How to download research papers for free: legal ways

Most journals still earn their money from subscriptions. A university library pays the publisher every year, and everyone on that campus can read the journal. Anyone else meets a paywall. Open access journals work the other way round: the paper is free for every reader, and the costs are met in other ways.

The useful thing to know is that a single paper often exists in more than one version, and the paywall usually guards only the last one.

Subscription journals usually do not let authors post the published version online. Many do allow them to share the accepted manuscript, sometimes after a waiting period. So when a paper is locked, there is often a legal copy of nearly the same text sitting somewhere else.

Before anything else, Dr. Jha sent Nazia to the university library. "You are not paying for these papers," he said. "The country is."

He meant One Nation One Subscription, or ONOS. The Union Cabinet approved it in November 2024 as a central scheme to give researchers across the country access to international journals, with about Rs 6,000 crore set aside for three calendar years from 2025. Access began on 1 January 2025. The subscription is coordinated by INFLIBNET, the UGC's inter-university centre in Gandhinagar.

• What it covers: about 13,000 e-journals from 30 publishers. The list on the Guru Gobind Singh Indraprastha University library page counts 13,137, including Taylor and Francis, Springer Nature, Elsevier, Wiley, Sage, IEEE and Oxford University Press.
• Who can use it: students, faculty and researchers at higher education institutions run by the central or state governments, and at central government R and D institutions.
• Who cannot, for now: private universities and colleges are not part of Phase I. Reports in September 2026 said the Centre is considering a second phase that would include private universities, but that is still a proposal.

On campus, access usually works through the university's internet address, so the full text simply opens when you are on the campus network. Off campus, it depends on how your institution has set things up; some give users a login with their university email. Ask your librarian how it works at your institution.

Nazia's university was on the list. Seven of her twenty papers opened the moment she logged in from the library computers.

For the papers that were still locked, Dr. Jha showed Nazia the two places to look on Google Scholar. Most students search there already, but few look past the first link.

1. Search the exact title.Put the title in quotation marks so the paper comes up first.
2. Look to the right of the result.A link marked [PDF] or [HTML] points to a free full text, often on a university repository or the author's page.
3. Click "All versions".It sits under the result. It lists every copy Google Scholar has found, and one of them is often free even when the first link is not.
4. Check what you have opened.Compare the title, authors and abstract with the journal's page, and note whether it is a preprint, an accepted manuscript or the final version.

Four more of Nazia's papers turned up this way, three as accepted manuscripts in university repositories and one on the author's own website.

If you are new to searching, our guide on how to search academic databases explains keywords, filters and following citations, which will bring you to better papers before you even worry about access.

Unpaywall is a free browser extension for Chrome and Firefox. When you open a paper's page, it looks for a legal open access copy, one placed online by the author, a repository or an open access journal. If it finds one, a green tab appears at the side of the screen, and one click takes you there. If it finds nothing, the tab stays grey. It does not use pirated copies.

The green tab found free copies of three more of Nazia's papers on her first evening with it. Some places are open by design, and are worth knowing by name:

Shodhganga crossed six lakh theses in May 2025, and the UGC asks universities to deposit theses there once a degree is awarded. For Nazia it was a gift. Three Indian theses on Maithili and Hindi in schools were there in full, each with a long literature review and a reference list she could follow.

Six papers were still locked. "Write to the authors," said Dr. Jha. Nazia was nervous. Would a professor abroad answer a first year scholar from Darbhanga? Many do. Researchers want their work to be read, and a request from someone who is clearly reading it closely is welcome.

Keep the email short and specific:

Dear Professor [surname], I am a first year PhD scholar in Linguistics at a university in Darbhanga, working on how children mix Maithili and Hindi in primary school classrooms. Your paper "[title]" in [journal, year] is central to my literature review, but I cannot access it through my library. Would you be able to share a copy for my research? Thank you for your time. Nazia Parveen

On ResearchGate, the "request full-text" button does much the same thing. The request goes to the authors, and only an author can choose to share a copy; ResearchGate itself never sends one. Whether an author may share it depends on the agreement they signed with the publisher, which is why some send the accepted manuscript rather than the journal PDF.

• find a current email on the author's university page or the paper's first page
• write to the corresponding author first
• send one reminder after a week, not more
• thank them, and cite the paper properly

Five of the six authors replied within a week. One sent a newer paper too, which turned out to be more useful than the one Nazia had asked for.

In her hostel, Nazia's friends told her there was a quicker way: Sci-Hub. Dr. Jha was clear that she should not use it.

Sci-Hub and similar shadow libraries share copies of paywalled papers without the publishers' permission. In 2020, Elsevier, Wiley and the American Chemical Society sued its founder, Alexandra Elbakyan, in the Delhi High Court for copyright infringement. In August 2025, the court ordered Sci-Hub, its mirror sites and a related site called Sci-Net to be blocked in India. The court found that she had broken an undertaking given in December 2020 not to upload new papers by these publishers, and held her prima facie guilty of contempt. That order was an interim one, made while the case continues.

There is also a quieter reason. A research career is built on doing things you can describe openly. For every paper Nazia read, she could say exactly where she got it.

A free copy is not always the final copy. A preprint may have changed a lot during peer review, and page numbers in an accepted manuscript will not match the journal. Before you rely on a paper, check which version you have.

1. Read whichever legal version you can get; for most purposes the accepted manuscript is close to the final text.
2. Cite the published version of record, with the journal, volume and DOI, unless your style guide tells you to cite the preprint itself.
3. Quote carefully from preprints, because the wording may have changed. If you can, check the quote against the final paper.
4. Check that the paper has not been retracted since it was posted. Our guide on what retraction means and how to check shows how.

Then save each paper the moment you find it, with its details. Nazia put every PDF into a reference manager, with a note saying where it came from and which version it was. Our guide on how to use Zotero shows how to set that up, and how to read a research paper shows how to get through twenty of them without drowning.

At the research committee meeting a month later, Nazia had read nineteen of the twenty papers in full and found five more of her own. The twentieth was a conference paper from 1998 that nobody seemed to have a copy of. "Write to the library," said Dr. Jha. "That is what they are there for."