MSSC2026 Summer School on

undefined

The Department of Chemistry and the Thomas Young Centre at Imperial College London and the Theoretical Chemistry Group of the University of Torino , in collaboration with the Computational Materials Science Group of the Science and Technology Facilities Council (STFC), are organising the

MSSC2026 Summer School on

Ab initio Modelling in Solid State Chemistry

The School is designed for Master and Ph.D. students, as well as for post-docs and researchers who have an interest in getting or strengthening a background in:

️ Computational Solid State Chemistry
🧠 Physics
🧱 Materials Science
Surface- and Nano-Science

The week-long School consists of morning lectures and afternoon hands-on tutorial sessions, where the formal framework and functionalities of the CRYSTAL electronic structure package will be explored.

While we strongly encourage in-person participation, we also offer the possibility to attend remotely through streaming of morning lectures and afternoon hands-on tutorials.

Participants will have the opportunity to present their research at a poster session.

You can register here!
Friday 8 May - Deadline for payment of early bird fees.

See the School website for further details.

undefined

Dear Jonas,

What you are seeing in your output is actually consistent with having three zero-frequency acoustic modes. The first entry of the table (modes 1-2) has irrep E, which is a two-dimensional irreducible representation, so it corresponds to two degenerate modes at zero frequency. The second entry (mode 3) has irrep B2, which is non-degenerate and represents the third zero-frequency mode.

This is standard for crystals with symmetry: degeneracies can make the number of printed lines smaller than the actual number of modes.

Hope this helps!

undefined

Hi Jonas,

When computing thermodynamic properties, we should set up a canonical partition function over all vibrational states of the lattice, which includes phonons at all k points within the first Brillouin zone in principle. The partition function, and thus thermodynamic properties, tend to converge as the sampling of phonons at different k points improves.

As an example, consider what happens for the specific heat and vibrational entropy of MgO in the picture below, where lines of increasing thickness correspond to calculations on larger supercells (i.e. to richer k sampling):

As you see, thermodynamic values change and eventually converge!

So, depending on the size of the original cell, phonon dispersion can have a very large effect on computed thermodynamic properties. This is particularly evident for MgO where the original cell is very small.

So, going back to one of your questions: "how large should the supercell be?" You should use the smallest cell that allows you to get converged results. How to determine it? By progressively increasing the size and by checking your results.

The rule of thumb you suggest of using supercells for different systems containing a similar number of atoms is a good guideline for consistent expansions across different systems.

Hope this helps,

undefined

Hi,

To analyze conduction bands I can recommend a couple of options:

Decomposition of selected bands into leading AO contributions. This option allows to characterize the nature of a band at a given k point in terms of the main AOs contributing to it. From the PROPERTIES module, refer to the ANBD keyword, see page 307 of the User's Manual.
3D plotting of crystalline orbitals. From the PROPERTIES module, refer to the ORBITALS keyword, see page 347 of the User's Manual.

Hope this helps,

undefined

Hi,

Indeed, columns of the TENS_RAMAN.DAT file correspond to xx, ..., zz components of the polarizability tensor $ \alpha $. However, rows correspond to the 3N atomic Cartesian displacements. Indeed, the Raman tensor reported in this file corresponds to the following quantity:
$$
R_{ac,ij} = \frac{\partial^3 E}{\partial u_{a,c} \partial \varepsilon_i \partial \varepsilon_j} = \frac{\partial \alpha_{ij}}{\partial u_{a,c} }
$$
where $ \varepsilon_i $ is a Cartesian component of the electric field, and $ u_{a,c} $ is an atomic displacement of atom $ a $ along the $ c$-th Cartesian direction. The values are reported in atomic units.

Hope this helps,

undefined

Hi,

With CRYSTAL (from the PROPERTIES module actually) one can output the Hartree+EN potential in the all-electron case in 2D or 3D grids with POTM and POT3 keywords, respectively. In the latter case, the output is in .cube format. But I am afraid that currently there is no keyword to plot the XC part of the potential on a grid.

undefined

Hi,

You can do this using the PROPERTIES module. The PBAN option [see CRYSTAL23 User's Manual at page 348] allows you to build a density matrix from a user-defined subset of electronic bands. This partial density matrix is used for subsequent calculations. The ECH3 option can then be used to evaluate the associated electron density on a user-defined 3D grid of points, to be stored in .cube format, which can then be plotted in 3D isosurfaces with standard visualizers, such as VESTA.

A template .d3 PROPERTIES input file for this would look something like:

NOSYMADA
NEWK
24 24
1 0
PBAN
1
14
ECH3
80
RANGE
-10 10
END

where in PBAN as an example I have selected just 1 band, number 14 in the list.

Hope this helps,

undefined

It was a great week! Let me share a group picture from the event:

undefined

Hi,

If I am not mistaken, "total atomic spins" are indeed reported in units of the Bohr magneton and are obtained from a Mulliken partitioning of the spin density (i.e. difference between the electron density of spin-up and spin-down electrons).

Cheers,

undefined

Hi Drmajouri2025,
When performing frequency calculations, it is essential that the geometry is fully optimized with the SCF properly converged. If the structure is not at a true stationary point of the potential energy surface, the computed second derivatives (Hessian) can lead to unreliable frequencies and response properties.

Regarding the *******************, these typically means that the value exceeds the field width allocated in the printing format. In other words, the number is too large (or not representable) within the fixed output format, so it is replaced by stars.

However, in your case, since the geometry did not converge properly, the appearance of stars is very likely due to numerical problem rather than just a harmless formatting issue.

I would strongly recommend:

First ensuring tight SCF convergence (you can also increse TOLDEE and/or TOLINTEG parameters).
Fully optimizing the geometry until forces are below the required thresholds.
Verifying that the optimized structure has no imaginary frequencies.
Only then performing the Raman calculation.

Developer

Posts

MSSC2026 Summer School on

Ab initio Modelling in Solid State Chemistry

Member List