Research Article ISSN 1998-0124 CN 11-5974/O4 2019, 12(12): 3069–3074 https://doi.org/10.1007/s12274-019-2553-y Angular dependence of nanoparticle generation in the matrix assembly cluster source Maria Chiara Spadaro1,†, Junlei Zhao2, William D. Terry3, Jian Liu3,4, Feng Yin5, Flyura Djurabekova2, and Richard E. Palmer1 () 1 College of Engineering, Swansea University, Bay Campus, Fabian Way, Swansea SA1 8EN, UK 2 Department of Physics and Helsinki Institute of Physics, University of Helsinki, P. O. Box43, FIN-00014, Helsinki, Finland 3 Nanoscale Physics Research Laboratory, School of Physics and Astronomy, University of Birmingham, Birmingham B152TT, UK 4 National Center for Electron Microscopy in Beijing, Key Laboratory of Advanced Materials (MOE), The State Key Laboratory of New Ceramics and Fine Processing, School of Materials Science and Engineering, Tsinghua University, Beijing 100084, China 5 School of Physics and Information Technology, Shaanxi Normal University, Xi’an 710062, China † Present address: Catalan Institute of Nanoscience and Nanotechnology (ICN2), CSIC and BIST, Campus UAB, Bellaterra, 08193 Barcelona, Catalonia, Spain © The Author(s) 2019 Received: 8 August 2019 / Revised: 22 October 2019 / Accepted: 27 October 2019 ABSTRACT The matrix assembly cluster source (MACS) represents a bridge between conventional instruments for cluster beam deposition (CBD) and the level of industrial production. The method is based on Ar+ ion sputtering of a pre-condensed Ar-M matrix (where M, is typically a metal such as Ag). Each Ar+ ion produces a collision cascade and thus the formation of metal clusters is in the matrix, which are then sputtered out. Here we present an experimental and computational investigation of the cluster emission process, specifically its dependence on the Ar+ ion angle of incidence and the cluster emission angle. We find the incidence angle strongly influences the emerging cluster flux, which is assigned to the spatial location of the deposited primary ion energy relative to the cluster into the matrix. We also found an approximately constant angle between the incident ion beam and the peak in the emitted cluster distribution, with value between 99° and 109°. KEYWORDS cluster beam deposition scale-up, nanoparticles, silver, ligand-free, green synthesis 1 Introduction thus only ~ μg/h [16], or below. Thus is generally sufficient for model studies, but cluster beam research also has the potential to contribute Cluster beam sources were developed in the early 1980s for to technological challenges, e.g., in biomedicine or fine chemicals fundamental studies of free cluster physics, leading, for example, to catalysis. To reach this goal, it is necessary to scale-up the deposition the discovery of magic numbers in metal and rare gas clusters [1, 2]. rate up to ~ mg/h or even ~ g/h rate. Within 10 years, these instruments were adapted to enable cluster A number of efforts to scale-up cluster beam deposition have been beam deposition (CBD) onto surfaces [3]. Such sources are based reported [17–20]. In this context, a completely new type of cluster on physical processes for material vaporisation and then cluster beam source, the matrix assembly cluster source (MACS), was invented (nanoparticles) aggregation, that highly differ from the chemical- by our group [21–25]. In this approach, evaporated metal atoms are based approaches. CBD sources are characterized by a number of co-condensed with rare gas atoms on a substrate cooled down to potential advantages, such as (i) purity, the nanoparticles are produced 10–20 K in vacuum. An Ar+ ion gun is then exploited to bombard with a solvent-free synthesis route and, therefore, are characterized by the matrix. Metal clusters are produced, ripened and sputtered out high surface activity as no ligands or reducing agents are attached to of the matrix as a result of the collision cascades initiated by successive their surfaces [4]; (ii) versatility, as the cluster can be produced from incident ions. The cold matrix acts as the refrigerant needed to assist any base material (metal, alloy, semiconductor), while conventional body collisions [21]. Early applications of clusters from the MACS chemical routes rely on the availability of different precursors [4–6]; include catalysis of the oxygen evolution reaction (OER) over Co (iii) tunable defects, as under controlled conditions it is possible to clusters in alkaline media [25] and the vapour phase selective synthesize defect-rich material which can broad on their potential hydrogenation of 1-pentyne to 1-pentene over Pd and Pd/Au clusters applications [7, 8]. Furthermore, by coupling the CBD source with [4]. Therefore, this novel cluster source can be used to produce any the mass selection stage, it is possible to select the size of the clusters kind of nanostructured material, both single elements or alloys, with [9]. The cluster produced following this route are exploited in many unprecedented deposition rates. Its development represent a serious diverse research fields, ranging from magnetism [10, 11] and step forward in the prospective exploitation of cluster properties in plasmonics [12, 13] to catalysis [4, 14] and biomedicine [15]. technological applications. The beam current from such CBD sources is typically limited to Understanding the cluster emission mechanism is fundamental the 0.1–1 nanoampere regime. The amount of material generated is to improve the MACS performance further. In this paper we present Address correspondence to r.e.palmer@swansea.ac.uk 3070 Nano Res. 2019, 12(12): 3069–3074 a systematic experimental study of the effect of the incident ion beam angle upon the cluster beam flux, and of the angular distribution of the clusters emerging. The experiments are complemented by large scale molecular dynamic simulations. 2 Experimental and computational methods The configuration of the MACS employed in the experiments is shown elsewhere [22, 24] and the incident and collection angles are defined in Fig. 1. The matrix support was a 15 mm × 40 mm copper plate, cooled below 20 K with a continuous flow of liquid helium. Ar gas was introduced into the chamber through a leak valve and the pressure Figure 2 (a) Cross-sectional view of the initial configuration of the MD set at ~ 5 × 10−6 Torr during the matrix condensation. The Ag simulations. The Ag cluster atoms are grey and Ar matrix atoms orange. The red deposition rate into the matrix was 0.05 A/s (measured with a quartz dashed circle indicates the initial location of the incoming Ar atom, (dx, 0, dz). The crystal microbalance), corresponding to a metal loading of 0.8%. off-centre distance I is the perpendicular distance from the centre of the cluster The thickness of the matrix grown was around 100 nm. The energy to the beam trajectory. (b) The initial configuration of the multi-irradiation of the Ar+ ion beam used to sputter the matrix was 2.5 keV, with simulation: 338 2-nm Ag clusters were randomly embedded in the Ar matrix. current at the matrix of 3.2 μA. Minor of the ion beam fluctuations were addressed by setting the total charge deposited onto the cold We noticed that a single crystalline structure for the Ag cluster could finger constant (2.2 × 10−4 C). The sputtering time (also the cluster lead to strong channelling effects [27] at certain incident angles. deposition time) was 30 s for each sample. Referring to Fig. 1(b), Therefore amorphous clusters were generated by the following process: it is possible to change the sputtering angle, α, by rotating the matrix (i) a quasi-spherical cluster was cut from the bulk fcc Ag, (ii) the support. For each sputtering angle (α) employed, cluster were cluster was heated to 2,000 K for 10 ps and quenched to 10 K for collected at a range of different angles (β) by mounting different 5 ps, so the cluster adopts a poly-crystalline structure, (iii) the cluster TEM grids on the same sample stage. The number and size distributions was rotated to a random axial angle, and (iv) the cluster was half- of the deposited clusters were obtained from high-angular annular embedded in the Ar matrix and the interface relaxed to the local dark field (HAADF) images in a probe-corrected STEM (JEOL minimum energy to avoid overlap between Ag and Ar atoms. After 2100F, with a CEOS spherical aberration corrector). The HAADF the cluster was inserted, the simulation cell was additionally relaxed images were acquired with inner and outer collection angles of 62 over 100 ps, so the initial system was in equilibrium at 10 K. An and 164 mrad (camera length 10 cm). incoming Ar atom was initially placed at (dx, 0, dz), where dx is the The molecular dynamics (MD) simulations performed to study displacement in the x direction and dz was equal to 50 Å in all cases. the cluster emission process focused on the removal by sputtering We defined the off-centre distance I as the perpendicular distance of cluster already formed in the matrix (described as simulation from the centre of the cluster to the ion beam. We ran the simulations Type S2 in our previous paper [26]). In the current work, a larger with different values of dx, ranging from −60 to +60 Å (the simulation cell (up to 432 Å × 432 Å × 320 Å) was constructed in corresponding I ranges from −42.4 to 42.4 Å). The neutral Ar atom order to simulate higher energy (2 keV) ion impact. We embedded is inserted instead of the positive charged ion, because Coulombic one Ag cluster of different sizes: 6,631, 1,965 or 321 atoms, with interaction between the ion and bulk matrix can cause the complex diameters of 6, 4 or 2 nm, respectively, in an amorphous Ar matrix. charge transfer which cannot be caught by classical MD. All simula- The total number of the atoms in the system is about 1.5 million. tions were conducted using the classical MD code LAMMPS [28] Unlike the previous simulations where the Ag cluster was embedded and results visualized using OVITO [29]. at a distance of 10 Å below the open surface [26], here the cluster In order to directly compare with the experimental statistics, we was only half–buried by the matrix, as shown in Fig. 2(a). The centre also launched a multiple-impact irradiation simulation with an array of the cluster is at (0, 0, 0). The choice of the current simulation of embedded clusters. As shown in Fig. 2(b), the initial structure model is based on the benchmarking simulations as shown in the consists of 338 2-nm Ag clusters which are randomly embedded Electronic Supplementary Material (ESM). The cluster which is in the matrix. The size of this cell was 432 Å × 432 Å × 320 Å. The embedded at a distance of 2 nm below the open surface, had very corresponding Ag load was 4.7 at.%. The structure was first prepared limited chance to escape from the crater with a 45-degree impact and and relaxed following the same process as above. The Ar atoms with mainly remains half-buried on the surface. Therefore, we choose kinetic energies of 2 keV were injected at random positions laterally this well-defined model to study the effect of the energy deposition. from a starting point 8 nm above the original matrix surface. The incident angle, α, was kept at 30°. During the simulation, all the sputtered atoms 150 Å above the surface were removed from the simulation every 10,000 MD steps. We simulated 100 ion impacts and obtained 32 emission events. 3 Results and discussion The yield and angular distribution of clusters emerging from the matrix for each four of incidence angles (10º, 15º, 35º and 45º) were obtained from the cluster density in the STEM-HAADF images. Figure 3 shows a set of STEM-HAADF images, and corresponding size distribution graph for clusters prepared at α =10° for three different collection angles. In each case, the size of the clusters spans the range from 50 to over 5,000 atoms/cluster and the peak size Figure 1 Definition of α, the incident angle of Ar+ beam, and β, the collection slightly shifts down as β increases. However, this peak always lies angle of the emitted clusters in the MACS. between 200 and just over 300 atoms/cluster. | www.editorialmanager.com/nare/default.asp Nano Res. 2019, 12(12): 3069–3074 3071 Figure 3 Examples of size distributions of clusters produced from the MACS Figure 4 Examples of size distributions of clusters produced from the MACS matrix for a sputtering angle of α =10° and different collection angle β. matrix for different sputtering angles, α, integrated over the collection angle, β. Figure 4 shows STEM-HAADF images and corresponding size distributions for cluster prepared at three different sputtering angles, α, of 10º, 35º and 45º, and integrated over the whole range of collection angles, β, ranging from 1º to 120º. Again the peak size lies between 200 and 300 atoms/cluster. Most clusters (over 65%) contain from 150 to 450 atoms/cluster. The results of Figs. 3 and 4 indicate that the cluster size distribution does not depend much at all on either the sputtering or collection angles. In Fig. 5(a), the number of cluster collected as a function of the collection angle β is reported for different sputtering angles α. The overall yield of clusters increases for smaller α (more grazing incidence). The collection angle β at which the highest intensity is measured, βmax, decreases as α increases. As shown in Fig. 5(b), the relationship between α and βmax is in fact close to linear. Furthermore, the angle  subtended between the incident angle α and the corresponding βmax, as calculated using Eq. (1), is approximately constant with a value between 99° and 109°, as evidenced in Table 1 (the value decreases slightly with increased incident angle α). γ = 180 – (α + βmax) (1) Figure 5 (a) Cluster intensity as a function of collection angle, β, for incident From Figs. 5(a) and 5(b) it is possible to observe that the highest ion beam angle, α, of 10°, 15°, 35° and 45°. (b) Relationship between the optimal cluster flux is obtained when α = 10° and βmax = 60°. We use MD collection angle βmax and the incident angle α. simulations to shed some light on the atomistic origin of the observed angular dependence. In the benchmarking simulations, we study the Table 1 Angle  subtended between each α and the corresponding βmax as crater formation in the pure Ar matrix under 2 keV Ar irradiation calculated from Eq. (1) as shown in Fig. S1 in the ESM. The result shows that the craters from 10 independent simulations, have the depth of 5 nm and the α (°) βmax (°)  (°) diameter of 10 nm in average. The further simulations with small 10 61 109 Ag cluster embedded at the different depths (0, −1, −2 and −3 nm) 15 58 108 indicate that the significant emission only occurs with shallowly 35 43 102 embedded cluster (i.e., 0 and −1-nm embedded depth). The deeply 45 36 99 buried clusters mainly stay at the bottom of the crater as shown in www.theNanoResearch.com∣www.Springer.com/journal/12274 | Nano Research 3072 Nano Res. 2019, 12(12): 3069–3074 Fig. S2 in the ESM. Therefore, we will focus on the emission of the half-buried clusters. We identify two mechanisms which lead to any distribution we observe. If the Ar ion is stopped by the Ag cluster, as in case in Fig. 6(b), the deposited energy will be confined within the cluster before transfer to the Ar matrix across the whole matrix interface. This results in rather uniform sublimation of the matrix around the cluster, later leading on to nearly vertical cluster emission (β ~ 90°), as shown in Fig. 6(b6). In case of Fig. 6(a), the impact point is positively off the centre of the Ag cluster (I = 10.4 Å), and the primary energy deposition is confined within the cluster, as shown in Fig. 6(a3). Therefore, the emission direction is again about vertical. By contrast, in case of Fig. 6(c), the negative off-centre impact causes the energy deposition to be focused on the other side of cluster/matrix interface, so the resultant collection angle β is close to 60°. This result indicates that the primary energy deposition distribution must be an essential factor in determining the angular emission. Therefore, we further tested the effect of the initial impact point on the emission process. As shown in Fig. 6, the initial process of energy deposition is finished within the first 1 ps. The distribution of the deposited energy can be clearly mapped by examining the kinetic energies of atoms after about 1 ps, as shown in Fig. 6 in the third frame of each sequence. The full table is attached in Fig. S3 in the ESM, where more impact Figure 7 A schematic summary of the possible cases in the MD simulation. points are included. The dependence of the emission direction on the The incident angle α is 45°. (a) and (d) The impact point is far away from the cluster and no emission happens. (b) The impact point is close to the interface localized primary energy deposition is consistent with the selected of the cluster and the matrix and causes the emission on the opposite side of the results shown in Fig. 6. surface normal (β = 45°–90°). (c) The impact point is right on the cluster and the In the schematic illustration of Fig. 7, we summarize four situations, emission direction is vertical to the surface (β ~ 90°). depending on the impact point of the ion with respect to a 6 nm Ag cluster; the orange colour intensity indicates the deposited energy confirm that the primary location of the sublimation zone determines trend. As shown in cases (a) and (d), if the impact point is far away the angular dependence of the cluster emission. However, for 2-nm from the cluster-matrix interface, the cluster will not detach from clusters (200–300 atoms/cluster) the energy confinement within the the matrix because of insufficient energy transfer to the nanocluster. cluster is negligible. Therefore, these simulations struggle to explain In cases (b) and (c), the cluster and the interface can both stay in the off-normal except for clusters larger than about 4 nm, and we need the high-temperature state for long enough to enable the emission to look for other explanations for the experimental distribution of process. In (b) the cluster is emitted on the opposite of the normal Figs. 3 and 4. One factor that may merit consideration is the local to the surface with respect to the incident ion. A key concept in (c) surface roughness of the matrix, which is likely to affect the emission is the shadowing effect of the cluster for the incoming ion; a large process (see Fig. S4(b) in the ESM). In fact, the matrix surface fraction of ions with positive I will be stopped within the cluster, roughness increases from 0.85 to 35 A after only 100 ion impact without initiating any cascade in the Ar matrix in front of the events, as shown in Fig. S5 in the ESM. cluster. The energy deposited from the ions in the cluster will be In an attempt to better mimic the experimental conditions, we transmitted uniformly to the matrix and subsequently promote conducted another kind of simulation, by constructing a new matrix vertical cluster emission. Therefore, the average emission direction containing 338 randomly embedded clusters (rather than just one), for α = 45° lies at β = 45°–90° for the 6 nm cluster. When we simulate which corresponds to 4.7% Ag loading in the solid Ar matrix. the same process with smaller clusters of 4 nm in diameter, the results Although the metal loading is higher than the experimental value, the cascade and crater generated by 2 keV Ar is within 5 nm in radius. Therefore, the well separated clusters can be considered as independent systems without interfering each other. The sizes of all the clusters are about 2 nm in diameter (190–220 atoms/cluster). A hundred successive 2 keV irradiation events were simulated with the angle α set to 30°, which is representative of the experimentally investigated range. In these simulations, we observed 32 cluster emission events. As shown in Fig. 8, we identified these emission events as either “hit” or “miss”, based on the value of the emission angle β measured from the horizontal plane of the initial surface. If β > 90° (red lines in Fig. 8(a)), the emitted cluster will “miss” the sample stage; if β < 90° (green lines in Fig. 8(a)), the cluster will likely be collected on the sample stage and counted in the statistics of the angular dependence of the emission. The histogram of the emission Figure 6 Two-dimensional cross-sectional views of the MD simulations. The Ag angles of all 19 “hits” events is shown in Fig. 8(b). The most probable and Ar atoms are shown as large and small circles, respectively. The off-centre emission angle, βmax, from the simulations is around 50°, and the distance I is equal to (a) 10.6 Å, (b) 0 Å and (c) −10.6 Å, respectively. The incident subtended angle between incident ion and βmax is thus 100°. These angle α is 45° in all cases. The dashed lines in the first frame of each sequence results are in good agreement with the experimental results; in effect, show the incoming Ar ion. The dashed lines in the final frames show the emission the surface roughness created by ion bombardment selects a privileged direction of the Ag cluster. The colour coding shows the kinetic energy of the set of emitted trajectories for collection in the experiment. atom from 0 to 0.05 eV. A full table with I range from −42.4 to 42.4 Å is attached Analysing in detail the dynamics of Ag cluster emission, we in Fig. S3 in the ESM. identified two kinds of cluster emission event. In the majority of | www.editorialmanager.com/nare/default.asp Nano Res. 2019, 12(12): 3069–3074 3073 Figure 8 (a) The final configuration with illustrated trajectories of the emitted clusters (for the full movie, see Movie ESM1). The red trajectories indicate β angles > 90°, while the green trajectories indicate β angles < 90°. (b) The histogram of cluster with β angles < 90°. The most probable emission angle, βmax, from the simulations is around 50°, and the subtended angle is 100°. events, as shown in Fig. 9 Type I, the ion impacts in the vicinity of a than 5,000, and most of the collected clusters have a few hundred Ag cluster and generates a collision cascade in the Ar matrix. The atoms, as shown in the Figs. 3 and 4; smaller clusters are less common expansion of the heated cascade region generates a strong shock as already discussed in our previous paper [24]. In this range of wave in the cold matrix; energy and momentum are transferred to dimension (200–300 atoms/cluster) the simulations show that the the Ag cluster. If the cluster is close to the surface, and the cascade cluster emission direction follows the thermal spike’s path, described has initiated beneath it (more likely in grazing incidence), the shock in [26]: clusters are emitted for β both smaller and bigger than 90°. wave is able to provide sufficient kick for the cluster to escape. On Because of the deposition system geometry, only clusters emitted the other hand, a small fraction of the clusters receives a direct hit by for β < 90° are collected on the sample stage (Fig. 1). Therefore, the the impacting ions, as shown in Fig. 9 Type II. In this case, a large formed shock wave causes the neighbouring clusters to be emitted fraction of the energy is deposited into these particles, dramatically with β angles < 90° and, as a consequence, the number of uncollected raising their temperature, and there is no momentum transfer in clusters decreases. Furthermore, as small cluster emission is guided favour of the emission at a specific angle. But the high temperature by the Ar flow direction and thus by the formed shock wave of the particle will heat up the surrounding Ar, possibly resulting propagation, the optimal emission direction subtends an angle of in the random emission of the Ag cluster. The escaping clusters in around 100° with the incident direction, in harmony with the value Fig. 9, Type I and Type II, are indicated using dashed circular lines. obtained experimentally (as indicated in Table 1). Averaging over the different kinds of emission, the subtended angles are statistically biased towards 100°, which is independent of the incidence angle, as observed in the experiments (Fig. 5(a)), but is 4 Conclusions closely related to the shock wave formation and propagation. Moreover, In this paper, we have reported the angular distribution of clusters due to the limits of the computational time required, the authors generated in the MACS and shown that the highest cluster flux is note that several experimental parameters, such as size distribution obtained when the sputtering ion beam is near grazing (α = 10°). of the clusters and grazing angle emission rate, are not studied The measured cluster size was largely independent from MACS computationally in this paper. Further (Monte Carlo) simulations deposition system’s geometry. We observed a correlation between are necessary to comprehensively investigate these parameters, which the angle of incidence α and the collection angle with the highest is beyond the scope of the present study. cluster flux, βmax. The angle subtended is approximately constant The measured cluster sizes range from tens of atoms to more with value between 99° and 109°. Figure 9 Two different impact processes from the continuous irradiation simulations. Type I: the ion impacts on the Ar matrix. The energy and momentum is transferred from the matrix to the cluster via the collective motion of the Ar atoms. Type II: the ion impacts on a cluster. The energy spreads from the cluster to the surrounding matrix. The colour coding represents the velocities of atoms along the x axis. www.theNanoResearch.com∣www.Springer.com/journal/12274 | Nano Research 3074 Nano Res. 2019, 12(12): 3069–3074 By MD simulations, we showed that direct ion impacts on a large [7] Spadaro, M. C.; Luches, P.; Bertoni, G.; Grillo, V.; Turner, S.; Van Tendeloo, metal nanocluster 4–6 nm in diameter lead to deposited energy G.; Valeri, S.; D’Addato, S. Influence of defect distribution on the reducibility confinement within the cluster and uniform matrix sublimation of CeO2−x nanoparticles. Nanotechnology 2016, 27, 425705. [8] Singh, A. V.; Vyas, V.; Patil, R.; Sharma, V.; Scopelliti, P. E.; Bongiorno, G.; around it. Such nanoclusters emit mainly normally to the surface Podestà, A.; Lenardi, C.; Gade, W. N.; Milani, P. Quantitative characterization of the substrate. The non-uniform fast sublimation of the Ar matrix of the influence of the nanoscale morphology of nanostructured surfaces from one side of the cluster, on the other hand, will promote biased on bacterial adhesion and biofilm formation. PLoS One 2011, 6, e25029. emission. These results suggest that the cluster size and the local [9] von Issendorff, B.; Palmer, R. E. A new high transmission infinite range surface configuration should regulate the emission of nanoclusters, mass selector for cluster and nanoparticle beams. Rev. Sci. Instrum. 1999, since for the large size clusters the energy is mainly deposited within 70, 4497–4501. [10] Skumryev, V.; Stoyanov, S.; Zhang, Y.; Hadjipanayis, G.; Givord, D.; Nogués, or near the cluster. The emission of small clusters 2 nm in diameter J. Beating the superparamagnetic limit with exchange bias. Nature 2003, from a cryogenic matrix is shown to be driven by the shock waves 423, 850–853. which are generated by the collision cascades ignited by incoming [11] Benelmekki, M.; Bohra, M.; Kim, J. H.; Diaz, R. E.; Vernieres, J.; ions beneath the near-surface nanoclusters. The direct comparison Grammatikopoulos, P.; Sowwan, M. A facile single-step synthesis of ternary of MD simulations and the experiments shows a good measure of multicore magneto-plasmonic nanoparticles. Nanoscale 2014, 6, 3532–3535. agreement between them. The MACS cluster emission mechanism [12] D’Addato, S.; Pinotti, D.; Spadaro, M. C.; Paolicelli, G.; Grillo, V.; Valeri, S.; investigated here should lead to more efficient cluster beam Pasquali, L.; Bergamini, L.; Corni, S. Influence of size, shape and core-shell interface on surface plasmon resonance in Ag and Ag@MgO nanoparticle generation/collection, as required by industrial applications. films deposited on Si/SiOx. Beilstein J. Nanotechnol. 2015, 6, 404–413. [13] Pelli Cresi, J. S.; Spadaro, M. C.; D’Addato, S.; Valeri, S.; Benedetti, S.; Acknowledgements Di Bona, A.; Catone, D.; Di Mario, L.; O’Keeffe, P.; Paladini, A. et al. Highly efficient plasmon-mediated electron injection into cerium oxide We thank the EPSRC, the Leverhulme Trust, Innovate UK and the from embedded silver nanoparticles. Nanoscale 2019, 11, 10282–10291. EU H2020 project CritCat (project ID: 686053) for their financial [14] Baxter, E. T.; Ha, M. A.; Cass, A. C.; Alexandrova, A. N.; Anderson, S. L. support. Research performed at the University of Helsinki was Ethylene dehydrogenation on Pt4, 7, 8 clusters on Al2O3: Strong cluster size supported by the EU project M4F (project ID: 755039). The authors dependence linked to preferred catalyst morphologies. ACS Catal. 2017, 7, 3322–3335. are also grateful to the Finnish IT Centre of Science, CSC, for grants [15] Carbone, R.; Giorgetti, L.; Zanardi, A.; Marangi, I.; Chierici, E.; Bongiorno, of computational time. The MACS concept is the subject of the G.; Fiorentini, F.; Faretta, M.; Piseri, P.; Pelicci, P. G. et al. Retroviral following patent applications: Palmer, R. E.; PCT Patent Application microarray-based platform on nanostructured TiO2 for functional genomics No. WO2013017870; Europe Patent Application No. EP2739565; and drug discovery. Biomaterials 2007, 28, 2244–2253. China Patent Application No. CN103857618; Japan Patent Application [16] Pratontep, S.; Carroll, S. J.; Xirouchaki, C.; Streun, M.; Palmer, R. E. No. JP6316748; U.S. Patent Application No. US10233545. Size-selected cluster beam source based on radio frequency magnetron plasma sputtering and gas condensation. Rev. Sci. Instrum. 2005, 76, 045103. [17] Wegner, K.; Piseri, P.; Tafreshi, H. V.; Milani, P. Cluster beam deposition: Electronic Supplementary Material: Supplementary material A tool for nanoscale science and technology. J. Phys. D Appl. Phys. 2006, (additional MD simulation and matrix surface roughness information) 39, R439–R459. is available in the online version of this article at https://doi.org/ [18] Llamosa, D.; Ruano, M.; Martinez, L.; Mayoral, A.; Roman, E.; Garcia- 10.1007/s12274-019-2553-y. Hernández, M.; Huttel, Y. The ultimate step towards a tailored engineering of core@shell and core@shell@shell nanoparticles. Nanoscale 2014, 6, Open Access This article is licensed under a Creative Commons 13483–13486. Attribution 4.0 International License, which permits use, sharing, [19] Martínez, L.; Lauwaet, K.; Santoro, G.; Sobrado, J. M.; Peláez, R. J.; Herrero, V. J.; Tanarro, I.; Ellis, G. J.; Cernicharo, J.; Joblin, C. et al. adaptation, distribution and reproduction in any medium or format, Precisely controlled fabrication, manipulation and in-situ analysis of Cu as long as you give appropriate credit to the original author(s) and based nanoparticles. Sci. Rep. 2018, 8, 7250. the source, provide a link to the Creative Commons licence, and [20] Kouznetsov, V.; Macák, K.; Schneider, J. M.; Helmersson, U.; Petrov, I. A indicate if changes were made. novel pulsed magnetron sputter technique utilizing very high target power The images or other third party material in this article are included densities. Surf. Coat. Technol. 1999, 122, 290−293. in the article’s Creative Commons licence, unless indicated otherwise [21] Palmer, R. E.; Cao, L.; Yin, F. Note: Proof of principle of a new type of in a credit line to the material. If material is not included in the cluster beam source with potential for scale-up. Rev. Sci. Instrum. 2016, 87, 046103. article’s Creative Commons licence and your intended use is not [22] Oiko, V. T. A.; Mathieu, T.; Cao, L.; Liu, J.; Palmer, R. E. Production of permitted by statutory regulation or exceeds the permitted use, you silver nanoclusters using a matrix-assembly cluster source with a solid will need to obtain permission directly from the copyright holder. CO2 matrix. J. Chem. Phys. 2016, 145, 166101. To view a copy of this licence, visit http://creativecommons.org/ [23] Ellis, P. R.; Brown, C. M.; Bishop, P. T.; Yin, J. L.; Cooke, K.; Terry, W. D.; licenses/by/4.0/. Liu, J.; Yin, F.; Palmer, R. E. The cluster beam route to model catalysts and beyond. Faraday Discuss. 2016, 188, 39–56. [24] Cai, R. S.; Jian, N.; Murphy, S.; Bauer, K.; Palmer, R. E. A new method to References prepare colloids of size-controlled clusters from a matrix assembly cluster [1] Knight, W. D.; Clemenger, K.; De Heer, W. A.; Saunders, W. A.; Chou, M. source. APL Mater. 2017, 5, 053405. Y.; Cohen, M. L. Electronic shell structure and abundances of sodium [25] Xu, J. Y.; Murphy, S.; Xiong, D. H.; Cai, R. S.; Wei, X. K.; Heggen, M.; clusters. Phys. Rev. Lett. 1984, 52, 2141–2143. Barborini, E.; Vinati, S.; Dunin-Borkowski, R. E.; Palmer, R. E. et al. [2] Echt, O.; Sattler, K.; Recknagel, E. Magic numbers for sphere packings: Cluster beam deposition of ultrafine cobalt and ruthenium clusters for Experimental verification in free xenon clusters. Phys. Rev. Lett. 1981, 47, efficient and stable oxygen evolution reaction. ACS Appl. Energy Mater. 1121–1124. 2018, 1, 3013–3018. [3] Palmer, R. E.; Pratontep, S.; Boyen, H. G. Nanostructured surfaces from [26] Zhao, J. L.; Cao, L.; Palmer, R. E.; Nordlund, K.; Djurabekova, F. Formation size-selected clusters. Nat. Mater. 2003, 2, 443–448. and emission mechanisms of Ag nanoclusters in the Ar matrix assembly [4] Palmer, R. E.; Cai, R. S.; Vernieres, J. Synthesis without solvents: The cluster source. Phys. Rev. Mater. 2017, 1, 066002. cluster (nanoparticle) beam route to catalysts and sensors. Acc. Chem. Res. [27] Nordlund, K.; Djurabekova, F.; Hobler, G. Large fraction of crystal directions 2018, 51, 2296–2304. leads to ion channeling. Phys. Rev. B 2016, 94, 214109. [5] Grammatikopoulos, P.; Steinhauer, S.; Vernieres, J.; Singh, V.; Sowwan, M. [28] Plimpton, S. Fast parallel algorithms for short-range molecular dynamics. Nanoparticle design by gas-phase synthesis. Adv. Phys. X 2016, 1, 81–100. J. Comput. Phys. 1995, 117, 1–19 [6] D’Addato, S.; Spadaro, M. C. Low pressure bottom-up synthesis of [29] Stukowski, A. Visualization and analysis of atomistic simulation data metal@oxide and oxide nanoparticles: Control of structure and functional with OVITO–the open visualization tool modelling. Simul. Mater. Sci. properties. Phys. Scr. 2018, 93, 033001. Eng. 2010, 18, 015012. | www.editorialmanager.com/nare/default.asp