Branched-gallium phosphide nanowires seeded by palladium nanoparticles

Palladium nanoparticles were produced by a chemical reagent-free and versatile method called spark ablation with control over particle size and density. These nanoparticles were used as catalytic seed particles for gallium phosphide nanowire growth by metalorganic vapour-phase epitaxy. Controlled growth of GaP nanowires using significantly small Pd nanoparticles between 10 and 40 nm diameter was achieved by varying several growth parameters. Low V/III ratios below 2.0 promote higher Ga incorporation into the Pd nanoparticles. Moderate growth temperatures under 600 °C avoid kinking and undesirable GaP surface growth. In addition, a second batch of palladium nanoparticles of concentration up to 1000 particles μm−2 was deposited onto the GaP nanowires. Subsequently, three-dimensional nanostructures evolved, with branches growing along the surface of the GaP nanowires. The GaP nanowires revealed a zinc blende structure with multiple twinning and a PdGa phase at the tip of the nanowires and branches.

A semiconductor material that possesses attractive properties in the form of nanowires (NWs) is gallium phosphide (GaP) [25][26][27]. One-dimensional GaP nanostructures including NWs have been investigated in similar applications to Pd NPs, namely in catalysis, hydrogen storage, and biomedicine [28][29][30][31][32], opening the prospect of synergy effects when combined with Pd. For instance, Pd-Ga phases exhibit long-term stability for alkane dehydrogenation [33], and Pd-Ga intermetallic compounds display notable selectivity and stability in the selective hydrogenation of acetylene [34]. Furthermore, GaP NWs have the potential to provide good support for Pd NPs, the high aspect ratio of NWs leads to high surface area of the GaP NW-Pd nanostructures and, by being biocompatible, GaP has an advantage over other semiconductor materials. Several studies have investigated the applications of Pd NPs and GaP NWs individually. However, to study possible synergy effects, the first step is to create stable GaP NW-Pd nanostructures. Therefore, the fabrication of one-to three-dimensional nanoarchitectures with topological properties of both Pd NPs and GaP NWs is worth exploring.
Looking at possible routes to fabricating these specific nanostructures, we noted that Pd NPs have been tested as an alternative catalyst to Au for the growth of GaAs NWs [35] and InAs NWs [36,37]. However, no studies on Pd-seeded GaP NWs have, until now, been carried out. Pd seeded III-V NW growth is a challenge due to the high growth temperatures and large group-III concentrations required to ensure a vapour-liquid-solid (VLS) growth. This mode of growth is known to promote high axial growth rates [38,39], stable growth direction [40], and crystal phase control [41]. In contrast, when temperatures lower than the eutectic melting points are used, the solid seed NPs lead to a vapour-solidsolid (VSS) growth mechanism, often resulting in lower growth rates than VLS [42] and unstable growth direction [43]. Interestingly, several studies on NW growth by Pd NPs have demonstrated the presence of both VLS and VSS growth modes [44,45] which are highly dependent on the particle size. Seed particles smaller than a critical diameter are liquid (i.e. VLS) and when larger than that diameter, are solid (i.e. VSS) during NW growth. Moreover, the growth mechanism can also affect the morphology of the sidewalls of the NWs, crystal structure, and composition of the NW tips with the formation of distinctive Pd-Ga phases. Hence, it is essential to study the interplay of the growth parameters for GaP NW seeded by Pd NPs to determine the optimal NW growth conditions.
Further enhancing the 1D nanostructures involves exploring the growth of branches serving as a 3D support, effectively increasing the surface area and facilitating greater exposure of the Pd-Ga NPs. This high surface area leads to the creation of numerous active sites and increased surface activity making branched NWs highly advantageous for a wide range of applications, particularly in catalysis.
In this work, Pd NPs were produced by spark ablation, a versatile and wet chemical synthesis-free method to produce high-purity Pd NPs with a tuneable control over particle size and number concentration [46][47][48]. Pd NPs of different diameters were selected as catalysts for epitaxial growth of GaP NWs by metalorganic vapour-phase epitaxy (MOVPE). The GaP NW growth from Pd NPs ranging from 10 to 40 nm offers a comprehensive investigation into the diameterdependent yield of vertically oriented GaP NWs, which is an indication of their mechanical stability, aiming to maximize surface area while achieving the smallest mechanically stable nanowire size possible. The crystallographic nature and growth mechanisms of these III-V NWs were studied for different growth parameters such as V/III ratio, temperature, and diameter in order to determine the best conditions for the growth of straight NWs. Additionally, 3D nanostructures were fabricated by depositing a second batch of Pd NPs at high particle number concentration onto the GaP NWs to induce branch growth and increase the surface area of the nanostructures. The branch growth was then performed at similar growth conditions to those used for the core NWs, and the growth direction, crystal structure, and composition of these branched-NWs were analysed. In this paper, we use the expression 'core NW' for a straight NW and for the original straight NW within a branched structure, and 'branched NW' for secondary growth from the core.

Catalytic Pd nanoparticles produced by spark ablation
Pd-rod electrodes of ca. 3.00 mm diameter were mounted on an anode and cathode of a spark discharge generator (SDG) as described by Messing et al [49]. The Pd NPs were sizeselected for 10 nm, 20 nm, and 40 nm diameter by using a differential mobility analyser (DMA) and sintered at 750°C into compact particles. Afterwards, monodispersed Pd NPs of a selected number concentration (1 particle μm −2 ) were deposited in a random distribution on a GaP (111) B substrate loaded in an electrostatic precipitator (ESP) [50].

Nanowire growth
GaP NW growth was seeded (or catalysed) by Pd NPs and performed via MOVPE at low pressure (100 mbar). Phosphine (PH 3 ) and trimethyl gallium (TMGa) {Ga(CH 3 ) 3 } were used as precursors in a hydrogen (H 2 ) carrier gas at a total flow of 6000 cm 3 min −1 into the reactor. The first step is the annealing of the GaP substrate to remove the native oxides and was performed at 650°C for 10 min. The second step is the NW growth performed at several temperatures (560°C-600°C). The V/III precursor ratios (1.30-3.03) were set by fixing the molar fraction of TMGa at 8.56 ×10 −4 and varying the PH 3 in the range of 1.11 × 10 -3 -2.59 × 10 -3 .

Branched nanowire growth
GaP NWs seeded by Pd NPs of 10 nm, 20 nm, and 40 nm diameter and grown at 560°C with a V/III ratio of 1.30 were selected and loaded into the ESP of the SDG. A second batch of Pd NPs of 10 nm diameter produced in the SDG was then deposited onto the GaP NWs at several particle number concentrations (100-1000 particles μm −2 ). Afterwards, branches were grown via MOVPE at the same conditions as written above (i.e. a growth temperature of 560°C and a V/ III ratio of 1.30) and for a growth time of 10 s.

Characterization
Scanning electron microscopy (SEM) was performed on a ZEISS LEO Gemini 1560 and a ZEISS Gemini 500 at an accelerating voltage of 15 kV to study the morphology of the Pd NPs and GaP NWs.
For the transmission electron microscopy (TEM) studies, the GaP NWs were dispersed on holey carbon film-coated Cu TEM grids. The grids were loaded into a double-tilt holder to allow the tilting of the NWs to the [110] zone axis. The data presented here were acquired using a TEM JEOL 3000F equipped with a field emission gun and operated at 300 kV. A high-angle annular dark-field (HAADF) detector coupled with an energy dispersive x-ray (EDX) spectrometer (Oxford Instruments) was used in scanning transmission electron microscopy (STEM) mode for elemental scanning profiles. High-resolution transmission electron microscopy (HRTEM) imaging and selected area electron diffraction (SAED) were used to determine the crystal structure of the NWs. The STEM/EDX data were used to determine the composition of the NWs via INCA software.

Effect of NW growth parameters
The Pd NPs of 10 nm, 20 nm, and 40 nm diameter produced by spark ablation are shown in figures 1(a)-(c), respectively. A high yield of straight GaP NWs grown in the 〈111〉B directions was achieved at a growth temperature of 560°C and V/III ratio of 1.30, as shown in figures 1(d)-(f). Indeed, a high Ga concentration (i.e. low V/III ratio) at moderate temperatures is required to ensure a kinetic-limited NW growth and avoid kinking due to excessive surface growth and possibly non-VLS NW growth mechanism. These criteria are based on the Ga-Pd phase diagram, which indicates high melting eutectic temperatures compared to Au-Ga (e.g. 55 at% Ga requires a eutectic temperature of ca. 440°C for Au-Ga and ca. 1000°C for Pd-Ga). Instead, lower growth temperatures demand high Ga concentration to ensure a VLS NW growth. Furthermore, the effect of varying growth temperature (560°C-600°C) and V/III ratio (1.30-3.03) was studied for a given seed particle size (figures S1-S3 of supplementary information), and the results showed a lower yield of straight GaP NWs for higher values of these parameters. The presence of kinked and curly NWs suggests a non-continuous growth, which can be caused by several factors. A partial solidification of the Pd NPs (i.e. VSS growth) leads to high supersaturation in the Pd-Ga droplet changing the surface energies to favour other growth directions [51]. Another factor is attributed to the competition between the nucleation rate at the Pd-Ga droplet and at the substrate surface; when nucleation at the substrate surface dominates, parasitic GaP growth in the form of 2D layers, agglomerated particles, and/or 3D islands occurs. This substrate surface growth may hinder NW growth by changing the regular 〈111〉 growth directions to unfavourable new growth directions [52][53][54].
The temperature dependence of the axial growth rate (GR) for several V/III ratios (1.30-3.03) is shown in figure 2(a), and for a given seed particle diameter in figure S4 of supplementary information. The Arrhenius behaviour of the data (R 2 − 〈0.94-0.99〉) shows a comparable slope, −E a /k B , for similar V/III ratios. From this slope, we calculate an activation energy (E a ) in the range of 〈28.2-65.9〉 kcal mol −1 (E a 〈1.22-2.86〉 eV) which indicates a kinetic-limited NW growth [55,56]. Furthermore, an increase of GR with temperature and V/III ratio was observed for all of the NWs with values ranging from 0.88 to 4.87 μm min −1 . The increased GR is due to a higher Ga adatom mobility and chemical potential difference of Pd-Ga in the droplet, causing a faster diffusion rate (i.e. longer diffusion length) and, consequently, increasing the NW axial growth [57]. Taking the NW diameter into consideration as shown in figure 2(b), two opposite trends can be identified for all of the temperatures: at low V/III ratio (1.30 and 1.87) the GR monotonically increases while the particle diameter increases and at high V/ III ratio (2.45 and 3.03) the GR monotonically decreases while the particle diameter increases. Regarding the first trend at low V/III ratio, the Gibbs-Thomson (G-T) effect [58][59][60][61] provides an explanation, i.e. a decrease in particle diameter leads to a lower supersaturation (the driving force of the NW growth) and a decrease in GR. The lower supersaturation occurs because vapour pressure of Ga in the liquid Pd-Ga droplet increases in smaller NW seed particle diameters, effectively hindering Ga to incorporate into the smaller particles. On the other hand, at a high V/III ratio, the opposite effect was previously reported [62,63], i.e. thinner NWs grow faster than thicker NWs. It is commonly assumed that group III is the surface-diffusing, rate-limiting species [63,64]. But in this case, the TMGa flow is kept constant during the NW growth, and TMGa thermally decomposes in the H 2 atmosphere at temperatures between ca. 375°C-490°C [65,66], which are significantly lower than the NW growth temperatures. Thus, all CH 3 radicals are removed from TMGa, and a constant number of Ga atoms are available during growth. Therefore, the adatom diffusion on the substrate surface and along the sidewalls of the NW is accelerated by the increase of the group V reactant (i.e. high V/III ratio). Since the droplet acts as a sink, its low chemical potential directs the adatom diffusion flux towards the interface of the seed particle and the NW [67]. In conclusion, surface diffusion influenced by the G-T effect is rate limiting for our case of NW growth. Consequently, at high supersaturation thinner NWs will experience a faster axial growth promoted by the surface diffusion, whereas at low supersaturation thinner NWs will grow slower than thicker ones due to the G-T effect [63,68].
A model that incorporates both the G-T effect and surface diffusion in the NW growth [58,63], as explained above, can be represented by the following expression Where dL/dt is the GR of the NWs, b is a kinetic proportionality constant, k B is the Boltzmann constant, T is the growth temperature, Δμ 0 is the difference in chemical potential (supersaturation) between the species in the vapour phase and in the solid NW with infinite diameter, Ω is the atomic volume of the seed particle, α is the surface energy of the seed particle (Pd-Ga alloy), λ is the diffusion length along the side of the NW, and d is the NW diameter. For Pd-Ga alloy the approximate values of Ω and α are 1.8 ×10 -29 m 3 and 1.4 J m −2 , respectively. This model fitted to the experimental values is shown in figure 2(b) for NW growth at 560°C. The trends in GR as a function of NW diameter at both low and high V/III ratios are consistent with the results of Soci et al [69]. From the fits, we find that both Δμ 0 and λ increase with increasing V/III ratio (as shown in figure S5 of supplementary information). The increase of λ with the V/III ratio is discussed in [70]. Higher values of λ indicate a greater contribution of mass transport through surface diffusion during NW growth at high V/III ratios. The straight GaP NWs obtained at 560°C, and a V/III ratio of 1.30 were subjected to further studies. The crystal structure of these NWs for different seed particle sizes is shown in figures 3(a)-(c). HRTEM images show multiple {111} twin planes in the NWs characteristic of twining [71,72]. Multiple twinned zinc blende (ZB) is observed in the GaP NWs for all of the diameters, and in some regions where the twinning is dense the structure is close related to wurtzite (WZ) as previously reported for GaP NWs seeded by Au [73]. The formation of WZ can be ascribed to the lower surface energy of the parallel side facets of WZ compared to that of ZB [73] and the interface energies at the VLS three-phase line (i.e. edge of the Pd NP-NW interface, where nucleation occurs) [74]. Thus, the WZ crystallization can be particularly favourable for thin NWs, whereas ZB is more stable for larger diameters [73] and, the number of stacking faults decreases with increasing diameter [75]. The SAED patterns were acquired along the [110] zone axis from a region that included several twinned segments. The spacing of the (111) lattice fringes, whose normal is parallel to the growth direction, was measured to be ca. 0.31 nm. This value is in good agreement with d 111 for ZB GaP (lattice parameter a = 0.545 nm).
Compositional information on the GaP NWs was obtained from a STEM/EDX scan profile along the core NW and seed particle, choosing the energy lines P Kα, Ga Lα, and Pd Lα, as shown in figure 4. Measurements show comparable Ga and P content along the segment of the core NWs, which confirms the Ga:P atomic ratio of 1:1. In the seed particle, the P content drops to negligible values demonstrating its limited solubility in solid Pd [76]. The composition of the seed particles of ca. 50 at% Pd and 50 at% Ga is close to the PdGa phase. The small deviation from 1:1 stoichiometry for the GaP NW seeded by 10 nm Pd NPs may be attributed to  several factors during the scanning process including electron beam damage, sample drifting and electron channelling at a zone axis [77]. For example, some of the electron probe incident on the seed particle scatters into the core NW and generates a P signal resulting in a combined reading for Ga and Pd of less than 100%. Nevertheless, point analyses from several of the 10 nm seed Pd NPs showed no P content and a similar atomic composition for Pd and Ga.

3D nanostructures
Complex nanostructures such as branched-NWs were examined. A second batch of 10 nm-Pd NPs with a number concentration of 700 particles um −2 was deposited on GaP NWs seeded by 10, 20, and 40 nm Pd NPs, (figures 5(a)-(c) followed by branch growth (figures 5(d)-(f)). The branch length is controllable, and during 10 s of growth at the same conditions as for the core NWs, a range of 77-86 nm in length was obtained. A rough estimation of the GR of the branches is in the range of 0.46-0.52 μm min −1 (assuming a linear growth over time), which is significantly lower than the GR of the core NWs (0.88 μm min −1 under the same growth conditions). The lower GR suggests a strong competition for precursor material between the branches, the core's sidewalls, and the core's tip, which also experienced a slight axial growth.
Taking the core NW growth in the [¯¯111] direction by convention, the branches grow preferentially along the other three 〈111〉B directions, i.e.
[111], [1 11], and [111] as previously reported for branched GaP NWs [78]. Because of the 60°rotation of the (111) plane of the core NW caused by the presence of the stacking faults (twining), the branches grow predominantly on the six facets of the groups of planes (112), (12 1 ), (2 11), and (112), (1 21), (211) denoted as {112} and {112} family planes, respectively [78,79]. The top views of the branched NWs in figures 5(g)-(i), show these side facets. Nonetheless, some branches seem to grow in other directions to the core NW, possibly due to the defects on the sidewalls or aggregation of Pd NPs which affects the growth direction. In addition, the effect of varying the particle number concentration (100 particles um −2 -1000 particles um −2 ) and the subsequent branch growth were studied (figures S6-S8 in supplementary information). It is worth noting that the number of Pd NPs deposited on the core NW is only a small fraction of the deposited particle concentration. The higher the deposition density of Pd NPs, the more aggregation of particles is expected on the tip and along the sidewalls. This aggregation leads to coalescence of NPs, possibly during annealing and/or branch growth, resulting in broadening of the tip of the NW as well as thicker branches. Also, shorter branches can be observed for higher particle densities as the competition for precursor material cause less Ga incorporation per branch. Furthermore, for higher deposition densities, more Pd NPs were observed on the substrate surface, and these Pd NPs do not lead to NW growth when compared to the branches. Twining on the sidewall facets of the core NWs favours a lower surface energy (i.e. {111} planes) than the GaP substrate [80], thus it is reasonable to expect the onset branch growth to occur before NW growth on the GaP substrate.
TEM and HRTEM micrographs in figure 6 clearly show the stacking faults extending smoothly along the branches and perpendicular to the 〈111〉B growth directions of the core NWs [79]. Figure 6(a), shows some branches grown in the atypical 〈110〉 directions perpendicular to the core. Likewise, the branch growth preserves the NW core's stacking faults independently of the branch's growth direction, as shown in figures 6(b), (d). Some branches grown in the 〈111〉 directions are shown in figures 6(c) and (d) with the characteristic continuity of the stacking faults. Finally, the composition of the branches obtained by STEM/EDX reveals an approximate Pd:Ga atomic ratio of 1:1 on the seed NP, suggesting the presence of a PdGa phase.

Conclusions
High yield of straight GaP NWs seeded by Pd NPs of 10, 20, and 40 nm diameter were grown at low V/III ratios of 1.30 and 560°C. Growth rates as high as 0.88 μm min −1 was attained for Pd-seeded GaP NWs. The crystal structure of the GaP NWs is predominantly ZB with a reduction of the multiple twinning for larger diameters. Furthermore, branched-GaP NWs were obtained within 10 s under the same growth conditions as for the core NWs achieving a GR as high as 0.52 μm min −1 . Therefore, we have demonstrated the successful growth of Pd-GaP NWs and 3D nanostructures which will open up the possibility to study synergy effects for Pd and GaP in various applications.

Acknowledgments
All nanofabrication was carried in Lund Nano Lab. We acknowledge financial support from the Swedish Foundation for Strategic Research (Grant No. FFL18-0282), MyFab, and NanoLund. The project has, furthermore, received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 945378.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).