A closed‑form reaction‑kinetics model for LET‑ and oxygen‑dependent hydroxyl radical availability under irradiation Authors: Ladan Rezaee Published: 2026 ABSTRACT The biological effectiveness of ionizing radiation depends not only on absorbed dose but also on radiation quality and microenvironmental factors, particularly linear energy transfer (LET) and oxygenation. Although Monte Carlo track-structure simulations can describe radiation–matter interactions in detail, such complexity often obscures the individual roles played by governing parameters. We present here a closed-form, analytical reaction–kinetics model for the indirect chemical stage of radiation action. The model introduces an LET-dependent source term for hydroxyl radical production based on experimentally and computationally established G-value trends and combines it with a balance equation derived from a directional Boltzmann-P1 approximation under spatial averaging. Macroscopic reaction rates incorporate the presence of oxygen and background scavengers, leading to an explicit, steady-state expression for hydroxyl radical density independent of biological response functions. The model systematically replicates suppression of indirect chemical activity with increasing LET and predicts the natural emergence of smoothly varying radiochemical regimes corresponding to production-limited, scavenging-limited, and track-structure-dominated behavior. These regimes emerge naturally from the model structure and the interplay between production, scavenging, and track‑density effects. The framework is not designed to predict biological damage or clinical outcomes but provides a physically transparent and analytically tractable description of the chemical stage of radiation action, one suitable for incorporation into multiscale models of heterogeneous irradiation.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-54958-3. Introduction Radiation-induced biological damage arises from a complex cascade of physical, chemical, and biological processes across a broad range of spatial and temporal scales. The physical stage, initiated by the deposition of energy from ionizing radiation, gives rise to a dense pattern of ionizations and excitations that subsequently evolves into a chemically reactive environment dominated by short-lived radical species. Of these, hydroxyl radicals (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document}) play the central role in mediating indirect Deoxyribonucleic Acid (DNA) damage through diffusion-driven reactions with biomolecular targets. It follows that understanding how radiation quality modulates hydroxyl radical production and loss establishes a mechanistic link between microscopic energy deposition and its downstream biological consequences. One major characteristic of radiation quality is its LET, an important factor in determining ionization track doses. Quite extensively, both experimentally and theoretically, it has been observed that as LET increases, there is an increase in intra-track recombination and radical-radical reactions, thus resulting in a systematic reduction in freely diffusing hydroxyl radicals compared with low LET irradiations1–3. Although this LET-dependent modulation in radiochemical yield is well accepted as an important component in differentiating indirect DNA damage, its impact in the chemical stage due to variations in LET is frequently incorporated implicitly in radiobiological modeling. Oxygen availability represents a second, equally important modulating factor in radiation-induced chemistry. Oxygen fixation is a critical factor in the biological consequence of radiation-induced damage. This occurs through the oxygen fixation hypothesis. In this well-established mechanism, oxygen does not play a direct role in the scavenging of hydroxyl radicals. This is because the reaction of the hydroxyl radical with biomolecules occurs too quickly. However, oxygen does play a direct role in the fixation of the damage. In the oxygen fixation hypothesis, oxygen reacts with biomolecular radicals, producing peroxyl radicals. This makes the damage less repairable, thus increasing the biological effectiveness of the radiation. In the chemical stage, the availability of radicals is affected indirectly through the modulation of the chemical environment. The role of oxygen in radiation responses has been extensively explored in relation to oxygen enhancement ratio effects and radio-resistance due to hypoxia1,4. However, in a number of modeling scenarios, oxygen-related processes are captured in a phenomenological fashion, ideally directly related to biological effects, but not at a radiochemical transport/kinetics level. Monte Carlo track structure simulations, for instance, enable a highly detailed and mechanistic approach to understanding the radiochemical reactions, from the nanometer scale down to individual molecules, through simulations such as PARTRAC, Geant4-DNA, and TOPAS-nBio2,5,6. While these simulations provide a level of physical detail that is unparalleled, their complexity can limit their practicality for use. The work herein serves as a supplement to these simulations, as a simpler analytical model that can be understood. There, thus, exists a definite need to derive simplified models, amenable to analytical analysis, which capture the fundamental physics/chemistries underlying LET/oxygen dependence of radicals. Various analytical or semi-analytical approaches have also been developed to fill the gap: these include diffusion-reaction equations for radical transport or simplified yield functions of radiolysis7,8. It should be noted that previous models generally observed LET or the role of oxygen in a simplified or isolated fashion through curve fitting of the results with biological observations or lacked the explicit transport description enabling the discrimination of production, scavenging, or absorption processes. Therefore, a simple analytical framework, in which the effect of the quality of the radiation, the effect of oxygen, and the effect of the transport of the radicals are all explicitly included, has not yet been fully developed, to our knowledge, and the purpose of the present work is to fill the gap. A brief comparison of some representative existing approaches for radiation-induced radical transport and chemistry is given in Table 1. Monte Carlo track-structure models provide detailed stochastic descriptions but at considerable computational cost, while multiscale methods have typically employed simplified oxygen treatments or numerical diffusion schemes. In this context, the present effort formulates a fully analytical transport-reaction framework explicitly incorporating oxygen-dependent scavenging and LET-dependent radical production within a closed-form approach. This comparison brings into view the unique methodological niche addressed by the model developed herein. In this study, we propose an analytical reaction-kinetics model for hydroxyl radicals, taking into account both LET-related radical yields and oxygen-related chemical scavenging reactions. We begin from a Boltzmann transport equation, but under spatial averaging we simplify it to a steady-state concentration balance in an analytically closed form, retaining dose-rate, LET, and oxygen partial-pressure dependencies. The LET dependencies remain in the form of a physically motivated hydroxyl G-value determined from experimental radiolysis data and Monte Carlo transport calculations. On the other hand, treatment concerning oxidative effects uses conventional kinetic reactions without adjustable parameters. In the present framework, we choose to limit our discussion to the chemical stage, where oxygen is introduced via kinetic rate constants for reactions involving \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document}. The biological fixation of damage by oxygen, the subject of the oxygen enhancement ratio, is not treated in the present framework, but is intended to be included in higher-level biological models, for which the output of our calculations for the chemical stage would be input. Table 1Comparison of existing models and the current analytical model.Model/StudyApproachOxygen modelingTransport mechanismSolution typeHypoxia-focusedComputational costNikjoo et al. and Alloni et al9,10.Monte Carlo (PARTRAC)Implicit, via track-structure chemistryStochastic diffusionNumerical/stochasticLimitedHighSurdutovich and Solov’yov7Multiscale, LET-awareSimplified or normoxic assumptionsDiffusion-basedSemi-analyticalLimitedModerateKaramitros et al.5Monte Carlo (Geant4-DNA)Indirect via chemical stageStochastic diffusionFully numericalPartialVery highCurrent studyAnalytical reaction-kinetics (spatially averaged)Explicit, oxygen- and scavenger-dependentSimplified balance from Boltzmann–P₁ under \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\nabla\:.J\approx\:0$$\end{document}Closed-form analyticalYesLow† “Low” computational costs in the proposed analytical approach can be obtained through disregarding detailed spatial structure, stochastic effects, and track structure information. This enables quick parametric investigations and the integration of scales, yet it cannot replace the level of detail available from Monte Carlo simulations. Table 1Comparison of existing models and the current analytical model.Model/StudyApproachOxygen modelingTransport mechanismSolution typeHypoxia-focusedComputational costNikjoo et al. and Alloni et al9,10.Monte Carlo (PARTRAC)Implicit, via track-structure chemistryStochastic diffusionNumerical/stochasticLimitedHighSurdutovich and Solov’yov7Multiscale, LET-awareSimplified or normoxic assumptionsDiffusion-basedSemi-analyticalLimitedModerateKaramitros et al.5Monte Carlo (Geant4-DNA)Indirect via chemical stageStochastic diffusionFully numericalPartialVery highCurrent studyAnalytical reaction-kinetics (spatially averaged)Explicit, oxygen- and scavenger-dependentSimplified balance from Boltzmann–P₁ under \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\nabla\:.J\approx\:0$$\end{document}Closed-form analyticalYesLow† “Low” computational costs in the proposed analytical approach can be obtained through disregarding detailed spatial structure, stochastic effects, and track structure information. This enables quick parametric investigations and the integration of scales, yet it cannot replace the level of detail available from Monte Carlo simulations. Comparison of existing models and the current analytical model. † “Low” computational costs in the proposed analytical approach can be obtained through disregarding detailed spatial structure, stochastic effects, and track structure information. This enables quick parametric investigations and the integration of scales, yet it cannot replace the level of detail available from Monte Carlo simulations. One of the important characteristics of the current formulation is its transparency to the structure. The solution allows for a clear separation between the roles of radical production, transport, and absorption, and it is possible to study LET, oxygen levels, and dose rates independently. This also allows an assessment of its robustness and qualitative reproduction across scales compared to Monte Carlo calculations, as well as its separation into radiochemical states without requiring models associated with biological response and parameters of damage fitting. With its analytically tractable yet physically informed description of radiochemistry with LET and oxygen, this work is intended to provide a useful complement to detailed track structure calculations and to represent a bridge scale that can be leveraged by more macroscopic models of radiobiology. The goal is not to make predictions about biological endpoints but to provide a tractable and chemically well-founded base upon which additional complexities of biology might be built in a controlled and informative way. This suggests that it is well suited to multiscale models of radiation response or to exploring radiation qualities in inhomogeneous irradiation fields. Results This model accounts for the hydroxyl radical concentration as a balance of their generation and disappearance. The dependence of generation upon LET was approximated using phenomenological saturation function \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{\mathrm{O}\mathrm{H}}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)={G}_{0}{[1+{(\mathrm{L}\mathrm{E}\mathrm{T}/{\mathrm{L}\mathrm{E}\mathrm{T}}_{c})}^{\alpha\:}]}^{-1}$$\end{document}, reflecting the diminishing yield of radicals with an increase in LET owing to intra-track recombination effects. Loss mechanisms were included into the macroscopic absorptive cross section \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varSigma\:}_{a}={{\Sigma\:}}_{\mathrm{D}\mathrm{N}\mathrm{A}}+\sum\:_{i}{k}_{i}{C}_{i}+{k}_{{\mathrm{O}}_{2}}H{p}_{{\mathrm{O}}_{2}}$$\end{document}, and the steady-state concentration of hydroxyl radicals equals \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\phi\:}_{0}={S}_{0}/{{\Sigma\:}}_{a}$$\end{document}, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{0}$$\end{document} is the volumetric generation term (the product of dose rate and G-value). Based on this model, the following subsections provide the results. Internal consistency and order‑of‑magnitude checks Before comparing with any outside benchmarks, we first ensure that the analytical reaction-kinetics model (spatially averaged) produces physically realistic results, which match radiochemical expectations. The steady-state hydroxyl concentration is the result of the balance between LET-dependent production and chemical loss processes as given by the equation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\phi\:}_{0}={S}_{0}/{{\Sigma\:}}_{a}$$\end{document} (Sect. 4). Here, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{0}$$\end{document} is a function of the LET-modulated G-value of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} radicals, while \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\Sigma\:}}_{a}$$\end{document} represents the macroscopic absorption coefficient that takes account of reactions with DNA, molecular oxygen, and endogenous cellular scavengers. This formulation is consistent with standard descriptions of water radiolysis and radical transport1,4. Table 2 presents order-of-magnitude calculations of the hydroxyl radical steady-state concentrations based on reaction rates and cross sections established from the literature. The calculated values are consistent with the ranges reported in experimental radiolysis studies2,3,11 without any need for adjustment or fitting of model parameters. These order-of-magnitude estimates provide a basic check on the model’s consistency, demonstrating that it behaves as physically expected within these parameter ranges, without any fitting of the model to the data. The most important aspect of the results is the distinction between LET-induced radical production and chemistry-controlled loss. LET affects \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{0}$$\end{document} via radiochemical yield, while oxygen impacts \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\Sigma\:}}_{a}$$\end{document}, such that differences in radical concentration are ascribed to physically different processes. This decoupling is critical for multiscale modeling and has been previously highlighted in numerical simulations5,7. Lastly, the applicability of the steady-state approximation is justified based on timescale arguments, as radical lifetimes (picoseconds to nanoseconds) are orders of magnitude shorter than irradiation times in biological systems1–4. Taken together, the results in Table 2 show that the model is capable of predicting physically correct values of radical concentrations under various LET and microenvironmental conditions. Table 2A priori model-predicted ranges for LET-dependent hydroxyl radical production and steady-state density.LET RegimeRadiation typeOxygenationRadiochemical stage\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\boldsymbol{G}}_{\mathbf{O}\mathbf{H}}\left(\mathbf{L}\mathbf{E}\mathbf{T}\right)$$\end{document} (molecules/100 eV)Dose-rate regimeRepresentative \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{\boldsymbol{D}}$$\end{document} (Gy s⁻¹)Source Term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\boldsymbol{S}}_{0}$$\end{document}† (nm⁻³ s⁻¹)Absorption Coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\boldsymbol{\Sigma\:}}_{\boldsymbol{a}}$$\end{document} (nm⁻¹)Steady-State Density \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\boldsymbol{\phi\:}}_{0}$$\end{document} (nm⁻³)Low LETγ-rays/fast e⁻OxygenatedChemical stage2.6–2.9Conventional RT10⁻² – 10⁻¹10⁻⁸ – 10⁻⁷10⁻² – 10⁻³10⁻⁶ – 10⁻⁵Low LETγ-rays/fast e⁻HypoxicChemical stage2.6–2.9Conventional RT10⁻² – 10⁻¹10⁻⁸ – 10⁻⁷10⁻³ – 10⁻⁴10⁻⁵ – 10⁻⁴Low LETγ-rays/fast e⁻OxygenatedEarly-time (ps)4.7–5.0Any—10⁻⁷ – 10⁻⁶—‡Upper boundModerate LETProtons (10–100 MeV)OxygenatedChemical stage2.3–2.8Conventional/Hypofractionated10⁻¹ – 110⁻⁷ – 10⁻⁶10⁻² – 10⁻³10⁻⁵ – 10⁻⁴Moderate LETProtons (10–100 MeV)HypoxicChemical stage2.3–2.8Hypofractionated10⁻¹ – 110⁻⁷ – 10⁻⁶10⁻³ – 10⁻⁴10⁻⁴ – 10⁻³Moderate LETProtons (Bragg peak)OxygenatedChemical stage1.5–2.3Hypofractionated1–1010⁻⁶ – 10⁻⁵10⁻² – 10⁻³10⁻⁴ – 10⁻³Moderate LETProtons (Bragg peak)HypoxicChemical stage1.5–2.3Hypofractionated1–1010⁻⁶ – 10⁻⁵10⁻³ – 10⁻⁴10⁻³ – 10⁻²Moderate LETProtonsAnyEarly-time (ps)4.2–4.8Any—10⁻⁶ – 10⁻⁵—‡Upper boundHigh LETC, O ionsOxygenatedChemical stage0.7–2.0Particle therapy1–1010⁻⁶ – 10⁻⁵10⁻² – 10⁻³10⁻⁴ – 10⁻³High LETC, O ionsHypoxicChemical stage0.7–2.0Particle therapy1–1010⁻⁶ – 10⁻⁵10⁻³ – 10⁻⁴10⁻³ – 10⁻²High LETC, O ionsAnyEarly-time (ps)3.8–4.5Any—≥ 10⁻⁵—‡Upper bound† \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{0}$$\end{document} calculated from the analytical source-term formulation using water density and LET-dependent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{OH}$$\end{document}; values normalized per unit dose rate.‡ Early-time yields are not propagated to steady-state densities and are included solely to constrain the physical upper bound of radical production. Table 2A priori model-predicted ranges for LET-dependent hydroxyl radical production and steady-state density.LET RegimeRadiation typeOxygenationRadiochemical stage\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\boldsymbol{G}}_{\mathbf{O}\mathbf{H}}\left(\mathbf{L}\mathbf{E}\mathbf{T}\right)$$\end{document} (molecules/100 eV)Dose-rate regimeRepresentative \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{\boldsymbol{D}}$$\end{document} (Gy s⁻¹)Source Term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\boldsymbol{S}}_{0}$$\end{document}† (nm⁻³ s⁻¹)Absorption Coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\boldsymbol{\Sigma\:}}_{\boldsymbol{a}}$$\end{document} (nm⁻¹)Steady-State Density \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\boldsymbol{\phi\:}}_{0}$$\end{document} (nm⁻³)Low LETγ-rays/fast e⁻OxygenatedChemical stage2.6–2.9Conventional RT10⁻² – 10⁻¹10⁻⁸ – 10⁻⁷10⁻² – 10⁻³10⁻⁶ – 10⁻⁵Low LETγ-rays/fast e⁻HypoxicChemical stage2.6–2.9Conventional RT10⁻² – 10⁻¹10⁻⁸ – 10⁻⁷10⁻³ – 10⁻⁴10⁻⁵ – 10⁻⁴Low LETγ-rays/fast e⁻OxygenatedEarly-time (ps)4.7–5.0Any—10⁻⁷ – 10⁻⁶—‡Upper boundModerate LETProtons (10–100 MeV)OxygenatedChemical stage2.3–2.8Conventional/Hypofractionated10⁻¹ – 110⁻⁷ – 10⁻⁶10⁻² – 10⁻³10⁻⁵ – 10⁻⁴Moderate LETProtons (10–100 MeV)HypoxicChemical stage2.3–2.8Hypofractionated10⁻¹ – 110⁻⁷ – 10⁻⁶10⁻³ – 10⁻⁴10⁻⁴ – 10⁻³Moderate LETProtons (Bragg peak)OxygenatedChemical stage1.5–2.3Hypofractionated1–1010⁻⁶ – 10⁻⁵10⁻² – 10⁻³10⁻⁴ – 10⁻³Moderate LETProtons (Bragg peak)HypoxicChemical stage1.5–2.3Hypofractionated1–1010⁻⁶ – 10⁻⁵10⁻³ – 10⁻⁴10⁻³ – 10⁻²Moderate LETProtonsAnyEarly-time (ps)4.2–4.8Any—10⁻⁶ – 10⁻⁵—‡Upper boundHigh LETC, O ionsOxygenatedChemical stage0.7–2.0Particle therapy1–1010⁻⁶ – 10⁻⁵10⁻² – 10⁻³10⁻⁴ – 10⁻³High LETC, O ionsHypoxicChemical stage0.7–2.0Particle therapy1–1010⁻⁶ – 10⁻⁵10⁻³ – 10⁻⁴10⁻³ – 10⁻²High LETC, O ionsAnyEarly-time (ps)3.8–4.5Any—≥ 10⁻⁵—‡Upper bound† \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{0}$$\end{document} calculated from the analytical source-term formulation using water density and LET-dependent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{OH}$$\end{document}; values normalized per unit dose rate.‡ Early-time yields are not propagated to steady-state densities and are included solely to constrain the physical upper bound of radical production. A priori model-predicted ranges for LET-dependent hydroxyl radical production and steady-state density. † \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{0}$$\end{document} calculated from the analytical source-term formulation using water density and LET-dependent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{OH}$$\end{document}; values normalized per unit dose rate. ‡ Early-time yields are not propagated to steady-state densities and are included solely to constrain the physical upper bound of radical production. Comparison with experimental radiolysis data Any LET-dependent radiochemical model is also required to be consistent with the yields of experimentally determined hydroxyl radicals. The yields of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} radicals, measured experimentally by scavenger techniques, pulse radiolysis, and track segment analysis, have all been observed to decrease monotonically with increasing LET due to increased intra-track recombination1,3. Figure 1(a) plots the calculated LET-dependent G-values from the model against the experimental ranges given in Supplementary Table S16,12–20. The analytical model captures both the mean and the variability of the experimental data over more than two orders of magnitude in LET without any empirical fitting. All parameters were fixed independently based on radiochemical and track structure models2,3,11. The calculated curve accurately reproduces the flat region at low LET, the gradual decrease at intermediate LET, and the saturation at high LET, as seen in radiolysis experiments. It should be noted that the agreement shown here is a result of the ability of the chosen functional form to follow the trend of the compiled experimental data. Comparison with monte carlo track-structure simulations Monte Carlo track structure simulations yield detailed descriptions of the interactions of radiation with matter on the nanometer scale, including the stochastic processes of energy deposition, radical formation, diffusion, and early chemical reactions2,3,5. Comparison with such simulations is a key test for the consistency of a reduced analytical model in terms of its ability to correctly reproduce the underlying physics and chemistry of hydroxyl radical availability. The analytical model does not specifically simulate track geometry, spatial clustering, or recombination in the early stages of damage formation, but instead includes all of these effects in the LET-dependent source term. The consistency of the results in the normalized LET dependence shows that the model captures the expected qualitative behavior of radicals at a coarse level but does not reproduce the detail of the track structure simulations. This is the approach taken in Monte Carlo codes such as PARTRAC, Geant4-DNA, TOPAS-nBio, and RITRACKS, which explicitly simulate individual ionization events and early-stage chemistry with very detailed spatial and temporal resolution2,3,5. In contrast, the present model treats these processes in a more macroscopic manner, grouping the effects of individual events into an LET-dependent source term and macroscopic absorption coefficients. A summary of estimates for the yield of hydroxyl radicals from representative Monte Carlo calculations for various qualities and energies of the incident radiation is given in Supplementary Table S2. Figure 1(b) compares the normalized dependence on LET of the steady-state density of hydroxyl radicals predicted by the analytical model and the Monte Carlo calculations. All values are normalized with respect to low-LET values to isolate the effects of radiation quality. In both calculations, there is a monotonic decrease in the availability of radicals with increasing LET, and the curves have similar shapes over more than two orders of magnitude. Although the number of Monte Carlo data points available in the low LET region is limited in the literature, the available data (e.g., for photons and fast electrons) do follow the low LET plateau of the model. In addition, a comprehensive list of the available benchmark data from various Monte Carlo calculations is provided in Supplementary Table S2. The consistency of the Monte Carlo data across the whole LET range support the consistency of the model that the analytical model correctly represents the radiochemical trends at a coarse-grained level. The analytical model also reproduces the familiar curvature seen in Monte Carlo calculations. At low LETs, the radicals are seen to have constant yields owing to the sparsity of the track structures. At intermediate LETs, the suppression is seen to be quite strong, and at high LETs, the suppression is seen to saturate, as reported in the literature for the track core effect2,3,11. These differences between the individual Monte Carlo results and the analytical curves are expected, as the analytical model does not explicitly address the geometry of the track, spatial heterogeneity, or DNA topology2,5. However, the level of consistency seen in the results of Fig. 1(b) demonstrates that the key radiochemical effects as a function of LET are captured with sufficient accuracy. Fig. 1Normalized hydroxyl radical yields as a function of LET. (a) The analytical model (shaded region, indicating physically plausible parameter variation) is compared to experimental radiolysis data compiled from the literature (symbols). See Supplementary Table S1 for further details. The analytical model reproduces the monotonic decrease from the low LET plateau to the saturation value at high LET. (b) The normalized steady state hydroxyl radical density (or yield) from the analytical model (shaded region) is compared to the results from Monte Carlo track structure simulations (symbols). In the low LET region, the published data from the latter method are relatively scarce, but the few data points available (e.g., for photons and fast electrons) do follow the low LET plateau of the analytical model. Additional benchmark data for the full LET range can be found in Supplementary Table S2. The overall monotonic decrease with increasing LET is correctly reproduced by the analytical model, thus confirming the cross-scale consistency between the two approaches. Normalized hydroxyl radical yields as a function of LET. (a) The analytical model (shaded region, indicating physically plausible parameter variation) is compared to experimental radiolysis data compiled from the literature (symbols). See Supplementary Table S1 for further details. The analytical model reproduces the monotonic decrease from the low LET plateau to the saturation value at high LET. (b) The normalized steady state hydroxyl radical density (or yield) from the analytical model (shaded region) is compared to the results from Monte Carlo track structure simulations (symbols). In the low LET region, the published data from the latter method are relatively scarce, but the few data points available (e.g., for photons and fast electrons) do follow the low LET plateau of the analytical model. Additional benchmark data for the full LET range can be found in Supplementary Table S2. The overall monotonic decrease with increasing LET is correctly reproduced by the analytical model, thus confirming the cross-scale consistency between the two approaches. Sensitivity and Robustness Another important condition for the credibility of the LET-dependent radiochemical model is its robustness in the presence of uncertainties in the model parameters. In the current model, the LET dependence is limited to the parameterization of the hydroxyl radical G-value, i.e., the transition LET scale (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{L}\mathrm{E}\mathrm{T}}_{c}$$\end{document}) and the curvature parameter (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\alpha\:$$\end{document}). Table 3 shows the ranges of the parameters that were chosen to cover the experimentally and computationally reported values for photons, protons, and light ions2,3,7,11. It should be noted that Fig. 1(a) illustrates the strict monotonicity of the normalized LET dependence of the hydroxyl radical source term for all tested combinations of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{L}\mathrm{E}\mathrm{T}}_{c}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\alpha\:$$\end{document}. The changes in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{L}\mathrm{E}\mathrm{T}}_{c}$$\end{document} result in the shift of the onset of the suppression along the LET axis, and α determines the “sharpness” of the transition. Figure 1(b) illustrates that these changes are propagated to the steady-state hydroxyl radical density, keeping the consistency with the Monte Carlo track-structure envelopes2,3. Notably, no compensatory tuning is present between LET-dependent parameters and those related to oxygen, since LET influences only the source term, whereas oxygen-dependent phenomena enter through the absorption coefficient and include the scavenge term. Accordingly, uncertainties in LET parameters do not influence oxygen dependencies. Overall, the sensitivity analysis described in Table 3 and depicted in Figs. 1(a) and 1(b) confirms that the LET-dependent reduction of hydroxyl efficacy is a matter of model structure and not a result of fine tuning of model parameters. Table 3LET-dependent model parameters and literature-based justification.ParameterSymbolAdopted rangePhysical meaning in modelLiterature basis and justificationSensitivity impactLow-LET hydroxyl radical yield \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{0}$$\end{document} 2.6–3.1 molecules/100 eVReference \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} yield under photon/fast-electron irradiationCanonical chemical-stage yields in liquid water reported by Buxton et al.14, Elliot & Bartels15, and ICRU-supported compilations; weak dependence on experimental protocol at low LETSets absolute normalization only; does not affect LET trendCharacteristic LET scale \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{L}\mathrm{E}\mathrm{T}}_{c}$$\end{document} 30–100 keV/µmLET at which intra-track recombination and spur overlap significantly suppress \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} productionInferred from heavy-ion radiolysis experiments13,17 and Geant4-DNA chemical-stage simulations19; consistent with onset of dense track coresControls transition region; shifts curve horizontally without altering monotonicityPhenomenological exponent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\alpha\:$$\end{document} 1–2Controls sharpness of LET-induced suppressionCaptures variability between gradual (α ≈ 1) and steep (α ≈ 2) reductions observed across ion species and time scales in both experiments and Monte Carlo studies6,12Modulates curvature; does not change ordering or trendEffective radical source term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{0}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)$$\end{document} Derived quantityLET-dependent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} generation rate used as input to transport modelComputed directly from \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{\mathrm{O}\mathrm{H}}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)$$\end{document}; no additional free parameters introducedFully constrained once \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{0}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{L}\mathrm{E}\mathrm{T}}_{c}$$\end{document}, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\alpha\:$$\end{document} are specified Table 3LET-dependent model parameters and literature-based justification.ParameterSymbolAdopted rangePhysical meaning in modelLiterature basis and justificationSensitivity impactLow-LET hydroxyl radical yield \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{0}$$\end{document} 2.6–3.1 molecules/100 eVReference \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} yield under photon/fast-electron irradiationCanonical chemical-stage yields in liquid water reported by Buxton et al.14, Elliot & Bartels15, and ICRU-supported compilations; weak dependence on experimental protocol at low LETSets absolute normalization only; does not affect LET trendCharacteristic LET scale \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{L}\mathrm{E}\mathrm{T}}_{c}$$\end{document} 30–100 keV/µmLET at which intra-track recombination and spur overlap significantly suppress \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} productionInferred from heavy-ion radiolysis experiments13,17 and Geant4-DNA chemical-stage simulations19; consistent with onset of dense track coresControls transition region; shifts curve horizontally without altering monotonicityPhenomenological exponent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\alpha\:$$\end{document} 1–2Controls sharpness of LET-induced suppressionCaptures variability between gradual (α ≈ 1) and steep (α ≈ 2) reductions observed across ion species and time scales in both experiments and Monte Carlo studies6,12Modulates curvature; does not change ordering or trendEffective radical source term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{0}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)$$\end{document} Derived quantityLET-dependent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} generation rate used as input to transport modelComputed directly from \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{\mathrm{O}\mathrm{H}}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)$$\end{document}; no additional free parameters introducedFully constrained once \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{0}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{L}\mathrm{E}\mathrm{T}}_{c}$$\end{document}, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\alpha\:$$\end{document} are specified LET-dependent model parameters and literature-based justification. Fully constrained once \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{0}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{L}\mathrm{E}\mathrm{T}}_{c}$$\end{document}, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\alpha\:$$\end{document} are specified Qualitative consistency with indirect damage trends As a further cross-scale verification, we can compare the LET dependence of the availability of the hydroxyl radicals predicted by the model with the qualitative trends known to exist in the availability of indirect DNA damage. A detailed mechanistic treatment of the yields of DNA damage from the concentrations of radicals would necessitate the modeling of DNA geometry, reaction probabilities, as well as the competition between damage and repair processes, which is beyond the scope of the present study. However, under the simplified assumption that the availability of indirect DNA damage is proportional to the steady-state availability of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} radicals, the LET dependence of the normalized availability of indirect DNA damage would follow the same trend as the source term. According to this approach, this relationship is given by1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{{G}_{OH}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)}{{G}_{0}}\propto\:\frac{{Y}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{t}}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)}{{Y}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{t}}\left(0\right)}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Y}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{t}}$$\end{document} is the yield of indirect DNA damage, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{0}$$\end{document} is the low LET reference hydroxyl radical yield. The proportionality indicates that the yields are related through their common dependence on the chemistry mediated by the radicals. It is important to stress that Eq. (1) provides an oversimplified proportionality for illustrative purposes only and does not take into account such factors as the spatial arrangement of DNA, lesion clustering, chromatin protection, and the more complicated overlapping of tracks that becomes relevant at high LETs. Damage production in biology is affected by such non-linear phenomena and, therefore, cannot be described using Eq. (1). The equation provides nothing but a simple bridge between the output of our chemical stage, namely, the steady-state radical density, and further biological modeling. Figure 2 presents the LET dependence of hydroxyl radical yields as computed by the model in comparison to experimental trends for indirect DNA damage yields due to different types of radiation1–3,11. Both the model and experimental data display similar behavior: gradual decrease with increasing LET, accompanied by significant suppression at intermediate to high LET values. Equation (1) is a representation of this proportionality, which is introduced here not as a predictive model for DNA damage, but as a conceptual link to discuss how this output might interface with a biological model. The proportionality should be viewed as a consistency argument. It should be noted that the model does not attempt to describe direct DNA damage, clustered damage, or high LET enhancement, as these are outside the realm of diffusive radical chemistry. This similarity between the model and experimental data serves as an important consistency check on the biological plausibility of the analytical radiochemical framework while maintaining the strict separation between chemical cause and biological effect. The analytical model calculates the dependence on normalized LET for indirect DNA damage by assuming that the yields for damage scale linearly with the density of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} at steady state (Eq. (1)). The data plotted in Fig. 2 for the Monte Carlo track structure calculations are based on detailed simulations that include the geometry of energy deposition and the chemistry and geometry of DNA at the nanometer scale (see Supplementary Table S2 for a compilation of the data and references). The scatter between the analytical calculation and the simulation data is a result of the fundamental difference in the level of modeling: the present model is a spatially averaged, steady-state approach that includes the complex track structure calculations into a phenomenological LET-dependent source term, while the Monte Carlo calculations include the full stochastic fluctuations and the time development of the chemical reactions. Thus, the comparison should be viewed as a qualitative consistency check, which demonstrates that both approaches exhibit a monotonic trend towards decreasing indirect damage yields as LET increases. However, they should not be expected to yield identical results. The direct experimental validation of indirect damage yields over a wide LET range is difficult, as most experiments, e.g., plasmid DNA breakage, integrate both direct and indirect effects, and often repair processes are included. Figure 2 is therefore meant to illustrate the cross-scale consistency between a coarse-grained analytical model and finer-scale computational simulations. Cross-scale credibility The above agreement procedures show that the current analytical framework is consistent across physical, chemical, and biological scales without parameter adjustment or phenomenological modeling. The LET-dependent hydroxyl radical source term is based on conventional radiolysis physics and is consistent with experimental G-values1,3,4, producing physically correct concentrations of radicals. At the simulation level, the model captures the LET-dependent behavior found in Monte Carlo track-structure simulations2,5,11 without simulating microscopic geometry. At the biological level, the calculated reduction in hydroxyl radical availability is consistent with experimentally measured decreases in indirect DNA damage with increasing LET1–3,21. This consistency is due to the clear distinction between LET-dependent production, transport, and scavenging, thus confirming the current analytical framework’s cross-scale validity. Normalized LET-dependent radical density In order to investigate the intrinsic effect of radiation quality on indirect chemical reactions, all results are presented in a normalized manner with respect to a reference radiation of low LET. This normalization procedure eliminates any trivial scaling effects with dose rate and material parameters and facilitates direct investigation of radiochemical effects related to LET. The volumetric source term related to hydroxyl radical formation is found to be proportional to the related G-value, as shown in Sect. 4. The normalization of the source term is then defined as2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{{S}_{0}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)}{{S}_{0}\left(0\right)}=\frac{{G}_{\mathrm{O}\mathrm{H}}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)}{{G}_{0}}$$\end{document} In this normalization, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{0}\equiv\:{G}_{OH}(\mathrm{L}\mathrm{E}\mathrm{T}\to\:0)$$\end{document} represents the canonical low-LET G-value. This normalization procedure eliminates any dependence on density, dose rate, and energy conversion factors and provides a universal reference to all radiation qualities. Equation (2) describes an important finding of this work: the dependence of the normalized hydroxyl radical source on LET arises solely from radiochemical effects. It is independent of the modality and rate of delivery and arises because of increased intra-track recombination and spur overlap at higher LET. Low-LET radiation leaves the indirect chemical potential virtually intact, while higher LET radiation reduces the fraction of chemically available radicals. In the context of the analytical reaction-kinetics model, the steady-state hydroxyl radical density will be directly proportional to the effective source term. The normalized steady-state radical density will have exactly the same dependence on LET as the normalized source, without any additional curvature, inflection points, or particle type. Because the loss processes appear as linear terms in the balance equation, the normalized LET dependence of the source term is preserved in the steady‑state radical density. This result illustrates the distinct separation of function in the model: LET controls the efficiency in the source term for radical production, and absorption transform this input into an available steady-state density in a manner that is independent of radiation quality. The normalized form thus serves as a physically transparent starting point for further investigations of oxygen effects, dose rate scaling, and combined parameter dependencies. Fig. 2Comparison of the normalized yields of indirect DNA damage versus LET predicted by the present radiochemical model (solid curve with uncertainty shading) against Monte Carlo track-structure calculations for various ion species. Comparison of the normalized yields of indirect DNA damage versus LET predicted by the present radiochemical model (solid curve with uncertainty shading) against Monte Carlo track-structure calculations for various ion species. Combined dependence on LET and Oxygen Although the Sect. 4 shows that the dependence of the steady-state hydroxyl radical density on LET is a direct consequence of the radiochemical source term, the general analytical solution enables the combined impact of oxygenation and dose rate to be explored in detail. By substituting the LET-dependent source term into the steady-state solution, the resulting radical density is a function of dose rate, LET, and oxygen partial pressure. This dependence can be written in normalized form as3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{{\phi\:}_{0}(\dot{D},\mathrm{L}\mathrm{E}\mathrm{T},{p}_{{\mathrm{O}}_{2}})}{{\phi\:}_{0}(\dot{D},\mathrm{L}\mathrm{E}\mathrm{T},0)}=\frac{{{\Sigma\:}}_{\mathrm{b}\mathrm{g}}}{{{\Sigma\:}}_{\mathrm{b}\mathrm{g}}+{k}_{{\mathrm{O}}_{2}}H{p}_{{\mathrm{O}}_{2}}}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\Sigma\:}}_{\mathrm{b}\mathrm{g}}={{\Sigma\:}}_{\mathrm{D}\mathrm{N}\mathrm{A}}+\sum\:_{i}^{}{k}_{i}{C}_{i}$$\end{document} is the LET-independent background scavenging of DNA and background cellular constituents. This expression captures the purely chemical competition between oxygen-mediated quenching and background radical scavenging. Figure 3 illustrates the normalized hydroxyl radical yield as a function of oxygen partial pressure for three example background scavenging rates. Strong, nonlinear inhibition is evident at low to moderate oxygenation levels, while higher background scavenging rates reduce oxygen sensitivity by shunting radicals down non-oxygen-mediated pathways. All reaction rates employed in these simulations are listed in Supplementary Table S3. It is noteworthy that the data in Fig. 3 are LET and dose rate independent, and this is a direct result of the hierarchical nature of the model, in which LET influences the efficiency of radical production, oxygen affects the lifetime of chemicals, and dose rate determines the intensity. This gives a mechanistic basis for understanding the combined effects of radiation quality and micro-environment on indirect chemical damage. Structural robustness and invariance Apart from agreement with external benchmarks, another important virtue of the analytical framework is its intrinsic robustness in terms of structure. The steady-state density of the hydroxyl radicals is defined as a ratio of production and absorption terms in which the LET, the oxygenation, and the dose rate are involved in a physically different manner. The LET is involved in the radiochemical source through the G-value of the hydroxyl radicals, the oxygen is involved in the survival of the radicals through the scavenging kinetics, and the dose rate scales the overall production intensity. This distinction is made explicit under normalized form, where the dependence on dose rate is removed and the separate influences of LET and oxygen are distinguished. It should be noted that this separability is a direct consequence of the multiplicative form of the analytical solution \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\phi\:}_{0}={S}_{0}/{{\Sigma\:}}_{a}$$\end{document} in which LET appears in the source term, oxygen in the absorption coefficient, and dose rate as a linear scaling factor. Therefore, what appears to be an orthogonality among these parameters is more a function of the model assumptions than an externally derived physical truth. Nonetheless, this separability offers a clear basis for understanding the relative contributions to the radiochemical state. Crucially, this invariance is seen as a direct consequence of the analytical form and not as a result of parameter adjustment, and this is seen as key to the robustness of the model under heterogeneous dose environments. Model-predicted radiochemical regimes The analytical model developed in this work allows for a natural grouping of irradiation conditions into separate radiochemical regimes that arise from the combined but analytically separable action of LET‑dependent radical production and oxygen‑dependent chemical scavenging. These regimes correspond to areas of the parameter space (LET, oxygen partial pressure) where a particular suppression mechanism dominates the steady-state radical density. They are not intended as a physical result, but rather as a useful interpretive aid, which follows immediately from the form of the normalized steady-state result. In this sense, the regimes offer a useful means of visualizing the interplay between radiation quality and microenvironment effects on radical availability. In normalized form, the steady-state density of hydroxyl radicals can be expressed as a product of independent factors related to LET-controlled production efficiency, oxygen-induced scavenging, and dose rate effects. Such a description makes it possible to determine areas in the LET-oxygen parameter space where one suppression mechanism is more important than others. Figure 4 presents these regimes as contours of the logarithm of the reduced hydroxyl radical density, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{log}}_{10}\left[{\phi\:}_{0}\right(\dot{D},\mathrm{L}\mathrm{E}\mathrm{T},{p}_{{\mathrm{O}}_{2}})/{\phi\:}_{0}(\dot{D},\mathrm{0,0}\left)\right]$$\end{document}, as a function of LET and partial pressure of oxygen. Contour levels at −0.5, −1.5, and − 3 represent orders-of-magnitude suppression with respect to the low-LET and anoxic reference state and are provided as a graphical aid to distinguish between regimes of weak, moderate, and strong suppression of indirect chemical activity. Fig. 3Normalized steady-state hydroxyl radical flux as a function of oxygen partial pressure for different biological scavenging conditions, showing maximal sensitivity in the hypoxic-to-normoxic transition region. Normalized steady-state hydroxyl radical flux as a function of oxygen partial pressure for different biological scavenging conditions, showing maximal sensitivity in the hypoxic-to-normoxic transition region. Fig. 4Contour map of normalized steady-state hydroxyl radical density across the parameter space of LET (log scale) and oxygen partial pressure, identifying distinct radiochemical regimes. Contour map of normalized steady-state hydroxyl radical density across the parameter space of LET (log scale) and oxygen partial pressure, identifying distinct radiochemical regimes. At low LET and low oxygen tension, the model indicates that there should be a production-dominated regime, with efficient production of hydroxyl radicals and low levels of quenching. In this region, indirect chemical effects are maximized, and changes in LET will scale the yield according to radiolysis efficiency with minimal effect from oxygen. With increasing oxygen tension at constant LET, there should be a smooth transition into an oxygen-controlled region, with increased levels of scavenging limiting the supply of radicals. In this region, the effect of LET on production remains, but the range of achievable radical density is reduced due to the effect of oxygen on quenching, reflecting the reduced indirect effect at higher oxygen tension. At high LET values, the model enters a regime dominated by track structure, largely independent of the oxygen level. In this regime, the high density of energy deposition and increased recombination within a track strongly inhibit the production of diffusible hydroxyl radicals, and thus the impact of oxygen scavenging appears less significant. This is a fundamental limitation of indirect chemistry at high LET. Crucially, the transitions between regimes are continuous and distinguished by order-of-magnitude changes in radical density, rather than empirical thresholds. This serves to emphasize the robustness of the decoupling and offers a physically interpretable framework within which radiochemical mechanisms can be associated with biological modeling on a higher level without confusing chemical causality with biological outcome. Discussion In this work, we have developed a physically based analytical model that explicitly links radiation quality to the indirect chemical phase of radiation interaction by incorporating an LET-dependent model of hydroxyl radical production and transport. The model offers clarity regarding the role of radiation quality in the radiochemical interaction by incorporating a minimal yet physically based formulation. A central outcome of this framework is that, within the model’s assumptions, the dependence on LET is seen to affect the chemical phase mainly by modulating the efficiency of hydroxyl radical production, quantitatively given by the LET-dependent G-values, with the tension in oxygen affecting the availability of radicals by well-established kinetics and dose rate scaling the total production. This separation follows the multiplicative form of the analytical solution \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\phi\:}_{0}={S}_{0}/{\varSigma\:}_{a}$$\end{document}. This result emerges naturally from the governing equations and provides a transparent separation of control parameters, a feature that may be less transparent in simulation-based approaches. The analytical structure permits complex multivariable behavior to be understood in a simple fashion. LET specifies the intrinsic efficiency with which energy deposition is converted to diffusible chemical agents, while oxygen regulates their survival through chemical quenching, and dose rate linearly scales radical flux without affecting normalized dependencies. This not only enhances clarity but also avoids over correlation with radiation quality, microenvironment, and dosimetry-related factors. The robustness of the framework with respect to parameter variation, as well as the absence of fine-tuning, also supports robustness and reproducibility. Comparison with experimental radiolysis data and Monte Carlo track structure simulations shows that the model indeed reproduces the expected qualitative behavior of hydroxyl yields as a function of LET and radiation type. These are provided as cross-scale consistency checks rather than verifications of the underlying radiochemical processes. The quantitative consistency is supported with respect to the monotonic behavior, order-of-magnitude reproduction, and suppression with increasing LET, which are the relevant consistency criteria for the model. The observed discrepancies with respect to the yields of the different particle species at low and very high LET values are due to the track structure effects beyond the one-parameter LET description. Instead, these discrepancies are consistent with the physical interpretation of the LET descriptor. A significant conceptual result of the model has been the identification of various radiochemical regimes that exist in the combined LET‑oxygen parameter space. These regimes are production-limited, scavenging-dominated, and track-structure-suppressed. They are determined by the structure of the model, where one of the processes that compete to determine the steady-state density of radicals becomes dominant. Such regimes are not offered as individual physical discoveries but as a succinct interpretative model which is directly implied by the mathematical expression. The transitions between regimes are smooth and continuous, reflecting the underlying physics and chemistry rather than any step function thresholds. Such an interpretation of the radiochemical behavior in terms of regimes has been found useful as a simple and portable way to describe complex radiochemical behavior involving many variables and to provide some insight into the nature of indirect chemical contributions. From a radiobiological point of view, the results support the interpretation that the chemical phase acts as an intermediary between the deposition of physical energy and the formation of biological damage. LET is represented in the model as not only a measure of the spatial clustering of energy but also a determinant of the efficiency of chemical conversion. The systematic reduction of hydroxyl radical production at higher LET values suggests that radiation quality is not only a determinant of the location of energy deposition but also of the efficiency of energy conversion to diffusible species that can initiate indirect damage. This is especially important in the intermediate LET region, where direct and indirect effects coexist and compete. Here, it is essential to distinguish between the chemical kinetics of oxygen within the radiolysis medium and the biological role of oxygen as a dose-modifying agent. Molecular oxygen does react with hydroxyl radicals in pure water or diluted solutions with a rate constant of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:1.9\times\:{10}^{9}{\mathrm{M}}^{-1}{\mathrm{s}}^{-1}$$\end{document}14. Yet, this reaction does not play a role quantitatively within the complex biochemical environment of the cell nucleus compared to the diffusion-controlled reactions of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} with DNA, proteins, and other organic molecules, which take place on a picosecond to nanosecond time scale1,4. The major effect of oxygen as a radio sensitizing agent does not act via the direct reaction of oxygen with hydroxyl radicals, but rather via the oxygen fixation hypothesis22,23. In this scheme, the involvement of oxygen is to react with the carbon-centered radicals on the DNA (DNA•), which were formed by the action of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} radicals or by direct ionization. This leads to the formation of DNA peroxyl radicals (DNA-OO•), which are chemically stable and cannot be repaired by cellular thiols. This is the fixation event. The analytical model accounts for the involvement of oxygen in the chemical event by incorporating the term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{k}_{{\mathrm{O}}_{2}}H{p}_{{\mathrm{O}}_{2}}$$\end{document} in the absorption coefficient. This term reflects the competition between oxygen and other solutes in the solution for the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} radicals. This is in agreement with standard conventions in radiation chemistry24. Nevertheless, we would like to emphasize that this kinetic component on its own does not capture the full biological oxygen effect; rather, this component captures a part of a multi-step process that involves radical production, induction of DNA damage, as well as oxygen fixation. When this chemical stage process is to be embedded within a multiscale approach, oxygen fixation needs to be taken into consideration at the biological level, as shown conceptually in Fig. 2. A more detailed approach to the oxygen fixation process at the level of DNA DSBs, as well as LET-dependent fixation kinetics and overkill effects, has been introduced in our recent work25, where this output at the chemical stage, as obtained within this or similar models, could be used as an input to simulate at the biological level. The scheme also has implications for the understanding of the variation of RBE with LET, oxygenation, and dose rate. The LET dependence of RBE variations cannot be explained solely in terms of increased direct damage or the clustering effect; the modulation of the chemical component is also involved. The model, with its explicit analytical description of the LET dependence of the suppression of the radical-mediated processes, may be used as a basis for understanding the failure of RBE to increase monotonically with LET, as well as the interaction between oxygen and dose rate effects. Considering the future perspective, the nature of analysis inherent in the present framework makes it highly suitable for integration into dynamic biological models of damage repair (e.g., WAM model). Within these models, the steady-state density of radicals \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\phi\:}_{0}$$\end{document}, which remains unchanged throughout the course of irradiation, must be viewed as a dose-rate dependent damage constraint or boundary condition, but not as a time-dependent stressor. It represents the total chemical insult, whose subsequent evolution is governed by repair and elimination mechanisms at the biological stage. Although the model’s strengths lie in its capabilities, the model’s domain is restricted by design. LET is employed as a leading-order descriptor for radiation quality, and details of nanoscale track structures are not resolved. The phenomenological LET dependence of the hydroxyl yield is only defined for LET values up to approximately 300 keV/µm; extrapolation to higher values would require a modified form that includes a residual yield. The steady-state model describes the chemical phase between intra-track recombination and biological processing but is not intended to predict the yield of DNA damage or its repair or biological outcomes. Oxygen scavenging is addressed using a macroscopic approximation that reproduces major physiological effects but ignores gradients and reaction complexity. These are not weaknesses but rather deliberate design choices that provide clear avenues for future improvement. Within these limits, the current work offers a credible, reproducible, and analytically solvable model of LET-dependent radiochemistry. By keeping the distinction between the chemical causes and biological effects sharp, this model offers a sound basis for the integration of radiochemistry with other models at different scales, while keeping the physical model physically interpretable and non-overextended. The comparisons made here are meant as cross-scale consistency checks, not as validation of the underlying radiochemical process, which remains parameterized phenomenologically. This work presents a physically transparent and mathematically tractable framework for describing the indirect chemical component of radiation action as a function of radiation quality (LET), dose rate, and microenvironmental oxygenation. By formulating hydroxyl radical production and scavenging within a unified steady-state approach, the model provides a clear separation between radiochemical source terms and macroscopic absorption processes. A central outcome of this framework is the illustration that, under the assumptions of the present framework, the LET dependence of the indirect chemical mechanism is represented through the hydroxyl radical G-value, oxygen tension influences radical availability via well‑established scavenging kinetics, and dose rate scales the overall radical flux. This has been a result of the multiplicative form of the analytical solution, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\phi\:}_{0}={S}_{0}/{\varSigma\:}_{a}$$\end{document}, as well as the hierarchical structure of the governing equations, which allows the individual parameter dependencies to be distinguished, a process that may be less transparent in simulation-based approaches. The model also distinguishes different radiochemical regimes in the parameter space defined by LET and oxygen. These are defined as regions in the parameter space based on the model behavior and provide a convenient way to express the interaction between radiation quality and oxygen availability. They are based on the structure of the analytical solution but not derived from first principles, and they are presented without reference to any model of biological response. The resulting transition regions between production-limited, scavenging-dominated, and track-structure-controlled behavior are continuous and physically interpretable. By grounding all chemical parameters in literature-established values and eschewing parameter fitting, the present framework emphasizes credibility, reproducibility, and conceptual clarity. Its analytic structure makes it readily extensible and suitable for embedding into multiscale radiobiological models and related studies that link radiochemistry to downstream biological processes, while making a clean separation between chemical cause and biological effect. In particular, the steady‑state radical densities provided by the model can serve as input quantities to time‑dependent biological repair models, providing a quantitative basis that may support downstream modeling of processes such as DNA repair and cell fate determination, rather than establishing the biological mechanisms directly. Methods The current investigation adopts a reduced analytical formalism for the modeling of hydroxyl radical reaction kinetics in nuclear microenvironments under spatial averaging for conditions representative of irradiation, attempting to capture the strong physical and chemical couplings that underlie indirect radiation damage with a manageable level of analytical complexity. It specializes the formalism for the chemical stage of the radiolysis, once the physical energy deposition processes are completed and prior to the fixation of biological damage. Radiochemical source term Hydroxyl radicals form as a result of radiolysis in ionized and excited water molecules due to ionizing radiation. On timescales below tens of femtoseconds, initially forming spurs and track structures occur; then, chemically active radicals appear and commence diffusion. On the scale relevant to this study, hydroxyl radicals’ production may be modeled using a source term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{0}$$\end{document}, referring to the amount of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} radicals created per time and per volume. For an absorbed dose rate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{D}$$\end{document} given to a medium of mass density \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\rho\:$$\end{document}, the source term is given by4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{0}\left(\dot{D},\mathrm{L}\mathrm{E}\mathrm{T}\right)=\rho\:\dot{D}{G}_{\mathrm{O}\mathrm{H}}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right){C}_{E}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{\mathrm{O}\mathrm{H}}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)$$\end{document} is the LET-dependent hydroxyl radical G-value, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{C}_{E}$$\end{document} is the energy-to-quanta conversion constant. The constant \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{C}_{E}$$\end{document} emerges strictly as a consequence of the definition of dose absorption and the relationships between joules and electron volts. With \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:1\:\mathrm{G}\mathrm{y}=1\:\mathrm{J}{\mathrm{k}\mathrm{g}}^{-1}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:1\:\mathrm{e}\mathrm{V}=e\:\mathrm{J}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{C}_{E}=(1/100\:\mathrm{e}\mathrm{V})\times\:(1\:\mathrm{J}/e)=6.24\times\:{10}^{15}$$\end{document} (100 eV quanta per Gy per gram). This constant has no biological or kinetic content; it depends neither on radiation quality nor on any biological hypothesis. A formally consistent expression of the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} source term has been found. The explicit dose rate and LET dependency correctly express established reduction factors for free radical yields with increasing track density as a result of intra-track recombination and spur overlap. Reaction and absorption of hydroxyl radicals Following formation, hydroxyl radicals diffuse, but we average over spatial scales larger than the diffusion length. They react with DNA (the primary target of indirect damage) as well as with cellular constituents such as thiols, glutathione, and proteins, all of which act as scavengers in the conventional radiochemical sense. Molecular oxygen is also involved in the reaction as discussed below. Rather than treating the microscale reaction steps individually, these reaction steps are described using a macroscopic absorption coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varSigma\:}_{a}$$\end{document}. This takes into account all the loss mechanisms that occur at the microscale. The transport of hydroxyl radicals in the nuclear microenvironment can be described using the continuity equation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\partial\:\phi\:/\partial\:t+\nabla\:.\mathbf{J}={S}_{0}-{{\Sigma\:}}_{a}\phi\:$$\end{document}, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\phi\:$$\end{document} is the local density of the radicals, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathbf{J}$$\end{document} is the flux, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{0}$$\end{document} is the volumetric production rate of the radicals, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\Sigma\:}}_{a}$$\end{document} is the macroscopic absorption coefficient that takes into account all the chemical losses. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\Sigma\:}}_{a}$$\end{document} is given as the sum of the contributions of different reaction channels:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varSigma\:}_{a}={{\Sigma\:}}_{\mathrm{D}\mathrm{N}\mathrm{A}}+\sum\:_{i}^{}{k}_{i}{C}_{i}+{k}_{{\mathrm{O}}_{2}}H{p}_{{\mathrm{O}}_{2}}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\Sigma\:}}_{\mathrm{D}\mathrm{N}\mathrm{A}}$$\end{document} is referred to as effective \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document}–DNA interactions, while \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\sum\:_{i}^{}{k}_{i}{C}_{i}$$\end{document} takes into consideration reactions involving non-oxygen scavengers. The last term is associated with oxygen-dependent scavenging and is expressed as the product of reaction kinetic constant \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{k}_{{\mathrm{O}}_{2}}$$\end{document}, Henry’s constant \:H, and oxygen pressure \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{{\mathrm{O}}_{2}}$$\end{document}4,24. All of the reaction rate constants are specified in microscopic form (nm[− 1 through implied conversion from kinetic expression based on a hydroxyl radical transport velocity. It is worth noting that, although molecular oxygen does react with hydroxyl radicals, the rate constant for this reaction is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:1.9\times\:{10}^{9}{\mathrm{M}}^{-1}{\mathrm{s}}^{-1}$$\end{document}, and this is not the main route by which oxygen carries out its radiosensitizing activity in the body. The extremely high concentration of biomolecules in the body and the high-rate constant for the diffusion-controlled reactions involving hydroxyl radicals make the scavenging activity of molecular oxygen on hydroxyl radicals a minor contributor. The term involving the oxygen molecule is still included in the expression for the absorption coefficient for completeness. The biological aspect will be discussed separately. Steady-state approximation and analytical solution The timescale of the chemical reactions of the hydroxyl radicals as well as the diffusion of the radicals in a biological environment is of the order of picoseconds to nanoseconds, while the irradiation time during conventional radiotherapy varies from milliseconds to minutes. This shows that the timescales are separated, indicating that the population of radicals rapidly reaches equilibrium with respect to the slowly varying source term, thereby allowing the quasi-steady-state approximation. It is important to note here that the quasi-steady-state approximation should not be taken as a description of the absence of time dynamics, but rather as a description of a coarse graining of the time dynamics of the concentration of radicals, where all time variations faster than the averaging window are ignored. The balance between production and absorption is dominant, and the time dependence is removed, thereby arriving at a simple equation describing the situation. Such an approach is commonly adopted in analytical radiation chemistry, where the focus is more on the description of the concentrations of the radicals rather than the time evolution of the concentrations of the radicals24,26. Characteristic time and spatial scales for hydroxyl radical sources, diffusion, and loss, following irradiation in nuclear microenvironments, vary over several orders of magnitude. As indicated in Supplementary Table S4, radical source rates are femtosecond processes, while diffusion and loss events continue into the nanosecond time domain for scales ranging from nanometers to submicrometers1,2,4,24,26. This unambiguous separation of time and spatial scales justifies a quasi-steady-state approximation for hydroxyl radicals, allowing us to neglect the explicit time dependence and focus on a balance between the volumetric source term and macroscopic absorption. In this spatially averaged picture, the net outcome of radical production and loss is captured without resolving the detailed spatial transport across the microenvironment. Under the first-order approximation to the Boltzmann equation (see Supplementary Text S5), the angular variation of the flux of radicals is expanded, leading to two coupled moment equations, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\nabla\:.\mathbf{J}+{{\Sigma\:}}_{a}\phi\:={S}_{0}$$\end{document} (conservation of particles) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathbf{J}=-\frac{1}{3{{\Sigma\:}}_{a}}\nabla\:\phi\:$$\end{document} (Fick’s Law). Plugging the second equation into the first equation results in a diffusion equation. Nonetheless, in cases where the characteristic length scale of the spatial gradient (for example, the size of the nucleus) is large compared to the mean free path of absorption, the term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\nabla\:.\mathbf{J}$$\end{document} can be disregarded relative to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\Sigma\:}}_{a}\phi\:$$\end{document}. At such length scales, the conservation equation reduces to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\Sigma\:}}_{a}\phi\:={S}_{0}$$\end{document}, and the particle density is6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\phi\:}_{0}=\frac{{S}_{0}}{{{\Sigma\:}}_{a}}$$\end{document} This equation proves to be the primary analytical result of the spatially averaged reaction-kinetics model. In it, there is a clear and rather direct relationship between radiation quality (through \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{\mathrm{O}\mathrm{H}}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)$$\end{document}), dose rate, oxygenation, and chemical scavenging. It should be emphasized that while the final form of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\phi\:}_{0}={S}_{0}/{{\Sigma\:}}_{a}$$\end{document} may be viewed superficially as a balancing equation between production and absorption, this form is a result of spatial averaging over distances greater than the radical diffusion length. The angular-dependent Boltzmann equation and P₁ approximation (Supplementary Text S5) provide the formal starting point, but after spatial averaging the explicit transport terms are eliminated, and their net contribution is incorporated into the macroscopic absorption coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\Sigma\:}}_{a}$$\end{document}. The approximation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\nabla\:.\mathbf{J}\approx\:0$$\end{document} in the averaged steady state reflects that within a sufficiently large volume element (e.g., the cell nucleus), the net flux of radicals out of this volume will be negligible compared to production and absorption. Thus, the macroscopic absorption coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\Sigma\:}}_{a}$$\end{document} will implicitly include information on transport processes via its dependence on reaction rates and diffusion coefficients26. Therefore, the ultimate expression of the steady state equation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\phi\:}_{0}={S}_{0}/{{\Sigma\:}}_{a}$$\end{document} can be considered as a spatial averaging of the reaction-diffusion balance, not a spatially resolved transport equation, since it does not consider spatial variations on smaller than the diffusion length scale. None of this modeling involves fitting parameters to a set of numbers or simulating a stochastic process. This representation adds a direct functional dependence of radical density on dose rate, LET, and oxygen availability. Putting it in Eq. (6):7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\phi\:}_{0}(\dot{D},\mathrm{L}\mathrm{E}\mathrm{T},{p}_{{\mathrm{O}}_{2}})=\frac{\rho\:\dot{D}{G}_{\mathrm{O}\mathrm{H}}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right){C}_{E}}{{{\Sigma\:}}_{\mathrm{D}\mathrm{N}\mathrm{A}}+{k}_{{\mathrm{O}}_{2}}H{p}_{{\mathrm{O}}_{2}}+\sum\:_{i}^{}{k}_{i}{C}_{i}}$$\end{document} In hypoxic environments (low \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{p}_{{\mathrm{O}}_{2}}$$\end{document}), a possible consequence of a decrease in absorption is an increased lifetime of the radicals and possibly elevated levels at steady state, which might compensate for the decrease in oxygen fixation. LET‑dependent radiochemical source term: parameterization, consistency, and model integration The process of hydroxyl radical production after ionizing radiation is, in essence, a function of the microscopic character of energy deposition in water. This process is best described in terms of LET, which is a measure of both the concentration of ions produced in a particle path and the attendant correlations between radiolytic products. Within our model, a dependence on LET is possible only through the radiochemical source term. The G-value for the hydroxyl radical, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{\mathrm{O}\mathrm{H}}$$\end{document}, is defined as the amount of hydroxyl radicals created per deposited energy and is known to decrease systemically with increasing LET. This is due to an increase in intra-track recombination and overlaps in densely ionizing radiations, resulting in reduced amounts of freely diffusible radicals. In order to describe such a behavior with retention of analytical tractability, the G value’s LET dependence has been represented by a monotonic function of the saturation type:8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{\mathrm{O}\mathrm{H}}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)={G}_{0}{[1+{\left(\frac{\mathrm{L}\mathrm{E}\mathrm{T}}{{\mathrm{L}\mathrm{E}\mathrm{T}}_{c}}\right)}^{\alpha\:}]}^{-1}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{0}$$\end{document} is the reference yield for low LET, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{L}\mathrm{E}\mathrm{T}}_{c}$$\end{document} a characteristic value of the LET, and α a shaping factor of the suppression function at large LET. This functional form reproduces the experimentally observed monotonic decrease in the yields of hydroxyl radicals with increasing LET and the saturation to a reduced level at high LET. It should be noted, however, that in the functional form given by Eq. (8), the yields approach zero asymptotically for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{L}\mathrm{E}\mathrm{T}\to\:\infty\:$$\end{document}, while experimental data and track structure calculations suggest the existence of a small, but non-zero, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} yield even at high density in the dense ion tracks, mainly resulting from the peripheral and low-density regions within the tracks12,18. The functional form is valid within the LET range relevant to therapeutic radiation, i.e., up to about 200–300 keV/µm for protons and light ions, and the model should be used within this LET range. The functional form reproduces the experimental data within the relevant LET range, as demonstrated in Fig. 1(a). The extension of the model to LET values beyond the relevant range should be made cautiously. These phenomenological parameterizations are commonly used in semi-analytical radiochemical calculations to facilitate the transition from numerical simulations on the microscopic level to macroscopic radiation chemistry treatments such as in Surdutovich et al. and Nikjoo et al.7,27. The parameter ranges for the LET‑dependent G‑value are physically motivated as follows. The values of parameters LETc and α are not freely fitted to a specific dataset; instead, they are constrained within physically plausible ranges derived from empirical radiolysis experiments and Monte Carlo simulations. The values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{L}\mathrm{E}\mathrm{T}}_{c}$$\end{document} are given to be in the scale of keV·µm-1, which roughly marks the beginning of the overlap region for heavy charged particles5,9. The value of the exponent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\alpha\:$$\end{document} is expected to be between 1 and 2, depicting a smooth but sharp transition between low- and high-LET regions. Rather, it emphasizes qualitative behavior that persists despite a range of variations within these physically motivated ranges. The key takeaway from such robustness is that the behavior in the regimes described in the Sect. 2 emerges due to structural characteristics of the models. This parameterization shows consistency with published experimental and simulation data. The LET-dependent suppression of hydroxyl radical yields so adopted is consistent with a wide body of experimental and computational literature. Reduced \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{O}\mathrm{H}}^{{\bullet\:}}$$\end{document} yields for high-LET radiations compared to low-LET photons and electrons have been reported consistently in time-resolved pulse radiolysis measurements and track-structure simulations when dose normalization was imposed12,28. This behavior is explicitly reproduced by the Monte Carlo codes, such as PARTRAC and GEANT4-DNA, by using a microscopic modeling of spur formation and radical recombination. The present analytical parameterization captures exactly the same physical effect reduced to a transparent form while not losing consistency with the established data. At the macroscopic scale of this model, the dependence on particle type, energy, and track structure is considered negligible. Though LET is only the unifying descriptor of radiation quality, it also implicitly contains information on particle species and energy in its relation to stopping power. The differences in track structure based on particle type are known to be of minor importance in radiochemical yields at the macroscopic scale of radical production, when averaged over nuclear dimensions9. Consequently, within this study, there is no dependence on particle type incorporated within the current model, aside from LET. This is considered a design choice of the present model, as it intends to predict and provide insight primarily into radiochemical phenomena of a first-order characterization. In its specific application within this scope, LET parameterization is justified. A temporal interpretation of G‑values and effective radical yields is provided here. In this regard, it should be noted that radiochemical G-values are necessarily time-dependent quantities, defined at specific observation times after energy deposition. As for experimental G-values for hydroxyl radicals, these would generally correspond to effective yields on a time scale ranging from picoseconds to nanoseconds after the completion of ultra-fast spur chemistry. In the current formulation, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{\mathrm{O}\mathrm{H}}\left(\mathrm{L}\mathrm{E}\mathrm{T}\right)$$\end{document} can thus be understood as an effective yield of freely diffusible hydroxyl radicals that have escaped the initial recombination events in the track. This is fully consistent with presently used analytical approaches in radiolysis simulations, which avoid mistaking the fast-track physical events for the longer-time scales of chemical dynamics, as pointed out in Pimblott and LaVerne12. The coupling of the source term with chemical loss processes is described in the following. The source-term that is a function of LET is introduced multiplicatively through the volumetric production rate in Eq. (4). Importantly, it is primarily the source-term that is influenced by LET and not the chemical loss coefficients (i.e., the macroscopic absorption terms). This structural decoupling guarantees a transparent interaction between LET-dependent radical productions and oxygen-dependent scavenging reactions, as well as other absorption events, which do not compete in an ad hoc fashion but rather in a well-separated manner, as shown in the Sect. 2, to allow a well-defined radiochemical behavior. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1