A One‐Pot Route to Faceted FePt‐Fe3O4 Dumbbells: Probing Morphology–Catalytic Activity Effects in O2 Reduction Catalysis

The design and synthesis of faceted nanoparticles with a controlled composition is of enormous importance to modern catalyst engineering. Faceted FePt‐Fe3O4 dumbbell nanoparticles are prepared by a simple, one‐pot technique that avoids the need for expensive additives or preformed seeds. The faceted product consists of an FePt octopod and a cubic Fe3O4 lobe, of mean diameter 13.6 and 14.9 nm, respectively. The mass normalized activity for electrocatalytic oxygen reduction shows that this new structure types outperforms related catalysts in alkaline media. This work illustrates the power of morphology control and tailoring crystal facet abundance at the nanoparticle surface for enhancing catalytic performance.


Introduction
The synthesis of metal nanoparticles (NPs) with precisely controlled shapes that express selected crystal faces is highly desirable. [1] Such morphology manipulation can be a powerful way of tailoring electronic, optical, catalytic, [2][3][4][5] and magnetic properties. [6][7][8] Anisotropic NPs show asymmetric morphology, with one recently reported subset being nanopods, [9] in which NPs demonstrate a distinct branched morphology. An octopod (OP) is a cube-like nanopod with overgrowth at the corners giving eight branches. [10] Considering the wealth of desirable properties of Pt and its base-metal alloys, they are ideal candidates for morphology development with the aim of generating catalytically active and/or magnetically recyclable nanopods. [2,[11][12][13] When developing anisotropic NP syntheses, morphology can be controlled [14] to minimize the surface free energy of a catalytic ORR activity by adding an iron oxide phase to small Pt NPs. [38] They also harnessed the electron-donating ability of the oxide lobe in FePt-Fe 3 O 4 dumbbells to improve sensitivity for the electrocatalytic detection and reduction of dopamine. [39] Some work has attempted to combine the points above. For example, NiPt hexapods synthesized by a complex oxidative etching and CO gas adsorption methodology [40] proved viable for ORR. Likewise, CoPt 3 OPs showed impressive activity in CO hydrogenation. [24] Meanwhile, although anisotropic FePt NPs have seldom been prepared, Pazos-Pérez et al. reported FePt OPs for theranostics. [41] Although this synthesis required neither an initial seed particle nor expensive metal additives to induce branching, compositional control proved difficult; in spite of using a 2:1 Fe:Pt molar ratio, mean OP composition was Fe 12 Pt 88 . Indeed, reports of cubic or pseudospherical FePt NP syntheses have typically used a 2-3 molar excess of Fe precursor. This helped offset the formation of unreactive iron(III) oleates, which caused a lack of Fe in the final product. [42] In this work, the synthesis is reported of a range of complex heterobimetallic NPs through the simultaneous polyol reduction of Pt(acac) 2 and thermal decomposition of Fe(CO) 5 , as previously reported by Sun and co-workers. [42] Uniquely, it is shown for the first time that control can be exerted over faceting and alloying within NPs that demonstrate a dumbbell heterodimer structure. A simple and reproducible one-pot methodology is employed, which avoids the need for acidic etching or expensive additives. Not only is the selective generation of a faceted heterostructure reported, but also it proves possible to systematically vary the morphology of either dumbbell component through simple changes to synthetic parameters. The new morphology incorporates a unique combination of traits considered beneficial for catalysis. The advantages offered are explored for ORR in alkaline media and structureproperty relationships for a range of morphologies are compared.

Nanoparticle Synthesis
To the authors' knowledge, varying the availability of Fe for alloy formation during NP synthesis through simple surfactant variations has not been explored. Given the simplicity of faceted NP formation demonstrated using an equimolar mixture of oleic acid (OA) and oleylamine (OAm), [41] the preparation of FePt alloys was targeted under a range of surfactant ratios. Simultaneous reduction of Pt(acac) 2 and thermal decomposition of Fe(CO) 5 in exclusively OA (see Figures S1 and S2, Supporting Information) immediately achieved a morphology drastically different to OPs expected following previous FePt syntheses that used 50:50 OA:OAm. [41] Transmission electron microscopy (TEM) analysis of the products obtained shows a mixture of pseudospherical (d ave = 4.7 ± 0.88 nm), cubic and rod (l ave = 11.3 ± 3.0 nm) NPs. The introduction of OAm (OA:OAm 75:25, Figures S3 and  S4, Supporting Information), gave only polydisperse pseudospheres (d ave = 4.8 ± 0.9 nm). Meanwhile, equimolar OA:OAm ( Figures S5 and S6, Supporting Information) gave particles consistent with previously reported OPs [41] (d ave = 14.2 ± 1.5 nm) with a Pt-rich composition, denoted OP Pt . Finally, using an excess of OAm (OA:OAm 25:75, Figures S7-S10, Supporting Information), remarkably gave faceted heterodimeric dumbbells, as shown by scanning TEM-high angle annular dark field (STEM-HAADF) imaging in Figure 1a. STEM-HAADF data reveal that the brighter phase has maintained OP morphology, while the darker phase is cubic (OP d ave (tip to tip across the diagonal OP face) = 13.6 ± 1.4 nm; cubic d ave = 14.9 ± 2.1 nm). This unprecedented morphology will be referred to as OPD c . Finally, when using exclusively OAm (Figures S11 and S12, Supporting Information), brightfield (BF) TEM revealed predominantly pseudospheres (d ave = 3.5 ± 0.4 nm) containing also some nanowires. In both structures, a lighter phase surrounds a darker particle, suggesting a Pt x Fe y core and a Fe 3 O 4 shell.
Overall, these data reveal a strong influence of surfactant ratio on both product morphology and composition. In terms of developing anisotropic NPs, either a 50:50 (OP Pt ) or 25:75 (OPD c ) OA:OAm ratio proves desirable. This outcome agrees with reports by Chou et al., [10] where a mixture of OA and OAm was required to modulate the relative growth between the (111) and (100) facets of FePt. Chou found that OAm preferentially bound to the (100) facet (cube face). This allowed significantly faster growth on the (111) facet (cube corners) of a cuboctahedral seed and so produced branches. Alternatively, OA was found to interact with both facets indiscriminately, and so to produce a highly symmetric cuboctahedral morphology. Similarly, Chen et al. previously noted the preferential formation of FePt nanocubes in OA:OAm mixtures with the sequential addition of OA and OAm. Nanocubes were exclusively formed, with OA addition first allowing the overgrowth of the (100) cube face. [43] However, further to the purely morphology-directing nature of the surfactant mixture, the current work now demonstrates that varying OA and OAm concentrations can also allow tailoring of the final NP composition within these faceted structures. Such enhanced composition control can be achieved using excess OAm when higher levels of Fe inclusion are required. Hence, when OAm is in excess (OPD c ) or is used exclusively there is evidence of a lighter iron oxide phase in a dumbbell or core@ shell structure, respectively. These data are expected based upon the limited or nonexistent capability of forming unreactive Fe oleates in OAm-rich systems, leading to a higher concentration of reactive Fe precursor in the reactions in spite of the initial Fe(CO) 5 concentration being constant in all syntheses.
The structure observed in OPD c particles provides a unique combination of both anisotropic NPs that offer an unusual variety of crystalline facets, and dumbbells that allow access to multiple chemical surfaces within single uniform NPs. From Figure 1a, the octopod and cubic lobes are lighter and darker, respectively. The former is logically monometallic Pt or bimetallic FePt, with the latter expected to consist solely of Fe, which passively oxidizes upon atmospheric exposure. High-resolution (HR) TEM data in Figure 1b show a representative OPD c particle with indexed atomic planes (also Figure S13, Supporting Information). The octopods and cubic lobes show atomic spacing consistent with fcc Pt or FePt [111] and Fe 3 O 4 [113], respectively. While it is challenging to distinguish microscopically between Fe 3 O 4 and γ-Fe 2 O 3 , passive oxidation will likely form Fe 3 O 4 , whereas γ-Fe 2 O 3 needs strongly oxidizing conditions. [44] Figure 1c,d shows the STEM energy dispersive X-ray (EDX) analysis of OPD c NPs (also Figures S14 and S15, Supporting Information). This establishes that, of the two possible This understanding led to the targeting of an OP structure with an ≈Fe 50 Pt 50 composition by repeating the OPD c synthesis by retaining a 25:75 OA:OAm ratio but using only half as much Fe(CO) 5 . Of the three rational outcomes of this experiment-1) Pt-Fe 3 O 4 OPDs, 2) FePt-Fe 3 O 4 OPDs containing less Fe, or 3) FePt OPs-the desired outcome was realized; Figure 1e shows the product to be octopods (OP FePt , d ave = 17.7 ± 2.0 nm; Figures S19 and S20, Supporting Information). This gives an indication of the OPD c growth mechanism, suggesting initial FePt octopod formation, with any remaining iron precursor forming the Fe 3 O 4 component of the dumbbell structure. Furthermore, manipulating the surfactant ratios makes it possible for the first time to create an FePt octopod with a controlled morphology and  Figure  S21, Supporting Information). This represents a significant increase in Fe contribution compared to that in OP Pt (Fe 25 Pt 75 , Figure S22, Supporting Information). Finally, Figure 1g compares the powder X-ray diffraction (PXRD) patterns of OPD c and OP FePt . These data illustrate a lack of Fe 3 O 4 within the octopod component of the former, supporting TEM analysis.
While focusing on parameters that can achieve the desirable OPD c structure, the influence of temperature on morphology and composition has also been investigated. Initial preparation of OPD c NPs involved heating a reaction mixture to the intermediate temperature of 240 °C for 1 h followed by 260 °C for 2 h (X = 240 in Figure 2a).  Figures S27-S31, Supporting Information) reinforce the conclusion of the temperature study above. Namely, reactions in which the concentration of active Fe is reduced, either by lowering intermediate reaction temperature or using less Fe(CO) 5 , yield products that incorporate a dendritic component. Indeed, dendrite formation within an Fe-depleted reaction mixture (see the Supporting Information) indicates a possible mechanism for DD c growth at 180 °C and is evidenced in Figure 2b. In this case, the Fe(CO) 5 is present within the reaction mixture, yet the low temperature restricts decomposition from generating reactive iron in situ. Thus, the DD c structure more closely resembles the morphology of Pt dendrites formed in the absence of Fe(CO) 5 . Upon increasing the temperature of the 180 °C system to allow reaction completion, the final DD c structure ( Figure S23, Supporting Information) is achieved, containing an intermixed FePt dendrite phase. Taken together, data illustrate how the synthetic method developed here can selectively manipulate the FePt component of a dumbbell to give the highly desirable faceted OP structure whilst also maintaining that component's ≈Fe 50 Pt 50 composition.

Oxygen Reduction Electrocatalysis
As noted earlier, the activity of modern Pt catalysts can be improved by faceting, increasing surface area, and creating heterostructures with modulated electronic properties and reactive heterojunctions. With this in mind, the potential the newly created OPD c morphology offered in terms of manipulating all these key traits was tested in electrocatalytic ORR. The use of alkaline media for ORR is an emerging area and offers higher stability for the Fe 3 O 4 component compared to traditionally employed acidic media. [45] To understand the influence of the unique morphology, chemical composition and dual surface of the OPD c , their activity was compared with those of structures exhibiting only one trait with the potential to improve ORR activity. To interpret the role of morphology within a dumbbell structure, pseudospherical FePt-Fe 3 O 4 dumbbells (denoted PSD) were made based on a reported method ( Figures S32  and S33, Supporting Information). [39] Interestingly, when this methodology was modified to use 25:75 OA:OAm, echoing the synthesis of OPD c , an OPD g product resulted ( Figures S34  and S35, Supporting Information). The key difference between OPD c and OPD g is in the morphology of the Fe 3 O 4 phase; that in OPD c is uniform and cubic, whereas that in OPD g is globular and indistinct. Meanwhile, a comparison of OP FePt and OPD c probed the effect of the Fe 3 O 4 lobe. In addition, the influence of octopod composition was studied by comparing OP Pt with OP FePt . Finally, a comparison of FePt-Fe 3 O 4 PSDs and OPD c elucidated the importance of a faceted structure. NPs used for electrocatalytic ORR testing are summarized in Figure 3. Figure 4a shows linear scan voltammograms for ORR at potentials of −0.05 to −0.3 V versus saturated calomel electrode (SCE; for conversion to reversible hydrogen electrode see the Experimental Section) for the five morphologies summarized in Figure 3 and also industry standard Pt. The electrocatalytic tests were performed in a three-electrode setup with a rotatingring disk working electrode in 1 m aqueous KOH under saturated oxygen conditions. Each sample was measured at room temperature using an identical Pt loading of 0.5 µg per disk. At −0.06 V, the onset potential of the OPD c is competitive with the commonly used industrial standard 40% Pt on C (Sigma Aldrich), and is at a substantially more positive potential than for the other samples. This points to a more active morphology that enables the reduction of oxygen while requiring less voltage at the same mass loading of platinum. The remaining NP morphologies show similar onset potentials, with the OPD g morphology being slightly more favorable. Taken together, these data point to OPDs being a promising morphology for ORR. Figure 4b shows mass activity normalized for Pt at −0.2 V versus SCE. The performance of OPD c exceeds those of OPD g and Pt. The 300% increase in mass activity suggests either that the morphology of the Fe 3 O 4 phase is a critical determinant of mass activity or that active sites are generated at the FePt-Fe 3 O 4 interface. Moreover, it is noteworthy that the activity of Fe-rich OP FePt is boosted by 400% by introducing a monodisperse cubic Fe 3 O 4 phase (OPD c vs OP FePt ). This can be explained by Fe within Fe 3 O 4 donating electron density to Pt, as shown by work on nonfaceted systems. [38] Similarly, Figure 4b also allows a comparison of OP Pt (high Pt content) and OP FePt (≈50:50 FePt), showing a significant increase in mass activity with increased Fe content. These data provide further support for the desirable composition of the octopod phase within OPD c .
Finally, FePt-Fe 3 O 4 PSDs [24] show relatively low activity. This is surprising since they are substantially smaller than the OPD particles, so for an equivalent mass of Pt a larger surface area and a higher activity would be expected. Further experimental data can be found in the Supporting Information ( Figures S36-S41, Supporting Information). ORR can proceed via a 2e − or 4e − reduction to yield H 2 O 2 or H 2 O as final products, respectively. The corrosive nature of H 2 O 2 renders its production undesirable in fuel cell technologies. ORR activity was therefore tested on a rotating-ring disk electrode to simultaneously measure peroxide production ( Figures S44 and S45, Supporting Information, for Faradaic efficiencies). [46] At −0.2 V, there is little difference between the selectivity of each morphology for a 4e − reduction. However, at more negative potentials, a greater selectivity for H 2 O is demonstrated by catalysts that incorporate a higher concentration of Fe. Overall, the OPD c particles show far superior activity with a minimal compromise in selectivity for H 2 O. A comparison of the mass activity of the new OPD c with both PSD Pt-Fe 3 O 4 [38] and industrial standard Pt shows that OPD c has a far greater mass activity at −0.2 V versus SCE. This superior morphology can be prepared by a one-pot technique that avoids the need to isolate and purify Pt seeds, ensuring both greater reproducibility and atom efficiency. A combination of density functional theory (DFT) and spectroscopic studies are being initiated to explore the effects of faceting and the nature of active sites in the OPD c morphology.

Conclusion
In conclusion, a facile one-pot reaction that achieves uniform and complex NPs is reported. These data reveal not only a bimetallic alloy, but also a dumbbell morphology that shows selective faceting on both lobes. Through temperature control, it is possible to selectively manipulate the morphology of the FePt lobe while maintaining a desirable Fe 50 Pt 50 composition. In ORR under alkaline conditions, this new material, FePt-Fe 3 O 4 octopod dumbbells, substantially outperforms a series of comparable samples, illustrating the importance of composition, morphology, and interface control as a means of enhancing catalytic activity.
PSD Synthesis: Pt(acac) 2 (0.25 mmol, 98.32 mg), oleic acid (2 mmol, 0.64 mL), oleylamine (2 mmol, 0.66 mL), and octadecene (5 mL) were mixed under Ar at room temperature. The temperature was increased to 105 °C for 10 min to remove H 2 O and ensure dissolution of the precursors. Fe(CO) 5 (0.1 mL) was added at 120 °C and the temperature increased to 300 °C at a heating rate of 5 °C min −1 , where it was maintained for 1 h before cooling to room temperature.
Materials Characterization: For TEM analysis, extensive washing of NPs by sedimentation in EtOH was followed by sonication in hexane and drop-casting on a lacey carbon copper grid. The grid was plasma cleaned for 40 s before TEM analysis was performed using a Thermo Scientific (FEI) Talos F200X G2 TEM or FEI Philips Tecnai 20, 200 keV with 70 µm objective aperture for brightfield TEM and pot size 6 STEM EDX acquisition. PXRD patterns were measured on a PANalytical Empyrean diffractometer fitted with an X'celerator detector and using a Cu-Kα 1 (λ = 1.5406 Å) source, using a step size of 0.002° and a scanning speed of 0.022° s −1 at 40 kV and 40 mA.
ORR Electrocatalysis: FePt/C ink was prepared following a published literature method for Pt-Fe 3 O 4 dumbbells to aid comparison. [38] Inductively coupled plasma-optical emission spectroscopy analysis was used to determine the concentration of Pt within the FePt/C inks, which were then diluted as necessary to generate a Pt loading of 0.5 µg µL −1 ink. 1 µL of FePt/C ink was applied to a freshly polished glassy carbon rotating-ring disk electrode (RRDE, Pt ring, Pine Research Instrumentation, E6R1 ChangeDisk) that was being rotated at 200 rpm with a mild N 2 gas flow to promote an even coverage of the disk (diameter 5 mm). The electrode was left to dry for a minimum of 3 h. O 2 gas was bubbled through ≈50 mL of 1 m aqueous KOH solution for 30 min before measuring for ORR activity. LSV measurements were taken at room temperature against a Ni foam counter electrode and saturated calomel electrode (SCE) reference electrode with a rotation speed of 1600 rpm and sweep rate of 5 mV s −1 . A minimum of five cyclic voltammograms were recorded to condition the electrode prior to LSV measurements. An Ivium Compactstat Bipotentiostat was used for all measurements. No iR compensation was performed, but the high conductivity of the electrolyte suggests that minimal potential shifts would be expected. [47] The industrial standard was prepared in the same manner, leading to 2.5 µg cm −2 Pt deposited on the electrode. A diffusion limited current density of −0.495 mA at 0.6 V versus reversible hydrogen electrode (RHE) was measured. Unsurprisingly, this is lower than observed in the literature [48] considering that the Pt loading is 1-2 orders of magnitude lower. The low loading was chosen in line with work on similar heterobimetallic dumbbell particles for ease of comparison. [38] SCE to RHE conversion was achieved using Equation (1

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.