By Peter Weiss
For centuries, optimistic inventors have proposed perpetual-motion machines. These would defy scientific law, of course, and none has ever worked as advertised. Now, a blueprint for a minuscule perpetual-motion machine has been found convincing enough by other scientists to get published in the October Foundations of Physics.
In the report, Daniel P. Sheehan of the University of San Diego and his colleagues describe their design: a square silicon doughnut about the size of a red blood cell. A narrow gap in the doughnut ring would harbor a strong electric field. This would develop, presumably, because ambient heat and the structure’s electronic properties would separate charges. By driving a tiny silicon piston within the gap, the device would perform work.
A desktop array of such devices could power an entire office “without plugging any appliances into wall sockets,” Sheehan’s group projects.
However, the second law of thermodynamics requires that heat-driven machines receive energy that raises their temperatures above that of the surroundings–so that the flow of heat from hot to cold can run the machines (SN: 10/7/00, p. 234: https://www.sciencenews.org/20001007/bob1.asp). By directly converting the heat-generated motion of particles into mechanical power, the proposed device would break that law, Sheehan contends.
Not so, counters Denis J. Evans of the Australian National University in Canberra, who rates the San Diego scheme “impossible.” He and his coworkers recently demonstrated that light-guided microspheres in water can briefly violate the second law but obey it over the long run (SN: 7/27/02, p. 51: Law and Disorder: Chance fluctuations can rule the nanorealm). While the San Diego team has done “good work,” says Daniel C. Cole of Boston University, the more likely benefit from it will be “qualifications on normal statements of the second law,” rather than practical payoffs.
****************
If you have a comment on this article that you would like considered for publication in Science News, please send it to editors@sciencenews.org.
