R&D tax credits for semiconductors and photonics

Semiconductor and photonics development is R&D-intensive by its nature: device physics, process engineering and packaging each carry uncertainty that published knowledge cannot resolve, and iteration is the working method. South Wales hosts a recognised compound semiconductor cluster, and we work on its doorstep, so these claims are familiar territory for us.

What counts as qualifying R&D in semiconductors and photonics?

Work seeking an advance in the field through uncertainty a competent professional could not readily resolve. An advance is an advance in overall knowledge or capability in a field, not in what one company happens to know (paragraph 6). At concept level, that regularly includes:

• Device development: new structures, materials systems and geometries whose performance cannot be predicted from published results.
• Process development: epitaxy, deposition, etch and lithography work where yield, uniformity or repeatability at specification is genuinely uncertain.
• Packaging and integration: thermal, optical and electrical behaviour of novel assemblies, including photonic integration.
• Test and characterisation methods, where existing techniques cannot measure what the device work requires.
• Design and modelling, where the design objective cannot be met without resolving technological uncertainty; design not contributing directly to that is not R&D (paragraph 41).

Routine yield improvement within an established process window is the honest boundary: optimisation using known methods does not qualify, while work pushing past what the process can demonstrably do often does.

Where does the uncertainty sit?

In two places, both addressed by the Guidelines. The first is knowledge held privately: process recipes, epitaxy conditions and packaging methods are among industry’s most closely held trade secrets, and published literature stops well short of production practice. Where an advance has been made but the details are not readily available, for example because they are a trade secret, work to achieve it can still be an advance in science or technology (paragraph 11). The test is what is readily deducible from publicly available knowledge by a competent professional (paragraph 20), not what your company has done before: work that simply brings a company into line with overall knowledge or capability is not an appreciable improvement, however new it is to that company (paragraph 24).

The second is integration, which meets the objection that every component was bought in. System uncertainty results from the complexity of a system rather than from uncertainty about how its components behave (paragraph 29), and uncertainty exists where a competent professional cannot readily deduce how the separate components should be combined to have the intended function (paragraph 30). A hybrid assembly whose coupling losses, drift and crosstalk cannot be predicted from the datasheets is that pattern; assembly to an established pattern is not.

Where does R&D start and stop on a pilot line?

The design, construction and testing of prototypes generally fall within the scope of R&D; once modifications reflecting the test findings are made and further testing satisfactorily completed, the uncertainty is resolved and further work is not R&D (paragraph 39). Construction and operation of a pilot plant is R&D while its operations are being assessed, until the uncertainty associated with the intended advance is resolved (paragraph 40) — the paragraph that governs a pilot line as much as a chemical plant. Test structures, split lots and shuttle runs sit inside the boundary while the question they were built to answer is open; qualification runs against a settled process sit outside it.

What does not qualify?

Paragraph 14 draws the line: improvements, optimisations and fine-tuning which do not materially affect the underlying science or technology do not constitute work to resolve scientific or technological uncertainty. Yield walking inside a qualified window, and tool commissioning to a vendor specification, are skilled and necessary, and they are not R&D.

Which costs carry these claims?

Staff costs dominate: device, process, integration and test engineers, apportioned on a basis a reader can follow. Agency staff enter as externally provided workers at 65% for unconnected providers, and only where the workers are within UK PAYE and Class 1 National Insurance; subcontracted development enters at 65% for unconnected parties. Connected parties are restricted instead to the lower of the payment and the other party’s own relevant expenditure. Wafers, substrates, precursors and process gases consumed in development runs qualify as consumables, except where they are absorbed into something sold in the ordinary course of business; scrap wafers and reclaim sold as waste are unaffected. The power and water the R&D consumes qualify too — apportion cleanroom and tool energy on tool hours or wafer starts, not a round percentage of the site bill. Software, data and cloud costs for simulation, TCAD and EDA can qualify, apportioned where the licences also serve production: see the full cost categories. Fabs, tools and cleanroom build are capital, and capital expenditure does not qualify for R&D tax relief. Capital spending on the R&D itself can attract research and development allowances instead, so the revenue and capital boundary needs setting early.

Which scheme fits?

Fabless and pre-revenue device companies with heavy engineering payrolls are often exactly the ERIS profile. A loss-making SME whose relevant R&D expenditure is at least 30% of its total relevant expenditure can receive up to 26.97p per £1 of qualifying spend, a £26,970 payable credit on £100,000. That denominator is broadly the trading costs in its accounts for the period, not just the R&D ones. Connected companies are counted on both sides of the ratio.

Profitable companies claim the merged scheme, worth £14,700 to £16,200 net per £100,000: £14,700 at the 26.5% marginal rate on augmented profits between £50,000 and £250,000, £15,000 at the 25% main rate, and £16,200 where the 19% small profits rate applies. Loss-makers outside ERIS also receive £16,200, in cash, because the notional tax on their credit is charged at 19% rather than 25%.

Payable credits under both schemes are capped at £20,000 plus 300% of relevant PAYE and National Insurance. A merged-scheme excess carries forward as a credit for the next period; under ERIS the credit does not carry forward, so the claim must be sized to the cap to keep the unsurrendered loss. Exemption from that cap needs two conditions together: relevant intellectual property created or managed by the company’s own employees, and connected-party contractor and worker spend within 15% of qualifying R&D expenditure. IP-rich businesses here frequently meet both, but it is argued on the Additional Information Form, not assumed. Start with the claim value calculator or which scheme applies.

Sector points worth knowing

• Grant funding, including Innovate UK programmes, no longer blocks or reduces relief under the current schemes: see the current grant position.
• Foundry and shuttle-run arrangements raise the contracted-out question: whether the customer intended or contemplated R&D of the sort actually undertaken, which HMRC reads as needing a specific appreciation of what R&D will be done, and therefore the ability to understand and specify it, rather than mere awareness that some R&D will happen.
• Overseas fab runs need checking against the overseas restrictions: for accounting periods beginning on or after 1 April 2024, conditions necessary for the R&D must be absent in the UK, present where the work is done and wholly unreasonable to replicate here; cost and worker availability are expressly excluded.
• Maintaining R&D equipment, and feasibility studies informing a specific R&D activity, are qualifying indirect activities in their own right (paragraph 31).
• Devices built to survive orbit tie this work to space and satellite systems; volume production raises the questions in manufacturing process development.

What evidence does a semiconductor claim need?

The fab already produces the right material, and it beats anything written a year later: run sheets and lot travellers, split-lot and design-of-experiment matrices, metrology and yield maps, characterisation data and failure analysis reports. A yield map that will not converge across a split describes an unresolved process uncertainty better than any narrative. Tie those records to a project boundary with a stated advance and uncertainty, and give the time apportionment a basis: HMRC accepts an estimated proportion of known expenditure where the estimate is arrived at using evidence and reason, with the methodology recorded. Write the Additional Information Form, mandatory since 8 August 2023, from those records rather than from memory.

Talk it through with a chartered adviser

LimestoneGrey is a firm of chartered tax advisers and chartered accountants specialising in R&D tax relief, regulated by ICAEW. Every claim is signed off by a chartered adviser, with enquiry support included as standard and the fee agreed before work starts. Two compliance points: first-time claimants, and companies that have not claimed in the three years ending with the notification deadline, must notify HMRC within six months of the end of the period of account or the claim is invalid, and HMRC checked around one in six claims in 2023-24, its latest published figure — see HMRC R&D enquiries.

If your company is developing devices, processes or photonic systems, we will give you a straight view on eligibility and scheme fit. Call 0330 223 4 223 or send us a message.

Sources

• Guidelines on the meaning of R&D for tax purposes — paragraphs 6, 11, 14, 20, 24, 29, 30, 31, 39, 40 and 41.
• Check what R&D costs you can claim — cost categories, consumables, software and cloud, and the 65% rule.
• CIRD161000 — the intended-or-contemplated test; CIRD151000 — the s1138A(3)(a) overseas conditions; CIRD151100 — the s1138A(3)(b) exclusion of cost and worker availability.
• GfC3: Recommended approach to claims and record keeping (part 5) — HMRC on estimates arrived at using evidence and reason, and on recording the methodology.