RocketCat sez

Here is your handy-dandy cheat-sheet of rocket engines. Use this as a jumping-off point, there is no way I can keep this up-to-date. Google is your friend!

I'll point out a few of the more useful items on the sheet:

  • Aluminum-Oxygen is feeble, but is great for a lunar base (the raw materials are in the dirt).
  • VASIMR is the current favorite among ion-drive fans. Use this with orbit-to-orbit ships that never land on a planet. It can "shift gears" like an automobile.
  • Solar Moth might be a good emergency back-up engine.
  • Nuclear Thermal Solid Core is better than feeble chemical rockets, but not as much as you'd expect.
  • Nuclear Thermal Vapor Core is what you design along the way while learning how to make a gas core atomic rocket.
  • Nuclear Thermal Gas Core Open-Cycle is a full-blown honest-to-Heinlein atomic rocket, spraying glowing radioactive death in its exhaust.
  • Nuclear Thermal Gas Core Closed-Cycle is an attempt to have the advantages of both nuclear solid core and gas core, but often has the disadvantages of both. It has about half the exhaust velocity of an open-cycle atomic rocket.
  • Orion Nuclear Pulse is a rocket driven by detonating hundreds of nuclear bombs. If you can get past freaking out about the "bomb" part, it actually has many advantages. Don't miss the Medusa variant.
  • Magneto Inertial Fusion This is the best fusion-power rocket design to date.
  • Zubrin's Nuclear Salt Water This is the most over-the-top rocket. Imagine a continuously detonating Orion drive. There are many scientist who question how the rocket can possibly survive turning the drive on.

There is a nice basic overview of propulsion systems here.

You can spend lots of time researching spacecraft propulsion systems. But you are in luck, I've got some data for you. Most of this is from Philip Eklund's out of print boardgame Rocket Flight, the impressive Spaceship Handbook, and the indispensable Space Propulsion Analysis and Design. The rest is from various places I found around the internet, and no, I didn't keep track of where I got them. Use at your own risk.

Philip Eklund has a new boardgame out called High Frontier, which has the Atomic Rockets seal of approval (be sure to get the expansion pack as well). It has even more cutting-edge but scientifically accurate propulsion systems, which will eventually find there way onto this web page. (more details here, here, here, and here.)

If you don't like the values in the table, do some research to see if you can discover values you like better. Also note that the designs in the list are probably optimized for high exhaust velocities at the expense of thrust. There is a chance that some can be altered to give enough thrust for lift-off at the expense of exhaust velocity. Or you can just give up and go beg Mr. Tyco Bass for some atomic tri-tetramethylbenzacarbonethylene. Four drops should do the trick.

Some engines require electricity in order to operate. These have their megawatt requirements listed under "Power Requirements". With these engines, the Engine Mass value includes the mass of the power plant (unless the value includes "+pp", which means the mass value does NOT include the mass of the power plant). The power plant mass can be omitted if the spacecraft relies on beamed power from a remote power station. Alas, I could find no figures on the mass of the power plant. If the plant is nuclear, it probably has a mass of around 0.5 to 10 tons per megawatt. If it is beamed power the mass is of course zero. Efficiency is the percentage of the power requirements megawatts that are actually turned into thrust. The rest becomes waste heat and has to be removed with heat radiators.

T/W >1.0 = Thrust to Weight ratio greater than zero? This boils down to: can this engine be used to take off from Terra's surface? If the answer is "no" use it only for orbit to orbit maneuvers. It is calculated by figuring if the given thrust can accelerate the engine mass greater than one gee of acceleration. As a rule of thumb, a practical spacecraft capable of lifting off from the Earth's surface will require a T/W of about 50 to 75.

Most propulsion systems fall into two categories: SUV and economy. SUV propulsion is like an SUV automobile: big and muscular, but the blasted thing gets a pathetic three miles to the gallon. Economy propulsion has fantastic fuel economy, but has trouble climbing low hills. In the world of rockets, good fuel economy means a high "specific impulse" (Isp) and high exhaust velocity. And muscle means a high thrust.

The only vaguely possible propulsion system that has both high exhaust velocity and high thrust is the Nuclear Salt Water Rocket, and not a few scientist have questions about its feasibility. Well, actually there is also Project Orion, but that has other problems (see below). In science fiction, one often encounters the legendary "fusion drive" or "torchship", which is a high exhaust velocity + high thrust propulsion system that modern science isn't sure is even possible.

With ion engines, chemical engines, and nuclear torches we're facing a classic Newton's Third Law problem. Somehow the exhaust needs to have sufficient momentum for the opposite reaction to give the ship a good acceleration.

Chemical rockets solve the problem by expelling a ton of mass at a relatively low velocity. (high propellant mass flow but low exhaust velocity: SUV)

Ion drives expel a tiny amount of mass, so to get anywhere they get it moving FAST, but even at gigawatts of power they get a measly 0.0001g. (low propellant mass flow but high exhaust velocity: Economy)

Torch drives take a small-to-moderate amount of mass and use nuclear destruction to get it moving insanely fast. (medium propellant mass flow and high exhaust velocity: Torch) They're the only ones (insert disclaimer) with enough power per unit of reaction mass to get 0.3g constant acceleration conveniently. Even a perfect ion drive would need a phenomenal (read: impossible) amount of power input to match the performance of a nuclear explosion.

(A low propellant mass flow and low exhaust velocity engine would be utterly worthless)

From On Torchships comment by Eric (2010)

The Drive Table

All drives listed in the table whose names end in "MAX" require some sort of technological breakthrough to to prevent the engine from vaporizing and/or absurdly large reaction chamber sizes.

If these figures result in disappointing rocket performance, in the name of science fiction you can tweak some of them and claim it was due to a technological advance. You are allowed to tweak anything who's name does not end in "MAX". You can alter the Thrust, Engine Mass, and/or the Eff, but no other values. If there is a corresponding "MAX" entry for the engine you are tweaking, you cannot alter any of the values above the "MAX" entry (i.e., you are not allowed to tweak NTR-SOLID-DUMBO's thrust above 7,000,000, which is the value in the NTR-SOLID MAX entry).

The engines are sorted by thrust power, since that depends on both exhaust velocity and thrust. So engines that high in both of those parameters will be towards the end of the list. This is useful for designers trying to make spacecraft that can both blast-off from a planet's surface and do efficient orbital transfers.

If one was trying to design a more reasonable strictly orbit-to-orbit spacecraft one would want the engine list sorted by exhaust velocity. And surface-to-orbit designers would want the list sorted by thrust.

I have also created a graph of the data below.

Propulsion SystemThrust Power
Exhaust velocity
Engine mass
Power req
CHEM: Solid rocket3,000yes
CHEM: UDMH-N2O43,300
CHEM: Kerosine-Oxygen3,500yes
CHEM: Methane-Oxygen3,700yes
CHEM: Hydrogen-Oxygen4,600yes
CHEM: Hydrogen-Fluorine4,700yes
ETHERM: Resistojet0.000000729000.50.00280%
NTR: Radioisotope0.00678001.5no
EMAG: VASIMR (high gear)0.006294,0004010+ppno1060%
EMAG: VASIMR (med gear)0.006147,0008010+ppno1060%
EMAG: VASIMR (low gear)0.00629,00040010+ppno1060%
ETHERM: ArcJet0.0000220,0002no0.135%
Monatomic-H MITEE0.01512,7502,3500.2yes
ETHERM: Hybrid
BEAM: Solar Moth0.0189,0004,0000.1noSunlight63%
Basic MITEE0.0759,81014,0000.2yes
ESTAT: Colloid0.1743,000800020no20085%
MPD: J x B Electric0.1974,0005,000110no21180%
NTR: Solid (H2)8,093
NTR: Solid (CH4)6,318
NTR: Solid (NH3)5,101
NTR: Solid (H2O)4,042
NTR: Solid (CO2)3,306
NTR: Solid (CO or N2)2,649
NTR: NERVA0.198-
see above49,00010no
Propulsion SystemThrust Power
Exhaust velocity
Engine mass
Power req
BEAM: Laser Thermal0.06540,00013,00020no920 laser30%
NTR: LARS0.219,62020,0001.0yes
Mass Driver0.330,00020,000150no35090%
NTR: LANTR (Nerva mode)0.3099,22167,000?yes
NTR: LANTR (LOX mode)0.5846,347184,000?yes
ESTAT: Ion1.05210,00010,000400no80096%
FUSE: D-T Fusion1.222,000108,00010yes
NTR: NERVA Deriv (H2)1.358085334,06110.1yes(1570)
Metastable He*1.443,00064,00010no
NTR: Pebble Bed (H2)1.599,530333,6171.7yes(1945)
NTR: Vapor Core (H2)1.69,800330,0006.83yes
NTR: Cermet (H2)2.039,120445,2679.0yes(2000)
AM: Solid max2.410,791440,000?yes
Fission Fragment2.614,990,0003449no
EMAG: MPD3.1314,00020,0001540no400079%
Metastable He IV-A?21,600?10?
AM: Gasmax? 24,500???
NTR: Gas/Closed (H2)4.520,405445,00056.8no
PULSE: ORION Fission5.743,000263,000200no
THS HI Fusion Pulse6300,00040,0004yes
THS HT Fusion Pulse6150,00080,0004yes
PULSE: ORION Fusion10.773,000292,000200no
NTR: Dumbo14.0-
see above3,500,0005yes
CHEM: Solid rocket153,00010,000,000yes
CHEM: Space Shuttle
CHEM: x1 Saturn-V F-1232,9827,740,500yes
Propulsion SystemThrust Power
Exhaust velocity
Engine mass
Power req
FUSE: H-B Fusion30980,00061,000300no
AM: Plasma/Water30980,00061,000500no
CHEM: Space Shuttle
x2 SRB
NTR: Solid MAX4212,0007,000,00015yes
NTR: Liquid max5616,0007,000,00070yes
NTR: Gas/Open (H2)6135,0003,500,00030-200yes
NTR: Gas/Open 2nd Gen10050,0005,000,00030-200yes
AV:T Fusion
Cruise Mode
CHEM: Saturn-V first stage
x5 F-1
PULSE: Mini-Mag Orion147157,0001,870,000133no
NTR: Gas MAX15098,0003,000,00015yes
NTR: Gas/Coaxial (H2)15717,65817,800,000127yes
FUSE: He3-D Fusion1927,840,00049,0001200no
AM: Plasma/Hydrogen1927,840,00049,000500no
FUSE: MC-Fusion MAX2008,000,00050,0000.6yes
NSWR 20% UTB42766,00012,900,00033yes
ESTAT: IBS Agamemnon1,095219,00010,000,000?no
PULSE: 1959 ORION 1st Gen1,60039,00080,000,0001,700yes
AV:T Fusion
Combat Mode
PULSE: 1959 ORION 2nd Gen24,000120,000400,000,0003,250yes
NSWR 90% UTB MAX31,0004,700,00013,000,000?yes
PULSE: ORION MAX39,0009,800,0008,000,0008yes
FUSE: IC-Fusion MAX500,00010,000,000100,000,0001000yes
FUSE: H->He Fusion MAX?30,000,000??yes
FUSE: H->Fe Fusion MAX?50,000,000??yes
AM: Beam MAX500,000100,000,00010,000,00010?
Propulsion SystemThrust Power
Exhaust velocity
Engine mass
Power req


Solid Core

AM: Solid
Thrust Power2.4 GW
Exhaust velocity10,791 m/s
Thrust440,000 n
T/W >1.0yes

Basically a NERVA design where a tungsten target replaces the reactor. 13 micrograms per second of antiprotons are annihilated. The gamma rays and pions are captured in the tungsten target, heating it. The tungsten target in turn heats the hydrogen. Produces high thrust but the specific impulse is limited due to material constraints (translation: above a certain power level the blasted tungsten melts)

Gas Core

AM: Gas
Exhaust velocity 24,500 m/s

Microscopic amounts of antimatter are injected into large amounts of water or hydrogen propellant. The intense reaction flashes the propellant into plasma, which exits through the exhaust nozzle. Magnetic fields constrain the charged pions from the reaction so they heat the propellant, but uncharged pions escape and do not contribute any heating. Less efficient than AM-Solid core, but can achieve a higher specific impulse. For complicated reasons, a spacecraft optimized to use an antimatter propulsion system need never to have a mass ratio greater than 4.9, and may be as low as 2. No matter what the required delta V, the spacecraft requires a maximum of 3.9 tons of reaction mass per ton of dry mass, and a variable amount of antimatter measured in micrograms to grams.

Well, actually this is not true if the delta V required approaches the speed of light, but it works for normal interplanetary delta Vs.

Plasma Core

AM: Plasma
Thrust Power30 GW
Exhaust velocity980,000 m/s
Thrust61,000 n
Thrust Power192 GW
Exhaust velocity7,840,000 m/s
Thrust49,000 n
Engine mass500 tonne
T/W >1.0no
Power req MW
Eff %

Similar to antimatter gas core, but more antimatter is used, raising the propellant temperature to levels that convert it into plasma. A magnetic bottle is required to contain the plasma.

Beam Core

AM: Beam
Thrust Power500,000 GW
Exhaust velocity100,000,000 m/s
Thrust10,000,000 n
Engine mass10 tonne

Microscopic amounts of antimatter are reacted with equal amounts of matter. The charged pions from the reaction are used directly as thrust, instead of being used to heat a propellant. A magnetic nozzle channels them. Without a technological break-through, this is a very low thrust propulsion system. All antimatter rockets produce dangerous amounts of gamma rays. The gamma rays and the pions can transmute engine components into radioactive isotopes. The higher the mass of the element transmuted, the longer lived it is as a radioisotope.

Positron Ablative

Positron Ablative
Exhaust velocity49,000 m/s

This engine produces thrust when thin layers of material in the nozzle are vaporized by positrons in tiny capsules surrounded by lead. The capsules are shot into the nozzle compartment many times per second. Once in the nozzle compartment, the positrons are allowed to interact with the capsule, releasing gamma rays. The lead absorbs the gamma rays and radiates lower-energy X-rays, which vaporize the nozzle material. This complication is necessary because X-rays are more efficiently absorbed by the nozzle material than gamma rays would be.

Drawbacks include the fact that you need 1836 positrons to equal the energy of a single anti-proton, and only half the positrons will hit the pusher plate limiting the efficiency to 50%.

This system is very similar to Antiproton-catalyzed microfission

Beamed Power

Laser Thermal

Laser Thermal
Thrust Power0.065 GW
Exhaust velocity40,000 m/s
Thrust13,000 n
Engine mass20 tonne
T/W >1.0no
Power req920 MW laser

Similar to Solar Moth, but uses a stationary ground or space-station based laser instead of the sun. Basically the propulsion system leaves the power plant at home and relies upon a laser beam instead of an incredibly long extension cord.

With the mass of the power plant not actually on the spacecraft, more mass is available for payload. Or the reduced mass makes for a higher mass ratio to increase the spacecraft's delta V. The reduced mass also increases the acceleration. In some science fiction novels, combat "motherships" will have batteries of lasers, used to power hordes of ultra-high acceleration missiles and/or fighter spacecraft.

Laser Sail

A Laser Sail is a photon sail beam-powered by a remote laser installation.

As an important point, the practical minimum acceleration for a spacecraft is about 5 milligees. Otherwise it will take years to change orbits. Photo sails can only do up to 3 milligees, but a laser sail can do 5 milligees easily.

Solar Moth

Solar Moth
Thrust Power0.018 GW
Exhaust velocity9,000 m/s
Thrust4,000 n
Engine mass0.1 tonne
T/W >1.0no
Power reqSunlight
Eff63 %

Solar thermal rocket. 175 meter diameter aluminum coated reflector concentrates solar radiation onto a window chamber hoop boiler, heating and expanding the propellant through a regeneratively-cooled hoop nozzle. The concentrating mirror is one half of a giant inflatable balloon, the other half is transparent. Propellant is hydrogen seeded with alkali metal. The advantage is that you have power as long as the sun shines. The disadvantage is it doesn't work well past the orbit of Mars. The figures in the table are for Earth orbit.

The solar moth might be carried on a spacecraft as an emergency propulsion system, since the engine mass is so miniscule.

The energy density of sunlight at a given distance from the sun is:

Sed = 1 / Sdist2


  • Sed = sunlight energy density (Earth orbit density = 1.0)
  • Sdist = distance from Sun (Astronomical Units, Earth = 1.0)

1.0 astronomical units is defined as 149,597,870,700 meters.

So in Earth's orbit, the density is 1.0, at Mars orbit it is 0.44 (44%), at Jupiter orbit it is 0.037, at Neptune orbit it is 0.001, at Mercury orbit it is 6.68

Multiply this by the solar constant to get the exact power density. The solar constant is about 1.361 kilowatts per square meter (kW/m2) at solar minimum and about 1.362 kW/m2 at solar maximum. For example, during solar minimum the energy density at Jupiter orbit is 0.05 kW/m2 and at Mercury orbit is 9.1 kW/m2.


A barely contained chemical explosive. Noted for very high thrust and very low exhaust velocity. One of the few propulsion systems where the fuel and the propellant are the same thing. There is a list of chemical propellants here


Chemical: Aluminum-Oxygen
Exhaust velocity2,800 m/s
T/W >1.0yes

Aluminum and oxygen are burned resulting in an unremarkable specific impulse of about 285 seconds. However, this is of great interest to any future lunar colonies. Both aluminum and oxygen are readily available in the lunar regolith, and such a rocket could easily perform lunar liftoff, lunar landing, or departure from a hypothetical L5 colony for Terra (using a lunar swingby trajectory). The low specific impulse is more than made up for by the fact that the fuel does not have to be imported from Terra. It can be used in a hybrid rocket (with solid aluminum burning in liquid oxygen), or using ALICE (which is a slurry of nanoaluminium powder mixed in water then frozen).

Of course the aluminum oxide in lunar regolith has to be split into aluminum and oxygen before you can use it as fuel. But Luna has plenty of solar power. As a rule of thumb, in space, energy is cheap but matter is expensive.


Chemical: Methane-Oxygen
Exhaust velocity3,700 m/s
T/W >1.0yes

Methane and oxygen are burned resulting in an unremarkable specific impulse of about 377 seconds. However, this is the highest performance of any chemical rocket using fuels that can be stored indefinitely in space. Chemical rockets with superior specific impulse generally use liquid hydrogen, which will eventually leak away by escaping between the the molecules composing your fuel tanks. Liquid methane and liquid oxygen will stay put.


Chemical: Hydrogen-Oxygen
Exhaust velocity4,600 m/s
T/W >1.0yes

Hydrogen and oxygen are burned resulting in close to the theoretical maximum specific impulse of about 450 seconds. However, liquid hydrogen cannot be stored permanently in any tank composed of matter. The blasted stuff will escape atom by atom between the molecules composing the fuel tanks.

Atomic Hydrogen

100% Atomic Hydrogen
Exhaust velocity20,600 m/s
15% Atomic Hydrogen in solid H2
Exhaust velocity7,300 m/s

Atomic hydrogen is also called free-radical hydrogen or "single-H". The problem is that it instantly wants to recombine. The least unreasonable way of preventing this is to make a solid mass of frozen hydrogen (H2) at liquid helium temperatures which contains 15% single-H by weight.

Metallic Hydrogen

Metallic Hydrogen
Exhaust velocity17,000 m/s

Hydrogen subjected to enough pressure to turn it into metal, then contained under such pressure. Release the pressure and out comes all the stored energy that was required to compress it in the first place. It will require storage that can handle millions of atmospheres worth of pressure. The mass of the storage unit might be enough to negate the advantage of the high exhaust velocity.

Metastable He*

Metastable He*
Thrust Power1.4 GW
Exhaust velocity43,000 m/s
Thrust64,000 n
Engine mass10 tonne
T/W >1.0no

Spin-polarized triplet helium. Two electrons in a helium atom are aligned in a metastable state (one electron each in the 1s and 2s atomic orbitals with both electrons having parallel spins, the so-called "triplet spin state", if you want the details). When it reverts to normal state it releases 0.48 gigjoules per kilogram. Making the stuff is easy. The trouble is that it tends to decay spontaneously, with a lifetime of a mere 2.3 hours. And it will decay even quicker if something bangs on the fuel tank. Or if the ship is jostled by hostile weapons fire. To say the fuel is touchy is putting it mildly. The fuel is stored in a resonant waveguide to magnetically lock the atoms in their metastable state but that doesn't help much. There were some experiments to stablize it with circularly polarized light, but I have not found any results about that.

Metastable He IV-A

Metastable He IV-A
Exhaust velocity21,600 m/s
Engine mass10 tonne

Meta-helium would be such a worthwhile propulsion system that scientists have been trying real hard to get the stuff to stop decaying after a miserable 2.3 hours. One approach is to see if metastable helium can be formed into a room-temperature solid if bonded with diatomic helium molecules, made from one ground state atom and one excited state atom. This is called diatomic metastable helium. The solid should be stable, and it can be ignited by heating it. The exhaust velocity is about half that of pure He* which is disappointing, but not as disappointing as a dust-mote sized meteorite blowing your ship into atoms.

Theoretically He IV-A would be stable for 8 years, have a density of 0.3 g/cm3, and be a solid with a melting point of 600 K (27° C). The density is a plus, liquid hydrogen's annoying low density causes all sorts of problems.

Electromagnetic (Plasma)

Electrodeless plasma

Helicon Double Layer (HDLT)

Magnetoplasmadynamic (MPD)

Thrust Power3.1 GW
Exhaust velocity314,000 m/s
Thrust20,000 n
Engine mass1540 tonne
T/W >1.0no
Power req4000 MW
Thrust Power0.033 GW
Exhaust velocity79,000 m/s
Thrust98 n
Num above enginesx6

Magnetoplasmadynamic thruster, a travelling wave plasma accelerator. Propellant is potassium seeded helium.

Pulsed Inductive (PIT)

Pulsed Plasma (PPT)


High Gear
Exhaust velocity294,000 m/s
Thrust40 n
Medium Gear
Exhaust velocity147,000 m/s
Thrust80 n
Low Gear
Exhaust velocity29,000 m/s
Thrust400 n
Thrust Power0.006 GW
Engine mass10 tonne + pp
T/W >1.0no
Power req10 MW
Eff60 %

Some classify this as an electromagnetic plasma, some as an electrodeless electrothermal

The variable specific impulse magnetoplasma rocket is a plasma drive with the amusing ability to "shift gears." This means it can trade exhaust velocity for thrust and vice versa. Three "gears" are shown on the table. There are more details here and here.

VASIMR has been suggested for use in a space tug aka Orbital Transfer Vehicle. A VASIMR powered tug could move 34 metric tons from Low Earth Orbit (LEO) to Low Lunar Orbit (LLO) by expending only 8 metric tons of argon propellant. A chemical rocket tug would require 60 metric tons of liquid oxygen - liquid hydrogen propellant. Granted the VASIMR tug would take six month transit time as opposed to the three days for the chemical, but there are always trade offs.



ESTAT: Colloid
Thrust Power0.17 GW
Exhaust velocity43,000 m/s
Thrust8000 n
Engine mass20 tonne
T/W >1.0no
Power req200 MW
Eff85 %

Similar to Ion, but utilizing tiny droplets instead of ions.

Field-Emission Electric (FEEP)

Field-emission electric propulsion, a type of Colloid thruster.

Hall Effect (HET)


Thrust Power1.05 GW
Exhaust velocity210,000 m/s
Thrust10,000 n
Engine mass400 tonne
T/W >1.0no
Power req800 MW

Gridded Electrostatic Ion Thruster. Potassium seeded argon is ionized and the ions are accelerated electrostatically by electrodes. Other propellants can be used, such as cesium and buckyballs. Though it has admirably high exhaust velocity, there are theoretical limits that ensure all Ion drives are low thrust.

It also shares the same problem as the other electrically powered low-thrust drives. In the words of a NASA engineer the problem is "we can't make an extension cord long enough." That is, electrical power plants are weighty enough to make the low thrust an even larger liability. A high powered ion drive will generally be powered by a nuclear reactor, Nuclear Electric Propulsion (NEP). Low powered ion drives can get by with solar power arrays, all ion drive space probes that exist in the real world use that system. Researchers are looking into beamed power systems, where the ion drive on the spaceship is energized by a laser beam from a remote space station.

If you are interested in the technical details about why ion drives are low thrust, read on.

And it suffers from the same critical thrust-limiting problem as any other ion engine: since you are accelerating ions, the acceleration region is chock full of ions. Which means that it has a net space charge which repels any additional ions trying to get in until the ones already under acceleration manage to get out, thus choking the propellant flow through the thruster.

The upper limit on thrust is proportional to the cross-sectional area of the acceleration region and the square of the voltage gradient across the acceleration region, and even the most optimistic plausible values (i.e. voltage gradients just shy of causing vacuum arcs across the grids) do not allow for anything remotely resembling high thrust.

You can only increase particle energy so much; you then start to get vacuum arcing across the acceleration chamber due to the enormous potential difference involved. So you can't keep pumping up the voltage indefinitely.

To get higher thrust, you need to throw more particles into the mix. The more you do this, the more it will reduce the energy delivered to each particle.

It is a physical limit. Ion drives cannot have high thrusts.

{ IBS Agamemnon }

IBS Agamemnon Ion
Thrust Power1,095 GW
Exhaust velocity219,000 m/s
Thrust10,000,000 n
T/W >1.0no

Fictional Interplanetary BoostShip Agamemnon from Jerry Pournelle's short story ""Tinker"". This fictional ship is a species of Ion drive utilizing cadmium and powered by deuterium fusion. Looking at its performance I suspect that in reality no Ion drive could have such a high thrust. The back of my envelope says that you'd need one thousand ultimate Ion drives to get this much thrust.



Thrust Power0.00002 GW
Exhaust velocity20,000 m/s
Thrust2 n
T/W >1.0no
Power req0.1 MW

Hydrogen propellant is heated by an electrical arc.

Microwave Electrothermal


Thrust Power0.0000007 GW
Exhaust velocity2900 m/s
Thrust0.5 n
T/W >1.0no
Power req0.002 MW

In a resistojet, ropellant flows over a resistance-wire heating element (much like a space heater or toaster) then the heated propellant escapes out the exhaust nozzle. They are mostly used as attitude jets on satellites, and in situations where energy is more plentiful than mass.


Magnetic Confinement

Thrust Power200 GW
Exhaust velocity8,000,000 m/s
Thrust50,000 n
Engine mass0.6 tonne
T/W >1.0yes

A magnetic bottle contains the fusion reaction. Very difficult to do. Researchers in this field say that containing fusion plasma in a magnetic bottle is like trying to support a large slab of gelatin with a web of rubber bands. Making a magnetic bottle which has a magnetic rocket exhaust nozzle is roughly 100 times more difficult.

Since the engine is using a powerful but tightly controlled magnetic field, it might be almost impossible to have a cluster of several magnetic confinement fusion engines. The magnetic fields will interfere with each other.


D-T Fusion
Thrust Power1.2 GW
Exhaust velocity22,000 m/s
Thrust108,000 n
Engine mass10 tonne
T/W >1.0yes

Fuel: deuterium and tritium. Propellant: lithium. 1 atom of Deuterium fuses with 1 atom of Tritium to produce 17.6 MeV of energy. One gigawatt of power requires burning a mere 0.00297 grams of D-T fuel per second.

Note that Tritium has an exceedingly short half-life of 12.32 years. Use it or lose it. Most designs using Tritium included a blanket of Lithium to breed more fresh Tritium fuel.


H-B Fusion
Thrust Power30 GW
Exhaust velocity980,000 m/s
Thrust61,000 n
Engine mass300 tonne
T/W >1.0no

Fuel is Hydrogen and Boron-11. Propellant is hydrogen. Bombard Boron-11 atoms with Protons (i.e., ionized Hydrogen) and you get a whopping 16 Mev of energy, three Alpha particles, and no deadly neutron radiation.

Well, sort of. Current research indicatates that there may be some neutrons. Paul Dietz says there are two nasty side reactions. One makes a Carbon-12 atom and a gamma ray, the other makes a Nitrogen-14 atom and a neutron. The first side reaction is quite a bit less likely than the desired reaction, but gamma rays are harmful and quite penetrating. The second side reaction occurs with secondary alpha particles before they are thermalized.

The Hydrogen - Boron reaction is sometimes termed "thermonuclear fission" as opposed to the more common "thermonuclear fusion".

A pity about the low thrust. The fusion drives in Larry Niven's "Known Space" novels probably have performance similar to H-B Fusion, but with millions of newtons of thrust.

It sounded too good to be true, so I asked "What's the catch?"

The catch is, you have to arrange for the protons to impact with 300 keV of energy, and even then the reaction cross section is fairly small. Shoot a 300 keV proton beam through a cloud of boron plasma, and most of the protons will just shoot right through. 300 keV proton beam against solid boron, and most will be stopped by successive collisions without reacting. Either way, you won't likely get enough energy from the few which fuse to pay for accelerating all the ones which didn't.

Now, a dense p-B plasma at a temperature of 300 keV is another matter. With everything bouncing around at about the right energy, sooner or later everything will fuse. But containing such a dense, hot plasma for any reasonable length of time, is well beyond the current state of the art. We're still working on 25 keV plasmas for D-T fusion.

If you could make it work with reasonable efficiency, you'd get on the order of ten gigawatt-hours of usable power per kilogram of fuel.

Professor N. Rostoker, et. al think they have the solution, utilizing colliding beams. Graduate Student Alex H.Y. Cheung is looking into turning this concept into a propulsion system.


He3-D Fusion
Thrust Power192 GW
Exhaust velocity7,840,000 m/s
Thrust49,000 n
Engine mass1,200 tonne
T/W >1.0no

Fuel is helium3 and deuterium. Propellant is hydrogen. 1 atom of Deuterium fuses with 1 atom of Helium-3 to produce 18.35 MeV of energy. One gigawatt of power requires burning a mere 0.00285 grams of 3He-D fuel per second.

{ AV:T Fusion }

AV:T Fusion
Cruise mode
Thrust Power102 GW
Exhaust velocity832,928 m/s
Thrust245,250 n
T/W >1.0no
Combat mode
Thrust Power2,540 GW
Exhaust velocity104,116 m/s
Thrust48,828,125 n
T/W >1.0yes

Fictional magnetic bottle fusion drive from the Attack Vector: Tactical wargame. It uses an as yet undiscovered principle to direct the heat from the fusion reaction out the exhaust instead of vaporizing the reaction chamber. Like the VASIMR it has "gears", a combat mode and a cruise mode. The latter increases specific impulse (exhaust velocity) at the expense of thrust.

In the illustration, the spikes are solid-state graphite heat radiators, the cage the spikes emerge from is the magnetic bottle, the sphere is the crew quarters and the yellow rectangles are the retractable power reactor heat radiators. The ship in the lower left corner is signaling its surrender by deploying its radiators.

{ THS Fusion Pulse }

Fusion Pulse low gear
Thrust Power6 GW
Exhaust velocity300,000 m/s
Thrust40,000 n
Fusion Pulse high gear
Thrust Power6 GW
Exhaust velocity150,000 m/s
Thrust80,000 n
Engine mass4 tonne
T/W >1.0yes

Fictional inertial-confinement fusion drive from the game GURPS: Transhuman Space. Like the VASIMR it has "gears", one increases specific impulse (exhaust velocity) at the expense of thrust.

Nuclear Thermal

These use the heat generated from a nuclear reaction to heat up propellant. The nuclear reaction is controlled by adjusting the amount of free neutrons inside the mass of fissioning material.

As a side effect, if you have a cluster of several such engines it is vitally important to have distance and neutron shields between them. Otherwise the nuclear reaction in each engine will flare out of control due to the neutron flux from its neighbor engines.

Solid Core

Solid Core NTR
3200° K
Exhaust velocity (H2)8,093 m/s
Exhaust velocity (CH4)6,318 m/s
Exhaust velocity (NH3)5,101 m/s
Exhaust velocity (H2O)4,042 m/s
Exhaust velocity (CO2)3,306 m/s
Exhaust velocity (CO or N2)2,649 m/s

Nuclear thermal rocket / solid core fission. It's a real simple concept. Put a nuclear reactor on top of an exhaust nozzle. Instead of running water through the reactor and into a generator, run hydrogen through it and into the nozzle. By diverting the hydrogen to a turbine generator 60 megawatts can be generated. The reactor elements have to be durable, since erosion will contaminate the exhaust with fissionable materials. The exhaust velocity limit is fixed by the melting point of the reactor.

Solid core nuclear thermal rockets have a nominal core temperature of 2,750 K (4,490° F).

Hydrogen gives the best exhaust velocity, but the other propellants are given in the table since a spacecraft may be forced to re-fuel on whatever working fluids are available locally (what Jerry Pournelle calls "Wilderness re-fuelling", Robert Zubrin calls "In-situ Resource Utilization", and I call "the enlisted men get to go out and shovel whatever they can find into the propellant tanks"). For thermal drives in general, and NTR-SOLID in particular, the exhaust velocity imparted to a particular propellant by a given temperature is proportional to 1 / sqrt( molar mass of propellant chemical ).

The value for "hydrogen" in the table is for molecular hydrogen, i.e., H2. Atomic hydrogen would be even better, but unfortunately it tends to explode at the clank of a falling dust speck (Heinlein calls atomic hydrogen "Single-H"). Another reason to avoid hydrogen is the difficulty of storing the blasted stuff, and its annoyingly low density (Ammonia is about eight times as dense!). Exhaust velocities are listed for a realistically attainable core temperature of 3200 degrees K for the propellants Hydrogen (H2), Methane (CH4), Ammonia (NH3), Water (H2O), Carbon Dioxide (CO2), Carbon Monoxide (CO), and Nitrogen (N2).

The exhaust velocities are larger than what one would expect given the molecular weight of the propellants because in the intense heat they break down into their components. Ammonia is nice because it breaks down into gases (Hydrogen and Nitrogen). Methane is nasty because it breaks down into Hydrogen and Carbon, the latter tends to clog the reactor with soot deposits. Water is most unhelpful since it doesn't break down much at all.

Dr. John Schilling figures that as an order of magnitude guess, about one day of full power operation would result in enough fuel burnup to require reprocessing of the fissionable fuel elements. (meaning that while there is still plenty of fissionables in the fuel rod, enough by-products have accumulated that the clogged rod produces less and less energy) A reprocessing plant could recover 55-95% of the fuel. With reprocessing, in the long term each totally consumed kilogram of plutonium or highly enriched uranium (HEU) will yield ~1E10 newton-seconds of impulse at a specific impulse of ~1000 seconds.

Dr. Schilling also warns that there is a minimum amount of fissionable material for a viable reactor. Figure a minimum of 50 kilograms of HEU.

One problem with solid-core NTRs is that if the propellant is corrosive, that is, if it is oxidizing or reducing, heating it up to three thousand degrees is just going to make it more reactive. Without a protective coating, the propellant will start corroding away the interior of the reactor, which will make for some real excitement when it starts dissolving the radioactive fuel rods. What's worse, a protective coating against an oxidizing chemical is worthless against a reducing chemical, which will put a crimp in your wilderness refueling. And trying to protect against both is an engineering nightmare. Oxidizing propellants include oxygen, water, and carbon dioxide, while reducing propellants include hydrogen, ammonia, and methane. Carbon Monoxide is neither, as the carbon atom has a death-grip on the oxygen atom.

Keep in mind that the oxidizing/reducing effect is only a problem with solid-core NTRs, not the other kinds. This is because only the solid-core NTRs have solid reactor elements exposed to the propellant (for heating).

A useful refinement is the Bimodal NTR.

Say your spacecraft has an honest-to-Johnny NERVA nuclear-thermal propulsion system. Typically it operates for a few minutes at a time, then sits idle for the rest of the entire mission. Before each use, one has to warm up the reactor, and after use the reactor has to be cooled down. Each of these thermal cycles puts stress on the engine. And the cooling process consists of wasting propellant, flushing it through the reactor just to cool it off instead of producing thrust.

Meanwhile, during the rest of the mission, your spacecraft needs electricity to run life support, radio, radar, computers, and other incidental things.

So make that reactor do double duty (that's where the "Bimodal" comes in) and kill two birds with one stone. Refer to above diagram. Basically you take a NERVA and attach a power generation unit to the side. The NERVA section is the "cryogenic H2 propellant tank", the turbopump, and the thermal propulsion unit. The power generation section is the generator, the radiator, the heat exchanger, and the compressor.

Warm up your reactor once, do a thrust burn, stop the propellant flow and use the heat exchanger and radiator to partially cool the reactor to power generation levels, and keep the reactor warm for the rest of the mission while generating electricity for the ship.

This allows you to get away with only one full warm/cool thermal cycle in the entire mission instead of one per burn. No propellant is wasted as coolant since the radiator cools down the reactor. The reactor supplies needed electricity. And as an added bonus, the reactor is in a constant pre-heated state. This means that in case of emergency one can power up and do a burn in a fraction of the time required by a cold reactor.

Pretty ingenious, eh?

And the Pratt & Whitney company went one step better. They took the Bimodal NTR concept and merged it with the LANTR concept to make a Trimodal NTR. Called the Triton, it can be used in LANTR mode when thrust is more important than specific impulse, NTR mode when specific impulse is more important than thrust, and in power generation mode while coasting.


Thrust Power0.198-0.065 GW
Exhaust velocitySee Table
Thrust49,000 n
Engine mass10 tonne
T/W >1.0no

Nuclear Engine for Rocket Vehicle Applications. The first type of NTR-SOLID propulsion systems. It used reactor fuel rods surrounded by a neutron reflector. Unfortunately its thrust to weight ratio is less than one, so no lift-offs with this rocket. The trouble was inadequate propellant mass flow, the result of trying to squeeze too much hydrogen through too few channels in the reactor.

NERVA Derivative

Thrust Power1.35 GW
Exhaust velocity8,085 m/s
Thrust334,061 n
Engine mass10.1 tonne
T/W >1.0yes


General Dumbo
Thrust Power14.0-4.6 GW
Exhaust velocitySee Table
Thrust3,500,000 n
Engine mass5 tonne
T/W >1.0yes
Dumbo Model A
Engine mass0.7 tonne
Thrust400,000 n
Propellant mass flow52 kg/sec
Exhaust velocity7,700 m/sec
Engine Height0.6 m
Engine Radius0.3 m
Engine Volume0.2 m3
Dumbo Model B
Engine mass2.8 tonne
Thrust3,560,000 n
Propellant mass flow460 kg/sec
Exhaust velocity7,700 m/sec
Engine Height0.6 m
Engine Radius1.0 m
Engine Volume1.8 m3
Dumbo Model C
Engine mass2.1 tonne
Thrust400,000 n
Propellant mass flow48 kg/sec
Exhaust velocity8,300 m/sec
Engine Height0.6 m
Engine Radius0.4 m
Engine Volume0.3 m3

This was a competing design to NERVA. It was shelved political decision that, (in order to cut costs on the atomic rocket projects) required both projects to use an already designed NEVRA engine nozzle. Unfortunately, said nozzle was not compatible with the DUMBO active cooling needs. Dumbo does, however, have a far superior mass flow to the NERVA, and thus a far superior thrust. Dumbo actually had a thrust to weight ratio greater than one. NASA still shelved DUMBO because [a] NERVA used off the shelf components and [b] they knew there was no way in heck that they could get permission for nuclear lift-off rocket so who cares if T/W < 0? You can read more about Dumbo in this document.

Note that the "engine mass" entry for the various models does not include extras like the mass of the exhaust nozzle, mass of control drums, or mass of radiation shadow shield.

Pebble Bed

Pebble Bed
Thrust Power1.59 GW
Exhaust velocity9,530 m/s
Thrust333,617 n
Engine mass1.7 tonne
T/W >1.0yes

Particle bed / nuclear thermal rocket AKA fluidized-bed, dust-bed, or rotating-bed reactor. In the particle-bed reactor, the nuclear fuel is in the form of a particulate bed through which the working fluid is pumped. This permits operation at a higher temperature than the solid-core reactor by reducing the fuel strength requirements . The core of the reactor is rotated (approximately 3000 rpm) about its longitudinal axis such that the fuel bed is centrifuged against the inner surface of a cylindrical wall through which hydrogen gas is injected. This rotating bed reactor has the advantage that the radioactive particle core can be dumped at the end of an operational cycle and recharged prior to a subsequent burn, thus eliminating the need for decay heat removal, minimizing shielding requirements, and simplifying maintenance and refurbishment operations.


NERVA mode
Thrust Power0.309 GW
Exhaust velocity9,221 m/s
Thrust67,000 n
T/W >1.0yes
LOX mode
Thrust Power0.584 GW
Exhaust velocity6,347 m/s
Thrust184,000 n
T/W >1.0yes

LOX-augmented Nuclear Thermal Rocket. One of the systems that can increase thrust by lowering Isp. This concept involves the use of a "conventional" hydrogen (H2) NTR with oxygen (O2) injected into the nozzle. The injected O2 acts like an "afterburner" and operates in a "reverse scramjet" mode. This makes it possible to augment (and vary) the thrust (from what would otherwise be a relatively small NTR engine) at the expense of reduced Isp


MInature ReacTor EnginE. MITEE is actually a family of engines. These are small designs, suitable for launching on existing boosters. You can find more details here.

Thrust Power0.075 GW
Exhaust velocity9,810 m/s
Thrust14,000 n
Engine mass0.2 tonne
T/W >1.0yes

The baseline design is a fairly conventional NTR. Unlike earlier designs it keeps the fuel elements in individual pressure tubes instead of a single pressure vessel, making it lighter and allowing slightly higher temperatures and a bit better exhaust velocity.

Monatomic H
Monatomic H MITEE
Thrust Power0.015 GW
Exhaust velocity12,750 m/s
Thrust2,350 n
Engine mass0.2 tonne
T/W >1.0yes

This advanced design works at a lower chamber pressure so that some of the H2 propellant disassociates into monatomic hydrogen, although the chamber temperature is only slightly greater. The drawback is that this reduces the mass flow through the reactor, limiting reactor power.

Thrust Power0.015 GW
Exhaust velocity17,660 m/s
Thrust1,700 n
Engine mass1-10 tonne
T/W >1.0no

The use of individual pressure tubes in the reactor allows some fuel channels to be run at high pressure while others are run at a lower pressure, facilitating a hybrid electro-thermal design. In this design cold H2 is heated in the high pressure section of the reactor, is expanded through a turbine connected to a generator, then reheated in the low pressure section of the reactor before flowing to the nozzle. The electricity generated by the turbine is used to break down more H2 into monatomic hydrogen, increasing the exhaust velocity. Since this is a once through system there is no need for radiators so the weight penalty would not be excessive.

Liquid Core

Liquid Core 1
Thrust Power56 GW
Exhaust velocity16,000 m/s
Thrust7,000,000 n
Engine mass70 tonne
T/W >1.0yes
Liquid Core 2
Exhaust velocity14,700 to 25,500 m/s

Nuclear thermal rocket / liquid core fission. Similar to an NTR-GAS, but the fissionable core is merely molten, not gaseous. A dense high temperature fluid contains the fissionable material, and the hydrogen propellant is bubbled through to be heated. The propellant will be raised to a temperature somewhere between the melting and boiling point of the fluid. Candidates for the fluid include tungsten (boiling 6160K), osmium (boiling 5770K), rhenium (boiling 6170K), or tantalum (boiling 6370K).

Liquid core nuclear thermal rockets have a nominal core temperature of 5,250 K (8,990°F).

The reaction chamber is a cylinder which is spun to make the molten fluid adhere to the walls, the reaction mass in injected radially (cooling the walls of the chamber) to be heated and expelled out the exhaust nozzle.

Starting up the engine for a thrust burn will be complicated and tricky, shutting it down even more so. Keeping the fissioning fluid contained in the chamber instead of escaping out the nozzle will also be a problem.


Thrust Power0.2 GW
Exhaust velocity19,620 m/s
Thrust20,000 n
Engine mass1.0 tonne
T/W >1.0yes

Nuclear thermal rocket / liquid annular reactor system. A type of NTR-LIQUID. You can find more details here

Vapor Core

Vapor Core
Thrust Power1.6 GW
Exhaust velocity9,800 to
11,800 m/s
Thrust330,000 n
Propellant mass flow30 kg/sec
Reactor thermal power1,400 to
1,800 MW
Total engine mass6.83 tonne
T/W >1.0yes
Fuel element mass total1.35 tonne
Forward reflector mass0.60 tonne
Aft reflector mass0.51 tonne
Radial reflector mass2.47 tonne
Radiation shield mass0.9 tonne
Total reactor mass5.83 tonne
Misc. engine
component mass
0.9 tonne

This is sort of an intermediate step in learning how to design a full-blown Gas Core Nuclear Thermal Rocket. It is basically a solid core NTR where the solid nuclear fuel elements are replaced by chambers filled with uranium235 tetrafluoride vapor. The engine is admirably compact with a nicely low critical mass, and an impressive thrust-to-weight ratio of 5-to-1. However the specific impulse / exhaust velocity is only slightly better than a solid core.

In other words, the system is not to be developed because it has fantastic performance, but because it will be an educational step to building a system that does.

The uranium fuel is kept physically separate from the hydrogen propellant, so the exhaust is not radioactive.

A 330,000 newton thrust NVTR would have a core with almost 4,000 fuel elements, with a core radius of 120 cm, core height of 150 cm, and 1,800 MW. Criticality can be achieved with smaller cores: a core volume five times smaller with radius of 60 cm, height of 120 cm, and power of 360 MW.

Data is from Conceptual Design of a Vapor Core Reactor Rocket Engine for Space Propulsion by E.T. Dugan, N.J. Diaz, S.A. Kuras, S.P. Keshavmurthy, and I. Maya (1996).

ForwardBeryllium oxide15 cm0.60 tonne
AftC-C Composite25 cm0.51 tonne
RadialBeryllium oxide15 cm2.47 tonne
CORE: 2000 fuel elements
Radius0.5 m
Height1.5 m
Fuel channel per element12 to 32
Hydrogen channel per element12 to 32
Critical mass20 kg
Hydrogen pressure100 atm
UF4 pressure100 atm
Fuel center temperature4,500 K
Design Values
Pump Flowrate (Total)75.20 lbm/s
Pump Discharge Pressure3,924 psia
Pump Efficiency80.01%
Turbopump RPM70,000 RPM
Turbopump Power (each)9,836 HP
Turbine Inlet Temperature481 deg-R
Turbine Pressure Ratio1.69
Turbine Flow Rate (each)33.77 lbm/s
Reactor Thermal Power1,769 MW
Fuel Element and Reflector Power1,716 MW
Nozzle Chamber Temperature5,580 deg-R
Chamber Pressure (Nozzle Stagnation)1,500 psia
Nozzle Expansion Area Ratio500:1
Vacuum Specific Impulse (Delivered)997.8 sec
Heat Loads
Nozzle-con (total)30.05 MW
Nozzle-div (total)22.97 MW
Reflector (total)35.0 MW
Typical NVTR Engine Parameters
Nozzle Area Ratio500
Fuel Pressure100 atm
Average Fuel Temperature4000 K
Maximum Element Heat Flux420 W/cm2
Nomial Element Length150 cm
Fuel Volume Fraction0.15
Coolant Volume Fraction0.15
Moderator Volume Fraction0.70
Fuel Element Power0.9 MWt
Element Heat Transfer Area2141 cm2
Reactor Core L/D1.5
Fuel Channel Diameter0.142 cm
Fuel Channel Sectional Area0.0158 cm2
Total Fuel Channel Area Per Element0.505 cm2
Fuel Element Sectional Area3.464 cm2
Element Diameter (across flats)2.2 cm
Coolant Channel Diameter0.142 cm
Coolant Channel Sectional Area0.0158 cm2
Total Coolant Channel Area Per Element0.505 cm2
Core Volume1.2 m3
Core Volume Density1,500 MW/m3
Fuel Element Mass, Total1.35 MT
Forward Reflector Mass0.60 MT
Aft Reflector Mass0.51 MT
Radial Reflector Mass2.47 MT
Radiation Shield Mass0.90 MT
Total Reactor Mass5.83 MT
Misc. Engine Components Mass0.9 MT
Total Engine Mass6.83 MT
Engine F/W5.0

Gas Core

Closed Cycle

Gaseous Core NTR closed 1
Thrust Power4.5 GW
Exhaust velocity20,405 m/s
Thrust445,000 n
Engine mass56.8 tonne
T/W >1.0no
Gaseous Core NTR closed 2
Thrust Power0.6 to 231 GW
Exhaust velocity10,800 to 31,400 m/s
Thrust117,700 to 14,700,000 n
Engine mass30 to 300 tonne
Engine T/W0.4 to 5.0
Operating Pressure400 to 1600 atm
NASA report nuclear lightbulb
Thrust Power3.7 GW
Engine Power4.6 GW
Exhaust velocity18,300 m/s
Thrust409,000 n
Engine mass32 tonne
Engine T/W1.3
Operating Pressure500 atm
Propellant mass flow22.3 kg/s

Closed-cycle gaseous core fission / nuclear thermal rocket AKA "Nuclear Lightbulb". Similar to an open-cycle gas core fission rocket, but the uranium plasma is confined in a fused quartz chamber. It is sort of like a child's classic Easy-Bake Oven. Except that there is propellant instead of cake mix and the light bulb is full of fissioning uranium instead of electricity.

You can read more about this on the Unwanted Blog in the posts here, here, and here.

The good news is that unlike the open-cycle GCNR it does not spray glowing radioactive death there is no uranium escaping in the exhaust. The bad news is that the maximum exhaust velocity is halved, as is the Delta-V. Yes, I did ask some experts if it was possible to make some kind of hybrid that could "shift gears" between closed and open cycle. Sorry, there would be no savings over just having two separate engines.

The maximum exhaust velocity is halved because you are trying to have it both ways at once. The higher the propellant's temperature, the higher the exhaust velocity and rocket Delta-V. Unfortunately solid core nuclear reactors have a distressing habit of vaporizing at high temperatures (as do all material objects). Gas core reactors are attempting to do an end run around this problem, by having the reactor start out as high temperature vapor. But adding the quartz chamber is re-introducing solid material components into the engine, which kind of defeats the purpose. The only thing keeping this from utter failure is the fact that quartz does not heat up as much due to the fact it is transparent.

Even with the restrictions, it seems possible to make a closed-cycle GCNR with a thrust to weight ratio higher than one. This would allow using the awesome might of the atom to boost truely massive amounts of payload into Terra orbit, without creating a radioactive wasteland with every launch. See the GCNR Liberty Ship for an example. The Liberty Ship can boost in one launch more payload than any given Space Shuttle does in the shuttles entire 10 year operating life. Then the Liberty Ship can land and do it again.

The ideal solution would be to somehow constrain the uranium by something non-material, such as a mangnetohydrodynamic force field or something like that. Alas, currently such fields can only withstand pressures on the order of the breeze from a flapping mosquito, not the 500 atmospheres of pressure found here. But since researchers are working along the same lines in their attempts to make a fusion reactor, this may change. And I did find a brief reference to something called an "MHD choke" in reference to slowing the escape of uranium into the exhaust stream of an open-cycle gas core rocket. I'm still trying to find more information on that.

The high pressure is to ensure the uranium vapor is dense enough to sustain a fission reaction.

NASA Report

The pictured engine is the "reference engine" described in the report Studies of Specific Nuclear Light Bulb and Open-Cycle Vortex-Stabilized Gaseous Nuclear Rocket Engines. I will give the high-lights from the report, but for all the boring nitty-gritty details you'll have to read the report yourself. Please note that as always, "down" is the direction the thrust is going (to the right in the blueprint) and "up" is the opposite direction (to the left). I apologize for the use of imperial units instead of metric. The report is in imperial, and it is too much of a pain to change everything.

The important statistics: Specific impulse is 1870 seconds (18,300 m/s), thrust 409,000 newtons, engine mass 32,000 kg, thrust-to-weight ratio 1.3.

The other stats. The total volume of the reaction chambers (cavities) is 170 cubic feet. There are seven cavities, each six feet long. The cavity pressure is 500 atmospheres. The specific impulse is 1870 seconds (report says can theoretically be from 1500 to 3000 seconds). The total propellant flow (including seed and nozzle transpirant coolant flow) is 49.3 pounds per second. The thrust is 92,000 pounds. The engine power is 4600 megawatts. The engine weight is 70,000 pounds. If one can design a variable-throat-area nozzle (instead of fixed-area) this will result in a major decrease in the required chamber pressure during startup.


The basic configuration is seven separate unit cavities surrounded by moderator-reflector material in between each cavity (beryllium oxide) and surrounding the entire cavity array (graphite). Each cavity is 6.0 feet long and the total volume of all seven cavities is 169.8 cubic feet. The cavity pressure is 500 atmospheres due to criticality and fuel density considerations.


In each lightbulb, a critical mass of gaseous uranium creates thermal radiation. The thermal radiation can pass through the transparent quartz crystal walls of the lightbulb, but the uranium vapor cannot. This means no lethal uranium enters the exhaust. Hydrogen propellant flowing over the lightbulb is heated to high temperatures by the thermal radiation and is expelled out the rocket nozzles, producing thrust. The hydrogen is "seeded" with tungsten dust because it too is ordinarily transparent to thermal radiation. The seeding makes it opaque, and allows it to be heated. Seven "lightbulbs" are used instead of one, since that increases the total lightbulb radiating area by about 2.2 times.

Transparent quartz walls

The transparent quartz wall of the lightbulb contains lots of coolant channels. This is because the quartz is mostly transparent to thermal radiation, but not totally. And fissioning uranium produces an awful lot of thermal radiation. I told you that nuclear lightbulb designers were trying to have it both ways. The coolant channels are marked "circumferential coolant tubes" in the diagram below.

Inside a lightbulb

Inside the lightbulb, neon buffer gas is used to create a vortex ring to suspend the gaseous nuclear fuel (a "radial inflow" vortex). The vortex ring looks like an elongated donut (I know it looks like two separate blobs above, that's due to the fact the diagram is a cross-section). One of the important jobs done by the neon buffer gas is to prevent the 42,000°R uranium plasma from making contact with the lightbulb walls. This would be very bad, as the walls would be instantly vaporized. The neon passes along the lightbulb walls, bends round the end caps, then travels down the long axis of the lightbulb (right down the center of the vortex ring). When it reaches the fore end cap, it is removed from the lightbulb through a port (marked "thru-flow" in diagram above).

The removed neon is very hot, and contains unburnt uranium and fission products. It is cooled by mixing with low-temperature neon, which condenses the unburnt uranium vapor into hot liquid uranium. The liquid uranium is separated from the neon by a centrifuge and sent back into the vortex (at point marked "fuel injection"). The neon is cooled further then it too is sent back into the vortex (at point marked "buffer gas injection"). While examining the blueprint, I noticed that the centrifuges, and indeed the entire uranium fuel delivery system is conspicuous by its absence. Probably classified.

Note that the centrifuges is a neat solution to the problem of fission fragments clogging up the fuel. In essence, this design has its own built-in nuclear fuel reprocessing plant. Of course the nasty fission fragments will have to be stored and eventually disposed of.

Lightbulb dimensions

The total volume inside all the lightbulbs is 84.9 cubic feet, which is 12.1 cubic feet per lightbulb. The radius of the uranium fuel containing region is 85% of the radius of the transparent wall. While the fissioning uranium fuel has a core temperature of 42,000° Rankine, the outer surface is only at 15,000° Rankine.

Propellant flow in a lightbulb

The propellant is assumed to exit with a temperature of 80% of the fuel temperature, or 12,000° Rankine. This is because the quartz transparent walls will reflect about 15% of the thermal radiation back inside. By some compilcated reasoning that you will find in the report, the total thermal radiation from the lightbulbs is 4.37 x 106 BTU/sec. The hydrogen propellant has an "enthalpy" of 1.033 x 105 BTU/pound at 12,000°R. So by dividing the two, one discovers that the entire engine can support a propellant flow rate of 42.3 pounds per second, which means 6.07 lb/sec for each of the seven cavities.

If that last paragraph confused you, let me explain. As a simple example, if a pound of hydrogen at 5°R contains 2 BTUs ("enthalpy"), and the engine puts out 6 BTU per second, then obviously the engine can heat up 6 / 2 = 3 pounds of hydrogen per second. Why do we care? If you multiply the propellant flow rate by the exhaust velocity you will discover the engine's thrust value. And that's a number we do care about.

The tungsten dust that the propellant is seeded with has a particle diameter of 0.05 microns. The seed density is 1.32 x 10-2 lb/ft3, which is about 3.9 percent of the inlet propellant density. This can probably be reduced if tungsten dust was in the form of thin flat plates instead of spherical particles.

The hydrogen propellant enters the pressure shells from the fore end (see "Primary Circuit Inlet" in pressure shell diagram below). A bit is bled off from small H2 flow ports in order to pressurize the interior of the shells, circulating to provide coolant to the engines and machinery. But most of it is fed into the turbopump, then injected into the cavities. Since the fore end of each cavity is almost blocked off by the butt end of the lightbulb, there is only a narrow rim to inject the hydrogen.

In the diagram to the right, you can see how the propellant is fed from the pink pipe into the pink-and-gold wedge-shaped injectors. I presume there are three injectors per cavity, spraying into the clear area between the transparent wall's coolant manifolds and buffer gas injectors.

Uranium fuel

The total fissioning uranium in all seven vortexes be about 25.2 pounds of uranium (about 3.6 pounds per cavity). You would ordinarily need more to ensure nuclear criticality, but the required amount is brought down by the beryllium oxide neutron reflector encasing each cavity. The average uranium fuel density is 0.409 lb/ft3. The total density of the neon-uranium mix inside the vortex is about 0.56 lb/ft3. A unit of neon gas will spend about 3.8 seconds inside the cavity. A unit of uranium will spend about 19 seconds inside the cavity. This implies a uranium fuel flow rate of 0.19 lb/sec per cavity.

According to my slide rule, if the array of seven cavities is producing 4,600 megawatts, it means that the array is burning a miniscule total of 0.055 grams (0.00012 pounds) of uranium fuel per second (0.0079 grams per cavity per second). It still needs the full 3.6 pounds per cavity to be present in order to burn the fraction of a gram.

The theoretical maximum specific impulse possible is 2230 seconds. Due to this designs incomplete expansion, transpiration coolant flow in the nozzle, presence of tungsten seeding, and friction losses the specific impulse is reduced to 84% or 1870 seconds. Total propellant flow (allowing for tungsten seeds and transpiration cooling) is 49.3 lb/sec. This would result in a thrust of 92,000 pounds force. For complicated reasons you can find in the report, this implies that the exhaust nozzles are 0.0875 feet in diameter at the throat expanding to 2.04 feet diameter at the exit.

Uranium refueling

Careful readers may have noticed how the description avoids mentioning the details on how one gets the uranium into the lightbulbs. This is because it is quite a difficult problem, and each of the proposed solutions has drawbacks. The basic problem is old reliable: all the atomic fireworks inherent in 235U will happen if you merely let too much of it accumulate in one place. You have to store it diffuse and somehow bring it together in the lightbulb.

Method #1 Store it as uranium hexafluoride gas. This would be in large tanks of low pressure (i.e., low density) and with the tanks full of neutron absorbing foam. Pump enough into the lightbulb, a chain reaction will start, and well before the reaction reaches 13,000°R the uranium will have separated from the fluorine.

The problem is that now you have the insanely dangerous task of dealing with 13,000°R fluorine gas. At room temperature the blasted stuff will violently react with any element in the known universe except helium and neon. A temperature of 13,000°R makes it about 13,000 times as deadly. It will explosively corrode away anything solid in its path like molten lead on facial tissue. Chemist Derek Lowe sarcastically notes that "At seven hundred freaking degrees, fluorine starts to dissociate into monoatomic radicals, thereby losing its gentle and forgiving nature." You can read more about the suicidal risk of dealing with hot fluorine in his amusing blog post.

Method #2 Store it as sub-critical chunks of uranium, melt them, and inject the molten uranium into the lightbulb. Uranium melts at 1403°K, which is difficult but not impossible. The plan is to somehow turn the molten uranium into a sort of aerosol mist suspended in hot neon.

The problem is that the molten uranium wants to plate itself all over the melter and the aerosol spray equipment. Which is annoying if the material in question is something like lead, but disasterous if the material is radioactive and fissionable.

Method #3 is to store the uranium cold as finely divided dust. As dust it is pumpable, injectable, and it will not plate over everything. Inside the lightbulb the uranium dust will be rapidly heated to vaporization by the nuclear reaction. This method does not have any major problems, except for the common problem of how to protect the transparent wall from being vaporized by the heat.

Again, the uranium delivery system seems to be totally missing from the blueprint. The only bit present is the short stub of the injector at the top of each lightbulb.

Pressure shells

The entire engine is encased in two nested pressure shells constructed of filament-wound fiberglass. The inside of the inner shell is pressurized to 500 atmospheres. Hydrogen propellant enters through a 0.5 foot diameter duct at the fore end (aka "Primary Circuit Inlet"). There are seven 0.4 foot diameter holes in the aft end for the engine nozzles, one at zero degrees off-axis, the other six at 60°. The pressure shell can be separated into two parts along the flange at the point of maximum diameter, to allow an engineer or waldo manipulator access to the engine interior. This point is also where the rear structural grid protrudes from the interior, this is where the engine bolts onto the structural frame of the spacecraft to transmit the engine thrust.

If you look at the large blueprint, you will see that parts of the rear structural grid penetrate the cavities to support the end-caps of the quartz lightbulbs.

Coolant system

The plumbing for the coolant system is rather complicated (translation: I don't understand it all). Click for larger image. You can use this diagram along with the large blueprint to attempt to puzzle out what all the pipes are for. Basically the propellant enters the system through the "Primary circuit inlet" (at lower left of plumbing diagram, and in the pressure shell diagram above) and leaves the system via the "Propellant injection" arrow, where the propellant is heated by the lightbulbs in the cavity and jets out the exhaust nozzles. In between, the propellant frantically threads its way over every single other engine component in a desperate attempt to cool them off.

Propellant flow overview

In the blueprints you can see how the pipes that feed the propellant injectors are originally fed from horns over the graphite moderators. Which is exactly as per the plumbing diagram.

This is my best guess at how the hydrogen propellant flows through the engine. It may be incorrect, use at your own risk. It starts with the green arrow at the left. This is the Primary circuit inlet at the nose of the engine, where the propellant enters the pressure shell. The pipe splits several ways (probably six ways, one for each outer cavity) and enters the base of the turbopump (arrows change color to Yellow).

Pipe runs to the inner shell, then I hypothesize that there is a connection between the two bumps on the inner shell. Propellant runs to the inner pipe array just on top of the cavities, then it is injected into coolant channels in the beryllium oxide moderator around the tie rods. After cooling the beryllium, it spurts out and enters the base of the graphite moderator surrounding the hexagonal beryllium array (arrows change color to orange). It passes through coolant channels in the graphite, and emerges at the top into the collector horns. There it enters the outer pipe array above the inner pipe array.

This feeds the three wedge shaped propellant injectors on each cavity. This injects the propellant around the edge of the transparent light bulbs (arrows change color to red). The propellant shoots aft while being heated by the thermal radiation from the light bulbs. The hot propellant then jets out the exhuast nozzles and thrust occurs.

Cross sections

Here are a set of cross sections through the cavities. The one on the left is zoomed in on the cavity interior, the other two gradually zoom out.

Open Cycle

Thrust Power157 GW
Exhaust velocity17,658 m/s
Thrust17,800,000 n
Engine mass127 tonne
T/W >1.0yes
Thrust Power0.495GW
Exhaust velocity22,000 m/s
Thrust45,000 n
Engine mass66 tonne

Gaseous core coaxial flow fission / nuclear thermal rocket.

Circa 1960 NASA-Lewis concept for a gas core nuclear rocket engine. Specific Impulse 2200 seconds (exhaust velocity 22,000 m/s). Thrust 45,000 newtons. Thrust to weight ratio 0.68 (engine mass 66,000 kilograms), reactor diameter 5 meters, overall reactor length 5 meters. The fuel would reach 20,000 degrees R, while the propellant would get to 10,000 degrees R. From The Unwanted Blog.

Open Cycle
Open Cycle
Thrust Power61 GW
Exhaust velocity35,000 m/s
Thrust3,500,000 n
Engine mass30-200 tonne
T/W 11.9 to 1.8
Open Cycle 2
Thrust Power100 GW
Exhaust velocity50,000 m/s
Thrust5,000,000 n
Engine mass30-200 tonne
T/W 17.0 to 2.5
Open Cycle 3
Thrust Power GW
Exhaust velocity25,000 to 69,000 m/s
Thrust19,600 to 108,000 n
Engine mass40 to 110 tonne
T/W0.05 to 0.10
Operating Pressure400 to 2000 atm
Open Cycle MAX
Thrust Power150 GW
Exhaust velocity98,000 m/s
Thrust3,000,000 n
Engine mass15 tonne
T/W 20.4

Gaseous core fission / nuclear thermal rocket AKA consumable nuclear rocket, plasma core, fizzler, cavity reactor rocket. The limit on NTR-SOLID exhaust velocities is the melting point of the reactor. Some engineer who obviously likes thinking "outside of the box" tried to make a liability into a virtue. They asked the question "what if the reactor was already molten?"

Gaseous uranium is injected into the reaction chamber until there is enough to start a furious chain reaction. Hydrogen is then injected from the chamber walls into the center of this nuclear inferno where is flash heats and shoots out the exhaust nozzle.

The trouble is the uranium shoots out the exhaust as well.

The reaction is maintained in a vortex tailored to minimize loss of uranium out the nozzle. Fuel is uranium hexaflouride (U235F6), propellant is hydrogen. However, in one of the designs, U235 is injected by gradually inserting into the fireball a long rod of solid uranium. The loss of uranium in the exhaust reduces efficiency and angers environmentalists.

Open-cycle gas core nuclear thermal rockets have a nominal core temperature of 21,000 K (37,340°F).

In some designs the reaction chamber is spun like a centrifuge. This encourages the heavier uranium to stay in the chamber instead of leaking into the exhaust. This makes for a rather spectacular failure mode if the centrifuge's bearings seize.

You can find more details here.

If used for lift off it can result in a dramatic decrease in the property values around the spaceport, if not the entire county. An exhaust plume containing radioactive uranium is harmless in space but catastrophic in Earth's atmosphere.

Amusingly enough, this is the best match for the propulsion system used in the TOM CORBETT: SPACE CADET books. However the books are sufficiently vague that it is possible the Polaris used a nuclear lightbulb. According to technical advisor Willy Ley, "reactant" is the hydrogen propellant, but the books imply that reactant is the liquid uranium.

Nuclear Salt Water
20% UTB
Thrust Power427 GW
Exhaust velocity66,000 m/s
Thrust12,900,000 n
Engine mass33 tonne
T/W >1.0yes
90% UTB
Thrust Power31,000 GW
Exhaust velocity4,700,000 m/s
Thrust13,000,000 n
T/W >1.0yes

This concept by Dr. Zubrin is considered far-fetched by many scientists. The fuel is a 2% solution of 20% enriched Uranium Tetrabromide in water. A Plutonium salt can also be used.

Just to make things clear, there are two percentages here. The fuel is a 2% solution of uranium tetrabromide and water. That is, 2 molecules of uranium tetrabromide per 100 molecules of water.

But the uranium tetrabromide can be 20% enriched. This means that out of every 100 atoms of uranium (or molecules of uranium tetrabromide), 20 are fissionable Uranium-235 and 80 are non-fissionable uranium. If it is 90% enriched, then 90 atoms are Uranium-235 and 10 atoms are non-fissionable. As a side note, 90% enriched is considered "weapons-grade".

The fuel tanks are a bundle of pipes coated with a layer of boron carbide neutron damper. The damper prevents a chain reaction. The fuel is injected into a long cylindrical plenum pipe of large diameter, which terminates in a rocket nozzle. Free of the neutron damper, a critical mass of uranium soon develops. The energy release vaporizes the water, and the blast of steam carries the still reacting uranium out the nozzle.

It is basically a continuously detonating Orion type drive with water as propellant. Although Zubrin puts it like this:

As the solution continues to pour into the plenum from the borated storage pipes, a steady-state conditions of a moving detonating fluid can be set up within the plenum.

He also notes that it is preferable to subject your spacecraft to a steady acceleration (as with the NSWR) as opposed to a series of hammer-blow accelerations (as with Orion).

The controversy is over how to contain such a nuclear explosion. Zubrin maintains that skillful injection of the fuel can force the reaction to occur outside the reaction chamber. He says that the neutron flux is concentrated on the downstream end due to neutron convection. Other scientists are skeptical.

Naturally in such a spacecraft, damage to the fuel tanks can have unfortunate results (say, damage caused by hostile weapons fire). Breach the fuel tubes and you'll have a runaway nuclear chain reaction on your hands. Inside your ship.

The advantage of NSWR is that this is the only known propulsion system that combines high exhaust velocity with high thrust. The disadvantage is that it combines many of the worst problems of the Orion and Gas Core systems. For starters, using it for take-offs will leave a large crater that will glow blue for several hundred million years, as will everything downwind in the fallout area.

Zubrin calculates that the 20% enriched uranium tetrabromide will produce a specific impulse of about 7000 seconds (69,000 m/s exhaust velocity), which is comparable to an ion drive. However, the NSWR is not thrust limited like the ion drive. Since the NSWR vents most of the waste heat out the exhaust nozzle, it can theoretically produce jet power ratings in the thousands of megawatts. Also unlike the ion drive, the engine is relatively lightweight, with no massive power plant required.

Zubrin suggests that a layer of pure water be injected into the plenum to form a moving neutron reflector and to protect the plenum walls and exhaust nozzle from the heat. One wonders how much protection this will offer.

Zubrin gives a sample NSWR configuration. It uses as fuel/propellant a 2% (by number) uranium bromide aqueous solution. The uranium is enriched to 20% U235. This implies that B2 = 0.6136 cm-2 (the material buckling, equal to vΣfa)/D) and D = 0.2433 cm (diffusion coefficent).

Radius of the reaction plenum is set to 3.075 centimeters. this implies that A2 = 0.6117 cm-2 and L2 = 0.0019. Since exponential detonation is desired, k2 = 2L2 = 0.0038 cm-2. Then k = U / 2D = 0.026 cm-1 and U = 0.03.

If the velocity of a thermal neutron is 2200 m/s, this implies that the fluid velocity needs to be 66 m/s. This is only about 4.7% the sound speed of room temperature water so it should be easy to spray the fuel into the plenum chamber at this velocity.

The total rate of mass flow through the plenum chamber is about 196 kg/s.

Complete fission of the U235 would yield about 3.4 x 1012 J/kg. Zubrin assumes a yield of 0.1% (0.2% at the center of the propellant column down to zero at the edge), which would not affect the material buckling during the burn. This gives an energy content of 3.4 x 109 J/kg.

Assume a nozzle efficiency of 0.8, and the result is an exhaust velocity of 66,000 m/s or a specific impules of 6,7300 seconds. The total jet power is 427 gigawatts. The thrust is 12.9 meganewtons. The thrust-to-weight ratio will be about 40, which implies an engine mass of about 33 metric tons.

For exponetial detonation, kz has to be about 4 at the plenum exit. Since k = 0.062 cm-1, the plenum will have to be 65 cm long. The plenum will be 65 cm long with a 3.075 cm radius, plus an exhaust nozzle.

Zubrin then goes on to speculate about a more advanced version of the NSWR, suitable for insterstellar travel. Say that the 2% uranium bromide solution used uranium enriched to 90% U235 instead of only 20%. Assume that the fission yield was 90% instead of 0.1%. And assume a nozzle efficency of 0.9 instead of 0.8.

That would result in an exhaust velocity of a whopping 4,725,000 m/s (about 1.575% c, a specific impulse of 482,140 seconds). In a ship with a mass ratio of 10, it would have a delta V of 3.63% c. Now you're talkin...

Ken Burnside: In my game universe, the engineers call the pumps that feed Uranium Tetrabromide solution into the reaction chamber "Wileys", reputedly after the engineer who first made them safe to use and maintain.

More than likely, it's after the coyote of the same name...

Winchell Chung: An appropriate name for what are basically atomic squirt-guns.

From a thread in SFConSim-l (2002)

You need much more propellant than fuel, 22,000 times more in the case of the Zubrin without open cycle cooling, and 44,000 times more if open cycle cooling is used.

The Zubrin drive exhaust (without open cycle cooling) contains 108 kg/sec of water, but only about 5 grams/sec of uranium.

(This is from a quick calculation: mass flow equals the Zubrin thrust (8.7 meganewtons) divided by the exit velocity (80 km/sec) = 108 kg/sec. But the fissioning energy can be estimated from the Zubrin total power of 427 GW divided by the energy content of Uranium 235 of 83 TJ/kg.)

Dr. Zubrin responded, and he defends the performance of the Zubrin drive as depicted in the game (as high thrust & high specific impulse rocket with low mass and low radiators).

1). In U235 fission, only about 2% of the energy goes into neutrons (unlike D-T fusion).

2). The design uses a pusher plate or open nozzle, like an Orion drive. Or magnetic confinement (since most of the energy is released as a plasma). Therefore, the opportunity to absorb heat is low.

3) Many of the neutrons that are intercepted would sail through the pusher plate, rather than be absorbed as waste heat.

4) No lithium should be in the outer water, because this would poison the fission reactions.

5). Because the design does not use a heat engine cycle, the radiators could be far hotter than ones in the game. He suggested graphite at 2500 K°. That would drop the required radiating area by a factor of 40 (2.5 to the fourth power), which means that the radiator could be the first wall itself.

Dr. Zubrin went on to say the chief disadvantage is the expense of the fuel (like He3-D and antimatter drives).

Philip Eklund, from a discussion on the High Frontier Yahoo group about the NSWR drive in High Frontier

"So anyway, we were passing through the outer Kirkwood Gap, totally the a** end of nowhere. I'm trying to catch some rack time, XO has the conn, nice boring trip to Europa." The CO of U.S.N.A.S. Saskatchewan tipped back another shot of Scotch and continued his story. "Totally routine, right? No problems at all. So then, all of the sudden, the whole ship gets racked. Meteoroid. Big one, too, maybe a centimeter across."

The captains seated around the table, two Americans including Fitzthomas, an Indian, three Chinese, and the South African, all clucked and groaned.

"Well, we got lucky and it missed the crew compartment, but by the time I get to command the chief engineer is screaming over the intercom that it holed tank one, busted three tubes, and we've got nuke juice pooling and we have to dump the tank. Problem is, we're running at top speed and if we dump the tank, we don't have enough propellant to stop at Europa. We'd have to ride all the way out to Neptune, sling around, and hope someone from the inner solar system has dispatched a tanker to intercept us on the return trip, and we don't have near enough consumables for that."

"So what did you do, mate?" said the South African.

"I told the chief he had to fix the tank or we'd all starve before we could stop the damn ship. Well, he screams some more that we don't have time, and I tell him his choice is fix the tank or die real fast in a runaway, because we're not going to die slow in the void. So he grabs a crew, stuffs them into suits, and crawls out onto the tank. They punch some holes in it to let the juice drain instead of pool, but it's still leaking like a f***** and the water's evaporating and leaving uranium crusted all over everything. So he radios command and says, 'It's still leaking, and all this uranium crud is going to accumulate into a critical mass somewhere, so we still have to drop the tank.' Meteoroid busted open three valves, you see. No way to stop the leak. And I tell him again, that's no good, and by now astrogation has confirmed it and the XO has tallied up the consumables and I know for sure we don't have enough for an unscheduled trip to Neptune.

"So he says something about how he's not a miracle worker, and I tell him he damn well has to be. Lo and behold, he and his crew go ahead and do something crazy and it works."

"What was that?" said the South African.

"They take torches to the tank. The plug up the broken pipes as best they can, and then they go ahead and cut away the smashed cells. Just cut it off and jettison it into space, and suddenly the propellant that's still leaking is leaking right into space. We have lousy flow through the tank and the braking burn is going to be real tricky, but we can make Europa. I put the chief up for a commendation medal for figuring that out on the fly and saving our asses."

The other captains nodded their approval at the chief's quick thinking. Good chiefs prevented accidents; great ones prevented disasters.

"Is the chief's name Mr. Scott, by any chance?" said one of the Chinese captains.

Commander George Allen, New Jersey's full blooded Cherokee XO, drifted into the command deck from astrogation, where he'd been monitoring the final approach to Hektor. He took his place at the copilot station and put on his headset. Fitzthomas toggled his direct channel to Allen's station.

"How was the approach?"

"We wasted too much propellant before the chain reaction started. I think Pennai should inspect the nozzles and pumps before we get underway again."

"What does Pennai say about it?"

Pause. "Pennai thinks the fuel is dirty."

"Is it?"

"It was certified 90% enriched at Roosevelt Station."

"Is there any way to test it here?"

"No sir. Not without a centrifuge."

"How does Pennai know, then?"

"Some engineering technobabble about neutron flux and reaction rate. I couldn't follow a tenth of what she said."

Fitzthomas considered that. "Have her inspect the pumps and nozzle alignment. If they pass, then we might have a fuel problem."

"Captain," said Allen, "Thought you'd like to know: Pennai just inspected the entire fuel line. Everything there is in order."

"So what are you telling me, George?"

"I think we have dirty fuel."

"What's her recommendation?"

"She wants to drain the tanks and top up with the good stuff. But I can't—"

"Write that request, I know. The CO has to. Where's Pennai now?"

"Racked out. She has the midwatch tonight."

"After her watch tonight, she has four days of leave."

"Sir, she's supposed to be OOW all day Wednesday."

"I'll take that shift. She was right, we were wrong. She deserves to be rewarded. When I get back I'll write up a request and have it to the fuelmaster by tomorrow AM."

"Do you have to return to your ship?"

"Yeah. Dirty fuel, God damn it. Wait until I get my hands on the fuelmaster at Roosevelt."

(Admiral Castro said) "Anyway, I saw your chief engineer's report. I passed it back to Fleet. The fuelmaster at Roosevelt Station is going to have a lousy day tomorrow. There's also a bulletin going out to the entire fleet. Everyone who tanked up at Roosevelt near the same time you did should keep a close eye on his reaction rate. Your Lieutenant Pennai might be up for a commendation letter in her file."

Duvalier left Ortiz main engineering and vaulted down the access tube to the reactor room. The tube ran down the ship's spine, surrounded by megaliters of water enriched with uranium salts in highly complex tanks made of neutron absorbing material. In his head, he knew the tube was the safest part of the ship, shielded from the worst the universe could throw at it by dozens of meters of water. In his head, he knew the fuel, so long as it didn't pool into a critical mass somewhere in the thousands of kilomters of pipes on all sides of him, emitted only low intensity alpha rays which couldn't penetrate his own skin, let alone the aluminum skin of the pressure tube. It was all perfectly safe, so far as anything in space could be safe. He knew that in his head.

His b***s, however, hadn't gotten the memo. His testicles tried to crawl up into his body every time he climbed through the hatch.

From The Last Great War by Matthew Lineberger (not yet published)

Fission Fragment

Fission Fragment

George Chapline
Exhaust velocity980,000 m/s

All of the other nuclear thermal rockets generate heat with nuclear fission, then transfer the heat to a working fluid which becomes the reaction mass. The transfer is always going to be plagued by inefficiency, thanks to the second law of thermodynamics. What if you could eliminate the middleman, and use the fission heat directly with no transfer?

That what the fission fragment rocket does. It uses the hot split atoms as reaction mass. The down side is that due to the low mass flow, the thrust is minuscule. But the up side is that the exhaust velocity is 5% the speed of light! 15,000,000 kilometers per second, that's like a bat out of hell. With that much exhaust velocity, you could actually have a rocket where less than 50% of the total mass is propellant (i.e., a mass ratio below 2.0).

Dr. Chapline's design use thin carbon filaments coated with fission fuel (coating is about 2 micrometers thick). The filaments radiated out from a central hub, looking like a fuzzy vinyl LP record. These revolving disks were spun at high speed through a reactor core, where atoms of nuclear fuel would undergo fission. The fission fragments would be directed by magnetic fields into an exhaust beam.

The drawback of this design is that too many of the fragments fail to escape the fuel coat (which adds no thrust but does heat up the coat) and too many hit the carbon filaments (which adds no thrust but does heat up the filaments). This is why the disks spin at high speed, otherwise they'd melt.

Dusty Plasma
Thrust Power0.16 GW
Thrust22 n
Thrust Power2.6 GW
Thrust344 n
Exhaust velocity15,000,000 m/s
Engine mass9 tonne

Rodney Clark and Robert Sheldon solve the problem with their Dusty plasma bed reactor (report).

You take the fission fuel and grind it into dust grains with an average size of 100 nanometers (that is, about 1/20th the thickness of the fuel coating in dr. Chapline's design). This does two things [A] most of the fragments escape and [B] the dust particles have such a high surface to volume ratio that heat (caused by fragments which fail to escape) readily dissipates, preventing the dust particles from melting.

The dust is suspended in the center of a reaction chamber whose walls are composed of a nuclear moderator. Power reactors will use beryllium oxide (BeO) as a moderator, but that is a bit massive for a spacecraft. The ship will probably use lithium hydride (LiH) for a moderator instead, since is only has one-quarter the mass. Probably about six metric tons worth. The dust is suspended electrostatically or magnetically by a containment field generator. The dust is heated up by radio frequency (RF) induction coils. The containment field generator will require superconductors, which will probably require a coolant system of its own.

The dust particles are slow and are relatively massive, while the fission fragments are fast and not very massive at all. So the magnetic field can be tailored so it holds the dust but allows the fission fragments to escape. Magnetic mirrors ensure that fragments headed the wrong way are re-directed to the exhaust port.

One valuable trick is that you can use the same unit for thrust or to generate electricity. Configure the magnetic field so that the fragments escape "downward" through the exhaust port and you have thrust. Flip a switch to change the magnetic field so that the fragments escape upward into deceleration and ion collection electrodes and you generate electricity. As a matter of fact, it is go efficient at generrating electricity that researchers are busy trying to adapt this for ground based power plants. But I digress.

The dust is only sufficient for a short period of critical nuclear reaction so it must be continuously replenished. The thermal energy released by fission events plus heat from collisions between fission fragments and dust grains create intense heat within the dust cloud. Since there is no core cooling flow, the reactor power is limited to the temperature at which the dust can radiatively cool itself without vaporizing. The interior of the reaction chamber walls will protected by a mirrored (95% reflection) heat shield attached to a heat radiator. The outer moderator layer will have its own heat shield.

Clark and Sheldon roughed out a propulsion system. It had six tons for the moderator, 2 tons for radiators and liquid metal cooling, 1 ton for magnets, power recovery, and coils, for a grand total of 9 tons. The reaction chamber will be about 1 meter in diameter and 10 meters long. The moderator blanket around the chamber will be about 40 centimeters thick. The thrust is a function the size of the cloud of fissioning dust, and is directly related to the power level of the reactor. There is a limit to the maximum allowed power level, set by the coolant system of the reaction chamber. Clark and Sheldon estimate that only about 46% of the fission fragments provide thrust while the rest are wasted. See the report for details.

In the table, the 550AU engine is for a ten year journey to the Solar gravitational lensing point at 550 astronomical units (so you can use the sun as a giant telescope lens). The 0.5LY engine is for a thirty year trip to the Oort cloud of comets. These are constant acceleration brachistochrone trajectories, the 550AU mission will need a reactor power level of 350 MW and the 0.5LY mission will need 5.6 GW. Don't forget that the engine power is only 46% efficient, that's why the table thrust values are lower.

Werka FFRE
First Generation
Thrust Power0.111 GW
Exhaust velocity5,170,000 m/s
Thrust43 n
Engine mass113.4 tonne
Reactor Power1.0 GW

Robert Werka has a more modest and realistic design for his fission fragment rocket engine (FFRE). He figures that a practical design will have an exhaust velocity of about 5,200,000 m/s instead of his estimated theoretical maximum of 15,000,000 m/s. His lower estimate is still around 1.7% the speed of light so we are still talking about sub 2.0 mass ratios. Collisions between fission fragments and the dust particles is responsible for the reduction in exhaust velocity.

Incidentally the near relativistic exhaust velocity reduces radioactive contamination of the solar system. The particles are traveling well above the solar escape velocity (actually they are even faster than the galactic escape velocity) so all the radioactive exhaust goes shooting out of the solar system at 0.017c.

The dusty fuel is nanometer sized particles of slightly critical plutonium carbide, suspended and contained in an electric field. A moderator of deuterated polyethylene reflects enough neutrons to keep the plutonium critical, while control rods adjust the reaction levels. The moderator is protected from reaction chamber heat by a heat shield, an inner layer composed of carbon-carbon to reflect infrared radiation back into the core. The heat shield coolant passes through a Braydon cycle power generator to create some electricty, then the coolant is sent to the heat radiator.

The details of Werka's initial generation FFRE can be found in the diagram below. The reaction chamber is about 5.4 meters in diameter by 2.8 meters long. The magnetic nozzle brings the length to 11.5 meters. The fuel is uranium dioxide dust which melts at 3000 K, allowing a reactor power of 1.0 GW. It consume about 29 grams of uranium dioxide dust per hour (not per second). Of the 1.0 GW of reactor power, about 0.7 GW of that is dumped as waste heat through the very large radiators required.

The second most massive component is the magnetic mirror at the "top" of the reaction chamber. Its purpose is to reflect the fission fragments going the wrong way so they turn around and travel out the exhaust nozzle. Surrounding the "sides" of the reaction chamber is the collimating magnet which directs any remaining wrong-way fragments towards the exhaust nozzle. The exhaust beam would cause near-instantaneous erosion of any material object (since it is electrically charged, relativistic, radioactive grit). It is kept in bounds and electrically neutralized by the magnetic nozzle cage.

Fission Sail

Fission Sail

Antimatter-Driven Sail

The sail is made of graphite and carbon-carbon fiber, infused with a tiny amount of uranium. It is subjected to a misting of antiprotons. These induce uranium atoms to fission, with the recoil pushing the sail. Since this is nuclear powered, the sail does not have to be kilometers in diameter, five meters will do. 30 miligrams of antiprotons could push the sail to the Kuiper Belt.



Fission Orion
Thrust Power5.7 GW
Exhaust velocity43,000 m/s
Thrust263,000 n
Engine mass200 tonne
T/W >1.0no
Fusion Orion
Thrust Power10.7 GW
Exhaust velocity73,000 m/s
Thrust292,000 n
Engine mass200 tonne
T/W >1.0no
1959 Orion 1st Gen
Thrust Power1,600 GW
Exhaust velocity39,000 m/s
Thrust80,000,000 n
Engine mass1,700 tonne
T/W >1.0yes
1959 Orion 2nd Gen
Thrust Power24,000 GW
Exhaust velocity120,000 m/s
Thrust400,000,000 n
Engine mass3,250 tonne
T/W >1.0yes
Orion MAX
Thrust Power39,000 GW
Exhaust velocity9,800,000 m/s
Thrust8,000,000 n
Engine mass8 tonne
T/W >1.0yes

Orion AKA "old Boom-boom" is the ultimate consumable nuclear thermal rocket, based on the "firecracker under a tin can" principle. Except the tin can is a spacecraft and the firecracker is a nuclear warhead.

This concept has the spacecraft mounted with shock absorbers on an armored "pusher plate". A stream of small (5 to 15 kiloton) fission or fusion explosives are detonated under the plate to provide thrust. While you might find it difficult to believe that the spacecraft can survive this, you will admit that this will give lots of thrust to the spacecraft (or its fragments). On the plus side, a pusher plate that can protect the spacecraft from the near detonation of nuclear explosives will also provide dandy protection from any incoming weapons fire. On the minus side I can hear the environmentalists howling already. It will quite thoroughly devastate the lift-off site, and give all the crew bad backs and fallen arches. And they had better have extra-strength brassieres and athletic supporters.

Mathematician Richard Courant viewed an Orion test and said "Zis is not nuts, zis is super-nuts."

Some environmentalists howl that Orion should never be used for surface-to-orbit boosts, due to the danger of DUNT-dunt-Dunnnnnnnn Deadly Radioactive Fallout. However, there is a recent report that suggests ways of minimizing the fallout from an ORION doing a ground lift-off (or a, wait for it, "blast-off" {rimshot}). Apparently if the launch pad is a large piece of armor plate with a coating of graphite there is little or no fallout.

A more sophisticated objection to using Orion inside an atmosphere is the sci-fi horror of EMP melting all our computers, making our smart phones explode, and otherwise ruining anything using electricity. But that actually is not much of a problem. EMP is not a concern unless the detonation is larger than one megaton or so, Orion propulsion charges are only a few kilotons (one one-thousandth of a megaton). Ben Pearson did an analysis and concluded that Orion charges would only have EMP effects within a radius of 276 kilometers (the International Space Station has an orbital height of about 370 kilometers). So just be sure your launch site is in a remote location, which you probably would have done anyway.

If you want the real inside details of the original Orion design, run, do not walk, and get a copy of Aerospace Projects Review issue Volume 2, Number 2. It has blueprints, tables, and lots of never before seen details. If you want your data raw, piled high and dry, here is a copy of report GA-5009 vol III "Nuclear Pulse Space Vehicle Study - Conceptual Vehicle Design" by General Atomics (1964). Lots of charts, lots of graphs, some very useful diagrams, almost worth skimming through it just to admire the diagrams.

The sad little secret about Orion is that the mission it is best suited for is boosting heavy payloads into orbit. Which is exactly the mission that the enviromentalist and the nuclear test ban treaty will prevent. Orion has excellent thrust, which is what you need for lift-off and landing. Unfortunately its exhaust velocity is pretty average, which is what you need for efficient orbit-to-orbit maneuvers.

Having said that, there is another situation where high thrust is desirable: a warship jinking to make itself harder to be hit by enemy weapons fire. It is also interesting to note that the Orion propulsion system works very well with the bomb-pumped laser weapons system.

Each pulse unit is a tiny nuclear bomb, encased in a "radiation case" that has a hole in the top. A nuclear blasts is initially mostly x-rays. The radiation case is composed of a material that his opaque to x-rays (depleted uranium). The top hole thus "channels" the flood of x-rays in an upwards direction (at least in the few milliseconds before the bomb vaporizes the radiation case).

The channeled x-rays then strike the "channel filler" (beryllium oxide). The channel filler transforms the atomic fury of x-rays into an atomic fury of heat.

Lying on top of the channel filler is the disc of propellant (tungsten). The atomic fury of heat flashes the tungsten into a jet of ionized tungsten plasma, traveling at high velocity (in excess of 1.5 × 105 meters per second). This crashes into the pusher plate, accelerating the spacecraft. It crashes hard. You will note that there are two stages of shock absorbers between the pusher plate and the spacecraft, preventing instant crew death.

The ratio of beryllium oxide to tungsten is 4:1.

The jet is confined to a cone about 22.5 degrees (instead of in all directions). The detonation point is positioned such that the 22.5 cone exactly covers the diameter of the pusher plate. The idea is that the wider the area of the cone, the more spread out the impulse will be, and the larger the chance that the pusher plate will not be utterly destroyed by the impulse.

It is estimated that 85% of the energy of the nuclear explosion can be directed in the desired direction. The pulse units are popped off at a rate of about one per second. A 5 kiloton charge is about 1,152 kg. The pulses are so brief that there is no appreciable "neutron activation", that is, the neutron from the detonations do not transmute parts of the spacecraft's structure into radioactive elements. This means astronauts can exit the spacecraft and do maintenance work shortly after the pulse units stop detonating.

The device is basically a nuclear shaped charge. A pulse unit that was not a shaped charge would of course waste most of the energy of the explosion. Figure that 10% at best of the energy of a non-shaped-charge explosion would actually hit the pusher plate, what a waste of perfectly good plutonium.

Each charge accelerates the spacecraft by roughly 12 m/s. A 4,000 ton spacecraft would use 5 kiloton charges, and a 10,000 ton spacecraft would use 15 kiloton charges. For blast-off, smaller charges of 0.15 kt and 0.35 kt respectively would be used while within the Terra's atmosphere. The air between the charge and the pusher plate amplifies the impulse delivered, so if you are not in airless space you can get away with a smaller kt yield.

How much weapons-grade plutonium will each charge require? As with most details about nuclear explosives, specifics are hard to come by. According to GA-5009 vol III , pulse units with 2.0 × 106 newtons to 4.0 × 107 newtons all require approximately 2 kilograms per pulse unit, with 1964 technology. It goes on to say that advances in the state of the art could reduce the required amount of plutonium by a factor 2 to 4, especially for lower thrust units. 2.0 × 106 n is 1 kiloton, I'm not sure what 4.0 × 107 n corresponds to, from the document I'd estimate it was about 15 kt. Presumably the 2 kg plutonium lower limit is due to problems with making a critical mass, you need a minimum amount to make it explode at all.

According to Scott Lowther, the smallest pulse units were meant to propel a small ten-meter diameter Orion craft for the USAF and NASA. The units had a yield ranging from one-half to one kiloton. The USAF device was one kiloton, diameter 36 centimeters, mass of 86 kilograms, tungsten propellant mass of 34.3 kilograms, jet of tungsten plasma travels at 150,000 meters per second. One unit would deliver to the pusher plate a total impulse of 2,100,000 newton-seconds. Given the mass of the ten-meter Orion, detonating one pulse unit per second would give an acceleration well over one gee. According to my slide rule, this implies that the mass of the ten-meter Orion is a bit under 210 metric tons.

Pulse UnitYieldMassDiameterHeightPropellantJet VelocityThrust
NASA Orion 10m plate - vacuum charge1 kt141 kg0.36 m0.6 m90? kg? m/s3.5 × 106 n
USAF Orion 10m plate - vacuum charge1 kt79 kg0.36 m0.6 m34.3 kg150,000 m/s2.0 × 106 n
Orion 20m plate - vacuum charge5? kt450 kg? m? m? kg? m/s1.6 × 107 n
4,000 ton Orion 41m plate - atm charge0.15 kt? kg0.81 m0.86 m? kg? m/s? n
4,000 ton Orion 41m plate - vacuum charge5 kt1,152 kg0.81 m0.86 m? kg39,000 m/s8.0 × 107 n
10,000 ton Orion 56m plate - atm charge0.35 kt? kg? m? m? kg? m/s? n
10,000 ton Orion 56m plate - vacuum charge15 kt? kg? m? m? kg120,000 m/s4.0 × 108 n
20,000? ton Orion ?m plate - vacuum charge29 kt1,150 kg0.8 m? m? kg? m/s? n

Tungsten has an atomic number (Z) of 74. When the tungsten plate is vaporized, the resulting plasma jet has a relatively low velocity and diverges at a wide angle (22.5 degrees). Now, if you replace the tungsten with a material with a low Z, the plasma jet will instead have a high velocity at a narrow angle. The jet angle also grows narrower as the thickness of the plate is reduced. This makes it a poor propulsion system, but an effective weapon. Instead of a wall of gas hitting the pusher plate, it is more like a directed energy weapon. The military found this to be fascinating, who needs cannons when you can shoot spears of pure nuclear flame? The process was examined in a Strategic Defense Initiative project called "Casaba-Howitzer", which apparently is still classified. Which is not surprising but frustrating if one is trying to write a science fiction novel or spacecraft combat game.

For more in depth calculations of an Orion rocket's specific impulse, read page 1 and page 2. But be prepared for some heavy math.

NASA has been quietly re-examining ORION, under the new name of "External Pulsed Plasma Propulsion". As George Dyson observed, the new name removes most references to "Nuclear", and all references to "Bombs."

For details about spacecraft using Orion propulsion, go here.

Mini-Mag Orion

Mini-Mag Orion
Thrust Power147 GW
Exhaust velocity157,000 m/s
Thrust1,870,000 n
T/W >1.0no

The Mini-MagOrion is a sort of micro-fission Orion propulsion system. The idea was to make an Orion with weaker (and more reasonably sized) explosive pulses, using pulse charges that were not self contained (the full Orion pulse units were nothing less than nuclear bombs). Subcritical hollow spheres of curium-245 are compressed by a Z-pinch magnetic field until they explode. The sacrificial Z-pinch coil in each pulse charge is energized by an a huge external capacitor bank mounted in the spacecraft. So the pulse units are not bombs.

The explosion is caught by a superconducting magnetic nozzle.

More details are in the Realistic Designs section.


Exhaust velocity490,000 m/s
to 980,000 m/s

Medusa is driven by the detonation of nuclear charges like Orion, except the charges are set off in front of the spacecraft instead of behind. The spacecraft trails behind a monstrously huge parachute shaped sail (about 500 meters). The sail intercepts the energy from the explosion. Medusa performs better than the classical Orion design because its pusher plate intercepts more of the bomb's blast, its shock-absorber stroke is much longer, and all its major structures are in tension and hence can be quite lightweight. It also scales down better. The nuclear charges will be from 0.025 kilotons to 2.5 kilotons.

The complicated stroke cycle is to smooth out the impulses from each blast, transforming it from a neck-braking jerk into a prolonged smooth acceleration.

Jondale Solem calculates that the specific impulse is a function of the mass and yield of the nuclear charges, while the thrust is a function of the yield and explosion repetition rate. In this case, the mass of the nuclear charge is the mass of "propellant".

Remarkably the mass of the spinnaker (sail) is independent of the size of its canopy or the number or length of its tethers. This means the canopy can be made very large (so the bomb blast radiation does not harm the canopy) and the tethers can be made very long (so the bomb blast radiation does not harm the crew). The mass of the spinnaker is directly proportional to the bomb yield and inversely proportional to the number of tethers.

Inertial Confinement

Thrust Power500,000 GW
Exhaust velocity10,000,000 m/s
Thrust100,000,000 n
Engine mass1000 tonne
T/W >1.0yes

A pellet of fusion fuel is bombarded on all sides by strong pulses from laser or particle accelerators. The inertia of the fuel holds it together long enough for most of it to undergo fusion.

Magneto Inertial Fusion

Magneto Inertial Fusion
Low Gear
Thrust Power0.0026 GW
Exhaust velocity50,420 m/s
Solar Power req27 kW
Thrust103 n
Delay between Fusion Pulses180 seconds
High Gear
Thrust Power0.35 GW
Exhaust velocity50,420 m/s
Solar Power req350 kW
Thrust13,800 n
Delay between Fusion Pulses14 seconds
Exhaust velocity23,940 m/s
to 56,110 m/s
T/W >1.0no

There are two main approaches to utilizing nuclear fusion, magnetic confinement and inertial confinement. Magnetic confinement uses titanic magnetic fields, inertial confinement is how fusion bombs explode (a third way would be stars shining by gravitational confinement, but we don't know how to generate artificial gravitational fields). As propulsion systems, both have major drawbacks.

Magnetic confinement requires huge (read: massive) electromagnets. The technique also has the problem of plasma instabilities (read: fusion plasma has thousands of different ways to wiggle out of the magnetic cage) which so far have defied any solution.

Inertial confinement works well in bombs, but trying to do it in a small controlled fashion (read: so the fusion reaction does not vaporize everything in a one kilometer radius) has also defied any solution. The compressing laser or particle beams have such low efficiencies that tons of excess power is required. Timing all the beams so they strike at the same instant is a challenge. Also, there is nothing in between the fusion reaction and the chamber walls, leading to severe damage to the walls.

Both approaches have a problem with getting the fusion reaction energy to heat the propellant. Magnetic confinement tries to use the actual fusion plasma as propellant, resulting in a ridiculously small mass flow and thus a tiny thrust.

Dr. John Slough and his associates have come up with a new technique that sort of combines the two conventional approaches: magneto inertial fusion (MIF). You can find their published papers on the subject here

A blob of FRC (field reversed configuration) plasma is created and injected axially into the chamber.

Simultaneously injected into the chamber is a "liner". The liner is a foil ring composed of lithium, about 0.2 meters in radius. Each liner will have a mass of 0.28 kg (minimum) to 0.41 kg.

As the liner travels axially down the chamber, electromagnets crush it down into a solid cylinder (the crush speed is about 3 kilometers per second, the cylinder will have a radius of 5 centimeters). This is timed so that the plasma blob (plasmoid) is in the center of the cylinder. The liner compresses the plasmoid and ignites the fusion reaction.

The lithium stands in between the reaction and the chamber walls, protecting the walls. It also absorbs much of the radiation, protecting the crew. The lithium is also the propellant. Since it is tightly wrapped around the reaction, it is very efficient at getting the fusion reaction energy to heat the propellant. The ionized lithium (plus the burnt fusion fuel) exits through a magnetic nozzle, providing thrust.

Since this is an open-cycle system, the exhaust acts as the heat radiator, so the spacecraft can get by with only a tiny radiator. The energy to run the magnets can be supplied by solar cell arrays. Since the compression is so efficient, this will work with several types of fusion fuel: D-T, D-D, and D-3He. D-D is probably preferred, since tritium is radioactive with a short half-life, and 3He is rare.

Please note that if you replace the magnetic nozzle with a magnetohydrodynamic (MHD) generator, the propulsion system is transformed into an electrical power generator. This could be used for ground based fusion power generators.

Dr. Slough et al worked up two spacecraft for a Mars mission. The first was optimized to have a high payload mass fraction. The second was optimized to have the fastest transit time. Both were capable of a direct abort and return. The "Low Gear" engine is the study author's opinion of an engine easily achievable with current technology (that is, achievable fusion yields). The "High Gear" engine is a bit more speculative, but requiring only modest incremental improvements in technology.

Fusion Drive Rockets (FDR)
High Mass Fraction
EngineLow Gear
Transit Time90 days
Initial Mass90 mT
Payload Mass Fraction65%
Specific Mass4.3 kg/kW
Shortest Transit Time
EngineHigh Gear
Transit Time30 days
Initial Mass153 mT
Payload Mass Fraction36%
Specific Mass0.38 kg/kW

Antimatter catalyzed

Nuclear fission pulse drives like Orion scale up well, since it is relatively easy to design a bigger bomb than the last one. However, physics seem to prevent the creation of a nuclear device with a yield smaller than about 1/100 kiloton (10 tons, 42 GJ) and a fissionable material mass under 25 kilograms. This is due to critical mass restraints.

However, if a tiny sub-critical bit of fissionable material is bombarded by a few antiprotons, it will indeed create a tiny nuclear explosion. The antiprotons annihilate protons in uranium atoms, the energy release splits the atoms, creating a shower of neutrons, and a normal chain reaction ensues. Using antiprotons, yields smaller than 1/100 kiloton can be achieved. This can be used to create Antimatter catalyzed nuclear pulse propulsion


Thrust Power0.016 GW
Exhaust velocity598,000 m/s
Thrust55 n
T/W >1.0no

Antiproton-initiated Microfusion. Inertial Confinement Fusion. See here.


Thrust Power14 GW
Exhaust velocity132,000 m/s
Thrust106,000 n
Propellant0.8 kg/s
Antiprotons30 ng/s
T/W >1.0no

Antiproton-catalyzed microfission, inertial confinement fission. See here.

Fuel pellets have 3.0 grams of nuclear fuel (molar ratio of 9:1 of Deuterium:Uranium 235) coated with a spherical shell of 200 grams of lead. The lead shell is to convert the high energy radiation into a form more suited to be absorbed by the propellant. Each pellet produces 302 gigajoules of energy (about 72 tons of TNT) and are fired off at a rate of 1 Hz (one per second). The pellet explodes when it is struck by a beam containing about 1×1011 antiprotons.

A sector of a spherical shell of 4 meters radius is centered on the pellet detonation point. The shell is the solid propellant, silicon carbide (SiC), ablative propellant. The missing part of the shell constitutes the exhaust nozzle. Each fuel pellet detonation vaporizes 0.8 kilograms of propellant from the interior of the shell, which shoots out the exhaust port at 132,000 meters per second. This produces a thrust of 106,000 newtons.

The Penn State ICAN-II spacecraft was to have an ACMF engine, a delta-V capacity of 100,000 m/s, and a dry mass of 345 metric tons. The delta-V and exhaust velocity implied a mass ratio of 2.05. The dry mass and the mass ratio implied that the silicon carbide propellant shell has a mass of 362 metric tons. The wet mass and the thrust implied an acceleration of 0.15 m/s2 or about 0.015g. It can boost to a velocity of 25 km/sec in about three days. At 0.8 kilograms propellant ablated per fuel pellet, it would require about 453,000 pellets to ablat the entire propellant shell.

It carries 65 nanograms of antiprotons in the storage ring. At about 7×1014 antiprotons per nanogram, and 1×1011 antiprotons needed to ignite one fuel pellet, that's enough to ignite about 453,000 fuel pellets.

The system is very similar to Positron Ablative.


Sail propulsion does not carry onboard reaction mass or does not use reaction mass. They are powered by a remote source, either the Sun or a satellite installation with a huge power supply and an equally huge laser/plasma beam.

Electric Sail

An E-Sail is a sail powered by solar wind.

Magnetic Sail

Magnetic Sail
Thrust per sail area0.001 N/km2
Thrust by Sol dist1/R2

A MagSail is a sail powered by the solar magnetic field.


Thrust per sail area0.001 N/km2
Thrust by Sol distConstant
Disk Inflates
as 1/R2
Plasma use0.25 kg/Day per N Thrust
Isp = 35,000

A Mini-magnetospheric plasma sail (M2P2) is a MagSail inflated by an injection of plasma, powered by the solar magnetic field.


A MagBeam is Mini-magnetospheric plasma sail beam-powered by a remote helicon plasma beam installation. Report here. Alternatively the spacecraft can use a plasma magnet instead of a M2P2 to intercept the beam. With the current design, the spacecraft mass cannot be larger than about 10,000 kg (10 metric tons).

The installation is called a High Power Platform (HPP). The HPP does not have much range, so the spacecraft will require a second HPP at the destination in order to slow down. For a Mars mission the HPP fires for about four hours before the spacecraft is out of range. By that time the spacecraft is travelling at about 20,000 m/s, which is fast enough to get to Mars in 50 days flat. The range is about 1×107 meters (ten thousand kilometers).

After boosting a spacecraft, the HPP rotates the MagBeam in the opposite direction and uses it as an ion drive to move back into position. Newton's laws still hold, the recoil from the MagBeam is going to push the HPP way off base.

And I'm quite sure that at short ranges the MagBeam can be used as a weapon. It would also be a nifity thing for a warship to mount, so it can use it to boost missiles to ferocious velocities.

The main advantages seem to be increased acceleration levels on the spacecraft, and that one HPP propulsion unit can service multiple spacecraft. There are certain maneuvers that are impossible for low acceleration spacecraft, such as sub-orbital to orbital transfers, LEO to GEO transfers, LEO to escape velocity, and fast planetary missions.

Plasma beams as a general rule have short ranges. However, the system can take advantage of the fact that both the HPP and the spacecraft have magnetic fields. The MagBeam uses magnetic fields to focus the beam and the spacecraft has a MagSail to catch the beam. If they start off close enough to each other, the two magnetic field merge ("magnetic reconnection"), and gradually stretch as the spacecraft moves. This creates a long magnetic tunnel to confine the plasma stream, making the stream self-focusing.

This will be a problem when the HPP is faced with the task of slowing down an incoming spacecraft, since initially there will be no magnetic link. The spacecraft will have to temporarily inflate its MagSail, which can be done because it is an M2P2. Once the magnetic connection is made the M2P2 can be deflated to normal size.

Plasma will probably be argon or nitrogen. The beam range will a few thousand kilometers if the HPP or the beam passes through the ionosphere, tens of thousands of kilometers if in the magnetosphere. This is because of the ambient plasma and magnetic fields in the ionosphere.

Since the spacecraft does not carry its propellant, the standard rocket equation does not apply. Instead:

HPPe = (0.25 * M * deltaV * Ve ) / HPPeff


  • HPPe = electrical energy expended by HPP (joules)
  • M = mass of spacecraft (kg)
  • deltaV = delta V applied to spacecraft (m/s)
  • HPPeff = efficiency of HPP at converting electricity into plasma energy (100% = 1.0, currently 0.6)

Mpb = HPPe / (0.5 * Ve2)


  • Mpb = mass of propellant expended in HPP beam (kg)
  • HPPe = electrical energy expended by HPP (joules)
  • Ve = velocity of HPP beam (m/s)

HPPpower = HPPe / Taccel


  • HPPpower = miminum power level of HPP power plant (watts)
  • HPPe = electrical energy expended by HPP (joules)
  • Taccel = duration of HPP beam usage (sec)

So if a HPP had to boost a 10,000 kg (10 metric ton) spacecraft to a deltaV of 3,000 m/s (3 km/s) using a plasma beam with a velocity of 19,600 m/s (2000 s) had only 300 seconds (5 minutes) to do so and had an efficiency of 0.6 (60%), then the electrical power used would be 2.5×1010 joules, the power plant would need a level of 82,000,000 watts (82 megawatts), and 127 kilograms of propellant would be expended.

Photon Sail

Photon Sail
Thrust per sail area9 N/km2
Thrust by Sol dist1/R2

A Photon Sail is a sail powered by solar photons. Commonly called a "solar sail", but that term does not make it clear if the sail is powered by solar photons, solar magnetic field, or solar wind.

Plasma Magnet

An plasma Magnet is a type of E-sail powered by solar wind.


{ Beer }

Thrust Power8 × 10-8 GW
Exhaust velocity83 m/s
Thrust84 n
T/W >1.0no

In The Makeshift Rocket (also known as A Bicycle Built for Brew), the old geezer cobbles together a crude rocket out of hogs-heads of pressurized beer in order to escape to an adjacent asteroid.

(ed note: I asked Rob Davidoff for an estimate of the performance of beer.)

Thrust = velocity * mass_flow

Assume we model the system as the fluid starting from stagnation (V-o = 0) under pressure P_o and accelerating to a vacuum pressure P_2 = 0 at velocity v_1. We can then employ Bernoulli's equation, which says the following once we knock out some irrelevant terms:

P_o = 0.5 * rho * (V_1)2

Solve for V_1:

V_1 = sqrt( 2 * P_o / rho)

So, what's a reasonable pressure? Sheesh, I dunno. A standard fuel-driven rocket engine operates at about 35 atm for a very low-pressure combustion, let's try that. Using the density of water (1000 kg/m3), I get...84 m/s. Isp of 8.5 seconds or so. The thrust will be this times the mass flow, so 1 kg/s would give 84 Newtons.

Is this any use? It's pretty crappy, but maybe it's good enough. Say he needs, oh, 150 m/s. That's a mass ratio of 6, which isn't terrible. To lift off from an asteroid, you basically need a T/W of anything non-zero, so it's workable. Of course, keeping beer pressurized to 35 atmospheres was the starting assumption, any maybe that was a little high.

However, the big issue is the density of the beer. Substitute in an air-like gas with a density of 1.4 kg/m2 instead of 1000, and you get to an Isp of ~220s, instead of 8. That's a lot more like it.

Rob Davidoff

Mass Driver

Mass Driver
Thrust Power0.3 GW
Exhaust velocity30,000 m/s
Thrust20,000 n
Engine mass150 tonne
T/W >1.0no
Power req350 MW

Mass drivers: magnetic buckets filled with packed rock dust are accelerated electmagnetically. Buckets are recovered for re-use. Propellant is rock dust or anything else you can stuff into the bucket. Popular with asteroid miners who want to nudge their claims into different orbits. However, their existence may prompt the creation of an Orbital Guard.

In Gerard O'Neill's plan for L5 colonies, mass drivers were used to deliver raw materials mined on Luna into orbit for colony construction. But instead of the mass driver being mounted on a cargo rocket, it was instead a ground installation near the lunar mine. The buckets were filled not with rock propellant, but instead with cargo cannisters of raw materials. The mass driver shot the cannisters into orbit. The cannisters were intercepted by a "catcher" at the colony site. So instead of needing a fleet of cargo rockets, you just needed a mass driver launcher and a catcher.

A mass driver is an electromagnetic mass accelerator that is optimized for propulsion. If you optimize it as a weapon instead, you have a coil-gun or rail gun. The weapons still have recoil and can be used as a crude propulsion system.

“Anjeä SysCon, this is VS Ardent Voyager, gated in-system from Loxix, identifying. Over.”

“Ardent Voyager, Anjeä SysCon, we have you arriving at 5173-09-14:7-51-11; squawk ident. Welcome to Imperial space, please specify your intentions. Over.”

“Anjeä SysCon, Ardent Voyager. Request through-clearance for immediate transit to Conclave System, minimum delta transfers. Over.”

“Wait one, Ardent Voyager… Voyager, please confirm your hull class and propulsion. Over.”

“Anjeä SysCon, we are a beehive habitat with reserve mass driver propulsion. Over.”

“In other words, Ardent Voyager, you’re flying an asteroid and moving by throwing rocks. With regret, please shut down all active drive systems immediately. You are denied transit permission under power. Over.”

“Anjeä SysCon, we are a diplomatic vessel and have the right of transit to Conclave System. Over.”

“Ardent Voyager, you have the right of transit, but that doesn’t exempt you from the rules of navigation. Over.”

“Anjeä SysCon, what’s your problem with us? Nowhere else has refused us transit. Over.”

“Ardent Voyager, this is a crowded system with too damn many loose rocks anyway, see? We don’t want any accidents, and a drive like yours is a flyin’ invitation to accidents, or a hefty cleanup bill. It’s a miracle you got clearance to transit this far. Over.”

“Anjeä SysCon, what are we supposed to do, then, just sit here? Over.”

“Ardent Voyager, hire a tug? Either to finish out your voyage or jump back out-system, but either way, you’re not runnin’ that hazard to navigation anywhere in our sky. SysCon, clear.”

- overheard on system space-control channel, Anjeä (High Verge)


Exhaust velocity299,792,458 m/s

The exhaust is not a stream of matter. Instead it is a beam of Electromagnetic radiation, basically a large laser. The advantage is that it has the maximum possible exhaust velocity and thus the highest specific impulse. The disadvantage is the ludicrously high power requirements.

The momentum of a photon is p = E/c, where E is the energy of the photon. So the thrust delivered by a stream of photons is ∂p/∂t = ∂E/∂t/c. This boils down to:

F = P/c

P = F * c


  • F = thrust in Newtons
  • P = power in watts
  • c = speed of light in a vacuum (3e8 m/s)

In other words, one lousy Newton of thrust takes three hundred freaking megawatts!!

Watch the Heat

From my limited understanding, the basic problem is how to keep the engine from vaporizing.

Fp = (F * Ve ) / 2


  • Fp = thrust power (watts)
  • F = thrust (newtons)
  • Ve = exhaust velocity (m/s)

The problem is that at high enough values for exhaust velocity and thrust, the amount of watts in the jet is too much. "Too much" is defined as: if only a fractional percentage of those watts are lost as waste heat, the spacecraft glows blue-white and evaporates. The size of the dangerous fractional percent depends on heat protection technology. There is a limit to how much heat that current technology can deal with, without a technological break-through.

Jerry Pournelle says (in his classic A STEP FARTHER OUT) that an exhaust velocity of 28,800,000 cm/s corresponds to a temperature of 5 million Kelvin.

As an exceedingly rough approximation:

Ae = (0.5 * Am * Av2) / B


  • Ae = particle energy (Kelvin)
  • Am = mass of particle (g) (1.6733e-24 grams for monatomic hydrogen)
  • Av = exhaust velocity (cm/s)
  • B = Boltzmann's constant: 1.38e-16 (erg K-1)

(note that the above equation is using centimeters per second, not meters per second)

A slightly less rough approximation:

Qe = (Ve / (Z * 129))2 * Pw


  • Qe = engine reaction chamber temperature (Kelvin)
  • Ve = exhaust velocity (m/s)
  • Z = heat-pressure factor, varies by engine design, roughly from 1.4 to 2.4 or so.
  • Pw = mean molecular weight of propellant, 1 for atomic hydrogen, 2 for molecular hydrogen

The interiors of stars are 5 million Kelvin, but few other things are. How do you contain temperatures of that magnitude? If the gadget is something that can be mounted on a ship smaller than the Queen Mary, it has other implications. It is an obvious defense against hydrogen bombs, for starters.

Larry Niven postulates something like this in his "Known Space" series, the crystal-zinc tube makes a science-fictional force field which reflects all energy. Niven does not explore the implications of this. However, Niven and Pournelle do explore the implications in THE MOTE IN GOD'S EYE. The Langson Field is used in the ship's drive, and as a force screen defense. The Langson field absorbs energy, and can re-radiate it. As a defense it sucks up hostile laser beams and nuclear detonations. As a drive, it sucks up and contains the energy of a fusion reaction, and re-radiates the energy as the equivalent of a photon drive exhaust.

(And please remember the difference between "temperature" and "heat". A spark from the fire has a much higher temperature than a pot of boiling water, yet a spark won't hurt your hand at all while the boiling water can give you second degree burns. The spark has less heat, which in this context is the thrust power in watts.)

If one has no science-fictional force fields, as a rule of thumb the maximum heat load allowed on the drive assembly is around 5 MW/m2. This is the theoretical ultimate, for an actual propulsion system it will probably be quite a bit less. For a back of the envelope calculation:

Rc = 0.12 * sqrt[H]


  • Rc = reaction chamber radius (meters)
  • H = reaction chamber waste heat (megawatts)

(this equation courtesy of Anthony Jackson)


Say your propulsion system has an exhaust velocity of 5.4e6 m/s and a thrust of 2.5e6 N. Now Fp=(F*Ve)/2 so the thrust power is 6.7e12 W. So, 6.7e12 watts divided by 1.0e6 watts per megawatt gives us 6.7e6 megawatts. Plugging this into the equation results in 0.12 * sqrt[6.7e6 MW] = drive chamber radius of 310 meters or a diameter of a third of a mile. Ouch.

As a first approximation, for most propulsion systems one can get away with using the thrust power for H. Science-fictional technologies can cut the value of H to a percentage of thrust power by somehow preventing the waste heat from getting to the chamber walls.

Only use this equation if H is above 4,000 MW or so, and if the propulsion system is a thermal type (i.e., fission, fusion, or antimatter).

Playing with these figures will show that enclosing a thermal torch drive inside a reaction chamber made of matter appears to be a dead end. Unless you think a drive chamber a third of a mile in diameter is reasonable.

Therefore, the main strategy is to try and direct the drive energy with magnetic fields instead of metal walls. The metal framework lets the heat escape instead of vaporizing the nozzle. The magnetic field cannot be vaporized since it is composed of energy instead of matter.

This is gone into in more detail here in the Torchship section.

And don't forget the Kzinti Lesson.

Calculating the performance of a spaceship can be complicated. But if the ship is powerful enough, we can ignore gravity fields. It is then fairly easy. The ship will accelerate to a maximum speed and then turn around and slow down at its destination. Fusion or annihilation-drive ships will probably do this. They will apply power all the time, speeding up and slowing down.(ed note: a "brachistochrone" trajectory)

In this simple case, all the important performance parameters can be expressed on a single graph. This one is drawn for the case when 90% of the starting mass is propellant. (ed note: a mass ratio of 10) Jet velocity (exhaust velocity) and starting acceleration are the graph scales. Distance for several bodies are shown. Mars varies greatly; I used 150 million kilometers. Trip times and specific power levels are also shown. "Specific power" expresses how much power the ship generates for each kilogram of its mass, that is, its total power divided by its mass. The propellant the ship will carry is not included in the mass value.

An example: Suppose your ship can produce 100 kW/kg of jet power. You wish to fly to Jupiter. Where the 100 kW/kg and Jupiter lines cross on the graph, read a jet velocity of 300,000 m/s (Isp = 30,000) and an initial acceleration of nearly 0.01g. Your trip will take about two months.

The upper area of the graph shows that high performance is needed to reach the nearest stars. Even generation ships will need, in addition to very high jet velocities, power on the order of 100 kW/kg. The space shuttle orbiter produces about 100 kW/kg with its three engines. The high power needed for starflight precludes its attainment with means such as electric propulsion.

Gordon Woodcock

Atomic Rockets notices

Welcome to the improved Atomic Rockets!

Atomic Rockets

Support Atomic Rockets

Support Atomic Rockets on Patreon