Precise AuxPt1−x Alloy Nanoparticle Array of Tunable Composition for Catalytic Applications

A 3-dimensional Block Copolymer Micellar nanoLithography (BCML) process was used to prepare AuxPt1−x alloy nanoparticles (NPs) monodisperse in size and composition, strongly anchored onto SiO2-particles (0.2 wt.% AuxPt1−x/SiO2). The particles possess a face-centered cubic (fcc) crystal structure and their size could be varied from 3–12 nm. We demonstrate the uniformity of the Au/Pt composition by analyzing individual NPs by energy-dispersive X-ray spectroscopy. The strongly bound AuxPt1−x NPs catalyzed the oxidation of CO with high activity. Thermal ageing experiments in pure CO2 as well as in ambient atmosphere demonstrated stability of the size distribution for times as long as 22 h.

Scientific RepoRts | 6:20536 | DOI: 10.1038/srep20536 spectroscopy (EDX), and inductively coupled plasma atomic emission spectroscopy (ICP-AES). CO oxidation studies of monodisperse Au, Pt, and Au x Pt 1−x alloy NPs were conducted to investigate correlations between the properties of the catalyst and its catalytic activity. Catalytic activity measurements of the NPs were performed by way of a cyclic temperature-programmed CO oxidation reaction.

Results
BCML of bimetallic alloy NPs. Au x Pt 1−x NPs were synthesized using a previously described block copolymer micellar nanolithography technique (BCML) [18][19][20][21][22] . The size of the NPs and the distance between individual NPs was tailored through adjusting the choice of the block copolymer, the concentration of the micellar solution, and the metal loading. The amount of metal salt in the micelles is a function of the loading parameter L: where m represents the mass, M the molar mass and [Units VP] the amount of vinylpyridine monomer. Block copolymers can form self-assembled monomicellar layers. Flat substrates can be spin-or dip-coated by a monomicellar film whereby the noble metal salt is concentrated within the core of the micelles. By a subsequent plasma treatment monodisperse noble metal NPs deposite in a quasi-hexagonal pattern at the interface of the substrate [18][19][20] . To generate bimetallic alloy noble metal NPs, the micelles are loaded by two different metal salts, a second loading step is applied after the first loading step (M A ). The metal ions of the second metal salt (M B ) also diffuse into the hydrophilic core of the micelle. M A M B NPs are formed during the following plasma treatment 23 .
Employing this procedure, we were able to synthesize alloy NPs with diameters between 3-12 nm and desired noble metal ratios Au x Pt 1−x . The following noble metal ratios were used for the catalytic measurements: Au 0.1 Pt 0.9 , Au 0.3 Pt 0.7 and Au 0.5 Pt 0.5 . Pure Au NPs and Pt NPs were also compared. The optimization of the diameter, particle size, and interparticle distance of Au x Pt 1−x alloy NPs based on the utilization of different block copolymers is shown in Fig. 1 using the example of Au 0.5 Pt 0.5 NPs.
For catalytic applications a sufficiently large number of noble metal NPs must be well-dispersed on a carrier with a high surface to volume ratio in order to enable sufficient chemical conversions. This can typically be realized as a washcoat. Washcoats are ceramic powders, comprised typically of SiO 2 , TiO 2 , Al 2 O 3 , or mixtures of these materials, which cover the walls of the honeycomb structures of catalytic converters 24,25 . Here, mesoporous silica powder is used as a substrate for the immobilization of the Au x Pt 1−x NPs. The total surface area was calculated from an adsorption-desorption isothermal curve obtained from H adsorption experiments. The silica powder had a total surface area of a BET = 141.1 m 2 /g. Gadomska et al. 26 established a method for creating well-ordered gold NPs on spheres called BCML. Employing their BCML method, they covered 75 μ m glass beads with a monolayer of quasi hexagonally-ordered micelles. Likewise, it is possible to coat the surface of mesoporous silica powder with Au x Pt 1−x alloy NPs. The SEM and CTEM characterization of the coated powder showed excellent dispersion of the Au x Pt 1−x alloy NPs on the surface of the ceramic powder particles (Fig. 2).
The following analysis and measurement of the catalytic activity in a CO oxidation reaction was performed on Au x Pt 1−x alloy NPs (approx. 6 nm) synthesized via BCML using the PS(227)-b-P2VP(99) block copolymer. For the size distribution of the Au x Pt 1−x NPs used for catalytic measurements and characterization see Supplementary Figs S1 and S2 online. The Au x Pt 1−x NPs synthesized using BCML were analyzed by high-resolution transmission electron microscopy (HRTEM), high-angle annular dark-field imaging (HAADF), energy-dispersive X-ray spectroscopy (EDX), and inductively coupled plasma atomic emission spectroscopy (ICP-AES). Our measurements revealed that Au and Pt atoms are randomly distributed inside the NPs (see Fig. 3). Concerning their atomic lattice parameters, the NPs are arranged along the < 110> direction and show two (111) lattice planes. The angle between the two (111) lattice planes is 70°, which indicates a face-centered cubic (fcc) crystal structure. The results are in agreement with recent work by Petkov et al. 27 who observed a random alloy structure in Au x Pt 1−x NPs (< 10 nm) synthesized by a two-phase method using aqueous solutions of the metal salts.
EDX measurements of single NPs and ICP-OES measurements of the coated silica powder were performed to quantify the atomic ratio of Au to Pt. The NPs designated for catalytic CO oxidation measurements (i.e., Au 0.1 Pt 0.9 , Au 0.3 Pt 0.7 , and Au 0.5 Pt 0.5 NPs) were analyzed. Results are listed in Table 1. The alloy composition in the left column states the originally used Pt and Au amounts. Most importantly, the measurement of the atomic ratio within the overall sample, shown in the column on the right, proves that the designated alloy composition is achieved not only in some but in all NPs. We believe that the precision of this production method with regard to obtaining a particular metal ratio in every single NP is outstanding and cannot be reproduced by any other method for preparing alloy nanoparticles.
Regarding the thermal stability of the Au x Pt 1−x alloy NPs EDX measurements were performed to verify that also no phase segregation of Au and Pt takes place. For this, Au 0.5 Pt 0.5 NPs were immobilized on Y:ZrO 2 (100) planar substrates and aged under CO 2 (435 mbar) and atmospheric conditions at 400 °C for 22 hours (Fig. 4). The EDX results reveal that phase segregation does not occur during the aging process. Furthermore, no changes in the size distribution of all used alloy (Au x Pt 1−x ) and pure noble metal (Au, Pt) NPs were observed.
ICP-AES analysis not only reveals the ratio of the compounds but also provides information about the quantity of noble metal immobilized on the surface of the silica powder. Au x Pt 1−x samples of each ratio were prepared using an identical procedure. Our measurements showed that in all cases the amount of noble metal dispersed on the surface of the silica powder constituted about 0.2 wt% of the sample.
Experiments pertaining to the catalytic activity in the CO oxidation reaction were performed in a fully automated set-up, thus allowing the measurement of transient kinetics. Helium was used as the balance gas. The reactor was operated with a theoretical oxygen excess of 100% for the reaction It is well known that Pt surfaces show a strong tendency for CO poisoning 28 ; therefore, CO-rich conditions were chosen in order to evaluate the possible contribution of the gold content in CO oxidation. Temperature-programmed oxidation was carried out cyclically with a total volume flow rate of 20 NmL/min and The ion current of CO was corrected by the theoretical ratio of the produced CO 2 (which equals 11,4%), because CO is a fragment of CO 2 . The results of the CO oxidation measurements are displayed in Fig. 5. The result of the first heating cycle of the 0.2%Au 0.5 Pt 0.5 /SiO 2 NPs is of particular importance (see Fig. 5a): during this first stage of heating (up to nearly 300 °C), the amount of CO decreased strongly. Simultaneously, the O 2 signal decreased to zero before slightly increasing to a constant value at higher temperatures. The CO 2 signal -the product of the oxidation reaction -reached a peak before arriving to a steady state regime. The turnover rate of CO oxidation reaches a limiting value determined by the rate of adsorption-desorption of the involved molecules 29,30 . Furthermore, two clearly separated H 2 O peaks are visible, one below 100 °C and one at approximately 400 °C. The first peak indicates a significant amount of weakly bound, physisorbed water, whereas the second peak, which is also lower in intensity, suggests strongly bound water. The water peaks only appear during the initial heating of the samples. For the sake of simplicity, only the CO 2 signal and the temperature profile of 5 cycles are plotted for the other Au x Pt 1−x NPs-based catalysts (Fig. 5b-f).
The catalytic behavior of all Au x Pt 1−x /SiO 2 NP samples was similar with the exception of the 0.23%Au/SiO 2 sample. The T 50 temperature, at which 50% of the CO streamed in is turned over, is a direct indicator of the catalytic activity. Table 2 gives an overview of the catalytic activity of the five measured cycles. During the first cycle, activation of the catalyst takes place. This activation is demonstrated by the observed removal of stored molecules and water from the Pt sample (see Fig. 5a). Therefore, the T 50 temperature of the first cycle is much higher than in later cycles, but reaches a constant value of catalytic activity with the second cycle. Additionally, the 0.17%Au 0.3 Pt 0.7 and 0.18%Au 0.1 Pt 0.9 samples show a higher CO 2 discharge during the first cycle compared to the other samples. This higher amount of CO 2 discharge could be due to the reaction of stored hydrocarbons on the samples, which also contribute to the signal. The stable T 50 values of the subsequent cycles indicate no ageing of the samples for temperature cycles up to 400 °C. The T 50 temperature of the fifth cycle for each sample is used to compare the catalytic performance in CO oxidation ( Table 2). The pure 0.15%Pt/SiO2 sample showed the highest catalytic activity of all the tested samples with a T 50 temperature of 242 °C after activation. In contrast, the 0.23%Au/SiO2 sample showed no catalytic activity under these experimental and evaluation conditions. We do not measure any advantage of Au on Pt-based NPs with respect to improving the catalytic performance during CO oxidation. The correlation between the T 50 temperature and NP composition is most evident: T 50 increases with increasing amounts of Au and decreasing amounts of Pt in the NPs (Fig. 6). In particular, a comparison of the 0.15%Pt and 0.18%Au 0.1 Pt 0.9 samples clearly shows that, although the amount of Pt is similar in both NPs, the T 50 temperature is decidedly different. This finding suggests that Au atoms compete with Pt atoms at the NP surface, which, in turn, reduces the overall catalytic activity of the NPs with increasing amount of Au. In general, the catalytic activity of the Au x Pt 1−x NPs is in agreement with the published activity of pure Pt catalysts 31,32 .
In other words, the observed T 50 for CO oxidation is closer to that of pure Pt catalysts than to the T 50 reported for pure Au/support catalysts in ambient temperature ranges 32 .

Discussion
BCML proved to be a powerful tool to synthesize monodisperse fcc alloy crystal structure NPs of desired composition, size, and interparticle distance with excellent precision. Most importantly, the ratio between Pt and Au that was measured in individual NPs was very consistent between the sampled NPs. It was possible to produce 0.2% Au x Pt 1−x /SiO 2 model catalysts by using a 3D BCML coating process. The catalytic activity of these Au x Pt 1−x / SiO 2 systems during CO oxidation was tested by temperature-controlled oxidation. The level of oxidation activity was determined on the basis of the T 50 temperature and the systems ranked according to their activity. The activity of each catalyst as prepared by BCML did not show any ageing effects for temperatures up to 400 °C. The T 50 temperature revealed further that catalytic activity correlates with the amount of Pt in the NP. We assume that the observed changes in oxidation activity are due to a higher surface activity of Au atoms on the surface of alloy NPs as Au atoms compete with Pt atoms for surface occupation. However, no core shell structure or other phase segregation structures could be discovered in our alloy NPs. The ability to selectively compose individual metal alloy NPs with high precision opens up new possibilities to study the catalytic performance of alloy noble metal catalysts. It greatly improves the understanding of the role that individual NP components play. Additionally, the role of the support could be the focus of future catalytic experiments, in which the interaction between the support and the noble metal NPs, known as the "Strong Metal-Support Interaction" (SMSI), and the reactivity of the perimeter interface could be further examined. Specifically, supports like TiO 2 or CeO 2 are known to enhance the catalytic activity of noble metal NPs 5,10,33,34 . With this knowledge further catalytic applications can be considered.   To immobilize the Au x Pt 1−x NPs on the silica powder (mean density of 300-400 g/l), 3 g of silica powder was mixed with 15 ml of the loaded micellar solution and sonicated for 30 seconds. The wet powder was then put into a glass column sealed with a PFTE frit (pore size 30 μ m) and pressed through the powder using an Argon flux. After drying, the powder was treated with pure H-plasma (45 min, 300 W, 0.4 mbar).
Annealing. The Au 0.5 Pt 0.5 NPs immobilized on Y:ZrO 2 (100) wafers were annealed in air or under CO 2 atmosphere for 22 h at 400 °C (heating rate of 5 K/min). The samples in the CO 2 atmosphere were evacuated in a glass tube filled with CO 2 (435 mbar/RT).
Characterization. The arrangement and order of the NPs on the substrates in 2D and 3D was analyzed by SEM (Zeiss Ultra 55, 2 and 5 kV). The in-lens and ESB detector were used for imaging. The size distribution of the Au x Pt 1−x NPs was measured using TEM (CM 200, Philips, 200 kV). For this, the NPs were immobilized on TEM Cu grids with a SiO 2 membrane (mesh 300, 20 nm, Plano GmbH). To analyze the NPs immobilized on silica powder, a small amount of the sample was mixed with Ethanol and 5 μ l was deposited onto TEM grids (Cu mesh 400) with a carbon film substrate. The TEM measurements to analyze the size distribution as well as the visual analysis of the powder structure were performed on a Philips 200 kV Transmission Microscope in BF, DF, and diffraction mode. EDX/HR-EDX measurements and HAADF imaging were performed on the SESAM (Sub-Electron-volt-Sub-Angstrom TEM, Zeiss, 200 kV) and ARM (JEOL-ARM200CF, 200 kV) microscopes. The EDX measurements of the NPs were performed on the same samples used for size distribution measurements. The crystal structure of the Au 0.5 Pt 0.5 NPs was measured by preparing cross-cut samples of single NPs on a Y:ZrO 2 (100) wafer employing a PIPS (precision ion polishing system). The composition of the NPs was also examined by EDX mapping, line scans, and spectra over single NPs. The TEM measurements were performed at the Stuttgart Center for Electron Microscopy (StEM).
Catalytic Measurements. The catalytic measurements were performed at RUBOKAT GmbH, Bochum.
The CO oxidation reaction was performed in a fully automated reactor capable of measuring transient kinetics (BELCAT B; BEL corp. Japan). Specifically, ion currents of CO 2 , CO, O 2 , and H 2 O molecular ions were measured. The reactor is attached to a gas mixing unit (BEL (J)) that uses helium as a diluent gas. The gas mixing unit was used to adjust the CO flow so that 1% CO gas was mixed with 1% O 2 gas. The reactor was operated with a theoretical oxygen excess of 100% for the reaction +  CO O CO 1 2 2 2 . Temperature-controlled oxidation was carried out cyclically, with a total volume flow rate of 20 Nml/min and a heating rate of 5 K/min from 25-400 °C. The initial weight was about 100 mg per catalytic measurement. The detector consisted of an online quadrupole mass spectrometer (Pfeiffer, GAM 400). The ion current of CO was corrected with the theoretical ratio of 11,4% of the produced CO 2 , because CO is a fragment of CO 2 .