I noticed that a lot of articles use DFT+U methods to calculate different formation energies when using regular DFT and DFT+U methods. For example, in the attached paper[1] on $\ce{In2O3}$, the enthalpy of formation was found to be $\pu{-9.86 eV}$ and $\pu{-10.41 eV}$ using LDA and LDA+U, respectively.

When calculating the formation energies using DFT+U, I am wondering if the energies of the elemental reference states ($\ce{In}$ and $\ce{O2}$ in the above case) also need to be calculated using DFT+U, or just from regular DFT, since they often do not have band gaps that DFT+U seeks to correct.

  1. Pakpoom Reunchan, Xin Zhou, Sukit Limpijumnong, Anderson Janotti, Chris G. Van de Walle, Vacancy defects in indium oxide: An ab-initio study, Current Applied Physics, Volume 11, Issue 3, Supplement, 2011, Pages S296-S300,DOI: 10.1016/j.cap.2011.03.051

2 Answers 2


As far as my knowledge, they must be calculated using DFT+U. When you calculate the energy of $\ce{In}$ in its bulk state using DFT, it won't lead to the same value using DFT+U.


To calculate formation energies from DFT+U, you must use a self consistent set of parameters. This is easy to visualize since you can imagine the reaction of InODFT -> InODFT+U, which will end up having some non-zero reaction energy. While this example is obviously incorrect, mixing energy differences with DFT+U corrections and non-corrected differences results in a similar picture, but is often less obvious.

There have been some attempts to mix these however and materials project has a small section on a mixing correction they have applied, however I do not know personally if this has been successful or not. Someone more familiar with materials project might know more.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.