BSSE issue: atoms with ecp basis

Hi,

I'm trying to run dft with BSSE correction.
Firstly, I wanted to test a simple system of Gd3+ and H2O interactions.
I seemed fine for the first four dft runs
such as [Gd3+--H2O], [G3+], [Gd3+--ghost H2O] and [H2O].

However, once it got to the point of calculation between H2O and ghost Gd3+,
it died with an error message of
"dft_inpana: negative no. of electrons ? -18".

I doubt that there are some problems in calculation with atoms with "ecp" basis,
since similar calculation with neutral or cationic Na has no problem in BSSE.

It would be very appreciated if someone helped this problem.





Input file ====================================
echo
title bssebq
start bssebq

scratch_dir ./scratch
permanent_dir ./scratch

charge 3
geometry noautoz

 symmetry c1

 Gd   0.00257245    -0.01251597    -3.12708552

 O   -0.00109882     0.00155421     1.05737837

 H    0.00208036     0.79190602     1.66072605

 H   -0.00355400    -0.78094426     1.67088586

end


basis "ao basis"

 Gd    library               "CRENBL ECP"

 O     library                3-21g

 H     library                3-21g

 bqGd  library      Gd       "CRENBL ECP" 

 bqO   library      O         3-21g

 bqH   library      H         3-21g

end
ecp

 Gd    library               "CRENBL ECP"

 bqGd  library      Gd       "CRENBL ECP"

end


dft
xc b3lyp
odft
mult 8
iterations 500
tolerances tight
decomp
mulliken
end

driver
maxiter 200
end

bsse

 mon metal 1

   charge 3

 mon water 2 3 4

   input "dft\; mult 1\; end"

end

task dft optimize

============================================