ground state d9 + d10L

Concerning the IONIC LIMIT (hybridization equal to zero) and ZERO coulomb interaction, a Metal-Ligand complex whose metal is a d9 system (like Cu in CuO) has as ground state the |d9> state. As far as I let tpd to be non zero the gs becomes A|d9> + B|d10 L> (L means a hole in the Ligand shell)

-- the first 10 states are the d states while the last 10 are the Ligand-p ones (linear combination of p orbitals with the same irrep D4h of the d ones)

# pre-factor Determinant

 1  -1.624686788288E-01       11111111111111111110
 2   6.882687767466E-01       11111111101111111111
 3   6.881132977867E-01       10111111111111111111
 4  -1.624503270680E-01       11111111111011111111

So the single-particle removal spectrum should be composed of three states d9L d10L^2 and d8. While I compute the related one-hole green function I should see three peaks each separated from the other by an amount of energy equal to the charge-transfer energy delta. Nevertheless, I see only two peaks: the d8 and the d9L (separated by delta) and there is no trace of the d10L^2 whose position should be at 2delta from the d8 (which is at zero energy due to the absence of Coulomb interaction.

If the gs calculation is correct, as you can see from the wave function I posted, why does the gf calculation not give me the correct spectrum?