Local B-site chemistry controls oxygen-vacancy energetics in Ca-Ce-Ti-Mn perovskites for thermochemical hydrogen production
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.
Kai: I'm Kai.
Mira: And I'm Mira, here with Lev.
Kai: We're looking at the paper: "Local B-site chemistry controls oxygen-vacancy energetics in Ca-Ce-Ti-Mn perovskites for thermochemical hydrogen production".
Mira: Here's the gist of the paper. Local B-site chemistry is identified as the dominant factor controlling oxygen-vacancy formation energy (E v) in Ca–Ce–Ti–Mn (CCTM) perovskites, rather than bulk stoichiometry alone.
Kai: This conclusion stems from a systematic resolution of all fifteen symmetry-distinct oxygen nearest-neighbor environments within the CCTM solid solution and the demonstration that local B-site chemistry dictates these energetics.
Mira: The core finding is quantified by the Crystal-Feature Model (CFM), which approximates E v as a linear combination of two physically motivated descriptors: E v = beta b E b + beta r V r + c (S1), where E b is the average M–O bond-dissociation energy and V r is the average crystal reduction potential.
Kai: The analysis reveals that varying the nearest-neighbor Mn fraction (B site) shifts E v by 1.0–1.5 eV, contingent upon local Ce content, whereas variations in A-site Ce contribute a smaller, Mn-dependent shift of 0.2–0.6 eV.
Paper summary: Kai: So we're looking at this paper, "Local B-site chemistry controls oxygen-vacancy energetics in Ca-Ce-Ti-Mn perovskites for thermochemical hydrogen production". The main idea here is that the local chemistry at the B site—that’s the Mn part—is what really drives how easily oxygen vacancies form, not just the overall ratio of stuff in the whole material.
Mira: Right, so they're saying that instead of just looking at which element you have in total, you need to look closely at how those specific B site cations interact locally to figure out the energy cost of creating an oxygen vacancy. It's about mapping out all fifteen different ways those oxygen atoms can be coordinated around the B site.
Lev: From an engineering standpoint, that makes sense because if you’re building something, you need to know exactly what’s happening at the atomic level when you try to pull electrons out of the material for hydrogen production. It moves it from a simple bulk calculation to something way more specific about structure and composition coupling.
Kai: They use this Crystal-Feature Model, or CFM, which is an approximation. They're using two main things to describe the energy cost of a vacancy: the average M-O bond dissociation energy and the average crystal reduction potential.
Mira: And what they find is that changing the fraction of manganese at that B site shifts that vacancy energy by one point zero to one point five electron volts, depending on how much cerium is around it locally.
Lev: That shift by one to one point five electron volts means you can tune the material's performance just by controlling the Mn content, which is a lot better than having to change the whole bulk formula for that kind of adjustment.
Kai: But they also noted that varying the cerium on the A site only gives a smaller shift, about zero point two to zero point six electron volts. So it’s clearly way more sensitive to what’s happening at the B site than what’s happening at the A site.
Paper summary: Mira: That hierarchy suggests that if you want to tune how fast this material can cycle for hydrogen production, establishing and keeping a specific local order of cations on the B site is a really practical design strategy.
Lev: It sounds like they are pointing toward a path where you don't necessarily have to change the main bulk composition, but instead control that short-range ordering to get better results in terms of redox performance.
Kai: The compositional screening they did pointed toward two specific candidates: CaCeTiMn with a ratio of zero point six seven for calcium, zero point three three for cerium, zero point three three for titanium, and zero point six seven for manganese.
Mira: And another one is CaCeTiMn with a ratio of zero point seven one for calcium, zero point twenty nine for cerium, four-two percent titanium, and five-eight percent manganese. These two seem to be the most promising because they predict cycle capacities that are comparable to or even better than pure ceria at much lower reduction temperatures.
Lev: Lower reduction temperatures mean less energy input for the reactor design, which is a huge practical win for scalability in hydrogen production.
Kai: To back up this local chemistry claim, they used two models: the CFM and a defect graph neural network, or dGNN. The dGNN was used to check for nonlinear structure composition defect coupling that the CFM didn't cover linearly.
Mira: And experimental validation confirmed the findings. They showed that measured values for delta delta, which is related to these vacancy energies, surpassed what their models predicted when they were tested under more strongly reducing protocols.
Lev: I’m interested in the computational checks too because usually those things are tricky at convergence. They recomputed the vacancy energies using a denser Monkhorst Pack mesh on the Gamma relaxed geometries and found only a small residual shift of sixty-two milli-electron volts, which stays well within the predicted range of three point four to three point nine electron volts.
Paper summary: Kai: It’s good that they showed that their calculated oxygen vacancy energies are stable against those sampling convergence issues, meaning the results aren't just artifacts of how much data they fed into the first-principles calculations.
Mira: The way they calculate those average bond energy and reduction potential descriptors is interesting too, because they calculate them as unweighted averages over six first nearest neighbor cations surrounding each oxygen vacancy site, which includes four A site cations and two B site cations.
Lev: That averaging over the specific set of neighbors is what makes the CFM physically motivated because it directly maps to the local environment a defect experiences.
Kai: So, to summarize this paper "Local B-site chemistry controls oxygen-vacancy energetics in Ca-Ce-Ti-Mn perovskites for thermochemical hydrogen production", they found that local B site chemistry dictates oxygen vacancy formation energy instead of just bulk stoichiometry.
Mira: The key finding is the Crystal Feature Model which shows how the average M O bond dissociation energy and the crystal reduction potential combine to define that vacancy energy.
Lev: It’s important because it gives us a specific recipe for tuning these materials to handle oxygen exchange at lower temperatures, which is what we need for real-world applications in hydrogen production.
Kai: The implication is that we can design perovskites by focusing on controlling the local Mn environment rather than just mixing in different amounts of elements overall.
Mira: These specific compositions like CaCeTiMn with a zero point six seven calcium, zero point three three cerium, zero point three three titanium, and zero point six seven manganese are the most promising ones they identified for this work.
Lev: For those of us building hardware, it tells us that if we can engineer that specific B site ordering in our materials, we have a clear target to aim for in terms of improving cycle life at lower thermal stress.
Kai: This research is really pushing the boundary on how we understand defect energetics in complex oxides and how that knowledge translates into practical material design.
Mira: It provides a systematic approach for screening these materials based on what the local chemistry dictates, rather than just relying on simple bulk measurements.
Conclusion: Kai: So we've seen how the specific chemistry at the B site, that Mn part, is really what controls how easily oxygen vacancies form in these Ca-Ce-Ti-Mn perovskites.
Mira: Yeah, and this paper is titled "Local B-site chemistry controls oxygen-vacancy energetics in Ca–Ce–Ti–Mn perovskites for thermochemical hydrogen production."
Lev: It means they're focusing on the local atomic arrangement around the Mn ions instead of just the total mixture of elements in the whole material.
Kai: Exactly. They’re using a model called the Crystal-Feature Model to show how that local environment dictates the energy cost for making an oxygen vacancy, which we call E-v.
Mira: It boils down to looking at two things: how much energy it takes to break those metal oxygen bonds, and some kind of crystal reduction potential.
Lev: And they show that changing the fraction of manganese right there on the B site can shift that vacancy energy by one point zero to one point five electron volts, depending on how much cerium is nearby.
Kai: That's a big deal because it means we can tune the material's performance just by controlling this short-range ordering, without having to change the whole bulk composition.
Mira: They identified two specific compositions that look really promising for things like solar thermochemical hydrogen production at lower temperatures than they used to think was possible.
Lev: Lower reduction temperatures mean less heat input for the reactor, which is a major practical win if we want this stuff to scale up.
Kai: But they also did some rigorous checks with two models, including a graph neural network, to make sure this isn't just an approximation or some artifact from how much data they fed into the first-principles calculations.
Mira: And the experimental validation confirmed their findings because the measured values actually surpassed what their models predicted when they were tested under more reducing conditions.
Lev: That’s good because it means this is a solid mechanism, not just a theoretical exercise on paper.
Kai: So, to sum up, the local chemistry at the B site dictates the vacancy energy in these perovskites, and they've found specific recipes for compositions that should perform better under lower temperatures.
Mira: And this opens up a new way to design materials by focusing on controlling that specific local environment instead of just mixing in different amounts of elements overall.
Lev: It gives us a clear target for how we might engineer these materials to improve their cycle life when they are actually running the hardware.
Kai: We'll look at how this kind of tuning capability applies to other perovskites next.
Department of Chemistry and Institute of Materials Science and Engineering, Washington University in St Louis · ASU LightWorks® and the School for Engineering of Matter, Transport and Energy, Arizona State University · Sandia National Laboratories · Department of Mechanical and Aerospace Engineering, The Ohio State University · School of Earth Sciences, The Ohio State University
cond-mat.mtrl-sci
Submitted: 2026-07-30
Updated: 2026-11-11
Comments: v2: revised after peer review; adds post-cycling XRD and a magnetic-moment test to the SI. 16 pages, 6 figures, 3 tables; SI appended (24 pages, 11 figures, 7 tables), 40 pages total. Manish Kumar and Natalia Ali contributed equally. Data and figure-generation scripts at https://github.com/wexlergroup/cctm-screening. VASP inputs and outputs at https://doi.org/10.17172/nomad.6d7e-hvb1
Code: https://github.com/mwitman1/cgcnndefect
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 90/100
The gist: Local B-site chemistry is identified as the dominant factor controlling oxygen-vacancy formation energy (E v) in Ca–Ce–Ti–Mn (CCTM) perovskites, rather than bulk stoichiometry alone.
Key concepts
- Oxygen Vacancy Formation Energy ($E_v$)
- This is a measure of how much energy is required to create an oxygen vacancy, which is a missing oxygen atom in the crystal structure. Lower $E_v$ generally means it's easier to form these vacancies, which is crucial for the material's redox (reduction/oxidation) activity.
- Crystal-Feature Model (CFM)
- This model approximates $E_v$ by combining two physical descriptors: the average M–O bond-dissociation energy ($\langle E_{b} \rangle$) and the average crystal reduction potential ($\langle V_{r} \rangle$). It provides a simple way to interpret how local atomic environments influence vacancy energetics.
- Local B-site Chemistry
- This refers to the specific arrangement and identity of the B-site cations (like Mn) immediately surrounding an oxygen vacancy. The study demonstrated that varying the fraction of Mn in this immediate neighborhood has a much larger impact on $E_v$ than changes in other parts of the crystal structure.
- Compositional Screening
- This is the process of testing different material compositions to find the best candidates for a specific goal, like high redox performance. The researchers screened various Ca-Ce-Ti-Mn ratios and identified two specific compositions, CCTM2112 and another variant, as having superior predicted cycle capacities.
Terminology
Summary
Local B-site chemistry is identified as the dominant factor controlling oxygen-vacancy formation energy (E v) in Ca–Ce–Ti–Mn (CCTM) perovskites, rather than bulk stoichiometry alone. p. 8 The implications of these complementary biases for fitted-model selection
in composition-space screening are addressed in the Discussion. Discussion
B-site chemistry as the primary control over vacancy energetics
Local B-site chemistry, not bulk stoichiometry alone, dominates
oxygen-vacancy formation energetics in CCTM perovskites. A predictive framework connecting
local cation chemistry to macroscopic redox performance through
interpretable descriptors would enable both rational composition
optimization and transferable design rules for related perovskite
families. The principal advance of this work is the systematic resolution
of all fifteen symmetry-distinct oxygen nearest-neighbor environments in the CCTM solid solution and the demonstration that
local B-site chemistry, rather than bulk stoichiometry, dominates
oxygen-vacancy energetics; the composition-space screening that
follows is an application of this framework rather than its central result.Where the paper says it
Where the paper says it
The core finding is quantified by the Crystal-Feature Model (CFM), which approximates E v as a linear combination of two physically motivated descriptors: E v = beta b E b + beta r V r + c (S1), where E b is the average M–O bond-dissociation energy and V r is the average crystal reduction potential. p. 2 9 eV window is thus accessible via two routes:
high local Mn content with high Ce, or moderate Mn with low Ce. These observations motivate a physics-informed linear model of
the form
Ev = βb⟨Eb⟩+βr⟨Vr⟩+β0, (1)
where ⟨Eb⟩ is the average M–O bond-dissociation energy and ⟨Vr⟩
is the average crystal reduction potential (the reduction potential
of a cation couple evaluated in the oxide crystal rather than in
aqueous solution34), both computed over all six nearest-neighbor
cations (four A-site, two B-site). We develop an
interpretable crystal-feature model (CFM) whose fitted coefficients
directly encode the underlying Born–Haber thermochemistry, and
we benchmark these predictions against a retrained defect graph
neural network (dGNN). This analysis shows that local B-site chemistry dominates vacancy energetics: varying the Mn fraction among nearest-neighbor
B sites shifts Ev by 1. Local B-site chemistry
dominates the oxygen-vacancy formation energy (Ev): varying the nearest-neighbor Mn fraction (B site) shifts Ev by 1. 5 eV depending
on local Ce content, whereas A-site Ce variation contributes a smaller, Mn-dependent shift of 0. 5 eV depending
on local Ce content, whereas A-site Ce variation contributes a smaller, Mn-dependent shift of 0. This hierarchy indicates that
short-range B-site cation order, if it can be established and kinetically retained through targeted processing, is a candidate means of
tuning redox performance without changing bulk composition.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
The compositional design space points towards Ca 0. p. 1 The
interpretable framework provides both compositional screening
capability and mechanistic design rules expected to transfer to
related perovskite families. Results
Compositional design space
Maximizing hydrogen yield in thermochemical water splitting requires oxygen-vacancy formation energies, Ev, that are low enough
to permit thermal reduction yet high enough to drive steam reoxidation, together with a phase-stable host that withstands repeated cycling. 9 eV range established in our earlier thermodynamic
analysis. 2,30 Experimental studies on the reference composition Ca2/3Ce1/3Ti1/3Mn2/3O3 (CCTM2112) confirmed hydrogen
productivities of ∼10 mmolH2/mol atomCCTM2112 per cycle at
1350 ◦C, with rapid and reversible kinetics. 67O3 (CCTM2112) and
Ca0. 58O3 as the most promising candidates
among the six compositions examined (Table 1), with predicted
cycle capacities comparable to or exceeding ceria at substantially
lower reduction temperatures; experimental measurements on
three compositions are consistent with these predictions. Experimental measurements on three CCTM compositions show ∆δ increasing monotonically
with Ce content under cycling protocols close to the model conditions, consistent with the qualitative trend from the fitted models. The
interpretable framework provides both compositional screening capability and mechanistic design rules expected to transfer to related
perovskite families for solar thermochemical hydrogen production.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
To rigorously establish the dominance of local B-site chemistry, the researchers employed a combination of first-principles defect thermochemistry and two complementary models: the CFM, which captures mean thermodynamic trends with transparent physical interpretation, and a finetuned defect graph neural network (dGNN), which models potential nonlinear structure–composition–defect coupling beyond the linear form of the CFM. p. 0 We develop a crystal-feature model
whose fitted coefficients directly encode the underlying Born–Haber thermochemistry, and we fine-tune a defect graph neural network to
model potential nonlinear structure–composition–defect coupling beyond the linear form of the crystal-feature model. Local B-site chemistry
dominates the oxygen-vacancy formation energy (Ev): varying the nearest-neighbor Mn fraction (B site) shifts Ev by 1. 044) falls at the lower edge
of the measured range, the closest of the three; the ordering is
consistent with experimental ∆δ being a lower bound on the equilibrium value, but the magnitude of the SCAN+U–experiment
gap is larger here than at the other two compositions. Under
Protocols A and B, measured ∆δ exceeds all three predictions
for all three compositions, consistent with the more strongly reducing residual-PO2
atmospheres of these protocols falling outside
the model conditions. 9 eV range are expected
to balance high-temperature reduction with steam reoxidation:
the upper bound of the window reflects the requirement that
Ev be low enough for thermal reduction to be accessible, and
the lower bound that it remain high enough to retain a thermodynamic driving force for steam reoxidation, the same competition formalized at finite temperature through the reduction and
water-splitting oxygen chemical potentials (eqn (6) and (8) in
Methods). 2 This 0 K Ev target window is distinct from the explicit
finite-temperature cycle-capacity evaluation, which uses the specific T–PO2
values in Methods. These observations motivate a physics-informed linear model of
the form
Ev = βb⟨Eb⟩+βr⟨Vr⟩+β0, (1)
where ⟨Eb⟩ is the average M–O bond-dissociation energy and ⟨Vr⟩
is the average crystal reduction potential (the reduction potential
of a cation couple evaluated in the oxide crystal rather than in
aqueous solution34), both computed over all six nearest-neighbor
cations (four A-site, two B-site). The fitted coefficients encode
the effective contributions of each term to vacancy formation.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Improved materials screening by using an interpretable framework that connects local cation chemistry to macroscopic redox performance, allowing for
rational composition optimization and transferable design rules for related perovskite families.
-
Ability to predict oxygen-vacancy formation energies Ev across a composition space using both a crystal-feature model (CFM) and a defect graph neural network (dGNN), providing complementary descriptions of mean thermodynamic trends and nonlinear structure–composition–defect coupling.
-
Systematic identification of the most promising compositions for solar thermochemical hydrogen production by mapping them to a Ce/Mn-balanced region where the predicted cycle capacity delta is
comparable to or exceeds the ceria benchmark at substantially lower reduction temperatures.
-
Design rules for processing materials, such as
Layerby-layer thin-film growth
orepitaxial strain,
to tune oxygen vacancy energetics by promoting specific short-range B-site cation order.
Abstract
Two-step thermochemical water splitting driven by concentrated solar heat is a scalable route to renewable hydrogen, but it requires oxides whose oxygen-vacancy formation energies balance facile reduction with favorable reoxidation. Perovskite solid solutions can tune this balance, but the relationship between bulk stoichiometry and local defect energetics remains poorly understood. Here we map oxygen-vacancy formation energetics across Ca-Ce-Ti-Mn (CCTM) perovskites by combining first-principles calculations with a coverage-constrained special quasirandom structure approach that realizes all fifteen symmetry-distinct oxygen nearest-neighbor environments, an interpretable crystal-feature model whose fitted coefficients directly encode the underlying Born-Haber thermochemistry, and a fine-tuned defect graph neural network. Local B-site chemistry dominates the oxygen-vacancy formation energy E v: varying the nearest-neighbor Mn fraction shifts E v by 1.0-1.5 eV depending on local Ce content, whereas A-site Ce variation contributes a smaller, Mn-dependent shift of 0.2-0.6 eV. Short-range B-site cation order, if it can be established and kinetically retained through processing, is therefore a candidate means of tuning redox performance without changing bulk composition. Composition-space maps identify a Ce/Mn-balanced region (29-33% Ce on the A site and 58-67% Mn on the B site) combining a high fraction of vacancy sites within the targeted E v window with phase stability and solubility, whose predicted redox cycle capacity matches or exceeds the ceria benchmark at 1350 C rather than the roughly 1600 C ceria requires. Measurements on three CCTM compositions show cycle capacity increasing monotonically with Ce content under protocols close to the model conditions. The design rules are expected to transfer to related perovskite families.
Sources
Related papers
- Structural contribution to light-induced gap suppression in Ta 2 NiSe 5
- Anatomy of fast current-induced skyrmion motion in synthetic antiferromagnets
- Magnetic control of an exciton-polariton condensate in a van der Waals magnet
- High-pressure elastic properties of GeO2 polymorphs up to 120 GPa
- Decoupling Strain-Rate Sensitivity and Deformation Length Scale Effects in Neutron-Irradiated Tungsten: A Coupled Nano-Indentation, HR-EBSD and Crystal Plasticity Study
- Magnetism modulation in Co3Sn2S2 by current-assisted domain wall motion