Fabrication of electrocatalytic Ta nanoparticles by reactive sputtering and ion soft landing

Nanoparticles (NPs) are scientifically interesting because of their tunable chemical and physical properties which result from variations in structure and reactivity at the nanometer length scale.¹ The size- and structure-dependent properties of NPs make them promising candidates for a number of technological applications. For instance, NPs have possible applications in plasmon-enhanced photovoltaic devices for efficient generation of solar energy² and in inks for fast and environmentally responsible printing of electronic circuit components.³ Catalysis, however, is where NPs are most widely used in the chemical sciences.⁴ This is because NPs have large surface areas and may be prepared from numerous elements with selected morphologies that express both catalytically active and selective surface facets.⁵ In many ways, NPs epitomize building blocks with tunable properties which may be incorporated into larger mesoscale structures⁶ that exhibit emergent behavior not observed in conventional bulk matter.

Synthesis of NPs in solution is generally regarded to be the most practical way to prepare scalable quantities of these materials for commercial applications.⁷ This approach requires the use of organic ligands to maintain NP solubility, arrest growth at a certain size, direct the formation of specific crystal facets, and prevent agglomeration.⁸ The ligands, however, must typically be removed to expose the active metal surface and enable immobilization of NPs on supports for applications such as catalysis, chemical sensing, and separations.⁹ Removal of ligands often involves treatments at elevated temperatures or the application of harsh oxidizing chemicals.¹⁰ In either case, this step increases the complexity and cost involved in using NPs and causes added uncertainty in the size and structure of the NPs that were so carefully prepared in the first place.

To address this challenge inherent to NPs prepared in solution, complementary physical techniques have been developed to synthesize bare NPs in the gas-phase without solvents or ligands. For example, physical NP preparation methods now include gas aggregation,¹¹ laser vaporization,¹² magnetron sputtering,¹³ and pulsed arc synthesis.¹⁴ These techniques are generally implemented at reduced gas pressures and coupled with ion guides and mass filters which allow selection of one ionic species from the larger distribution of charged NPs produced at the source. Until recently, mass-selected ion currents produced by physical techniques were in the range of picoamperes (10⁻¹² A) which restricted their use to preparing well-defined model systems for fundamental research in surface science and catalysis. However, advances in ion production, mass-selection, and transmission have enabled currents of up to 6 nanoamperes (10⁻⁹ A) of mass-selected cluster ions to be delivered to surfaces.^15–17 The resulting increase in the rate of deposition now provides an opportunity for soft landing of mass-selected ions to enter the manufacturing of high-value components.

Magnetron sputtering combined with gas aggregation, in particular, has evolved to be a versatile approach for synthesizing bare ionic NPs over a range of sizes and compositions not obtainable in solution. Haberland and co-workers introduced the original design in 1991 and subsequently used it to study the preparation of thin films on surfaces by energetic cluster ion deposition.¹³ An updated design was presented by Palmer and co-workers in 2005 which enabled radiofrequency (RF) sputtering of targets as well as a variable gas aggregation distance.¹⁸ This source expanded the elements which could be sputtered to include insulators and provided greater control over the size and morphology of NPs formed in the aggregation region.¹⁹ In recent years, modified versions of this source design have appeared which enable pulsed sputtering,²⁰ pulsed gas aggregation,²¹ and sputtering of multiple independent targets in the same region of gas aggregation.^22–24 Collectively, these sources have been used to prepare a range of nanomaterials on surfaces for studies in catalysis,^25–27 photovoltaics,²⁸ magnetism,²⁹ memory,^30,31 cluster-surface interactions,³² hydrophobic coatings,³³ and “nanoportals” for hydrogen storage.³⁴ In addition, sputtering of multiple independent targets in the same region of gas aggregation has been used to prepare Janus and core-satellite Si-Ag NPs²³ as well as stable Pd-core MgO-shell NPs for catalytic methanol oxidation.³⁵ This approach, however, produces NPs with a distribution of sizes and compositions that must be filtered in the gas phase to obtain precisely defined NPs on surfaces.

Soft landing of mass-selected ions provides unparalleled control of the size, composition, and charge state of complex ions deposited onto surfaces.^32,36–39 The idea of utilizing low energy ions to selectively modify surfaces was demonstrated by Cooks and co-workers in 1977 by depositing small sulfur-based ions onto metal substrates.⁴⁰ In the intervening years, several groups have demonstrated that complex ions including peptides,⁴¹ proteins,^42,43 protein assemblies,⁴⁴ organometallic complexes,^45–47 metal clusters,^15,48,49 and NPs^22,50 may be soft landed onto substrates intact with preservation of their properties. For instance, in our laboratory, we provided evidence that both the size and charge state of cationic phosphine-ligated gold clusters may be controlled by soft landing of mass-selected ions onto self-assembled monolayers (SAMs) on gold terminated with different functional groups.^51,52 In addition, we showed that the charge retention of anionic polyoxometalate clusters on surfaces is controlled by their electron binding energy rather than the electron transfer properties of the SAMs.⁵³ Mass-selection allows ionic species with well-defined charge and chemical composition to be delivered to surfaces. In addition, for species that are deposited from solutions, electrospray ionization combined with ion soft landing avoids any contamination resulting from the presence of neutral molecules, residual reactants, solvent, and counterions. Furthermore, it allows the deposition of large non-volatile species that may not be amenable to traditional atomic and molecular layer deposition.⁵⁴ Employing soft landing of mass-selected ions, it is possible to study the size, composition, coverage, and support-dependent properties of complex ions such as NPs and clusters on surfaces with a high level of control over the deposition process, which is useful for establishing structure-function relationships in applications such as catalysis and energy storage.

The oxygen reduction reaction (ORR) is an important catalytic process that takes place at the cathode in proton exchange membrane fuel cells (PEMFCs) and currently limits their commercial viability.⁵⁵ One factor limiting the efficiency and commercialization of PEMFCs is the large expense associated with the precious Pt-based catalysts required to promote the sluggish ORR. As a result, substantial effort has been expended to understand the mechanism of the ORR and identify cheaper Pt-alloy and non-noble metal catalysts for this reaction.⁵⁶ For example, on the fundamental side, soft landing of mass-selected ions has been used to examine the size-dependent activity of sub-nanometer metal clusters^57,58 as well as the composition-dependent behavior of larger NPs toward the ORR.⁵⁹ As early as 1973, Boudart and co-workers reported a seminal study in which tungsten carbide was found to exhibit similar catalytic properties to Pt.⁶⁰ The following year it was shown that the electronic density of states near the Fermi level of tungsten carbide more closely resembles that of Pt than W.⁶¹ These discoveries suggested the possibility of using earth-abundant metal carbides as cheaper substitutes for expensive precious metals. Correspondingly, numerous investigations were launched into the catalytic properties of these and related materials including studies of Ta-oxide based compounds by Ishihara and Seo.^62,63

In this study, reactive magnetron sputtering of Ta in the presence of different gaseous hydrocarbons was evaluated as a new physical route for the synthesis of non-noble metal NP catalysts for the ORR. Organic-inorganic hybrid NPs were prepared through sputtering of Ta in a partial pressure of 2-butanol, heptane, or m-xylene and soft landed onto carbon electrode surfaces. Pure Ta NPs were also prepared for comparison with the hybrid species. Cyclic voltammetry (CV) experiments revealed that the Ta-heptane and Ta-xylene NPs are active toward O₂ reduction at acidic conditions and stable toward repeated potential cycling at room temperature. In comparison, the pure Ta and Ta-butanol NPs are not active. Height and size measurements of the supported NPs on graphite and carbon grids obtained with atomic force microscopy (AFM) and transmission electron microscopy (TEM) revealed narrow size distributions with some agglomeration but little coalescence into larger species. High resolution TEM images showed that the Ta-butanol and Ta-heptane NPs have crystalline cores surrounded by amorphous regions. The Ta-xylene NPs, in comparison, were found to be amorphous throughout without the surrounding material that was observed for the other reactively sputtered NPs in this study. X-ray photoelectron spectroscopy (XPS) indicated that while the percent composition of Ta, C, and O in the NPs varies for the different hydrocarbons the chemical states of the elements are similar. The difference in reactivity of the NPs is attributed to their Ta/C ratios which are 0.15, 0.46, and 0.54 for the Ta-butanol, Ta-heptane, and Ta-xylene NPs, respectively, with a larger Ta/C ratio associated with high catalytic activity. Overall, the results suggest that reactive sputtering combined with gas aggregation and ion soft landing constitutes a promising approach for the preparation of lower cost catalytically active and durable non-noble metal NPs without the use of solvents and organic ligands.

A comprehensive description of the customized Nanogen NP source and Q-Prep 500 deposition system obtained from Mantis Deposition Ltd. (Oxfordshire, UK) is presented in a preceding publication.⁶⁴ Concisely, the NP source utilizes one 2″ diameter sputtering target mounted on a linear translator within a region of gas aggregation. Organic-inorganic hybrid NPs were prepared by reactive direct current (DC) magnetron sputtering of a Ta target (circular, 2″ diameter, 0.15″ thick, 99.95% purity) obtained from Plasmaterials (Livermore, CA) in a controlled flow of ultra-high purity (UHP, 99.999%) Ar and He gas purchased from Matheson Tri-Gas (Newark, CA). To accomplish reactive sputtering, a flow of 2-butanol, heptane, or m-xylene (ACS reagent grade, Sigma-Aldrich, St. Louis, MO) was introduced into the aggregation region from a sealed glass tube using a heated flow controller. Prior to use, the liquid reagents were degassed through multiple freeze-pump-thaw cycles. The DC sputtering current applied to the Ta target, the flow rates of Ar, He, and the reactant gases as well as the position of the linear translator in the aggregation region were optimized to maximize the production (total anion current) of organic-inorganic hybrid NPs delivered to the surfaces. The values of the source parameters are provided in Table I.

After exiting the aggregation region, the anionic NPs pass through a 5 mm inner diameter skimmer and into the high-mass MesoQ frequency scanning quadrupole mass filter. The narrower distribution of NPs (typically full width at half maximum (FWHM) ∼1 nm) mass-selected from the overall distribution (FWHM ∼2 nm) is transmitted by the MesoQ filter, extracted, and focused using a set of two custom-built einzel lenses in series which direct the beam of NPs to a circular spot on the surface approximately 25 mm in diameter. The spatially focused NPs are soft landed onto substrates connected to ground potential through a conductive backing plate. The substrate holder, which contains multiple surfaces, is rotated at 20 rotations per minute to ensure even coverage and isotropy of the samples. This configuration allows the same NPs to be deposited simultaneously onto different substrates that may be used for surface, structural, and electrochemical characterization. The uniformity of NPs across the deposited spot depends both on the nature of the surface and the total NP coverage as has been discussed in detail in a previous publication.⁶⁴ The kinetic energy of the NPs is imparted primarily by the expansive gas flow through the exit aperture of the aggregation region and is estimated to be <1 eV which ensures gentle deposition of NPs without fragmentation.⁶⁵ The quantity of soft landed material is determined by monitoring the ion current during deposition both at a wire mesh located at the exit of the quadrupole filter and at the surface with a Keithley 6485 picoammeter (Cleveland, OH). The wire mesh captures ∼5% of the ion current that is measured at the surface. It is assumed that most of the anionic NPs produced in these experiments are singly charged based on the previous findings of Haberland⁶⁶ and Biederman.⁶⁷ 

Square 2 mm thick glassy carbon (GC) substrates, 10 mm × 10 mm, with both sides lapped and one side diamond polished to <50 nm root mean square (RMS) surface roughness were purchased from HTW Hochtemperatur-Werkstoffe GmbH (Thierhaupten, Germany). All surfaces were ultrasonically washed in 99.5% acetone obtained from Sigma-Aldrich for 30 min. The surfaces were then rinsed with pure methanol, dried with a stream of N₂, and placed in an ultraviolet (UV) cleaner (Boekel Scientific, Feasterville, PA) for 20 min. The surfaces were then loaded into the soft landing instrument, degassed in vacuum (base pressure ∼10⁻⁹ Torr) at a temperature of 800 °C for 30 min, and allowed to return to room temperature in vacuum overnight. The surfaces were sputtered with Ar RF plasma at a power of 50 W in a pressure of 10⁻² Torr of UHP Ar for 1 min immediately before starting NP deposition. Circular (diameter 3.05 mm) lacey thin carbon film coated copper microscopy grids (400 mesh) were purchased from Pacific Grid Tech (San Francisco, CA) and used as received. Square highly oriented pyrolytic graphite (HOPG) surfaces, 10 mm × 10 mm, 1.5 mm thick, ZYB quality were obtained from NT-MDT (Moscow, Russia) and cleaved using adhesive tape to expose a fresh surface prior to soft landing of NPs.

An Asylum MFP-3D BIO AFM (Santa Barbara, CA) was used to characterize the surface coverage and monodispersity in height of soft landed organic-inorganic hybrid NPs on HOPG. The probe was operated in tapping mode with 330 kHz tips that were purchased from Nanosensors (Neuchatel, Switzerland). The AFM data were converted to 3- and 2-dimensional (3D, 2D) color map images and line profiles while topography distributions were extracted using the WSxM software available for free from Nanotech Electronica (http://www.wsxmsolutions.com, Madrid, Spain).^68–70 

TEM and high-angle-annular-dark-field (HAADF)-STEM were used to characterize the size and morphology of NPs soft landed onto carbon microscopy grids. Images were collected with a probe-corrected FEI Titan 80-300 microscope operated at 300 kV and prepared using the Gatan Digital Micrograph software (Pleasanton, CA). The particle size distributions were determined by manually measuring the diameters of at least 50 randomly selected NPs.

XPS measurements were performed with a Physical Electronics (PHI) Quantera Scanning X-ray Microprobe (Eden Prairie, MN), which employs a focused monochromatic Al Kα X-ray (1486.7 eV) source for excitation and a spherical section analyzer. The instrument has a 32 element multichannel detection system. A 100 W electron source generates an X-ray beam which is focused to 100 μm diameter and rastered over a 1.2 × 0.1 mm rectangular area of the sample. The X-ray beam was incident normal to the sample and the photoelectron detector was at 45° off-normal. High-energy resolution spectra were collected using a pass energy of 69.0 eV with a step size of 0.125 eV and a ±20° analyzer acceptance angle. The binding energy scale was calibrated using the Cu 2p_3/2 feature at 932.62 ± 0.05 eV and Au 4f_7/2 at 83.96 ± 0.05 eV for known standards. XPS measurements were acquired at different spots on the HOPG surfaces with multiple measurements obtained at each spot. XPS spectra were referenced to an O 1s binding energy of 530.7 eV in all cases. To quantify the elemental concentrations, the peak areas were determined after background subtraction using a function to subtract the peaks from the spectral background. The spectra of adventitious adsorbates were estimated by performing XPS measurements on an HOPG surface without NPs. All analyses were performed at room temperature.

CV measurements were implemented with a Princeton Applied Research VersaStat4 potentiostat (Princeton, NJ) in a three-electrode configuration with Ag/AgCl as a reference electrode in saturated KCl, a Pt wire as the counter electrode, and the NPs on GC as the working electrode. The GC working electrode was placed in a modified plate material evaluation cell (catalog no. 011951, ALS, Japan). The liquid electrolyte consisted of 0.5 M H₂SO₄ (Sigma-Aldrich, St. Louis, MO) in deionized water obtained from an Ultrapure Milli-Q system (resistivity: 18.2 MΩ-cm at 25 °C). The electrolyte solution (∼1 ml) was purged with UHP Ar or O₂ for 15 min before and during CV measurements. A minimum of 10 CV cycles were acquired for each of the surfaces at a scan rate 100 mV s⁻¹ in the potential range of 1.5 to 0.0 V versus the regular hydrogen electrode (RHE). All CV experiments were conducted at room temperature.

To evaluate the ability of reactive DC sputtering of metals in the presence of hydrocarbons to produce novel organic-inorganic hybrid NPs with catalytic properties, soft landing experiments were performed using Ta and three model molecules as the reactant gases: 2-butanol, heptane, and m-xylene. 2-butanol was selected as a representative secondary alcohol, heptane was chosen as a typical saturated linear alkane, and xylene was employed as a model aromatic hydrocarbon. In addition, pure Ta NPs sputtered with only Ar and He were prepared for comparison to the NPs produced by reactive sputtering. The pure Ta NPs were prepared by DC sputtering of a Ta target at a current of 85 mA in a flow of 20 standard cubic centimeters per minute (sccm) of Ar sputtering gas and 85 sccm of He carrier gas. The linear translator was set at a position of 80 mm in the aggregation region with respect to the exit aperture. These optimized source conditions produced a negative ion current of ∼950 pA of mass-selected anionic 5 nm diameter Ta NPs (∼Ta₃₆₄₀, m/z ∼ 660 000) that was measured at the substrates for 10 min resulting in coverage of 3 × 10³ ions μm⁻².

The Ta-butanol NPs were synthesized by DC sputtering of a Ta target at a current of 80 mA in a flow of 20 sccm of Ar, 7.5 sccm of He, and 0.4 sccm of 2-butanol reactant gas. The 2-butanol reactant molecules, which have a vapor pressure of 1.67 kPa at a temperature of 20 °C, were introduced into the sputtering zone through a 6″ long, 1/8″ diameter stainless-steel tube with its opening near the exit aperture of the gas aggregation region. Introducing the reactant gas at this position downstream from the sputtering zone helped to prevent poisoning of the target surface which can terminate the deposition process. Introduction of the reactant gas typically results in an increase of the sputtering voltage of the pure metal in Ar (∼50 V) which plateaus after some time (∼5 min) and may be maintained throughout the deposition by minor adjustments provided that the flow rate of reactant gas is optimal. If the flow rate is too high the deposition current will decrease with time as the target becomes poisoned. The linear translator was set at a position of 55 mm in the aggregation region with respect to the exit aperture. These optimized source conditions produced a negative ion current of ∼280 pA of mass-selected anionic Ta-butanol NPs that was measured at the substrates for 20 min resulting in a surface coverage of 3 × 10³ ions μm⁻². The negative ion current of the NPs obtained by reactive sputtering was maintained within a limited range of ±5% by small adjustments to the Ar and He gas flow rates (±2 sccm) and sputtering current (±5 mA) throughout the deposition. Based on the diameter of the NPs measured with AFM and TEM, and the elemental composition determined by XPS (described in detail later), it is possible to estimate an approximate molecular formula of ∼Ta₆₁₀C₃₉₀₀O₁₃₅₀⁻ for the 2-butanol NPs mass-selected for deposition (m/z ∼ 178 800).

In the next set of reactive sputtering experiments, Ta-heptane NPs were synthesized. Heptane has a vapor pressure of 5.33 kPa at a temperature of 20 °C, approximately three times that of 2-butanol (1.67 kPa), and was selected as a longer-chain saturated hydrocarbon that does not contain oxygen. With the linear translator located 75 mm from the exit aperture of the aggregation region a negative ion current of ∼240 pA was focused at the surfaces for 40 min resulting in soft landing of 6 × 10³ ions μm⁻². While an identical flow rate of Ar (20 sccm) and sputtering current (80 mA) were used for the synthesis of both Ta-butanol and Ta-heptane NPs, the flow rate of He carrier gas and heptane was set lower at 5 sccm and 0.2 sccm, respectively, in order to optimize the total ion current of Ta-heptane NPs. In addition, the position of the linear translator was increased from 55 mm for Ta-butanol to 75 mm for the Ta-heptane NPs. Decreasing the flow rate of He carrier gas is expected to increase the residence time of the NPs in the aggregation region which may result in the formation of larger NPs. At the same time, decreasing the He flow rate may increase the mean free path of the NPs resulting in fewer collisions leading to cooling, nucleation, and growth.⁷¹ As will be shown later, for the Ta-heptane NPs, the latter effect was dominant leading to the formation of slightly smaller NPs with reduced He gas flow compared to the Ta-butanol NPs. Concomitantly, increasing the aggregation length may result in longer NP residence times and more cooling collisions which typically causes formation of larger NPs. In the case of the Ta-heptane NPs, however, it appears that any effect of the longer aggregation region was offset by the influence of the lower flow rates of both He and heptane resulting in formation of smaller NPs. The quadrupole mass-filter was set to transmit NPs with m/z ∼ 109 000 assuming that the particles have an estimated molecular formula of ∼Ta₅₀₀C₁₀₅₀O₃₇₀⁻ based on their measured size and elemental composition described later.

M-xylene was selected as another reactant gas to investigate the potential of aromatic hydrocarbons for the synthesis of catalytic NPs. M-xylene has a vapor pressure of 1.2 kPa at a temperature of 20 °C which is similar to that of 2-butanol (1.67 kPa) and substantially smaller than that of heptane (5.33 kPa). With the linear translator set at a position 70 mm from the exit aperture of the aggregation region, an ion current of ∼300 pA was directed at the substrates for 30 min resulting in a total coverage of 3 × 10³ ions μm⁻². It is worth noting here that to produce an ion current of Ta-xylene NPs similar to that of the Ta-butanol and Ta-heptane NPs it was necessary to sputter Ta more aggressively at a current of 140 mA. While the reason for this higher sputtering current is not known, it is possible to speculate that the chemical composition of the surface of the Ta target may evolve from metallic Ta to substoichiometric Ta_xO_y and Ta_xC_y in the early stages of deposition depending on the composition and flow rate of the reactant gas. The influence of reactant gases on the nucleation and growth of metal NPs in a similar aggregation source has been examined by Polasek and co-workers among others.^72–75 Their study revealed that a small flow of O₂ may substantially increase the NP deposition rate by providing more stable metal oxide seeds for further nucleation while a large flow of O₂ may instead result in excessive oxidation of the target (poisoning) and reduced NP formation. Analogous behavior was identified in compound thin films prepared by reactive magnetron sputtering. Specifically, Safi and co-workers explained that at low flow rates of reactant gases, most of the molecules are gettered by the sputtered metal and the partial pressure of reactant gas remains stable. However, if the flow of reactant gas exceeds the gettering rate of the sputtered metal the composition of the target changes which can result in reduced sputtering yield.⁷⁶ A survey of the literature reveals that the sputtering yield of Ta₂O₅ is around 1.2 atoms/ion for 500 eV Ar⁺ ions^77,78 while that of metallic Ta is approximately 0.6 atoms/ion at 600 eV.⁷⁹ Therefore, oxygen present in 2-butanol may cause partial oxidation of the Ta target which may increase the overall rate of O sputtering and metal oxide NP formation. In comparison the sputtering yield from tantalum carbide is 6 × 10⁻³ atoms/ion for Ta and 9 × 10⁻² atoms/ion for C employing He⁺ ions at around 500 eV.⁸⁰ The lower sputtering yield of Ta compared to C from tantalum carbide, therefore, may explain the need for a higher sputtering current to generate an equivalent Ta deposition rate for the Ta-xylene NPs. The NPs were deposited with the mass-filter set to transmit NPs with m/z ∼ 99 000 assuming an estimated molecular formula of ∼Ta₄₅₀C₈₀₀O₄₇₅ based on the measured NP size and elemental composition described in the section on compositional analysis.

Individual GC surfaces were coated with a predetermined coverage of pure Ta, Ta-butanol, Ta-heptane, and Ta-xylene NPs. As GC is a common and well-characterized electrode material, it serves well as a support for the NPs so that their electrochemical activity may be measured in solution using CV. The CVs obtained for the reference pure Ta NPs on GC are presented in Fig. 1(a) and reveal only minor features during purging with Ar consistent with the presence of a small number of oxygenated functional groups on the surface of GC. In comparison, during purging with O₂, a small but observable reduction current (∼8 μA) was recorded at around 0 V potential. The reduction potential and current for the pure Ta NPs, therefore, are both low indicating little catalytic behavior. The CV for Ta-butanol NPs on GC is presented in Fig. 1(b). During purging with O₂, a small reduction current (∼15 μA) was recorded at around 0 V potential. The reduction potential and current observed for the Ta-butanol NPs, therefore, are slightly larger than that of the pure Ta NPs but still low overall indicating minimal catalytic behavior. In the case of the Ta-heptane and Ta-xylene NPs, small features consistent with the presence of oxygenated functional groups on the surface of GC were observed again during purging with Ar. When the purge gas was switched to O₂, however, strong reduction currents (∼250-275 μA) were measured at a half-wave potential of ∼0.4 V as shown in Figs. 1(c) and 1(d). The CV curves were also stable toward repeated cycling over a potential range of −0.1 to 1.5 V indicating that the organic-inorganic NPs produced by reactive sputtering of Ta with heptane and xylene are both active and robust catalysts toward the ORR at room temperature. While the reduction potentials determined for the Ta-based NPs are modest compared to values reported previously for state-of-the-art Pt NPs (∼0.8 V),⁸¹ it is nevertheless encouraging to observe that sputtering of an unreactive base metal in the presence of different hydrocarbons produces NPs with reactivity reminiscent of more costly and rare precious metals. For comparison, Ta-oxide based catalysts prepared recently on carbon supports exhibited reduction potentials around 0.65 V.^62,63 It is expected that additional optimization of the reactive sputtering process will increase the O₂ reduction potential of the Ta-heptane and -xylene NPs.

To gain insight into how the composition and structure of the NPs influence their catalytic properties, a suite of characterization techniques was employed. A representative high magnification (5μm²) 3D tapping mode AFM color image is presented in Fig. 1 in the supplementary material that reveals the formation of a partial layer of Ta-butanol NPs on HOPG. In the AFM image, the Ta-butanol NPs are observed to bind almost exclusively along the step edges of HOPG creating parallel extended linear chains of NPs and leaving the terrace sites largely vacant. Anionic NPs that arrive from the gas phase with random trajectories, therefore, are mobile on the HOPG surface at room temperature and diffuse from the weakly binding terraces to the step edges where they become more strongly immobilized. This linear arrangement of the Ta-butanol NPs is consistent with our previously reported results for bimetallic Pt/Ti and V/Ti NPs soft landed on HOPG.⁶⁴ In addition, similar findings were described previously by Bowen and co-workers⁸² for mass-selected anionic Mo_n and Palmer and co-workers for Ag_n clusters soft landed onto HOPG.⁸³ This linear arrangement is caused by the Smoluchowski effect which results in a localized increase in charge density along the top and a depletion along the bottom of steps on conductive surfaces. The charge redistribution causes preferential alignment of adsorbed molecules and clusters along step edges.

The surface topography (height) distribution obtained with AFM over a 5 μm² region of the HOPG surface is mono-modal and centered at 10 nm in height with a FWHM of ∼4 nm, as shown in Fig. 2(a). The topography distribution, however, is not symmetrical with there being a slightly higher abundance of larger NPs in the 10-20 nm size range compared to smaller species 1-10 nm in diameter. As discussed previously, the variation in NP heights measured with AFM may result from different isomers of the NPs as well as distinct binding geometries to specific sites (defects) on the surface.⁶⁴ The 2-D (5 μm²) AFM color image presented in Fig. 1 in the supplementary material confirms the low coverage of Ta-butanol NPs on HOPG and reveals the presence of smaller and larger features that are not as evident in the 3-D image. An AFM line profile obtained through selected features on the surface is also presented in Fig. 1 in the supplementary material revealing the presence of 5 adjacent NPs with similar heights of ∼10 nm along with a smaller feature that is ∼4 nm as well as another peak that is ∼9 nm. Previous characterization of a large number of soft landed pure metal NPs on HOPG indicates that the height (z) observed with AFM is generally consistent with the size selected using the quadrupole mass filter.^24,64 This suggests effective mass filtering and little transmission of larger neutral species to the surface. The in-plane (x,y) diameter obtained with AFM, however, is typically not a reliable measure of size due to the tip-convolution effect which artificially increases the NP diameter. The narrow distribution of NP heights observed with AFM suggests that substantially larger or smaller NPs that may be produced at the source do not comprise a sizable amount of the material that reaches the substrates. A recent work by Palmer and co-workers demonstrated that changing the gas aggregation length and magnetron sputtering power of a similar NP source enables the controlled production of ionic Au clusters with different structural motifs.¹⁹ Therefore, a distribution of isomers originating at the source may be partially responsible for some of the outliers in height observed in Fig. 2(a). In another publication, the same authors showed that Au NPs subjected to electron beam bombardment in a microscope undergo structural evolution over time.⁸⁴ This indicates that several energetically close lying isomers may exist for a given size NP. Another factor complicating the interpretation of the AFM images is the possible mobility of the NPs on the HOPG surface which may result in their migration to step edges and potential agglomeration into larger features. Vertical stacking may also explain the slightly higher abundance of larger features observed in Fig. 2(a) for the Ta-butanol NPs. Despite these possibilities the majority of NPs observed with AFM in Fig. 2(a) are centered around 10 nm in height which demonstrates the ability of the magnetron sputtering gas aggregation approach combined with ion soft landing to prepare well-defined NPs on surfaces. It is worth noting that the nascent size distributions of NPs produced in the current study by reactive sputtering are, in general, narrower compared to the NP distributions reported in previous studies which were formed by conventional inert gas condensation.^24,64 This observation is based on the change in the mass spectra of Ta NPs as each reactant gas is added to the sputtering zone. Reactive sputtering, therefore, may offer a way to circumvent mass-selection by creating intense and inherently size-focused beams of NPs for soft landing.

A representative high magnification (5 μm²) AFM 3-D color image of Ta-heptane NPs on HOPG is presented in Fig. 2 in the supplementary material. Similar to the spatial arrangement observed for the Ta-butanol species, the Ta-heptane NPs preferentially orient themselves along the step edges of HOPG. However, due to the higher coverage of the Ta-heptane NPs (6 × 10³ ions μm⁻²) compared to the Ta-butanol NPs (3 × 10³ ions μm⁻²), fewer open terrace sites are observed with AFM. The higher coverage of the Ta-heptane NPs in the AFM image in Fig. 2 in the supplementary material may also result from selection of a particularly dense region of the surface for analysis. The surface topography distribution is plotted in Fig. 2(b) and reveals a monomodal height distribution that is consistent with slightly smaller ∼7 nm diameter Ta-heptane NPs (m/z ∼ 100900) compared to the larger 10 nm Ta-butanol species (m/z ∼ 178800). The topography distribution of the Ta-heptane NPs is also narrower and more symmetrical (FWHM = ∼ 3 nm) than that for the Ta-butanol species which has a slightly increased population of larger NPs. The 2-D AFM image presented in Fig. 2 in the supplementary material confirms the higher coverage of Ta-heptane NPs and their alignment along the step edges of HOPG. The white line on the lower right side of the 2-D AFM image designates the path of a line profile obtained through selected features which is also presented in Fig. 2 in the supplementary material. In agreement with the overall topography distribution, the line profile reveals around seven NPs with heights ranging from 5 to 6.5 nm. Similar to the situation for the Ta-butanol NPs, it is proposed that the variance in height observed with AFM for the Ta-heptane NPs may result from different structural isomers and binding sites on the HOPG surface.

A representative high magnification 3-D (2 μm²) AFM image of Ta-xylene NPs soft landed on HOPG is presented in Fig. 3 in the supplementary material. The NP coverage in this image is similar to that of the Ta-heptane NPs and higher than that of the Ta-butanol species. In agreement with the results for the other organic-inorganic hybrid NPs examined in this study, the AFM image reveals the arrangement of the soft landed Ta-xylene NPs along the step edges of the HOPG surface. However, here a larger fraction of open terrace sites is observed at higher magnification (2 μm²) compared to the AFM image obtained for the Ta-heptane NPs (5 μm²). The surface topography distribution of the Ta-xylene NPs is presented in Fig. 2(c) and reveals a bimodal distribution of heights centered at 6.5 and 12 nm. While the majority of the NPs are centered at a height of 6.5 nm which corresponds with the selected mass (m/z ∼ 99000), there is a small fraction of species with larger size. Based on the fact that a feature 12 nm in height is almost consistent with the result of vertical or somewhat diagonal stacking of two 6.5 nm NPs, it is proposed that formation of a second NP layer begins at relatively low coverage at the step edges of HOPG before the vacant terrace sites are occupied. A 2-D AFM image is presented in Fig. 3 in the supplementary material which confirms the linear alignment and tightly focused size of the Ta-xylene NPs on HOPG. A line profile through selected features in the 2D image (white line in upper left corner) is also presented in the supplementary material which reveals the presence of 4 adjacent NPs with almost identical height.

Representative TEM and HAADF-STEM images of a surface prepared by soft landing 3 × 10³ ions μm⁻² of mass-selected anionic ∼10 nm Ta-butanol NPs onto a continuous carbon microscopy grid are presented in Fig. 4 in the supplementary material. A low magnification phase-contrast TEM image is provided that reveals the presence of a partial layer of Ta-butanol NPs on the surface of the flat continuous carbon grid. Unlike the results observed on the stepped HOPG surface with AFM, the NPs bind randomly to the surface of the carbon grid with some aggregation but little coalescence into larger species. Additionally, the NPs are found to exhibit darker centers surrounded by regions of lower phase contrast. In agreement with the height distribution obtained on HOPG using AFM (∼10 nm) and the mass of the ions selected for deposition (m/z ∼ 178800), the NP diameters obtained with TEM are centered at ∼9 nm with a FWHM of ∼2 nm, as shown in Fig. 3(a). Therefore, AFM and TEM images on two different substrates (HOPG and carbon grid) both exclude soft landing of a sizable number of substantially smaller or larger NPs that may be produced by the sputtering source. The images also exclude coalescence of the NPs into larger species. A total of 50 NPs were randomly selected and manually measured to produce the size distributions plotted in Fig. 3. The diameters were measured from the outer edges of the regions of lighter contrast on each NP (including core and shell). A representative high magnification TEM image of a single selected Ta-butanol NP is presented in Fig. 4(a) which reveals the presence of defined lattice fringes indicating a crystalline core surrounded by an amorphous shell. A high resolution HAADF-STEM image is also shown in Fig. 4 in the supplementary material that provides additional support for a higher z-contrast (brighter) region at the core of the NP compared to the shell. Atomically resolved STEM imaging of the Ta-butanol NPs was challenging due to the rapid build-up of carbon material under the tightly focused high energy electron beam. Cleaning procedures such as plasma treatment and heating in vacuum were attempted but were not able to sufficiently remove the carbon contamination without also causing changes to the sample. Nevertheless, the TEM and STEM images are consistent with Ta-butanol NPs that have a crystalline core surrounded by an amorphous carbide or carbon shell.

Representative TEM and HAADF-STEM images were also acquired for Ta-heptane NPs. An intermediate magnification TEM image is presented in Fig. 5 in the supplementary material that reveals sub-monolayer coverage of Ta-heptane NPs with diameters centered at ∼7 nm and a FWHM of ∼2 nm, as shown in Fig. 3(b). The TEM data, therefore, confirms the smaller size of the Ta-heptane NPs (m/z ∼ 100900) compared to the larger Ta-butanol species (m/z ∼ 178000) selected for deposition and observed on HOPG with AFM. The high resolution TEM image in Fig. 4(b) reveals four distinct higher contrast regions containing lattice fringes indicating crystalline material. These higher contrast crystalline regions are surrounded by lower contrast amorphous material that forms a continuous network connecting the Ta-heptane NPs together. The high-resolution STEM image presented in Fig. 5 in the supplementary material shows a NP core with higher z-contrast than the surrounding shell region. Again, due to the carbon content of the NPs, it was challenging to obtain atomically resolved STEM images. While the coverage and size of the Ta-heptane NPs differ from the Ta-butanol species, both reactively sputtered NPs have similar spherical morphologies with crystalline cores surrounded by an amorphous matrix.

Representative TEM and HAADF-STEM images of the Ta-xylene NPs are shown in Fig. 6 in the supplementary material. The low magnification TEM image reveals a partial coverage of NPs with high contrast arranged randomly on the carbon surface without any substantial agglomeration. The size distribution is presented in Fig. 3(c) and reveals that the NPs are centered at ∼6 nm in diameter with a FWHM of ∼3 nm. This is in reasonable agreement with the mass selected for deposition (m/z ∼ 99000) and the 6.5 nm average height measured on HOPG using AFM. A higher magnification TEM image presented in Fig. 4(c) indicates that the Ta-xylene NPs have spherical morphology with small higher contrast spots within each NP. Unlike the TEM images obtained for the Ta-butanol and Ta-heptane NPs, however, there is no evidence of a crystalline core surrounded by an amorphous layer for the Ta-xylene NPs. A high resolution STEM image is presented in Fig. 6 in the supplementary material which reveals a higher contrast region at the core of the NPs.

In order to characterize the elemental composition and chemical state of the elements comprising the NPs, XPS spectra were obtained on HOPG substrates and are presented in Fig. 5. The spectra for Ta-butanol NPs appear as black lines centered at 27.6 and 25.8 eV for the Ta 4f_5/2 and 4f_7/2 features, respectively. In addition, the O 1s and C 1s regions of the XPS spectra are shown in Figs. 5(b) and 5(c) revealing electron binding energies of 530.4 and 284.1 eV, respectively. To interpret the XPS spectra, it is instructive to compare the results in Fig. 5 with those reported previously in the literature for Ta oxides and carbides, both of which may form during the sputtering and soft landing processes as well as when the samples are exposed to the ambient environment prior to characterization with XPS. For example, in an early XPS study, Choi and co-workers carburized Ta₂O₅ using CH₄ in a temperature-programmed experiment.⁸⁵ The Ta 4f_5/2 XPS binding energies were reported to be 25.2, 26.9, and 28.8 eV for TaC_x, TaO₂, and Ta₂O_5, respectively, while the Ta 4f_7/2 binding energies were assigned as 23.1, 25.2, and 26.9 eV. In a subsequent study, the same authors examined porous Ta carbide crystallites and found the Ta 4f_5/2 binding energies to be 25.2, 27.0, and 28.8 eV for TaC, TaO₂, and Ta₂O₅, respectively.⁸⁶ The Ta 4f_7/2 values for these crystallites were 23.1, 25.1, and 27.1 eV. The C 1s binding energy was reported to be 284.4 eV which is indicative of graphitic rather than carbidic carbon that appears instead at 282.6 eV. Another XPS study by Kerrec and co-workers reported Ta 4f_7/2 binding energies of 23.6 and 26.3 eV for lower and higher oxidation states of Ta, respectively.⁸⁷ This is consistent with the findings of Moo and co-workers who measured the Ta 4f_7/2 energies to be 23.9, 25.3, and 26.7 eV for TaO, TaO₂, and Ta₂O₅, respectively.⁸⁸ Based on a comparison of the current data to these previous XPS results, it is proposed that the spectra presented in Fig. 5 are consistent with partially oxidized (substoichiometric) Ta NPs that may form when the surfaces are removed from vacuum and exposed to laboratory air. Furthermore, the C 1s peak located at 284.1 eV in Fig. 5(c) indicates the presence of carbon in the graphitic form while the O 1s peak at 530.4 eV is consistent with substoichiometric TaO₂.⁸⁵ The percentages of Ta, O, and C were calculated using the XPS data to provide the values presented in Table II. For the Ta-butanol NPs the percentage weights of Ta, C, and O are 10%, 64%, and 21%, respectively.

The XPS spectra obtained for the Ta-heptane NPs on HOPG are presented as red lines in Fig. 5. As shown in Fig. 5(a), the 4f_5/2 and 4f_7/2 binding energies observed at 27.8 and 26.0 eV are similar to those obtained for the Ta-butanol NPs at 27.6 and 25.8 eV (black lines). Therefore, it is proposed that the Ta-heptane NPs are also primarily substoichiometric TaO_x with the oxygen coming from the atmosphere when the NPs are brought out of vacuum. The O 1s XPS spectrum of the Ta-heptane NPs shown in Fig. 5(b) appears at 530.6 eV which is close to the value observed for the Ta-butanol NPs (530.4 eV). The C 1s peak is present at 284.1 eV for both the Ta-butanol and Ta-heptane NPs suggesting that carbon is present in the graphitic form for both species. The percentage weights of Ta, C, and O in the Ta-heptane NPs were estimated to be 25%, 54%, and 19%, respectively (Table II). Compared to the XPS results for the Ta-butanol NPs, therefore, the Ta-heptane species contain more than twice as much Ta and approximately 10% less carbon. It is evident that the flow rate and type of reactant gas influence the amount of metal and carbon incorporated into the organic-inorganic hybrid NPs. The increased time for condensation and NP formation resulting from the longer aggregation region may also explain the higher amount of Ta observed in the Ta-heptane compared to the Ta-butanol NPs.

The composition and chemical state of elements in the Ta-xylene NPs soft landed on HOPG were also characterized with XPS. The resulting spectra are presented in Fig. 5 as blue lines. The Ta 4f_5/2 and 4f_7/2 peaks are present at 27.9 and 26.1 eV, respectively. These values are essentially the same as those measured for the Ta-heptane NPs (27.8 and 26.0 eV) but slightly higher than the binding energies for Ta-butanol NPs (27.6 and 25.8 eV). The Ta-xylene O 1s peak appears with a binding energy of 530.9 eV in Fig. 5(b) which is higher than that of both the Ta-heptane (530.6 eV) and Ta-butanol NPs (530.4 eV). The energy of the Ta-xylene C 1s peak is 284.3 eV which is slightly larger than the values of the other NPs that both appear at 284.1 eV. The percentage weights of Ta, C, and O in the Ta-xylene NPs are also provided in Table II. It is interesting to note that the percentage of Ta in the Ta-xylene NPs (24%) is essentially the same as in the Ta-heptane NPs (25%) but substantially larger than that of the Ta-butanol NPs (10%). This is surprising given the fact that the Ta target was sputtered much more aggressively at 140 mA for the Ta-xylene NPs compared to 80 mA for the Ta-butanol and Ta-heptane NPs. In addition, the percentage of C in the Ta-xylene NPs (44%) is lower than that measured for the Ta-heptane (54%) and Ta-butanol (65%) NPs. While the flow rates of the reactant gases were controlled, the mole fraction of each element available in the plasma for reaction (assuming atomization of a large portion of the molecules) also depends on the stoichiometry of the hydrocarbon. For instance, 2-butanol (C₄H₁₀O) contains approximately half the number of carbons per molecule compared to heptane (C₇H₁₆) and m-xylene (C₈H₁₀). For this reason, approximately twice the flow rate (0.4 sccm) of 2-butanol was employed in an effort to provide a consistent amount of atomized carbon to all of the NPs formed during synthesis. In the case of the Ta-heptane and Ta-xylene NPs, the flow rates were lower (0.2 sccm) which served to balance the larger number of carbons in each molecule compared to 2-butanol. Despite these efforts, the percentage of C in the NPs varies considerably. In addition, the percentage of oxygen was found to be the largest for the Ta-xylene NPs where the reactant gas does not contain oxygen! A possible explanation for this observation is that the Ta-xylene NPs have a higher Ta/C ratio (0.54) than either the Ta-butanol (0.15) or Ta-heptane (0.46) NPs. The larger fraction of Ta may getter more oxygen from the environment when the NPs are removed from vacuum which may explain the higher percentage of oxygen in the exposed Ta-xylene NPs.

Previous studies have characterized the influence of different sputtering parameters and reactant gases on the size, morphology, and composition of NPs prepared by magnetron sputtering and gas aggregation.^{19,22,89–92} Furthermore, due to their often unique compositions and morphologies compared to NPs prepared in solution as well as their large surface-to-volume ratios, the catalytic activity of these NPs has been evaluated toward a number of commercially relevant reactions.^15,26,35,48 The electrochemical ORR is one process that occurs at the cathode in PEMFCs and limits the efficiency of these potential clean energy devices.⁵⁵ In recent years the ORR has been used as a model reaction to explore the size-dependent activity of sub-nanometer Pt clusters on electrode supports.^57,93,94 Larger Pt and Pt-alloy NPs containing other transition metals have also been evaluated as catalysts for promoting the ORR.^24,59 One motivation for the current study was to determine whether reactive sputtering of base transition metals in the presence of different hydrocarbons may produce organic-inorganic hybrid NPs with catalytic properties reminiscent of noble metals. This concept was introduced by Boudart in 1973 when tungsten carbide surfaces were shown experimentally to exhibit similar catalytic properties to Pt.⁶⁰ A year later it was discovered that the electronic density of states near the Fermi level of tungsten carbide closely resembles that of Pt.⁶¹ Therefore, small well-dispersed metal carbide NPs with high surface area may be promising candidates for non-noble metal catalysts.⁹⁵ In addition, it appears from this study that the introduction of carbon around the NPs stabilizes them toward agglomeration and sintering during repeated potential cycling similar to the metal-oxide shells employed previously by Sowwan and co-workers³⁵ and others.^96,97 The difference in reactivity between Ta NPs sputtered in the presence of Ar and 2-butanol compared to heptane and xylene is proposed to result from the substantially larger concentration of Ta (25%) in the latter NPs. In comparison, NPs sputtered with 2-butanol had a lower Ta concentration of 10% while pure Ta NPs rapidly oxidized forming a less reactive surface layer. The Ta-butanol NPs also contained a much larger concentration of carbon (65%) compared to the Ta-heptane (54%) and -xylene (44%) NPs. The concentration of oxygen was found to be similar between each type of reactively sputtered NP (20%-25%). This is surprising considering that 2-butanol contains oxygen while heptane and xylene do not. As mentioned earlier, all of the reactively sputtered NPs likely evolve toward a similar extent of oxidation following exposure to the ambient atmosphere.

In this contribution, the potential of reactive magnetron sputtering and gas aggregation as a preparation technique for catalytically active hybrid organic-inorganic NPs was demonstrated. Reactive DC sputtering of Ta in the presence of 2-butanol, heptane, or m-xylene combined with gas aggregation resulted in the synthesis of novel NPs that were mass-selected and soft landed. The catalytic activity of the NPs was evaluated using a GC surface coated with NPs as the working electrode in an electrochemical cell. Both the Ta-heptane and Ta-xylene but not the Ta-butanol or pure Ta NPs exhibited large reduction currents consistent with the ORR during purging of the solutions with O₂. Furthermore, the CV curves remained stable following repeated potential cycling indicating that the NPs are both active and robust at room temperature. Future experiments will probe the stability of the supported NPs in acidic media at elevated temperatures. AFM images revealed the preferential alignment of all three types of NPs along the step edges of HOPG. Topography distributions calculated from AFM data and line profiles through selected NPs on the HOPG surfaces showed the NPs to be centered at 10, 7, and 6.5 nm in height for the Ta-butanol, Ta-heptane, and Ta-xylene species, respectively. In addition, XPS spectra obtained from the NPs on HOPG revealed similar electron binding energies for the Ta 4f_5/2, Ta 4f_7/2, O 1s, and C 1s peaks of each type of NP. These binding energies are consistent with substoichiometric Ta_xO_y and graphitic C. The XPS data also showed that the percentage weights of Ta, C, and O varied considerably depending on the stoichiometry and flow rate of the reactant gas employed. The differences in reactivity, therefore, are attributed to the Ta/C ratios of the NPs which are 0.15, 0.46, and 0.54 for the Ta-butanol, Ta-heptane, and Ta-xylene NPs, respectively, with a larger Ta/C ratio associated with higher catalytic activity. Images obtained with TEM and HAADF-STEM demonstrated that all three types of NPs bind randomly the carbon surface of the grids with agglomeration but no coalescence. The TEM images also revealed that the Ta-butanol and Ta-heptane NPs have higher contrast crystalline cores surrounded by lower contrast regions comprised of amorphous material. In comparison, TEM images of the Ta-xylene NPs showed them to be of higher contrast throughout compared to the Ta-butanol and Ta-heptane species. The NP size distributions obtained from analysis of the TEM images were in agreement with the mass to charge ratios selected for deposition and the heights determined on HOPG using AFM. Reactive sputtering of a base metal combined with gas aggregation and ion soft landing, therefore, offers a versatile physical approach for the preparation of non-noble metal NPs for catalytic applications.

See the supplementary material for 2 and 3-dimensional atomic force microscopy images and line profiles of soft landed nanoparticles. Transmission electron microscopy and scanning transmission electron microscopy images of soft landed nanoparticles.

The research described in this paper is part of the Chemical Imaging Initiative, at Pacific Northwest National Laboratory (PNNL). It was conducted under the Laboratory Directed Research and Development Program at PNNL. G.E.J. and J.L. acknowledge support from the US Department of Energy, Office of Science, Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences, and Biosciences of the U.S. Department of Energy (DOE). This work was performed using EMSL, a national scientific user facility sponsored by the Department of Energy’s Office of Biological and Environmental Research and located at PNNL. PNNL is operated by Battelle for the U.S. DOE.