Wood Modification: Thermal, Acetylation and Furfurylation

Why this matters in the sawmill

For a sawmill, wood modification is first of all a raw-material topic. Modification plants do not buy "boards"; they buy a controlled input: a defined species, cross-section, moisture state and sorting. A mill that understands what happens inside a thermal kiln or impregnation vessel can sell into a niche that pays for consistency rather than volume, and avoid lots being rejected because boards behave unevenly during treatment.

This article overviews thermal modification, acetylation, furfurylation, resin impregnation and densification, and what the sawmill must deliver. A note on sources: figures come from open papers, a review, an industry handbook (ThermoWood Handbook, April 2021 edition) and a producer web page, all re-opened on 26 October 2026. Nothing in this article relies on US Forest Products Laboratory documents. Thermal figures are mainly European (Finnish/Nordic handbook, a Norway spruce study, Polish and Turkish studies; the encyclopedia table of oriental spruce and Scots pine strength has no stated country); the acetylation and furfurylation review (Mantanis, 2017) does not state a single geography, so treat those as international commercial figures dated 2017 to 2018.

The basic idea: why modify wood at all

Untreated wood takes up and releases water as humidity changes. That sorption causes swelling and shrinkage, and it keeps the moisture content high enough in wet service for decay fungi to become active. The handbook of the Finnish ThermoWood association puts the critical threshold at 20% moisture content for fungal activity. Modification attacks this at the cell-wall level. Either hydroxyl groups are chemically blocked (acetylation), the cell wall is bulked with a polymer (furfurylation, resin impregnation), water-binding hemicelluloses are partly destroyed (thermal modification), or the wood is compressed (densification, which addresses hardness more than moisture).

None of these routes is free. Each trades something. Thermal modification buys stability and some durability at the cost of strength and toughness. Acetylation gives very high stability and durability, and the review page read does not report a strength penalty (the abstract page does not address strength, so this was not re-verified), but it needs a chemical plant. Furfurylation gives hardness and durability but is reported to make the wood more brittle (this brittleness statement was not visible on the review's abstract page and was not re-verified).

Thermal modification

Process temperatures and steps

Thermal modification (TM) heats wood in a low-oxygen atmosphere, usually steam or an inert gas, to temperatures that are far above normal kiln drying but below the point of charring. A 2019 study on Norway spruce (van Blokland, Olsson, Oscarsson and Adamopoulos) notes that thermally modified wood is commonly produced at about 160 to 240 °C (European). A table of industrial processes in an encyclopedia entry (adapted from Sandberg and co-authors; the entry's own text is on weathering of thermally modified wood) lists several branded routes: ThermoWood (steam, atmospheric pressure, roughly 30 to 70 hours in total, with a modification phase at 185 to 215 °C), Le Bois Perdure (steam, 200 to 230 °C, 12 to 36 hours), Retification (nitrogen or another gas with at most 2% oxygen, 160 to 240 °C, 8 to 24 hours), Plato (a four-stage process using saturated steam and heated air) and OHT (hot vegetable oil in a closed system, 180 to 220 °C, 24 to 36 hours).

The ThermoWood handbook (Finnish, European) describes the sequence most sawmill people meet in practice. Phase one is high-temperature drying: the chamber is brought quickly to 100 °C and then raised gradually, and moisture is driven down to close to zero. Phase two is the actual modification, held at the class temperature with the wood essentially dry. Phase three is cooling and conditioning, where water spray lowers the temperature and steam raises the moisture content back to 4 to 7%. The handbook does not give numeric phase durations because they depend on class, species, starting moisture and dimension. In the spruce study cited above, Thermo-D boards reached a maximum of 212 °C for 3 hours within a total schedule of three days.

Classes, if you want to name them

The ThermoWood handbook defines two classes. Thermo-S, aimed at stability, uses a maximum of 190 °C for Nordic softwoods and radiata pine and 185 °C for hardwoods, each with a tolerance of plus or minus 3 °C. Thermo-D, aimed at durability, uses 212 °C for Nordic softwoods, ayous, frake and ash and 230 °C for radiata pine. The same document sets the minimum thermal modification temperature at 160 °C. Treat these as one European (Finnish) industry scheme, not a general standard; the figures are from the April 2021 edition of the handbook.

Effects on moisture, dimensions, durability, strength and colour

Equilibrium moisture content (EMC) falls markedly. The handbook states that at about 220 °C the EMC is roughly half that of untreated spruce (the statement in the draft that it stays well below 20% across the humidity range was not found in the handbook text read and is removed). A 2015 Turkish study on wild cherry (Aytin and co-authors) treated by the ThermoWood method at 190 and 212 °C for about 1 and 2 hours reported EMC reductions of about 43 to 51% under four conditioning environments for the most intense schedule, and dimensional-change values 51 to 54% lower in tangential and radial directions and 53% lower volumetrically (the page does not state outright that these are reductions, so confirm in the full paper). Treat these as examples for one hardwood, not as universal figures.

Durability rises with temperature. The 2019 spruce paper cites earlier literature indicating that roughly 200 to 220 °C for about 3 hours is generally needed before decay resistance becomes sufficient. The handbook assigns Thermo-D pine, spruce, ash, ayous and frake to durability class 2 (durable) for outdoor weathered use such as cladding, and Thermo-S pine and spruce and Thermo-S and Thermo-D hardwoods to class 3 (moderately durable) for covered outdoor use such as saunas and roofed structures (the handbook wording for Thermo-D hardwoods overlaps with class 2 for ash, ayous and frake, so check the handbook table before quoting). Interior cladding falls in class 4 (little durable, dry indoor use). The handbook does not recommend use in direct contact with soil or water.

Strength is where thermal modification costs most. In a 2019 test of 100 Thermo-D spruce boards against 100 unmodified controls, mean bending strength fell from 45 MPa to 26 MPa (minus 42%), the 5th-percentile (characteristic) bending strength fell from 23.9 to 11.7 MPa (minus 51%), work to maximum load fell by 72%, and the coefficient of variation increased from 26% to 34%. Stiffness (local modulus of elasticity, mean 12.7 GPa in both groups) did not change significantly. Those boards were tested at 4.6% moisture against 12.9% for controls, which makes direct comparison imperfect, and the authors state that today's thermally modified products are not permitted in structural applications. A table in the encyclopedia entry (modulus of rupture; country not stated) gives smaller reductions at lower temperatures: Scots pine at 190 °C, from about 89.5 to 74.2 MPa; Oriental spruce (Picea orientalis) at 190 °C, from about 75.2 to 72.5 MPa. By our own calculation these are about 17% and 4% losses. The handbook adds that impact strength fell about 25% in spruce at 220 °C for 3 hours and splitting strength falls by roughly 30 to 40%. The draft's statement that parallel-to-grain compression is not reduced was not found in the handbook passages read and is treated as unverified. Thermally modified timber is for cladding, decking, joinery, sauna and interior use, not graded structural members.

Density drops. The handbook gives densities of 490 kg/m³ for reference pine against 430 for Thermo-S and 420 for Thermo-D; spruce goes from 460 to 430 and 420; ash (conditioned at 20 °C and 65% relative humidity) from 625 to 560 and 554. Colour changes through the cross-section to a brown that is darker for Thermo-D than for Thermo-S; the claim that surfaces still turn grey outdoors under UV is general knowledge and was not re-checked in the handbook. The surface also becomes more acidic (pH about 4 against 4.5 to 5 for standard timber), so stainless fasteners are recommended (A2 is the most common stainless grade, A4 the most common acid-resistant grade; the handbook table requires at least A2).

What the mill delivers to a thermal modification plant

The handbook states that Nordic softwood is taken from heartwood, lists the species (pine, spruce, radiata pine, birch, aspen, ash, ayous and frake) and names certification systems (PEFC, FSC and OLB); whether certification is mandatory for each plant should be asked of the plant. Typical lengths are 2.7 to 5.7 m for softwoods and 1.8 to 4.2 m for hardwoods. Final dimensions matter: the handbook notes that a 25 by 125 mm nominal board ends up around 24 by 121 mm, roughly 3% smaller, with permitted tolerances of minus 2 to plus 4 mm in width and minus 1 to plus 3 mm in thickness. A Finnish producer's grading page (Thermoholz) for A and AB softwood grades quotes 6 ± 2% final moisture content and ± 0.5 mm width and thickness deviation for both, no knots or defects for A (ThermoRadiata) and knots up to 50 mm for AB (ThermoPine and ThermoSpruce); this is one company's specification, not a standard. Minimum finished thicknesses from the handbook are 19 mm for outdoor wall and ceiling cladding, 26 mm for outdoor patios and similar structures and for floorboards, and 14 mm for indoor wall and ceiling cladding. So the mill should saw to a green dimension that leaves room for planing after treatment and for the shrinkage and tolerance above, and that is a point to agree with the plant in writing.

Acetylation

Acetylation reacts wood with acetic anhydride, replacing hydroxyl groups with acetyl groups. A published review of the scaled-up technologies (Mantanis, 2017, BioResources; abstract page re-read) reports process temperatures of 100 to 120 °C without catalyst, with acetyl weight gain of 16 to 19% (commercial Accoya is described as averaging about 20%). At around 20% loading the fibre saturation point falls below 15%. The review reports swelling and shrinkage reduced by 70 to 75% against untreated wood, durability class 1 under EN 350-2, hardness increased by 15 to 30%, and acetylated pine at above 20% acetyl gain resisting marine borers after 11 years of field exposure. The species named are radiata pine and alder. Output was about 40,000 m³ in 2017, with capacity planned to reach 60,000 m³ in 2018 (the company, Accsys, as reported in the 2017 review; the draft wrongly gave 60,000 m³ as 2018 output). The draft also stated a teak/ipé/merbau comparison and no negative strength effect; neither appears on the abstract page and both are unverified. The review gives no numeric EMC reduction, so none is quoted.

For the sawmill, acetylation means high uniformity requirements and relatively thick, permeable material. Acetic anhydride has to penetrate the whole cross-section, so permeable sapwood-rich softwood is favoured; No maximum thickness or incoming moisture requirement is quoted in the sources opened. Ask the plant for its written intake specification.

Furfurylation

Furfurylation impregnates wood with furfuryl alcohol, a bio-based chemical derived from agricultural residues, and then cures it by heat so that it polymerises inside the cell wall. According to the same review, Kebony Clear reaches a mean weight percent gain (WPG) of about 35% and Kebony Character about 20%. At 30 to 35% WPG decay resistance over nine years of testing was comparable to CCA-treated pine; anti-shrink efficiency is about 60% at roughly 35% WPG; durability class 1 is stated. Species named are radiata pine, southern yellow pine and maple (Clear) and Scots pine (Character). The statements that hardness, modulus of elasticity and modulus of rupture increase while impact resistance does not improve and the material is more brittle were not visible on the abstract page and are unverified. Marine-borer resistance was only clearly achieved at WPG above 50%; results at 30 to 35% after 16 years were not fully satisfactory (the draft's 'slight to moderate damage' wording was not confirmed). Output was about 22,000 m³ in 2017 (Kebony, as reported in 2017). The review page gives no concentrations or curing temperatures.

For sawmills, this implies sapwood-rich pine free of checks; that is our inference, not a statement from the review, so confirm it with the plant.

Resin impregnation and densification in brief

Resin impregnation with synthetic thermosetting resins bulks the cell wall similarly to furfurylation. No source with verified weight gains or properties for these was used, so no numbers are given. The only verified statement comes from a densification study (Han et al. 2022): resin impregnation and thermal post-treatment reduced set recovery and water uptake compared with densified wood alone, and the resin-impregnated group showed a considerable increase in modulus of elasticity.

Densification compresses wood perpendicular to the grain in a hot press. In a 2022 study on Scots pine sapwood (Han, Kutnar, Couceiro and Sandberg; Slovenia and Sweden; European), kiln-dried specimens of about 480 kg/m³ were compressed radially from 20 mm to 10 mm (50%), with platens pre-heated to about 170 °C, raised to about 200 °C for 2 minutes, then cooled to about 60 °C under compression, at 4 MPa and a 3-minute hold at the end stop. Density roughly doubled, modulus of rupture rose by about 64%, stiffness was not significantly changed, and creep compliance was 40 to 55% lower in a 14-day test. The cell-wall density of about 1,500 kg/m³ sets the practical ceiling. The weakness is set recovery: densified wood can spring back when it gets wet and warm, so fixation by heat or resin is essential. The statement in the draft that the authors say densification mainly benefits low-density softwoods such as Norway spruce was not confirmed in the abstract and is unverified. For the mill, the input is straight-grained, clear softwood of uniform thickness.

Which species and products suit which route

In the sources opened: thermal modification is specified for Nordic pine and spruce, radiata pine, ash, birch, aspen, ayous and frake; acetylation is commercial on radiata pine and alder; furfurylation on radiata and southern yellow pine, maple and Scots pine; densification is demonstrated on Scots pine. Clear or nearly clear boards sell best into visible uses (cladding, decking, flooring); knottier boards fit lower cladding grades.

Practical rules of thumb for sawmills

• Ask for the plant's written intake specification (species, heartwood or sapwood, moisture, thickness, length, knot limits) and sort to it; do not infer it from another plant.
• Dry to the moisture window the plant asks for. Thermal plants dry the wood themselves in the first phase (the handbook describes high-temperature drying as phase one, without giving a required starting moisture; ask the plant), but uneven moisture inside a lot creates uneven results.
• Saw oversize. Plan for roughly 3% size loss in thermal modification plus the tolerances in the handbook, and add planing allowance.
• Keep species, heart/sap and grades in separate labelled lots.
• Provide chain-of-custody documents (PEFC or FSC) when the plant requires them.
• Do not sell thermally modified timber, and do not let your customers use it, as structural timber.
• Request feedback on rejects after each run.

What is uncertain or varies

Process parameters differ between plants and between branded processes; the tables in reviews are simplifications. Strength losses depend on species, temperature, duration and test moisture, so single numbers (42% in one spruce test; 10 to 20% quoted as background in an introduction to a Polish thermogravimetric paper and not measured there) should not be transferred. The acetylation and furfurylation figures come from one 2017 review of commercial technologies and are partly manufacturer-reported. EN 350 class assignments were not verified in the standard text, nor the certification schemes for each branded product, nor any ThermoWood class beyond the handbook. No numbers for resin impregnation are given. Volumes cited are for 2017 and 2018 and are not current market figures.

Key points

• Modification trades one property for another: stability and durability against strength (thermal), brittleness (furfurylation) or process cost (acetylation).
• Thermo-S (190 °C) and Thermo-D (212 °C) are ThermoWood classes for Nordic softwoods; thermally modified timber is not structural.
• Thermal modification roughly halves EMC in some species and lowers bending strength by amounts ranging from small to about 40%.
• Acetylation (about 20% acetyl gain) and furfurylation (about 20 to 35% WPG) claim durability class 1 in the reviewed source.
• Densification doubles density in Scots pine but needs fixation against set recovery.
• The sawmill's value is in uniform, specified, certified input: right species, size allowance, moisture and sorting.