Analyzing Natural Bond Orbitals (NBO) results from Gaussian ’03 & ’09

I’m posting this white paper “as is” but I will keep on coming back to update it; the thing is that I haven’t had the time to post anything else lately and if I don’t do it like this then I will never get it out. Please be kind on your comments and ratings!

For a more thorough discussion of NBO analysis please check out the references.

Requesting an NBO population analysis on the route section of a Gaussian job (via the simplest option pop=nbo) generates a lot of information on the output file which sometimes is not too straightforward to read. Please refer to previous posts in this blog (all under the NBO category) for more info on NBO calculations. The first part of this post deals with the analysis of the default information obtained from an NBO calculation; the second part deals with some useful/popular options of the NBO analysis which are controlled through the pop=nboread keyword available in NBO3.1 which is the version incorporated in Gaussian.


Perhaps the most common issue when reading NBO results is making sense of the hybridization shown. Take the example below as taken from the NBO manual for Methylamine:

(Occupancy)   Bond orbital/ Coefficients/ Hybrids


1. (1.99858)  BD  (1)  C 1 – N 2

(40.07%)    0.6330*  C1 s(21.71%)p 3.61(78.29%)

-0.0003  -0.4653  -0.0238  -0.8808  -0.0291

-0.0786  -0.0110  -0.0000  -0.0000

(50.93%)     0.7742* N  2  s(30.88%)p 2.24(69.12%)

-0.0001  -0.5557  0.0011  0.8302  0.0004

0.0443  -0.0098  0.0000 0.0000

The first line shows the occupancy (between 0.00 and 2.00 electrons), then the label of the NBO (BD = Bonding (2 centers); CR = Core (1 center); LP = Lone pair; RY = Rydberg; BD* = Antibond), after that a ‘serial number‘ which corresponds to the connectivity (or idealized bond index between the atoms, i.e., single double triple bond), and finally the atom(s) to which the NBO belongs.

The next line describes the natural atomic hybrids of which the NBO is composed, giving the percentage (100|cA|^2) of the NBO on each hybrid (40% the C hybrid orbital and 60% the N hybrid orbital), the polarization coefficient cA (Nitrogen is more electronegative than Carbon, hence the polarization coefficients look like 0.633 for C and 0.7742 for N, i.e. this NBO is more polarized towards N), the atom label (self explanatory) , and a hybrid label showing the sp^x composition: for the C hybrid orbital it is sp^3.61 while for the one on N is sp^2.24. The number in parenthesis is just the percentage in composition of the hybrid (75%p and 25%s corresponds to an ideal sp^3 hybridization) Now, for the tricky part: Sometimes you’ll get numbers larger than 3 for an sp^x hybrid or even s^yp^x hybrids (for y>1 and x>3), this has to do with the basis set employed and the number of functions used to describe each atomic orbital. (EXAMPLE SOON, IM LAZY)

Below the hybridization we find the set of coefficients that specify how the NHO is written as a linear combination of Natural Atomic Orbitals. The previous NHO for C has the largest coefficients from the second (-0.4653) and fourth (-0.8808) natural atomic orbitals, corresponding to a rough description like: NHOc = -0.4653(2s)c -0.8808(2px)c


To control the options in the Natural Bond Orbital population analysis the line $NBO keywords $END must be included at the end of the input file (after the infamous blank line located after the molecule specification)

Wiberg bond indexes

This is a feature I use a lot when dealing with adducts with closed shell molecules. It is achieved through the BNDIDX keyword. Two indexes are obtained: The Wiberg bond index which is presented as a matrix; and the Wiberg bond index total which is the summation of all Wiberg indexes for every atom. The latter index roughly resembles the number of covalent bonds each atom forms. When analyzing this total index on a given atom, it should be compared to a well defined one in another part of the molecule just to make sure that all numbers are reflecting the same trends. Wiberg’s is not the only bond index provided by NBO analysis: the BOAO keyword generates bond index on the basis of the natural atomic orbitals.

Some common errors and possible solutions

->> Concerning NBO Deletion analysis

***** WARNING ***** The variational principle has been violated and the above deletion energy is invalid!!

This usually caused by one of the following:

1) Wavefunction symmetry breaking: Use the Nosymm option on the route section of your input file.

2) The use of DFT methods: Strong deletions often lead to densities that derail the density functional.  Safest is to use HF, where the variational theorem can be counted on, and
DFT artifacts are averted.  The FIXDM keyword (available only in NBO 5.0) may also correct some numerical problems associated with large basis sets

3) Deletion of some degenerated orbitals which breaks the symmetry of the wavefunction. Sometimes it could be that although there are no formal degenerate states accidental degeneracies might occur, deletion of which will cause this error.

->> Also on NBO Deletion Analysis


This is one weird error. I have solved it by changing the molecule specification from Z-Matrix to Cartesian coordinates. Also, if you are running a DFT calculation in an older revision of Gaussian 03, try using the IOp(5/48=10000). Later revisions of Gaussian 03 and the more recent Gaussian 09 claim to have fixed this problem.

->> NBO Analysis

WARNING: Population inversion found on atom X#

This is not a problem! NBO lists orbitals within each atom according to their energy and their population afterwards. This warning only states that some low energy lying orbitals are less populated than others with higher energy. It cannot be “fixed” since this is a natural consequence of the orthonormalization process of the NBO generation.


NBO5.0 Home. Site of the creator of NBO and NBOView (Dr. Frank Weinhold)

FAQ‘s on NBO Home. I can’t recommend it enough!!!

Population Analysis Keyword in Gaussian


About joaquinbarroso

Theoretical chemist in his early thirties, in love with life and deeply in love with his woman. I love science, baseball, literature, movies (perhaps even in that order). I'm passionate about food and lately wines have become a major hobby. In a nutshell I'm filled with regrets but also with hope, and that is called "living".

Posted on November 11, 2009, in Computational Chemistry, Gaussian, NBO, White papers and tagged , , . Bookmark the permalink. 86 Comments.

  1. I am currently investigating transition state structures of Diels-Alder and 1,3-dipolar cycloadditions. I have performed an NBO anaysis on a few of these calculations (QST2 or QST3 in G03). I have some problems interpreting the output but I am better after reading some posts here and elsewhere.

    However, in my readings I have come to realize that one of the most important criteria used to identify bond development i.e. asynchronousity in the TS, is via analysis of Wiberg bond indices. Is there any literature, examples, tutorials, etc that allows one to teach themselves to read a Gaussian output and obtain the relevant information?

    In general, I am concerned TS calculations and their structures. What insight can an NBO anlysis furnish oneself? Specifically, what keywords would one want to include in their calculation that afford insight into transition states?

    References and tutorials in any capacity are welcome.

    Best regards,

    Glenn Abbott

  2. Hello!

    First let me appologize for my late response, I was on vacations and i completely neglected real life for a while.
    In order to obtain Wiberg Indexes in your input you have to use the BNDIDX keyword in the $NBO section at the end of your input file.I don’t know of any tutorial about reading Wiberg bond indexes in Gaussian outputs, except for the NBO manual which I think you can consult directly at:

    Wiberg’s original reference is:
    Wiberg, K.B. Tetrahedron, 24, (1968) 1083.

    Now, reading wiberg’s indexes is very easy, you just have to search for the string ‘Wiberg’ in your input and that will take you to the Wiberg bond index (WBI matrix section in which you can read the WBI for any par of atoms by just crossing them in the matrix. The next place where the string is found is the “Wiberg Bond Index Total by atom” This section adds ALL of the wbi’s displayed above for each atom, which roughly resembles (or should resemble) the number of total bonds formed by each atom.

    About the keywords suitable for TS analysis I think it depends on what kind of information you want to get. Since NBO is a Lewis reinterpretation of the wavefunction I think its use for TS structures would allow you to understand which electron pairs are being broken and possibly the Second Order Perturbation Analysis (obtained with the E2PERT keyword) would also yield the corresponding energy to such breaking process. I don’t have any experience with NBO and TS structures, I think it’s interesting and worth the research so please keep me posted about your results; I will sure do some trials of my own and will post them here.

    Best regards and thanks for reading!

    • I forgot something…

      If you are performing Geometry Optimizations you have to be careful what NBO analysis you are looking at, since it gets performed both at the starting geometry and at the optimized one; so of course you want to take a look at the final NBO analysis, or in other words you will find the ‘Wiberg’ string four times in your file (the firs two corresponding to the analysis of the starting geometry)

      I hope this helps and as usual I’m available for any other thing you need

      Best wishes!

  3. Hi,
    I have the following case and I’ll be grateful for any suggestion(s).
    – I have crystal structure, in which there is interesting connection between two atoms and I want to study this bond by studying the NBO and to check the atomic orbitals that make this connection.
    The problem is I tried many optimization methods and changed the bassis sets and played with the charge and multiplicity and in every time after optimization I lost the new bond. So, how to solve the optimization problem and is there any way to make single point calculation, followed by NBO?
    – The best optimization method gave me 0.2A difference with the crystal structure however there is some modification in the bond conformation and that made me worried that the new structure is different. So, is that divination accepted?
    – I’ll be happy to listen to any idea or suggestion.
    Thank you very much
    Best regards

    • Hi Kem!
      When you take a crystal structure and you optimize it you do it in a vacuum, that means you are taking any other interaction or force field out of the equation. Out of the crystal field your molecule might look quite different so maybe you should try to use a single point calculation unless you find too many imaginary frequencies. You may play with the basis set and method (a.k.a. level of theory) but playing with the charge and multiplicity makes no sense, you are removing/adding electrons to your structure! you can’t expect the wavefunction to behave the same way when you are in fact calculating different electronic configurations. If you still insist in optimizing it then try the option MaxStep=N where N relates to the size of the step as 0.01N bohr or radians (default N=30) so the optimization is more constrained. Another approach you might use is to freeze the bond; use opt=modredundant and after the molecule specification and before the NBO section type atomnumber1 atomnumber2 valueinangstroms F
      The “F” will freeze that bond.
      So in a nutshell, differences between optimized and crystal structures are to be expected. Maybe you are finding out that such a bond is an artifact produced within the context of the crystal field! (cant wait to read your paper 😉 ) try calculating its properties without optimizing. Do not play with the charge and multiplicity, your molecule or ion has the number of electrons it has and that is that!

      Have a nice day and thanks for reading

  4. dear sir,

    i am karthick research scholar in the field of vibrational spectroscopy. our team tried to get potential energy scan from Gaussian 03. but scan option in 03 requires input variables. but our doubt is “where the input variables to be attached?”. from your website we studied a lot in our research basics. we hope your favorable reply will help us a lot and best coordination between ours.

    thanking you,

    • Hello Karthick!
      I’m glad to know the site has been helpful to you. About your question, take a look at this other post of mine under this link.
      The variables are placed at the end of the file after the molecule specification (coordinates), their syntax depends on the kind of scan you want to perform. I hope this answers your question but if it doesn’t then let me know and I will try to do it again.

      Have a nice day!

      • sir,
        i am Nijesh, research scholar doing phd in computational chemistry. your blog is very helpfull to me. sir how can get i overlap matrix/overlap population using gaussian09

        Nijesh K.
        Dept of Chemistry
        NIT calicut

      • Hi Nijesh,

        Sorry for the lateness of my response. Just include IOP(3/33=1) in your route section.

        Have a nice day

  5. Mohammad Chahkandi

    I want to study the nucleophilic attack of bromide to some complexes and oxidation of bromide to HOBr. I think maybe investigation about HOMO of Br- and LUMO of related complexes are useful. Lower gap energy between these two orbitals indicate the faster bromide oxidation reaction. Maybe survey the orbital size and spreading is helpful. Actually, I am not sure which kind of caclucaltions with Gaussian program are suitable for this situation.
    Your guidance is my pleasure.

    Best regards,

    Mohammad Chahkandi

    Inorganic and Analytical Chemistry II
    Friedrich-Schiller-Universität Jena, Germany

    • Hi Mohammad!
      I think you are on the right path, analyzing the gap between frontier orbitals is a good approach. What you need to do then is to run a population analysis, NBO is a robust and suitable method for your needs so give it a try! use pop=NBO in your route section and then proceed as described in this post. I also encourage you to run a Fukui indices analysis (find the post in this same blog)

      Have a nice day!

  6. dear sir,
    I am saran. Im working in metal -organic molecule interaction studies. how to define the K-points for a particular system for eg.,Al(100) plane.

  7. Концерты, песни, хиты, треки, видеоролики на Concertus.Info

  8. Dear sir,
    Iam Mrs.Rajamanikumaran research scholar in the field of vibrational spectroscopy.i wish to get your suggestions for my work.i wish to calculate hyperpolaraizability and NBO.i got 10 components for hyperpolarizability. output i got from gauss03w was Final packed hyperpolarizability:
    K= 1 block:
    1 0.170195D+03
    K= 2 block:
    1 2
    1 0.150704D+03
    2 -0.150383D+02 -0.229488D+03
    K= 3 block:
    1 2 3
    1 -0.514043D-01
    2 -0.215172D-01 -0.114820D-01
    3 0.660429D+02 -0.636988D+02 -0.457934D-01 how to take beta components from this and how to convert it to esu.To calculate NBO i gave #rhf/6-311+G(d,p) pop=nbo in the route section.i didnt get the nbo output.please give me your suggestion regarding this.
    your suggestions are very valuable for me and this site is very useful to me.
    Thanking you

  9. Dear sir,
    I am devi reseach scholar in the field of computational chemistry. I want to know how to calculate wibergs index, and how to give input file for finding wibergs index in guassian 03,and also i request you to give the details beta components of hyperpolarisibility in guassian 03 output file, for example beta value is 0.660429D+02 like that what is the value of D here.

    • Hi Devi

      You can calculate the wiberg bond index by incluiding pop=NBORead in your route section and then at the end of the file leave a blank line followed by:
      This will calculate the Wiberg bond index.
      About the D in 0.6604D+02 it is a way to write scientific notation, i.e., it means 10^2 hence 0.6604D+02 = 0.6604×10^2 = 66.04
      A negative sign after the D would mean a negative exponent for example 12.34D-02 = 12.34×10^-2 = 0.1234

      Hope this helps

  10. Dear sir,

    I am Devi, Thank you sir, your valuable suggestions helped me a lot. I have one more query in wibergsindex, how to write the input file in route section for finding wiberg index in guassian 03. I does not know about the route section basics. I request you to give the suggestions for me.
    Thaking you,
    Yours sincerely,

  11. Thank you so much for this helpful blog. I have run the NBO analysis, and have the data in hand. However it seems that the output from the NBO reorders my MOs from what I view them in with gaussview. So instead of NBO 1 being the first surface I view, it may be something like #17 or something like that.

    I’m talking about output like this:

    Natural Bond Orbitals (Summary):

    Principal Delocalizations
    NBO Occupancy Energy (geminal,vicinal,remote)
    Molecular unit 1 (C6H4)
    1. BD ( 1) C 1 – C 2 1.99006 -0.89691 124(g),125(g),50(v),92(v)
    2. BD ( 2) C 1 – C 2 1.93227 -0.36735 128(v),134(v),124(g),125(g)

    These two orbitals when I view them in gaussview actually correspond to #7 and #17. Is there any easy way to correlate these two without either comparing the energy values or the output from the “sorting of NBOs”?

    Thanks again for this great blog!

    • Hi,

      Thanks for your kind words about the blog.
      It would seem like the orbitals were reordered but remember the basis was re-orthonormalized. Plus the NBO sorting uses a different criteria than the canonical MOs. I wouldn’t worry about this, for they only resemble in shape but not in energy.

      Hope this helps!

      Best wishes

  12. Respected Sir

    Natural Bond Orbitals (Summary) consist numerous values. From which how we interpret the calculation results. Sir any common methods is used for sorting these results.

    Senior research fellow
    Dept of chemistry
    University of calicut

  13. Great blog!
    Quick question. We are trying to determine the relative interaction with a protein and a inhibitor. We have crystal structure data (from a .pdb) for the protein and are attempting to put different inhibitors in the pocket (which we assume we know where it is). We have optimized ONIOM calculations for several inhibitors and now want to look at NBO’s to distinguish what interactions are important.

    However when i run a NBO with the following (obtained from your site, many thanks!)
    #P opt rhf/6-31G(d,p) pop=(full,nboread) gfoldprint geom=connectivity

    blank iine

    i am seeing the structure of our cut out piece of protein and inhibitor move with the different cycles. This seems alarming to me as we spent much time trying to get their optimal structure already. Is this the correct way to go about gleaming interactions between a protein and a inhibitor? And should i be worried about the movement of the molecules?

    • Hi Steve!

      Thanks for your kind words about the blog!
      This seems to me like a very thorough way of performing a docking analysis. It could get very expensive if you try to probe many different substrates, however if you can manage the calculations in reasonable amounts of time then go for it!
      From what I see in your input you are asking gaussian to optimize the structure again. If you already did it with the ONIOM method then you can drop the ‘opt’ keyword and just run the NBO analysis of the intermolecular interactions and the obtained results will correspond to the ONIOM structure.
      So in short. This is a correct, although probably unorthodox, way to probe the interactions between a substrate and a protein; I hope I can take a look at your paper when its done! About the movement of the molecules, this is not a problem but it will for sure take a long time to re-optimize the full system (which already took a long time optimizing with the ONIOM method, right?)

      I hope this helps; if it doesn’t, please don’t hesitate to contact me again . Thanks for reading!

  14. Hi,
    This blog really is very helpful. I’m a grad student. I have been recently trying to use NBO analysis to try and get an indication of which orbitals are interacting the strongest. I do a lot of CASSCF and CASPT2 work to try and find transition states for electrocyclic isomerization reactions where I break sigma bonds. A lot of them are very bi-radical and I know that the HF wave function fails to describe them correctly. My question is do you think that the info I get from the perturbation theory analysis of the Fock matrix is relevant if the HF wave function is doing a poor job? I was hoping that this method would still have some qualitative use to help rationalize why several activations barriers were lower in energy than I expected. Any advice would be greatly appreciated.

    • Hi Jeff,

      This is a very interesting question! I think this is a very good way to at least probe the space of those orbitals that might interact more strongly than others. I think that at least qualitatively you are safe and I see no problem in doing so if you save a lot of time on the CASSCF and CASPT2 calculations. Of course a systematic study would ensure this approach is sound and theoretically consistent but in the mean time I think you are ok. Nice work!

      Thanks a lot for your kind words about the blog. I’m glad you’ve found it useful

      Have a nice day!


    I am having Gaussian 03, and i want to know how to calculate NBO, internal coordinates, and symmetry coordinates with model calculation

  16. Hi Jeff,

    This is a very interesting question! I think this is a very good way to at least probe the space of those orbitals that might interact more strongly than others. I think that at least qualitatively you are safe and I see no problem in doing so if you save a lot of time on the CASSCF and CASPT2 calculations. Of course a systematic study would ensure this approach is sound and theoretically consistent but in the mean time I think you are ok. Nice work!

    Thanks a lot for your kind words about the blog. I’m glad you’ve found it useful

    Have a nice day!

  17. Hi sir
    I performed a NBO analysis (with gaussian program) of an
    open shell molecule employing the MP2 density (density=mp2) and
    the following message appeared:
    A bond orbital with an occupancy of – 0.00002 electrons was found!
    Please check you input data.

    • Hi Zahra,

      I would need to see your input data to be able to help you. An electron occupancy below zero is just wrong, I don’t know what is going on.

      Have a nice day

      • Hi Joaquin
        Thanks very much for your input is:# mp2/6-311++g(d,P) pop=(saveNBO) Gen 6D scf=(maxcyc=1000) density
        and gave this error:A bond orbital with an occupancy of -0.00009 electrons was found!

        Best regard for you

  18. Hi Dr.joaquin
    I performed an optimization on Cu with this input
    # opt=(maxcycle=300,calcALL,steep,gdiis) b3lyp/6-31g(d,p) scrf=(solven
    t=water,read,pcm) scf=(maxcyc=2000,vshift=800) integral(grid=ultrafine
    and give this error
    GPN1st: Unable to complete.
    Error termination via Lnk1e in /share/apps/g09/l701.exe at Sat Apr 21 14:26
    How I can solve this problem?
    Have a nice day

  19. Not to say that i know what i am doing. But it does appear that you have two max cycles at different numbers, 300 following opt and 200 following scf?

  20. sir i run my compound using gaussian03,i got one negative frequency and 77 positive how to modefy negative frequency.

    • Hi
      negative frequency means that there is a transition state

    • Hi Praba,

      You need to visualize that particular vibration with some visualization software like GaussView, then see how the molecule is getting distorted, this will tell you in which direction some atoms are trying to accommodate themselves. Then modify your input in that direction so its closer to the actual minimum.

      I hope this helps. Have a nice day!

  21. my input #p b3lyp/6-31G(d,p) opt freq=raman
    got one nagative freq and 77 positive freq

    • Prabhakar Sharma

      if you are looking for TS then its is good neither change some coordinate and reoptimize it
      Prabhakar Sharma
      Devi Ahilya University Indore

  22. i did’t understand

  23. Hi Doctor
    I should say only Thank you ,for your good blog.
    thanks a lot

  24. Dear Sir,

    I am an Italian student and I am trying to setting up a simple nbo calculation with g09. However, the following input doesn’t produce any NBO output and no errors appear. I need to avoid the use of the POP=NBO keyword.



    0 1
    C 0.622637 0.000000 -0.931455
    H 1.663280 0.000000 -1.216741
    C -0.622637 0.000000 -0.931455
    H -1.663280 0.000000 -1.216741

    $NBO $END

    thank you in advance,


    • Buon giorno, Luca!

      You are asking the program to read some keywords but there aren’t any! May I ask, why do you need to avoid pop=nbo? anyway, you can place a keyword like E2PERT which will calculate the 2nd order perturbation analysis which is performed by default anyway. You could type BNDIDX and get wiberg’s bond orders which are most of the time helpful.
      If you need your log file to be stripped from mostly anything other than the basic NBO analysis you can use the keyword NPA which will only perform the Natural Population Analysis

      Spero che questo servirá! 🙂

      • I have also tried with E2PERT but it doesn’t work. I also wanted to perform some deletions and I have tried:

        # HF/cc-pvdz NOSYMM POP=NBODEL


        0 1
        C 0.622637 0.000000 -0.931455
        H 1.663280 0.000000 -1.216741
        C -0.622637 0.000000 -0.931455
        H -1.663280 0.000000 -1.216741

        $NBO $END

        but I’ve got the same results (the output stops without problem but it runs only the SCF procedure, without NBO analysis). It works only with the POP=NBO keyword but I need also to carried out some deletion analysis.

        grazie mille per l’aiuto 🙂

      • What version of Gaussian are you using? Is it properly installed? Have you been able to perform deletion calculations with this configuration?
        It is weird indeed. Send me more information so I can do a better job helping you.


  25. I have run the same calculation with other programs. Now I am using the last version of G09 but I have also tried a previous version and the problem it was still the same. The program are proprerly installed and widely used, but I am the first in our group trying to perform NBO
    analysis, so I don’t know if I should compile something else.

    It seems like it cannot recognize the line $NBO $END, since if I run the previous calculation (the one with the deletion) it gives me this error: NtrOpn-Old failed and, if I run the job with the NBOREAD and $NBO E2PERT $END it stops after the SCF without errors but without performing NBO analysis.

    It is the first time I use gaussian so most likely I did some trivial error.

    • I think this calls for Gaussian’s help desk; they should be able to help you. My guess is that the NBO program (NBO3.1 as it is supplied with gaussian) is not properly installed.

      Mi dispiace, ma non so che cosa e male (I forgot the italian word for ‘wrong’) 😦

  26. Hi, I am shova from Nepal. I m doing Msc thesis on adsorption of chlorine on H-passivated graphene. Mainly my job is to calculate binding energy, HOMO-LUMO gap, charge transfer and so on by using gaussian 03W. What is the problem is that i got unusual binding of halogen with different no. of carbon atoms. I would be happy if you suggest me about that .

    • Hello Shova
      Could you please be more specific about what do you mean by unusual binding? Are you optimizing your structure first? What level of theory are you using and what kind of population analysis have you chosen?
      I would need more information to help you out, sorry.

      Have a nice day!

  27. Hi,
    This blog really is very helpful. I’m a Phd student. Now I am using the last version of Gaussian 09. I want to calculate on-site Coulomb repulsion U between electrons (such as 3d orbitals of cobalt atom ) for my system. Is it possible with NBO? or Are there any keywords?

    • Hello!

      I’m glad you’ve found it useful. Thanks for reading it!
      I’m not aware of any method to perform what you are asking but it sure sounds interesting. If you are in a hurry and need a quick answer, please try to contact they will promptly get back at you. If possible, let me know what they said, ok?

      Have a nice day!


    i am try to calculate the band gap of different polymer how to calculate it

  29. hello, can you explain to me how to calculate the WIBERG indexes with a small example please.(imidazole molecule)

    • Hi Sophia!

      Use the pop=NBORead option in your route section then write the line
      at the end of your input file (leaving a single blank line before and after it!)
      Try looking for other NBO tutorials here in this blog, they will be most helpful to you.
      I hope this helps

  30. Whichever is correct in the calculation of charge by nbo in gaussian software. using minimization or single point calculation after minimization of geometry ?

    • If you perform a geometry optimization calculation, the NBO analysis will be performed both before and after, so be careful to find the charges after the geometry has been optimized.
      You want to always optimize your system before performing any analysis on it.

      Have a nice day!

  31. Thank you very much

  32. Hello, I would like to ask how to change for Second Order Perturbation Theory Analysis of Fock Matrix in NBO Basis the Threshold for printing of 0.50 kcal/mol to a lower value.

    Thank you in advance
    Isabela Man

  33. #NBO change treshold for printing

    Hello, I would like to ask how to change for Second Order Perturbation Theory Analysis of Fock Matrix in NBO Basis the Threshold for printing of 0.50 kcal/mol to a lower value. I mention that I work in Gaussian for windows (both G03 or G09)

    Thank you in advance
    Isabela Man

  34. Hi Joaquín,

    I’m phD student in computational chemist and recently I used NBO, but I had a little problem and it would be nice if you could help me. I’m doing NBO in a several systems between one cation and other neutral molecule, so the total charge of the system it would be 1. All the input files are like this:
    # m052x/6-311++g** scf=tight scrf(pcm,read) density=current integral=ultrafine pop=nboread


    1 1
    N -4.263076 0.139549 0.279740
    C -2.794101 0.370468 0.452987
    …. (coordinates)


    $nbo bndidx e2pert $end

    My problem are the results. The NBO calculations ends without apparent problems, but when I see the Total charge of the system, it changes in some of them, is not 1, in particular, in those systems in which tryptophan is implied. I have looked for a solution during days, but maybe I couldn’t see what I did wrong. Would you know where is the problem? Thanks in advance Joaquín!

    PD. Muy buen blog!Gracias por todo

    • Hello Ana

      This is indeed a very weird error! What is the total natural charge you are observing? It is an integer, right? Even if tryptophan is becoming protonated the total electron count should remain unchanged throughout the calculation. Sometimes this kind of errors are due to an extra proton luring somewhere around the rest of your molecule (it doesn’t add any electrons but changes the total charge). If you are using an incorrect Trp fragment in all your calculation you might be dragging an error in all of them.
      I think I’d need to look at the specific log file (and corresponding input file) in order to try giving you an answer.

      Thanks for your kind words about this little blog of mine, I’m glad you find it useful 🙂

      Have a nice day

      • Hello Joaquín,

        Thanks for your answer :). The Total natural charge I observed was different in every system, one example is 0.93827 (between guanidinium cation and tryptophan). I thought at the beggining the same as you, that my molecule could be wrong, with an extra proton or something similar, but I checked it, and also I drew the system again.

        I was looking for possible solutions, and I found one site in which they had the same problem as me (, and they found a solution including: pop=npa IOp(6/76=8), because they have seen this with some other calculations with diffuse basis sets and sometimes tightening the criteria for linear dependence will help. Thus, the new command, which changes it from 10^-6 to 10^-8, will still be stable and yet correct the removal of important contributions.
        Ok, I did that, and the charge total is perfect now, the new problem is that I need the Natural bond orbital analysys too.

        If you want I can send you the input ant .log files where you tell me 🙂

        Thanks again and have a nice day!

      • Hi again Joaquín,

        I forgot to tell you that I solved already the problem 🙂 Anyway, thank you so much for you interest in my problem.

        Best regards,


    • Man Isabela,

      I am also looking for the same option. To change the threshold of E2 in NBO. Have you already found the answer? If so, can you please share it?

      • Unfortunately there is no way to modify it without going into the source code. Sorry

      • Hi, I have found the way to do this the keyword is E2PERT = 0.1 and there is the way I have used in my script:

        $NBO RESONANCE bndidx E2PERT=0.1 $END

        All the best!
        Isabela Man

      • Thanks Isabela. After the input co-ordinates I am using a blank line and then the following line

        $NBO E2PERT=0.1 $END

  35. I know what you mind, but i’m so confuse about that sabil

  36. At this time it sounds like BlogEngine is the preferred blogging platform available
    right now. (from what I’ve read) Is that what you are using on your blog?

  37. Hi sir,

    How to find the type of symmetry from the NBO output of transition metals?

  38. distributedscience

    Dear Dr. Joaquin,

    A very useful blog. Thanks for your contributions, they have really helped. With more of a background in biochemistry & stat mech I’m struggling with getting to grips with NBO analysis of a peptide fragment that I’ve been asked to study using DFT (wb97xd/lanl2dz). Prior to introducing Cu(2+), my structure seems reasonable enough yet when Cu(2+) is introduced at a particular site, the C-C bond prior to the carboxylate (we knocked off the hydrogen from the COOH group due to its absence from our mass spec data) group yields a distance of 1.88 angstrom. The structures have been tightly optimised, and I have tried this out on a different control with the same result.

    My question is how do I go about understanding whether there is a bond, or no bond at all using NBO. I have looked at my NBO output and found the BD and BD* entries:

    1. (0.97584) BD ( 1) C 1 – C 2
    ( 47.03%) 0.6858* C 1 s( 25.63%)p 2.90( 74.37%)
    0.0001 -0.4963 0.0999 0.5583 0.0288
    0.6491 0.0140 0.0981 0.0049
    ( 52.97%) 0.7278* C 2 s( 11.14%)p 7.98( 88.86%)
    -0.0001 -0.3314 0.0399 -0.6464 -0.0426
    -0.6385 -0.0367 -0.2441 -0.0184

    289. (0.15350) BD*( 1) C 1 – C 2
    ( 52.97%) 0.7278* C 1 s( 25.63%)p 2.90( 74.37%)
    0.0001 -0.4963 0.0999 0.5583 0.0288
    0.6491 0.0140 0.0981 0.0049
    ( 47.03%) -0.6858* C 2 s( 11.14%)p 7.98( 88.86%)
    -0.0001 -0.3314 0.0399 -0.6464 -0.0426
    -0.6385 -0.0367 -0.2441 -0.0184

    I am not a pure chemist to any degree, but are their wave functions out of phase, thus removing any potential bond.

    Many thanks

    • Hello Anthony,

      Sorry for the lateness of my response. If you can find a BD between atom1 and atom2 then you have a chemical bond. In the case above, you have a C1 C2 bond with a population of 0.97 electrons which is already interesting since that number should tend to 2. then you get the percentages of each atomic orbital which is only to get the hybridization of orbitals (sp3 sp2 etc.) but I think this is not what interests you. Those bonds with a star in front means they are empty states or virtual as per the regular SCF methodology. Those are empty although they also contribute some to the total electronic structure albeit in a destabilizing manner.

      There are a couple of posts on NBO in this blog that may help you understand the meaning of the output. May I suggest you go to the home site for NBO6.0 in which many wonderful tutorials are archived.

      I hope this helps!

  39. Hi Dr Joaquin,

    I am trying to calculate biradical character for a molecule by using “opt=tight casscf(2,2)/6-31g(d,p) nosymm guess=(mix,save) pop=nbo geom=connectivity”. After getting the .log file, how can I go about doing it?

    Looking forward to your reply.

    Happy holidays!


    • Hi Dexter,

      Sorry for the lateness of my response. I’m not sure what you mean by biradical character, could you please ellaborate? I’m kinda loosing touch with that part of chemistry, sorry. On the other hand if you have already calculated (with such a high accuracy) the structure of your molecule successfuly then may I suggest to perform a Fukui analysis for which you’ll need to repeat your calculation without optimizing but adding/removing an electron and changing the multiplicity accordingly.
      There is a post here on Fukui that might help you.

      Looking forward to know what a biradical is.

  40. Respected sir, I have perform different properties in wfn file like electron density, laplacian of electron density, global maxima and minima, but i didn’t understand physical meaning of global maxima and minima, also its relation with Potential energy surface.
    I am looking forward your response.

    • Maxima on a PES relate to Transition states, whereas Minima relate to stable states or chemically accesible states that you may actually isolate. These are the structures that make sense to us chemists.
      I hope this helps!

  41. Hello sir….I am doing my thesis on the aqua complexes of transition metals..while observing the second order stabilization energy, i need to have a look at the case in which charge has been transferred from the oxygen to the metal ion [ in the paper i have been provided with, the calculations are written for the charge transfer from the oxygen atoms to n(star) orbitals]. Now I am confused….what are these n(star) orbitals?? How should I carry my calculations?

  42. S. N. Balasurahmanyam

    This is not a comment but a matter that needs clearing up for me who a somewhat old-fashioned chemical understanding. Is it valid to think in terms of Pi- and Sigma-ovelap when considering the Wiberg bond index? One or the other is stronger and similar quenrtions are invoved here.
    Can the calculated IR frequency of a given bond DECREASE with higher Wiberg bond index calculated for it?

  43. Dear Prof. Barroso

    I want to estimate the metal–ligand back-bonding contribution in Pt(II) complexes (es: PtL3PMe3), through nbo analysis. How can I do??

    Now I am using second order perturbation theory (E(2) stabilization energy) to do this, but I must consider only the contribution of higher energy or should I do the sum of all the energy contributions?

    Best regards

  44. Hello Joaquin,

    I have some difficulties to get the natural localized molecular orbital NLMO of ruthenium complexes and I would like you to help me handle that issue.
    when I perform the calculation by entering the keywords like $NBO NLMO $END or $NBO PLOT $END or $NBO DIPOLE $END with POP=NBOread for any complex, I get the same message displayed below.


    Highest occupied NBOs are not at the beginning of the NBO list;
    The NLMO program is not currently set up to handle this.

    I don’t understand what it means. However, when I perform the same calculation with the same complex by only replacing ruthenium by iron Fe or osmium Os that share the same periodic group with Ru, the result is successful.
    would you help me understand what is happening?
    My best regards

  45. Juan Sebastián

    I have the same problem: Highest occupied NBOs are not at the beginning of the NBO list

    • This is an issue of NBO3.1 and other versions of NBO programs. Given the fact that NBOs come from the orthonormalization of canonical orbitals some mixing is expected, therefore the program chooses to order your resulting NBOs by energy instead of occupation. This usually is not a problem with any chemical interpretation derived from it.

      I hope this helps

  46. Behnam dehbandi

    What is the cause nbo Star in computing ?I am using the g09 and nbo3.1.
    1. BD (1) N1 – B3 /509. RY*(1) B24 0.93 1.35 0.032
    1. BD (1) N1 – B3 /510. RY*(2) B24 1.09 1.24 0.033
    1. BD (1) N1 – B3 /***. BD*(1) N1 – B24 5.52 1.18 0.072
    1. BD (1) N1 – B3 /***. BD*(1) N1 – H94 1.50 1.12 0.037

  47. Hello Joaquin,

    I am facing a problem, which I mentioned below. Please suggest me to overcome this error.

    The total number of electron is not an integer: 239.64807
    Please check your input data.
    Illegal range for read: IUnit= 7 Address= -256 Length= -256 File length= 17825792.
    Error termination in NtrErr:
    NtrErr called from NtrIO.

  1. Pingback: The Chemical blogspace (Cb) has ranked this blog! « Dr. Joaquin Barroso's Blog

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: