Ultra-slow aging dynamics of glass and its application to geological dating Authors: Yong Zhao, Huaping Zhang, Liming Xu, Ziqiang Chen, Rui Zhao, Bo Zhang, Haiyang Bai, Wei-Hua Wang Published: 2025 ABSTRACT AbstractThe ultra-slow relaxation dynamics of glasses at ambient temperature provide a promising alternative for dating glasses with extremely low isotopic content that cannot be dated using traditional radiometric methods. However, these ultra-slow, nonlinear aging dynamics remain poorly understood due to the lack of accurate theoretical models and long-term experimental validation. Existing equilibrium-based dynamics models substantially overestimate relaxation times at temperatures far below the glass transition temperature, making it difficult to model and quantify non-equilibrium aging over geological timescales. We address this challenge by formulating an empirical equation that quantifies the non-equilibrium effective relaxation time (τeff) for various glasses, including metallic glasses, organic amber, and lunar glasses. Our findings demonstrate a universal nonlinear aging dynamics governed by a single τeff, which follows a robust empirical relation parameterized by aging temperature and material-specific fragility. Employing this equation, we propose a universal glass kinetics dating method, conceptually analogous to radioactive decay, where τeff serves as a material-specific decay constant. This approach enables dating of glassy materials over timescales spanning decades to billions of years. This work bridges a fundamental gap in glass aging theory and establishes a practical framework for dating geological and planetary glasses.Subject terms: Glasses, Structure of solids and liquids, Nonlinear phenomena Introduction Glasses, although macroscopically solid and apparently static, are fundamentally out-of-equilibrium systems that evolve continuously over time1–3. This structural evolution, known as physical aging, arises from an extremely slow relaxation of the disordered structure toward lower energy configurations. At temperatures near the glass transition temperature (Tg), the aging process has been extensively studied, with clear manifestations such as enthalpy recovery4,5, volume relaxation6, and mechanical stiffening7,8. However, at temperatures far below Tg—such as ambient or geological conditions—the dynamics of aging slow so dramatically that relaxation becomes effectively undetectable within conventional experimental timeframes. Yet mounting evidence challenges the long-held assumption that glasses are entirely “frozen” in this regime. Observable aging effects have been reported over timescales ranging from decades to billions of years, in systems including metallic glass (MG)5, polymer9–11, and silicate glasses12,13. This recognition raises a central question in non-equilibrium physics: how can we quantitatively describe and predict the dynamics of disordered systems evolving over geological timescales? One of the most remarkable features of glass aging dynamics is its inherent nonlinearity8, 14, 15. In equilibrium supercooled liquids, the relaxation behavior in response to perturbations (e.g., subtle temperature jump) can be well described by a stretched exponential Kohlrausch-Williams-Watts (KWW) function16, 17: ϕt=exp−t/τeqβ, where ϕt=pt−p∞/p0−p∞ is the normalized relaxation function, and p(t) is the measured quantity (e.g., enthalpy, density, modulus). And τeq here is the equilibrium relaxation time which does not depend on aging time. Unlike equilibrium supercooled liquids, non-equilibrium glasses exhibit progressively slower and history-dependent relaxation. To describe this nonlinear aging dynamics, the concept of material time is introduced14,15,18,19, and a nonlinear integral function is required8: where ξt=∫0tdt1/τneqt1 is the dimensionless material time, and τneq is the instantaneous relaxation time which is not fixed but evolves continuously with aging time. This nonlinear integral form introduces substantial complexity, as the time-dependent nature of τneq(t) prevents a closed-form analytical solution, requiring numerical integration or approximation methods. This mathematical complexity reflects a deeper conceptual challenge: accurately capturing the nonlinear, time-dependent nature of τneq(t) in theoretical models. However, most widely used models—including the Arrhenius, Vogel-Fulcher-Tammann (VFT)20–22, Parabolic23, and MYEGA relations24—were developed to capture equilibrium or near-equilibrium dynamics. When extrapolated to deep sub-Tg regimes, they often yield vastly overestimated relaxation times, in some cases even exceeding the age of the universe. Even frameworks explicitly designed for non-equilibrium behavior, such as the Narayanaswamy-Moynihan (NM) model18,25, treat relaxation as quasi-equilibrium and fail to fully account for the nonlinear growth of τneq(t). As a result, theoretical predictions consistently overestimate effective relaxation times by several orders of magnitude compared to experimental observations. For example, long-term studies on MGs, polymers, and oxide glasses—including those by Duan et al.26, Zhao et al.5, Perez-De Eulate et al.9, and Corning Incorporated12—have shown that the measured relaxation times are several orders of magnitude shorter than NM-based predictions. These discrepancies highlight a critical limitation in current theoretical frameworks: the failure to completely incorporate nonlinear effects that dominate ultra-slow aging behavior. This fundamental knowledge gap in glass physics has profound implications for Earth and planetary sciences. Natural glasses, such as obsidian, tektites, organic amber, and lunar glasses, are widespread in geological and extraterrestrial environments. They are often the only available amorphous records of thermal, impact, or volcanic events. However, some lunar glasses contain K or Pb at concentrations too low to produce measurable isotope signals27–30, while amber contains 14C but its short half-life (~5730 years)31 restricts dating to at most several tens of thousands of years, far shorter than the millions of years typically required. As a result, glasses with extremely low isotope abundances cannot be reliably dated using traditional radiometric techniques, which calls for the development of alternative dating methods. The continuous evolution of physical properties in these aging glasses suggests an alternative: a kinetics-based dating approach that estimates age through intrinsic structural relaxation dynamics rather than isotopic decay. Such a method could unlock the chronology of otherwise undatable glassy materials, providing new tools for reconstructing geological history, tracing planetary surface processes, and interpreting extraterrestrial sample returns. In this study, we address these challenges by formulating an empirical equation that quantifies the effective relaxation time (τeff) of glasses in non-equilibrium aging regimes at deep sub-Tg temperatures. Through systematic analysis of diverse glassy materials—including Ce-based MG, fossilized amber, and lunar glasses—we demonstrate that their nonlinear relaxation dynamics can be unified through a single τeff, governed by aging temperature and material-specific fragility. The resulting empirical relation captures essential features of ultra-slow aging across multiple classes of glasses and reveals a universal behavior that serves as a kinetic fingerprint. Based on this framework, we propose a glass kinetics dating method, conceptually analogous to radioactive decay, in which τeff serves as a material-specific decay constant. This approach provides a quantitative chronometer for glassy materials, applicable across timescales spanning decades to billions of years. Results Glass kinetics decay method: nonlinear features of aging The introduction of material time is crucial for understanding and analyzing the aging dynamics of glass14,15,18,19. In the special case where τneq(t) is constant, the material time ξ(t) scales linearly with the real time t, i.e., ξ(t)∝t. However, during aging process, τneq(t) typically increases with t, hence ξ(t) is nonlinear with t. Struik32 systematically studied the change of creep relaxation time τneq(t) with aging time t in glasses, and observed a power-law relationship between them, i.e., τneqt∝tμ, where μ is the Struik shift factor. In this case, the material time is related to the aging time by a power law, i.e., ξ(t)∝t1−μ. And the overall relaxation function still remains a KWW form, i.e., where βeff = β(1−μ) is the apparent stretching exponent, τeff corresponds to the time when ϕ(t) decays to e-1, which represents a characteristic time scale of glass aging. In such case, the material time can be reconstructed as ξt=t/τeff1−μ, and the Struik relation can be deduced as τneqt=tμ×τeff1−μ/(1−μ). The Eq. (2) suggests that even in nonlinear aging regimes, if the Struik relation is valid, the relaxation dynamics can still be described using a single effective time scale and a modified exponent. Even in systems where the Struik power-law relation does not strictly apply, the Eq. (2) can still provide an empirical approximation over finite time windows33–35. Consequently, the KWW function is generally applicable to relaxation in diverse glass systems, and even in other many-body interacting systems36,37. The broad validation of the KWW equation in modeling the aging dynamics of various glasses can be evidenced from the data collected in Fig. 1a. Metallic glasses such as Ce-based5, Pd-based6, and Zr-based7 glasses, as well as polymer glasses like poly(vinyl-chloride) (PVC)8, polystyrene (PS85K)38, polycarbonate (PC)38, and poly(methyl methacrylate) (PMMA)38, and silicate glasses12,39 have been shown to follow the KWW equation closely during their aging processes. In all cases, the initial glassy states were prepared by quenching from their supercooled liquids, and the as-cast samples were subsequently annealed isothermally below Tg. The temporal evolution of their physical properties, such as recovered enthalpy, hardness, creep modulus, fictive temperature, and strain, was then monitored during the approach toward equilibrium. Despite the different thermal histories and the diversity of measured properties, the KWW equation can accurately capture the kinetics of these changes, allowing for the determination of the non-equilibrium effective relaxation time τeff and the stretching exponent βeff. In Fig. 1b, by normalizing the aging time with τeff and βeff, all the decay curves can collapse into a single master curve. This collapse strongly supports the universality of the KWW equation in modeling the non-equilibrium aging kinetics of a wide range of glassy systems, irrespective of their composition or physical properties. Universality of the KWW function in describing the relaxation dynamics of various glassy materials. a Relaxation behaviors of various glassy systems are well captured by the KWW equation. These systems include metallic glasses (Ce-based5, Pd-based6, Zr-based7), polymer glasses (PVC8, PS85K38, PC38, PMMA38), and inorganic silica glasses12,39. The relaxation dynamics are extracted from diverse physical quantities, including recovered enthalpy, hardness, creep modulus, fictive temperature, and strain, measured under different aging protocols (the fitted values of τeff and βeff are listed in Supplementary Table 1). b By reducing the aging time with the non-equilibrium effective relaxation time τeff and the βeff index obtained in (a), the variations in different physical properties of various types of glass can be unified into a single master curve, demonstrating the effectiveness of the KWW empirical formula in describing the kinetic decay of relaxation. Empirical equation for the decay of non-equilibrium glasses To characterize the temperature dependence of the effective relaxation time (τeff) in aging glasses, we begin by reviewing existing theoretical models. Numerous phenomenological equations have been proposed to describe the evolution of equilibrium relaxation time τeq with temperature in supercooled liquids. The simplest is the Arrhenius equation τeq=τ0expEa/RT, where τ0 is the prefactor and Ea is the apparent activation energy. The Arrhenius equation only works for several strong glass-forming systems (like silica), but fails for most other systems where apparent activation energy Ea varies with temperature. To address this, various non-Arrhenius models, such as the VFT Ea=BRT/T−T020–22, parabolic Ea=CRTT0/T−1223, and MYEGA Ea=KRexpC/T24 have been developed. However, these models20–24 are only valid for equilibrium supercooled liquids and break down in the non-equilibrium glassy state. A more general framework for modeling non-equilibrium glass involves the concept of fictive temperature (Tf), defined as the effective thermodynamic temperature at which a glass’s instantaneous properties match those of the equilibrium liquid40. During aging, Tf continuously decreases from the temperature that glass is frozen (Tg) toward the ambient temperature (T). The Narayanaswamy–Moynihan (NM) model18, 25 relates the growth of the instantaneous non-equilibrium relaxation time τneq(t) to the evolution of Tf(t), as: where Eg is the activation energy around Tg, x is the weighting factor. And the τneq(t) in turn determines the decaying rate of fictive temperature Tf(t) and related physical properties p(t), i.e., Due to the inherent complexity of nonlinearity, obtaining an exact analytical solution for these equations is challenging. To make analytical progress, we provide an approximating solution under two assumptions: 1) the τeff is defined as the time when relaxation function decays to e−1 (i.e., ϕτeff=e−1); 2) a simple KWW function can be used to approximate relaxation function ϕ(t) near t=τeff, i.e. ϕ(t)=exp−t/τeffβeff with βeff = β(1−μ). Consequently, the non-equilibrium relaxation time τneq(t) near t=τeff could be deduced as a power law relationship τneqt=tμτeff1−μ/(1−μ). Combining these simplified equations with Eq. (3) and Eq. (4) yields the following equation: where Tft=T+Tg−Texp−t/τeffβeff. In such case, an analytical solution for τeff can be deduced as τeff=1−μτ0expxEg/RT+1−xEg/RTn, where Tn=Tfτeff=T+Tg−Te−1 is the fictive temperature at t=τeff, and the parameter 1−μ=1+1−xyT−1 with yT=βEgTn−T/RTn2. Considering that Eg is connected to fragility index m (i.e., Eg=mln10RTg) and that τg=τ0expEg/RTg, the non-equilibrium effective relaxation time of glass τeff can be reconstructed as with Tn=T+Tg−Te−1 and yT=βmln10TgTn−T/Tn2. The upper limit of the τeff is related to the case of x = 1, i.e., logτeffup/τg=mTg/T−1, which is a simple Arrhenius equation. Whereas the lower limit of τeff is related to the case of x = 0, i.e., logτefflow/τg=mTg/Tn−1−log1+yT. The lower limit of τeff has a limited value even at zero temperature, i.e., logτefflowT=0/τg=me−1−log1+eβmln10