Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Gaussian09 or VASP:adsorption of big molecules on surfaces/comp. req. for Gaussian09

  1. Oct 18, 2011 #1
    Hi Everyone,

    I am a PhD student working on different collector molecules used in flotation process (an important mineral separation process). One of the important part of my project will be to study the reactivity of different types of collectors towards some particular mineral(s) by using the DFT methods. However, this area is new to me, though I have made substantial progress through reading reference books and papers that have been published in the past.

    Unfortunately, the field of computational chemistry is kind of new to our group and I am the only who will be working in this area. It is my job to decide on the software to be used. My work will essentially involve big collector molecules adsorbing on the mineral surfaces. Through my understanding I have narrowed down my search for suitable software tool on Gaussian09 and VASP, but I am unable to choose one. As both of these softwares are available commercially so I have to be careful.

    Could someone please suggest which one would be better for my purpose of use. I am kind of inclined towards Gaussian, but when I went through the website I learned that it will be using atmost 2 GB of my computer memory (as I will be buying a single user license).

    CAn someone please advise whether the single user license version will be very slow (assuming i ll be running on a decent comp - anyways in the website it is mentioned that even if i am using a high end computer the software can use only a maximum of 2 GB RAM)

    I have i5 processor with 4GB RAM on my desktop.

    Any help from you guys who are certainly far more experienced that me, regarding choosing the software and computer configuration related question for Guassian would be highly appreciated.

    Thank you.

    Best Regards.
  2. jcsd
  3. Oct 19, 2011 #2


    User Avatar
    Science Advisor

    Re: Gaussian09 or VASP:adsorption of big molecules on surfaces/comp. req. for Gaussia

    First, what you are interested in calculating is a difficult issue, since DFT is intrinsically unable to describe dispersive interactions, and dispersive interactions often play a major role in adsorption processes. There are empirical corrections (e.g., Grimme's DFT+D), but these are often hit and miss. Without applying them, you will definitely not get reasonable numbers, but even if you do, you may well miss your accuracy goals. And you may need to do compromises regarding the size of the molecules you want to treat.

    Regarding the software: If I understand you correctly, you will be dealing with periodic surfaces. VASP is *much* better at periodic DFT than Gaussian.

    About the computer: If you are concerned about 4 GB of RAM and an i5 processor, you are not thinking in the right dimensions for doing computational chemistry. With computational chemistry (and in particular quantum chemistry) it is easily possible to bring any of the most modern computer platforms to their knees, and the system you describe is rather on the low end of what would be a sensible setup to do calculations today. In particular, you want to have lots of RAM, lots of processor speed, and for many applications also lots of disk space on fast discs. There is no real "good enough". What you have (together with your patience) decides over what you can or cannot calculate.
  4. Oct 19, 2011 #3


    User Avatar
    Science Advisor

    Re: Gaussian09 or VASP:adsorption of big molecules on surfaces/comp. req. for Gaussia

    by the way: if you have access to some computing center at your university (or at another place), you may find your solution there. These often have a variety of chemistry software installed.
  5. Oct 20, 2011 #4
    Re: Gaussian09 or VASP:adsorption of big molecules on surfaces/comp. req. for Gaussia

    Thanks alot Cgk,

    Like I said i am quite new to the computational chemistry area. I have found that our university servers have Gaussian 09 installed which I could access, so I guess the computer configuration problem is solved now..

    Our University doesn't have VASP though, so may be I will have to work with Gaussian only. Regarding the difficulties in studying the adsorption of molecules, right now I don't have any clue about dispersive interactions and empirical corrections, but I think I will learn more once i start working on the software. Actually I am chemical engineer and initially my project was supposed to be experimental only, but unfortunately (or fortunately? i don't know) I finished most of the experimental part much earlier than my supervisor had expected me to. So he suggested to include these calculations into my project...that explains why i had such type of questions...but i am working on the theoretical part by following some good reference books and once i will start working on Gaussian i think i will be in a better situation.

    I have one more question, I was looking for a user manual for the software and found an online version on the Gaussian website but it is not very descriptive one. Then on google books i came to know they have also published one book (Gaussian 09 User's Reference - 400+ pages) but i couldn't find from where i can buy the book.

    Does someone has idea whether the online user manual and this book will be same. And the user manual that can be purchased from Gaussian's website in some 60 dollars - is this manual is actually the same book?

    Could someone please tell me whether for a amateur like me purchasing this book will be any help or not?

    Thank you all for the time and consideration.

  6. Oct 25, 2011 #5
    Re: Gaussian09 or VASP:adsorption of big molecules on surfaces/comp. req. for Gaussia

    Whoa whoa whoa! Stop right there. I will go further than cgk on this one and say that Gaussian will not work for your system. Gaussian has pretty much all of bells and whistles that anyone could want for doing calculations or analyzing the results, but it scales really poorly for anything over eight or so processors and simply can't be used for large jobs. This is a computational chemistry/computer science buzz phrase, but it basically means that if you use sixteen processors instead of one, you won't get anything close to sixteen times the performance out of Gaussian. This isn't relevant for small systems that don't require many processors, but for large molecules and certainly surfaces or periodic systems you'll need to use something else. Other codes like NWChem and NRLMOL are free DFT codes that scale much MUCH better than Gaussian, but are more of a pain to use and get support for. Even those though won't work for what you're doing. For your stuff you'll really need a plane wave DFT code (plane wave codes use certain tricks to express the density that make them MUCH faster and describe periodic systems perfectly). VASP is the gold standard right now, but it's kind of expensive. There are alternatives however like Qunatum Espresso, GPAW and CP2K that are all very fast and free. You'll need to get someone at your school's cluster to install them (doing it yourself if you're not experienced will be maddening) and then it will take you a while of working with someone more experienced, trial and error and emailing questions to the listserve before you can get the results you want. There are people whose entire careers much less just their PhD research is doing the calculations you want to do. It's not small task. However, using the wrong software package will ensure failure.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook