Multiwfn official website: http://sobereva.com/multiwfn. Multiwfn forum in Chinese: http://bbs.keinsci.com/wfn
You are not logged in.
Development of Multiwfn 3.8 starts from 2020-Aug, during the last 5 years it was very actively developed, now it is the right time to release its formal version, it has been very stable and no bug is known. Now it can be downloaded from "download" page of Multiwfn website. All new functions, improvements and bug fixes with respect to Multiwfn 3.7 can be found in "Update History" page on the Multiwfn website.
From 2026, new naming convention of Multiwfn will be applied. For example, Multiwfn released on Jun-4, 2026 will be named as Multiwfn 2026.6.4, and binary package of Windows version will be Multiwfn_2026.6.4_bin_Win64.rar. The new convention will make users clearly recognize the date of update. No developement version will be released in the future, there will only be formal versions, fully distinguished by version names (directly corresponding to release date).
Thanks all users for supporting Multiwfn and citing Multiwfn original papers in their publications!
Dear Sid,
Cube files of hole and electron are exported separatedly by the hole-electron analysis module of Multiwfn. You should load both of them into VMD, and make them visible as isosurface style, but with different colors (e.g. green for electron and blue for hole).
Best,
Tian
Thank you for the reply,
If you don’t mind, could you help me understand this in a little more detail? Because I thought that in a charge-transfer state where certain atoms (that received charge) are significantly negatively charged, at least those atoms should need to be treated with diffuse functions.
CT excitation doesn't necessarily lead to heavily negatively charged atoms. However, when it is indeed true, adding diffuse functions for those atoms may improve result.
How can I plot the potential energy curves and dipole moment curves and know the transition of dipole moment of excited states in Gaussian 16 software?
If you perform potential energy surface scan with an excited state calculation method (for example, using "scan" together with "TD(root=x)" keyword), you will be able to obtain energy and transition dipole moment of state x for every point of the scan coordinate. If you also need dipole moment of the excited state, also try to add "pop=always".
You do not need to add diffuse functions for common CT states. Only Rydberg states, and excitation state calculations for negatively charged system, need diffuse functions.
Hello,
This is completely beyond the scope of wavefunction analysis, I don't have intention to implement these features in Multiwfn. There have been many available codes to print Huang-Rhys factor for electronic transition, including Gaussian (please search "HuangRhys" in manual), ESD module of ORCA, Dushin, FCclasses.
Dear Saeed,
Happy new year
This observation implies that the automatically determined bonding doesn't fulfill your expectation. You need to use $CHOOSE keyword to customize bonding relationship, please check NBO manual. All atoms directly or indirectly connected by "BD" will be regarded as a single fragment.
Best,
Tian
If the output file was obtained under Powershell, you need to change the encoding of this file from Unicode (default) to ASCII/ANSI, or run the command like D:\orca611\orca test.inp | out-file test.out -encoding ascii. If it is not the case, please open the ORCA output file via text editor, and check if excited state information has been normally printed.
Example:
# MP2/cc-pVTZ density out=wfx
...
D:\test.wfx
Then you can use the test.wfx as input file of Multiwfn to perform various wavefunction analyses.
If you are using Gaussian >=G09 C.01, then "density" can be omitted, because "density" is the default setting in the case of "out=wfx".
Multiwfn doesn't directly provide a function to calculate EET, and I don't know what is the coefitions (coefficients?) you referred to. PS: Frankly speaking, I am not familiar with EET calculation.
There is an update of Multiwfn in 2025-Dec-7, important for ORCA users:
Now the electronic excitation analyses that rely on orbital wavefunctions and configuration coefficients, such as hole-electron, IFCT, CTS, NTO analysis, etc. can be exactly carried out in combination with TDDFT calculation of ORCA since version 6.1.1. See Section 3.21.A.2 of Multiwfn manual for details, and it is suggested to check blog article "Method of performing hole-electron and relevant analyses via Multiwfn in combination with TDDFT calculation of ORCA" (http://sobereva.com/758).
Note: Previously, TDDFT of ORCA could not be exactly combined with Multiwfn for hole-electron analysis, IFCT, CTS, NTO, etc., because ORCA cannot separately output excitation and deexcitation configuration coefficients. This update of Multiwfn, which, combined with json file of ORCA 6.1.1, completely solves this long-standing problem!
I would like to suggest using CASSCF to perform geometry optimization when some likely reliable DFT functionals gave very different geometries. Note that spin density cannot be evaluated for CASSCF wavefunction, but you can use Multiwfn to calculate odd electron density (OED) based on CASSCF wavefunction to characterize distribution of unpaired electrons, and you may compare it with |spin density| calculated by different functionals.
To anyone else who wants to do something similar, plotting electrostatic potential over the electron density surface:
Take a molden, .wfn, or .fchk file, load, and use
5 Ouput and plot specific property within a spatial region (calc. grid data)
1 Electron density (rho)
11 Select a set of atoms, set extension distance around them and grid spacing2.3 A around a central atom of the NO3 ligand worked well.
Density saved to a cube file as density.cub
Repeated with option 5 -> 12 Total electrostatic potential (ESP), saved as totesp.cub
Loaded density.cube into VMD, and totesp.cub isolated just the NO3 atoms I wanted, then colored the isosurface according to volume of totesp.cub.
Color scale set to RGB to have red be the negatively charged areas and blue be positive.
Right, your steps are correct. Thanks for sharing it with other people.
I don't quite understand your question. But I would like to mention it is possible to obtain volume of a fragment (e.g. a ligand in coordinate) via AIM basin analysis. Please check Section 4.17.1 of Multiwfn manual, as you can see via basin analysis, one can obtain volume of each atom enclosed by 0.001 a.u. electron density isosurface.
I don't understand your question.
Excitation energies are dependent of geometry. For example, the excitation energies calculated at S0 minimum and S1 minimum geometries are different.
This is not necessarily emission wavelength.
It is emission wavelength only if the two conditions are satisfied:
(1) The current geometry is the minimum of potential energy surface of actual emission state.
(2) The calculated 1st excited state indeed corresponds to the actual emission state.
For fluorescene, usually the emission state corresponds to S1.
Please enter main function 0 and carefully check your atom coordinates. If the function on the line is zero everywhere, or the line is outside the spatial range of provided grid data, then the map cannot be generated.
Dear Jonghwan Lee,
I did a test, and my input file is attached: benzene_S1.wfn. I didn't find the same issue on both Windows and Linux versions. Perhaps "ulimit -s unlimited" didn't take effect, please check.
Best,
Tian
You should set "iuserfunc" in settings.ini to -1 (linear interpolation) or -3 (B-spline interpolation), then when plotting line map in main function, select "100 User-defined function". In this case the function to be plotted corresponds to interpolated function from the loaded grid data.
Unfortunately there is no relevant options to tune the text size in GUI of Multiwfn.
However, if the graphical window is too small, you can enlarge "plotwinsize3D" in settings.ini.
The current isovalue is negative, however, electron density cannot be negative. So, try to change isovalue to a positive value, e.g. 0.001. If the problem still exists, please check Section 6.5 of Multiwfn manual and verify if the wavefunction has been properly loaded into Multiwfn.
There are many possible ways to facilitate SCF convergence, I have a blog article to discuss this point: http://sobereva.com/61 (written in Chinese, you may use Google translator)
I cannot clearly answer this question without more information. Please show me any error if you encountered, and show me commands you inputted in Multiwfn since booting up Multiwfn.
scf=xqc should not be used. The sobEDA.sh script cannot properly extract data from output of quadratic convergence (QC) algorithm.
8: It is impossible to calculate free energy at MECI and the fourth point, which are not stationary point on PES.
9: No.
Additional Q9: You can ask program to calculate larger number of roots, which is not directly related to increasing active orbitals/electrons.
(new): With state-specific solvation treatment, you don't have analytical derivatives, the cost of evaluating fully numerical Hessian is quite expensive (usually computationally prohibitively expensive).
Connolly surface is defined for molecular surface, it is not a general surface for real-space functions. You can study isosurfaces of ESP.
About question 1: DFT-D3(BJ) correction is calculated fully based on geometry and irrelevant to electronic structure, therefore it affects geometry optimization process but doesn't directly affect vertical excitation energy. If you believe dispersion interaction notably affect geometry of your system, then adding DFT-D3(BJ) would be safer.
About question 5: The only advantage of SMD compared to the default IEFPCM is that the former explicitly considers non-polar contribution. This contribution doesn't directly affect excitation energy and electronic structure. If you hope the calculated energy of a given state includes full solvation energy (including non-polar contribution), then SMD is preferred over IEFPCM.
About question 8: As your purpose is not to accuractly calculate excitation energies, but reaction energies/barriers, w-tuned LC-wPBE doesn't have ANY advantage compared to many general functionals. Using wB97XD is recommended for both ground state and excited state reaction energies/barriers. PBE0-D3(BJ) or B3LYP-D3(BJ) cannot correctly represent evident CT state (even geometry)
About question 9: There are dedicated MECI optimization algorithms, see e.g.
Chem. Eur. J., 10, 2327 (2004)
Mol. Phys., 104, 1039 (2006)
JCTC, 6, 1538 (2010)
Gaussian can optimize MECI at CASSCF level, ORCA can optimize MECI at TDDFT and SF-TDDFT level.
Relaxed scan can rarely used to locate MECI.
1 DFT-D3(BJ) doesn't directly affect result of TDDFT
2 No
3 Can be
4 Electron excitation may directly cause or induce bond-formation or bond-cleavage. You can perform geometry optimization or calculate reaction path/barrier on excited state PES to verity the anticipation.
5 No. SMD doesn't have advantage in this regard.
6 No. If you geometry essentially has symmetry, even if you use "nosymm", the calculated atomic forces are still symmetric with respect to geometry.
7 It is too cumbersome. Simply use:
(1) wB97XD/def2SVP TD(Nstates=5,Root=2) opt freq SCRF(SMD,solvent=dimethylsulfoxide)
(2) wB97XD/def2TZVP TD(Nstates=5,Root=2) SCRF(SMD,solvent=dimethylsulfoxide) guess=read geom=allcheck
Set "E=" in settings.ini of Shermo to the 2nd excited state energy obtained from step (2), then boot up Shermo and load output file of step (1), you will obtain reasonable free energy of the 2nd excited state at its minimum.
8 CAM-B3LYP frequently overestimates excitation energy of local excitations. However, if your purpose is to explore reaction path over excited states, then CAM-B3LYP doesn't have well-known flaw.
9 If TS-2 is a transition state on excited state PES, you can certainly optimize it. Note that there may be no TS between B* to C, you may need to locate minimum energy conical intersection (MECI) to study the process (if you are not familiar with it, look at materials about computational photochemistry studies)
I didn' find similar problems. If you can send me your full input and output files (including fch), I will test.
No. They use numerical basis set, Multiwfn is impossible to support them. In addition, they are not free-of-charge, I even cannot become their user.