electrostatic potential based population analysis techniques

From: "Langenaeker, Wilfried [JanBe]" <WLANGENA(-(at)-)janbe.jnj.com>
Subject: electrostatic potential based population analysis techniques
Date: Thu, 24 Jun 1999 08:13:39 +0200

Hi,
 We are using the electrostatic potential based population analysis
 techniques on a number of organic systems and encountered a problem with
 benzene.
 The symmetry is not present in the charges: we found two types of C-atoms
 and H-atoms (both in SPARTAN and Gaussian). Increasing the number of points
 used in the procedure does solve the problem but also increases the cpu-time
 dramatically. Can anybody explain why this effect is so large (in SPARTAN at
 the HF/3-21G* level with 1 point per atomic unit the two charges for carbon
 are -0.163028 and -0.118392) and tell us if it is only a problem in highly
 symmetrical systems? Does anyone know about a systematic study of this
 phenomenon?
 Best regards,
 Wilfried Langenaeker