Gas-Phase Synthesis of Nanoparticles: present status and perspectives

Abstract

There is an increasing interest in the generation of well-defined nanoparticles (NPs) not only because of their size-related particular properties, but also because they are promising building blocks for more complex materials in nanotechnology.

Here, we will shortly introduce the gas phase synthesis technology that has evolved rapidly in the last years and allows the fabrication of complex NPs with controllable and tuneable chemical composition and structure while keeping very good control over the size distribution. We will also address some limitations of the technology (stability over time, production yield…) and discuss possible solutions.

Introduction

The production of nanoclusters by gas phase synthesis has been developed and widely used since the 80’s and 90’s by groups interested in studying their properties and their interaction with surfaces [1]. With the advent of the nanotechnology the gas phase synthesis technology has evolved to the fabrication of well-controlled nanoparticles. All variants of the technology are based on the atomization of a material, followed by the controlled coalescence of the atoms into nanoparticles that are collected [2]. The different variants differ mainly in the way in which the material is atomized and they have rapidly evolved in the last decade giving rise to new experimental apparatus that can produce a wide variety of nanoparticles [3]. The most popular nanoparticle source is probably the one based on magnetron sputtering because it is relatively easy to use and it produces the largest proportion of charged nanoparticles [4] which allows their mass selection and deflection. This probably explains why this type of nanoparticle source became commercially available in 2001.

In this letter, we will address three different issues that improve the versatility, stability and production rates of sputter gas aggregation sources (SGAS). First, we will show how the single magnetron based nanoparticles source has been adapted to the Multiple Ion Cluster Source (MICS) to produce a wider variety of nanoparticles. In a second step, we will face the question of SGAS stability; although magnetron based cluster sources are stable (in terms of nanoparticle size and deposition rates) over short periods of time (tens of minutes), instabilities for longer production times must be addressed for the further scaling-up and mass production. In that sense the limitations induced by the well-known race track formation on the magnetron target is discussed and a possible solution to overcome this issue is presented. Finally, we address the not well-understood issue of the quality of the residual vacuum in the nanoparticle sources. In particular, we demonstrate its importance for the controlled formation of nanoparticles in terms of size and synthesis rate for prolonged operation times.

1). The Multiple magnetron approach in gas phase synthesis of nanoparticles

In the seminal works of H. Haberland et al. [4,5] a magnetron based nanoparticle source has been proposed. In the following years the single magnetron SGAS was extensively used to produce a wide variety of nanoparticles and nanoparticles based systems. One of the major advantages of such NPs source resides in the fact that the stoichiometry of the fabricated NPs is identical or nearly identical to the stoichiometry of the sputtering target placed into the magnetron. However, this advantage became also a major drawback as an increasing number of studies requested a continuous scanning of the chemical composition of the nanoparticles and this could not be achieved without a physical change of the magnetron target for each chemical composition; which in turns, implies the aperture of the vacuum vessels that is time and economical consuming. In order to overcome such limitation, few multi-magnetron approaches have been proposed in order to combine the different elements sputtered by more than one magnetron. Two of the proposed designs reside in the combination of a magnetron-based NPs source coupled with one or more magnetrons that in a second stage allow the formation of a shell on the NPs generated by the NPs source [6,7]. Another design incorporates 3 sputtering targets in the same magnetron head that is inserted in an aggregation zone [8,9]. In 2010 our group [Low dimensional Advanced Materials group at the ICMM-CSIC] proposed a new design, licensed to Oxford Applied Research Ltd., were the magnetrons are completely independent therefore giving more flexibility for the fabrication of well-controlled alloyed and core@shell, core@shell@shell nanoparticles. The new design is based in the replacement of standard 2” diameter magnetron by 3 smaller magnetrons mounted in the same flange and, thus, compatible with the already existent aggregation sources. The magnetrons are completely independent regarding their power supply and argon feed but most importantly that can be placed at different positions in the aggregation zone in an independent manner. Such design has been called Multiple Ion Cluster Source (MICS) as it has multiple magnetrons and it has been proven to be very powerful and flexible for the formation of alloyed nanoparticles with well-controlled chemical composition and size [10] and also for the engineering of core@shell and core@shell@shell nanoparticles [11,12,13]. Figure 1 illustrates the different configurations that are used for the synthesis of the variety of nanoparticles. In figure 1a) we illustrate the formation of single element NPs that is achieved by switching on only one of the 3 magnetrons as it is routinely done in standard NPs sources. In figure 1b) the fabrication of alloyed NPs is illustrated were 2 of the magnetrons are placed in the same aggregation length in the aggregation zone (i.e. at the same vertical plane). In this configuration the plasmas of both magnetrons intermix giving rise to the formation of alloyed nanoparticles [10]. Figure 1 b) illustrates the formation of binary nanoparticles but the configuration can be extended to a ternary alloy by simply placing the third magnetron in the same plane as the other two. Thanks to the intermixing of the plasmas generated by each magnetron, this configuration allows the synthesis of homogeneous alloyed nanoparticles. The size of the NPs can be controlled through the positioning of the magnetrons inside the aggregation zone and/or by injecting helium like in other gas phase NPs sources [14]. On the other hand the chemical composition of the nanoparticles is controlled by the densities of atoms and ions generated by each magnetron that are monitored by the applied power and/or the argon flux injected in each magnetron [15,16]. Figure 1 c) displays the configuration that is used for the generation of core@shell (CS) nanoparticles. The formation of CS NPs is based on the fact that NPs formed from a first magnetron (red NPs in Figure 1c) are covered by a second material (yellow in Figure 1c) as they pass through the plasma generated by a second magnetron. In such configuration the synthesis of NPs does not rely anymore on chemical reactions (oxidation and reduction potentials for example) but only on the sticking coefficient of the atoms generated by the second magnetron on the nanoparticles generated by the first magnetron. This approach not only simplifies the synthesis process but also allows the fabrication of nanoparticles not accessible by chemical methods. In the supplementary file S1 we present an explicative movie of the different fabrication procedures.

Figure 1.

schematics of the operation of the MICS for the production of a) single element NPs, b) alloyed NPs and c) core@shell type NPs.

Therefore this design has demonstrated to be very flexible for the synthesis of a wider variety of nanoparticles. In collaboration with Oxford Applied Research Ltd., our group has developed a scaled-up variant of the MICS that is composed of 3 magnetrons of 2” diameter each in a bigger aggregation zone, for the generation of more intense NPs fluxes for fundamental studies in astrophysics [17].

Although such design of the MICS extends the variety of nanoparticles that can be synthesized by SGAS, there are some technical problems related to the evolution of the magnetron and instabilities that still need to be addressed for the fabrication of NPs over long (hours) periods of time. In the next section we discuss the race track formation, its influence on the formation of nanoparticles and a possible solution based on the Full Face Erosion magnetron.

2). Race track influence and Full Face Erosion magnetron design

Among the possible origins of instabilities of SGAS, the race track formation is the most commonly identified and can be considered as the major source of instability. Hence, it is important to address this issue to access prolonged operation and reliable production of nanoparticles.

The race track is well-known to form on the target surface [18] and its formation has been extensively reported in the literature even in SGAS [19,20]. In their works, Hippler et al. have analyzed and discussed the effect of the race track formation on the generation of nanoparticles. In particular, they have clearly shown that the formation of the race track has an influence on the nanoparticle size and synthesis rate [19,20]. Note that the synthesis rate is referred to the number of nanoparticles that are formed into the aggregation zone (by unit time) and the deposition rate to the number of nanoparticles that are collected (by unit time and surface) outside the aggregation zone (typically on a flat surface). While the deposition rate can be precisely measured by Atomic Force Microscopy for example, the synthesis rate is hardly measurable inside the aggregation zone. In that sense, the synthesis rate is proportional to the deposition rate that in turn depends on the sticking coefficient of the NPs on the chosen surface (Silicon wafer, TEM grid, etc). In Figure 2 we illustrate the effect of the race track formation on the synthesis rate of gold nanoparticles. Both curves represent the evolution of the ion current (that is proportional to the number of fabricated nanoparticles) measured with a quadrupole mass filter in the case of a new and flat gold target (upper curve) and for a used target were the race track is well formed (bottom curve). In both cases, all operation parameters were kept identical (applied power, gas fluxes, target thickness, aggregation length, etc). As can be clearly observed, the flat target allows the production of a more intense NPs beam that is nearly one order of magnitude more intense than for the used target. Note that in terms of mass loss, the used target has approximately 95% of the mass of the new target which means that the useful proportion of the target is no more than 5%. Although the target can be used on both sides (which allows its 10% mass use), it clearly appears that the usage of a sputtering target in a SGAS is very low. And most importantly, the evolution of the target morphology that affects the sputtering yield has also an effect on the size of the nanoparticles that drastically limits the stable production of well-controlled nanoparticles to short periods of time (tens of minutes typically).

Figure 2.

comparison of ion current (that is proportional to NPs fabrication rate) as a function of NPs diameter, measured from a new and flat magnetron target (upper curve) and from a used (with race track) target.

In order to overcome the instabilities generated by the race track formation, we have developed in collaboration with Nano4Enegy SLNE a Full Face Erosion (FFE) magnetron adapted to our standard 2” SGAS. The FFE magnetron is based on the movement of the magnets in order to sweep the whole surface of the target. Although this concept is not new in standard sputtering processes [21,22], the challenge we faced was to reduce its size to fit it in a standard NPs source and to test its validity for the generation of NPs. The new FFE magnetron developed in collaboration with Nano4Enegy SLNE was fitted in a standard aggregation zone and mounted with a gold 2” target for testing. Figure 3 compares the surface morphology of two targets, one used in a standard magnetron (left side) and the other in the developed FFE magnetron (right side). The target used in the FFE magnetron has been operated during more than 12 hours at an average power of 90 W and its weight consumption is approximately 20%. The target used in a standard magnetron has a weight consumption of 5% approximately and has been removed from the NPs source when the production rate of NPs reached nearly zero NPs due to the formation of the race track.

Figure 3.

surface images of targets after a) 5% use in standard magnetron and b) more than 20% use in a Full Face Erosion magnetron; c) ⋅d microscope image of target used in standard magnetron; d) 3D image of target used in a Full Face Erosion magnetron; e) depth profiles comparison between targets used in standard magnetron and FFE magnetron. Profiles are performed on lines indicated in figures 3 c) and 3 d).

As can be observed in Figures 3 a) and 3 b) the erosion of both targets is very different. The FFE target displays a much more pronounced erosion close to the centre of the target. Note that the erosion of this target did not induce variations neither in the synthesis rate nor in the NPs size during its whole operation period. 3D microscope images (Figure 3c and 3d) also evidence the very different erosion process but the differences in surface morphologies are better appreciated in Figures 3e) where we display a depth profile of the targets extracted from the microscope images. In the supplementary file S2 we present the animated images of both targets obtained by the numerically reconstruction of the recorded 3D images. Although the design of the FFE magnetron might be improved even more, we have demonstrated for the first time its possible use in a SGAS with the subsequent advantage of stability of NPs production over time and extended lifetime of the target.

Thanks to the design of the FFE magnetron, first order instabilities of the SGAS linked to the racetrack formation have been addressed and solved, which made possible to address second order instabilities like those associated with traces of gases.

3). Effect of trace gases in the synthesis of nanoparticles. Test-case of water vapour

Although the effect of gas traces for the generation of NPs has been reported in several studies, little is understood and both experimental and theoretical studies are needed to fully understand the observed phenomena. For example it has been demonstrated that controlled doses of oxygen injected into the ICS could increase significantly the synthesis rate of Cu, Ti, Co and W nanoparticles [23,24,25]. More recently, it has been also shown that the introduction of H₂ or CH₄ in small proportions could favour the formation of alloy or core@shell structure of Mg and Ti [26]. Most of the studies are focused on the deposition rate behaviour as gas traces were introduced into the SGAS, and Prof. Haberland stated that “…diatomic molecules are very helpful to start the clustering process. Moreover it is observed experimentally that a tiny flow of N₂, O₂, CO₂, and so on, introduced into the aggregation tube, will lead to much larger clusters..…If plastic or Teflon tubing is used in the gas inlet system, this should be regarded with extreme suspicion.” [3]. Hence, it clearly appears that traces of gases can have important effects on the clustering process that induces changes in cluster size and synthesis rates. As an example we present in Figure 4 two deposits performed using the Full Face Erosion magnetron with the same deposition parameters except for the base pressure. Figure 4a) is an Atomic Force Microscopy (AFM) image corresponding to the deposit performed starting from a base pressure less 8.10^-9 mbar, while Figure 4b) corresponds to the deposit performed with a base pressure of 2.10^-6 mbar obtained by injecting in a controlled way (using a mass flow controller) a tiny flow of air (0.1 sccm). Although the proportion of air that injected into the SGAS represents only 0.12% of the total injected gas (the rest of the gas is 80 sccm of argon), it drastically affects the synthesis rate as the number of nanoparticles in Figure 4b) is much higher than in Figure 4a) (by approximately a factor of 2.7). While Figure 4a) corresponds to a deposit less than a monolayer of NPs (density was found to be 380 NPs/μm²), deposit of Figure 4b) has a higher density than a monolayer of NPs that makes the determination of the NPs height (some NPs are lying on top of each other) and density (found to be approximately 1032 NPs/μm²) difficult. Nevertheless, both AFM images and corresponding profile lines displayed in Figure 4c) clearly illustrate the effect of gas traces on the synthesis rate for the simple case of Au NPs in presence of air impurity.

Figure 4.

Atomic Force Microscopy (AFM) images. a) AFM image of an Au nanoparticles deposit on Si wafer with starting base pressure of 8. 10^-9 mbar; b) similar AFM image of a deposit of Au nanoparticles with starting base pressure of 2.10^-6 mbar due to controlled air leak of 0.1 sccm; c) line profiles following the lines displayed in Figure 4 a) and 4 b). Both deposits have been performed with the following parameters: argon flow: 80 sccm, applied magnetron power: 90 W; deposition time: 140 seconds.

The air is composed of N₂ (78.08%), O₂ (20.95%), Ar (0.93%), water vapour and gases in very small concentrations (Ne, Kr, Xe, CO₂, CH₄, H₂). In first approximation we can therefore consider only the N₂, O₂, Ar and water vapour. Since Ar is injected in large amounts in order to generate the plasma, it can also be discarded. While preliminary studies performed injecting controlled amounts of N₂ and O₂ have not shown evident influence of these gases on the synthesis of gold NPs (results to be published elsewhere), the addition of tiny amounts of water vapour resulted in drastic modifications of both, synthesis rate and NPs size as can be observed in figure 5. Experiments summarised in Figure 5 were performed using the FFE magnetron in the GAS, i.e. avoiding therefore any influence of the race track formation and evolution upon time. The applied power was 90 W, the argon flux was 80 sccm and no helium was added in the gas mixture. As can be observed in Figure 5, at the initial stages of water vapour injection (5.10^-3 sccm), the NPs average size (extracted from the log-normal fitting of the height distribution measured by AFM) is drastically reduced from approximately 9 nm (that corresponds to 0 sccm of water vapour) to a value close to 4 nm (red curve in figure 5). Note that the injected water vapour represents only 0.6% of the injected gases. Despite such low amount of water vapour, the NPs mean size is reduced by a factor close to 2. While higher amounts of water vapour do not modify substantially the NPs mean size, the effect on the deposition rate is very important as it is increased by a factor of approximately 15 (black curve in Figure 5). These results clearly show that the presence of small amounts of water molecules in the SGAS have important effects in the synthesis of gold nanoparticles and it is inferred that such effects could also be crucial in the synthesis of NPs of other metals. Inversely, the progressive diminution of water molecules concentration in the residual vacuum that will occur with time as the vacuum pumps continuously remove the water molecules from the vacuum vessel could explain the experimentally observed decrease of NPs deposition rate upon time. Also, it is expected that upon running time, the aggregation zone will be depleted from water molecules which will induce a decrease in the synthesis rate of NPs, as the group of Prof. Faupel reported [24]. Hence, it clearly appears that ultra-high vacuum conditions are not sufficient to avoid evolutions of synthesis rates and NPs size in SGAS, but that a precise monitoring and adjustment of the gas traces in the aggregation zone are necessary for a fine control of the NPs synthesis.

Figure 5.

evolution of the deposition rate and mean height of Au NPs as a function of the injected flow of water vapour.

More experimental studies in combination with theoretical simulation are needed to unravel the mechanisms that govern the tuning of the size and synthesis rate of the gold nanoparticles as a function of water concentration and other gases in the gas mixture. In particular seed effect and/or heat dissipation mechanism could be responsible and need to be identified to allow a better understanding of the effect of gas traces addition on the synthesis of nanoparticles.

Conclusions

We have presented recent developments in the synthesis of nanoparticles by the gas phase approach using the magnetron sputtering. The Multiple Ion Cluster Source that is based in the replacement of the standard single magnetron by three independent magnetrons has been developed in order to widen the spectrum of possible nanoparticles that can be fabricated. In that way, single element, alloyed and core@shell nanoparticles can be synthesised in a single step while keeping control on the stoichiometry, diameter, core diameter and shell thickness. Thanks to this design, nanoparticles not accessible by chemical routes can be fabricated as chemical reactions are not involved in the synthesis.

The Full Face Erosion magnetron adapted to the SGAS has been also presented. Such design eliminates the known fluctuations in terms of nanoparticle size and synthesis rate as the erosion of the target is more uniform and the racetrack formation is avoided. Hence the synthesis of nanoparticles is much more stable and the lifetime of the target is prolonged. Both facts open the use of SGAS in industrial applications.

Finally we address the other origin of instability of the SGAS that is related to the residual gases present in the vacuum vessel. We show that tiny traces of water vapour can drastically modify the fabrication of gold nanoparticles in terms of nanoparticles size and synthesis rate. More studies are needed in order to understand the mechanisms that are involved but our results clearly show that a stable fabrication of nanoparticles is related to the control of the residual gases in the vacuum vessel.

Supplementary Material

S1

video displaying the different configurations of the Multiple Ion Cluster Source that allow the synthesis of pure element NPs, alloyed NPs, core@shell NPs and inverse core@shell NPs in a single step.

S2

video displaying in different positions the erosion shape of a target used in a standard magnetron and a target after use in a Full Face Erosion magnetron adapted to the Ion Cluster Source.

S3

figure displaying the arrangement of the magnets inside the Full Face Erosion magnetron (left) and the shape of the plasma as a function of the positions of the magnets (right).