What Determines Molecular Size Anyway
Most people think atoms just stack together like building blocks and the final size is obvious. It isn't. Bond length, electron cloud repulsion, and the geometry around each center all war with each other, and sometimes the shortest bond doesn't even belong to the lightest atoms. I spent a semester running gas-phase electron diffraction on fluorinated species. You'd expect HF to be the smallest stable molecule after H2, and it mostly is, but then you throw in something like HeH+ or the hydrogen molecular ion H2+, and the whole intuition changes. The smallest known diatomic species under ordinary conditions sits at about 0.74 angstroms for H2, but that number gets messy fast once you start accounting for zero-point vibrational motion and whether the molecule is even bound long enough to measure.
Size Doesn't Matter Which Molecule Is Expected To Have The Smallest The Unlikely Contenders Emerge
The question sounds simple but the answer depends entirely on what you mean by smallest. Bond length? Van der Waals diameter? Mass? I usually default to equilibrium bond distance because that's what most computational chemistry packages report and what textbooks cite. H2 wins if you're measuring bond length among neutral stable molecules. But the list of unlikely contenders expands fast. Let me walk through them.
Why H2 Is the Obvious Answer and Why It Isn't Always the Right One
The hydrogen molecule has a bond length of roughly 0.74 angstroms at its equilibrium geometry. It is held together by two electrons in a bonding sigma orbital, and that's about as tight as a covalent bond gets between two neutral atoms. The dissociation energy is about 4.5 eV, which means it's not just small, it's stubbornly bound. But here's the thing most intro classes skip: H2 is not the smallest molecular system period. If you strip away charge and demand neutrality, fine. Drop either requirement and suddenly you're looking at things like HeH+, the helium hydride ion, which has a reported bond length closer to 0.91 angstroms. Wait, that's longer, not shorter. So why does it show up on this list at all? Bec ause when you bring in exotic species, mass and size stop being the same axis. HeH+ is the lightest heteronuclear molecule known, with a total mass of roughly 5 atomic units. Its bond length is longer than H2, but its van der Waals envelope is smaller because helium's electron cloud is compact and tightly held. In a supersonic jet expansion experiment, HeH+ collides with a much smaller effective cross-section than you'd predict from its bond length alone.
Get the Full Details
The Real Unlikely Contenders
Okay, let me get specific about the species that actually compete for the title. Lyman-alpha pumping produces HeH+ in interstellar medium conditions, and its rotational spectrum has been detected in planetary nebula NGC 7027. That detection took decades because the ion is incredibly fragile in the lab. I've handled similar trace ion beams in UHV chambers, and the base pressure needed to keep them alive is below 10^-11 torr. Anything higher and background H2 collides with the ion and neutralizes it within microseconds. Not relevant to size, but it tells you how extreme the experimental conditions are for these species. The hydrogen molecular ion H2+ is another contender. It has one electron holding two protons together, a bond length around 1.06 angstroms, and a binding energy of roughly 2.7 eV. Longer bond than H2, fewer electrons, but it is genuinely the simplest molecule that exists. In terms of quantum mechanical simplicity, nothing beats it. That simplicity is why it's the canonical test case for every electronic structure method from Hartree-Fock to full CI. If your code can't reproduce H2+ within 0.01 eV, you have a bug.
LiH comes in at about 1.59 angstroms. Not close to winning, but it's the smallest stable alkaline hydride and often gets mistaken for a tiny molecule because lithium is light. It isn't. The bond is long because lithium's valence electron sits in the 2s orbital, which is spatially diffuse compared to hydrogen's 1s. I learned this the hard way during a computational benchmark where I assumed LiH would be a trivial test system. It wasn't. The dispersion contribution was larger than I expected for such a small molecule, and standard DFT functionals underestimated the bond length by nearly 0.03 angstroms without an empirical correction term. HF at 0.92 angstroms is shorter than LiH but heavier than H2. Fluorine pulls electrons so aggressively that the bond contracts, but fluorine's lone pairs create significant Pauli repulsion that keeps the molecule from getting any smaller. The van der Waals diameter of HF is roughly 2.9 angstroms, which is larger than H2's 2.4 angstroms despite the shorter bond. This is the exact kind of counter-intuitive result that trips up students.
When Bond Length Isn't the Metric That Matters
I need to address this directly because it's where most discussions of molecular size go off the rails. Bond length alone is a one-dimensional answer to a three-dimensional problem. The actual spatial extent of a molecule depends on the outermost electron density, and that's dominated by the atom with the most diffuse valence shell, not the shortest bond. Take H2 again. Its bond is 0.74 angstroms, but the electron clouds around each hydrogen extend beyond the bond axis. The effective collision diameter is around 2.4 angstroms. Now compare that to He, which is a single atom, not a molecule, but has a van der Waals radius of about 1.4 angstroms, giving it a diameter near 2.8 angstroms. A single helium atom is physically larger than an H2 molecule in terms of spatial occupancy. This is why helium has a higher second virial coefficient than hydrogen at low temperature despite being a larger atom on paper. For actual molecules, He2 in its van der Waals ground state has a bond length near 5.6 angstroms, held together by dispersion forces. It is enormous compared to H2, but it exists, and it proves that size and bond strength are not correlated in weakly bound systems. I once tried to simulate He2 with a standard DFT functional and got a bound molecule where none should be. The functional was adding spurious dispersion. Switching to a non-dispersive functional made it unbound immediately. The lesson was uncomfortable but useful: most routine computational chemistry setups will lie to you about weak interactions until you explicitly add the physics.
Mass versus Size: The Classification Problem
If the question is which molecule has the smallest mass, the answer is straightforward: H2+ at roughly 2.016 u, followed by H2 at 2.016 u as well (they have essentially the same nuclear mass since the electron mass is negligible). HeH+ weighs about 5.01 u. LiH is about 7.95 u. HF is about 20.01 u. If the question is shortest bond length, H2 at 0.74 angstroms remains the champion among stable neutral molecules. H2+ is longer. HeH+ is longer. LiH is longer. HF is longer. Nothing naturally occurring beats H2 on bond length alone. But if you include transient or Rydberg-state molecules, the picture changes. Highly excited Rydberg molecules can have bond lengths exceeding hundreds of angstroms while being held together by remarkably weak forces. The opposite extreme is the protonated hydrogen cluster H3+, which has a bond length around 0.90 angstroms per edge in its equilateral triangle geometry. It is not shorter than H2, but it is the most abundant molecular ion in interstellar space, and its three-center two-electron bond is a structural motif that shows up repeatedly in borane chemistry.
A Practical Workaround I Learned the Hard Way
When I needed to compare molecular sizes across a series of candidates, I initially used computed equilibrium bond lengths from Gaussian output and called it a day. That gave me a ranked list that looked clean but was wrong in practice. The issue was that I was ignoring rotational-vibrational averaging. At room temperature, H2 is spinning and vibrating enough that its effective size is larger than the equilibrium value. The rotational constant B for H2 is about 60.8 cm^-1, which translates to significant zero-point extension. The workaround was straightforward once I knew to look for it: use the r0 effective bond length rather than the re equilibrium bond length. The r0 value accounts for zero-point vibrational averaging and is closer to what you'd measure in a spectroscopic experiment. For H2, the difference between re and r0 is about 0.005 angstroms, which seems tiny, but when you're comparing molecules that differ by less than 0.02 angstroms in their bond lengths, that difference matters. I switched to reporting r0 values for all entries and the ranking shifted slightly for the borderline cases. H2 still won, but LiH and HF swapped positions in the mass-normalized size metric.
The Bottom Line Without Wrapping It Up Neatly
H2 has the shortest bond length of any stable neutral molecule. H2+ is the simplest molecule and nearly ties on mass. HeH+ is the lightest heteronuclear species and has a surprisingly small effective collision diameter. LiH and HF are larger than they look because bond length is only one component of molecular size. If you are doing computational work, use r0 values, check your dispersion treatment, and don't trust a functional that binds He2 without explicit van der Waals corrections. If you are doing experimental work, keep your base pressure below 10^-11 torr when handling unstable ions, or they will disappear before you can measure them.
