Generating neat orbitals/surfaces from molden/wfn-files

Question

I usually use Gaussian/GaussView and don't have to bother with other programs as I think that GaussView is capable of producing nice orbital/surface pictures like this one of one of uracil's Kohn-Sham-orbitals.

But when it comes that I have to use other programs than Gaussian - such as Turbomole, ORCA or Molcas - I am confronted with molden- or wfn-files that GaussView cannot handle.

---

Avogadro 1.1.1 (Lib 1.1.1, OB 2.3.2, Qt 4.8.5)

I quite like Avogadro because it can produce as fancy pictures as GaussView but especially for molden files at least on my computer it seems to fail hard. As you can see the orbital is reproduced perfectly from the file but Avogadro somehow cannot manage to show all atoms.

I hope that I'm not the only one around who can't manage this to work properly.

Thanks to Geoff's comments I learned that wfn-files can be imported through the Extensions $\rightarrow$ QTAIM menu which allows AIM analysis.

---

gmolden 4.x or 5.x

I have absolutely no idea which version this is.

What seems to be easier to open molden-files with molden? I guess that's something to try out. But for me molden stops being an alternative since I am unable to arrange my molecules in space as much as I want. Also molden produces only those stacked-layer orbitals that are not really up to date. Molden can produce pretty good orbitals after one gets how it works. Thank you hokru!

I am admittedly not a big fan of it and didn't try to get more out of it. I was always searching for better alternatives. I'm still no big fan but as those things that I want are implemented and I just have been unable to click the right buttons molden is getting an appropriate way to visualize those orbitals.

---

Gabedit 2.4.8

Now Gabedit is fancy ... there are a lot of input- and output-files that are supported - luckily molden files are part of it.

At least I can create orbitals somehow acceptable for me and arrange my molecule in space as much as I want it. There is also the povray-output stuff but I was not very successful with this approach, yet.

Now what stops me from using Gabedit at a regular basis. It is that there is everytime this point-and-click-adventurish menu -> orbitals -> selection -> again orbital selection -> quality settings -> isovalue settings -> wait and look. The settings were wrong? That's bad ... do the most things again and choose better values. Is there a simple way to compare orbitals fast? Not really, because the orbital slide show is a little bit tricky, too.

Did I say something about wfn-files? I think they are not supported.

---

Jmol 14.2.4_2014.08.03

Jmol is another program that can open molden files and I am again unable to create nice orbital images with this program. Even rising the so called "resolution" up to a reasonable degree only produces hexagonal-shaped orbitals.

Wfn-files? Who needs this? I guess no one except me.

---

Multiwfn 3.3.7

Now there is this free fancy little program from china. A big 400 pages manual/tutorial-pdf comes with it. With it one can quickly have a look on orbitals with a possibility to quickly change isovalues. It lacks unfortunately my favored free arrangement in space plus there are no proper lighting settings ... and the orbitals and structures are not really that beautiful.

But(!) it can read molden files and even wfn-files. It is also able to convert them to cube files that I afterwards can read in with GaussView to create what I want. It can also create those cube files for nearly all other surfaces that it can calculate. I like it much for this simple ability.

---

Conclusion

Well ... as I can use Multiwfn to convert the files that GaussView cannot read in by itself, there is always this curious feeling that there has to be a simpler alternative to all this. A program that can produce nice images without all the lackings and misbehaviours.

How do you do this stuff for visualisation of molden-/wfn-files to produce nice images of your orbitals and molecular surfaces?

Answer 1

Often different quantum chemical programs require different analysing tools. Unfortunately there is no universal standard (program), that can do it all. If there were, someone would come along and just invent something new.

---

I personally prefer ChemCraft for producing publication ready images. It is a commercially distributed program. It mainly supports any Gaussian and GAMESS file formats (and Firefly by extension). It can visualize quite a few more formats now. What I especially like about the program is that it can read the files generated by the NBO programs. It makes visualisation of all kinds of localised and canonical orbitals much simpler.
There are some more features of the program I really got attached to, like symmetrizing molecules or building molecules from scratch. It has the big advantage of writing the symmetry adapted data group for GAMESS for custom basis sets, which I have not been able to find anywhere else. But this is certainly not the target of this post.
Unfortunately it is not able to read molden or wfn^[1] files.

The program comes with a large number of visualisation schemes, which can easily be modified to produce a personalised style. Below is a small demonstration.

Most of the times I would use one of the above, depending on whom I have to impress. Often it is far more beneficial to have monochrome images, like the ones below.

A quite recent addition came with styles that allow you to generate simple 2D like structure. It is still possible to overlay these structures with orbital plots. (Unfortunately the transparency is hard-coded and cannot be set explicitly.)

The caveat is obviously, that it cannot read molden or wfn^[1] files, not even Turbomole output.

A feature that I am still missing is the batch production of orbital pictures. You still always have to save them by hand, which can be incredibly tedious for large molecules (and large active spaces). (But I have written a feature request for that.)
Support for the program is quite good and the available help files are even helpful. Most of the features can be understood intuitively, however, as always, it takes time to get used to working with it.
Another caveat is, that the windows version outperforms its little linux brother by a lot. If you already run windows somewhere, that is probably fine for you. If you run linux almost everywhere, then you have to deal with some missing features (or recently it can be ported via wine, making it slow though).

---

I do use molden (Xwindow) quite a lot for intermediate visualisation, since it is one of the fastest interfaces available. It is very robust and has low graphics; you can easily use it whit a rather slow internet connection. It is free for private and non-commercial usage. Since my first days as a computational chemist involved the use of 56k-modem for checkup routines, this was one of the only go to programs. It will always be the first program I install on a work-dedicated computer. One simply does not forget his or her oldest friend.
A very big plus is the very well structured and written postscript files produced by molden. Some of the purists might even like it. Here are a few converted samples, but you can download the source here (via tinyupload.com).

Such files can become extremely handy, when preparing posters (especially when using LaTeX). I have on occasion seen molecular plots generated by molden in publications; I am even guilty of that. However, nowadays I believe there are more elegant methods.

---

Visualising Turbomole output has been, is and will probably always be a painful experience. For some of the deeper features you should probably be an expert in gnuplot, too. For the basics of structure and orbitals, the current recommendation is to use TmoleX, the GUI to turbomole. If it is done by the same people who designed define, then it's will be quite painfully to use. When I was a heavy user of turbomole, this did not exist, so I should probably shut up.
Judging from the current (version 7.0) manual the workflow I used is still somewhat recommended. It is to let the native modules create the binary *.plt files, which are gOpenMol input. Unfortunately, what was once a very aspiring program is nowadays not developed any more. You can still obtain an unsupported copy. The source code is available, but the current license state of the program is somewhat unclear. Since VMD (see Brians post) and TmoleX do now support gOpenMol files, the necessity of learning a deprecated program is probably not useful any more. One of the prime features of this program however was that it has a tcl interface and hence was fully scriptable - apparently something that VMD now also has.
It is well capable of producing nice orbital pictures, you might have seen them on this page already. (I tweaked the Gaussian output a bit, since I do not have Turbomole available.)

Although it supports quite a few file formats, molden input and wfn^[1] is unsupported.

---

^[1] Let's come to the last part: wfn files. However you produce them, they are probably not up to date. The main (and probably only) purpose for those files is the analysis with AIMPAC - a program that is no longer developed. They have one very big disadvantage: they cannot deal with functions higher than of f type. If you are using a triple zeta basis set, you probably have those functions. Another disadvantage is, that they don not include information about possible effective core potentials.
For the purpose of analysing the electron density with the quantum theory of atoms in molecules (QTAIM) the much superior version wfn extended (*.wfnx or *.wfx) should be used. I am currently only aware of Gaussian to create such files correctly. The main program which uses these files is AIMAll. It is also a commercially available program. You can learn more about these files on the specification page there.
The only valuable point of creating molecular orbitals with this program is, if you want to overlay it with the bond paths instead of just sticks. Just like in this example:

Apart from this usage, the process is very tiring and needs many steps to be completely performed. Working the program non-local is very, very slow. And to be perfectly honest, what is the purpose of the above other than impressing a few organic chemists?
Another negative aspect of the program is the very high demand of memory and storage, especially for large molecules.

In conclusion: If you are not doing QTAIM analysis you should probably refrain from *.wfx files for visualisation purposes. For every QC program there usually comes a workable interface/visualisation program. With any program there is the activation barrier of learning how to do things to overcome. In any case, I suggest having a few ready to work with, you never know where you go and you might have to learn a new program any time.
I think the biggest loss of any visualisation tool is the neglect of writing good tutorials or even a basic documentation.

Answer 2

for completeness, gmolden can actually produce high quality pictures like these:

You need to click the button in the top right in density mode to get the opengl rendering. It is also fully rotatable. While i really like molden for its quick response opening structure files, usability is not its strength. Also it is slow to generate the orbitals.

---

Another program is Molekel, which uses openbabel (i think) to read various file formats. It works fine with molden-type files. Much faster than molden to make the orbitals and more flexible (color, resolution, bounding box). I never tested wfn files.

Answer 3

VMD is my current go-to for molecular visualization. As with any software, it takes some getting used to, but the wealth of options, what seems to be a robust feature set, and generally flexible configurability make it the king of visualizers in my book.

I've only ever tried it with ORCA output, and I've only been able to make it work with "gOpenMol binary" files as generated by the orca_plot utility (ORCA's Molden files appear not to be completely read), but I'm quite satisfied with the results. The steps I typically follow are:

1. Run the ORCA computation to get the .xyz and .gbw files
2. Run orca_plot basename.gbw -i to generate the gOpenMol .plt binary file of the desired function (atomic/molecular orbital, electron/spin density, etc.)

3. Run VMD and load the .xyz and .plt files

   • File > New Molecule to reach the 'open' dialog
   • Browse... to the desired .xyz file; Determine file type should automatically detect the XYZ format
   • Load to pull the geometry into VMD
   • Change 'Load files for' to New Molecule
   • Browse...' to the desired '.plt' file;Determine file type` should automatically detect the 'gOpenmol plt'
   • Load to pull the binary information into VMD
   • Close the 'open' dialog

4. Change the molecule display mode

   • Click Graphics > Representations
   • Under 'Selected Molecule' choose the .xyz file
   • Under 'Drawing Method' choose the desired method (I usually use 'licorice')
   • Run up the Sphere Resolution and Bond Resolution, and tweak the Bond Radius as desired

5. Apply the desired formatting to the isosurface(s)

   • Again in the Graphical Representations dialog (don't have to close after setting the geometry drawing method), select the .plt from the Selected Molecule dropdown
   • If plotting a quantity with negative regions (orbital wavefunctions, possibly spin density, etc.), click 'Create Rep' once, to get a second 'representation'.
   • Change 'Coloring Method' to ColorID, and the number in the dropdown that appears to whatever color desired for one phase (positive or negative)
   • I usually set Material:Transparent (so as to see the molecule under the isosurface) ... Drawing Method:Isosurface ... Draw:Solid Surface ... Show:Isosurface ... Step:1 ... Size:1
   • Set the Isovalue appropriately for the signedness, positive or negative, by typing the number into the box and pressing Enter; I usually start at a magnitude of 0.02
   • Repeat the above for the opposite-signed portion of the plot, if required.

A quick example is below -- the SOMO of gas-phase monoaquo-Cu(II):

The three keyboard shortcuts I use the most are those that change the mouse behavior: R to rotate, S to scale (zoom), and T to translate.

VMD can apparently export to a number of 3-D file formats, as well as PostScript vector graphics. The direct bitmap export is fine if you don't need to zoom in, but pixelates pretty badly if you do. The PostScript is pretty blocky in gswin64, but might do better in, e.g., Inkscape. I don't have IS installed at the moment, so I can't tell if a better renderer does a better job of things.

One further nifty feature of VMD is that it can render in stereo. Even if you don't have 3-D glasses, you can still look at things Magic Eye style by clicking Display > Stereo > SideBySide. (Turn Display > Stereo Eye Swap on if you need to cross your eyes to see it, instead of looking into the distance.)

You would probably have to fiddle around with the different file types generable by each software package to find the best one to interface with VMD. I don't have a copy of either Molcas or Turbomole to test them out, unfortunately. VMD does support Gaussian CUBE files, so worst-case you can dump a (possibly huge) grid and plot it that way.

Answer 4

I use Avogadro obviously.

In Avogadro v1.1.x you can not only use wfn files, but perform QTAIM analysis. Open it under the QTAIM menu.

Send me the Molden file if you can't read all the atoms, that's a weird bug. It works on any Molden file I've tried.

Answer 5

It is possible to use Multiwfn to convert molden files into fchk files that could then be read by GaussView.

The way that satisfied me enough is not the most efficient, I guess, but as it is scriptable, it is acceptable easy and gives nice images.

The way to go: QC program > molden file > Multiwfn > cube file > UCSF Chimera.

1. Generate your molden file

2. Use Multiwfn to create the MO cubes:

   200 Other functions (Part2)
   3 Generate cube file for multiple orbital wavefunctions
   29,30
   3 High quality grid
   1 Output the grid data of these orbitals as separate cube files
   

   This will give you orb000029.cub to orb000030.cub to work with.

3. Generate a python session file with Chimera by opening an xyz file of your molecule, orienting it as you want it and File > Save Session As.

4. Write a program to generate something like the following input, which can be fed into Chimera.

   open session.py
   preset apply pub 2; color byelement
   copy file geometry.jpeg supersample 4 dpi 100 width 1450 height 1068
   open orb000030.cub
   preset apply pub 2; color byelement
   volume #1 level -0.04 color 0,.5,.6 level 0.04 color .9,.7,.1 step 1
   copy file orb000030.jpeg supersample 4 dpi 100 width 2175 height 1602
   close #1
   …
   stop
   

   This input would be called with chimera input.cmd which at this point only shows the geometry of the molecule because it was saved in the session file. My script then saves an image of this geometry and afterwards opens all cube files to produce nice images of them.