Coreduction methodology for immiscible alloys of CuRu solid-solution nanoparticles with high thermal stability and versatile exhaust purification ability†

This study provides a coreduction methodology for solid solution formation in immiscible systems, with an example of a whole-region immiscible Cu–Ru system. Although the binary Cu–Ru alloy system is very unstable in the bulk state, the atomic-level well-mixed CuRu solid solution nanoparticles were found to have high thermal stability up to at least 773 K in a vacuum. The exhaust purification activity of the CuRu solid solution was comparable to that of face-centred cubic Ru nanoparticles. According to in situ infrared measurements, stronger NO adsorption and higher intrinsic reactivity of the Ru site on the CuRu surface than that of a pure Ru surface were found, affected by atomic-level Cu substitution. Furthermore, CuRu solid solution was a versatile catalyst for purification of all exhaust gases at a stoichiometric oxygen concentration.


Mass transfer calculations.
Thiele theory has been used to evaluate the influence of mass transfer factor.
The Thiele modulus is: R0 is the radii of catalyst particle (ca. 100 nm), CNOs is the NO concentration at particle surface, De is the effective diffusion coefficient.
When pore size is relative large, De can be calculated as ̅ is the mean velocity of gas molecule, is mean free path, θ is porosity and is bending factor of pore. Finally, from equation (2), we obtain Thiele modulus as s = 1.3×10 -7 . The corresponding effectiveness factor is then estimated as = 1 based on the -relation. Since Thiele modulus for NO reduction is extremely small, we can neglect the influence of mass transfer.
Set d as the heat conduction distance in Al2O3, tNP as the temperature of CuRu NP, <t> as the heat-convection equivalent average temperature of the A3.
We set the forced-convection heat transfer coefficient of He in our experimental condition, hHe = 200 W m -2 K -1 .

Details of reaction time measurements
Fig. S1 General procedures of reaction time measurements.
Reaction time measurement of Ru acetylacetonate (Ru(acac)3): 100 ml triethylene glycol (TEG) solution containing 1.1g polyvinylpyrrolidone (PVP) was heated to 240 ˚C under ambient condition (named as solution A). A precursor solution with 199.2 mg (0.5 mmol) Ru(acac)3 dissolved in 10 ml TEG was dropwise added into solution A, the reaction system was stirred for 10 min under 240 ˚C. Samplings were taken at different time points as 1 min, 2 min, 3 min, 5 min and 8 min, respectively, followed by icy bath to cool the samples immediately. Transmission electron microscopy (TEM) measurements were carried out for each sample, reaction time was estimated by particle size change shown in Fig. 2a.
Reaction time measurement of Cu acetylacetonate (Cu(acac)2): 100 ml TEG solution containing 1.1g PVP was heated to 240 ˚C under ambient condition (named as solution A). A precursor solution with 130.9 mg (0.5 mmol) Cu(acac)2 dissolved in 10 ml TEG was dropwise added into solution A, the reaction system was stirred for 3 min under 240 ˚C. Samplings were taken at different time points as 10 s, 20 s, 40 s, 60 s and 90 s, respectively, followed by icy bath to cool the samples immediately. TEM measurements were carried out for each sample, reaction time was estimated by particle size change shown in Fig. S2.
Reaction time measurement of Cu acetate monohydrate (Cu(OAc)2•H2O): 100 ml TEG solution containing 1.1g PVP was heated to 240 ˚C under ambient condition (named as solution A). A precursor solution with 99.8 mg (0.5 mmol) Cu(OAc)2•H2O dissolved in 10 ml TEG was dropwise added into solution A, the reaction system was stirred for 10 min under 240 ˚C. Samplings were taken at different time points as 10 s, 20 s, 40 s, 1 min, 1.5 min, 2 min, 3min, 5min and 10 min, respectively, followed by icy bath to cool the samples immediately. Reaction time was estimated by colorimetry shown in Fig. 2b.

Reaction time measurement of Cu formate (Cu(HCOO)2•4H2O)
: 100 ml TEG solution containing 1.1g PVP was heated to 240 ˚C under ambient condition (named as solution A). A precursor solution with 76.8 mg (0.5 mmol) Cu(HCOO)2•4H2O dissolved in 10 ml TEG was dropwise added into solution A, the reaction system was stirred for 10 min under 240 ˚C. Samplings were taken at different time points as 10 s, 20 s, 40 s, 1 min, 1.5 min, 2 min, 3min, 5min and 10 min, respectively, followed by icy bath to cool the samples immediately. Reaction time was estimated by colorimetry shown in Fig. 2c.

CuRu alloy synthesis by combination of Ru(acac)3 and Cu(acac)2
Synthetic procedure: To synthesize CuRu alloy NPs, a TEG solution (4 ml) containing Ru(acac)3 (19.9 mg, 0.05 mmol) and Cu(acac)2 (13.1 mg, 0.05 mmol) was added into a mixture solution of PVP (220 mg) and TEG (10 ml) at 240 ˚C under ambient condition. The reaction was kept for 10 min and cooled to room temperature. The black powder was collected by same post-treatment as fcc-Ru NPs.

Fig. S6
Schematic illusion for Cu@Ru core-shell structure formation. According to coreshell formation mechanism, reaction time of Cu(acac)2 was estimated to be much shorter than that of Ru(acac)3, which is consistent with reaction time measurements of Cu(acac)2 and Ru(acac)3.

CuRu alloy synthesis by combination of Ru(acac)3 and Cu(HCOO)2•4H2O
Synthetic procedure: To synthesize CuRu alloy NPs, a TEG solution (10 ml) containing Ru(acac)3 (199.2 mg, 0.5 mmol) and Cu(HCOO)2•4H2O (112.8 mg, 0.5 mmol) was added into a mixture solution of PVP (1.1 g) and TEG (100 ml) at 240 ˚C under ambient condition. The reaction was kept for 10 min and cooled to room temperature. The black powder was collected by same post-treatment as fcc-Ru NPs.

CuRu alloy synthesis by combination of Ru(acac)3 and Cu(OAc)2•H2O
Synthetic procedure: To synthesize CuRu alloy NPs, a TEG solution (10 ml) containing Ru(acac)3 (199.2 mg, 0.5 mmol) and Cu(OAc)2•H2O (99.8 mg, 0.5 mmol) was added into a mixture solution of PVP (1.1 g) and TEG (100 ml) at 240 ˚C under ambient condition. The reaction was kept for 10 min and cooled to room temperature. The black powder was collected by same post-treatment as fcc-Ru NPs.

CuRu alloy synthesis by combination of Ru(acac)3 and Cu(OAc)2•H2O with anhydrous solvent
Synthetic procedure: The procedure was the same with CuRu alloy synthesis by combination of Ru(acac)3 and Cu(OAc)2•H2O except for solvent pretreatment. Solvent TEG in this synthesis was pretreated with activated molecular sieves overnight to remove water, and freshly treated anhydrous TEG was soon used in synthesis after pretreatment.

Atmosphere factor optimization.
Synthetic procedure: The procedure was the same with CuRu alloy synthesis by combination of Ru(acac)3 and Cu(OAc)2•H2O except for N2 bubbling condition compared with ambient condition. Before reaction, TEG solvent was pre-bubbled with N2 for 1 h. The N2 bubbling was continued during reaction and was stopped until the reaction cooled down to room temperature.

Reducing agent and solvent factor optimization.
Synthetic procedure: Diethylene glycol (DEG) solvent: Before reaction, a DEG solvent was pre-bubbled with N2 for 1 h. To synthesize CuRu alloy NPs, a DEG solution (10 ml) containing Ru(acac)3 (199.2 mg, 0.5 mmol) and Cu(OAc)2•H2O (99.8 mg, 0.5 mmol) was added into a mixture solution of PVP (1.1 g, 10 mmol) and DEG (100 ml) at 240 ˚C under N2 bubbling. The N2 bubbling was continued during reaction and was stopped until the reaction cooled down to room temperature. The reaction was kept for 10 min and cooled to room temperature. The black powder was collected by same post-treatment as fcc-Ru NPs.

Glycerol solvent:
The procedure was the same with CuRu alloy synthesis procedure with DEG solvent above except for glycerol solvent.    Scheme S1 XRD patterns of CuRu alloy NPs synthesized with Cu(OAc)2•H2O in diethylene glycol (DEG, red), TEG (purple) and glycerol (green), compared with fcc-Ru NPs (black) and Cu NPs (blue) at 303 K. The radiation wavelength was 1.54056 Å.
Product synthesized in DEG shows similar lattice constant compared with product obtained in TEG. On the other hand, multiple components product was obtained in glycerol (Fig.  S21). In general, the reducing abilities of polyol are enhanced with the increasing of hydroxyl (-OH) group density. Stronger reducing ability provides faster reducing velocity. Thus, the reducing velocity ranking of the solvents is glycerol > DEG > TEG (Scheme S1). This velocity ranking did not fit with the quality ranking of nanoparticles synthesized in the solvents, which were DEG > TEG > glycerol. In this case, the qualities of different products are considered to highly correlate with the viscosities of solvents, which has also been reported by Park and coworkers in case of size control 2 . Considering that viscosity of glycerol (1.41 Pa•s) is nearly 2-order higher than those of DEG (3.57 × 10 -2 Pa•s) and TEG (4.90 × 10 -2 Pa•s) under room temperature (Scheme S1) 1 , slower diffusion of metal atoms in glycerol would be expected, which resulted in inhomogeneous aggregation even with same reaction times. In contrast, DEG with lower viscosity can obtain the best quality of product among them.

Solvent oxygen content factor optimization.
Synthetic procedure: The procedure was the same with CuRu alloy synthesis by combination of Ru(acac)3 and Cu(OAc)2•H2O except for 3-day-liquid-N2 degassed DEG compared with N2 bubbled DEG.   For degassed DEG sample, from the Le Bail fitting result of PXRD pattern, only slight decrease on lattice constant could be observed compared with N2 bubbled DEG sample. Furthermore, by STEM-EDX mapping measurements, pure Cu elements locating on particle surface were found in N2 bubbled DEG sample (Fig. S23), meanwhile most of Cu elements in liquid N2 degassed DEG sample located inside CuRu alloy nanoparticles. This gives a direct evidence to better mixing for liquid N2 degassed DEG than that of N2 bubbled DEG. Solvent remaining oxygen by liquid N2 degas could be fully removed, however N2 bubbling cannot.

Stirring speed factor optimization.
Synthetic procedure: The reaction with Ru(acac)3 and Cu(OAc)2•H2O was carried out at stirring speed of 1200 rpm compared with 400 rpm under degassed condition. Very obvious right shift was observed for sample synthesized at stirring speed of 1200 rpm compared with that of stirring speed of 400 rpm. Better alloy formation was achieved at higher stirring speed. Combining the previous discussion for viscosity effect of solvents, it is clear that the diffusion process of metal atoms is of high importance in alloy formation for immiscible systems.     18. Rietveld refinement Results on in situ synchrotron PXRD patterns at 773 K and select area analysis of STEM-EDX mapping images after thermal stability test.

Fig. S34
Select area analyses of each particle in STEM-EDX mapping of CuRu alloy NPs after heating at 773 K. The atomic ratio of Cu:Ru is nearly equal to 1:1, which is same to the nominal ratio. No difference was found before and after heating up to 773 K. S34 Explanation to Fig. S35. In some synthetic process for bimetallic alloys, even the reduction velocities of each precursor are not same, by some techniques such as rapid reduction, when the reduction times of precursors is much shorter than diffusion time of metal atoms in solution, the random alloy structure could also be obtained. In that case, since the reduction velocities are not same, the random alloy structure is close to a multiple domain mixing structure shown in Fig. S35B, which is considered to be more unstable compared with atomic level mixing structured alloy obtained from well-optimized coreduction condition (Fig. S35A).

Fig. S39
Five cycles three-way catalytic durability test for CuRu solid solution.