LAST September, a team of engineers donned protective goggles and gathered
carefully around a lab bench at Los Angeles Air Force base. They were anxious to
find out whether their latest creation鈥�15 of the tiniest rocket thrusters
ever made鈥攚ould go off without blowing apart or fizzling out like damp
fireworks. At first, they were disappointed: the engines simply refused to
light. Finally, after reloading them with a more violent fuel, the team stood
back and tried again. With a pop and a bright flash, the rockets fired. 鈥淲e
realised鈥攈oly cow, now we have to make them work in space,鈥� says Erik
Antonsson, a microelectromechanical engineer at the California Institute of
Technology in Pasadena.
If Antonsson and project manager David Lewis, a rocket engineer at the
Ohio-based aerospace company TRW, succeed, their microscopic rockets will solve
a pressing problem. Many of today鈥檚 expensive communications satellites鈥�
5-tonne monsters鈥攃ould soon be replaced by swarms of 鈥渕icrosatellites鈥�.
These weigh just a few kilograms or less and offer a cheaper and more flexible
way to route communications or to observe the Earth and space.
But how do you manoeuvre these tiny satellites, point their sensors in a
different direction, for instance, or alter their orbit? There鈥檚 simply not
enough room on board for the fuel tanks, valves and pumps of conventional
thrusters.
Advertisement
So, with $3.5 million from the US Defense Advanced Research Projects
Agency, Lewis and Antonsson are constructing tiny rockets and loading each one
with just enough fuel to give a single blast of thrust. Cram millions together
onto a single silicon wafer and you can generate any amount of thrust, from a
quick squirt to a long blast, by firing them one at a time or in large groups.
Even better, you can stack the wafers onto the outside of small satellites, one
layer of thrusters on top of another. Use all the thrusters in one wafer and it
simply falls away like a discarded sweet wrapper, leaving a fresh set of
thrusters beneath.
And why stop there? Other researchers believe that these intricate rockets
will have uses nearer Earth, too. They hope to use them to power tiny flying
robot insects and to blast microscopic 鈥渟mart dust鈥� sensors into the atmosphere
to explore conditions inside tornadoes and thunderstorms or spy on enemy troop
movements.
Until Lewis and Antonsson, most engineers had tried to build small rockets
simply by miniaturising conventional thrusters. These are powered by chemical
propellants such as hydrazine and nitrogen tetroxide. The propellants are stored
in pressurised tanks, and thrust is controlled by opening valves and squirting
the fuel into a combustion chamber where it ignites. The hot gases expand
rapidly out of the rocket nozzle, pushing the satellite forward.
But miniature valves tend to leak. 鈥淎 satellite the size of a baseball would
last about two days before you鈥檇 lose all your propellant through these leaky
valves,鈥� says Siegfried Janson, an engineer working with Lewis.
Miniature valves are also difficult to open and close fast enough to allow
the delicate control required to place a microsatellite just where you want it
in space.
No moving parts
So Lewis and Antonnson have stolen a trick from the microchip industry. In a
radical departure from conventional satellite engineering, they are creating
simple rockets with no moving parts by carving components out of silicon wafers.
The result is a triple-decker sandwich with igniters in one layer, propellant in
another and rocket nozzles sunk into the surface.
This technology is relatively simple. First, the researchers take silicon
wafers protected on both sides by a 500-nanometre thick layer of silicon nitride
which is impervious to the etching chemicals used in making microchips. Next,
they mark out the components they require, using a programmable laser to burn
off the silicon nitride layer in areas they want to etch out. Finally, they
immerse the chips in potassium hydroxide to eat away exposed silicon, leaving
the other material untouched.
To build the thrusters, they carve tiny cylindrical combustion chambers just
1 millimetre long in a glass wafer. In another, they etch small, inverted
pyramids. The etching process stops when the potassium hydroxide reaches the
silicon nitride coating on the bottom of the wafer. In one move, the engineers
create two critical parts: the inverted pyramids are the rocket nozzles and the
silicon nitride coating is a flimsy diaphragm鈥攁bout fifty times thinner
than the width of a human hair鈥攖hat seals the propellant into the
combustion chamber until it is fired.
In a third wafer, they etch the igniters鈥攖iny electrical resistors less
than half a millimetre long. Finally, to complete the assembly process, they
load powdered rocket fuel into the combustion chambers by hand and glue the
layers together (see Diagram).
To fire the thruster, the designers simply pass a large current through the
resistors in the base layer. These become white hot: 鈥淭hey literally explode,鈥�
says Lewis. Then fragments of hot material create a shock wave which ignites the
solid propellant, adds Antonsson. In a split second, hot gases burst out through
the diaphragm creating a tiny puff of thrust.
The researchers describe their creation as 鈥渄igital propulsion鈥�. Just as the
1s and 0s of binary data represent a number of any magnitude, the thrusters can
be fired singly or in combination to produce almost any amount of power. And
while the thrusters are far less powerful than their larger counterparts, each
one containing only a few tiny grains of propellant, this is an advantage when
it comes to manoeuvring microsatellites that may weigh as little as a few
kilograms.
Engineers describe the shortest burst that a rocket produces as an 鈥渋mpulse
bit鈥�. Conventional thrusters must open and close their fuel valves as quickly as
possible to get the small impulse bits needed by microsatellites. 鈥淭en
milliseconds is about the quickest they can do it,鈥� Janson says. But Lewis鈥檚
valveless thrusters will fire for less than a millisecond, says Janson.
Eventually, they are aiming for impulse bits 10 000 times smaller than
conventional thrusters can produce.
So far, Lewis and his team have actually built wafers just 6 millimetres
across by 4 millimetres deep, containing 15 rocket thrusters. But they are
already designing a wafer 10 centimetres square that will carry more than a
million thrusters. And by reducing the size of the thrusters still further, they
believe that it might be possible to cram in more than ten times that
number.
Exploding chips
Panels of wafers could eventually be stacked on top of each other and
sloughed off by the spacecraft as it uses them up, says Lewis. These wafers
could be attached to the surface of a satellite, perhaps offering a huge saving
in weight by forming its exterior walls.
These thrusters must be extremely reliable if they are to remain in orbit for
years. In early tests, however, the thrusters stubbornly refused to light. If
this happened for real, a microsatellite would become a liability鈥攑acked
with propellant that could explode at any moment. 鈥淕od forbid the fuel doesn鈥檛
burn at all. Then what you have is a bomb,鈥� Lewis says.
In other tests, the engineers heard the familiar pop of detonation, but it
was followed by a tinkling sound. Peering gingerly over the top of the bench,
they realised that the chip had shattered, blasting the upper layer of silicon
more than a metre across the lab. The thin diaphragm between the combustion
chamber and the exhaust nozzle should rupture when the gas pressure reaches ten
times atmospheric pressure鈥攊nstead, the whole chip was disintegrating.
Eventually, they discovered where the problem lay. They were joining the
silicon layers together with glue and a tiny amount was leaking onto the
diaphragms. 鈥淵ou don鈥檛 want to get any glue on the diaphragms because that will
make them too strong,鈥� explains Janson. 鈥淓nough pressure builds up inside to
blow the whole chip apart.鈥�
Discovering a glue with exactly the right properties has proved a major
headache. It must be fluid enough to spread evenly between the layers, but not
so runny that it gums up the diaphragms鈥攁nd when set, it must be very
strong yet slightly flexible. 鈥淭o get into orbit, it has to survive a good deal
of vibration,鈥� Janson says. If the glue is too brittle, the violent buffeting
during launch will cause tiny fractures. When the thrusters fire, hot gases will
squeeze through these cracks and the chip could shatter. The researchers are
testing all sorts of epoxy resin glues in the hope of finding the perfect
adhesive, but eventually, admits Antonsson, they may need to find a completely
new bonding technique.
Even when the chips don鈥檛 blow themselves apart, the researchers have
discovered other, more subtle problems. When Lewis recorded the behaviour of the
experimental thrusters with a high-speed camera, he noticed that when the
thrusters fired, they spewed out a lot of unburnt fuel. Janson explains: 鈥淭hink
of a firecracker at the bottom of a tube filled with confetti. The confetti gets
blown out along with the hot gases produced by the firecracker.鈥� As a result,
the devices were squeezing just 10 per cent of the maximum thrust out of the
propellant鈥攃ompared with the 90 per cent efficiency of conventional rocket
engines.
The root of the problem appears to be that the propellant starts to burn at
the bottom of the combustion chamber, just above the igniter. This reaction is
so fast that the diaphragm bursts before all the propellant can burn and the
exhaust gases blast what is left out of the thruster. Putting the ignition
system near the top of the thruster should solve the problem, says Janson.
鈥淭here鈥檚 no physical reason we couldn鈥檛 get the same efficiency as the big
guys,鈥� Lewis says.
Squeezing millions of thrusters into a small space also has its risks. As
each thruster fires, the exhaust gases inside reach more than 1500 掳C. These
hot gases are separated from the propellant in neighbouring thrusters by walls
just fractions of a millimetre thick. If the heat reaches this propellant, a
whole array could go up like a string of firecrackers.
The best way to prevent this is to make sure that the propellant burns fast.
If the reaction is quick enough, the surrounding walls simply don鈥檛 have time to
heat up. The researchers are already using lead styphnate, a powerful compound
which burns so rapidly that there鈥檚 no time for heat to transfer to the other
chambers. But this makes the ignition harder to control, increasing the risk
that the rapidly rising pressure will blow the chip apart. 鈥淲e鈥檙e exploring the
use of a combination of fast and slower burning propellant to raise pressure in
a less violent way,鈥� says Antonsson.
Smart dust
If these problems can be solved, these miniature thrusters could
revolutionise the way military planners and scientists use satellites in the
21st century. Clouds of microsatellites might be used by the Pentagon to pick
off ballistic missiles, for instance. 鈥淲e could fly satellites like a flock of
birds,鈥� says Alok Das, a microsatellite expert at Edwards Air Force Base in
California. 鈥淲e could look for a downed pilot over Iraq one day and reconfigure
them to look for weapons the next day,鈥� he says. Communications companies could
use them as versatile antennas鈥攃hanging their formation or altering their
orbit to fulfil different roles. Or with the right sensors, these
microsatellites could be used to stare into outer space like a giant telescope a
kilometre across.
Kris Pister at the University of California at Berkeley has equally ambitious
plans for the tiny thrusters. He hopes to use them to power minute robotic
insects. Making flapping wings that give enough lift to carry objects around is
a difficult problem, says Pister鈥斺�漛ut it鈥檚 relatively easy to get things
to burn鈥�. So he is using silicon to build tiny 鈥渋nsects鈥� just millimetres
across, complete with their own antennas and transmitters. Attached to a tiny
thruster, he hopes these tiny insects will fly under their own power.
So far, he has made tiny thrusters that burn for about 2 seconds, but he
hopes to increase this to 20 seconds. Next, he will add thermoelectric
converters鈥攖iny electrical components that sit close to the thruster and
convert some of the heat from the rocket to electricity. This will power the
minute sensors, radio transmitters and receivers on the insect. 鈥淚f I can make
one, I can make a whole bunch of these things,鈥� he claims. 鈥淭hen I can start
thinking about making a colony, to find out how to get them to move and interact
迟辞驳别迟丑别谤.鈥�
Pister is also making 鈥渟mart dust鈥�, tiny clusters of sensors built on a
silicon chip less than a millimetre across that can measure conditions such as
wind speed or temperature in the atmosphere, and beam the information to ground
stations or nearby aircraft. Disperse them in the air and they could stay aloft
for hours, says Pister. 鈥淥ne idea is that aircraft will spill out little clouds
of these things behind them,鈥� he says. 鈥淭hey could record the current
atmospheric conditions so that other planes flying through that area will have a
warning if there鈥檚 some turbulence.鈥�
The US Department of Defense is attracted by another kind of smart dust.
Sprinkle it on the ground behind enemy lines, and when something interesting
passes by, say, a tank or a lorry, it fires its thruster and hops aboard. Once
there, miniature radio transmitters on the dust could relay its position to
base, giving away the position of enemy forces.
Smart dust may take a little time to develop, but Lewis is already planning
to send his thruster chips into orbit. In November they will get a test when
they shoot into space aboard an uncrewed Microcosm rocket. Lewis and the team
will wait nervously for sensors on the rocket to report whether their tiny
thrusters have fired, more than 100 kilometres up. Lewis is optimistic: the
physics of these tiny devices is on his side, he says, and this test should
prove it.