Lattice parameters and ICHARGE problem

[vaspman]

Hello VASP users
As a new VASP user, I am doing some practice calculations and I did without any problem and now I am doing calculations for wurthzite ZnO. Before doing calculations I want to find the lattice parameters and indeed I found it. I have a question which I don’t understand at the moment that why such thing is happened. When I use ICHCHARGE=0 (calculate the charge density from the initial wave functions) for different K-meshes (Gama centered )and it gives me the following parameters (For 443 mesh)
E(eV) a b c
-10.128658 3.250 3.250 5.476
I used different K-meshes and the lattice parameters are in the same range but the but changes was observed in the total energy i.e. Energy is not well converged.
Then I carried out other new calculations and I used ICHARGE=2 (Take superposition of charge densities ) and use different K-meshes (G-centered) and gave me the following results (for 443 mesh)
E(eV) a b c
-21.126424 3.19 3.19 5.17
And in this case I also used different K-meshes and gave me almost same results and the Energy was well converged w.r.t K-points. These values are in agreement with previous VASP calculations and the experimental parameters are

a b c
3.250 3.250 5.207

I will say ICHARGE=2 is okay because the Enrgy is well converged w.r.t. k-points and the lattice parameters are not so bad.
But my question is Why such behavior ? here I can compare my lattice parameters with either with the previous published calculations or with the experimental values. But if some one is optimizing new materials then how we will say which one way (ICHARGE= 1 or ICHARGE=2 ) is good.
I hope someone will guide me.
Thank you.

[admin]

If you don't start from a valid WAVECAR file, ICHARG is reset to 2 anyway, therefore I think something else was done severely wrong in your first calculation. (the energy difference is much too large to be reasonable) I hardly can believe that the results (energies) you show above are converged (electron, ion.) results, just differing with respect to the k-meshes alone .

I would strongly suggest to read the vasp-workshop toturial files and some basics in the online handbook before you start with your calculations.