Liquid fluorine has great potential as a rocket fuel; per the Encyclopedia Astronautica, LF2/LH2 has a specific impulse of 470 seconds. http://www.astronautix.com/l/lf2lh2.html Lithium/LF2/LH2 can get you over 500 seconds, but requires you to have molten lithium at over 450K stored near cryogenic liquid hydrogen.
Krypton difluoride (KrF2) is a compound with some interesting properties, aside from being a noble gas compound. Annoyingly, it breaks apart at temperatures above about 195K. More importantly for the purpose of using it as a rocket propellant, it is an incredibly strong oxidizing agent. In fact, it is a more powerful oxidizer than fluorine gas, a consequence of the extremely dissociation energy (delta Hf) of the krypton-fluorine bond (54 kJ/mol, vs 157 kJ/mol for F2, via https://labs.chem.ucsb.edu/zakarian/armen/11---bonddissociationenergy.pdf )
Hydrogen fluoride (HF), the product of burning hydrogen with fluoride compounds, has a bond dissociation energy of 568 kJ/mol. Modelling the combustion of KrF2 and H2, we get the following equation;
KrF2 + H2 = Kr + 2HF
-(2*55 + 436) = -2*569 + E
for a net energy production of 592 kJ/mol. Compared to 545 kJ/mol for hydrogen/fluorine combustion, this is slightly better. Also, it has the advantage that your fuel is at a higher temperature (195K for KrF2 vs. 85K for LF2), and KrF2 is more dense than liquid fluorine. Unfortunately, KrF2 is a solid with no known liquid phase, and as mentioned, is unstable above 195K.
A better option might be xenon fluorides. The xenon-fluorine bond has a bond energy of a mere 13 kJ/mol, and xenon compounds are much more understood than krypton compounds. Likely the best option is Xenon Hexafluoride (XeF6). Xenon hexafluoride melts at 322K, and boils at 348K. A fairly narrow temperature range, and unfortunately one that would require heating, but likely not insurmountable. (Sadly, I cannot find information on the liquid density of XeF6).
XeF6 would react with H2 according to the following formula.
XeF6 + 3H2 = Xe + 6HF
-(6*13+3*436) = -6*569 + E
for a net energy production of 2028 kJ/mol
Granted, this combination would almost certainly have lower specific impulse than LH2/LF2 due to the large size of the xenon atom (average mass of exhaust products is 35.8 vs. 19, so exhaust velocity would be roughly 37% less). Assume roughly 300 isp assuming flame temp is equal to LH2/LF2, possibly more if the flame temp for xenon hexafluoride is higher. However, xenon hexafluoride fuel would give a higher thrust, in addition to being more dense and making your first stage smaller. Although at that point, why not use kerolox and ditch liquid hydrogen altogether. Still, an interesting theoretical exercise, and I'd be keen to see data on xenon hexafluoride flame temps if anyone has it.