A comparative assessment of the accuracy of different quantum mechanical methods for evaluating the structure and the cohesive energy of molecular crystals is presented. In particular, we evaluate the performance of the semiempirical HF-3c method in comparison with the B3LYP-D* and the Local MP2 (LMP2) methods by means of a fully periodic approach. Three benchmark sets have been investigated: X23, G60, and the new K7; for a total of 82 molecular crystals. The original HF-3c method performs well but shows a tendency at overbinding molecular crystals, in particular for weakly bounded systems. For the X23 set, the mean absolute error for the cohesive energies computed with the HF-3c method is comparable to the LMP2 one. A refinement of the HF-3c has been attempted by tuning the dispersion term in the HF-3c energy. While the performance on cohesive energy prediction slightly worsens, optimized unit cell volumes are in excellent agreement with experiment. Overall, the B3LYP-D* method combined with a TZP basis set gives the best results. For cost-effective calculations on molecular crystals, we propose to compute cohesive energies at the B3LYP-D*/TZP level of theory on the dispersion-scaled HF-3c optimized geometries (i.e., B3LYP-D*/TZP//HF-3c(0.27) also dubbed as SP-B3LYP-D*). Besides, for further benchmarking on molecular crystals, we propose to combine the three test sets in a new one denoted as MC82.
Assessment of different quantum mechanical methods for the prediction of structure and cohesive energy of molecular crystals
CORNO, Marta;
2016-01-01
Abstract
A comparative assessment of the accuracy of different quantum mechanical methods for evaluating the structure and the cohesive energy of molecular crystals is presented. In particular, we evaluate the performance of the semiempirical HF-3c method in comparison with the B3LYP-D* and the Local MP2 (LMP2) methods by means of a fully periodic approach. Three benchmark sets have been investigated: X23, G60, and the new K7; for a total of 82 molecular crystals. The original HF-3c method performs well but shows a tendency at overbinding molecular crystals, in particular for weakly bounded systems. For the X23 set, the mean absolute error for the cohesive energies computed with the HF-3c method is comparable to the LMP2 one. A refinement of the HF-3c has been attempted by tuning the dispersion term in the HF-3c energy. While the performance on cohesive energy prediction slightly worsens, optimized unit cell volumes are in excellent agreement with experiment. Overall, the B3LYP-D* method combined with a TZP basis set gives the best results. For cost-effective calculations on molecular crystals, we propose to compute cohesive energies at the B3LYP-D*/TZP level of theory on the dispersion-scaled HF-3c optimized geometries (i.e., B3LYP-D*/TZP//HF-3c(0.27) also dubbed as SP-B3LYP-D*). Besides, for further benchmarking on molecular crystals, we propose to combine the three test sets in a new one denoted as MC82.File | Dimensione | Formato | |
---|---|---|---|
hf3c_cutini_jctc_2016_publ.pdf
file disponibile solo agli amministratori
Descrizione: Articolo principale
Tipologia:
Altro materiale allegato
Licenza:
DRM non definito
Dimensione
1.92 MB
Formato
Adobe PDF
|
1.92 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.