TICC 2026 · Invited Talk
Subsurface Control of Active-Site Distributions in Pt-Skin HEA Electrocatalysts
How buried atoms tune — and help predict — hydrogen adsorption across 96 distinct sites
96 DFT H sites
3 SQS Pt-skin slabs
≈0 ΔG H* (eV)
5 elements (HEA)
H*
Pt skin
HEA subsurface
Chen-Cheng Liao (廖振成) Department of Chemistry, Fu Jen Catholic University, New Taipei City, Taiwan DFT & machine learning · 165804@mail.fju.edu.tw
HER design starts from hydrogen adsorption
The free energy of adsorbed H, ΔG H* , is the central catalytic descriptor — too strong or too weak both lose; near zero wins
Key message HER catalyst design reduces to one goal: engineer a surface with near-optimal H binding .
Pt is the benchmark — but not the endpoint
Pt binds H near-optimally, yet it is costly and offers little tunability — motivating Pt-skin alloys
Key message Keep a Pt-like surface , let buried atoms tune it , and use less Pt .
HEA catalysis is not a single-site problem
Even under a pure-Pt skin, each hollow sits above a different subsurface — so adsorption becomes a distribution
Key message On a Pt-skin HEA, activity is a distribution of sites — an active-site population problem.
Pt-skin: a controlled platform for the buried effect
The adsorbate always meets a Pt-rich surface; the variation comes from the composition underneath
Key message The Pt skin isolates one clean question: how does the buried environment tune surface Pt ?
The central question of this work
Can local composition and coordinates predict hollow-site H adsorption — turning DFT into calibration data?
Key message If descriptors predict adsorption, DFT becomes calibration data , not an endpoint census.
Scope: what this work is — and is not
A controlled hollow-site calibration study and predictor framework — not a full search, kinetics, or MLP
Key message A descriptor-based local adsorption-energy predictor — the seed for active-site population estimation.
06
Methods overview
Fe–Co–Ni–Cu–Pt Pt-skin (111): from random structure to descriptors
1
SQS bulk
random alloy in a finite cell
2
Pt-skin slab
4×4×6 · 96 atoms · pure-Pt top
3
96 hollow sites
all FCC + HCP, 3 slabs
5
Descriptors
local environment · d-band · surrogate prediction
DFT settings
VASP · GGA-PBE
Free energy
ΔG H* = ΔE ads + 0.24 eV
computational hydrogen electrode
SQS gives finite periodic slabs whose local statistics approximate a random alloy.
Interactive SQS explanation →
Key message A controlled 96-site census across three Pt-skin HEA surfaces.
06 / 18
What is a special quasirandom structure (SQS)?
A small periodic cell built to imitate an infinite random alloy
An SQS reproduces the local atomic statistics of a random alloy — in a cell small enough for DFT.
1
The problem
A real random alloy is effectively infinite — it cannot enter a DFT calculation directly.
→
2
The constraint
DFT needs a finite, periodic cell — whatever it contains repeats forever.
→
3
The trick
Choose the arrangement whose neighbour statistics match a random alloy.
Key message An SQS is statistically random — not merely random-looking.
Choosing an SQS: match the neighbours
Same composition, different arrangement — pick the one closest to random
Count each atom's neighbours: the fractions should match the bulk composition.
Minimise the mismatch — the Warren–Cowley parameter α → 0 — over the nearest shells.
See it live the interactive companion walks through this step by step.
Interactive SQS explanation →
Key message The SQS is chosen by neighbour statistics , not by eye.
Our model: a Pt-skin (111) high-entropy slab
Top view and side view of the 4×4×6 slab — 96 atoms, six layers
Top view — Pt skin (16 surface Pt)
H
hollow
4×4 surface cell · FCC + HCP 3-fold hollow sites (96 in total)
Side view — 6 layers
vacuum
H
Pt skin
HEA
subsurface
Pt skin on top · Fe–Co–Ni–Cu–Pt below (top 3 relaxed, bottom 3 fixed)
Key message A pure-Pt skin over the HEA subsurface — the surface looks like Pt; the layers beneath do the tuning .
What we count: the 6 atoms beneath a hollow site
An x-ray view from the Pt skin into the subsurface ring
Pt
Fe
Cu
Co
Ni
Pt
H
An H atom adsorbs in a 3-fold hollow site on the Pt skin.
Make the skin transparent: directly beneath sit 6 subsurface atoms — the local ring.
Descriptor ls_X = counts in that ring → here Pt 2 · Fe 1 · Co 1 · Ni 1 · Cu 1.
Key message Each site's descriptor is the element makeup of the 6 atoms under its hollow .
07
The site population sits on the Sabatier optimum
ΔG H* across all 96 sites — tight and near-thermoneutral
Figure 1. ΔG H* distribution over 96 FCC + HCP hollow sites (3 SQS slabs); Pt(111) reference −0.09 eV.
+0.013 eV
mean ΔG H* — essentially thermoneutral
0.055 eV
standard deviation — a narrow spread
92 %
of sites within |ΔG H* | ≤ 0.10 eV
Key message The Pt-skin HEA naturally creates a near-optimal population of H-binding sites.
07 / 18
08
A trap hides in the simplest analysis
The subsurface counts are not independent — they add to a fixed total
site
6 subsurface neighbours — fixed total
Add one Fe and you must remove another element. The five counts are compositionally coupled .
Caution A raw correlation for one element absorbs everyone else's effect — correlation, not cause .
The fix is partial correlation : hold the other counts fixed, then ask what each element does.
Key message In a constrained alloy, naïve correlation can point at the wrong element .
08 / 18
09
The apparent ranking is misleading
Raw Pearson correlation of each subsurface count with ΔG H* (n = 96)
Apparent ranking (raw r vs ΔG H* )
ls_Pt −0.64
ls_Fe +0.54
ls_Cu −0.22
ls_Co +0.20
ls_Ni +0.02
Bars left of centre = strengthen H binding · right = weaken. Fe is flagged.
ls_Pt is genuinely the strongest signal (r = −0.64, p < 10−12 ).
Trap Fe looks important (+0.54) — but this is mostly Pt's complement : "few Pt here," not "Fe acting."
Cu and Co look weak here — only a method that removes the coupling can tell.
Key message At face value Fe rivals Pt — a classic collinearity artifact .
09 / 18
10
The corrected hierarchy follows chemistry
Partial correlation — after controlling for the other counts
Controlling for the other counts unmasks the true ranking
Fe collapses from apparent hero to weakest; Cu rises to second. The order tracks electronegativity (χ) , not arithmetic.
More-electronegative subsurface neighbours pull charge from surface Pt → weaker H binding.
Key message A stable, electronegativity-ordered descriptor hierarchy: Pt > Cu > Ni ≈ Co > Fe.
10 / 18
11
Subsurface Pt is the dominant handle
More Pt beneath the skin → weaker H binding, ΔG H* rising through zero
Figure 3. ΔG H* vs subsurface Pt count; colour = local Fe count. Linear fit r = −0.64.
−0.64 r
monotonic trend in subsurface Pt count
0 → 3 subsurface Pt walks a site's ΔG H* from positive toward and past the optimum.
Sign convention: higher ΔG H* = weaker H binding.
Key message One countable quantity — subsurface Pt — predicts which way H binding moves.
11 / 18
12
The electronics confirm the mechanism
A site-projected d-band-center effect, measured atom by atom
Figure 6. (a) ΔG H* vs surface-Pt d-band center (r = −0.86). (b) d-projected DOS for high vs low subsurface Pt.
−0.86 r
ΔG H* vs surface-Pt d-band center
subsurface Pt ↑ → d-band center ↓ → Pt–H weakens → ΔG H* → 0
Key message The descriptor is not a fit — it is the d-band center doing the work.
12 / 18
13
The mechanism in one line
A closed causal chain, each link independently measured
1
Buried atoms
subsurface Pt count
→
2
d-band center ↓
surface-Pt states shift down
→
3
H binding weakens
ΔG H* rises toward 0
→
4
Sabatier population
sites land on the optimum
Key message Composition, electronics, and energetics tell one consistent story .
13 / 18
14
A compact local-environment predictor is already possible
Seven regression algorithms converge to the same error
Figure 4. Predicted vs DFT ΔG H* ; seven models, leave-one-out cross-validation.
27–31 meV
LOO-CV MAE — the band all seven models share
Ridge, LASSO, Random Forest, GBRT, SVR, GPR, KRR — same accuracy. The 96 sites are a compact training set for adsorption prediction.
Key message DFT learns the rule; descriptors predict the rest .
14 / 18
From DFT census to adsorption predictor
DFT is the calibration data, not the endpoint
1
Local composition + coordinates
each hollow site's environment
2
Local descriptors
subsurface counts · geometry · d-band
3
DFT-trained surrogate model
96 sites as the training set
4
Predicted ΔG H*
for uncalculated sites
5
Active-site population
μ, σ, Popt
DFT learns the rule — it does not enumerate every site.
The current model is a local-environment adsorption predictor , not yet a full MLP.
Goal rapid estimation of active-site populations across HEA composition space.
Key message The 96 DFT sites are not just results — they are training data for a fast adsorption predictor .
15
Design rule: shift the distribution with composition
Engineer the ΔG H* population — centre and spread — not one site
ΔG_H* (eV)
ΔG = 0
starting alloy
tuned to optimum
Tune the subsurface Pt / Cu / Ni / Co / Fe balance to slide the whole population toward ΔG H* = 0.
The d-band center is the transferable bridge from composition to reactivity.
Key message Choose a subsurface composition that centres the site population on the optimum.
15 / 18
16
Scope, honestly — and where it goes next
Firm within these compositions; the frontier is generalization
Established / limits
Current claim: the predictor is validated within the studied Pt-skin hollow-site dataset (96 sites)Only three compositions: cross-composition transfer not yet proven
So far, only the electronic d-band descriptor transfers across compositions
Outlook
Next: test cross-composition transfer — binary → ternary → quaternary → quinary hierarchyCarry the subsurface framing to OER / ORR / CO2 RR
Toward descriptor-guided, autonomous catalyst discovery
Key message Within-composition claims are firm; cross-composition generalization is next.
16 / 18
17
Three things to take home
Subsurface control of active-site distributions in Pt-skin HEA
1
HEA catalysis is a distribution problem
96 Pt-skin sites form a near-optimal population (μ = +0.013 eV; 92% within ±0.10 eV) — study the distribution, not one site.
2
Subsurface composition controls H adsorption through the d-band
An electronegativity-ordered hierarchy (Pt > Cu > Ni ≈ Co > Fe); subsurface Pt lowers the surface-Pt d-band center.
3
Local environments can predict adsorption energies
Seven algorithms converge to 27–31 meV, suggesting the HEA adsorption landscape is descriptor-compressible .
Key message HEA site heterogeneity is, at heart, a subsurface problem — and a solvable one.
17 / 18
Thank you · Questions welcome
Buried atoms are a design handle for surface reactivity
distribution on the optimum → electronegativity-ordered hierarchy → d-band mechanism → descriptor-limited ceiling
At a glance
System Fe–Co–Ni–Cu–Pt Pt-skin (111)
Data 96 DFT sites · 3 SQS slabs
Centre ΔGH* = +0.013 eV · 92% optimal
Driver subsurface Pt → d-band (r = −0.86)
Ceiling 7 models · 27–31 meV
Chen-Cheng Liao (廖振成) · Department of Chemistry, Fu Jen Catholic University165804@mail.fju.edu.tw · Supported by NSTC 114-2113-M-030-015-MY2 With students Peggy P. M. J. and Yu-Huan Huang (黃宇桓)
