How to Find Open Access Research
Open access is free online availability of scholarly work, usually with reuse rights. The Budapest Open Access Initiative (14 February 2002) still supplies the definition researchers quote. This page is the map of gold, green, hybrid, bronze, diamond, and Subscribe to Open, then the legal search order under rules that are now in force. It is not how to get a library PDF, not how to search databases, and not how to decide if a source is credible.
By CitationEasy Editorial Team21 min read

On this page
Quick Answer: How to Find Open Access Research
Start with a DOI or exact title, then look for a legal copy: the journal's open version of record, an author accepted manuscript in a repository, or a subject archive such as PubMed Central or the NSF Public Access Repository. Install Unpaywall, check DOAJ for fully open journals, and read the journal's self-archiving rule in Open Policy Finder (Jisc; the database formerly called Sherpa Romeo). Do not treat a ResearchGate PDF as proof the file is licensed. Cite the article the same way you cite any journal paper; the access route is not a citation style.
Copy-ready search order (stop at the first legal file)
1. Publisher open copy:
If the journal is in DOAJ, or the article carries a Creative Commons badge on the publisher page, download the version of record and keep the DOI.
2. Unpaywall / oaDOI:
Paste the DOI at https://unpaywall.org/ or use the browser extension. Prefer publishedVersion over acceptedVersion when both appear.
3. Subject or funder repository:
PubMed Central and Europe PMC for biomedicine; NSF-PAR for NSF-funded papers; arXiv for physics, math, and computer science; a field archive only when you can name it.
4. Institutional repository:
Search the author's university repository for the accepted manuscript. Expect different pagination from the published PDF. Check Open Policy Finder before you assume the publisher PDF is the file you may keep or post.
5. Library or interlibrary loan:
If no legal open copy exists, go back to accessing academic journals. Do not pay the publisher Buy PDF button until you have checked the library.
What this page is for
Use this URL when the job is which open-access route applies, and where is a legal copy? How to access academic journals is the library-login and interlibrary-loan sequence after you already have a citation. How to find scholarly sources is where to search. Library databases is Boolean and peer-review filters. Google Scholar is operators and library links. Grey literature is reports and theses that were never journal articles. Keep those jobs off this page.
The official definition of open access
The Budapest Open Access Initiative FAQ reprints the 2002 definition. Two things matter in it: the file is free on the public internet, and reuse is allowed for any lawful purpose if the author is acknowledged. A publisher page that is merely free to read this month is not that definition.
Budapest Open Access Initiative (2002), FAQ reprint
By 'open access' to this literature, we mean its free availability on the public internet, permitting any users to read, download, copy, distribute, print, search, or link to the full texts of these articles, crawl them for indexing, pass them as data to software, or use them for any other lawful purpose, without financial, legal, or technical barriers other than those inseparable from gaining access to the internet itself. The only constraint on reproduction and distribution, and the only role for copyright in this domain, should be to give authors control over the integrity of their work and the right to be properly acknowledged and cited.
That is why librarians separate gratis (free to read) from libre (free to reuse). Creative Commons Attribution 4.0 (CC BY) is the licence Plan S names as the default: you may share and adapt, including commercially, if you give credit. A bronze PDF with no licence is gratis only. Open access in the glossary on this site follows the same split.
The routes students actually meet
Colour names are a map, not a quality ranking. Gold and green are the two BOAI strategies. Hybrid, bronze, diamond, and Subscribe to Open are later labels for cases the two colours alone do not cover. Piwowar and colleagues (2018) treated the categories as exclusive and let the publisher-hosted copy win, so a paper that is both in a gold journal and in a repository counts as gold.
Gold open access
The journal itself posts the version of record on day one, usually under a Creative Commons licence. Many gold journals charge an article processing charge (APC) that a funder or university often pays. You, the reader, do not pay. Listing in DOAJ is the practical check that the journal claims a real open licence and immediate free full text.
Real gold OA article (PeerJ, CC BY 4.0)
APA 7 reference (formatted from the article, not an invented source):
Piwowar, H., Priem, J., Larivière, V., Alperin, J. P., Matthias, L., Norlander, B., Farley, A., West, J., & Haustein, S. (2018). The state of OA: A large-scale analysis of the prevalence and impact of Open Access articles. PeerJ, 6, Article e4375. https://doi.org/10.7717/peerj.4375
In text:
(Piwowar et al., 2018)
Where the file lives:
https://doi.org/10.7717/peerj.4375 and PubMed Central PMC5815332. Unpaywall (queried for this rewrite) returns is_oa true, oa_status gold, publishedVersion, license cc-by.
Green open access
The article appears in a subscription journal. The author (or the university) deposits a copy in an institutional repository or a subject repository. That copy is often the accepted manuscript — same words, different pagination, no publisher typesetting. Older publisher contracts still impose an embargo. Newer funder rules, including Plan S, the 2024 NIH policy (in force for acceptances on or after 1 July 2025), and NSF PAPPG 24-1 Supplement 2 (awards on or after 22 January 2026), require the repository copy immediately. Check Open Policy Finder for the journal's self-archiving policy before you assume the publisher PDF is the file you may post.
What green usually is, and is not
- Usually is. The accepted manuscript in a university repository, PubMed Central, Europe PMC, or NSF-PAR, with a different header from the published PDF.
- Sometimes is. The version of record, if the publisher allows that file in the repository.
- Is not. A typeset PDF an author uploaded to ResearchGate against the contract. Piwowar et al. (2018) excluded academic social-network copies from their OA counts for that reason.
Hybrid open access
One article inside a subscription journal is opened after an APC. The rest of the issue stays paywalled. For a reader, hybrid looks like gold on that one article. For a library, the campus may still be paying a subscription on the same title. Plan S never supported ordinary hybrid APCs. The implementation guidance treated transformative arrangements as a temporary path and said that funding would cease on 31 December 2024. In January 2024 cOAlition S confirmed the sunset. Individual funders may still join a national transformative agreement; that exception is not the default.
cOAlition S, confirmation that transformative-arrangement funding ends after 2024
After careful consideration of the outcomes of transformative arrangements, the leadership of cOAlition S reaffirms that, as a principle, its members will no longer financially support these arrangements after 2024.
Bronze open access
The publisher lets you read the PDF and does not grant an open licence. Access can vanish. You may cite what you read; you may not assume you can republish figures. Piwowar et al. found bronze was the most common free-to-read category in their 2018 samples.
Piwowar, Priem, Larivière et al., PeerJ (2018), abstract (Europe PMC)
Notably, the most common mechanism for OA is not Gold, Green, or Hybrid OA, but rather an under-discussed category we dub Bronze: articles made free-to-read on the publisher website, without an explicit Open license.
Diamond / platinum open access
The journal is free for readers and charges no APC. A society, library, university, or grant pays the platform. Diamond is gold without the author invoice. It is common in some humanities and regional journals and rarer in the large commercial STM portfolios. After the 2024 transformative-arrangement sunset, cOAlition S said it would direct effort toward community-led models, including diamond. Absence of an APC is not a quality badge by itself; still check peer review and DOAJ.
Subscribe to Open
Subscribe to Open (S2O) is a library-subscription flip, not an author invoice. If enough libraries renew, that year's volume opens; if the target is missed, the volume can stay paywalled. DOAJ's Guide to applying now has a dedicated S2O test: the current year's target must already be met, the journal must have published a new issue or at least five OA research articles under S2O, and the site must state the transition year, the status of the backfile, and the reuse rights. Once a year's content is open, DOAJ requires it to stay open. Do not treat last year's open volume as a promise that this year's issue is open.
| Route | Where the file usually lives | Licence | Main failure mode |
|---|---|---|---|
| Gold | Publisher site on day one | Usually CC BY or similar | Predatory journal that charges an APC without review |
| Green | Repository; often the accepted manuscript | Varies; funders now want CC BY | Embargo leftover, or you downloaded the publisher PDF you may not post |
| Hybrid | Publisher site for that one article | Usually CC BY after an APC | The rest of the journal is still subscription; Plan S no longer funds ordinary hybrid |
| Bronze | Publisher site, for now | None stated | The PDF disappears; reuse is not granted |
| Diamond | Publisher site, no APC | Usually an open licence | Small journal with thin indexing |
| Subscribe to Open | Publisher site if that year's target is met | Must be stated; DOAJ requires an open licence | Last year's open volume does not guarantee this year's |
How to find a legal open copy
Do not start by typing the topic into a random PDF search engine. Start from an identifier, then ask a service that tracks legal locations. Unpaywall, PubMed Central, NSF-PAR, DOAJ, and Open Policy Finder are built for that. Sci-Hub and similar shadow libraries are not a legal access route and are not covered here.
Unpaywall and the DOI
Unpaywall is a nonprofit database from OurResearch. The browser extension unlocks a legal copy when you land on a paywall. The Simple Query Tool accepts a DOI. The same data powers many library resolvers and sits under OpenAlex. Prefer the published version when the record offers one. If Unpaywall shows nothing, that is a finding — not a cue to use an illegal copy. Go to the library sequence on access academic journals, which also records how OpenAlex ranks a best_oa_location.
Google Scholar, used as a locator
Scholar is good at surfacing a PDF link. It is not a licence checker. A right-hand PDF may be the version of record, an accepted manuscript, a preprint, or a file someone should not have posted. Open the file, read the first-page banner, and keep the DOI. Turn on library links in Scholar settings so a campus resolver appears beside results; the setup is on the Google Scholar guide.
PubMed Central, Europe PMC, and NSF-PAR
For health and life sciences, search PubMed, then open the Free PMC article link, or search PubMed Central and Europe PMC directly. NIH-funded author accepted manuscripts accepted on or after 1 July 2025 must appear in PMC with no embargo. Older NIH papers may still sit behind the 2008 policy's twelve-month delay. Europe PMC is the deposit target Wellcome names for its own articles. For NSF-funded journal articles and juried conference papers, search NSF-PAR as well. Do not assume every U.S. grant uses PubMed Central.
DOAJ for fully open journals
The Directory of Open Access Journals indexes journals that applied and passed its criteria. It is not a publisher and not a complete list of every free PDF on the web. The Guide to applying states the definition DOAJ uses. A journal that only calls itself open access is not the same claim as a DOAJ listing.
DOAJ, Guide to applying, The type of open access
We define these as journals where the copyright holder of a scholarly work grants usage rights to others using an open license (Creative Commons or equivalent). This allows for immediate free access to the work and permits any user to read, download, copy, distribute, print, search, or link to the full texts of articles, crawl them for indexing, pass them as data to software, or use them for any other lawful purpose.
DOAJ also requires no embargo, no registration wall to read, a public fee schedule (including a statement that there is no fee), and a documented peer-review process. Use those criteria when a journal is not yet listed. DOAB is the sibling directory for open-access books.
Open Policy Finder, then a repository
Open Policy Finder is Jisc's current name for the database most library guides still call Sherpa Romeo. Search the journal to see which version you may deposit, where, and after how long. The same platform absorbed the old SHERPA/Juliet funder records and the OpenDOAR repository directory. A subject repository is often faster than a general search once you know the field. An institutional repository is where green copies and theses actually live.
- PubMed Central / Europe PMC — peer-reviewed biomedical literature and NIH or Wellcome deposits.
- NSF-PAR — NSF-funded journal articles and juried conference papers at https://par.nsf.gov/.
- arXiv — preprints and accepted manuscripts in physics, mathematics, computer science, quantitative biology, and adjacent fields.
- bioRxiv and medRxiv — biology and medical preprints. These are not journal versions.
- SSRN, SocArXiv, PsyArXiv, EdArXiv — working papers and preprints in social science, psychology, and education.
- Zenodo and OSF — general research outputs, including papers and data, with DOIs.
- University DSpace / ePrints / Digital Commons sites — accepted manuscripts and dissertations from that campus.
Failure modes students actually hit
The cases that waste an afternoon
- Bronze mistaken for gold. The PDF opened, so you assumed CC BY. There is no licence line. You may read and cite; you may not drop a figure into your slides under an open-licence theory.
- Accepted manuscript vs version of record. Your quotation's page number does not match the published PDF. Cite the version you used, or switch to the version of record and re-paginate.
- Embargo leftover. A 2019 publisher contract still hides the accepted manuscript for twelve months even though NIH (acceptances on or after 1 July 2025) and NSF (awards on or after 22 January 2026) now forbid that delay for newly covered papers.
- ResearchGate or Academia.edu. The author uploaded the publisher PDF. That is often a contract breach, not green OA. Ask for the accepted manuscript or use Unpaywall.
- Predatory gold. An APC was charged and the paper appeared in days. Check DOAJ, the named editors, and Think. Check. Submit. before you cite it as peer-reviewed.
- Last year's S2O volume. The 2024 volume is open; the 2026 volume missed its target. Check this year's issue, not the journal's reputation for having flipped once.
- Unpaywall miss. No legal copy is indexed. That is when you use the library or interlibrary loan, not a shadow site.
Funder and government rules that removed the embargo
The old student heuristic — NIH papers become free after a year — is no longer the current rule for new acceptances. Public texts now set a default of immediate deposit and no embargo. Check the acceptance date or the award date, not the year you happened to search.
Plan S (cOAlition S)
cOAlition S funders require articles from calls published from 1 January 2021 (or earlier at a member's choice) to be open immediately. The three original compliance routes were an open journal or platform, a repository copy of the version of record or the accepted manuscript, or — as a transitional path that funded APCs only through 31 December 2024 — a transformative arrangement. Hybrid APCs outside those arrangements were never funded. The licence default is CC BY 4.0.
Plan S principle (cOAlition S, Principles)
With effect from 2021*, all scholarly publications on the results from research funded by public or private grants provided by national, regional and international research councils and funding bodies, must be published in Open Access Journals, on Open Access Platforms, or made immediately available through Open Access Repositories without embargo.
2024 NIH Public Access Policy (in force 1 July 2025)
The 2008 NIH policy allowed up to twelve months before PubMed Central release. The 2024 policy, moved forward to 1 July 2025 by NOT-OD-25-101, removes that embargo for manuscripts accepted on or after that date. The overview page, last updated 23 June 2025, states that the 2024 policy is now the one in force.
NIH Public Access Policy Overview (updated 23 June 2025)
the 2024 NIH Public Access Policy requires Author Accepted Manuscripts accepted for publication in a journal, on or after July 1, 2025, to be submitted to PubMed Central upon acceptance for publication, for public availability without embargo upon the Official Date of Publication.
NSF PAPPG 24-1 Supplement 2 (awards on or after 22 January 2026)
NIH is not the only U.S. agency that dropped the twelve-month wait. The NSF notice takes precedence over PAPPG 24-1 for financial assistance awarded on or after 22 January 2026. It removes the 12-month publication embargo from PDF product types and requires the accepted manuscript in NSF-PAR at or before publication. Awards made earlier can still follow the older twelve-month terms. Paying a hybrid APC is allowed on an NSF award but is not required for PAR deposit.
NSF, PAPPG 24-1 Supplement 2 (effective 22 January 2026), AAM deposit
Final versions of manuscripts accepted for publication after peer-review in either scholarly journals or juried conference papers or workshop proceedings (collectively referred to as author's accepted manuscripts or AAMs) must be deposited in NSF's Public Access Repository (PAR) at or before the time of publication.
OSTP Nelson memorandum (25 August 2022)
The White House Office of Science and Technology Policy told U.S. federal agencies to update public-access plans so publications and supporting data from federally funded research are public without an embargo. Agency plans were due for implementation no later than 31 December 2025. NIH and NSF are two implementations with different clocks — acceptance date versus award date. Other agencies have their own repositories. Do not assume every U.S. grant uses PubMed Central, and do not assume an award from 2024 inherited the 2026 NSF rule.
OSTP memorandum, 25 August 2022 (Nelson), recommendation 1
Update their public access policies as soon as possible, and no later than December 31st, 2025, to make publications and their supporting data resulting from federally funded research publicly accessible without an embargo on their free and public release;
Wellcome (page last updated 2 January 2025) already required immediate Europe PMC deposit under CC BY for original research articles from 1 January 2021, with CC BY-ND only by exception. The six-month Wellcome embargo you may still see in older handouts is the previous policy. From 1 January 2025 Wellcome funds APCs only for fully open journals or platforms indexed in DOAJ.
Evaluating an open-access journal before you cite it
Open access is a delivery model. It is not a substitute for peer review. Excellent journals (PLOS Medicine, eLife, many society titles) are open. So are predatory journals that sell an APC and skip review. Judge the journal, not the colour.
Think. Check. Submit., About
Think. Check. Submit. helps researchers identify trusted journals and publishers for their research.
The journals checklist asks whether you or your colleagues know the title, whether the publisher is contactable, whether peer review and fees are stated, and whether the journal is indexed or archived in a dedicated service. Submit — or, as a reader, trust — only if you can answer yes to most items. Think. Check. Submit. does not issue a quality mark and does not publish an approved-journal list.
Checks that actually move the decision
- DOAJ listing. The journal applied and passed. Self-description as open access is not the same thing.
- Named editors who know they are listed. Search an editor plus the journal title. A stolen masthead is a classic predatory tell.
- Fee page dated and complete. Submission, page, colour, and waiver rules should be public before you submit or cite.
- Unrealistic speed. Review promised in days is a warning, not a service perk.
- Indexing claims. Web of Science, Scopus, or Medline can be checked on those sites. A badge from an invented impact-factor company cannot.
- For more on credibility generally. Use How to tell if a source is credible.
Preprints are open. They are not the journal.
A preprint is a manuscript posted before journal peer review. BOAI already included unreviewed preprints that authors choose to share. arXiv, bioRxiv, medRxiv, and PsyArXiv are the usual homes. Cite a preprint when it is the version you used, label it as a preprint, and look for a later journal version. During fast-moving episodes (COVID-19 was the obvious case) the preprint may be the only copy; that is a reason to date-stamp your notes, not a reason to treat comments on the preprint server as peer review.
Real preprint later published as a conference paper
APA 7 preprint reference (arXiv:1706.03762):
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., & Polosukhin, I. (2017). Attention is all you need [Preprint]. arXiv. https://doi.org/10.48550/arXiv.1706.03762
What to do next:
Search the title plus NeurIPS or NIPS 2017 for the peer-reviewed conference version before you treat the arXiv file as the version of record.
How to cite an open-access article or repository copy
Citation styles describe the work, not the paywall. A gold PeerJ article is cited as a journal article with a DOI. A green accepted manuscript is cited as the journal article if you used the published version, or as the repository version if that is what you read. APA 7, MLA 9, and CMOS 18 all prefer a DOI when one exists. Do not invent a retrieval date for a journal article that has a DOI. For the DOI lookup itself, use Find a DOI or How to use DOIs.
Same gold article, three styles (real source: Piwowar et al., 2018)
APA 7:
Piwowar, H., Priem, J., Larivière, V., Alperin, J. P., Matthias, L., Norlander, B., Farley, A., West, J., & Haustein, S. (2018). The state of OA: A large-scale analysis of the prevalence and impact of Open Access articles. PeerJ, 6, Article e4375. https://doi.org/10.7717/peerj.4375
MLA 9:
Piwowar, Heather, et al. "The State of OA: A Large-Scale Analysis of the Prevalence and Impact of Open Access Articles." PeerJ, vol. 6, 2018, article e4375, https://doi.org/10.7717/peerj.4375.
Chicago notes-bibliography:
Heather Piwowar et al., "The State of OA: A Large-Scale Analysis of the Prevalence and Impact of Open Access Articles," PeerJ 6 (2018): e4375, https://doi.org/10.7717/peerj.4375.
If your instructor wants the repository copy identified, add the repository name and handle or DOI after the journal citation, or cite the repository item when no journal version exists (a thesis, a working paper, a dataset). Do not cite Unpaywall, OpenAlex, or Google Scholar as the publisher. They located the file; they did not issue it.
Frequently Asked Questions
- Is open access research lower quality than subscription journals?
- No. Quality tracks the journal's peer review, not the business model. PLOS Medicine, eLife, and many society journals are open and selective. Predatory journals also use gold OA. Check DOAJ and Think. Check. Submit. rather than assuming either direction.
- Can I cite an open-access article in a paper?
- Yes. Cite it as a journal article, book, or preprint — whichever it is. The fact that you did not hit a paywall does not change APA, MLA, or Chicago. Keep the DOI.
- Are preprints acceptable citations?
- Only if your instructor or journal allows them, and only if you label the preprint status. Prefer the later peer-reviewed version when it exists. Vaswani et al. 2017 on arXiv is a real example of a preprint that should be checked against the NeurIPS 2017 conference paper.
- What is the difference between open access and free to read?
- Open access in the Budapest sense includes reuse rights. Free to read (bronze) is a publisher PDF with no open licence and no promise it will stay up. Plan S and DOAJ both require an open licence, not merely a readable file.
- Is a ResearchGate or Academia.edu PDF legal open access?
- Not automatically. Those sites host whatever an author uploaded. Piwowar et al. 2018 excluded academic social-network copies from OA counts. Prefer Unpaywall, PMC, NSF-PAR, or the university repository. If only ResearchGate has the file, email the author for the accepted manuscript.
- How do I spot a predatory open-access journal?
- Use the Think. Check. Submit. journals checklist. Absence from DOAJ, flattering spam, review promised in days, an editorial board that does not know it is listed, and a fake impact factor are the usual cluster. See also the predatory journal glossary entry.
- Do I have to pay an article processing charge to read open-access papers?
- No. An APC is an author-side publishing charge. Readers do not pay it. Diamond journals charge no APC at all. Subscribe to Open is paid by libraries, not by the reader. If a site asks you for money to read an article, you are on a subscription or hybrid paywall, not on a gold reader path.
- Does NIH still allow a twelve-month embargo?
- Not for Author Accepted Manuscripts accepted on or after 1 July 2025. Those must go to PubMed Central with no embargo. Papers accepted earlier can still follow the 2008 policy's delay. Check the acceptance date, not the year you happened to search.
- Does NSF still allow a twelve-month embargo?
- Not for financial assistance awarded on or after 22 January 2026. PAPPG 24-1 Supplement 2 requires the accepted manuscript in NSF-PAR at or before publication. Older awards can still use the previous twelve-month terms. NIH uses the acceptance date; NSF uses the award date. Do not mix the clocks.
- Should I cite the repository PDF or the journal version?
- Cite the version of record when you used it. If you only have the accepted manuscript, cite the journal article and be honest about pagination, or cite the repository item if no journal version exists. Do not cite Unpaywall as the source.
- What if Unpaywall and PMC both find nothing?
- Then there is no indexed legal open copy. Search NSF-PAR if the work is NSF-funded. Otherwise use your library resolver, request interlibrary loan, or email the corresponding author. That sequence is on How to access academic journals. A paywall is not permission to use a pirate site.
Cite the article, not the paywall
Once you have a legal copy, format the citation from the article's DOI and publication details. Open access does not create a new citation style.