I calculated the bandgap from band structure using DFT. However, the difference between my estimated bandgap and experimentally obtained bandgap is 0.7eV. Can I reported this bandstructure and bandgap value if it shows variation with experimental studies? If no, what corrections can I make during DFT calculations to get correct bandgap value?
Hi Priyanka Roy ,
It depends the exchange correlation functional you have employed for your calculations. The most common LDA and GGA functional causes the bandgap to be underestimated. This is a well known fundamental problem in DFT with LDA/GGA functional ("The Bandgap problem"). If 0.7 eV comes within the range of underestimation expected with your choice of functional, there is no harm in reporting it.
Most of the time, getting an accurate bandgap is essential, especially if your aim is to calculate optical properties. In those situations, the way forward are
All the best!
You can use the hybrid function like HSE-06. Also you can use GW method in your calculation to solve your problem.
Priyanka Roy You should use HSE06-D3 method as this method is quite good for calculating the band structures. Also try to use fine k-mesh grid which will give you more accurate result.
Dear Priyanka Roy ,
It is clear that the DFT underestimates your bandgap. The colleagues gave some solutions to solve this problem.
You can also use the method given the paper: An Empirical, yet Practical Way To Predict the Band Gap in Solids by Using Density Functional Band Structure Calculationsat the link:Article An Empirical, Yet Practical Way to Predict the Band Gap in S...
I myself found a formula to estimate the bandgap you can follow it in the link:Article A new theory for band gap shrinkage
Best wishes
Dear Roy,
I already answer this questions several times, here in RG. Particularly, the last time was about one week ago, for Sruthil. In breif, the DFT is well known to underestimate the bandgap of semiconductors and dielectric materials. The ground state which is found by DFT ab initio method doesn't comply with the optical and electrical gaps, which are found in excited states. The solution is to use, alternatively, a many body perturbation method, such as the GWA method. More details can be found in my answer to Surthil.
In addition to variant ab initio methods (other than DFT), you can of course use other empirical methods, such pseudopotential and tight binding methods or even empirical formular. However, such empirical methods make use of fitting parameters which should be otherwise found by experiment
Abdelhalim abdelnaby Zekry sir and Muhammad El-Saba sir thank you for your help.
Dear Priyanka,
The best way to get a correct bandgap is by performing the GW+BSE calculations. This method gives you the imaginary and real parts of the dielectric function including excitonic effects (electron-hole interaction). The first peak of the absorption spectrum gives you the optical bandgap. This must be compared with those obtained by the experiments.
- Badges
- Science topic