Is citing only high-impact articles in my publication ethical?
No, omitting work you know to be relevant simply because its publication venue does not have as high an impact factor is not ethical.
The purpose of citations is to point your reader to the relevant additional literature, give credit to the work that came before you, and position your work in the broader space. Indexing the most prominent venues (and thereby reinforcing their prominence) is not one of their purposes.
If you read the paper, and it materially impacted your work, you should cite it. And you should cite other papers that are relevant. You don't have to cite every thing that has ever been written on the topic (and as Googlebot mentioned, review articles are a great thing to cite), but quality of individual papers, not prestige of their publication venue, should be the deciding criteria.
This is refutable by saying "Interesting work always gets cited and gets popular, irrelevant of the journals impact factor."
I do not believe that refutation, but even if it were true, it would not be true if authors approached citation as you are proposing.
- If your purpose is to cite general references for a topic, it is better to cite a recent review article (if available; as suggested by the OP), as summarizes all recent works on that topic. Not only this makes your manuscript and its references better weighted but also properly guides the readers (random references are not useful).
- If a relevant review article is not available, cite the references which better fit your statement to be referenced. If there is no other preference, it is better to cite references from the leading journals of the field (regardless of the impact factor), which are read by most of the readers.
- If you cite a specific finding, then the reference should be cited regardless of the host journal.
As a side note, for the general references, it is better to cite references you may use in the rest of your manuscript too.
A decision to cite someone is not always, but very often based on the catchy title, on friendship, or on the hopes to get reviewed by authors of the cited manuscript, to do research with them if they are rich, to collaborate, to push one's PhD students to them later, to get invited by them, etc. Citation counts do not have as much to do with the actual scientific paper quality as you think. Having said that, it is not only unethical, it is simply stupid to cite only high-impact papers.