Those of us who make nanoparticle catalysts all know an uncomfortable truth: more than half of the platinum you pay for never takes part in the reaction at all.

In platinum particles a few nanometres across, only surface atoms can reach the substrate; the platinum atoms buried in the core never do a single turn of work from start to finish. The smaller the particle, the better that ratio becomes — but once you are down to a few nanometres it is very hard to push further. Go smaller still and the particles become unstable, agglomerating and growing the moment they are heated.

Single-atom platinum is a different idea altogether: don't make particles at all. Break the platinum apart completely and pin it to the support atom by atom. Dispersion is then 100%, every platinum atom is at the surface, and not one is wasted.

It sounds like the ultimate answer. In practice, the trouble lies precisely there.

1. What Actually Counts as "Single-Atom"

The definition has to be pinned down, because the term is used rather loosely these days.

A genuine single-atom catalyst (SAC) must satisfy this: the metal exists as isolated atoms, with no direct bonding between atoms, each separated from the others by the support. It is not "a very small particle" but a centre in the coordination-chemistry sense — the platinum atom is coordinated to oxygen, nitrogen or similar atoms on the support, forming a well-defined coordination structure such as Pt–O₄ or Pt–N₄.

Two boundaries must be drawn clearly here:

• It is not a nanoparticle. Even at 1 nm, with only tens of atoms, if Pt–Pt bonds exist between the atoms it is a cluster, not a single atom. Clusters and single atoms differ greatly in their properties; conflating them leads directly to a wrong selection.
• It is normally not zero-valent metal. Single-atom platinum mostly exists as Pt(II) or Pt(IV) in an ionic state, held on the support by coordination bonds. It lacks the continuous electronic structure of metallic platinum — which is the fundamental reason certain reactions cannot be carried out on it.

A common compromise form is the single-atom alloy: platinum single atoms dispersed in a matrix of another metal such as palladium, copper or silver. The platinum is isolated, but the surrounding matrix is metallic and can supply electrons and a hydrogen spillover pathway; its performance differs markedly from that of single atoms on an oxide support.

2. Why the Effort Is Worth It: the Ceiling on Cutting Platinum Consumption

The point of the single-atom route is, in essence, to push platinum atom utilization to its theoretical maximum.

In electrocatalytic applications such as fuel cells, this bears directly on cost. Platinum loading in the membrane electrode assembly is the single largest item in fuel cell cost, and the industry has been pressing it down for years — from fractions of a milligram per square centimetre in the early days to the order of hundredths of a milligram today. On that path, simply making particles smaller soon hits a bottleneck (the smaller the particle, the higher the specific surface energy, and the more readily it dissolves, migrates and agglomerates during potential cycling); single atoms and core–shell structures are the two main ways out.

In thermal catalysis, the benefit is either lower platinum usage at equal activity, or higher activity at equal platinum loading. For applications where platinum consumption is the core cost pressure and the reaction type permits it, the economics of this route are attractive indeed.

There is a further layer of value that is often overlooked: the active sites are highly uniform. A nanoparticle surface carries terraces, steps, edges and kinks, each with different activity, so what is measured macroscopically is an average and mechanistic study is very difficult. Single-atom sites approach a single structure, characterization data are clean, and kinetic models are easier to build. That is a real gain in R&D efficiency.

3. The First Hurdle You Cannot Avoid: Thermodynamically It Shouldn't Exist

It has to be said plainly: isolated platinum atoms are thermodynamically unstable.

The surface energy of an individual atom is extremely high, and the system is inherently driven to bring them together into particles, because that lowers the total energy. A single-atom catalyst is therefore essentially a metastable state held down by kinetics — the strong coordination bonds between the platinum atoms and the anchoring sites on the support suppress the driving force for agglomeration.

What are these anchoring sites? Mainly these classes:

• Defect sites on the support surface (oxygen vacancies, carbon vacancies, edge sites);
• Surface functional groups such as hydroxyl and carboxyl;
• Pyridinic and pyrrolic nitrogen on nitrogen-doped carbon — at present the most mainstream and most mature anchoring approach;
• Coordinatively unsaturated metal sites on metal oxides.

The conclusion is direct: the support determines whether single atoms can be made at all, and whether they last. The same platinum source, applied to untreated activated carbon, will agglomerate into particles immediately; applied to nitrogen-doped carbon, it may stabilize as single atoms. The technical barrier for this class of product therefore lies not in the platinum but in the preparation and surface engineering of the support.

4. How It Is Made: Five Mainstream Routes

The biggest difficulty in scale-up is batch consistency. These processes are sensitive to temperature profile, atmosphere, heating rate, precursor concentration and pH; if any single parameter drifts, the product can shift from "mostly single atoms" to "single atoms mixed with a raft of small clusters." The performance gap between those two is large.

Worse still: that kind of contamination is very hard to detect by routine testing. The next section goes into it.

5. How to Prove It Really Is Single-Atom

This is where procurement and quality control most need to be rigorous — and where the industry contains the most hot air.

First, aberration-corrected HAADF-STEM — the decisive evidence. A high-resolution image should show individual, isolated bright points distributed evenly across the support, with no particle-like clusters of bright spots visible anywhere. This is the most intuitive evidence, and the one most often produced.

But note two limitations: it surveys only an extremely small field of view and cannot represent the uniformity of the whole batch; and sub-nanometre clusters can be confused with single atoms in the image. A report offering only a few handsome micrographs is therefore not persuasive.

Second, XAFS / EXAFS — quantitative hard evidence. Look at the coordination environment of the platinum:

• If the fit yields no Pt–Pt coordination path, only Pt–O or Pt–N paths, with a coordination number around 4, that is the key criterion for single atoms.
• Once a clear Pt–Pt coordination path appears, with a coordination number approaching that of bulk metal, particles or clusters are present.

This method is more representative of the whole sample than electron microscopy, and is the most powerful means of judging the degree of single-atom character.

Third, CO-DRIFTS — a chemical fingerprint, and very practical. The infrared spectrum of adsorbed carbon monoxide distinguishes site types:

• Single atoms give only the peak for linearly adsorbed CO (around 2090 cm⁻¹).
• Once a peak for bridge-bonded CO appears (around 1850 cm⁻¹), two or more adjacent platinum atoms are present — meaning clusters or particles.

This method is cheap, fast and sensitive; it is a routine item in our internal quality control and is especially good at catching "clusters mixed in."

Fourth, XPS for oxidation state, ICP for content, TPD / chemisorption for the number of sites. An oxidation state deviating from zero is normal for single atoms; platinum content must be determined accurately, otherwise no normalization of activity is possible at all.

Recommended acceptance combination: aberration-corrected STEM + EXAFS fit results + CO-DRIFTS. The three corroborate one another. Any one alone can be fooled by a sample that "looks beautiful locally and fails as a whole."

6. What It Can Do, and What It Cannot

This boundary matters more than the advantages.

Suited to:

• Low-temperature CO oxidation and PROX (preferential oxidation to remove CO): single-atom platinum performs outstandingly here; it is one of the most thoroughly studied areas.
• Selective hydrogenation: where only one functional group is to be reduced without touching neighbouring groups, the site uniformity and the "no contiguous platinum ensemble" character are advantages, and selectivity can be made very high.
• Electrocatalysis: oxygen reduction (ORR), hydrogen evolution (HER), formic acid / methanol oxidation. The main battleground for cutting platinum consumption.
• Photocatalysis and photocatalytic water splitting: single atoms anchored on a semiconductor as co-catalysts markedly suppress recombination of photogenerated charge carriers.
• Low-temperature catalytic oxidation of VOCs: high site density gives good low-temperature activity.

Not suited to (the key limitations):

• Reactions requiring dissociation of the hydrogen molecule. H₂ dissociation normally needs two adjacent metal atoms working together, which an isolated ionic platinum atom cannot provide. So in hydrogenation systems involving H₂ activation, single atoms usually have to rely on hydrogen spillover — the support or an adjacent second metal dissociates the hydrogen, which then migrates to the platinum. Pure single-atom platinum is generally weaker than nanoparticles for hydrogenation under a hydrogen atmosphere.
• Reactions requiring C–C bond cleavage (hydrogenolysis, severe reforming, coke gasification): these demand a larger contiguous metal ensemble, which the single-atom structure simply does not have.
• High-temperature duty. Above a few hundred degrees Celsius the anchoring bonds begin to fail, single atoms rapidly agglomerate into particles, and performance degrades to the level of an ordinary supported catalyst — irreversibly.

7. Three Real Problems Encountered on Site

Problem one: stability — the hardest hurdle.

There are three main deactivation mechanisms: failure of anchoring sites at high temperature, leading to migration and agglomeration; complexation and leaching of platinum by the reaction medium (particularly in halogen-, ammonia-, cyanide-bearing or oxidizing media); and dissolution–redeposition under potential cycling in electrochemical duty.

The symptom is non-linear decay of activity with time — slow at first, then a cliff after a certain point. Durability testing must therefore run long enough, or cover enough cycles; looking only at initial activity is meaningless. When we evaluate a single-atom sample, initial activity carries only a small part of the weighting; the decay curve is the core.

Problem two: clusters are present in the product, and the clusters may be what actually does the work.

This is a point under active debate in academia as well. In some systems, single atoms transform in situ into small clusters under reaction conditions, and what really catalyzes the reaction is the cluster formed during operation, not the single atom as charged. In such cases the label "single-atom catalyst" describes the state of the raw material, not the working state.

The engineering lesson is this: characterization should be done in situ, as close as possible to actual reaction conditions, rather than inferring operating performance from data taken under vacuum or inert atmosphere. Surface composition and structure drift with atmosphere — a point that holds for all platinum catalysts.

Problem three: scale-up and cost.

Aberration-corrected electron microscopy and synchrotron XAFS are affordable for research institutes, but costly and slow for production quality control. Yet, as noted above, batch variation in single-atom products is precisely what these methods are needed to identify.

This is one of the practical reasons it has still not displaced nanoparticle catalysts at scale: not that it cannot be made, but that making it stably, in bulk, and verifiably is extremely hard.

8. Practical Advice for Buyers

1. Confirm first whether the reaction type permits it. If H₂ dissociation, C–C cleavage or operating temperature above a few hundred degrees is involved, rule it out immediately — don't waste time evaluating.
2. Ask for three characterization datasets, not just micrographs. Aberration-corrected STEM + EXAFS (look specifically for the presence or absence of a Pt–Pt path) + CO-DRIFTS (look specifically for a bridge-bonded CO peak).
3. Ask for the decay curve, not just initial activity. Measure under your actual temperature, atmosphere and impurity levels, with enough cycles or duration.
4. Ask for batch data. At least three batches compared on content, dispersion and activity, to judge the supplier's process control capability. What is most valuable in a single-atom product is reproducibility, not handsome numbers from one batch.
5. Include recovery when calculating net cost. Supports for single-atom platinum are mostly carbon or oxides, so the recovery path is close to that of conventional supported catalysts; the scrap remains a platinum-bearing asset, and assay protocols at handover should be agreed in writing up front.
6. Don't use it to replace a mature process that already runs well, unless there is a clear cost or selectivity driver. At the present stage it is better suited to new project designs and to applications with hard targets on platinum consumption or selectivity.

9. The One-Sentence Summary

Single-atom platinum takes the question of platinum utilization to its theoretical limit — 100% dispersion, uniform sites, strong design freedom — and is one of the most promising routes for cutting platinum consumption.

But it is a metastable state held down by kinetics: the support decides whether it works, the anchoring bonds decide how long it lasts, and the characterization methods decide whether what you bought is genuine. And its structural limitation — having no continuous metal surface — means it is destined to be a sharp tool for specific reactions, not a general replacement for nanoparticles.

For a producer, the priority right now is not achieving a higher initial activity in the laboratory, but stabilizing batch consistency and lengthening the decay curve. Once those two things are done, it can genuinely move from the paper onto the production line.



Diethylenetetramethyldisiloxane platinum complex (platinum catalyst) MY 8115