At the University of California at Berkeley, post-doctoral fellow Shaoyi Jiang, formerly Gubbins' graduate student, is studying the potential to use adsorption to reduce pressure in gas storage tanks. "At present," Gubbins says, "the methods of storage are to liquefy and store natural gas in refrigerated containers, which is very expensive, or to store it in high-pressure gas cylinders at room temperature." The latter technique is used for natural-gas fueled busses and cars, for instance, but it's not very desirable. "People don't want to drive around with a high-pressure tank in the back, and it also limits the range of the vehicle."
In a conventional high-pressure storage tank, such as a propane tank used for a backyard barbecue, gas is forced into the tank -- the more gas, the more pressure. "What we propose," Jiang says, "is that if someone puts some microporous materials into the tank, you can store the same amount of natural gas in the same tank, but at lower pressure."
Using the Connection Machine CM-2 at Pittsburgh, Jiang and Gubbins simulated how parallel layers of carbon atoms can adsorb methane atoms. "There is a force exerted by the carbon atoms inside the pore," Jiang says, "and this force attracts a lot of gas molecules into the pore so that the amount of gas in the bulk is reduced. As a result, the pressure of the tank can be kept low while maintaining a high density of methane in the pores."
The optimal pore size, Jiang and Gubbins discovered, is the width of two methane molecules. After the first layers of methane atoms line up along the pore's sides, carbon's attractive forces fall off rapidly. "Thus," Gubbins says, "the adsorbed methanes that are not in the contact layer on the wall will be much less tightly adsorbed because the forces will be much weaker. A rough analogy might be iron filings attracted to the pole of a magnet. The first one or two layers will be tightly bound to the surface, and subsequent layers will be more loosely bound and less dense."
In the future, Gubbins says, supercomputing will help address other adsorption issues that now are impossible on today's machines. For instance, it's currently possible to simulate only a few billionths of a second in real time, but some processes in nature require more time. "I don't think supercomputing ever will eliminate the need for experimental work on materials," Gubbins says, "but it should be able to pinpoint the kinds of materials that should be studied for particular purposes."60 KB 66 KB