The Other Half of DARPA's Release 6: Three STTR Topics, $5.95M, and a 30 Percent University Requirement Almost Nobody Is Reading

When DARPA pre-released Release 6 of its FY26 solicitation on September 2, 2026, the coverage went where the drama was: seven SBIR topics, five of them from the Biological Technologies Office, four of them describing one continuous battlefield-medicine problem. We wrote that up ourselves — five of seven topics came from one office.

The STTR half of the same release got almost nothing, and it is arguably the more interesting document. Three topics, all from the Defense Sciences Office, all Direct-to-Phase-II with no Phase I option, totaling $5.95 million in base-plus-option authority. Together they describe something DARPA rarely states this plainly: the agency has decided that three of its hardest measurement and materials problems cannot be solved by a small business alone, and it has used the STTR instrument — which mandates a university or research-institution partner — to force the structure it wants.

The structural point most bidders miss

STTR is not "SBIR with a professor attached." The work-split rules are contractual eligibility conditions, and on these three topics they are the design:

• The small business must perform at least 40 percent of the work.
• The research institution partner must perform at least 30 percent.

That leaves 30 percent floating between them, which is where the real negotiation happens. On DPA26TZ06-DV005, the fuel-flexible propulsion topic, DARPA describes the intended division explicitly: the research institution leads gas-species-specific fundamental research — propellant chemistry, ionization behavior, erosion mechanisms — while the company handles productization and integration. That is not a suggestion buried in a compliance annex. It is the topic telling you which half of the problem it thinks is unsolved.

A team that treats the university partner as a 30 percent line item to satisfy a rule will produce a proposal that reads exactly that way. A team that assigns the partner the physics it actually cannot do in-house is responsive to what the topic asks for.

The calendar, and a numbering hazard

Topics pre-released Wednesday, September 2, 2026. They open for submission Wednesday, September 23. Technical questions to DARPA close Wednesday, October 14. Proposals close Wednesday, October 21 through the Defense SBIR/STTR Innovation Portal at dodsbirsttr.mil.

Two documentation problems are worth ten minutes of your attention before they cost you a submission.

First, the date. DARPA's individual STTR opportunity pages currently display an October 23 submission deadline, while the topic write-ups and the DSIP listings show October 21. This is the mirror image of the discrepancy on the SBIR side of the same release. Treat October 21 as operative, verify on DSIP the day you open your cost volume, and build no plan that depends on a two-day cushion that may not exist. DARPA does not accept late submissions and says so in every topic.

Second, the topic numbers. DARPA's SBIR/STTR topic index page lists these three under a DPA26TZ05 prefix; the individual opportunity pages and the solicitation itself use DPA26TZ06 — DV004, DV005, and DV006. If you search on the index-page string you will find nothing, and if you cite the index-page string in a proposal you will look careless. Use DPA26TZ06.

DPA26TZ06-DV004 — SHIELDER

Award: $1,500,000 base over 24 months, plus a $450,000 option over 12 months. Maximum $1,950,000 across 36 months. Direct-to-Phase-II only.

SHIELDER — Scalable Hard-mask materials with Improved Etch resistance and Low Degradation for Extreme-aspect-Ratio fabrication — wants nanofabrication hard-mask materials with substantially better plasma etch resistance than conventional lithography masks, enabling extreme high-aspect-ratio structures with clean pattern transfer, minimal sidewall roughness, and tight dimensional control. The application space DARPA names is broad on purpose: semiconductor, photonic, MEMS, and quantum device platforms.

The numbers are the topic. To bid, you must document current performance at etch selectivity greater than 50:1, aspect ratio greater than 25:1, and sidewall roughness under 5 nm RMS. The Phase II end state is selectivity above 150:1, aspect ratio of 100:1, and sidewall and line-edge roughness under 2 nm RMS.

Read those two sets side by side. DARPA is asking for roughly a 3x improvement in selectivity, a 4x improvement in aspect ratio, and a halving of roughness, in 24 months, from a starting point you must already have reached. That is not a research proposal; it is a scale-up proposal with a materials discovery risk attached. Teams whose selectivity numbers came from a single hero coupon will struggle with the word "scalable," which appears in the acronym for a reason.

Encouraging detail for smaller shops: no topic-level ITAR or EAR restrictions, and a projected CMMC Level 1 requirement — the lowest tier. Compared to most DARPA hardware topics, the compliance overhead here is modest.

DPA26TZ06-DV005 — Fuel-Flexible Spacecraft Electric Propulsion

Award: $1,000,000 base over 12 months, plus a $1,000,000 option over 12 months. Maximum $2,000,000 across 24 months. Direct-to-Phase-II. Up to $25,000 in Technical and Business Assistance on top.

DARPA wants one electric propulsion product line that runs on whatever is available: air (20–50 percent oxygen, balance nitrogen), water, water/CO₂ mixtures of the sort produced by hydrocarbon combustion, and nitrogen/hydrogen mixtures from ammonia or hydrazine decomposition — at better than 30 percent electrical efficiency, using as much shared componentry across propellants as possible.

The strategic logic is not subtle. Xenon and krypton supply chains are concentrated and fragile, and a thruster that can consume in-situ or scavenged propellant changes what a spacecraft has to carry and where it can be refueled. The engineering logic is brutal: oxygen-bearing propellants eat cathodes and channel walls, which is precisely why the research institution's share of this topic is the gas-species-specific erosion and chemistry work.

The feasibility gate is the most specific of the three, and it is a genuine filter:

• Anode thruster efficiency exceeding 30 percent on air or water — not on xenon, with a plan to migrate.
• 100+ cumulative hours on a single test article at that efficiency.
• Vacuum thrust stand data from a facility below 3 × 10⁻⁵ Torr, using in-situ calibrated stands.

The facility requirement alone excludes most bidders. Sustaining that pressure while flowing air or water through a thruster demands serious pumping capacity, and "in-situ calibrated" rules out thrust numbers inferred after the fact. The Phase II end state is TRL 6 hardware — thruster head, flow systems, power processing unit — plus 1,000-hour wear test data. In a 24-month program with the option exercised, a 1,000-hour test is roughly six weeks of continuous operation that has to be scheduled, not hoped for.

Compliance here is heavier than SHIELDER: the topic is ITAR-restricted, a CMMC Level 2 self-assessment is required, and foreign national disclosure is mandatory. That last one is where university partnerships get complicated fast. If your intended lab's propulsion group is staffed largely by international graduate students, resolve the disclosure and access questions with your sponsored programs office before you write the teaming agreement, not after selection.

DPA26TZ06-DV006 — Freestream Disturbances in Hypersonic Wind Tunnels

Award: $750,000 base over 12 months, plus a $1,250,000 option over 12 months. Maximum $2,000,000 across 24 months. Direct-to-Phase-II.

This is the quietest topic on the release and the one with the widest downstream consequences.

Every hypersonic ground test carries an asterisk. Wind tunnels generate their own acoustic and flow disturbances, the freestream in the test section is noisier than actual flight, and boundary-layer transition — the single measurement that most drives hypersonic vehicle design — is exquisitely sensitive to that noise. The result is a decades-old credibility gap between tunnel data and flight data that every hypersonics program pays for in margin, in redesign, and occasionally in failed flight tests.

DARPA's ask is to localize, characterize, and model the upstream noise sources, specifically in the places you cannot put a window: the driver, the reservoir, and the nozzle throat and walls. Then tie those measurements, through multi-fidelity simulation, to the freestream disturbance field that contaminates the test section downstream.

Note what this topic is not. It is not a better test-section probe — that problem has been worked for years. It is non-intrusive diagnostics in optically inaccessible, high-temperature, high-pressure hardware, coupled to a modeling chain that propagates what you measure forward. A proposal that is 90 percent sensor and 10 percent simulation misses half the ask; so does the reverse. The natural team shape is a diagnostics company with a computational aerodynamics group at the university — and this is the topic where the 30 percent research institution floor is likely to be the minimum, not the target.

The back-loaded award structure is a tell. A $750,000 base with a $1,250,000 option means DARPA expects year one to establish whether the measurement approach works at all, and is holding the majority of the money against that result. Propose a year-one plan with an honest go/no-go, not a smooth 24-month ramp.

How to decide, in the three weeks you have

The pre-release window exists so you can make a bid/no-bid call before spending proposal money. For Direct-to-Phase-II topics, that call is almost entirely about evidence you either already have or do not.

Ask the disqualifying question first. For SHIELDER: can you document 50:1 selectivity at 25:1 aspect ratio with sub-5 nm roughness, from a process that plausibly scales? For DV005: do you have 100 cumulative hours above 30 percent efficiency on air or water, measured on a calibrated stand below 3 × 10⁻⁵ Torr? For DV006: can you demonstrate a non-intrusive measurement in a driver or throat environment, and do you have a simulation partner who can carry it downstream?

If the answer is no, the answer is no. Direct-to-Phase-II topics do not reward a compelling narrative about how you would get there — the feasibility documentation is a gate, not a scoring dimension.

If the answer is yes, spend the window on three things. Lock the teaming agreement and the work-share percentages in writing, because 40/30 is an eligibility condition your contracts office will need to certify. Resolve foreign national access on DV005 before it becomes a schedule problem. And send every ambiguity to DARPA before technical questions close October 14 — after that date you are writing against whatever the topic text happens to say.

One consolation if the evidence is not there: DARPA pre-releases new topics the first Wednesday of every month, so Release 7 lands in early October. Under the monthly cadence, missing a cycle costs about four weeks, not a year. That makes the disciplined choice — generating the data the gate demands instead of forcing a thin proposal — much cheaper than it used to be. Our SBIR and STTR deadline calendar tracks the cadence across agencies, and matching a real capability to the right topic before you commit is exactly the problem Granted was built to shorten.

Sources: DARPA SBIR/STTR topics · DPA26TZ06-DV004 (SHIELDER) · DPA26TZ06-DV005 · DPA26TZ06-DV006 · DARPA SBIR/STTR program overview · Defense SBIR/STTR Innovation Portal