Faults specific to PM7

The following faults in PM7 were discovered after publication of the method. Users are advised to be aware of these faults when using PM7. Please report faults in PM7 that are not on this list to openmopac@gmail.com

Fault No.	Date reported	Compound	Fault	Implications
1.	February 28, 2013	HAl(CH₃)₂	Charge on Al is unreasonable: -2.45 C-Al distance is ~0.5Å too long	Calculations where a hydrogen atom is near to an aluminum atom should be avoided. Use PM6 instead.
2	December 29, 2013	Watson-Crick Guanine-Cytosine base pair	Optimized geometry is not planar	This error is small but definite. The planar form is 0.83 kcal/mol higher than the non-planar form. Probably it will not affect results for proteins or DNA-type systems significantly. The error is caused by the hydrogen bond approximation used in PM7.
3	March 11, 2014	Donor-acceptor complex B(C6F5)3-P(CH3)3	A P-F covalent bond is predicted	The P-F interaction is not repulsive enough at VDW distances. This might be due to the core-core terms. An attempt will be made to correct this fault.
4	June 23 2014	DNA-type systems are best modeled using PM6-DH+. PM7 gives N-H - N hydrogen bonds that are too long.
5	May 22 2015	Phenylalanine	Cα-Cβ-Cγ angle is 110.0°, should be 113.6°	Registers as an error with MolProbity. Fault common to aromatic residues.
6	May 22 2015	Proteins	Non-bonding residues are too close together	Registers as an error with MolProbity.
7	November 30 2015	Cl-Cu-P(Ph)3	Cu - P distance too large	Results for systems with Cu bonded to P will be very inaccurate - use PM6 instead.
8	April 20 2016	hexachlorobenzene crystal and in crystalline Cl₂	Cl - Cl non-bonded distances are too small	The heat of sublimation is incorrect.
9	May 20 2016		Silver - silver interactions are inaccurate.
10	June 22 2017	(CH₃)₃Si-O-Si(CH₃)₃	Si-O-Si angle is 180°	Geometries of Si-O-Si molecules will be faulty. Most minerals are not affected because of other strong interactions.