Mysterious Excited State PES Appearing for BH+

I am trying replicate the PES curves for the ground state and the first four excited states of BH+. Here is my input file:

molecule BH {
    1 2
    B
    H 1 R
}

set {
    basis aug-cc-pcvqz
    reference ROHF
    roots_per_irrep [2,0,3,0]
}

rvals = [###] # lots of numbers

for r in rvals:
    BH.R = r
    property('eom-ccsd', properties=['oscillator_strength'])

From this I was able to obtain the PES curves for the first 6 states.

I was comparing these states to ones in literature. I realized that one of the PES curves I had did not appear in the paper. Specifically the curve produced from the 2B1 irreducible representation does not appear in any literature paper. In addition, the energy does not converge at the minimum of this curve (around 1.8 A).

This mysterious curve is similar to literature results for the quartet state, but since I specified a multiplicity of two, I don’t think that is the answer.

Why is the mysterious curve appearing and what could it be?

ROHF-EOM-CCSD states are not spin adapted, so even if your SCF reference wave function is a doublet, neither the ground CCSD state nor the excited states need be true doublets. Quartet contamination is common. However, PSI does not have (yet) the ability to calculate $\langle \hat{S}^2\rangle$ expectation values, we can’t characterize the approximate spin state.