Re: Why can't aluminium chloride have one double and two single bonds?

Andrew, hold onto your hat, because you have asked a really deep question. I am
going to give you a thorough and honest answer, but keep it a secret, because
it is only going to confuse your teachers.

The main reason why your answer is wrong is because it is not what the textbook
says, and if you put it in an exam you will lose marks!

Nature does not know anything about bonds and Lewis structures. Nature only deals
with atoms and electrons. We currently believe that the behaviour of atoms and
electrons is well described by a set of equations that we call quantum mechanics.
But if we want to look at something as complicated as a molecule, even a fairly small 
molecule, we cannot easily solve the equations, and we have to use large computer 
programs to get the answers from quantum mechanics (and sometimes we simply 
cannot get them).

But when we are doing chemistry, we cannot just quote six pages of numbers when
we need to talk about what aluminium chloride is doing. So chemists have invented 
a special language of bonds and Lewis structures and similar ideas. This language 
was actually developed as the result of our experience of how chemicals behaved,
long before we knew much about quantum mechanics.

The ideas that we get from bonds and Lewis structures fit pretty well with what 
those pages of numbers from the computer calculations tell us, but not perfectly.

Now to get to specifics:
When I talk to my first year university class about bonding in carbon monoxide, I
tell this story: carbon has four valence electrons, and oxygen has six. If a carbon
atom and an oxygen atom were to share two pairs of electrons, then the oxygen
atom would have a completed Lewis octet in its valence shell, but the carbon atom
would only have 6 electrons in its valence shell. But if there were an unequal 
sharing arrangement, where oxygen contributed 4 electrons to the shared pool, 
but carbon only contributed 2, then there would be 3 pairs of shared electrons,
and each of the atoms would have a completed Lewis octet. So carbon monoxide
has a triple bond, but one of the three bonds is a dative or co-ordinate bond,
where both shared electrons are formally contributed by the same atom.
Interestingly, oxygen is a much more electronegative (electron-greedy) atom
than carbon, so although it contributes 4 electrons to the sharing arrangement,
it gets a roughly 2/3 share of the 6 shared electrons, coming out very close
to square. Carbon monoxide is an almost non-polar molecule.

Now there seems to me to be no good reason why a similar story should not be 
told about aluminium chloride. If one of the three chlorine atoms were to 
contribute an extra pair of electrons to a sharing arrangement to make a 
double bond, the aluminium could have a complete Lewis octet. Not only that,
but there are three chlorine atoms, which are symmetrical with one another,
so the task could be dynamically shared around. And of course chlorine is
more electron-greedy than aluminium, so the chlorine atoms would recover 
their share in the unequal sharing arrangement.

The main differences that I see between the two cases are that 
(1) carbon monoxide is a discrete and unreactive molecule, well described with
a triple bond. But aluminium chloride mostly exists as a dimer -- Al2Cl6 -- 
where a chlorine atom does indeed donate a pair of electrons to an unequal  
sharing arrangement, but to an aluminium from a different AlCl3 unit.

    Cl
    |
Cl-Al-Cl
    |  |
   Cl-Al-Cl
       |
       Cl

(2) The structure you are proposing is just one of a set of three
rsonance structures, adding an extra complication.
(3) Genuine double bonds are rather uncommon once you get
beyond the first row of the periodic table. (Si, P, S do not form
double bonds as readily as C, O, N).

If you want to follow up on the more philosophical aspects of
this answer, check out an article 'Chemical Laws and Theories
do not Obey the Rules', in Chapter 2 of  'Of Minds and Molecules' ed. Rosenfeld & Bhushan, Oxford UP, 2000.