Industries / Food, farming and forestry / Forestry and logging / Timber and sawmills / Lumber treating
Fire retardants
Soaking wood under pressure with salts or resins that make it slower to catch fire and slower to spread flame across its surface, and proving that the effect lasts.
7
Milestones achieved, from 1895 to 2026.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1895
The Navy specifies fireproofed wood
Fire-retardant treatment fills wood with chemicals that slow ignition and the spread of flame. Richard Gardner of Koppers, a wood-treating company, told a 1971 University of Southern California symposium that US Navy documents specified fire-retardant-treated wood as early as 1895 and that New York City wrote it into its building code by 1898. The market stayed small, with one or two treating plants, until the Second World War. By 1963 all US model building codes accepted it, and in 1969 46 plants treated 45 million board feet (about 106,000 m³) of lumber with it, by Gardner's figures. These early treatments were salts such as borates, sulfates and phosphates dissolved in water.
LessonSalts that dissolve in water wash out in the rain, so these treatments protect only wood kept indoors and dry.
Credit
the US Navy; the City of New York
1950
A tunnel test becomes standard
Albert Steiner developed a 25-foot (7.6 m) tunnel furnace in the 1940s, in which burners at one end drive flame along a sample on the tunnel's ceiling, and ASTM adopted it in 1950 as standard E84, by the account of the testing company Intertek. The test scores a flame spread index, on a scale where asbestos cement board is 0 and red oak 100 (Gardner). US codes sort interior finishes into Class A (index 0 to 25), B (26 to 75) and C (76 to 200). Most domestic woods score 90 to 160, so wood needs treatment to reach Class A, by the US Forest Products Laboratory's (FPL) Wood Handbook. To count as fire-retardant-treated wood under the International Building Code, a product must score 25 or less and show no significant progressive burning when the 10-minute test runs 20 minutes longer.
LessonThe Handbook says treatment slows flame on the surface but does not make wood noncombustible or change its fire resistance in a structure.
Credit
Albert Steiner; ASTM
1966
Retardant that stays outdoors
By Gardner's account, the Bel Air fire of 1961 in Los Angeles, where burning wood shakes and shingles blew from roof to roof and more than 500 homes were lost, led Koppers to look for a fire retardant that rain would not wash out. In late 1966 Underwriters Laboratories approved one and let Koppers label treated red cedar shakes and shingles as a Class C roof covering. Unlike the older salts, the chemical goes in as a water-soluble monomer, a small molecule that kiln heat links into an insoluble polymer, so it stays in the wood and draws little moisture from the air.
Credit
Koppers Company; Underwriters Laboratories
1980s
Roof sheathing crumbles in attics
In the mid-to-late 1980s some fire-retardant-treated plywood used as roof sheathing, and some treated roof truss lumber, failed in service, mostly in the eastern United States. Sun-heated roofs, moisture and acidic retardant chemicals broke down the wood's carbohydrates, and the plywood darkened, turned brittle and crumbled, so roofs had to be replaced. Jerrold Winandy of FPL writes that up to 75 million panels were involved and that replacement was first estimated at more than $2 billion, a figure later cut to 25 to 50 percent of that. A ten-year FPL program found that phosphoric acid and monoammonium phosphate caused the largest strength losses at 66 °C, and borax with boric acid the least. It produced two ASTM test methods, D5516 for plywood and D5664 for lumber, and a practice for design adjustments, D6305, which the American Wood Protection Association (AWPA) made mandatory for fire-retardant formulations and treated products.
LessonA treatment can pass a fire test and still weaken wood slowly in heat, so strength over time has to be tested too.
Credit
Jerrold E. Winandy, Susan L. LeVan and the FPL team
2016
Weathering washes out protection
Birgit Östman and Lazaros Tsantaridis of SP Wood Building Technology in Stockholm, now part of RISE, reported up to ten years of tests on fire-retardant-treated wood panels left outdoors near Stockholm, and on panels aged in the laboratory with cycles of water spray and drying. Several commercial products lost most of their improved fire performance during weathering. Products kept it only at high chemical loadings, usually under paint, and the laboratory ageing matched at most about five years outdoors. In Europe's reaction-to-fire classes (Euroclasses), untreated wood is class D, and the treated products reached B or C before weathering. Europe has since set classes for how long fire performance lasts (DRF, durability of reaction to fire), first as a technical specification, CEN/TS 15912, and now as standard EN 16755, whose EXT class covers outdoor use.
LessonA fire class measured on new wood says little about wood that has stood in the rain.
Credit
Birgit Östman and Lazaros Tsantaridis, SP (now RISE)
2025
Britain recalls fire-rated plywood
In February 2024 the UK Office for Product Safety and Standards (OPSS) reported that a birch plywood from Poland, sold as fire-retardant and declared Euroclass B-s1,d0, may not consistently reach class B when tested to EN 13501-1, and the maker withdrew it from the UK market. In June 2025 OPSS published a recall of Sentrin FRI plywood, marketed in the UK as Euroclass B, after local Trading Standards found it had no compliant declaration of performance or evidence of initial type testing.
LessonA fire class printed on a board is only a claim until test reports and factory checks stand behind it.
Credit
OPSS and local Trading Standards officers
2026
A weathered facade fails
In summer 2025 a fire in a four-storey solid-timber apartment building in Upplands-Bro, Sweden, spread up its fire-retardant-treated wooden cladding to several apartments in 24 minutes, and the building was destroyed. Uppsala Fire and Rescue Service found the fire spread much faster than expected, and SBI tests (the Single Burning Item test used for Euroclasses) on cladding from the building gave class D, not the expected B. In a report published by Lund University in September 2026, researchers ran two full-scale SP FIRE 105 facade fire tests on cladding that had weathered about three years on the building, and both failed the limit on fire spread along the facade. The report notes that such cladding is normally approved after laboratory ageing to EN 16755 EXT, questions whether that ageing matches real weather, and could not establish why the material lost its performance.
LessonThree years of real weather undid a treatment that laboratory ageing is meant to vouch for.
Credit
Nils Johansson, Björn Källander, Robert McNamee, Axel Mossberg and Konrad Wilkens Flecknoe-Brown (Lund University, RISE, Bengt Dahlgren Fire Research and WoodHolz); Uppsala Fire and Rescue Service
Watch for
Class B cladding without repainting
Lunawood, a Finnish maker of thermally modified wood, says its fire-retardant ThermoWood cladding, pine or spruce treated with a deep-penetrating retardant, reaches Euroclass B-s1,d0 for flat tongue-and-groove boards (B for limited contribution to fire, s1 for very limited smoke, d0 for no flaming droplets), needs no maintenance repainting, and maintains its fire classification outdoors.
Achieved whena laboratory with no tie to the company tests boards that have weathered unpainted on real facades for several years and finds they still meet class B.
Claimed by
Lunawood, March 2026
Open
01
A public fire-test table
An open table of published fire test results for fire-retardant-treated wood, one row per test: species, product, type of chemical, retention (kilograms of chemical per cubic meter of wood), paint, ageing method or years outdoors, test method (cone calorimeter, SBI or ASTM E84) and results such as peak heat release rate, time to ignition or flame spread index. Add a simple model that predicts whether a product keeps its class after a given ageing, tested on studies it was not fitted to.
The briefCredit · unclaimed02
Weathered cladding, sampled and tested
A survey of fire-retardant-treated wooden cladding already on buildings. With owners' permission, swap out boards of known product and age from at least 20 buildings, have an accredited laboratory measure the retardant left in them and run cone calorimeter tests (ISO 5660), and compare each with the same product new and after EN 16755 EXT laboratory ageing.
The briefCredit · unclaimed
Open 01 · the brief
A public fire-test table
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open table of published fire test results for fire-retardant-treated wood, one row per test: species, product, type of chemical, retention (kilograms of chemical per cubic meter of wood), paint, ageing method or years outdoors, test method (cone calorimeter, SBI or ASTM E84) and results such as peak heat release rate, time to ignition or flame spread index. Add a simple model that predicts whether a product keeps its class after a given ageing, tested on studies it was not fitted to.
Why it is still open
The results sit in papers and reports, a few products at a time. The US National Institute of Standards and Technology (NIST) keeps an open repository of cone calorimeter tests on GitHub; its files cover untreated species and coated and uncoated red cedar, and their names show no fire-retardant-treated series. Searches of Zenodo, Hugging Face and GitHub in November 2026 found no table of treated-wood fire results, and Lund University's 2026 report asks for documentation of long-term performance.
Built with what exists
The NIST cone-db data format as a starting layout; Crossref to find papers; WebPlotDigitizer to read values off published charts; an AI document reader to pull numbers from PDFs, with every number checked by hand (the Note "Seven more flagged" covers an AI reviewer of long documents); Python with pandas and scikit-learn; Zenodo to host the data.
First step
Enter every result in Östman and Tsantaridis's ten-year report and in the 2026 Lund University report into the cone-db layout.
Done looks like
At least 300 test rows from at least 25 sources, each linked to its source and released under an open license, with the model's accuracy at predicting a product's class after ageing reported on held-out studies.
Open 02 · the brief
Weathered cladding, sampled and tested
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A survey of fire-retardant-treated wooden cladding already on buildings. With owners' permission, swap out boards of known product and age from at least 20 buildings, have an accredited laboratory measure the retardant left in them and run cone calorimeter tests (ISO 5660), and compare each with the same product new and after EN 16755 EXT laboratory ageing.
Why it is still open
The 2026 Lund University report tested cladding from one building, and Östman and Tsantaridis's outdoor data come from one test field near Stockholm. Anders Dragsted of DBI, the Danish fire research institute, said in September 2026 that the response should be field studies and documentation of the actual long-term performance of these products, and DBI proposed a Nordic research effort. Searches found no multi-building survey.
Built with what exists
Building permit records and the reference projects that cladding makers publish, to find facades and their ages; a handheld moisture meter; replacement boards from the cladding maker; accredited fire laboratories for the cone calorimeter tests; an accredited chemical laboratory for phosphorus and boron content. All burning is done by the laboratories.
First step
List 30 buildings in Sweden, Finland and Norway with fire-retardant-treated cladding and a known completion year, and ask three owners for permission to replace a few boards.
Done looks like
Results for at least 60 boards from at least 20 buildings aged 1 to 15 years in at least two countries, published with each product's declared class and set against the same products after EN 16755 EXT ageing.