializes in surface analysis, vacuum coating technology and material modification using energetic ions. Dr. Geoffrey Dearnaley (right) is a retired SwRI Institute scientist who remains a consultant to SwRI in the area of coatings and surface modification technologies. Dr. Bruce Lanning (left), formerly of SwRI, is now at ITN Energy Systems in Littleton, Colo.

Coal gasification and fuel cells are two of our nation’s most promising technologies for the efficient production of clean electrical power. For example, fuel cells have the potential to reduce dependence on oil for propulsion of cars and trucks by using hydrogen derived from plentiful U.S. coal supplies. At the heart of both of these technologies is hydrogen, but the ability to produce it in a pure form has posed problems for researchers. Hydrogen is usually obtained as a mixture with other gases, and the cost of separating it from them can be unacceptably high.

An affordable, tough and selective hydrogen-separating membrane could significantly reduce this cost, especially if it can be integrated into the hydrogen generation system because local generation of hydrogen reduces the need for its storage and distribution. It is in this area that a team of Southwest Research Institute scientists, in collaboration with scientists at the Colorado School of Mines and at IdaTech LLC, a leading manufacturer of self-contained fuel cell and gas separation systems, is developing an ultra-thin palladium alloy membrane for hydrogen gas separation. This three-year, $1 million development effort, funded by the U.S. Department of Energy, began in September 2003.

SwRI researchers are fabricating ultra-thin palladium alloy membranes for hydrogen separation.

Membrane Types

Using membranes to separate substances is common practice. A coffee filter, for example, is a paper membrane that separates the brewed liquid from the grounds. More challenging to achieve, however, is a membrane that is permeable to hydrogen — a small molecule — but that strongly holds back other gases. It must also allow a high and stable rate of permeation.

Polymer membranes are economical for some applications, but they cannot tolerate high temperatures. Hydrogen produced at chemical plants is often at a high temperature, so a costly and bulky cooling system would be needed.

Microporous ceramic membranes, which are more heat-resistant, have been developed. However, in manufacturing these membranes it is difficult to control the extremely fine pore size that is required over large areas. Also, ceramics are brittle and not easily joined to other materials to form a gas-tight seal.

Metal membranes have the advantages of being tough and able to withstand high temperatures, and they also have superior joinability. Very few metals can dissolve and be permeable to hydrogen. However, outstanding in this respect is the noble metal, palladium.

Palladium as a Membrane Material

To pass through a dense, metal membrane, hydrogen must undergo a complex series of reaction steps, some of which must be catalyzed to proceed rapidly. Hydrogen must first be adsorbed on the membrane surface, then dissociated into its atomic form. It is this atomic hydrogen that can diffuse rapidly through the membrane to the other side. There, the atomic hydrogen must recombine into hydrogen molecules and be released or desorbed. These processes are influenced by the membrane’s surface topography, which governs the true area of the metal, and also by the microstructure of grains within the membrane.

After repeated hydrogen adsorption and desorption, palladium becomes somewhat brittle, so it must be toughened, typically by being alloyed with silver, copper or ruthenium. Adding copper to the alloy improves tolerance to sulfur impurities that are often present after coal gasification. Palladium is expensive, and alloying it with copper reduces the cost. There is obviously a limit, however, to how thick a membrane can be if it is to be affordable. An estimated 5 microns is the maximum thickness allowable for a fuel-cell membrane used for the reforming of methanol. Thinner membranes allow more rapid permeation. The economics improve substantially if the palladium can be recycled, and for this it is best if the membrane is free-standing rather than in a composite structure with other metals or ceramics.

In the past, thin metal membranes have been fabricated by rolling between precision rollers, but the potential for pinhole defects limits this method for palladium-alloy membranes to 25 microns (0.001 inch). Difficulty in controlling deformation across the length of the press rolls used in forming the membranes limits the practical width of the membranes to approximately 10 cm. Other methods, such as traditional “thick film” coating techniques, have been used to fabricate self-supporting membranes. Coating methods, such as electroplating, have been demonstrated, but these have significant concerns with contamination from organic carbon and the ability to keep a controlled and consistent bath chemistry over multiple cycles and large areas.

Even though palladium usage has increased for self-supporting membranes — membranes capable of standing alone without a support structure made of another material — further reductions in thickness are needed to make this technology economically viable. Hence, a method to fabricate thinner, large-area palladium-alloy membranes in a continuous or even semi-continuous manner would represent a significant breakthrough in the development and commercialization of hydrogen purifiers for fuel cells and coal gasification.

A large area vacuum coating system is used to deposit coatings onto hard surfaces, such as oxidized silicon. SwRI scientists chose silicon because of its excellent release properties for deposited palladium.

Continuous Production of Self-Supported Foils by Vacuum Deposition

Vacuum deposition is used in a wide variety of industries, including semiconductors, machine tools, razor blade manufacture and packaging. Typically, the required material is vaporized in vacuum and allowed to condense on the items to be coated. In another method, called “sputtering,” a directed flux of ions eject material, without the need for heating, by atomic collisions occurring at or near the surface. This is the preferred method for deposition of alloys or mixtures of controllable composition. At a manufacturing level, vacuum-based processes can be cost-effective.

A web coater, or roll coater, is designed to treat material in a thin film form, and it is the most efficient means for treating large areas. In the case of web coaters, flexible polymer and metallic substrates have been coated with a variety of metal and ceramic materials for use in capacitors, magnetic media, thin film batteries and food packaging. In a vacuum web coating system, materials are typically condensed out of a physical vapor by either thermal evaporation or magnetron sputtering processes onto a continuous moving web such as plastic film.

SwRI scientists have been able to produce free-standing, ultra-thin palladium alloy membranes using vacuum deposition methods in combination with a proprietary sequence of treatments developed at SwRI that allows the deposited membrane to be detached or lifted off with greater ease. The approach consists of vapor-depositing palladium alloys onto a temporary, or “sacrificial,” substrate material and then detaching or releasing the film from the flexible substrate. A key to the reproducibility of the method is the relative flatness and surface energy of this temporary substrate. Using this method, palladium alloy films have been deposited and released from plastics such as polystyrene and polyvinyl alcohol, and also from aluminum foil.

Semi-Continuous or Batch Production

The semiconductor industry has developed highly efficient methods for vacuum-processing silicon or other “wafers” of up to 12 inches in diameter. Oxidized silicon has excellent release properties for deposited palladium, and a suitably engineered metal film, even as thin as 1 to 3 microns, can be pulled away very reliably to obtain a free-standing film that is remarkably robust. Less expensive alternatives to silicon, such as polished quartz or soda-lime glass may be used. However, some of SwRI’s best-performing membranes so far have been produced on silicon substrates.

Membrane performance is also a function of the quality of the film and, in particular, the presence of defects in the film. As the SwRI team attempts to produce ultra-thin membranes over larger areas, surface contaminants and particulates, even in the sub-micron range, become problematic. Efforts are under way to reduce defects by optimizing vacuum processing parameters, especially prior to and during deposition. At present, the SwRI team can produce 12-inch-diameter palladium-copper foils, 3 microns thick, without significant through-thickness defects or pinholes. The team will use this process knowledge to work on larger areas and to use the flexible polymer substrates referred to earlier.

SwRI scientists use a vacuum roll coating system to coat flexible materials hundreds of feet in length with metals or ceramics.

Hydrogen Permeation Test Results

Permeability of hydrogen through a metal is a function of both diffusivity and solubility. Diffusion takes place because of a concentration gradient, or the difference in partial pressure across the metal. In most cases, diffusion in the bulk is the rate controlling process for permeation through a metal membrane. That is, the migration of hydrogen through the membrane is the slowest of the transfer processes, another reason for preferring thin membranes. Hydrogen concentration, or solubility of atomic hydrogen in the lattice (bulk), is proportional to the square root of hydrogen pressure. This power dependency of hydrogen, where the exponent n = 0.5, can be used to assess the nature of the separation process.

As an example, results for a 10-micron-thick palladium-copper membrane showed that when the hydrogen flux was plotted against pressure, the value of n was about 0.59. This indicated that hydrogen transport was not entirely governed by diffusion of dissolved gas, but may have been affected by a surface contaminant such as carbon. The hydrogen to helium separation factor was determined to be more than 100,000 indicating an excellent hydrogen selectivity.

In another case, SwRI scientists used a proprietary heat treatment to heat the membrane to around 400 degrees centigrade, which increased permeation dramatically. When normalized to the specifications of the DOE, Office of Fossil Energy, the 10-micron alloy membranes achieved 155 standard cubic feet per hour per square foot (scf/h/ft2), well exceeding the DOE 2007 target of 100 scf/h/ft2. A thinner, 6-micron-thick membrane after similar heat treatment achieved 187 scf/h/ft2, approaching DOE’s target for 2010.

The effect of the treatment is primarily due to surface restructuring, which increases the surface area of the membrane. This was demonstrated using SwRI’s atomic force microscope. The surface “hillocks” are 2 to 3 microns in size.

Current Work: Alloy Composition

The membranes tested so far have had palladium contents between 61.5 percent and 63 percent, which is not ideal. There is a steep rise in hydrogen permeability at a composition of 60 ± 0.2 percent palladium. This preferred composition has now been achieved by adjusting the makeup of the sputtering target. Ultra-thin films, only 3 microns thick, are now being evaluated at IdaTech LLC and at the Colorado School of Mines.

Future Prospects

In this on-going project, SwRI researchers plan to improve the performance of the membrane even further, well beyond present goals, by engineering the through-thickness composition of the material; that is, by functionalizing it. The unique features of vacuum deposition may be used to tailor the optimum surface catalytic properties by incorporating specific atoms such as ruthenium to enhance hydrogen molecule dissociation, and the optimal diffusion characteristics, by adding elements that increase the hydrogen solubility. Vacuum processing enables such atomic-scale control in a way that is clearly impossible with conventional foil fabrication techniques such as cold rolling. Membranes are now being made in various sizes to be fitted into systems for real-world applications.

Through this project and others, SwRI researchers are helping industry meet the growing energy needs of our country.

Comments about this article? Contact Arps at (210) 522-6588 or


The authors acknowledge the support of SwRI employees Research Engineer Craig Engel and Senior Technician Douglas Czaja, both of SwRI’s Materials Engineering Department, and of Professor Doug Way and Omar Ishteiwy from the Colorado School of Mines. This project has been supported by the U.S. Department of Energy’s National Energy Technology Laboratory in Pittsburgh, Pa., Dr. Arun Bose, technical advisor.

Published in the Spring 2006 issue of Technology Today®, published by Southwest Research Institute. For more information, contact Joe Fohn.

Spring 2006 Technology Today
SwRI Publications SwRI Home