If the error is simply a typo, or two, or a wrong sub-result which does not fully invalidate the findings, don't bother: the chances of publishing are almost zero. (E.g., in mathematics, around 80% of results are expected to be wrong in the strict sense of the word, e.g., not checking whether your claim "for all n≥0 ..." really includes the case n=0. I don't know about the applied physics, though.)
Thus, let's now assume you have found a serious error. A typical procedure is submitting a corrigendum. Corrigenda get seldom published, though. Read the submission guidelines of the journal in question, and, if it doesn't help, ask the editor.
Usually, the chances of getting such a corrigendum published oneself are slim unless one collaborates with the original authors ‒ otherwise, they would be asked to evaluate the corrigendum.
An aside: If methods mean methodology (rather than, say, algorithms), my personal approach to refereeing the paper would be checking whether the new authors
justify why the published part is wrong as it is,
improve the wrong part,
and re-do it completely, i.e., improve the methodology and apply it.