Journal of Nanoparticle Research

, Volume 12, Issue 1, pp 247–259

Generation of mixed metallic nanoparticles from immiscible metals by spark discharge

Authors

  • N. S. Tabrizi
    • Nanostructured Materials, Faculty of Applied SciencesDelft University of Technology
  • Q. Xu
    • Laboratory for Material Science, National Centre for HREMDelft University of Technology
  • N. M. van der Pers
    • Department of Materials Science and Engineering, Faculty of 3mEDelft University of Technology
    • Nanostructured Materials, Faculty of Applied SciencesDelft University of Technology
Research Paper

DOI: 10.1007/s11051-009-9603-4

Cite this article as:
Tabrizi, N.S., Xu, Q., van der Pers, N.M. et al. J Nanopart Res (2010) 12: 247. doi:10.1007/s11051-009-9603-4

Abstract

Using a spark discharge system, we synthesized Ag-Cu, Pt–Au and Cu-W mixed particles a few nanometers in size. These combinations have miscibility gaps in the bulk form. The microsecond sparks between electrodes consisting of the respective materials, form a vapour cloud. Very fast quenching of the mixed vapour results in the formation of nanoparticles. To investigate the morphology, size, composition and structure of the particles, TEM, XRD analyses and EDS elemental mapping were performed on the samples. The average compositions were measured by ICP and the specific surface areas were determined by the BET. Our method produces Ag-Cu and Au–Pt mixed crystalline phases that do not exist in macroscopic samples. For Cu-W, alloying is not observed, and the metals are mixed on a scale of about 1 nm.

Keywords

NanoparticlesImmiscible metalsSpark dischargeSynthesis method

Introduction

Metallic nanoparticles are interesting among others because their electronic structure may undergo a major deviation from that of the bulk and they possess the ability to store excess electrons (Lahiri et al. 2005). Bimetallic nanoparticles are of even greater interest, since they generally show different physicochemical properties as compared to their individual constituents (Wu and Lai 2004). Flexible mixing of materials on a nano scale, down to the atomic scale, would give access to an enormous variety of new material properties tunable not only through size but also through mixing ratios and the scale of mixing. This should lead to myriad applications in areas such as optoelectronics, catalysis, batteries, solar cells, fuel cells, hydrogen storage, magnetic materials and sensors (Wu and Lai 2004). In principle, nanomixing offers additional degrees of freedom for tailoring properties to match the application, and pure materials or stoichiometric mixing ratios are likely to become exceptions in functional materials of the future, if suitable methods of mixing are found.

In the present study, we produced Ag-Cu, Au–Pt and Cu-W nanoparticles. Ag-Cu has many interesting applications (Cagran et al. 2006). For instance, the high electrical conductivity of Ag and low electrical migration of Cu make Ag-Cu nanoparticles suitable as conductive fillers in electrically conductive adhesives (Jiang et al. 2005). Au–Pt nanoparticles can be used in various catalytic reactions (Zeng et al. 2006; Devarajan et al. 2005) and exhibit specific catalytic activity and selectivity in hydrogenation (Patel et al. 2005). The high thermal and electrical conductivity of Cu and the low thermal expansion coefficient of W make Cu/W composites attractive as heat sinks in electronic packages (Kang and Bong Kang 2003).

There have been many attempts to generate mixed nanoparticles through different methods such as vapour quenching, co-deposition sputtering, mechanical alloying and ion-beam mixing or irradiation (Almtoft Pagh et al. 2007; Radic and Stubicar 1998). In the present study, we used spark discharge ablation of electrodes in an inert gas, which was introduced by Schwyn et al. in 1988. This method falls under the category of vapour quenching. Short sparks locally evaporate material, which leads to small vapour clouds that are very rapidly cooled. In comparison to other methods used, it is extremely simple, cost-effective and flexible. Moreover, it has the potential of being scaled up. The size and concentration of particles can be controlled via the energy and repetition rate of the spark. While a recent study has focused on the potential of the method in mixing metals that easily form alloys in macroscopic systems (Tabrizi et al. 2008), the objective of the present study is to investigate the feasibility of generating internally mixed nanoparticles from constituents that are immiscible in the macroscopic case.

Since a large fraction of atoms composing a nanoparticle are situated at the surface and contribute to the excess Gibbs free energy, the phase diagram of the particulate binary system may be modified with respect to the macroscopic systems (Lahiri et al. 2005). In addition, reduction in the melting point of the nanoparticles (Buffat and Borel 1976; Ding et al. 2004) and the presence of defects at the interface in bimetallic nanoparticles enhance inter-diffusion of the metals and modify the alloying characteristics (Lahiri et al. 2005; Birringer 1989). Using Monte Carlo simulations Christensen et al. (1995) investigated the size dependence of phase separation in small bimetallic solid clusters. For the Ag-Cu system, which shows immiscibility in the bulk for a broad range of compositions up to the melting temperature, they found that the maximum temperature where phase separation can occur is strongly size dependent. They also found the absence of phase separation for clusters smaller than a critical size of about 270 atoms. By taking into consideration that electric charge can energetically affect the free energy of the formation of clusters, Ouyang et al. (2006a, b) proposed a charge-dependent thermodynamic model to address the phase transformation between miscible and immiscible for nano-sized alloying particles. By applying a thermodynamic model and an analytic embedded atom method, Xiao et al. (2006) recently showed that the heat of formation of alloy nanoparticles is not only composition but also size dependent and for bulk immiscible systems with a positive heat of formation, a negative heat of formation may be found for alloy nanoparticles of small size and in particular for a dilute solute component.

Experimental method

Our spark discharge system consists of a chamber in which two opposing cylindrical electrodes are mounted at an adjustable distance. The electrodes are connected to a high voltage source and parallel to a variable capacitor, which is periodically charged by a constant current to the electric breakdown voltage of the gap between the electrodes. The spark energy is adjustable by the capacitance but kept constant at 20 nF for the present study. The gap distance is kept at 1 mm (see Fig. 1). A very high current from the rapid discharge causes evaporation of the electrode material. A high temperature is reached in the spark for a short time, followed by a rapid cooling caused by radiation, expansion and mixing with the carrier gas, which results in the formation of nanosized primary particles. The nonequilibrium conditions may lead to the enhanced solid mixing and modified defect density with respect to equilibrium.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig1_HTML.gif
Fig. 1

Schematic of the experimental set-up

For production of the Ag-Cu and Cu-W particles, two pairs of sintered electrodes of (Ag72/Cu28) and (W72/Cu28) were utilized while for the Au–Pt system, a pure Au electrode was combined with a pure electrode of Pt (99+% purity). Moreover, to produce monometallic particles a pair of identical electrodes of the respective material was used. The inert gas carrying the particles was Ar (99.999% purity).

High resolution electron microscopy was performed on a Philips CM30-UT-FEG and a FEI Tecnai-200FEG. EDS data were acquired on the FEI Tecnai-200FEG using the STEM mode with a spot size of ~0.3 nm, which guaranteed that the measured spectra only came from the local area. The spectra acquisition, drift correction and data analysis were all processed using the software TIA (Tecnai Imaging & Analysis). The TEM grid was a Ni 200 mesh coated with a carbon film.

The average compositions of the samples were measured by ICP-OES on a PerkinElmer Optima 5300. The X-ray diffraction (XRD) measurements were performed on a Bruker–AXS D5005 diffractometer, equipped with a Huber CuKalpha-1 Ge monochromator in the incident beam and a Braun Position Sensitive Detector PSD-50 M in the diffracted beam. N2 adsorption isotherms were measured on a Quantachrome Autosorb-6B for the determination of the specific surface areas of the particles. A home-built scanning mobility particle sizer (SMPS) measured the particle size distribution on-line.

Results and discussion

Ag-Cu

Electron micrographs of Ag-Cu particles at two magnifications can be seen in Fig. 2. The particles are found to be crystalline, surrounded by an amorphous layer probably of oxide phase(s) (see the close-up). Neck formation is not observed and most of the particles are a few nanometers in size, which points to the absence of sintering and coalescence, contrary to the pure silver particles which showed significant sintering (Tabrizi et al. 2008).
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig2_HTML.jpg
Fig. 2

Electron micrographs of the Ag-Cu particles

Very few particles that are much larger (up to 200 nm) are present. The large size difference with respect to the small ones implies a different formation mechanism (Gray and Pharney 1974). We assume that these particles are sputtered from local patches of molten metal at the electrode surface, as we have previously reported for the case of pure metals (Tabrizi et al. 2009). EDS analysis reveals that the large particles are almost pure Ag. This points to phase separation in microscopic molten patches on the electrode surface. The average composition of the nanoparticles measured at different regions of the sample is about 30 wt% Cu and 70 wt% Ag. This agrees with the composition of the electrodes, which is 28 wt% Cu and 72 wt% Ag.

The average composition of the sample analysed by ICP confirms this agreement (see Table 1). This indicates that there is no preferential evaporation of the more volatile constituent in spite of a significant difference in the boiling points (2,435 K for Ag and 3,200 K for Cu). Moreover, it shows that partial melting of the electrode surface has not distorted the composition at the electrode surface so severely that the lower surface tension material becomes dominant at the surface. The surface free energies of Ag and Cu at their melting points are 1.046 and 1.576 J m−2, respectively (Mezey and Giber 1982).
Table 1

Compositional analysis of the Ag-Cu sample measured by ICP

Material

Cu/Ag wt ratio in the electrode

Cu/Ag wt ratio in the sample

Ag-Cu electrodes

0.388

0.380

An XRD pattern of the sample is shown in Fig. 3. Silver and copper both have face-centred cubic (fcc) structures. In addition to the reflections of Ag, a set of strong fcc Bragg peaks is present, which fits a lattice parameter somewhat smaller than Ag (by a factor of 0.9792). This can be attributed to an AgCu alloy with Cu in solid solution with Ag. Moreover, there is another set of Bragg reflections corresponding to a lattice parameter slightly larger than pure Cu (by a factor of 1.0187). This can be ascribed to an AgCu alloy with Ag in solid solution with Cu. For Ag and Cu, the atomic radii are 0.1445 and 0.1278 nm, respectively. Thus, Cu atoms are smaller than Ag atoms and dissolution of Cu in the Ag matrix leads to contraction of the lattice and similarly dissolution of Ag in Cu matrix leads to expansion of the lattice (Ceylan et al. 2006). Applying the data for the lattice parameters (Predel and Madelung 1998), the AgCu alloys in the sample are estimated to contain about 21.5 at.% Cu in Ag (a hypoeutectic alloy) and about 12 at.% Ag in Cu (a hypereutectic alloy). According to the phase diagram of the macroscopic Ag-Cu system, the maximum solubility of Cu in Ag is around 14.1 at.% and the maximum solubility of Ag in Cu is about 4.9 at.% at the eutectic temperature of 1,052 K. These results show that the alloying behaviour is significantly modified in our nano-scale system. This is in agreement with theoretical study carried out by Hajra and Acharya (2004), who found a remarkable decrease in the melting points of the metals, and the solid–solid and solid–liquid transition and the eutectic temperature with a consequent increase in the solubilities of the nano terminal phases of the Ag-Cu system.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig3_HTML.gif
Fig. 3

X-ray diffraction pattern of the Ag-Cu sample

For the above mentioned compositions of the Ag-Cu alloys, the heat of formation at 700 K is positive around 4.1 and 2.8 kJ/mol, respectively (Najafabadi and Srolovitz 1993). The sharp silver reflections are explained by the big particles mentioned above, one of which is shown in the TEM micrograph of Fig. 2. The reflection of Ag and Ag-Cu clearly show the difference in broadness, indicating that the former peak is mainly due to large Ag particles and the latter is due to nanoparticles.

We do not see characteristic peaks corresponding to crystalline oxide phases. This is probably due to the fact that the oxide phase, covering the particle surfaces as observed in the electron micrographs is amorphous. The undefined reflections are attributed to unknown impurities.

The specific surface areas of Cu, Ag and Ag-Cu samples measured by the BET method (Rouquerol et al. 1999) and the corresponding BET diameters calculated for spherical shape (shape factor 6) are listed in Table 2. Copper nanoparticles present a rather large surface area consistent with the particle size that we have previously observed in TEM micrographs of pure copper. This may be due to the role that oxygen plays in stopping coalescence. Silver presents a small specific surface area. The discrepancy between the pure silver primary particle size and the large calculated BET diameter can be attributed to the cold sintering and coalescence of silver particles on the substrate, as has previously been observed for pure gold nanoparticles (Tabrizi et al. 2009). The Ag-Cu particles show a similar BET diameter as the Cu particles, indicating that the presence of Cu in the particles predominantly consisting of Ag hinders sintering.
Table 2

Specific surface area of the Ag-Cu particles

Material

BET specific surface area (m2 g−1)

Calculated BET diameter (nm)

Ag

<10

>57

Cu

130

5.1

Ag-Cu

97

6.3

Size distributions of Cu, Ag and Ag-Cu particles were measured on-line by SMPS. It can be seen in Fig. 4 that the produced aerosols contain very small particles of 4–5 nm modal diameters. For Cu and Cu-Ag, this is in fair agreement with the BET diameters, while the discrepancy for Ag confirms cold sintering on the substrate.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig4_HTML.gif
Fig. 4

Size distributions of Ag, Cu and Ag-Cu nanoparticles (C = 5 nF, d = 1 mm, f = 10 Hz, Q = 5 lpm Ar)

Au–Pt

TEM images of pure Au and pure Pt presented in Fig. 5a and b show different morphologies of the particle assemblies. Gold particles have significantly sintered on the contact points and platinum particles seem to remain more separated, although some particles are interconnected. Au–Pt particles produced from one electrode of Au and the other one of Pt are depicted in Fig. 5c. The unattached primary particles are rather spherical and a few nanometers in diameter. They represent the primary particle size typically produced by our spark process. In some regions, the particles are sintered and joined by necks. Neck growth occurs by surface diffusion, dislocation motion and grain rotation of clean particles (Koch 2007). The tendency of sintering is highest in noble metallic nanoparticles because of their clean surfaces, since any oxide layer represents a barrier for sintering.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig5_HTML.jpg
Fig. 5

TEM images of Au (a), Pt (b) and Au–Pt (c) nanoparticles

In order to study the composition of the particles, a number of single particles in a narrow size range (5–12 nm) were analyzed by EDS. Table 3 lists the elemental compositions of nine particles with an average Au/Pt weight ratio of 0.97. The range of Au compositions in the measured particles is 22–65 wt%. According to the bulk phase diagram of the Au–Pt system, this compositional range falls in the two-phase region at an ambient temperature.
Table 3

EDS compositional analysis of single particles [Au(+)Pt(−)]

Au/Pt wt ratio

AVE

STDEV

1.86

1.59

1.10

0.77

0.54

0.29

0.64

1.09

0.87

0.97

0.50

To investigate the compositional homogeneity within a particle, elemental maps were recorded for a number of single particles using EDS in the STEM mode. A typical result in the form of a 7 × 7 grid on the particle 5.5 nm in diameter highlighted by the solid frame in Fig. 6a can be seen in Fig. 6b and c. The drift was checked and corrected for by using the agglomerate in the dash lined frame in Fig. 6a as a reference. The compositional mappings based on the Au L edge and the Pt L edge are shown in Fig. 6b and c, respectively. The patterns are very similar for the two metals, in particular at the outer rim. This indicates a rather good degree of mixing with no surface segregation. The result is in agreement with a computational study by Xiao et al. (2006) according to which Au–Pt particles not exceeding 7,000 atoms (about 6 nm in diameter) have a negative heat of alloy formation for the full range of concentrations. This can be seen as a surface effect making the alloying of Au and Pt thermodynamically easy in small particles. Figure 7 shows the elemental maps of a significantly larger particle (largest dimension ca. 12 nm, see solid line frame) using a 10 × 10 grid. Again the dash lined frame identifies the reference to compensate for drift. It is clear from Fig. 7b and c that the distributions of Au and Pt within the particle are quite alike. The mean concentration of Pt in the particle is 20 at.% with a standard deviation of 9.7 at.%. For the smaller particle (5.5 nm) mentioned above, the mean Pt concentration is 34 at.% with a standard deviation of 10 at.%.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig6_HTML.jpg
Fig. 6

EDS compositional mapping of the specified Au–Pt particle ~6 nm in size (a), Au L (b) and Pt L (c)

https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig7_HTML.jpg
Fig. 7

EDS compositional mapping of the specified Au–Pt particle (largest dimension ca 12 nm) (a), Au L (b) and Pt L (c)

Although the results of Xiao et al. (2006) indicate mixing in full range of concentrations for particle sizes up to 6 nm, our results indicate mixing also for the larger particle in Fig. 7. We assume that this is because the particle is twinned consisting of smaller units. As the whole sample consists of primary particles hardly larger than 6 nm, we do not expect any de-mixed cases either for the larger aggregates.

The EDS spectrum of a small portion of the particle (0.7 nm recorded in 15 s) in Fig. 8 confirms that both Au and Pt are present in comparable concentrations.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig8_HTML.gif
Fig. 8

A typical EDS spectrum of a small portion of a Au–Pt particle (~0.7 nm recorded in 15 s)

Average compositions of the particles produced with different electrode polarities were measured by ICP and listed in Table 4. It can be concluded that for a specific material, the cathode is more strongly eroded than anode. This is because the majority of charge carriers in the plasma are positive ions and electrons. The positive ions have a much higher mass and thus dissipate more kinetic energy in colliding with the attracting electrode than the electrons, despite the higher mobility of the latter. The fact that we do see substantial erosion of the positive electrode is due to temporary voltage reversal during the discharge process (Tabrizi et al. 2009).
Table 4

Compositional analysis of Au–Pt samples measured by ICP

Electrodes

Au/Pt wt. ratio in the sample

Au (+)Pt (−)

0.97

Au (−)Pt (+)

1.75

X-ray diffraction patterns of the Au(+)Pt(−) and Au(−)Pt(+) samples are presented in Fig. 9. For both arrangements, formation of an intermediate phase is confirmed by Bragg reflections positioned between the peaks of pure gold and pure platinum, which is in agreement with the EDS maps. The inset shows a close-up of the 220 plane reflections on the d-axis scale, which directly allows an estimate of the composition of the intermediate phase. Au and Pt both have fcc structures with lattice constants of 0.40782 and 0.39242 nm, respectively. Vegards’s law predicts a simple linear relationship between alloy composition and lattice spacing for metals with the same crystal structures and close atomic sizes (Barret et al. 1996). For Au(−)Pt(+) and Au(+)Pt(−) arrangements, the alloy phases were approximated to contain around 75 and 60 at.% Au, respectively. According to the bulk phase diagram of Au–Pt, these compositions are in the miscibility gap below the temperatures of 1,100 and 1,400 K. Our results are in accordance with Luo et al. (2005), who reported that their bimetallic nanoparticles, synthesized by wet chemistry, displayed alloy properties in contrast to the miscibility gap known for the bulk phase diagram.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig9_HTML.gif
Fig. 9

X-ray diffraction patterns of the Au–Pt samples

The broadness of the XRD peaks is determined by the particle size and lattice imperfections. In the case of mixed phases, the XRD peaks should additionally be broadened, if there is a distribution of compositions rather than one fixed composition. Slight amounts of pure metals were also detected in the samples (see inset). Here the peak sharpness indicates a large particle size, probably due to solidified droplets ejected from the electrode surface, as these have been seen before in the case of Ag above.

Table 5 lists the BET surface areas of Au, Pt and Au–Pt nanoparticles. Au and Pt both showed specific surface areas corresponding to BET diameters much larger than typical primary particle diameters. In the case of Pt and Pt–Au, the BET diameters are also larger than the one observed in the micrograph in Fig. 5b and c. This is explainable, because sintering and coalescence on substrate is limited through the particle density, while much larger amounts of material in multiple layers was collected for the BET measurement. The Au–Pt particles showed a significantly larger specific surface area than the pure metals, indicating reduced cold sintering between the particles. Reduced sintering of the mixed phase had also been observed in our previous work (Tabrizi et al. 2008) for systems which were miscible in bulk scale.
Table 5

Specific surface area of the Au–Pt particles

Material

BET specific surface area (m2 g−1)

Calculated BET diameter (nm)

Au

14

20

Pt

<10

>28

Au–Pt

32

9–10

Cu-W

Copper-tungsten nanoparticles were produced from a pair of sintered electrodes. The TEM image of the particles in Fig. 10a shows that the primary particles have rather a broad size distribution. The inset with higher magnification reveals that the particles are covered by a surface layer. Due to enhanced chemical reactivity of a nanosized particle, small amounts of impurities in the carrier gas, in particular oxygen, may be sufficient for chemical modification of a surface layer, and we assume that an oxide layer has formed. The SEM image of the particle assemblies in Fig. 10b shows that the particles are mainly nanosized and that few large round particles are also present.
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig10_HTML.jpg
Fig. 10

TEM image of the Cu/W nanoparticles (a), SEM image of the Cu/W particles with scale bar representing 1 μm (b)

The atomic compositions of a number of nanosized particles (5–12 nm) were determined by EDS. Table 6 shows the composition of 10 particles. Each particle contains both Cu and W. The average atomic composition agrees with the elemental composition of the electrodes (weight: W72/Cu28 or atomic: W47/Cu53). In addition, fixed point EDS analysis with a beam profile smaller than 0.3 nm was performed on each particle at six points. The drift correction was performed by comparing to the highlighted regions in the solid frames as references. The EDS line scans on six particles in Fig. 11 show that the particles are not compositionally homogeneous. A typical EDS spectrum of a point is shown in Fig. 12.
Table 6

Typical atomic compositions of the Cu-W particles

 

Cu

W

 

64.298

35.701

63.543

36.456

0.527

99.472

44.565

55.434

76.221

23.778

28.356

71.643

59.172

40.827

56.989

43.01

41.655

58.344

43.527

56.472

Average

47.8853

52.1137

https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig11_HTML.gif
Fig. 11

Micrographs of the Cu-W particles with EDS line scans

https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig12_HTML.gif
Fig. 12

A typical EDS spectrum recorded from a small portion of a Cu-W particle (~0.7 nm recorded in 15 s)

X-ray photoelectron spectroscopy (XPS) was applied to determine the average elemental composition of the sample surface (Table 7). The ratio of the metal compositions agreed well with the composition of the sintered electrodes. The XPS results also show a remarkable amount of oxygen and carbon. The spectra reveal the presence of CuO, Cu(OH)2, WO2 and WO3 which is explained by the fact that exposure of the sample to air could not be avoided before the analysis.
Table 7

Compositional analysis of Cu-W sample measured by XPS

Material

Cu/W wt ratio in the electrode

Cu/W wt ratio on the surface of the sample

Cu-W sintered electrodes

0.388

0.396

With a large positive heat of mixing (ΔHf = +36 kJ/mol), Cu and W are immiscible even in the liquid state (Gladyszewski et al. 1993; Ouyang et al. 2006a; Haubold and Gertsman 1992). Cu and W have fcc and bcc (body-centred cubic) structures with atomic radii of 0.1278 and 0.137 nm, respectively. If W dissolves in the Cu lattice, the lattice parameter of Cu should increase and similarly a decrease in lattice parameter is expected for W if Cu dissolves in the W lattice (Raghu et al. 2001). Equilibrium phases in the phase diagram of bulk Cu-W calculated from a thermodynamic model include the fcc terminal solid solution (Cu) with extremely limited solid solubility of W, and the bcc terminal solid solution (W) with extremely limited solid solubility of Cu (Massalski et al. 1990). On the basis of first principles calculations, Shu et al. (2003) suggested that the noble metal atoms (Ag, Au and Cu) would like to occupy the vacancy sites of the W(001) surface to form the substitutional surface alloys despite the fact that they hardly form an alloy in the bulk. Dirks and van den Broek (1985) produced Cu-W alloy films by simultaneous vapour deposition of the elements on unheated substrates. They found that at least 10 at.% W may be dissolved in fcc Cu films and 40 at.% Cu may be accommodated in the bcc W alloy films. They concluded that the vapour quenching technique could lead to the formation of homogeneous one-phase alloys over a wide range of compositions. They estimated that the metastable two-phase (fcc + bcc) coexistence region should lie between 40 and 60 at.% tungsten. This is almost in the compositional range of the measured particles in Table 6. Metastable mutual solid solubility in the ball milled Cu-W nanocrystallites was also reported by Raghu et al. (2001). According to Xiao et al. (2006), there is a competition between size effect and compositional effect on the heat of formation of immiscible system. When the formation enthalpy reduces to a smaller value than the interface energy of the system, because of the size effects, interface alloying can occur (Liang et al. 2005).

The XRD pattern of the sample in Fig. 13 shows characteristic peaks corresponding to the reflections of the initial constituents as well as W3O with A15 structure. The sharpness of the peaks indicates that they are due to the large particles as seen in the SEM image (Fig. 10b). It is interesting to note that XRD does not show any indication of a nanocrystalline phase, which should manifest itself as broadened lines. On the other hand, the existence of nanoparticles is clearly confirmed by TEM as well as the BET analyses. We conclude that the particles are amorphous or the scale of ordering is below the nm range. Indeed, the TEM micrographs in Fig. 10 reveal contrast changes on a very small scale (see the inset). We ascribe the darker contrast to W (atomic no. 74) and the lighter one to Cu (atomic no. 29).
https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig13_HTML.gif
Fig. 13

X-ray diffraction pattern of the Cu-W sample

The electron diffraction pattern in the inset of Fig. 13 was obtained from an area, where only nanoparticles were present. It exhibits some small diffraction spots arranged along the rings, indicating very small crystallites to be present. The dominance of smeared continuous rings gives further support to an amorphous or quasi-amorphous phase.

Together with the EDS line scans of Fig. 11, we conclude that we have mixed Cu and W on a scale larger than the lattice constants but significantly smaller than the particle diameters. The EDS line scan on the upper left side of Fig. 11 would be consistent with a homogeneous mixing, but the line scan on the lower left side shows that very significant changes in composition occur on the scale around one nanometer.

The specific surface areas of Cu, W and Cu-W nanoparticles are listed in Table 8. The calculated BET diameters are consistent with the mobility particle sizes shown in Fig. 14.
Table 8

Specific surface area of the Cu-W particles

Material

BET specific surface area (m2 g−1)

Calculated BET diameter (nm)

Cu

130

5.1

W

72

4.3

Cu-W

109

3–6

https://static-content.springer.com/image/art%3A10.1007%2Fs11051-009-9603-4/MediaObjects/11051_2009_9603_Fig14_HTML.gif
Fig. 14

Size distributions of Cu, W and Cu-W nanoparticles (C = 5 nF, d = 1 mm, = 10 Hz, Q = 5 lpm Ar)

Conclusion

Using sintered electrodes of Ag/Cu in the spark discharge generator, we produced nanoparticles which exhibited enhanced solid solubility of Cu in Ag and Ag in Cu with respect to the macroscopic case. Effective mixing is also possible by using two electrodes of different compositions, as the circuit applied leads to field reversal during the discharge process. Au–Pt particles produced this way show intermediate phases with compositions which lie in miscibility gap in bulk phase diagram at ambient temperatures. In agreement with this, EDS elemental mapping of a 6 nm particle reveals good mixing. This is in agreement with computation by Xiao et al. (2006). Since gold and platinum are both noble metals with low chemical reactivity, cleanness of the particle surfaces leads to cold sintering and low specific surface areas for pure particles. Admixing of Au and Pt results in an increase in specific surface area of the mixed particles. Applying sintered electrodes of Cu/W, we produced mixed particles a few nanometers in size containing both Cu and W. EDS line scans together with TEM, XRD and electron diffraction give evidence of mixing on a subnanometer scale but above the scale of atomic mixing. In conclusion, spark discharge ablation is a powerful technique for producing new mixed nanoparticulate phases that do not exist in macroscopic systems. Beside using the method for basic studies, the possibility of scaling it up by using multiple discharges bears the potential of producing new materials on a larger scale.

Acknowledgements

The authors would like to express their gratitude to Miren Echave Elustondo for carrying out particle size distribution measurements and Sander Brouwer for his assistance in BET measurements. The Project is partially funded by the Delft Center of Sustainable Energy (DISE).

Copyright information

© Springer Science+Business Media B.V. 2009