Correction to: Nature https://doi.org/10.1038/s41586-019-1901-0 Published online 15 January 2020

In the version of the article initially published, erroneously, a 10-based logarithm instead of a natural logarithm had been applied in the derivation of the activation energy in the band conduction regime (250–295 K). The correctly derived activation energy gives a value of 310 meV instead of 130 meV, which makes the proposed increased ionization energy due to dopant deactivation less likely.

This correction does not affect any of the experimental results nor the conclusions of the paper.