A variety of different rocket programs use a variety of different fuel types. Researching this, the main fuel combinations I saw were: RP-1 / LOX (SpaceX's Falcon rockets, early stage Atlas and Saturn rockets), methane / LOX (SpaceX's Starship), HTPB (ISRO's PSLV), PBAN (NASA's SLS boosters), LH2 / LOX (upper stage Ariane rockets and SLS) and UDMH/N204 (ISRO's GSLV). Nevertheless, these fuels aren’t used solely for their specific impulse because other factors such as toxicity and cost are considered.

According to Wikipedia

The highest specific impulse for a chemical propellant ever test-fired in a rocket engine was 542 seconds (5.32 km/s) with a tripropellant of lithium, fluorine, and hydrogen. However, this combination is impractical.

Theoretically, what other chemical propellant combinations would yield high specific impulse values and why would those combinations have a high specific impulse?

Edit: Thanks to Organic Marble for pointing out that I missed the LH2/LOX combination and pointing out that I hadn't thought of ion drives (though this question is limited to chemical rockets).

  • $\begingroup$ You left out the best "commonly" used propellants: liquid hydrogen/liquid oxygen, giving Isp in the mid 400s in vacuum. Forget the solids, and what about ion drives? Is your query restricted to chemical rockets? $\endgroup$ Commented Aug 9, 2022 at 11:24
  • 1
    $\begingroup$ Good question, the answer will be a trade-off between energy from the chemical reaction (that is why LH2-LOX is nice) and the mass of the exhaust (RP-1s specialty). I am really looking forward to see the answers. $\endgroup$
    – CallMeTom
    Commented Aug 9, 2022 at 13:32
  • 8
    $\begingroup$ "this combination is impractical" Just slightly. ;) On a closely related note, Clark mentions $\rm ClF_3$ in Ignition!: "It is, of course, extremely toxic, but that's the least of the problem. It is hypergolic with every known fuel, and so rapidly hypergolic that no ignition delay has ever been measured. It is also hypergolic with such things as cloth, wood, and test engineers, not to mention asbestos, sand, and water-with which it reacts explosively". science.org/content/blog-post/sand-won-t-save-you-time $\endgroup$
    – PM 2Ring
    Commented Aug 9, 2022 at 15:57
  • 2
    $\begingroup$ Highly related: space.stackexchange.com/q/17129/6944 $\endgroup$ Commented Aug 9, 2022 at 19:15
  • 1
    $\begingroup$ The ISP for a perfectly efficient lithium-fluorine rocket engine is around 700 s. $\endgroup$ Commented Feb 27 at 10:25

4 Answers 4


From the historical NASA document Space Handbook: Astronautics and Its Applications, by Robert W. Buchheim (1959).

Super high-energy bipropellants

300 to 385, ISP

  • Fluorine-Hydrogen
  • Fluorine-Ammonia
  • Ozone-Hydrogen
  • Fluorine-Diborane

But fluorine is very difficult to handle. Doing a combustion experiment back in the mid 1960s, I had a regulator burn up that was apparently not properly passivated that released a whole bottle of fluorine to the atmosphere. No injuries, luckily.

  • 2
    $\begingroup$ No discussion of fluorides as rocket fuel would be complete without a link to John Clark's description of chlorine trifluoride (towards the bottom of the linked post.) $\endgroup$ Commented Aug 11, 2022 at 17:23
  • 3
    $\begingroup$ It seems like that list may be oddly organized given that this doesn't include hydrogen-oxygen which has demonstrated 450 $\endgroup$
    – ikrase
    Commented Aug 12, 2022 at 3:43
  • 1
    $\begingroup$ @ikrase I suspect it's the age of the reference. The first LH2 fueled stage didn't fly until 1962, although certainly work was being done prior to that. "Taming Liquid Hydrogen" picks up in 1958 history.nasa.gov/SP-4230.pdf $\endgroup$ Commented Aug 12, 2022 at 13:50
  • $\begingroup$ Acetaline-Ozone should get an ISP over 450 but it has not been successfully tested. Pure Ozone is tricky to work with. $\endgroup$
    – Joshua
    Commented Mar 28 at 3:32
  • $\begingroup$ @MichaelSeifert I commented on Vivek's post re ClF3 and only then saw your comment. My input was, of course, based on Clarke's tome :-) $\endgroup$ Commented Apr 17 at 13:22

TL;DR: Pentaborane or LH2/LOX

From a purely specific impulse standpoint, pentaborane is perhaps one of the most potent rocket fuels, providing excellent specific impulse (about 340/365 seconds in atm/vac, see RD-270M) in the atmosphere. In addition, when combined with nitrogen tetroxide (RD-270M again), it can be used as a storable propellant as it is liquid at room temperature. For higher-energy combustion oxygen difluoride could also be used but it is very unstable and reacts with many materials. However, the reason why it isn't used as rocket fuel is because of its high toxicity and deadly exhaust products.

The lithium/fluorine/hydrogen combination does give 542 seconds of specific impulse, which is very good for a chemical propellant, but it is relatively impractical due to the difference of temperatures, with the addition of a toxic exhaust product and poor storability (lithium metal is liquid at 180 C, fluorine reacts with almost everything, and hydrogen liquifies near absolute zero, making the temperature differences near-impossible to fix). Plus, the tankage and properties of such a tripropellant may be a plumber's nightmare.

On the other hand, LH2/LOX (aka hydrolox) is a great rocket fuel for vacuum-optimized engines, providing up to 460-470 seconds of impulse in a vacuum. Many upper stages use this propellant mix, while a few launch vehicles use it as a lower stage (Ariane 5, Delta IV, Space Shuttle) but the engineering for a lower stage hydrolox engine may be more difficult due to expansion of the exhaust.

What about stuff that isn't a chemical propellant?

Again, this is from a specific impulse standpoint and not from practicality. Also this is based upon current to very near-future technology (maybe up to 50 years ahead), so antimatter-powered rockets, fusion drives, and whatnot are out of the question. If you want the best specific impulse here, use ion engines.

Ion engines (and the VASIMR) mainly use electricity (lots of it, up to several MW) and consume very little propellant for very high exhaust velocities at the cost of very low thrust, which only makes it applicable for vacuum operation. These engines can potentially reach up to 20000 seconds of specific impulse, and can operate for years without malfunctioning. However, ion engines are usually only used in small spacecraft as the tiny thrust cannot provide much acceleration for larger payloads.

The similar VASIMR may have applications for larger spacecraft or even manned spaceflight as it can trade efficiency for thrust, ranging from 1000 to 30000 seconds of impulse as thrust decreases. However, this is still in development and has not seen flight yet.

The last form of propulsion that I'll be talking about is nuclear thermal rockets, where a nuclear reactor heats up liquid hydrogen to very high temperatures and ejects it out of a nozzle, potentially giving up to 800-1000 seconds of specific impulse as well as relatively high thrust (up to 250 kN). However, such an engine is very heavy and brings the risks of reactor meltdown/failure and radioactive contamination. Although prototypes were built in the 1960s, they never saw spaceflight.

Hopefully this gives enough details about efficient rocket propellants. If you want me to give details on any other fuel, let me know.

And if you want supreme efficiency, use a kraken drive.

  • 10
    $\begingroup$ If you want supreme efficiency, use a photonic drive -- shining a flashlight out of your rocket will give you 30,570,000 seconds of specific impulse, and an abysmal thrust level. $\endgroup$
    – Mark
    Commented Aug 9, 2022 at 21:03
  • 4
    $\begingroup$ We abandoned borides in the ‘60s. You get boron-rich particles condensing in the exhaust. This is bad for the same reason H-rich exhaust molecules are good, and H-exhaust even better: Lightweight exhaust species convert thermal energy to kinetic energy more efficiently, and KE = Isp (“characteristic velocity”). Condensates, by comparison, are macromolecules. This is aside from the issue of plating the engine with deposits. In response, we tried diluting the boron-based fuel with conventional (hydrocarbon) fuel. The dilution ratio grew so big there was no more point even adding the B. $\endgroup$ Commented Aug 10, 2022 at 12:50

Very theoretically, metallic hydrogen has a predicted Isp of 1700 s.

(Or lower if it has to be diluted to not melt a rocket engine with a 6000 K reaction temperature.)

Under extremely high pressure, there's some experimental evidence that hydrogen does phase-change into a conductive metallic state. It's a very light atom, and has a lot of energy locked up in chemical bonds in that state, which is ideal for Isp. There are no heavier atoms involved. So this is probably the theoretical max as far as chemical energy propulsion.

It's hypothesized to be metastable, i.e. it may stay metallic if you release the pressure, but AFAIK not confirmed. But with the right nudge, it will release that energy as it reverts to gaseous form. Hopefully stable enough to not be set off by vibration in a fuel tank...

When I say high pressure, we're talking about millions of atmospheres (350 to 600 GPa), produced with diamond "anvils" (article with photo), with tips as small as 5μm in diameter.

So we're not making much metallic hydrogen at a time, to say the least. It's still a big deal when any lab manages to make any tiny amount and do some measurements to confirm that it's actually metallic. In terms of practically being able to produce quantities useful for rocketry, we might be a little closer than with anti-matter, but tech readiness level 1 at best. We're only at a "might be theoretically possible" stage.

And I don't think we've verified that it could maintain that phase if the pressure is released (i.e. metastability), which is necessary unless you want your fuel tanks to be diamond-tipped mountains, way too heavy for a rocket to ever lift.

Nonetheless, a 2010 paper by Silvera and Cole, Metallic hydrogen: The most powerful rocket fuel yet to exist, discusses the possibility. (Other researchers have criticized it as being overly optimistic for discussing rocketry applications when first announcing the possibility of having observed metallic hydrogen. It wasn't until later experiments at even higher pressures that people, including Silvera, were able to confirm metallic hydrogen). Anyway, its abstract includes:

Metastable metallic hydrogen would be a very light-weight, low volume, powerful rocket propellant. One of the characteristics of a propellant is its specific impulse, Isp. Liquid (molecular) hydrogen-oxygen used in modern rockets has an Isp of ~460s; metallic hydrogen has a theoretical Isp of 1700s!

If pure metallic hydrogen is used as a propellant, the reaction chamber temperature is calculated to be greater than 6000 K, too high for currently known rocket engine materials. By diluting metallic hydrogen with liquid hydrogen or water, the reaction temperature can be reduced, yet there is still a significant performance improvement for the diluted mixture.

  • $\begingroup$ Is there any actual evidence that metallic hydrogen can be made to be stabilized in a container that isn't incredibly heavy? $\endgroup$
    – ikrase
    Commented Aug 12, 2022 at 3:44
  • 1
    $\begingroup$ @ikrase: Not that I know of. As I said in the answer, it's hypothesized to be metastable, but I don't think that's been experimentally confirmed. The articles I found about it didn't mention any effort to test that. $\endgroup$ Commented Aug 12, 2022 at 4:29
  • $\begingroup$ Makes me wonder if it's only hypothetically metastable because the alternative is much less conductive to receiving research grants... $\endgroup$
    – Dragongeek
    Commented Sep 2, 2022 at 12:09
  • $\begingroup$ @Dragongeek: You mean nobody wants to check in case it's not metastable? That's possible but pretty cynical :/ It might still require some pressure and/or temperature range to stay metallic, so maybe people have tried but only ruled out a few possibilities? I didn't google a lot about what experiments have been done or published. I hope if anything did find it was metastable, the wiki article would be updated. $\endgroup$ Commented Sep 2, 2022 at 12:21

Ideal specific impulse for a given chemical reaction can easily be obtained from the energy density of the reaction.

If $d$ is the energy released by the chemical reaction per mass of reactants, then the effective exhaust velocity is simply

$$v_e = \sqrt{2\cdot d}$$

And the specific impulse is

$$I_{sp}=v_e / g_0$$

where $g_0$ is the standard gravity, $9.80665 \, \mathrm{m \cdot s^{-2}}$.

(The formula for $v_e$ is easily derived from the kinetic energy formula, $E_K=\frac{1}{2}mv^2$.)

For example, for an ideal engine burning a stoichiometric mixture of lithium and fluorine, assuming the enthalpy of formation is $23.75\, \mathrm{MJ\cdot kg^{-1}}$, we get

$$I_{sp}=v_e / g_0 = \frac{\sqrt{2\cdot 23.75\, \mathrm{MJ\cdot kg^{-1}}}}{9.80665 \, \mathrm{m \cdot s^{-2}}} \approx 703\, \mathrm{s} $$

According to this answer about highest enthalpy of formation, burning beryllium with oxygen (though rather impractical) would release slightly more energy per mass, as the stated entropy of formation of beryllium oxide is $23.96\, \mathrm{MJ\cdot kg^{-1}}$. But this only slightly increases $I_{sp}$ to about $706\, \mathrm{s}$.

In practice, the storage temperature and pressure of the reactants could make a much larger difference in $I_{sp}$ than just the three second difference between these ideal computed values.

See also these notes about ideal $I_{sp}$ for various fuel mixtures.

  • $\begingroup$ Using ozone instead of oxygen would further increase the ISP. $\endgroup$ Commented Feb 27 at 11:32
  • $\begingroup$ And further reduce practicality, concentrated ozone being a very unstable explosive. $\endgroup$
    – SF.
    Commented Feb 27 at 12:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.