Microscopic foundation of size-dependent hydriding phase transformations in palladium nanocrystals: deriving the surface phase preference and building beyond-Ising predictive models

Palladium nanocrystals absorb hydrogen in a first-order phase transformation between a dilute interstitial solid solution ($\alpha$-PdH$_x$) and a concentrated hydride phase ($\beta$-PdH$_x$), and experiments on monodisperse nanocubes show that midpoint (coexistence) pressures, hysteresis widths and transformation kinetics all depend strongly on particle size. A 2013 study explained these trends with a coarse-grained three-dimensional Ising lattice model in which the particle boundary carries a phenomenological energetic preference $\Delta J = J_+ - J_-$ for one phase, and flagged as limitations of that description the merely indicative functional form of $\Delta J(T)$, the omission of explicit hydrogen degrees of freedom and of lattice elasticity, the exclusion of shape- and facet-dependent energetics — in particular [111]-terminated particles below ~10 nm — from its cube-only size comparison, and the possibility that extrinsic perturbations could mask the intrinsic scaling. Work from 2013 to 2026 on each of these facets — single-particle thermodynamics, facet- and vertex-controlled kinetics, atomistic diffusive-molecular-dynamics and first-principles cluster-expansion treatments, and coherent-interface versus solid-state-wetting theory — has resolved substantial parts, but a microscopically derived surface phase preference together with explicit-hydrogen, elasticity- and facet-resolved predictions of midpoint pressures, nucleation barriers and kinetics, covering the shapes and sizes the original comparison excluded, remains open.

Background

Palladium–hydrogen is the prototypical metal–hydride system: dissolved hydrogen atoms occupy interstitial sites of the Pd lattice, forming a dilute $\alpha$ phase at low H concentration and a concentrated $\beta$ phase (PdH$_x$, $x \approx 0.6$–$0.7$) at high concentration. Across the two-phase region the pressure–composition isotherm exhibits a flat plateau, and loading versus unloading plateaus differ (hysteresis) because the transformation between phases requires nucleation — the fluctuation-driven formation of a critical nucleus of the new phase, whose size and free-energy barrier follow classical nucleation theory and diverge at bulk phase coexistence. Nanocrystalline Pd is central to hydrogen storage, sensing and catalysis, but whether its sorption thermodynamics and kinetics are genuinely size-dependent — and intrinsically so, rather than through substrates, ligands or measurement transducers — was long unsettled because experiments averaged over ensembles or perturbed the particles.

Bardhan, Hedges, Pint, Javey, Whitelam and Urban (Nature Materials 12, 905–912, 2013) measured hydriding and dehydriding of monodisperse, ligand- and antenna-free Pd nanocubes of side 14, 32, 65 and 110 nm at 22, 53, 82 and 110 °C using an in situ luminescence probe whose normalized signal directly reports the phase fraction (normalized luminescence 1 = metal, 0 = hydride), with samples equilibrated at each pressure before recording and all profiles normalized by the deposited nanocrystal mass. The reported thermodynamic observable is the midpoint (coexistence) pressure, defined by $\ln P_{\mathrm{mid}} = (\ln P_{\mathrm{hyd}} + \ln P_{\mathrm{dehyd}})/2$ evaluated at luminescence 0.5 and referenced to a bulk Pd film through $\beta\Delta G_{\mathrm{norm}} = \ln(P_{\mathrm{mid}}/P_{\mathrm{mid}}^{\mathrm{film}})$; hysteresis widths are quantified by the loop separation $\ln(P_{\mathrm{hyd}}/P_{\mathrm{dehyd}})$. Smaller cubes showed lower midpoint pressures, narrower hysteresis loops and faster transformations. To interpret this, the authors simulated a three-dimensional Ising lattice model — a statistical-mechanical model of spins on a lattice with nearest-neighbor coupling $J$ and field $h$, here a coarse-grained representation with Hamiltonian $H = -J\sum_{\langle ij\rangle} S_i S_j + h\sum_i S_i$, in which $S_i = +1$ (up) denotes the metal ($\alpha$) phase and $S_i = -1$ (down) the hydride ($\beta$) phase, the scaled field $2h/T$ plays the role of $\ln P$, the scaled magnetization $\rho = (1/2)(1 + N^{-1}\sum_i S_i)$ (1 in pure metal, 0 in pure hydride) is the luminescence proxy, and the finite cube of side $L$ is enclosed by six walls of non-flippable spins. Up- and down-spins adjacent to a wall interact with each nearest-neighbour wall spin with energies $J_+$ and $J_-$ respectively, so $J_+$ couples the wall to the metal ($\alpha$) phase and $J_-$ to the hydride ($\beta$) phase; the surface preference $\Delta J \equiv J_+ - J_-$ ($\Delta J = 0$ meaning inert walls, $\Delta J > 0$ walls favoring the hydride phase) is a phenomenological parameter whose temperature dependence was fixed only qualitatively from the measured size and temperature trends, the simulations using the form $\Delta J \propto T^3$, which the authors described as indicative only. Umbrella sampling (a biasing technique that reconstructs free energies along an order parameter) showed that at/near coexistence the nucleation barrier in a cube of side $L$ is finite and scales as $\Delta E / k_B T \propto c_1 L + c_2 L^2$ with $c_1, c_2 > 0$ (the reported fit is $\Delta G = -2.17 + 1.64\,L + 1.13\,L^2$ in units $k_B T = 1$, with $\Delta G = G(N) - G(1)$), consistent with a simple scaling argument in which the critical nucleus is a two-dimensional film of side $x$ on an interior face of the cube, with $\Delta E(x) = 2x(\sigma + \Delta J) + x^2(\sigma + \Delta J - 2h)$ maximized at $x = L$ near coexistence; far from coexistence the barrier becomes $L$-independent. Near the phase boundary the critical nucleus is therefore comparable in size to the nanocrystal itself: 'thermal nanoconfinement' makes transformation rates unavoidably size-governed. X-ray diffraction showed no noticeable size-dependent strain trend (strain increases of 18.76–36.63% upon hydriding for the 14–110 nm cubes, versus 118–221% for a substrate-clamped 65 nm film), supporting an intrinsic, thermally driven mechanism rather than strain-induced transformation.

Beyond these results, the paper makes several separate, differently-scoped statements about what its framework does not establish. First, the functional form of $\Delta J(T)$ used in the simulations ($\Delta J \propto T^3$) was 'indicative only', without microscopic justification. Second, 'the Ising model is a prototypical model of phase change, and does not resolve details of the system in question': it resolves neither individual hydrogen molecules and their interactions, nor lattice elasticity (misfit strain and coherent elastic interactions between phases); how these details modify midpoint pressures, nucleation barriers and kinetics was left unquantified. (The paper did verify that, in the subcritical nucleation-dominated regime it probed, long-range hydrogen–hydrogen interactions of the kind implicated near the bulk critical point — represented by an added long-range mean-field elastic term — do not qualitatively change its conclusions, and it noted that near-critical analysis is classically established; near-critical long-range interactions therefore stand as a consistency requirement for any successor model rather than as an effect the 2013 study left unexamined.) Third, the size comparison covered {100}-facetted cubes only: at sizes below ~10 nm Pd typically attains a spherical shape terminated by [111] planes, which have very different H$_2$ dissociation and diffusion energetics, and such particles were explicitly not compared, so the effect of facet-specific energetics on the size-scaling laws in that regime was unexplored (a smaller-size luminescence study was noted as underway). Fourth, as a caveat on the paper's own claim, extrinsic perturbations (defects, strain clamping, surface ligands, engineered plasmonic structures) 'could mask the scaling behaviour uncovered here', because the associated energy scales can far exceed the thermal energy that drives intrinsic phase transformations. Together these point to one research direction: a microscopically grounded, quantitative theory of hydriding phase transformations in nanocrystals as a function of size, shape and temperature, replacing the coarse-grained description and validated against equilibrated pressure–composition isotherm measurements of the kind the 2013 study performed.

Problem Statement

The problem is to replace the phenomenological coarse-graining of hydriding phase transformations in Pd nanocrystals used by the 2013 study with a microscopically grounded, quantitative description addressing the gaps that study itself identified: (1) derive from molecular-scale physics (surface and interface thermodynamics of the $\alpha$ and $\beta$ phases) the energetic preference of the nanocrystal boundary for one phase — the quantity encoded as $\Delta J(T)$ in the 2013 Ising representation — including its quantitative temperature dependence, rather than assuming an indicative functional form such as $\Delta J \propto T^3$; (2) represent the physical details the 2013 model omits — explicit hydrogen degrees of freedom and their interactions, and lattice elasticity (misfit strain and coherent elastic interactions between phases), consistently with the source's own verified finding that a long-range mean-field elastic term does not qualitatively change the subcritical, nucleation-dominated picture — in models that predict quantitatively the midpoint (coexistence) pressures, nucleation free-energy barriers and transformation rates beyond the two-phase Ising description; (3) determine how shape- and facet-dependent behavior alters the intrinsic size-scaling relations, covering the regime the cube-only comparison excluded — in particular [111]-terminated particles below roughly 10 nm, whose H$_2$ dissociation and diffusion energetics differ strongly from those of {100} facets; and (4) ensure that predictions remain distinguishable from extrinsic perturbations (surface ligands, substrate clamping, defects, engineered plasmonic structures), whose energy scales can far exceed the thermal energy driving intrinsic transformations and which could therefore mask or mimic the intrinsic size-scaling laws. The scientific target is an account in which the surface preference $\Delta J(T)$ is derived from Pd–H surface and interface thermodynamics rather than fitted to an indicative form, explicit-hydrogen and elastic physics are represented, and predictions of midpoint pressures, hysteresis widths, nucleation barriers and transformation rates as functions of size, shape and temperature agree with equilibrated pressure–composition isotherms of the kind the 2013 study performed — measurements on monodisperse, ligand- and antenna-free Pd nanocrystals referenced to a bulk Pd film, which the source argues are representative of individual cubes — and hold across the shapes and sizes, including the [111]-terminated sub-10 nm regime, that the original comparison left outside its scope. The four items (1)–(4) are jointly required. For the equilibrium comparison, 'agree' means that predicted midpoint pressures and hysteresis widths match the source's equilibrated isotherms over its measured cube sizes and temperatures to within the experimental scatter of those measurements, with no surface-preference parameter fitted to the same isotherms it is tested against; nucleation-barrier and transformation-rate predictions must follow the corresponding measured size- and temperature-dependent trends. For the concrete excluded regime — [111]-terminated (near-spherical) particles below roughly 10 nm — the account must give a facet-resolved prediction stating how replacing {100} facets by [111] facets alters the size-scaling relations relative to the cube-only case; since no equilibrated isotherms exist for that regime, the prediction must be consistent with the trend the model gives for the measured cubes. For item (4), the account must show that the intrinsic size-scaling signature is distinguishable from the signature obtained when a named extrinsic perturbation (ligand, clamping, defect or engineered plasmonic structure) is present at the same size, shape and temperature — through a predicted, checkable difference between the two signatures, or through an identified size/temperature window in which the intrinsic scaling dominates the extrinsic energy scales.

References

1. Rizia Bardhan, Lester O. Hedges, Cary L. Pint, Ali Javey, Stephen Whitelam, Jeffrey J. Urban, 'Uncovering the intrinsic size dependence of hydriding phase transformations in nanocrystals', Nature Materials 12, 905–912 (2013). DOI: 10.1038/nmat3716, https://doi.org/10.1038/nmat3716
2. Svetlana Syrenova, Carl Wadell, Ferry A. A. Nugroho, Tina A. Gschneidtner, Yuri A. Diaz Fernandez, Giammarco Nalin, Dominika Świtlik, Fredrik Westerlund, Tomasz J. Antosiewicz, Vladimir P. Zhdanov, Kasper Moth-Poulsen, Christoph Langhammer, 'Hydride formation thermodynamics and hysteresis in individual Pd nanocrystals with different size and shape', Nature Materials 14, 1236–1244 (2015). DOI: 10.1038/nmat4409, https://doi.org/10.1038/nmat4409
3. Ronald Griessen, Nikolai Strohfeldt, Harald Giessen, 'Thermodynamics of the hybrid interaction of hydrogen with palladium nanoparticles', Nature Materials 15, 311–317 (2016). DOI: 10.1038/nmat4480, https://doi.org/10.1038/nmat4480
4. Katherine Sytwu, Fariah Hayee, Tarun C. Narayan, Ai Leen Koh, Robert Sinclair, Jennifer A. Dionne, 'Visualizing Facet-Dependent Hydrogenation Dynamics in Individual Palladium Nanoparticles', Nano Letters 18, 5357–5363 (2018). DOI: 10.1021/acs.nanolett.8b00736, https://doi.org/10.1021/acs.nanolett.8b00736
5. Noah J. J. Johnson, Brian Lam, Benjamin P. MacLeod, Rebecca S. Sherbo, Marta Moreno-Gonzalez, David K. Fork, Curtis P. Berlinguette, 'Facets and vertices regulate hydrogen uptake and release in palladium nanocrystals', Nature Materials 18, 454–458 (2019). DOI: 10.1038/s41563-019-0308-5, https://doi.org/10.1038/s41563-019-0308-5
6. Noah J. J. Johnson, Brian Lam, Rebecca S. Sherbo, David K. Fork, Curtis P. Berlinguette, 'Ligands Affect Hydrogen Absorption and Desorption by Palladium Nanoparticles', Chemistry of Materials 31 (21), 8679–8684 (2019). DOI: 10.1021/acs.chemmater.9b02193, https://doi.org/10.1021/acs.chemmater.9b02193
7. X. Sun, M. P. Ariza, M. Ortiz, K. G. Wang, 'Atomistic modeling and analysis of hydride phase transformation in palladium nanoparticles', Journal of the Mechanics and Physics of Solids 125, 360–383 (2019). DOI: 10.1016/j.jmps.2019.01.006, https://doi.org/10.1016/j.jmps.2019.01.006
8. Xingsheng Sun, Pilar Ariza, Michael Ortiz, Kevin G. Wang, 'Long-term atomistic simulation of hydrogen absorption in palladium nanocubes using a diffusive molecular dynamics method', International Journal of Hydrogen Energy 43 (11), 5657–5667 (2018). DOI: 10.1016/j.ijhydene.2018.01.169, https://doi.org/10.1016/j.ijhydene.2018.01.169
9. Xingsheng Sun, Rong Jin, 'Size and shape dependence of hydrogen-induced phase transformation and sorption hysteresis in palladium nanoparticles', Modelling and Simulation in Materials Science and Engineering 32, 085012 (2024). DOI: 10.1088/1361-651X/ad89e3, https://doi.org/10.1088/1361-651X/ad89e3 (preprint arXiv:2405.08929)
10. J. Magnus Rahm, Joakim Löfgren, Paul Erhart, 'Quantitative predictions of thermodynamic hysteresis: Temperature-dependent character of the phase transition in Pd–H', Acta Materialia 227, 117697 (2022). DOI: 10.1016/j.actamat.2022.117697, https://doi.org/10.1016/j.actamat.2022.117697 (preprint arXiv:2110.07883)
11. Yong Li, Jörg Weissmüller, 'Size-dependent phase change in energy storage materials: Comparing the impact of solid-state wetting and of coherency stress', The Journal of Chemical Physics 162, 024704 (2025). DOI: 10.1063/5.0247515, https://doi.org/10.1063/5.0247515
12. Jörg Weissmüller, 'Coherent Phase Change in Interstitial Solutions: A Hierarchy of Instabilities', Advanced Science 11 (21), 2308554 (2024). DOI: 10.1002/advs.202308554, https://doi.org/10.1002/advs.202308554
13. A. Ulvestad, M. J. Welland, W. Cha, Y. Liu, J. W. Kim, R. Harder, E. Maxey, J. N. Clark, M. J. Highland, H. You, P. Zapol, S. O. Hruszkewycz, G. B. Stephenson, 'Three-dimensional imaging of dislocation dynamics during the hydriding phase transformation', Nature Materials 16, 565–571 (2017). DOI: 10.1038/nmat4842, https://doi.org/10.1038/nmat4842
14. Daewon Lee, Sam Oaks-Leaf, Hyeonjong Ma, Jianlong He, Zhiqi Wang, Yifeng Shi, Eonhyoung Ahn, Karen C. Bustillo, Chengyu Song, Stephanie M. Ribot, Rohan Dhall, Colin Ophus, Mark Asta, Jiwoong Yang, Younan Xia, David T. Limmer, Haimei Zheng, 'Nanocrystal Geometry Governs Phase Transformation Pathways in Palladium Hydride', ACS Nano 20 (21), 15266–15278 (2026). DOI: 10.1021/acsnano.6c01302, https://doi.org/10.1021/acsnano.6c01302 (preprint arXiv:2601.00093)
15. Daewon Lee, Sam Oaks-Leaf, Sophia B. Betzler, Yifeng Shi, Siyu Zhou, Colin Ophus, Lin-Wang Wang, Mark Asta, Younan Xia, David T. Limmer, Haimei Zheng, 'Atomic Evolution of Hydrogen Intercalation Wave Dynamics in Palladium Nanocrystals Revealed by Liquid-Phase Transmission Electron Microscopy', Journal of the American Chemical Society 148 (25), 25829–25841 (2026). DOI: 10.1021/jacs.6c03937, https://doi.org/10.1021/jacs.6c03937 (preprint arXiv:2404.02416)
16. Lucy J. T. Metzroth, Elisa M. Miller, Andrew G. Norman, Sadegh Yazdi, Gerard Michael Carroll, 'Accelerating Hydrogen Absorption and Desorption Rates in Palladium Nanocubes with an Ultrathin Surface Modification', Nano Letters 21 (21), 9131–9137 (2021). DOI: 10.1021/acs.nanolett.1c02903, https://doi.org/10.1021/acs.nanolett.1c02903
17. Xingsheng Sun, 'Exploring Solute-Defect Interactions in Nanosized Palladium Hydrides across Multiple Time Scales', arXiv:2310.03177 (2023). https://arxiv.org/abs/2310.03177
18. Arnaud Viola, Ivan Saldan, Sousa Javan Nikkhah, Muhammad Umer, Apinya Ngoipala, Matthias Vandichel, Jan M. Macak, Frédéric Maillard, 'Influence of the Size and Shape of Palladium Nanoparticles on Their Electrochemical Hydrogen Sorption Capacity', Journal of the American Chemical Society 148 (9), 9548–9562 (2026). DOI: 10.1021/jacs.5c19692, https://doi.org/10.1021/jacs.5c19692