On the Electronic Structure of NLi 2 and PLi 2 , Ground and Low-Lying Excited States Demeter Tzeli, Aristotle Papakondylis, and Aristides Mavridis* Laboratory of Physical Chemistry, Department of Chemistry, National and Kapodistrian UniVersity of Athens, P.O. Box 64004, 15710 Zografou, Athens, Greece ReceiVed: September 12, 1997; In Final Form: December 3, 1997 The ground states of the isovalent molecules NLi 2 (X ˜ 2 Π u ) and PLi 2 (X ˜ 2 B 1 ), along with some low-lying excited states ( 2 B 2 , 4 Σ g - , 2 Σ g - , 4 Σ u - , 2 Σ u - , and 2 A 1 ), have been examined using ab initio CISD, CASSCF, and MRCI methods in conjunction with relatively large correlation consistent basis sets. We report total energies, geometries, binding energies, Mulliken charges, energy gaps, and for certain states, potential energy curves. All states examined are bound with respect to the ground-state atoms N or P( 4 S) + 2Li( 2 S), while the mean binding energies N-Li and P-Li of NLi 2 and PLi 2 are 42.5 and 40.2 kcal/mol, respectively. 1. Introduction In the present report we examine via ab initio methods the electronic structure of the ground state and of some low-lying excited states of the isovalent molecules dilithium nitride (NLi 2 ) and dilithium phosphide (PLi 2 ). To the best of our knowledge there are no experimental or theoretical results in the literature for NLi 2 . The PLi 2 system has been observed in the gas phase by Knudsen-effusion mass spectrometry, 1 along with other phosphorus-lithiated species. We are also aware of some, as yet unpublished, ab initio results by Kudo 2 on the ground states of PLi and PLi 2 molecules (vide infra). On the other hand, experimental results on lithiated N- and P-species in the solid state have existed for quite a few years. 3 The interest in these systems is due to their ionic nature and the role they could potentially play as solid-state ionic conductors. Using SCF, CISD(+corrections), CASSCF (complete active space SCF), CASSCF+1+2 (CASSCF + single + double replacements ) MRCI), and MRCI+corrections methods in conjuction with rather large correlation consistent basis sets, we have examined the ground states X ˜ 2 Π u (NLi 2 ) and X ˜ 2 B 1 - (PLi 2 ), as well as the low-lying excited states 2 B 2 , 2 Σ g - , 2 Σ u - , 2 A 1 , 4 Σ g - , and 4 Σ u - for both title molecules. In an effort to better understand the nature of the chemical bond in these unusual species, we have also constructed potential energy curves (PECs) for certain states; in particular for the ground 2 Π u state of NLi 2 we constructed PECs referring to both dissociation channels, NLi 2 f NLi + Li and NLi 2 f N + 2Li. The present report is a continuation of our work on lithiated species; 4 we believe that the simplicity of the Li atom (one active electron), in parallel with its low-lying first excited state (ΔE( 2 Pr 2 S) ) 1.85 eV), presents an ideal case for the study of the chemical bond. 2. Basis Sets and Methods For all atoms, the correlation consistent polarized basis sets of Dunning were used, cc-pVnZ, where n is a cardinal number characterizing the basis set quality. 5,6 One of the nicest properties of the cc-bases is their potential of improving in a well-defined and systematic way the quality of calculation and their convergence toward the “complete basis set” limits of various molecular properties within the methodology employed. 7 For the N and P atoms the quadruple n ) 4 (QZ) quality basis was employed but with the functions of g-symmetry removed (cc-pVQZ-g). For the N atom only and in conjuction with the CISD method, the augmented cc-pVQZ-g (cc-pVQZ + one diffuse function for each symmetry present ) aug-cc- pVQZ-g) was used. For the Li atom the n ) 3(TZ) basis was selected. For instance, the PLi 2 basis set reads as follows: (16s11p3d2f/(11s5p2d1f) 2 ) f [6s5p3d2f/(4s3p2d1f) 2 ], com- prised of 110 contracted spherical Gaussians (i.e., five d and seven f functions). The reason for selecting these particular bases, e.g., n ) 4 for the N and P atoms and n ) 3 for the Li atom, will be discussed further in the next section. As was already mentioned, the theoretical methodologies employed are SCF, CISD(SCF+1+2), CASSCF, and CASS- CF+1+2 (MRCI). In the CASSCF calculations, the Li 2s-, N 2s2p-, and P 3s3p-like orbitals were included in the active space. The Li 1s-, N 1s-, and P 1s2s2p-like orbitals were optimized but always constrained to be doubly occupied. Given those restrictions, our CASSCF space for the triatomic species contains seven active orbitals and seven (valence) electrons. Depending on the symmetry of the state, the size of the CAS space ranges from 112 ( 4 Σ g - ) to 404 ( 2 A 1 ) CSFs, with 988 665 and 1 501 544 CSFs, respectively, in the MRCI space. No corrections for basis set superposition errors were applied, assuming that the size of the basis sets was large enough. It should also be mentioned that the size extensivity error in all MRCI PECs studied was less than 0.1 mhartree. Our computations were performed with the COLUMBUS 8 suite, with some CISD results checked by the MELD 9 code. Also, the MOLPRO 10,11 code was used for certain calculations on the diatomic molecule NLi. 3. Results and Discussion a. The Diatomics NLi and PLi. Recently we have reported on the ground and low-lying states of NLi 4b and PLi 4a molecules. With the purpose of selecting basis sets of adequate size for the triatomic species NLi 2 and PLi 2 , while at the same time keeping our calculations manageable, we have reexamined the dissociation energies (D e ) and bond lengths (r e ) of NLi and PLi ground states as a function of basis set size. Our results at the MRCI level, along with literature results, are reported in Table 1. Considering the D e and r e as the most sensitive parameters 2223 J. Phys. Chem. A 1998, 102, 2223-2230 S1089-5639(97)02998-8 CCC: $15.00 © 1998 American Chemical Society Published on Web 02/28/1998