CCL:G: generate electrostatic potential map problem in Gaussian

Hi Everyone,

I experience a problem in generating electrostatic potential map with Gaussian calculation. I have tried small molecules (2 water molecules). After single point calculation, the "Total Density" type of cube can be generated in "GView 3 -- Results -- Surface -- Cube action" with fchk file. Then with this cube, the "ESP type" new mapped surface can be generated in "surface action", which is what I expect.

But for a relative larger system (about 20 atoms), in the "cube action" step, an error message appears:
"surface build failed.
No data to plot.
Isovalue my be out of range."

So I tried cubegen to generate potential cube file from fchk file. Then the following error message appears:

PGFIO-F-231/formatted read/unit=20/error on data conversion.
File name = 3lbw-DP-DoD1-NMR-fullQM.fchk    formatted, sequential
access   record = 297544
In source file fchkrd.f, at line number 41

I have the following questions:
(1) why the electrostatic potential surface need to be generated based on "total density" type of cube instead of "ESP" type of cube? I also tried the latter cube, but a strange surface can be obtained.

(2) Is there any additional keyword needed in the root section of gaussian calculation for ESP calculation?

(3) I do not understand why only ESP of small molecule (< 10 atoms) can be generated? And why I fail in medium size system? Is there any atom size limitation for cubegen in both g03 and g09? My friend and I tried both version, but neither of them can generate potential cube file from fchk file. The command we used is "cubegen 0 potential a.fchk a_pot.cube 80"
 
Any suggestion is appreciated!

Best,
Hao