It is expected that understanding the formation mechanism(s) of nanomaterials will be a key to develop more controlled syntheses ²² . This in turn will lead to nanomaterials designed with tuned properties to best match the requirements for a given application. Certainly, materials like Au nanoparticles have been intensively investigated and a relatively clear picture of the nanoparticle formation has been proposed ^{23, 24} . Nevertheless, several questions remain to understand and control how atoms of metal in a complex form larger nanomaterials, e.g. even for the case of well-studied metals like Pt ²⁵ . It can be observed that metals for which the synthesis can easily be followed by simple techniques, such as ultraviolet-visible spectroscopy (UV-vis) for Au or Pt, have been more intensively studied. It is therefore tempting to explain the relatively limited knowledge on Os nanoparticles by the challenging characterization of the related materials. Importantly, the risk of forming the toxic OsO ₄ ¹⁸ is also a bottleneck in the investigation of Os nanoparticles compared to Au or Pt.

A specific feature of Os nanomaterials is to lead to relatively small (<2 nm, see Table 1) nanostructures, regardless of the synthesis approach used. For this size range, most characterization techniques, until recently, are not easily implemented to evaluate size, shape and structure or to follow the formation mechanism of Os nanomaterials. Recently, using a combination of complementary in situ X-ray diffraction (XRD), quick X-ray absorption fine-structure (QXAFS) and X-ray photoelectron spectroscopy (XPS) performed at synchrotron facilities, the formation at high temperature of PdOs nanoparticles from [Pd(NH ₃) ₄][OsCl ₆] was studied ⁴⁴ . Such advanced studies are much needed to better understand the formation of nanomaterials but remain scarce for Os and Os based materials. Another example is the use of X-ray total scattering with pair distribution function (PDF) analysis, also requiring access to synchrotron facilities, where Os _xCl _y intermediates structures were suggested for the formation of Os nanoparticles in a colloidal approach ⁴² . Despite a relatively poor understanding on how Os nanomaterials form, and few reports focusing on the formation mechanism of Os nanoparticles, a range of successful syntheses have been reported and are illustrated in Table 1.

As opposed to wet chemical syntheses detailed below, where the formation of Os nanoparticles proceeds in the liquid phase, a range of high temperature dry syntheses are reported for Os nanoparticles. Typically a support material is needed to stabilize the nanoparticles ² . An overview of different syntheses is proposed and an example of synthesis is the thermal decomposition of Os precursors ⁴⁵ . Pyrolysis leads to different nanoparticle size depending on the ligand structures of the precursors ²⁷ and needs to be performed at relatively high temperature, e.g. 300 °C, when the precursor is an Os carbonyl complex ²⁶ . Hydrogen (H ₂) reduction is also an option ²⁹ . Magnetron sputtering has been reported for Os films ⁴⁶ . Alternative methods include wet incipient impregnation ⁴⁷ , freeze drying ⁴⁸ or atomic layer deposition (ALD) of Os films and particles ²⁸ . However, in this last approach and in this specific study, where osmocene and molecular oxygen were used as precursors, the challenge was the formation of highly toxic OsO ₄ at high temperature.

Wet chemical or colloidal syntheses are very popular synthetic approaches to obtain a range of nanomaterials directly relevant for multiple applications ^{1, 7, 49} . The formation of nanoparticles proceeds in a solvent via the reduction of a metal complex in an oxidized state in the presence of a reducing agent ⁵⁰ , followed by the growth of the nanoparticles ⁵¹ . In most cases, the syntheses do not require a support material. This is an advantage to truly investigate support and loading-related effects in catalysis since the nanoparticle formation and control over the nanoparticle size and other properties is independent of the presence of a support ⁵² . Typically, the syntheses are performed in presence of a range of additives like surfactants to stabilize the nanoparticles.

Os nanoparticles can be obtained from a variety of solvents and reducing agents summarized in Table 1. Surfactants typically added for the synthesis are for example heparin ⁴⁰ , (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) (HEPES) ³⁴ , 3-[4-(2-hydroxyethyl)-1-piperazinyl] propanesulfonic acid (EPPS) ³⁴ , piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES) ³⁴ , 2-ethanesulfonic acid (MES) ³⁴ , polyvinylpyrrolidone (PVP) ¹⁷ , sodium dodecyl sulfate (SDS) ³⁸ , DNA ^{37, 53} , sodium citrate ⁵⁴ and various precursors are suitable to obtain Os nanoparticles. OsCl ₃ remains a common precursor. OsCl ₃ can be reduced at room temperature (RT) using a strong reducing agent like LiEt ₃BH (superhydride) in tetrahydrofuran (THF) ³⁹ . The nanoparticles are circa 1.3±0.2 nm. NaBH ₄ is also a suitable reducing agents for RT synthesis ³⁸ . In a range of other syntheses, temperature between 80 to 200 °C are typically used depending on the solvent selected, see Table 1. In methanol-water in presence of PVP, sub-nanometer nanoparticles are obtained ¹⁷ . Ionic liquids are also suitable to obtain nanoparticles for instance from the metal carbonyl precursor Os ₃(CO) ₁₂ ^{55, 56} . The reaction of OsO ₄ in aqueous solution of cetyltrimethylammonium bromide (CTAB), 2,7-dihydroxynaphthalene (2,7-DHN) and NaOH, leads to nanoparticles circa 1-3 nm ³⁶ . Adjusting the concentration of CTAB, different morphologies made of individual nanoparticles, chain-like or aggregated clusters were obtained. Chains of Os nanoclusters are also obtained using ascorbic acid (AA) as a reducing and capping agent in an aqueous medium to lead to nanoparticles in the size range 1-1.5 nm with properties suitable for surface-enhanced raman spectroscopy (SERS) ³⁵ .

Os nanoparticles are typically small (<2 nm) across different syntheses ³⁴ . This therefore questions the actual need to stabilize the small nanoparticles. Developing surfactant-free colloidal syntheses, although it is challenging, is possible ⁵⁷ . Surfactant-free nanoparticles with a more accessible surface to reactants are directly relevant for catalysis. Surfactant-free nanoparticles are also more simply modified, for instance with dedicated ligands and molecules towards bio-medical applications. Examples of surfactant-free nanoparticles include the polyol synthesis ³³ , typically performed in alkaline ethylene glycol, or recently reported mono-alcohol synthesis ¹ , performed in alkaline methanol or ethanol. In the latter case, it was actually shown that high precursor concentrations up to 100 mM ⁴³ and even without the need for a base ⁴² , leads to the formation of small size <3 nm Os nanoparticles, see Figure 3. Such small size nanoparticles were obtained across a large parametric study investigating the time of synthesis from hours to weeks, nature and concentration of precursors, solvent composition and reducing agent (methanol or ethanol) as well as base concentration.

A recent work showed that face-centered cubic ( fcc) nanoparticles instead of the expected hexagonal close packed ( hcp) structure could be obtained by careful choice of the precursor, reducing agent and solvent, see illustration in Figure 4. - Iridium is the neighbor transition metal of Os and adopts the fcc structure. The difference in total energy between the hcp and fcc structures of Os is expected to be small and so it should be possible to obtain fcc Os nanoparticles. In presence of ethylene glycol and PVP using Os acetylacetonate (Os(acac) ₃), fcc nanoparticles were obtained whereas hcp nanoparticles were obtained with OsCl ₃ in water using NaBH ₄ as reducing agent ⁴¹ . The change in structure is attributed to the role of the acac ligand that can stabilize the nearest-neighbor Os–Os bond length (ca. 2.67 Å) in a close-packed plane of Os, that is close to the O–O length (2.74–2.93 Å) of the acac ligand. This leads to nanoparticles with a different crystal structure. The question of whether or not this would happen is the synthesis was performed under exactly the same conditions (same precursor concentration, reducing agents and solvents) but only changing the precursor remains open. Size selected nanoparticles were obtained in a two-phase (water-toluene) approach from OsO ₄ and tetrabutylammonium borohydride (TBABH ₄): 1±0.2 nm, 10–30 nm, 22±2 nm and 31±3 nm nanoparticles were synthesized by changing the concentration ratio of the metal precursor and the amount of reductant ³⁷ .

Compared with Os nanoparticles, Os complexes have been more studied to date ^{20, 58– 60} . For instance, Os metal carbonyl complexes have been widely investigated ⁶¹ . The group of Professor Gates intensively studied Os _n clusters ^{62, 63} . In particular Os carbonyls clusters were widely investigated on various support like gold ⁶⁴ , MoS ₂ ⁶¹ , zeolite ⁶⁵ , MgO ^{62, 63, 66, 67} with a focus on conversion from complexes to clusters. Carbonyls clusters were investigated by ¹²⁹Xe nuclear magnetic resonance (NMR), where [HOs ₃(CO) ₁₁] ^- or [H ₃Os ₄(CO) ₁₂] ^- were found to formed in zeolites ⁶⁵ and [Os ₃(CO) ₁₂] on MgO ⁶⁶ .

Barry et al. used Os atoms and complexes as their model system for various studies, e.g. to build up 3D nanocrystals to observe, study and quantify crystal growth at the atomic scale controlled in real time ^{68, 69} , see the illustration in Figure 5. The experiments were conducted using the electron beam of a transmission electron microscope (TEM) and a micelle-stabilized complex of [Os( p-cymene)(1,2-dicarba- closo-dodecarborane-1,2-dithiolate)] ³⁰ . The same precursor under microwave irradiation leads to supported Os nanoparticles circa 1 nm in diameter ³¹ . Os was used to show the temperature dependent nucleation and growth kinetics of precious metal nanocrystals supported on silicon nitride by aberration corrected TEM ³² . Barry et al. for that purpose used homemade Os complexes in that study. The growth rate was found to be dependent on the temperature ( circa 2.5 times faster at 100 °C than at 20 °C). No effect of the temperature on the crystal structure of the nanocrystals was observed, ’’although the sizes of the crystals (<2 nm) and the very small number of atoms per crystal render clear elucidation of the structures extremely difficult’ ’ ³² . The challenging characterization of Os nanoparticles by routine equipment indeed remains a bottleneck.

Os nanomaterials found applications in a wide range of fields and a broad overview is proposed here.

Chemical synthesis. Recent reports suggest that Os nanomaterials might have specific properties for hydrogenation reactions compared to other precious metals ⁷⁰ . Os based materials have been used as catalysts for dihydroxylation ⁷¹ , cyclohexene hydrogenation ^{17, 37} , citral hydrogenation ⁷² . Other reactions include oxidation of benzyl alcohol with relatively low yield compared to Ir ³⁹ or reduction of 4-nitro aniline ³⁶ . Using HEPES protected nanoparticles, the conversion of p-methylbenzylalcohol ³⁴ , p-methoxybenzylalcohol ³⁴ , p-bromobenzylalcohol ³⁴ , 3-phenyl-2-propanol ³⁴ , 3-phenyl-2-propenol (cinnamylalkohol) ³⁴ , 1-phenylethanol ³⁴ , lead in all cases to the aldehyde or ketones in relatively high yield. On silica doped with zirconium, Os nanoparticles show reactivity for hydrogenation and hydrogenolysis/hydrocracking of tetralin from aqueous K ₂[OsCl ₆] ⁴⁷ . A high Os content displays weak hydrogenation activity but very good hydrogenolysis/hydrocracking activity. Os nanomaterials are also suitable for the synthesis of various 1,2-cis-diols ⁷³ , 1,2/3-triols synthesis from the allylic hydroperoxides ⁷⁴ , syn-dihydroxylation of alkenes ⁷⁵ , reduction of p-nitroaniline into p-phenyldiamine ⁷⁶ , oxidations reaction ³⁹ , CO oxidation ⁷⁷ , ammonia synthesis ⁷⁸ or Fischer–Tropsch synthesis ⁷⁹ . Under aerobic condition, oxidation of activated, unactivated and heteroatom containing alcohols to carbonyl compounds lead to high activity and selectivity even under mild conditions ⁸⁰ .

6.2.5.2. Electrochemistry. OsO _x materials have been shown to be suitable for a range of electrochemical reactions including hydrogen evolution reaction (HER) ⁸¹ , oxygen reduction reaction (ORR) ⁸² or as direct borohydride polymer electrolyte membrane fuel cell anodes ⁸³ . Freeze drying was used to obtain Os/Si nanowires and the corresponding nanoparticles by etching the Si nanowires ⁴⁸ . In this comparative study with Rh, Pt, Pd, Re, Ru, Au or Ag nanocomposites, Os was found to give the higher activity for the HER, a small onset potential of -25 mV and long term stability ⁴⁸ . Using magnetron sputtered Os it was found that the high activity of OsO _x for the HER in acidic media was correlated with poor stability ⁸⁴ . Nanoparticles based on Os are easy to de-alloy, e.g. Pt ₂Os to from quasi core-shell Os@Pt for ORR in acidic media ⁸⁵ . This property can be used to develop high surface area materials by de-alloying, e.g. to develop improved porous-electrodes for the oxygen evolution reaction (OER), see Figure 6 ⁸⁶ . Os itself is expected to show very high activity for the OER but suffer from poor stability ⁸⁴ .

6.2.5.3. Other applications. Os nanomaterials are less studied than other precious metals for medical applications ⁶⁰ or pollution management ³⁷ . However, Os nanoparticles found recent applications in sensing. Os nanoparticles protected by heparin as the protecting/stabilizing agent were used as a heparinase sensor ⁴⁰ . Bovine serum albumin is an efficient protective shell to give Os nanoparticles an antifouling property regarding various ions (e.g., Hg ^2+, Ag ⁺, Pb ²⁺, I ⁻, Cr ⁶⁺, Cu ²⁺, Ce ³⁺, S ²⁻, etc.), saline (0−2 M), or protein (0−100 mg/mL) conditions. A colorimetric sensor was developed for H ₂O ₂ detection with improved properties compared to Au or Pt based sensors ⁸⁷ , see the illustration in Figure 7. Other examples include glucose and pyruvic acid detections ⁵⁴ , folic acid detection ⁸⁸ colorimetric sensors for heavy metal ions discrimination (Cu ^2+, Ag ⁺, Cd ^2+, Hg ^2+, and Pb ²⁺) ⁸⁹ . Os nanoparticles also show SERS properties ^{36, 37, 76} .

Less work has been performed on Os nanoparticles than Ir ⁹⁰ or Pt ²⁵ nanoparticles but some theoretical work can be found in the literature ^{70, 91– 93} . For instance, Os was suggested to be a suitable catalysts for ammonia production ⁹⁴ . While being less investigated than Ir, Os _xn clusters were studied by density functional theory (DFT) for instance in light of their interaction with MgO for n=4,5 ⁹¹ . By analogy with what is available for Ir or Pt, it can be expected that theoretical work will be valuable to clarify why small size nanoparticles are easily obtained, which might be related to the formation of ‘magic number’ nanoparticles with specific sizes ⁹⁵ and/or sintering resistance properties ⁹⁶ . Equaly theoretical work could be relevant to explore further the properties of Os based nanomaterials, in particular towards improved stability.

In addition to the examples already mentioned above, for instance in Figure 6, various alloyed nanoparticles have been reported such as IrOs ¹⁴ , PtOs ⁹⁷ , OsB ₂ ⁹⁸ , in particular for their improved mechanical properties. PdOs nanoparticles were reported as catalyst for carbon nanotube synthesis ²⁶ . Other examples include NiOs ₄ reported for the improved hydrogenation of cinnamaldehyde ⁹⁹ , PtOs for the methanol oxidation reaction ¹⁰⁰ , CuOs for the methanol oxidation reaction and ORR ¹⁰¹ or OsTe nanorods for cancer therapy ¹⁰² .