Pyrolyse: Thermal Decomposition & Applications
- Pyrolyse is the thermal decomposition of materials under oxygen-limited conditions, with applications ranging from biofuel production to engineered carbon and ceramics.
- It encompasses diverse regimes—torrefaction, slow pyrolysis, and fast pyrolysis—each defined by specific temperatures, heating rates, and feedstock properties that influence yields.
- Advanced diagnostics and computational modeling reveal complex kinetics and structural evolution, enhancing process control in biomass, polymer, and ceramic conversions.
Searching arXiv for recent and foundational papers on pyrolysis to ground the article in the literature. Pyrolyse, more commonly written as pyrolysis in English-language thermochemical literature, is the thermal decomposition of matter under inert or oxygen-limited conditions. In biomass studies it is commonly represented in global form as , while in polymer and carbon-materials research it denotes the transformation of organic precursors into carbon-rich or ceramic products through staged bond scission, volatilization, crosslinking, and structural reorganization (Nyakuma, 2015). The term covers a wide severity range, from torrefaction at $250$–$300\,^\circ\text{C}$ to hydrocarbon shock-tube pyrolysis at $1450$–, and it underpins processes as diverse as biofuel upgrading, charcoal production, glass-like carbon formation, polymer-derived ceramics, carbon black synthesis, and ultrathin pyrolytic carbon growth (Nyakuma et al., 2015, Rezaeian et al., 29 Apr 2026, Clark et al., 15 Mar 2026).
1. Regimes, feedstocks, and thermochemical scope
Pyrolyse is not a single operating condition but a family of thermochemical regimes distinguished by temperature, heating rate, residence time, and precursor class. In lignocellulosic systems, torrefaction is explicitly framed as a mild pyrolysis or devolatilization process, typically around $200$–$300\,^\circ\text{C}$, in which biomass loses moisture and part of its volatile matter while retaining most of its combustible energy in a more carbon-rich solid (Nyakuma et al., 2015). At higher severity, slow pyrolysis of woody residues such as Ginkgo biloba pruning waste at $400$–$600\,^\circ\text{C}$ yields charcoal with mass yield of $27.33$–$250$0, approximate volume shrinkage of $250$1–$250$2, and an increase in calorific value from $250$3 to $250$4 (Paudel et al., 28 Jul 2025). Fast pyrolysis, by contrast, is described for lignocellulose as a high-temperature, fast-heating, oxygen-free route to bio-oils in high yields, and has been modeled at the elementary-reaction level for $250$5-D-xylopyranose as a representative hemicellulose motif (Lupi et al., 2023).
Polymer pyrolyse defines a second major domain. Glassy carbon is described as a graphene-rich form of elemental carbon obtained from pyrolysis of polymers; in one in situ transmission electron microscopy study, pyrolysis encompassed gradual heating of SU-8 above $250$6 under inert atmosphere, followed by cooling to room temperature (Sharma et al., 2018). A distinct polymer route converts preceramic coatings into SiOC ceramics: material-extrusion 3-D printed plastics were soaked in a polysiloxane precursor and then pyrolyzed in nitrogen at $250$7, leaving hollow ceramic skins thinner than $250$8 (Kulkarni et al., 2019). In hydrocarbon chemistry, shock-tube pyrolyse extends to millisecond, high-temperature gas-phase conversion, as in toluene pyrolysis at $250$9–$300\,^\circ\text{C}$0 and methane pyrolysis at $300\,^\circ\text{C}$1–$300\,^\circ\text{C}$2 (Rezaeian et al., 29 Apr 2026, Clark et al., 15 Mar 2026).
These examples show that the governing severity window determines whether pyrolyse is best understood as devolatilization, carbonization, soot inception, ceramic conversion, or gas-phase cracking. A plausible implication is that the term denotes a common thermochemical logic rather than a single product class.
2. Kinetic structure and reaction pathways
Pyrolyse kinetics are commonly formulated through the conversion rate law
$300\,^\circ\text{C}$3
with $300\,^\circ\text{C}$4 the degree of conversion, $300\,^\circ\text{C}$5 the frequency factor, and $300\,^\circ\text{C}$6 the activation energy (Nyakuma, 2015). For melon seed husk pyrolysis under nitrogen, thermogravimetric analysis combined with Kissinger and Flynn–Wall–Ozawa methods gave $300\,^\circ\text{C}$7 and $300\,^\circ\text{C}$8 by Kissinger, while the FWO method over $300\,^\circ\text{C}$9 to $1450$0 yielded $1450$1 from $1450$2 to $1450$3, with average $1450$4, average $1450$5, and average $1450$6 (Nyakuma, 2015). The fastest decomposition occurred at $1450$7 and the slowest at $1450$8, indicating conversion-dependent activation barriers rather than a strictly uniform single-step process.
At molecular scale, the early-stage chemistry can be pathway-specific. For $1450$9-D-xylopyranose under fast pyrolysis conditions, the lowest-barrier initial event was ring opening to acyclic D-xylose, with a computed barrier of about 0 at 1; competing ring-contraction channels were much higher, roughly 2–3, and dehydration channels likewise higher than ring opening (Lupi et al., 2023). Downstream of xylose formation, the model identified dehydration, cyclization, C–C bond fission, and isomerization routes leading to products such as furfural, glycolaldehyde, anhydro-D-xylopyranose, and dihydrofuran-3(2H)-one (Lupi et al., 2023).
Polysiloxane pyrolyse illustrates a different kinetic issue: multiple parallel channels can operate even when the precursor is often treated as thermally inert. Temperature-programmed desorption mass spectrometry and DFT analysis indicated that PDMS and its oxide-filled composites generate dimethylsilanone over a broad temperature range, in some cases starting at 4, and that the kinetics are best described by a first-order model with a highly ordered transition state and negative activation entropy, consistent with concerted four-centered elimination rather than radical scission (Kulyk et al., 2016). The study proposed three formation channels—direct elimination from PDMS, gas-phase decomposition of HMCTS, and gas-phase decomposition of TMCDS—thereby rejecting a single-pathway description (Kulyk et al., 2016).
Data-driven kinetics has also entered the field. A Chemical Reaction Neural Network trained on cellulose thermogravimetric data learned a sparse mechanism with 4 species and 6 reactions, reached mean absolute error about 5 overall, about 6 in pyrolysis, and about 7 in oxidation, and translated the learned reactions into classical Arrhenius-form kinetics (Ji et al., 2021). This suggests that pyrolyse kinetics can be inferred autonomously while preserving mechanistic interpretability.
3. Diagnostics and direct observation of pyrolytic transformation
Pyrolyse has been characterized through a broad diagnostic stack that spans TGA/DTG, Raman spectroscopy, FTIR, TPD-MS, TEM, XPS/UPS, laser extinction, and microscopy. The choice of probe determines whether the process is read out through mass loss, bond-specific spectral markers, evolving electronic structure, or morphology.
In situ high-resolution TEM of SU-8 pyrolysis provided direct visualization of microstructural evolution from 8 to 9. The observed carbon was not static: highly mobile graphene fragments separated from larger masses, merged into neighboring fragments, and changed in basal-plane size; one neck-like constriction appeared around $200$0, and another fragment entered the field of view around $200$1 and merged into a larger structure by $200$2 (Sharma et al., 2018). The resulting glassy carbon was interpreted as a heterogeneous assembly of randomly shaped graphene fragments with curvature, stacking, bonding, and voids, rather than a simple ribbon network or a structure composed mostly of closed fullerenes (Sharma et al., 2018).
Photoelectron spectroscopy resolves a different level of transformation. Real-time XPS/UPS of SU-8 3005 pyrolyzed in ultra-high vacuum up to $200$3 showed oxygen concentration falling from about $200$4 to $200$5 between $200$6 and $200$7, matching the nominal chemistry more faithfully by $200$8, then disappearing as a measurable O 1s signal by $200$9 (Astley et al., 18 Jul 2025). The C 1s peak shifted from about $300\,^\circ\text{C}$0 to about $300\,^\circ\text{C}$1, the asymmetry factor rose sharply from about $300\,^\circ\text{C}$2 onward to about $300\,^\circ\text{C}$3 at $300\,^\circ\text{C}$4, and UPS showed a zero binding energy state at $300\,^\circ\text{C}$5, together identifying the transition from semiconducting SU-8 to conducting glass-like carbon (Astley et al., 18 Jul 2025).
Gas-phase hydrocarbon pyrolyse has been probed by combined optical and spectroscopic methods. In toluene shock-tube pyrolysis, in situ extinction at $300\,^\circ\text{C}$6, Raman spectroscopy, FTIR, and TEM identified $300\,^\circ\text{C}$7 as the phase-transition reaction temperature: extinction became measurable, the first clear D and G Raman bands appeared, and poorly defined structures disappeared in TEM (Rezaeian et al., 29 Apr 2026). A second landmark at about $300\,^\circ\text{C}$8 marked the ordering threshold, where primary particle diameter reached a maximum and Raman disorder indicators decreased (Rezaeian et al., 29 Apr 2026). For methane pyrolysis, laser absorption of CH$300\,^\circ\text{C}$9, C$400$0H$400$1, and C$400$2H$400$3, together with multiwavelength extinction and TEM/HRTEM, showed that models can reproduce small-molecule speciation yet still mispartition carbon between particle number and particle size, making simultaneous gas and particle diagnostics necessary (Clark et al., 15 Mar 2026).
These studies correct a common simplification: pyrolyse is not fully captured by endpoint composition or final char yield. Intermediate structures, transient radicals, and evolving optical or electronic signatures are often the decisive evidence.
4. Products, carbon materials, and functional solids
The products of pyrolyse are application-specific and range from fuels to advanced solids. In biomass energy systems, feedstock quality controls whether pyrolyse is optimized for condensable liquids, charcoal, or biochar. Melon seed husk was reported with carbon $400$4, volatile matter $400$5, ash $400$6, fixed carbon $400$7, and HHV $400$8, supporting its evaluation as a potential solid biofuel feedstock for pyrolysis, biochar production, and thermochemical clean energy systems (Nyakuma, 2015). Oil palm empty fruit bunch briquettes torrefied at $400$9 for 1 h reached HHV $600\,^\circ\text{C}$0, with mass yield decreasing from $600\,^\circ\text{C}$1 to $600\,^\circ\text{C}$2 and energy yield from $600\,^\circ\text{C}$3 to $600\,^\circ\text{C}$4 as temperature increased from $600\,^\circ\text{C}$5 to $600\,^\circ\text{C}$6 (Nyakuma et al., 2015). For urban woody waste, Ginkgo biloba charcoal met first-grade Korean wood charcoal standards at $600\,^\circ\text{C}$7 and $600\,^\circ\text{C}$8, but elevated Zn and some other heavy metals led to recommendations for pretreatment and further research on copyrolysis (Paudel et al., 28 Jul 2025).
Pyrolyse is also a route to engineered carbon. Ultrathin pyrolytic carbon films were grown on a $600\,^\circ\text{C}$9 nickel film on $27.33$0 by methane CVD at $27.33$1, followed by sudden cooling by pulling the quartz tube out of the hot zone within a few seconds (Umair et al., 2013). Ultra-fast cooling was identified as crucial for controlling carbon segregation on nickel and improving uniformity; Raman area mapping over $27.33$2 gave mean in-plane crystal size $27.33$3 with standard deviation $27.33$4 (Umair et al., 2013). In another carbon-forming route, methane shock-tube pyrolysis co-produced H$27.33$5 and carbon black; TEM showed geometric mean primary particle diameter decreasing from $27.33$6 at $27.33$7 to $27.33$8 at $27.33$9, while graphitic nanostructure generally increased (Clark et al., 15 Mar 2026).
Ceramic conversion is a further branch. In the SiOC replica process, FFF-printed polymer cubes with $250$00 infill were soaked for 30 min in a 50:50 by weight acetone/polysiloxane solution containing a Pt catalyst, dried 24 h in air at room temperature, and pyrolyzed in flowing nitrogen at $250$01 for 1 h after a 45 min purge (Kulkarni et al., 2019). All tested filaments extruded at $250$02 produced ceramic skins of less than $250$03, with pyrolysis causing weight loss of $250$04 to $250$05 and volume shrinkage of $250$06 to $250$07 (Kulkarni et al., 2019). In plasma-assisted hydrocarbon cracking, optimized n-hexane pyrolysis in an Ar–H$250$08 reactor produced CFD-predicted C$250$09H$250$10 selectivity above $250$11, compared with about $250$12–$250$13 in conventional cracking benchmarks (Choi et al., 11 Jun 2025).
5. Process control, scale effects, and computational modeling
Because pyrolyse couples transport, reaction, shrinkage, and evolving material properties, its predictive modeling is intrinsically multiscale. In carbon-fiber reinforced plastics converted to C/C-SiC, pyrolysis was modeled as thermally driven reduction of spring rest lengths in a disordered lattice representing the $250$14 ply. Matrix shrinkage, restrained by fibers, generated tensile stresses that first caused distributed microcracking, then segmentation cracks, and finally a saturated state limited by crack interaction and micro-delaminations (Wittel et al., 2015). The main pyrolysis stage was identified experimentally as $250$15–$250$16, where strong matrix shrinkage and large defect generation occur (Wittel et al., 2015).
For biomass particles, a single-grid Eulerian Volume-Of-Fluid model resolved both the surrounding gas and the shrinking porous particle without empirical interface-transfer correlations. The model introduced a closure function $250$17 to partition reaction-induced change between porosity increase and macroscopic shrinkage, demonstrated mass conservation and numerical convergence, and matched experimental mass loss, temperature histories, volatile trends, and shrinkage profiles, with average steady-state diameter error about $250$18 in one wet-wood sphere case (Caraccio et al., 20 Oct 2025). This indicates that moving-boundary pyrolyse can be simulated without prescribing a fixed particle boundary, although the internal structural evolution of biomass remains a closure problem (Caraccio et al., 20 Oct 2025).
Surrogate and inverse models face a different limitation: transferability. For PMMA, inverse modeling with FDS used TGA, MCC, and cone calorimeter data, multiple effective Arrhenius reactions, and a surrogate fuel mixture to reproduce small-scale mass loss and heat release (Hehnen et al., 2023). Micro- and bench-scale agreement was good, but real-scale parallel-panel predictions diverged among parameter sets. The study’s main conclusion was explicit: small-scale calibration alone is not sufficient to guarantee real-scale transferability (Hehnen et al., 2023). Physics-informed neural surrogates have likewise been proposed for polymer pyrolysis and ablation; a knowledge-driven PINN predicted temperature distributions and degree of burning in 1D and 2D examples, and in one Shen-model benchmark reported temperature prediction error below $250$19 compared with MATLAB numerical results (Ghaderi et al., 2022).
A consistent theme across these models is that pyrolyse cannot be reduced to chemistry alone. Internal stress, porosity evolution, radiative heating, wall losses, mixing, and unresolved morphology can dominate the observed trajectory even when the kinetic submodel is nominally accurate.
6. Classification, limitations, and other technical uses of the term
Feedstock differentiation remains central to pyrolyse practice. A comparative study of 16 aromatic seed residues concluded that no single thermochemical route is optimal for all residues: some compositions are more suitable for pellets or briquettes, others for pyrolysis bio-oil, and others for biochar (Roig-Madrid et al., 12 Dec 2025). High volatile matter and favorable H/C ratios supported bio-oil-oriented conversion, while high ash and fixed carbon pushed materials toward biochar-oriented use; low ash was preferred because minerals catalyze secondary cracking and can reduce bio-oil yield and stability (Roig-Madrid et al., 12 Dec 2025). This classification logic parallels the Ginkgo case, where higher temperature improved fuel quality but also concentrated problematic metals, and the PDMS case, where oxide fillers did not act as inert reinforcements but changed the reaction landscape (Paudel et al., 28 Jul 2025, Kulyk et al., 2016).
Several misconceptions are therefore untenable. Pyrolyse is not merely “heating without oxygen”; it is frequently multi-stage, structurally heterogeneous, and path-dependent. It is not always desirable to maximize temperature, because higher severity can lower mass yield, overcrack valuable intermediates, or amplify contaminants. Nor is the final solid necessarily graphitic in a simple sense: glassy carbon remained non-graphitizing and disordered even as conductivity and short-range ordering increased (Sharma et al., 2018).
In a wholly distinct technical usage, “PYROLYSE” is also the name of an audit tool for overlapping IPv4, IPv6, and TCP reassembly policies: “Protocol bYte stReam OverLapping ambiguitY reaSsembly tEsting” (Aubard et al., 1 Aug 2025). That tool is unrelated to thermochemical conversion, but its existence is relevant because the uppercase form can denote a network-security framework rather than a heating process. It exhaustively tested overlap cases, observed from 14 to 20 different behaviors out of 23 tested implementations depending on protocol, and reported eight errors affecting one OS, two NIDSes, and two embedded stacks (Aubard et al., 1 Aug 2025). In scholarly usage, context therefore determines whether “pyrolyse” refers to thermal decomposition or to this specialized reassembly-testing system.
