How are the 10 vertical bonds in Ferrocene (sandwich structure) connected, by what type of bonds?

enter image description here

What is the real structure of ferrocene? Concerning the 10 vertical extra long bonds?

  • 4
    $\begingroup$ What do you mean by 'real structure'? If I'm not mistaken, we're looking right at it. $\endgroup$
    – M.A.R.
    Commented Sep 20, 2015 at 20:19
  • 13
    $\begingroup$ The "bond lines" in the picture you show are only there to hint that there is some bonding between the rings and iron. But they are not to be interpreted in the sense of bonds in Lewis structures, so don't take them literally. For molecules containing transition metals Lewis structures are not very accurate in describing the actual bonding (this might be interesting in that context). $\endgroup$
    – Philipp
    Commented Sep 20, 2015 at 20:33
  • 1
    $\begingroup$ Yes, but the question I am asking is how are the rings in 3-D SPACE ATTACHED TO THE Iron atom. The discovers & others for Ferrocene openly admitted they could not discover the actual COVALENT bonding type and the bond angles were not discoverable from their x-Ray diffraction patterns. I am looking for replays that can postulate the bonding TYPE, bonding angles and bond lengths, what fraction of the bond is from Fe & from Carbon. Might it be possible that the bonding is of the Pauling "bent-bonding" type. Also, what is the 3-D geometrical shape of the Iron Atom within the so-called sandwich? $\endgroup$ Commented Sep 21, 2015 at 19:00
  • 1
    $\begingroup$ How does Iron form 10 or more bonds of any type? Can Iron form 20 bonds to Iron? What is the maximum # of bonds Iron can form from one iron atom? What is the geometrical 3D structure of the Iron atom within Ferrocene? $\endgroup$ Commented Sep 21, 2015 at 19:28
  • 1
    $\begingroup$ @ChuckBoldwyn We know the structure of ferrocene. The C5 units are flat and symmetric and the iron forms the filling in a sandwich where the rings are parallel to each other on either side of the iron atom. You seem to be trying to impose a bonding structure made up of single bonds that doesn't (and probably can't) explain this structure. Only a more complex view of the bonding makes sense involving delocalised pi-orbitals from the rings and d-orbitals from the iron. No simple picture with single bonds captures this. $\endgroup$
    – matt_black
    Commented Sep 22, 2015 at 0:10

3 Answers 3


The "real" structure of ferrocene is an iron atom sandwiched between two flat, parallel, pentagonal C5H5 rings. The diagram above shows the atomic positions from a crystal structure but the bonds are merely a convenience and don't accurately summarize the way the bonding works though the picture does accurately summarize the atomic positions in the crystal structure.

The real bonding requires some understanding of molecular orbital theory and how the d-orbitals of iron interact with the delocalized pi-orbitals in the unsaturated ring. Some pictures of the orbitals contributing to the bonding are shown here.

Ferrocene creates some flurry of new thinking about bonding theory. Observations, for example, show the rings rotate easily in the plane of the rings and this very clearly shows that any picture with 10 sigma bonds is not a good explanation for the overall structure. Plus the chemical reactivity is far more similar to benzene than to an unsaturated system which, again, implies it isn't all sigma bonds.

  • $\begingroup$ You are assuming all "p" orbitals & their bonding are only being used in bonding to Iron.....How about all sp3 orbitals & bonding coming from the carbon atoms attaching to whatever orbitals are extended from the Iron atom all as bent bonds. Could a structure, just proposed be possible or likely. Could this be a result of all free-radical bonding resulting in the Pauling proclaimed " banana bonds or bent bonding type? $\endgroup$ Commented Sep 21, 2015 at 19:12
  • 3
    $\begingroup$ I didn't say anything about p-orbitals. The only way to really explain the structure is to think of molecular not atomic orbitals for the whole ring (which has delocalised pi-orbitals which interact with iron's d-orbitals). And we are not proposing a structure; we know the structure from x-ray and NMR evidence. All the carbons are equivalent and the structure looks like a sandwich with an iron filling. $\endgroup$
    – matt_black
    Commented Sep 22, 2015 at 0:14
  • $\begingroup$ Are not pi orbitals made up of "p" orbitals overlapping vertically? I think when the cyclopentadieyl ring converts from sp2 carbon atoms to all sp3 carbon atoms, one would get simple cyclo pentane forming 5 sp3 bonds with the multi- valent Iron atom $\endgroup$ Commented Sep 22, 2015 at 4:56
  • 1
    $\begingroup$ @ChuckBoldwyn The ring bonding doesn't "convert" to sp3 and the ring is aromatic with delocalised electrons. Pictures of the bonding that ignore this make no sense of the stability of structure of the molecule. $\endgroup$
    – matt_black
    Commented Sep 22, 2015 at 7:56
  • 1
    $\begingroup$ @Maurice Why not posting this as a new question; preferentially setting sandwich complexes with 4, 5, and 6 carbon atoms on each side next to each other (same transition metal and oxidation state), and a check if these are are Hueckel aromatic. $\endgroup$
    – Buttonwood
    Commented Aug 10, 2020 at 17:28

The pictures we use can inspire questions that are more complicated than they have to be. I like the hamburger picture of ferrocene:

enter image description here

OK, there are some people who like a picture of electron lobes:

enter image description here

Without going into the mathematics of all the orbitals, you can see that the filled orbitals on cyclopentadienide ion will have two lobes to donate into d-orbitals on ferrous ion. And there will be an electrostatic attraction between the cyclopentadienide ions and the ferrous ion.

The use of points (atoms) and lines (bonds) connecting them distorts the filled-space reality of the actual molecule. Counting out 10 Fe-C bonds doesn't suggest the best explanation of the bonding. Of course, when you look at the filled-space models, you can't see exact bond lengths and angles, but it may give you a better idea of atoms fitting together, with molecular orbitals providing some sort of glue.

  • 2
    $\begingroup$ The second figure you show is not electron density, just one random MO $\endgroup$
    – Greg
    Commented Aug 2, 2021 at 15:16
  • 1
    $\begingroup$ @Greg: Thanks. You are quite correct. I will change it to say "lobes", unless you suggest a better word - and change the picture, too! $\endgroup$ Commented Aug 2, 2021 at 18:01

Since the OP is talking about a VB-based approach to describing the bonding in ferrocene, I'm going to ignore any purely MO-based tactics in this answer.

Ferrocene is made by combining two cyclopentadienide anions and one ferrous cation. Each cyclopentadienide anion has one lone pair(which actually corresponds to a Hückel bonding orbital; still writing "lone pair" since I'm ignoring any purely MO-based tactics such as Hückel theory), and these two pairs can each be shared to the ferrous cation.

By now, the iron has 10 (6 from the itself, 2 each from the two bonds) valence electrons near itself. But wait- iron wants 18 electrons so it can be like krypton. Luckily, there are exactly 8 more electrons total nearby- the total of 4 pi-bonds of the two cyclopentadienide rings. Each pi-bond turns into a 3-center-2-elctron bond.

By now, there are 4 bonds near each of the 8 carbons (one C-H, two C-C sigma and one C-C-Fe 3c-2e) with no lone pairs, and 6 bonds near the iron (two C-Fe 2c-2e each, and four C-C-Fe 3c-2e each) with three lone pairs. Since (4 times 2) means octet and (6 times 2) + (3 times 2) means decaoctet, everybody is happy. Note that, since nobody has a formal charge here(dative bonds do not affect formal charge), this resonance canonical satisfies the Pauling rule of electroneutrality. In short, there are two 2c-2e carbon-iron bonds and four 3c-2e carbon-carbon-iron bonds.

Now, there initially are two cyclopentadienide anions, and the lone pair on each ring can reside on any of the five constituent carbon atoms of the ring. Therefore, there are 25 such canonicals, the combination of which constitutes the majority of the ferrocene molecule.

Explaining the pi-backbonding in the ferrocene molecule using VB theory is more complicated, but it can be intuitively understood by the following scheme: starting from a resonance canonical described above, a lone pair on the iron, and a 3c-2e bond involving both the iron and two or the carbons(total number of electrons involved : 4), magically rearrange to two iron-carbon bonds(total number of electrons involved here : 4).

Before: 4 bonds near each of the 2 carbons (one C-H, two C-C sigma and one C-C-Fe 3c-2e) with no lone pairs, and 6 bonds near the iron (two C-Fe 2c-2e each, and four C-C-Fe 3c-2e each) with three lone pairs

After: 4 bonds near each of the 2 carbons (one C-H, two C-C sigma and one C-Fe) with no lone pairs, and 8 bonds near the iron (four C-Fe 2c-2e each, and four C-C-Fe 3c-2e each) with two lone pairs

Again, the two carbons and the iron each satisfy the octet/decaoctet, and again, the dative Fe->C bond does not change the zeroness of the formal charges here, and hence the results are also equally valid resonance canonicals(have to add an MO note here: since the Fe->C bond is dative, the oxidation numbers do not change even after the formal donation of the iron's electrons to the more electronegative carbon).

All of this is an explanation purely in basic VB theory. In modern VB theory, this slightly fails. For example, NBO calculations appear to indicate that there is one half-filled "lone pair" on each cyclopentadienyl ring and no bonds between the ring carbons and the iron, and with the two remaining electrons being used up as parts of partially-filled "lone pair"s of antibonding character on the central iron atom. However, since ferrocene must contain at least one N-center-2-electron bond, it is likely that the intuitive approach above matches the purely VB-model of the bonding in ferrocene more than the NBO calculations do.


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.