Research Proposals and GRFP-Style Writing

PHYS 4604 · Lecture 6

Andrew J. Steinmetz

2026-10-02

Upcoming

What When Points
Research proposal, GRFP style Oct. 9, 11:59 PM ET 15%
NSF GRFP reference letters Oct. 16, 8:00 PM ET –
NSF GRFP deadline – Physics & Astronomy Oct. 23, 8:00 PM ET –

The goal of today’s lecture: turn the research proposal due Oct. 9 from a topic you like into two pages a reviewer can score.

The two review criteria

Every reviewer scores you on exactly two things.

Intellectual Merit

Potential to advance knowledge.

  • Is the question worth asking?
  • Is the approach sound and feasible?

Broader Impacts

Potential to benefit society.

  • Outreach, teaching, mentoring
  • Broadening participation in STEM
  • Applications beyond your subfield

Broader Impacts is where most physics applications fail. It is not a paragraph you append at the end. Write it into both documents.

(U.S. National Science Foundation 2026b)

A proposal is not a summary of a field

A literature review tells the reader what is known. A proposal names one open question and commits to attacking it.

A topic

I want to study dark matter using galaxy rotation curves.

What would you actually doon a workday?

A question

Can archival rotation curves tell a cored halo from a cuspy one?

Names the data, the sample, and the measurement.

Can’t say what you’d measure/calculate? Then it’s a topic, not a question.

The structure that fits in two pages

Section What it does
Title Names the question, not the subfield
Motivation / Background Why this problem, why now
Research Question or Objective The one thing you commit to
Approach / Methods How you would actually do it
Intellectual Merit Required heading
Broader Impacts Required heading
References Count against the two pages

Use the exact headings Intellectual Merit and Broader Impacts, kept separate, in both statements.

(U.S. National Science Foundation 2026a, 2026b)

Where your two pages go

Section Rough budget
Title and Motivation ~1/2 page
Research Question 2–3 sentences
Approach / Methods ~3/4 page
Intellectual Merit ~1/4 page
Broader Impacts ~1/4 page
References ~1/4 page

Over two pages? Cut in this order:

  1. Background the reviewer already knows
  2. References that only pad
  3. Method details beyond what shows feasibility

My rule of thumb: methods get the most space. Never cut the question or Broader Impacts.

Writing Intellectual Merit

The reviewer’s question: is this worth doing, and can you do it?

  • The gap: what is not known, and a citation showing it is not
  • The approach: why your method can close that gap
  • The scope: feasible for an early graduate student with a mentor
  • The payoff: what the field learns if it works, and if it does not

Feasible beats ambitious. A project that can work in three years scores better than a field-changing idea that cannot.

Writing Broader Impacts

Done beats promised. “I will do outreach” is weak. “I have tutored for three years and will extend that to …” is strong.

Common sources in physics

  • Outreach and public engagement
  • Open software, tools, and data
  • Mentoring and teaching
  • Applications: energy, health, infrastructure

Make it concrete

  • Who benefits?
  • How many, and how often?
  • How would you know it worked?
  • Why are you the one to do it?

Tie it to your research and to what you have already done.

Write for a reviewer outside your subfield

Reviewers know your broad field, not necessarily your subfield. NSF advises: do not be overly technical; explain your rationale.

Do

  • Say why it matters, in plain words
  • Define every acronym
  • Explain why before how

Don’t

  • Use your group’s shorthand
  • Assume they know the key paper
  • Spend lines on derivations

Could a physicist from another subfield explain your question back to you?

(U.S. National Science Foundation 2026a; Broad Research Communication Lab, n.d.)

Make the page skimmable

A reviewer reads many applications. Assume the first pass is a skim.

  • Bold the research question so it can be found in seconds
  • Start each paragraph with its point; details follow
  • Short paragraphs and white space, not a wall of text
  • Use the section headings from the structure slide
  • One figure at most, and only if it replaces a paragraph

If a reviewer read only your headings, your bold text, and your first sentences, would they know what you are proposing?

More in the technical writing guide.

Sentence-level style

Instead of Write
It is hoped that this research could potentially help… This work will test whether…
Measurements will be performed on the samples. I will measure the samples’ resistivity from 4 to 300 K.
A very novel and extremely important approach An approach that resolves X, which Y cannot
A large dataset 12,000 archival spectra
  • First person, active voice. You are the one doing the work.
  • Future tense for the plan; past tense for what you have already done.
  • Numbers and names beat adjectives.

Example: antibiotic resistance

Two pages of a winning NSF GRFP research proposal on a microfluidics system for studying antibiotic resistance, with handwritten-style annotations in the margins. Annotations note that it states Intellectual Merit and Broader Impacts right away, calls out Intellectual Merit with an explicit heading, sets out a knowledge gap, states a hypothesis, uses an information-dense figure, has an explicit Broader Impacts title, and spends only half the document on methods.

Example: viral photosensitivity

Two pages of a winning NSF GRFP research proposal on photocaged proteins for light-controlled cell therapies, with margin annotations. Annotations note that the title and keywords set the niche, the introduction narrows quickly from genetic disease to photochemistry, the hypothesis is stated explicitly, the research plan is detailed but organized, control experiments and pitfalls add credibility, and Intellectual Merit and Broader Impacts are labeled in bold within their paragraphs.

Physics example: neutrino mass with Project 8

Two pages of a winning NSF GRFP Graduate Research Plan proposing CCA+, an upgraded resonant-cavity readout for the Project 8 neutrino mass experiment. It has a title line, then bold headings for Background, Intellectual Merit, and Broader Impacts, a year-by-year plan with bold run-in labels, key sentences in bold, and a short numbered reference list.

Astrophysics example: exomoons

Two pages of a winning NSF GRFP Graduate Research Plan on detecting exomoons around free-floating planets using JWST and Roman infrared lightcurves. Underlined headings for Intellectual Merit, Methods, Future Work, and Broader Impacts; a four-stage timeline with months per stage; a two-panel lightcurve figure with a caption; and a compact run-in reference list.

Common failure modes

  • A topic, not a question. There is nothing for a reviewer to score.
  • Methods that could not fit in a PhD. Ambition without feasibility.
  • Broader Impacts bolted on at the end. One generic paragraph.
  • No references, or references that only pad.
  • Jargon that only your research group would follow.
  • Two statements that do not connect. NSF advises a consistent theme across both.

Each of these is fixable.

(U.S. National Science Foundation 2026a)

What NSF enforces

Rule Requirement
Length 2 pages, including references and figures
Page 8.5 × 11 in., PDF only
Margins 1 in. on all sides, with nothing in them: no header, footer, name, or page number
Font Times New Roman or Computer Modern, ≥ 11 pt; Arial, Courier New, or Palatino Linotype, ≥ 10 pt
Spacing At least single-spaced (about 6 lines per inch)
Exceptions Smaller text only in equations, figures, tables, and captions
Headings Intellectual Merit and Broader Impacts, separate

LaTeX puts a page number in the bottom margin by default. That breaks the margin rule. Turn it off.

(U.S. National Science Foundation 2026a, 2026b)

Setting it up in LaTeX

\documentclass[11pt,letterpaper]{article}
\usepackage[margin=1in]{geometry}
\pagestyle{empty}  % no page numbers

\begin{document}
...
\section*{Intellectual Merit}
...
\section*{Broader Impacts}
...
{\small
\begin{thebibliography}{9}
...
\end{thebibliography}}
\end{document}

Do not cheat the page limit

  • No \linespread below 1
  • No negative \vspace
  • No body text below 11 pt

Reviewers notice a crammed page.

Start from the starter template. New to LaTeX? Learn LaTeX in 30 minutes.

Before you leave today

  1. Write your research question in one sentence.
  2. Trade with a neighbor.
  3. Your neighbor says back, in their own words, what you would measure or calculate.
  4. If they cannot, revise the sentence and try again.

Bring a full draft on Oct. 9. We will workshop it in class, and it is due that night at 11:59 PM ET.

Advice and examples from past applicants

Rules change. Check every example against the current NSF requirements before you copy its format.

References

Broad Research Communication Lab. n.d. “NSF GRFP Research Proposal.” Accessed October 2, 2026. https://mitcommlab.mit.edu/broad/commkit/nsf-research-proposal/.
U.S. National Science Foundation. 2026a. “How to Apply: NSF Graduate Research Fellowship Program.” https://www.nsf.gov/funding/initiatives/grfp/how-to-apply.
U.S. National Science Foundation. 2026b. “NSF 26-526: NSF Graduate Research Fellowship Program (GRFP).” https://www.nsf.gov/funding/opportunities/grfp-nsf-graduate-research-fellowship-program/nsf26-526/solicitation.