Basically, you probably want to do a multi calculation, such as
{multi;
wf,20,1,0; state,1; lquant,2;
wf,20,4,0; state,1; lquant,2;
}
Note the lquant,2 says it is a delta state, the 1 and 4 are the A1 and A2 symmetries of the C2v point group used in Molpro (Molpro doesn't use full linear molecule symmetries), the '20' should be replaced by the real number of electrons and the '0' should be replaced by 2*S (number of unpaired electrons).
Then you want to do a MRCI calculation. For this step, you need to make you include all states of lower symmetry with 1 or 4 symmetry (i.e. Sigma, Pi, Delta and Gamma states), otherwise the energy will be very far off.
Try something like:
{ci; wf,20,1,0; state,1;}
{ci; wf,20,4,0; state,1;}
where you replace the 1 by the number of states of '1' and '4' symmetry equal to or lower in energy than the delta state. Note that you don't use the lquant,2 in the ci part (this is why you need to include all states of 1 and 4 symmetry respectively, otherwise, you will not get a delta state properly). You can check the lz expectation value (which should be 2) by adding