The aerospike engine is a new design concept for
rocket engines. It was intended to be used in the
X-33 spaceplane, but it now appears that that project is
pushing up daisies.
In any case, the primary characteristic of the aerospike is that it does not waste as much thrust as a conventional belled rocket engine. In a conventional engine, the combustion chamber is connected to the outside through a nozzle, and the nozzle is surrounded by a roughly parabolic cone - the bell. This is because the exhaust gases continue expanding after leaving the nozzle; as they expand inside the engine bell, they exert force (thrust) against the inside of the bell itself.
All well and good so far. The problem is that the optimum shape of the bell is determined to a great deal by the ambient pressure around the outside of the rocket engine. At high pressures (when the vehicle is low in the atmosphere) the bell may in fact be too large, because the exhaust will not expand far enough against the ambient pressure to make a difference. If you observe a rocket engine in such a state, the exhaust flare will be streaming out of the back in a straight path, or even compressing in on itself as it leaves the bell. In this case, the bell is doing no good and is simply a drag on the craft.
In low ambient pressure (or its extreme, vacuum) the exhaust flare will not only fill the bell but will continue to expand outward past the lip of the bell. In such a case, thrust is being wasted on horizontal expansion; obviously, the bell is not big enough to take full advantage of the available thrust.
Now the aerospike. In a linear aerospike engine, there is one solid wedge of metal or ceramic that is linear in one dimension. In other words, take a rectangular block and pinch one long end so that it meets in a point and, in addition, has a concave curve to both sides. Ideally, a cross-section of this long edge (or spike, hence aerospike) would look exactly like the cross section of an engine bell if it was cut in half and the two sides laid back to back.
The combustion happens externally, along the long edges and just beneath the top (wide edge) of the spike, flowing down the sides. In this case, the aerospike receives thrust from the expanding exhaust pressing upwards against its concave edges, as it would press against the bell. However, the exhaust flare is shaped not by this spike, but by ambient pressure. In high ambient pressure, the flare is contained by the ambient pressure itself without the need for a heavy, dragging bell. In low pressure, the combustion point can be brought back to just below the edge of the spike, and at worst it will still collect as much thrust (approximately) as the bell would in vacuum due to having the same (inverse) surface area. In fact, it will receive more thrust, because the spike is only inefficient (like the edge of the low-pressure bell) on 50% of it' shape, the outside edge, whereas the bell is inefficient (loses potential energy) around its entire perimeter.
Thus the aerospike is more efficient in both environments - in high pressure because the bell isn't being carried, and in low because more surface area is available to collect thrust energy (exhaust gas pressure). The surface of the spike can be much larger than the aggregate surface area of bells because of the simpler structure; it doesn't have to withstand tensile (expansive) stresses like the bell does, so less materials and weight are needed per unit of thrust.
One of those things I wish I'd thought of.
Update: Jurph checks in with corrections, noting that the proper term for a liquid-fuelled rocket is 'engine', not 'motor' as I had been using in some places (the latter is reserved for solid-fuelled rockets). Also, he points out that there is in fact a 3-D aerospike, resembling an inversely tapered cone, in addition to the 2-D 'wedge' described above.
Update: On Sept. 21, 2003, a liquid fuelled aerospike engine flew atop a test rocket built by CSULB and Garvey Spacecraft Corp. The rocket, the Prospecter 2, launched off a rail under aerospike propulsion and continued upward for several seconds before veering horizontal and finally impacting the desert. The malfunction appears to be related to graphite structures in the engine eroding asymetrically, not due to any flaw with the aerospike concept itself.