From schrecke -A_T- t12.lanl.gov Tue Sep 30 15:51:48 1997 Received: from mailhost.lanl.gov for schrecke[ AT ]t12.lanl.gov by www.ccl.net (8.8.3/950822.1) id PAA10630; Tue, 30 Sep 1997 15:16:12 -0400 (EDT) Received: from [128.165.22.209] (machgs.lanl.gov [128.165.22.209]) by mailhost.lanl.gov (8.8.7/8.8.7) with SMTP id NAA21615 for ; Tue, 30 Sep 1997 13:16:11 -0600 (MDT) Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Date: Tue, 30 Sep 1997 13:19:06 -0700 To: CHEMISTRY $#at#$ www.ccl.net (CCL) From: schrecke -x- at -x- t12.lanl.gov (Georg Schreckenbach) Subject: summary: solid state/band structures Dear CCL readers, some two weeks ago, I posted a query about solid state methods. Here now is the promised summary. Thanx to those who replied. I didn't get too many answers although they were certainly useful. I would like to also mention again the older summary on the same subject, CCL archives, Feb. 20, 96. In a sense, it complements this new posting. If anybody wants to comment further, please do so! Best regards, Georg Here is the original question: I am interested in solid state (band structure) programs for a research proposal that I am just starting to think about. A CCL archive search shows that this question pops up regularly in one form or another. The last time was just a few days ago! However, the same archive search revealed that there are, unfortunately, mostly questions, and not many answers: The only related summary that I could find is dated 20 Feb. 1996. Some of the relevant programs mentioned briefly in this summary are: - ADF-BAND (http://www.scm.com/) - WIEN97 (http://www.tuwien.ac.at/theochem/wien97/) - and then there are the Car-Parinello type codes, including "PAW" of P. Blochl et al. So far for the background, and here are my questions in more detail. Since this is a very early state of a possible future project, I am trying to gain a broad overview. Basically what I want to learn is the state of the art in the field of doing solid state calculations from a chemical point of view. My questions are, e.g.: - What kinds of programs are out there? (relevant literature?) - What methods do they use (e.g., plane waves or other basis sets; DFT or ab initio -- I am not interested in semiempirical approaches in the moment). Which method might be superior over another? - What functionality is available (ECPs or other ways to include relativistic effects, geometry optimizations, properties ...)? - General experience of users of, e.g., the mentioned programs (source code available?, numerical stability, user friendliness, number of bugs that one runs into, cost ...) - Any other info that you would deem useful. 1) Reply by Keith Refson Georg Schreckenbach writes: > A CCL archive search shows that this question pops up regularly in > one form or another. The last time was just a few days ago! However, the > same archive search revealed that there are, unfortunately, mostly > questions, and not many answers: The only related summary that I > could find is dated 20 Feb. 1996. > Some of the relevant programs mentioned briefly in this summary are: > - ADF-BAND (http://www.scm.com/) > - WIEN97 (http://www.tuwien.ac.at/theochem/wien97/) > - and then there are the Car-Parinello type codes, including "PAW" of P. = Bl=3D > =3DF6chl > et al. A few other programs to add to your list: CRYSTAL (http://www.dl.ac.uk/TCSC/Software/CRYSTAL/). This does periodic Hartree-Fock calculations on solids using standard Gaussian basis sets. No forces or geometry optimization. CASTEP (http://www.msi.com/info/products/modules/CASTEP.html, http://www.tcm.phy.cam.ac.uk/castep/) - Unfortunately now available only to but from MSI, this is Mike Payne's plane-wave pseudopotential code. Also from MSI, the Cornell PWPP code: PLANE_WAVE (http://www.msi.com/info/products/modules/I2_QP.html) CETEP: The parallel version of CASTEP used by the UK Car-Parrinello consortium. Collaboration may be possible. (http://www.dl.ac.uk/TCSC/projects/UKCP/ukcp.html) VASP ( http://tph.tuwien.ac.at/~vasp/) The Vienna ab-initio simulation package. One of the best and most highly developed plane-wave pseudopotential codes. Not free but the price is modest. There's Ali Alavi's FEMD The most widely avaliable LAPW code is WIEN97 which you already have. The other thing to mention is Methfessel's FP-LMTO program. I don't have a WWW address, but I may be able to dig out a literature reference. There are also other 3 or 4 other plane-wave pseudopotential codes which aren't generally avaliable - Parrinello's group amongst others. > > Basically what I want to learn > is the state of the art in the field of doing solid state > calculations from a chemical point of view. You could do a lot worse than look at the UKCP publications list and work in progress, see the URL above. > - What kinds of programs are out there? (relevant literature?) You should find this from the URLs listed. > - What methods do they use (e.g., plane waves or other basis sets; Most of the ones I listed are plane-wave + pseudopotential codes and use DFT. The exception you listed was WIEN97 which uses the LAPW basis set. I believe there is a version which uses pseudopotentials but it's primarily an all-electron method. The exception in my list is CRYSTAL which is a Hartree-Fock program using the usual Gaussian basis sets. > DFT or ab initio Be aware that physicists don't share the chemists' view of using ab-initio to refer only to Harfree Fock and methods based on it. DFT is definitely ab-initio in the physics lexicon. > Which method might be superior over another? DFT is always superior [ :) ]. This is more seriously true in the solid state world (a) because no periodic program implements any higher-level of Hartree-Forck - not even MP2. The best that can be done is the non-self-consistent density-functional correlation correction in CRYSTAL. (b) Hartree-Fock gets the metallic state so badly wrong that there's no point in even trying, so it has a bad name in the physics world. > - What functionality is available (ECPs or other ways to include > relativistic effects, CRYSTAL has ECPs. All plane-wave codes use pseudopotentials, which in the fashionable form are ab-initio, (sometimes) norm-conserving and nodeless. I can't remember whether WIEN97 handles relativistic effects or not but there is an experimental version which uses pseudopotentials. > geometry optimizations, All but CRYSTAL have forces and do geometry optimizations + molecular dynamics. VASP has a pretty extensive list of properties, but you'll have to check on the web sites for the rest. > - General experience of users of, e.g., the mentioned programs > (source code available?, numerical stability, user friendliness, number o= f > bugs that one runs into, cost ...) If you make up a table I can add comments on those I know about. FYI the UKCP is currently running a comparitive evaluation of FEMD, CASTEP/CETEP and VASP. I don't know whether the results are being made public but I will try to determine that in our meeting on Monday and let you know if it's allowed. There's certainly a comprehensive "features" list for them. At the moment most of the excitement is still with the plane-wave pseudopotential methods because of the O(N**2) scaling with system size. O(N) methods are under development (including the UCKP code) but I don't know of any which are in a useable state for general research as yet. There are also many post-LDA developments foing on such as SIC, LDA+U, exact exchange and others which I can't recall at the moment which all promise a much better approximation to "true dft" and hopefully to reality. I'm sure we will find out which of them deliver in the next few years. Hope this info is helpful. sincerely Keith Refson -- ----------------------------------------------------------------------------= -- Email: Keith.Refson # - at - # | Tel: +44 1865 272026 | Dr Keith Refson, = | earth.ox.ac.uk | Fax: +44 1865 272072 | Dept of Earth Sciences = | | Parks Road, Oxford OX1 3PR, U= K| ----------------------------------------------------------------------------= -- 2) Another reply by Keith Refson: Ome more program I nearly for got - "fhi93cp" which is Matthias Scheffler's plane-wave pseudopotential code. It's available from the Computer Physics Communications program library: - see http://www.cpc.cs.qub.ac.uk/cpc/ Keith Refson 3) Yet another reply by Keith Refson, answering a specific question of mine: I have talked to David Sherman of Bristol Uni who has tried out ADF-BAND and who spoke highly of its useability. However as far as I know it can not calculate forces and therefore can't do geometry optimizations which rather rules it out for most things I'd want it for. 4) From: d3g359(+ at +)fido.pnl.gov (John Nicholas) (I asked him about the results of a similar posting earlier on the CCL) Georg, I'm not sure how easy it would be to get all the replies together, but I can pretty much tell you what's available: Crystal (RHF and DFT), DSolid, CASTEP, Bloch-Demon, VASP, QUEST, DECAPO are all availble in various forms. I think the plane wave methods have such an advatage (speed) over the other basis-sets that they are far preferred. Also amsterdam-DFT. Regards, John -------------------------------------------------------------------- John Nicholas Office: (509) 375-6559 Senior Research Scientist FAX: (509) 375-6631 Environmental Molecular Sciences Laboratory Pacific Northwest National Laboratory Mailstop K1-96 Richland, WA 99352 -------------------------------------------------------------------- (End of summary) -- =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D Dr. Georg Schreckenbach Tel: (USA)-505-667 7605 Theoretical Chemistry T-12 FAX: (USA)-505-665 3909 M.S. B268, Los Alamos National E-mail: schrecke ^%at%^ t12.lanl.gov Laboratory, Los Alamos, New Mexico, 87545, USA Internet: http://www.t12.lanl.gov/~schrecke/ =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D