Since Dr. Nam Suh and his colleagues patented a technique for producing microcellular foam in 1987, there has been an international flurry of activity in this area. This technical note reviews some of the boundaries of the current and expected technologies.

Microcellular Foam – A Definition

In a recent publication [J.L. Throne, Thermoplastic Foams,Sherwood Publishers, Hinckley OH, 1996, Chapter 11, “Newer Foaming Technologies”], foams were classified in terms of their nominal dimension across the cell, BET surface areas, and number of cells per unit volume, as:

Foam Type Cell Size BET surface area [m2/g] cells/cm3
Commercial Foam 100 – 500 0.1 – 1 105
Crosslinked Polyolefin Foam 20 -100 10 10 6
Microcellular Foam 1 -10 10 – 20 107 – 109
Ultramicrocellular Foam 0.1 100 – 400 109 – 1012

In that same publication, several methods for manufacturing microcellular foams [MC] and ultramicrocellular foams [UMC] were detailed. Some of these methods are detailed below.


Crosslinked polyolefin foam is produced commercially worldwide. Electron beam and chemical peroxides are the two techniques used for crosslinking. Typically, for electron beam irradiation, a an extruded polymer sheet containing the chemical foaming agent is subjected to intense electron beam energy which serves to dehydrogenate the polymer, an action that results in crosslinking. The polymer is heated further to activate the foaming agent. For chemically crosslinked polyolefins, the crosslinking agent, usually a peroxide, is compounded into the polymer along with the foaming agent. The extruded polymer sheet temperature is sufficient to crosslink the polymer but not activate the foaming agent. After most of the crosslinking is completed, the polymer sheet is heated further to activate the foaming agents. Crosslinking density is about every thousand carbons. As a result, the polymer, such as LLDPE or PP, has sufficient molecular mobility to move under blowing pressure of the decomposing foaming agent, but sufficient crosslinking to prevent extensive bubble coalescence. The result is a very fine celled foam.

Phase Inversion

There are two approaches to phase inversion technology. In the first, extremely fine particles of polymer are dispersed in highly agitated non-solvent that has been heated above the melting or softening temperature of the polymer. As more and more particles are added, the polymer forms gossamer tendrils, then three-dimensional networks. As the oil mixture is cooled, the oil forms into discrete droplets inside the three-dimensional polymeric network. An alternative to this employs quiescent hot oil with a specific gravity less than that of the molten polymer. The polymer particles are added at a rate that allows them to form first into molten globules, then settle through the oil to contact previously added globules. Once the proper dosage has been reached, the oil mixture is slowly cooled and the oil forms into discrete droplets inside the interconnected globules. In both cases, the foam is made without foaming agents.

Aerogel Process

This process, developed originally with inorganics such as sodium silicate, relies on supercritical vitrification of a solid. The technique has recently been used with crosslinking polymers. Basically a low molecular weight expanding agent such as methanol for sodium silicate or carbon dioxide for resorcinol or epoxy, is the solvent or dispersing agent for the polymer or inorganic glass. A solution or emulsion is created at room temperature, then pressure is applied to a level greater than the critical pressure of the expanding agent. The temperature is then increased to above the critical temperature of the expanding agent, to a point where vitrification, crosslinking, or polymerization occurs. The pressure is slowly lowered to allow the expanding agent to diffuse from the rigidified structure. This supercritical drying step is critical, since if the pressure is lowered too quickly, the internal gas pressure will rupture the solid structure. Once the expanding agent has fully diffused from the rigidified structure, the temperature is lowered. As an aside, this technology has been used for decades to produce dried, powdered coffee. A patented variation of this technology involves solvent-non-solvent interchange at elevated pressure. Basically, the initial expanding agent is a solvent or theta-solvent for the polymer. Once expansion and vitrification has taken place, the first expanding agent is replaced, by molecular diffusion, with a second having much poorer solubility than the first. This expanding agent replacement can involve as many as four solvent-non-solvent interchanges.

These are just a few of the ways of producing MC and UMC foams. In addition, the technique of sintering micron-sized polymer powder particles yields a very fine open celled foam. Spinodal and binodal decomposition of a gas-laden polymer-polymer system yield sub-micron “voids” at the phase-separated interfaces, and so on.

The MIT Process and Variants

There appears to be two approaches to what is now termed “The MIT Process”. The first, called the batch process, is the technology espoused by Dr. Vipin Kumar in Seattle. The second, called the continuous process, is the technology promoted by Axiomatics/MuCell/Trexel.

The Batch Process

Very simply, a roll of extruded polymer is placed, at room temperature, in a pressure vessel. Nitrogen, carbon dioxide, or any other physical foaming agent, is added to the pressure vessel and the pressure vessel is pressurized to a level sufficient to force an appropriate amount of the foaming agent into the free volume of the polymer. The amount of gas dissolved in the polymer is directly proportional to the applied pressure, according to Henry’s law. Increasing the temperature on the roll of polymer increases the rate of diffusion but decreases the amount of gas that can be dissolved at a given pressure. Gas solubility in the crystal structure is substantially less than that in the amorphous region of any given polymer. As a result, the batch process seems to lend itself best to foaming of amorphous polymers. Once the gas has thoroughly saturated the polymer, the polymer roll is removed from the pressure vessel and is quickly heated to a softened or molten state. Keep in mind that the melting temperature or glass transition temperature of a polymer can be dramatically depressed by solution of small molecules into the polymer. Typical cell dimensions are less than 10 microns for polystyrene. There are a couple of advantages to the batch process. First, gas is dissolved by molecular diffusion, thus ensuring very small initial nuclei for bubble formation. Then, bubble coalescence is controlled by the way in which the sheet is heated. And, an integral skin-foam structure akin to a structural foam is relatively easy to achieve, simply by letting the sheet out-gas for some time prior to heating. The primary disadvantages are that it takes hours to diffuse gas into the sheet, but the sheet must be foamed in seconds to minutes after removal from the pressure vessel. The disparity in times leads to many, many pressure vessels for each foam expansion line.

The Continuous Process

It is apparent that the better alternative to MC foaming would be a continuous process. To this end, Axiomatics/MuCell/Trexel is developing and licensing a continuous foam process in which the foaming gas, again nitrogen, carbon dioxide or some other physical foaming gas, is dissolved in the polymer at a pressure greater than the solubility pressure given by Henry’s law and at a pressure greater than the critical pressure of the foaming gas. In one method, the pressure of the gas-laden melt is then dropped very rapidly by flow through a long-land orifice or slit. In another, a static mixer is used in place of the orifice or slit. And in a third, a gear pump, running backwards, is used in place of the orifice or slit. The resulting foamed structure is then shaped in a second, in-line die to produce the desired sheet or rod. The current continuous long-land orifice or slit die process is reminiscent of the Dow Frostwood and Sekesui Woodlike processes, wherein a perforated spinneret plate is placed at right angles to the flow of the gas-laden melt [F.A. Shutov, Integral/Structural Polymer Foams: Technology, Properties and Applications, Springer-Verlag, Berlin, 1986, pp. 118-119] . The gas-laden melt flows through the plated as jets or filaments that foam separately, then fuse into a structure that can then be shaped in a secondary shaping die. According to MIT literature, the key to MC foaming is the very high pressure used to propel the gas-laden melt through the nucleating and bubble growth region. With proper handling of the extrudate, bubble coalescence is limited to 50X or so. The resulting mostly closed-cell foam has nominal cell dimensions of 10 to 20 microns. The continuous nature of the process is a decided advantage over the batch process. However, since the foam cell structure is relatively uniform throughout the thickness, the foam does not have the high gloss that can be achieved in the batch process. And it appears in most samples that the cell structure in the continuous process is much larger than that in the batch process. This is undoubtedly due to the very high bubble coalescence and collapse in the high shear region of the orifice or slit. It was noted sometime ago [J.L. Throne and R.C. Progelhof, “Observations on the Transfer Rates of Plastics Containing Gases”, Polym. Proc. Engng,1 (1983), pp. 231-280] that gas-laden melts could be transferred at extreme rates with very little shear at the interface between the flowing polymer and the confining metal surfaces. The reason for this was believed to be a gas layer at the metal surface which acts like a gas bearing. It would seem that the problem of shear-induced coalescence would benefit through the deliberate addition of a low-viscosity lubricant at the entrances to both the high pressure-drop capillary die and the shaping die. In fact, a US patent describes high-speed pelletizing of molten polymer achieved through the deliberate addition of gas through porous metal rings at the entrance to the spinneret plate of a pelletizer. And high-speed extrusion of extremely high viscosity polymers such as UHMWPE and PTFE has been accomplished by bleeding in low-viscosity silicone lubricants at the entrance to the extrusion die.

Properties of MC Foams

As stated elsewhere [J.L. Throne, Thermoplastic Foams, Sherwood Publishers, Hinckley OH, 1996, Chapter 9, “Mechanical Design of Foams”], there are three levels of foam mechanical response to applied load. Short term response is typified by impact resistance. Long term response is typified by fatigue and creep. And moderate term response is typified by modulus and strength.

It is hypothesized that bubbles in an otherwise monolithic polymer structure act as stress concentrators or stress risers, in much the same way as inorganic fillers and organic occlusions. According to the Griffith crack hypothesis, mechanical failure typically occurs at these inhomogeneities. As the occlusions are made smaller and smaller, the stress concentration at the occlusion becomes less and less and as a result, the structure can bear greater and greater loads before failing. In short, if the cell sizes in the foam become very small, certain mechanical properties should improve. This section examines which properties are expected to improve.

The modulus of a material is determined by the slope of the stress-strain curve at zero strain. It has been demonstrated theoretically [L.J. Gibson and M.F. Ashby, Cellular Solids: Structure and Properties, Pergamon Press, 1988, pp. 130-131] and experimentally [A.M. Kraynik and W.E. Warren, “The Elastic Behavior of Low-Density Cellular Plastics”, in N.C. Hilyard and A. Cunningham, Eds., Low Density Cellular Plastics: Physical Basis of Behaviour, Chapman & Hall, London, 1994, p. 217] that for foams having reduced densities greater than about 0.1, the modulus of a foam is proportional to the square of the foam density:

Ef / Eo = (ρf / ρo)2

Below reduced density values of about 0.1, it has been theoretically determined that the cell wall acts to reinforce the structure [Gibson and Ashby cited above]. Practically, however, the above equation appears to overestimate the actual modulus, probably due to cell wall folds and broken membranes [Kraynik and Warren cited above]. In other words, the above equation should be considered as valid for all foams, regardless of the size of the cell structure, MC and UMC foams included. Claims to the contrary should be challenged.

On the other hand, extremely fine cell structure foams should show substantial improvement in any property that requires the polymer to be stressed to the limit, such as impact strength, flexural fatigue failure, and elongation at break. As an example, MC and UMC foams should show yield strengths greater than the theoretical yield strength [Gibson and Ashby, pp. 144-145]:

σy,f / σy,o = C(ρro) 3/2

Impact strength of untoughened polymers such as PMMA and PS should show even greater influence from reduced cell dimension. And in fact, these characteristics are seen in some of the MC data, albeit quite limited at this point. As one might anticipate, however, the impact strength dependency on cell size is certainly not a linear relationship. Instead, one might anticipate essentially no change in impact strength until cell sizes approach the critical Griffith crack dimension, which is in the range of 1 to 5 microns for unmodified polystyrene and PMMA. For foams having cell sizes below this range, however, the impact strength should jump immediately to the value of the unfoamed polymer. To date, this effect has not been adequately demonstrated in the open literature. Of course, it must be realized that for a foam to exhibit significantly improved impact strength, all cell dimensions must approach the critical value. It is insufficient to have some or even a very small minority of the cells above the critical value, since cracks will initiate at these larger cells.


There are many ways of manufacturing very fine celled foams, including those produced by the MIT process. The real question is not whether it is possible to produce MC or UMC foams, but whether it is possible to produce MC or UMC foams that are competitive with current foaming technologies. Of course, simple raw cost is not the proper criterion. Instead, the focus needs to be the almost trite “cost/performance ratio”. In other words, what industry applications can be benefited if an MC foam has superior impact strength, say, albeit at a much higher cost? If the best that can be achieved is an MC foamed egg carton, the technology will remain academic. However, in certain niche-oriented instances, extremely fine cell sizes may produce a product having unique properties. For instance, when cells are sub-micron in size [below 0.4 micron], the cells do not interfere with the transmission of visible light. UMC foams produced by the aerogel process are typically transparent. It has been estimated that there are 109 ft2 of single-glazed windows in the US, thus providing for an obvious application for flexible, easily applied, transparent insulating films 1. Even though UMC foams are typically open-celled, gases remain trapped in the cells, owing to an effect known as Knudsen diffusion. The gas molecules interact more with the cell walls than each other and so do not migrate, even though there is a bulk concentration gradient that should move the gas molecules from the foam cells. UMC foams with slightly smaller cells [0.1 micron or so] show great promise as bacteria filters, renal filters, and in-situ and implant drug substrates. And certainly organic UMC foams, when dehydrogenated, produce pure carbon foams of great surface area, with applications as neutron and noxious and toxic gas absorbers.

Leave a Reply