Plasma-assisted gas-phase aggregation of clusters for functional nanomaterials

Plasma-enhanced chemical vapor deposition of various gaseous compounds including plasma polymerization of organic precursors, RF magnetron sputtering of polymers, and DC and RF magnetron sputtering of various solids have been adopted as powerful tools for the production of functional nanostructured materials. In addition, intensive research is going to employ nanoparticles (NPs) prepared using plasma-assisted gas-phase synthesis. Such NPs have been considered for numerous potential applications including optoelectronic energy conversion, electrocatalysis for fuel cells, or novel plasmonic materials superior to Au and Ag. Electroluminescence, macromolecular self-assembly, supramolecular chemistry, and drug delivery are considered as well. Significant progress has been achieved in this area, although little experimental evidence has been found about the processes occurring during NPs nucleation, growth, and transport to the receiver. Another drawback lies in the relatively low production (deposition) rate of NPs, to be attractive for the industry. In this review, we concisely summarize the state-of-the-art plasma-based gas aggregation cluster sources, their current limitations, and possible directions for future progress. NPs deposited by “slow landing” on various substrates and the deposition into liquids are considered.

Nanotechnology, i.e., the ability to control the creation and arrangement of materials on the nanometer scale, is currently one of the most dynamically developing fields of human research. The ever-growing interest in nanomaterials including NPs is primarily motivated by their unique properties associated with their dimensions. This applies both to the enormous specific surface, i.e., a property essential for applications such as (photo)catalysis, sensing, energy storage, and harvesting, and to their unique physicochemical properties, which are not achievable with bulk materials. Last but not least, the dimensions of NPs are comparable to some biological objects (proteins, viruses), which represents a fascinating possibility to use nanomaterials to influence biological processes, or to prepare hybrid functional organic–inorganic units.

Due to the great application potential of NPs, it is, therefore, not surprising that ways are being sought for their effective and reliable preparation. In the last three decades, considerable interest has been paid to the vacuum-based plasma-assisted sources of NPs. Among many proposed concepts of NP sources, those based on gas aggregation have become the most common. In these sources, the supersaturated vapor undergoes spontaneous nucleation and subsequent growth of formed proto-particles by atom attachment or coalescence. The first technical realization of this concept (however, without plasma assistance at that time) dates back to the 1970s and 1980s, when the evaporation of metal in a cold inert gas atmosphere was proposed by Hogg and Silbernagel,¹ Sattler et al.,² Frank et al.,³ and Takagi et al.^3,4 in a system based on the evaporation cell with a thick lid containing a tube-like orifice. Vapors of the metal were compressed when passing the orifice and then underwent an adiabatic expansion that cooled them down in space with very low pressure (UHV). This provided the cooling effect needed to sustain created clusters.

One should keep in mind that NPs produced in the sources based on material evaporation were not electrically charged, making it difficult to manipulate them to perform mass selection, deflection, and/or acceleration for fast landing.⁴ Charging has been performed additionally by letting the neutral cluster beam pass through an electron-impact ionizer.^5,6 Takagi’s idea was to deposit films with an energetic cluster impact. The charging problem was simply solved by Haberland and his co-workers, who used the plasma of a planar magnetron for material sputtering, instead of conventional evaporation.⁷ The first design in 1991 consisted of two planar magnetrons facing each other [Fig. 1(a)]. Later in 1994, one planar magnetron was used within an aggregation chamber that ended with an exit orifice [Fig. 1(b)].⁸ However, Haberland et al.⁸ could employ negatively charged clusters for ion cluster energetic impact deposition, whereas positively charged and neutral clusters were wasted. At present, researchers use all clusters (NPs) for deposition, i.e., positively and negatively charged, and neutral in most cases, as ion cluster energetic impact for this purpose is rarely applied; however, it is still used for cluster beam etching.⁴ 

This type of source has become very popular since then. Different terminology and abbreviations have been used, including sputter gas aggregation source (SGAS), magnetron-based gas aggregation source (m-GAS), multiple-ion cluster source (MICS), and others. For simplicity, “GAS” will be used in the following paragraphs of this review, implying that it is based on magnetrons.

At the same time, NP sources based on different principles appeared. Ishii and co-workers propose the so-called flow sputtering that is based on hollow cathode discharge [Fig. 1(c)].⁹ By these sources, many different types of NPs, e.g., Fe NPs,¹⁰ and the so-called granular films have been deposited.¹¹ 

Another concept of the gas aggregation cluster source, the so-called laser vaporization cluster source (LVCS), was introduced by Maruyama et al.¹² and Milani and deHeer.¹³ In this source, which is schematically depicted in Fig. 2(a), the plasma produced by laser pulses vaporizes a material for NP creation. Although laser ablation assures excellent stability of the NP LVCS systems, these sources are characterized by a low production rate of NPs. This limitation was solved by the Milano group in a pulsed microplasma cluster source (PMCS).¹⁴ The working principle of PMCS is based on spatially confined pulsed plasma discharge ablation of a target placed in a condensation chamber. The vaporized species are quenched by a pulse of inert gas and condense to form NPs. A schematic representation of such a NP source is shown in Fig. 2(b).

Pilch et al.^15,16 proposed using high-power pulsed hollow cathode (HC) discharge, where a high degree of ionization of the sputtered species can take place [Fig. 2(c)]. NPs are formed by sputtered vapor in the gas phase by nucleation, coagulation, and accretion by the attachment of atoms and ions. When NPs grow larger, they attain a negative charge that attracts metal positive ions. This is supposed to speed up NPs growth and substantial enhancement of the production of NPs. The drawback of this method is that the cathode must be made of a conducting material.

At this point, we should also mention plasma-enhanced chemical vapor deposition (PECVD) methods for the preparation of NPs. This goes back to plasma polymerization (PECVD of organic compounds), e.g., plasma polymerization of ethylene when in 1973M. Shen and co-workers investigated the powder appearance.¹⁷ In addition, the PECVD of silanes by Mangolini et al.¹⁸ for Si nanocrystals’ production and the recent preparation of TiN nanoparticles with plasmonic properties underlined the importance of these investigations.¹⁹ However, these studies gave rise to a huge field of dusty plasma that is behind of the scope of this paper.

The latest interesting addition to the family of gas aggregation cluster sources is Matrix Assembly Cluster Source (MACS).²⁰ The matrix is created by the condensation of atomic metals and noble gases (such as Au and Ar) on a cryogenic support grid held at 15 K. Subsequently, Ar ions at an energy of 1 keV bombard the matrix and produce clusters by sputtering. The main advantage of this concept of NP production is their excellent production rate that can reach values higher than several tens of mg/h.²¹ 

Although various concepts of gas aggregation sources of NPs were suggested during the last three decades, the majority of researchers have employed GAS systems similar to those first introduced by the Freiburg group in 1994. Different variants were developed intended for the synthesis of heterogeneous (core–shell) NPs using single magnetrons with segmented targets,²² multiple magnetrons,²³ or systems based on the in-flight coating/modification of NPs produced by conventional GAS.²⁴ In Secs. III and IV, we will briefly highlight the advantages of such NP sources (Sec. III) as well as their current limitations and future perspectives (Sec. IV).

The ongoing popularity of gas aggregation sources and, in particular, those based on magnetron sputtering relates to their relative simplicity and unique features that make them highly competitive to other means of production of NPs. First of all, unlike more commonly used approaches based on chemical synthesis, the synthesis of NPs by GAS systems is a fully solvent-, ligand-, and linker-free deposition process. This allows the production of high-purity NPs without the need for time-consuming and laborious purification steps. In addition, vacuum-based NP growth avoids particle agglomeration usually present in the wet chemical process. Furthermore, the physical and “dry” way of NP production is in accordance with the current requirement for “green” technology.

The second important feature is the high versatility of the GAS deposition technique in terms of materials from which NPs may be prepared. Although this was traditionally related to metallic NPs including the ones with high melting points (e.g., Au,^23,25 Ag,^26–29 Pt,^30–32 Mo,³³ Cu,^34–37 Ta,³⁸ Pd,^39–41 Fe,⁴² Co,^43–45 Ni,⁴⁶ W,^47,48 Ti,^33,49–51 Mn,⁵² V,^33,53,54 Nb,^55–57 and Ru⁵⁸), GAS was also successfully applied to produce NPs of metal oxides,^59–61 oxynitrides,⁶² semiconductors,^63–65 or carbides.⁶⁶ Numerous studies proved the applicability of this type of NP source also for the production of temperature-sensitive and nonconducting materials, such as plasma-polymer NPs.^67,68 Typically, GAS uses a planar magnetron of diameter from 1 to 3-in installed in the cylindrical aggregation chamber. The diameter of the aggregation chamber is 100 or 150 mm. The distance between the magnetron and the exit orifice, named the aggregation length, is usually tuneable from 70 to 250 mm. Magnetron is operated in the DC mode with the current ranging from 0.5 to 3 A at a working gas pressure between 20 and 200 Pa and flow from 2 to 100 SCCM. However, some research groups use 10 times higher flow rates. In the case of nonconducting materials, such as polymers and some oxides, RF-powered magnetrons must be used with RF powers typically from 50 to 300 W. The selection of materials for NPs production using GAS is basically similar to conventional magnetron sputtering.

Furthermore, GAS systems based either on composite/segmented targets or utilizing multiple magnetrons in a single aggregation chamber are reported to also enable synthesizing alloy NPs^23,69,70 or more complex two- or even three-component ones with architectures ranging from core–shell, core–shell–shell or multi-core–shell structures to NPs with a Janus-like or core-satellite character^71–78 (see Fig. 3).

Another important feature of the gas aggregation sources of NPs relates to the fully vacuum-based character of the production of NPs. This makes it possible to in-flight modify/coat NPs before they reach the substrate, e.g., by an auxiliary plasma located in between the output orifice of the aggregation chamber and substrate to be coated⁷⁹ or, as will be discussed in Sec. IV C by their in-flight coating by PVD or PECVD techniques. In addition, the fact that NP production occurs in the aggregation chamber and is, thus, fully separated from the main deposition chamber, into which NPs enter in the form of a beam, makes it also possible to use more GAS systems in one deposition chamber simultaneously or combine GAS set-ups with other vacuum-based deposition techniques. The latter already paved the way for the production of a wide range of functional nanocomposite coatings, such as metal/plasma-polymer antibacterial coatings,^80–82 nanocomposites with tailor-made plasmonic properties for optical applications,^83,84 broadband thin-film absorbers,⁸⁵ super-wettable coatings for water–oil separation,⁸⁶ nanostructured coatings with improved biocompatibility,⁸⁷ or memristive nanocomposites.⁸⁸ 

The final two key advantages are good directionality of the deposition process and the possibility to coat NPs onto any type of substrate that withstands the low-vacuum conditions. While the latter will be discussed in Sec. IV E for the case of a liquid substrate, the beam-like character of the deposition process makes it possible to fabricate 1D or 2D gradient arrays of NPs,^89,90 prepare patterned NP films or, as shown recently, produce columnar coatings composed of individual NPs⁵³ (see Fig. 4).

As shown in the previous sections, GAS systems offer numerous advantages and experienced tremendous progress in the last three decades. Therefore, it might be rather surprising that the use of these NP sources on an industrial scale remains limited. In this section, we will try to identify and discuss several issues that hamper the broader use of gas aggregation sources. Before doing so, it has to be emphasized that the intention is not to provide a comprehensive list of obstacles in the way of GASes from the laboratory to industry, but rather to outline, at least from our point of view, the highly relevant ones related to vacuum science and technology that have to be addressed in order to increase the attractiveness of GAS systems to industrial use. These relate to

• insufficient deposition rate;

• stability and reproducibility of the deposition process;

• lack of experimental data needed for understanding the formation and growth of NPs;

• limited adhesion of deposited NPs to the substrate;

• possible use of GAS systems for the production of novel nanomaterials, such as nanofluids; and

• uniformity, control of size distribution, etc.

The first factor that represents a serious obstacle to the wider spread of GAS systems is their rather low production yield. This relates not only to the relatively low deposition rate that reaches several tens of mg/h at maximum but also to an inadequately low fraction of sputtered material that is consumed for NP growth.

Concerning the low production yield, it represents a serious obstacle in the case of costly materials such as gold, platinum, or palladium. The low deposition rate and yields are related to the mechanism of NPs formation and their transport out of the aggregation source. According to the basic scenario, the first step of metal NP formation is due to the spontaneous nucleation of supersaturated vapor close to the erosion zone of the magnetron that leads to the production of proto-particles. The proto-particles further grow by the attachment of atoms or coalescence and are directed by the flow of the working gas out of the aggregation chamber. The working gas also provides cooling. Although this description is factually correct, it represents an oversimplification of the real processes taking place in the aggregation chamber.

First, it does not take into account the gas flow pattern that influences the movement of NPs. This can lead to the trapping of NPs as shown in Fig. 5(a) or eventually to their redeposition onto the magnetron, especially when gas vortexes are formed. Backward NP deposition and their loss on the walls were experimentally observed by Nikitin et al.⁹¹ who reported on a substantially lower number of NPs that leave the aggregation source. To overcome this unfavorable situation, a rational design of gas flow patterns in the aggregation chamber, including the gas inlet position, is needed.^92,93

Drag force and Brownian diffusion are the main factors that influence NP transport after they leave the plasma zone.⁹⁴ Because the particle size has a huge influence on the Brownian force as compared to the drag force and due to the strong dependence of the drag force on the gas velocity, smaller particles and gas flows favor the Brownian motion. The situation becomes critical if the sizes of particles become less than 5 nm and the gas flow velocity drops below 10⁻² m/s. As it was shown by Zhang et al.⁹⁵ for planar magnetron, the Brownian force creates a great difference in NPs’ trajectory [see Fig. 5(b)].

Speaking about the plasma zone, NPs become charged and, thus, experience other forces, including thermophoretic force given by the temperature gradients, ion drag force induced by the momentum transfer from directed fluxes of ions and electrostatic forces. Due to the different directions of forces acting on NP, it might stay to be trapped in a region close to the central part of the magnetron as was confirmed by the experiments using in situ SAXS^96,97 or UV–Vis spectrophotometry.^91,92 Such trapped particle further grows either by the attachment of new atoms or coalescence. However, as the acting forces are dependent on the NP size, as soon as the size of the NP reaches a certain critical value, the forces in the direction of the output orifice surpass the ones acting in the opposite direction. At this moment, NPs are expelled from the trapping zone and are directed by the flow of the working gas out of the aggregation chamber.

NPs trapped in the regions close to the magnetron experience not only an influx of incoming atoms but are also heated. This may lead to their partial or entire evaporation and, thus, to the reduction of the production yield. Furthermore, the aforementioned scenario may, under certain conditions, lead to the onset of cyclic instabilities in terms of both deposition rate and plasma parameters. Such instabilities are reported both in the case of the production of plasma-polymer NPs⁷⁸ [see Fig. 6(a)] and metallic NPs.⁹⁷ However, due to the significantly different densities of metallic and plasma-polymer NPs, the frequencies of oscillations are markedly different; while in the case of plasma-polymer NPs, the instabilities have a frequency of tenths of Hz, the reported frequencies in the case of metallic NPs is much lower and close to 1 Hz and, thus, easily overlooked.

The above-mentioned trapping/growth/release character of the deposition process has an important consequence; it can be presumed that the release of trapped NPs might be controlled by pulsing the magnetron plasma, i.e., by “switching” on and off the electrostatic forces. As demonstrated in Fig. 6(b), switching off the plasma indeed results in the release of NPs from the trapping zone and the corresponding pulse in the deposition rate.⁹⁷ This may be used to tailor the residence time of NP in the aggregation zone and, thus, its final size.

Another issue, which relates to the performance of GAS systems, is the temporal instability of the production of NPs, both in terms of their production rate and size distribution. As it was highlighted by Ganeva et al.,⁹⁸ the principal parameter that influences the production rate/size distribution of NPs is the evolution of the erosion zone during the long-term operation of a planar magnetron. The continuous erosion modifies the geometry of the erosion track and if the trajectory of ions does not change the incident angle θ of ions that bombard the target surface increases. A small increase of θ from 0°, which is the initial value for the newly installed target, causes an increase of the sputtering yield due to its angular dependence, and, thus, higher numbers of free metal atoms are produced. This, in turn, results in a larger number of NPs and/or the formation of larger NPs. However, as soon as the critical value of the incident angle (∼60°) is reached, a further increase in the incidence angle causes a dramatic decrease in the sputtering yield and, consequently, a significant reduction in the number and/or size of produced NPs. This scenario further explains the experimental observation that the formation of NPs stops considerably earlier than the completion of target sputtering, i.e., before the depth of the erosion zone reaches the target thickness.

Apart from the variation in the number/size of produced NPs during the life cycle of a target, another important drawback connected with the use of conventional planar magnetrons is the limited utilization/consumption of the target. This leads to the need for its frequent replacement, i.e., the procedure is not favorable for industrial use. This situation, which is common for sputtering using planar magnetrons, is even worsening in the case of GAS systems, as the higher aggregation pressure necessary for reaching the super-saturation conditions leads to the narrowing of the erosion track. The possible solution proposed by Huttel et al.⁹⁹ relies on the use of a so-called full-face-erosion (FFE) magnetron, i.e., a magnetron, in which the movement of the magnetic circuit enables to sweep off the whole surface of the target. This not only substantially enhances the target consumption (e.g., for a 2-in. magnetron, Au target consumption increased from 5% up to 20%,⁹⁹ see Fig. 7) but also enables to overcome the above-discussed issues related to the time evolution of the shape of the erosion groove.

Recently, Nikitin et al.¹⁰⁰ proposed a novel experimental design in which a post(cylindrical) magnetron with a rotating magnetic circuit is placed in the aggregation chamber instead of a planar magnetron [Fig. 8(a)]. This configuration, which improves the target consumption, was found to be also capable to limit the cyclic trapping/release of NPs due to electrostatic trapping by increasing the rotation speed of the magnetic circuit. In addition, the gas flow pattern may better collect and transport NPs without the loss on the walls [Figs. 8(b) and 8(c)]. This GAS solution has the potential of further tuning its parameters such as magnetic circuit design and position of the postmagnetron.

To conclude this part, from the practical standpoint, further optimization of the production yield is possible by carefully designed geometries of the aggregation chambers with well-tailored gas flow patterns and by developing and utilizing novel magnetron configurations, e.g., such as shown in Figs. 8(b) and 8(c). This is of paramount importance also from the point of view of the upscale of GAS systems required by the industry.

As the functional properties of NPs are strongly related to their size, shape, composition, and structure, another highly relevant issue for the application of GAS systems is tailoring and controlling these characteristics. Although the formation of NPs is a complex process, their growth is generally accepted to proceed in the initial phase by the attachment of atoms to the already formed proto-particles. These can be in the first approximation assumed to have a close-to-spherical shape. However, the further growth of NPs may lead to different shapes ranging from octahedral or tetrahedral to cubic or even more complex shapes [for selected examples, see Figs. 9(a)–9(e)]. Furthermore, even for the same deposition system, various shapes of the produced NPs are readily observed with variations in the deposition conditions. An often-reported example of this is presented in Fig. 9(e), where TEM images of Fe NPs prepared at a constant aggregation pressure but with different magnetron powers are presented. The reason for this lies in an interplay of the number of incoming adatoms (i.e., the deposition rate), their diffusion that can be significantly different on different crystallographic facets of growing NPs, and the temperature of the system. Such a complex behavior is, nowadays, investigated by computer modeling that significantly improved our understanding of the processes responsible for the structure of formed NPs. For instance, based on the comparison of numerical simulations and experimental results, Zhao et al. suggested a “phase” model that accounts for different behaviors of the Fe atoms deposited on {100} and {110} surfaces at different temperatures.⁴² This model succeeded in explaining the shape transformation of Fe NPs from spherical to cubic shape.

Naturally, the situation becomes even more complicated as soon as two (or even more) component systems are considered. In these cases, additional factors such as a mutual interaction between the atoms of different materials, their phase separation, or site-specific wetting, must be taken into account in order to explain the various structures of fabricated NPs.^101–104 However, although the proposed theoretical models and simulations have become valuable and powerful tools to investigate and predict NP structures, they should still be considered an approximate description of the real systems. Besides a certain level of simplifications that are due to the inherent complexity of the process of NP formation (e.g., neglect of the complicated movement of NPs including the aforementioned trapping of NPs in the aggregation chamber), the principal limitation of simulations lies in the insufficient knowledge of the plasma/gas characteristics and state of growing NPs depending on the deposition conditions. This is particularly true for the temperature of the gas and NPs, densities and energies of charged and neutral species in the plasma, as well as their spatiotemporal evolutions. Because of this, further development in the understanding of NP growth is not possible without targeted experiments pointed to the evaluation of the aforementioned characteristics using advanced in situ diagnostics techniques.

The situation is complicated not only by the relatively high pressure but also by the necessity to perform the measurement in a closed and not easily accessible aggregation chamber. The first attempt in this direction has been done by Kousal et al.¹⁰⁵ and Gauter et al.¹⁰⁶ These authors utilized either a Langmuir or calorimetric probes for mapping the densities and energies of charged particles and heat fluxes in the aggregation chamber. Based on the analysis of the experimental results, they have come to the conclusion that regions with steep gradients of the plasma potential appear in the positions, where trapping of NPs was observed, and that the integral energy flux is predominantly driven by the contribution due to film condensation and reflected neutrals, with the latter being strongly dependent on the mass ratio between the gas and target atoms. However, in situ measurement of the temperature evolution of growing NPs still remains rather challenging.

As already mentioned in Sec. II briefly, one possible strategy to produce two-component NPs is based on the in-flight coating of NPs before they are collected on a substrate. Pioneering works in this direction were done by Balasubramanian et al.¹⁰⁷ and Cassidy et al.²⁴ While the first mentioned article reports on the possible production of core–shell NPs by the in-flight coating of TiO₂ NPs by evaporated paraffin, the later study investigated the formation of Si NPs decorated by silver nanoclusters produced by an auxiliary magnetron introduced between the aggregation chamber and the main deposition chamber. Since then, numerous alternative solutions were proposed and tested. For instance, Cu/plasma-polymer NPs are synthesized by coating Cu NPs with plasma polymers produced by an auxiliary RF source located in between the aggregation source of Cu NPs and the main deposition chamber [Fig. 10(a)].¹⁰⁸ GAS connected to the modification chamber with two 3-in. planar magnetrons facing each other with the axis of magnetrons perpendicular to the beam of NPs employed for the fabrication of Ag/Ti NPs [Fig. 10(b)].¹⁰⁹ Ni or Ag NPs is successfully coated by Cu using a tubular magnetron attached to a gas aggregation source [Fig. 10(c)].¹¹⁰ Strawberry-like silver/plasma polymers are synthesized by sputter deposition of Ag onto plasma-polymer NPs [Fig. 10(d)].^77,111

As compared to the systems based on the segmental targets or multiple magnetrons in a single deposition chamber, the spatiotemporal separation of core and shell productions may be seen as a significant simplification to some extent. Despite the apparent simplicity of such systems, two principal open questions may be identified.

The first one is related to the velocity of “primary” (core) NPs as they pass through the modification zone. According to several experimental studies published in the last decade, the typical velocity of NPs in the modification zone ranges from tens to hundreds of m/s.^112–114 Considering the length of the modification zone that is in order of several tens of cm, such relatively high velocities mean that the residence time of NPs in the coating zone is less than 0.1 s. Such a low residence time is too short for the formation of the conformal shell layer. Because of this, “primary” NPs must be significantly slowed down. NP deceleration can be achieved by the collision with the working gas and, theoretically, NP speed may be as low as the drift velocity of the working gas. To make this possible, the pressure in the modification zone must be significantly increased as a higher pressure means a lower drift velocity and, thus, a shorter deceleration distance.¹¹⁵ However, the higher pressure leads to a decrease in the deposition rate of the second material (i.e., “shell” material) in the case of magnetron sputtering. As a result, a balance between collisional-induced deceleration and the deposition rate of the shell needs to be found for an optimized deposition. Moreover, as the pressure in the modification zone reaches a critical value, the modification zone starts to behave as an aggregation chamber, and NPs from the second material (i.e., “secondary” NPs) begin to be formed. Such “secondary” NPs are then deposited alongside the primary ones, leading to the formation of heterogeneous NP films composed of NPs of two types, instead of core–shell NPs.⁷⁸ 

The second issue connected with the in-flight coating of NPs by plasma-based deposition sources is the interaction between NPs and the plasma; “primary” NPs that enter the plasma zone are readily recharged and their trajectory starts to be affected by the plasma in the modification zone. NPs–plasma interaction under certain conditions leads to the trapping of NPs as evidenced, for instance, in the study of Košutová et al.¹¹⁶ This effect prolongs the residence time of NPs in the modification zone and, thus, might be seen as positive in terms of the possibility of reaching the desired thickness of the shell. In addition, the trapping and, hence, the thickness of the deposited shell were reported to be controllable, e.g., by an external magnetic field applied to the tubular magnetron.¹¹⁰ This controlled trapping of NPs in the coating zone, thus, opens the way to produce shells with thicknesses variable in a certain range. However, the trapping of NPs in a deposition zone also has possible drawbacks, which are connected to the onset of cycling instabilities in the deposition rate,¹¹⁰ possible loss of NPs on the walls of the modification zone, or excessive heating of growing NPs that might slow down or even fully suppress the shell formation. Here again, more targeted experiments and simulations focused on the behavior of NPs in the auxiliary plasma zone are urgently needed to optimize the performance of the in-flight coating of NPs.

In the case of common GAS systems with no acceleration of NPs, the NP deposition proceeds in a soft-landing regime. In this case, the kinetic energy per atom E_at is below the binding (cohesive) energy of NP constituents. Under such impact conditions, the composition, as well as the shape and structure of deposited NPs does not undergo significant distortions and deformations.¹¹⁷ Furthermore, the deposition process can be described by the ballistic-aggregation or deposition-diffusion-aggregation (DDA) models. While the ballistic-aggregation model is, in general, applicable for larger NPs, the DDA model also accounts for the possible diffusion of NPs deposited in the initial phase of NP film formation. This assumes the mobility of deposited NPs and, thus, is appropriate only for the small NPs composed of up to thousands of atoms. In addition, the mobility of NPs on a substrate is strongly dependent on the strength of substrate–NP interaction, temperature, or the presence of surface defects or steps.¹¹⁸ However, even in the case of relatively low landing velocities typical for the soft-landing deposition regime, the incoming NP might rebound from the surface.¹¹⁹ As shown by numerical simulation, the transition between the adhesion and reflection of NP occurs as the Weber number, i.e., the ratio of the kinetic energy to the adhesion energy, passes through unity.¹²⁰ Naturally, rebounding depends strongly on the ability of NP to undergo plastic deformation on impact, but also on the velocity of deposited NPs. For instance, our recent results indicated that plasma-polymer NPs, especially, have to be slowed down to suppress their rebounding. Similar to the case of in-flight coating discussed in the previous section, this can be achieved by increasing the pressure in the deposition chamber. This is why in the so-far performed depositions of plasma polymers, relatively high pressures in the deposition chamber have been used.^67,78 In other words, knowledge of the velocity of NPs upon their impact on a substrate as well as NP–substrate interaction is crucial for the optimization of the deposition process.

There have been attempts to assess or even exactly measure NPs velocity behind the exit orifice. Ganeva et al.¹¹³ found—for both positively and negatively charged cluster ions (size 5–10 nm)—that the most probable velocities measured after the clusters left the aggregation region through a small orifice of a diameter of 3 mm are in the range of 80–180 m/s. A QMS filter was used for these measurements. Another technique was applied by Kousal et al.¹¹² In a simple experimental arrangement using electrostatic deflection plates, experimental values of the velocity of Ag nanoparticles (positively and negatively charged, as well as neutral) for a particular setup were found to be from 30 m/s for 50 nm particles to >200 m/s for particles smaller than 7 nm. TEM micrographs and numerical modeling supported the results. Generally speaking, the size and density of nanoparticles and the pressure difference between the aggregation and deposition chambers are vital for determining NPs velocity. The sophisticated velocity filter recently designed by Solar et al.^53,114,115 allowed more precise measurements of NPs velocity. Opposite to low energetic clusters not damaging the substrates, there are also more energetic clusters (with kinetic energies of 2.5–40 eV), which can sputter the substrate surface making them useful for the depth profiling of polymers.¹²¹ 

Although the soft-landing regime is highly advantageous for the production of porous NP films, its possible drawback is the poor adhesion of deposited NPs on the substrate. The common strategy to overcome this is to fix deposited NPs on a substrate by an overcoat film or embed them into a simultaneously deposited matrix. However, these strategies are not applicable to applications such as (bio)sensing or for the production of antibacterial/antiviral coatings, in which NPs have to be in direct contact with the surrounding media. This issue was addressed by Popok and co-workers in the case of polymer substrates.^79,122–125 These authors proposed a technique that is based on heating the polymer coated with NPs to a temperature close to its glass temperature T_g that increases the polymer chain flexibility and facilitates NP indentation [Fig. 11(a)]. This strategy opens the way for the formation of thin polymer films with either partly or fully embedded NPs, with a well-controlled filling factor of NPs, which is given by the initial surface coverage of the polymer by the particles.¹²³ In addition, the combination of this technique with electron beam lithography may be used for the production of the designed patterns of NPs [Fig. 11(b)].¹²⁴ The possible limiting factor is the possible occurrence of heat-induced oxidation of metallic NPs that can compromise their functionality.¹²³ To overcome this limitation, an alternative approach was followed by Kumar et al.¹²⁶ These authors heated C:H:N:O plasma-polymer films in situ during the deposition of metallic NPs. As demonstrated, such produced Ag/C:H:N:O films withstand ultrasonic baths without decay in their functionality.

As previously mentioned, patterns can be obtained by introducing additional steps, such as electron beam lithography. However, in most cases, NPs prepared by gas-phase plasma-assisted techniques are randomly scattered across the surface of a substrate. While this lack of order might be advantageous for certain applications on one side, for example, for creating random particle-based networks facilitating memristive switching phenomena,¹²⁷ the randomness is often considered a serious drawback that hinders the rate of integration of GASs into the existing production sector. To overcome the challenge, multiple strategies have been developed. The first approach relies on using shadow masking, which allows for the fabrication of passive electrical components on a planar substrate.¹²⁸ The primary advantage of this method is that it enables the manipulation of electrically neutral NPs, while maintaining a decent lateral resolution (reported to be below 1 μm). As a step for the improvement of resolution, nanoxerography, a directional self-assembly mechanism driven by electrostatic forces has been introduced.¹²⁹ It relies on the transfer of localized nanoscale surface charges to the substrate where NPs of opposite charge are to be deposited. Electrostatic patterning can be performed by multiple methods, for example, using atomic force microscopy tip.¹³⁰ Unlike the previous methods, nanoxerography has been demonstrated to be capable of producing charge patterns over a large area with a resolution below 100 nm.¹³¹ Another very different strategy for micropatterning takes advantage of thermophoretic forces acting on NPs landing onto a heated substrate to create nanoscale electrode assemblies.¹³² Finally, when working with magnetic NPs, one can opt for the magnetic extraction/collection of sputtered NPs. For instance, Ekeroth et al.¹³³ reported on utilizing this approach for the deposition of Pt_xNi_1−x NP-nanonetwork structure with high hydrogen evolution reaction activity.

To date, several works have considered the morphogenesis of nanomaterials facilitated by capillary forces. The tendency of minimizing the total free energy via reducing the area and/or energy per unit area at the NP surface/interface is shown to be responsible for the mechanism of self-organization of gas-phase synthesized Mg NPs on room temperature substrates.¹³⁴ 

As mentioned in the previous sections, NPs are typically deposited on solid substrates. Unlike conventional solids, liquids have rarely been used as substrates in vacuum-based depositions. The pioneering work of Yatsuya was published in 1974¹³⁵ who deposited thermally evaporated Ag onto silicon oil. Later, magnetron sputtering onto liquids was also realized.¹³⁶ For example, Ye with co-workers first applied magnetron sputtering of Ag onto silicon oil drops to prepare patterned metal films.^137,138 Wagener and co-workers^139,140 discovered the formation of fine silver and iron NPs inside the volume of mineral oil after the magnetron sputtering, instead of the metal film on the surface. This finding opened a new chemical-free route of synthesis of nanofluids, i.e., colloidal suspensions of NPs. The main limitation of the method is related to the use of liquids with sufficiently low vapor pressure and, therefore, compatible with the vacuum. During the last 20 years, magnetron sputtering has been successfully applied for the fabrication of gold and silver NPs in ionic liquids;^141,142 gold, silver, copper, platinum, palladium, and titanium NPs in liquid polymers;^143–147 gold and silver NPs in castor oil;^148,149 and platinum NPs in glycerol.¹⁵⁰ Besides monometallic NPs, more complex multicomponent nanofluids can also be synthesized. Alloy NPs Au/Ag,¹⁵¹ Pt/Cu,¹⁵² and even Cr/Mn/Fe/Co/Ni¹⁵³ were loaded into liquids using the sputtering of customized multicomponent targets. Another approach for the fabrication of complex NPs consists of the utilization of several independent magnetrons. The alloys of tuneable chemical composition, Au/Cu¹⁵⁴ and Au/Ag,¹⁵⁵ can be obtained by the co-sputtering of individual metal targets onto liquids. Moreover, sequential deposition onto the same liquid can lead to the decoration of initially prepared NPs by other metals, as has been demonstrated in the case of Au NPs coated by Pt.¹⁵⁶ Finally, the addition of reactive gases to argon during the sputtering process makes the approach suitable for the synthesis of metal oxide¹⁵⁷ or nitride (see below) nanofluids.

Despite the multiple advantages of the approach based on magnetron sputtering, several fundamental drawbacks must be considered. The independent tuning of the NP size, shape, and crystallinity is challenging. Since nucleation and growth occur in the liquid, the properties of the liquid are crucial for the control over the NPs characteristics. As already shown, chemical composition¹⁵⁸ or temperature^143,159 can influence the size of NPs. For example, an increase in the temperature of liquid polyethylene glycol (PEG) from 10 up to 110 °C during the sputter deposition of Au leads to the increase of NPs diameter by a factor of 4. However, this regulation is not accurate and convenient because magnetron plasma itself can heat the surface layer of the liquid as demonstrated in a recent study by Patel with his co-workers.¹⁵⁹ Moreover, plasma interacting with the surface of the liquid may induce its chemical transformations or polymerization.

An innovative approach to the chemical-free synthesis of nanofluids was presented in 2021 by Choukourov with his co-workers, who used the sputter-based GAS to produce NPs and deposit them in a vacuum-compatible liquid.¹⁶⁰ Here, the main advantage is the production of NPs in the gas phase, i.e., independently of the substrate. The authors deposited Cu NPs into liquid PEG. After the initial sedimentation of agglomerates, nanofluids became transparent and acquired green color, which was due to the localized surface plasmon resonance (LSPR). Moreover, Cu nanofluids demonstrated photoluminescence that could be beneficial in future applications.

To date, other colorized plasmonic nanofluids have been produced covering the whole spectrum in the visible range. Green Cu, yellow Ag, purple Au, and blue ZrN nanofluids have been synthesized using PEG as a substrate [Fig. 12(a)]. It is worth noting that ZrN NPs were synthesized using reactive magnetron sputtering. All the nanofluids exhibit plasmonic properties as demonstrated by UV–Vis spectra in Fig. 12(b).

In general, colloidal stability remains a fundamental question in the field of nanofluids. For GAS-derived nanofluids, partial agglomeration and sedimentation of NPs were observed by the naked eye during their storage. In more detail, colloidal stability was studied on Au/PEG nanofluids using in-liquid SAXS.¹⁶¹ Au NPs demonstrate a bi-modal size distribution immediately after the preparation [Figs. 13(a) and 13(b)]. The mean size of both populations is found to remain constant during storage in air; however, the relative volume ratio occupied by NPs substantially decreases with time [Figs. 13(c) and 13(d)]. Such behavior was attributed to the partial agglomeration of NPs into larger aggregates (undetectable by SAXS) followed by their sedimentation.

Not only metal nanofluids but also more sophisticated transition metal nitride nanofluids can be obtained using this method, as can be seen in the example of ZrN/PEG nanofluid shown in Fig. 12(a). Accurate control over the NP size, crystallinity, chemical composition, and concentration is possible by tuning the experimental conditions. Moreover, the use of two independent GASes for the simultaneous deposition of different types of NPs allows for obtaining bi-metal nanofluids as was recently proved by Biliak et al.¹⁶² Thus, the use of GAS demonstrates high potential for the synthesis of colloidal media for innovative applications including plasmonics or heat transfer.

Vacuum-based methods of NPs preparation are not replaceable for certain applications. Especially, plasma-assisted gas aggregation cluster sources equipped with a magnetron are now overwhelmingly used as shown above. However, there are fundamental understanding issues related to NPs birth and sustainability. Reactive magnetron sputtering-based processes applied in GAS will also need further attention. In addition, a stable process that will provide a very high yield of NPs production with a good potential for scale-up is still being looked for. This will be the main challenge for greater industrial applications in spite of the fact that some promising approaches exist, e.g., postcylindrical magnetron in various geometrical configurations.

The authors D.N, P.P., and O.K. are indebted to the Czech Science Foundation, Grant No. 21-12828S for partial support. The authors are also grateful to A. Choukourov and L. Martinu for reading the manuscript and for their helpful comments.

The authors have no conflicts to disclose.

O. Kylián: Conceptualization (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (equal). D. Nikitin: Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). J. Hanuš: Visualization (equal); Writing – original draft (equal). S. Ali-Ogly: Visualization (equal); Writing – original draft (equal). P. Pleskunov: Visualization (equal); Writing – original draft (equal). H. Biederman: Conceptualization (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (equal).

The data that support the findings of this study are available within the article as well as from the corresponding author upon reasonable request.