At this point, I just stopped citing any article paywalled by elsevier, and only their arxiv counterpart.

If anyone ask, I will argue that I cannot obtain the version published by elsevier and have only read the arxiv version; hence citing the elsevier version would be disingenuous.