Periodic Boundary Conditions in Gaussian



 hello --
 
We're trying to learn how to use the PBC routines in Gaussian 03, and we are reaching a stumbling block with our test case of diamond. If anyone has had much success with PBC in gaussian03 we would appreciate your advice.
 
Our input file, a test case of diamond, with a tiny basis set. (Before we do any real calcs we'd like to understand the simple case.)
 -------------------------------
 %chk=diamond2G.chk
 %mem=1024MB
 #HF/STO-3G PBC(NKPoint=48,CellRange=20) Pop(Full)
 diamond test
 0 1
  C         0.445875   0.445875  0.445875
  C        -0.445875  -0.445875 -0.445875
  TV   1.7835   1.7835   0.0000
  TV   1.7835   0.0000   1.7835
  TV   0.0000   1.7835   1.7835
 ----------------------------------
 
First, the gaussian output says it turns symmetry off (even if you turn it on, it cannot identify any symmetry).
 We varied the CellRange from 20 to 50 and obtained the following results:
 						NClRep
 CellRange	Energy		Time (min)	MaxNCR	NMtPBC    Convg
 10		-74.8774038316	   8.01		285	285   0.3482D-04
 15		-74.8788470298	   7.40		591	591   0.3388D-04
 20		-79.7834971028	  335.50 	1357	1357  0.1513D-02
 25		-77.9532971257     47.78	1717	2179  0.1892D-05
 30		-78.4667709573     26.46 	1721    3185  0.1067D-04
 35		-78.8229423341	   32.52	1721    4501  0.7331D-05
 40		-78.9686326165     43.85	1721	6775  0.2602D-05
 45		-86.8332484908   1479.41 	1721	8903  0.2117D-03
 50		-83.4676895141	 1724.68  	1721   11387(convergence failure)
 
(the Cellrange=50 ended with an abnormal termination due to failure to converge)
 
The ones that took a long time to run (20 and 45) not only had large Convg values in the SCF, but also in the population analysis had inequivalent carbons.
 
I'd assume that the energy oscillations are due to the tiny basis set; my chief concerns are with the convergence behavior and understanding the methods.
 My questions are, and I'd appreciate any assistance with these,:
 1) Can one enforce symmetry (space group) in gaussian with PBC? If so, how?
 2) What are MaxNCR, NClRep, and NMtPBC ?
 
3) Why are these parameters be the same for Cellranges <=20, but afterwards the first two level off, while NMtPBC increases between quadratic and cubic 4) Why would the convergence behavior be so erratic? i.e,. some cell ranges seem okay, but some take considerably longer in a seemingly random fashion, with odd results. 5) Does anyone know of any publications or white papers discussing the implementation of PBC in gaussian03? There seems to be virtually no documentation, or citations, in the G03 manual.
 6) any suggestions on how to run PBC calcs in G03?
 thanks for any help you can provide!
 cheers
 Fred
 Dr. Freddie Salsbury, Jr.
 Assistant Professor
 Department of Physics
 Wake Forest University
 Winston Salem, NC 27109
 http://www.wfu.edu/~salsbufr
 http://csb.wfu.edu/(Center for
 Structural Biology)