Ion Implantation for the Fabrication of Plasmonic Nanocomposites: A Brief Review

Metal nanoparticles (NPs) have been the subject of intense research in recent years, largely due to their characteristic surface plasmon resonance (SPR), a phenomenon arising from the collective oscillation of their conduction electrons. The SPR characteristics of metal NPs have been exploited to utilize them in medical diagnostics (Allain & Vo-Dinh, 2002), immunoassay (Hirsch et al., 2003a; Cui et al., 2006), cancer treatments (Hirsch et al., 2003b; Zhang, 2010), biological or chemical sensing (Sun & Xia, 2002; Alivisatos, 2004; Cao et al., 2009), and surface enhanced Raman scattering (SERS) (Jackson & Halas, 2004). Large thirdorder optical susceptibility and picosecond optical response of metal NPs made them excellent candidates for non-linear optical devices (Mazzoldi et al., 1996; White et al., 1998; Pavesi et al., 2000; Borsella et al., 2001). Strong SPR-induced optical absorption of these nanostructures has also been utilized to enhance their photocatalytic activity (Tihay et al., 2001) and to increase the efficiency of dye-sensitized solar cells (Catchpole & Polman, 2008; Atwater & Polman, 2010; Ding et al., 2011).

Frequently these NPs are incorporated into dielectric or semiconducting matrices to fabricate dielectric-metal or semiconductor-metal nanocomposites with the aim of manipulating their functionalities, exploiting the characteristics of both the matrix and the metal NPs. Among the nanocomposites, those in which the SPR effect is prominent are called plasmonic nanocomposites (PNCs). Frequently these nanocomposites are fabricated by incorporating nanoparticles of noble metals like Au, Ag and Cu in insulating matrices like silica (a-SiO 2 ), quartz (c-SiO 2 ), and other metal oxides like zinc oxide (ZnO), alumina (Al 2 O 3 ), and titania (TiO 2 ). Though, in principle, several other metal particles can be used for the fabrication of PNCs this is seldom done, due to their high proneness to oxidation or interband transitions in the usable spectral range. Consequently, in this short review we will mainly focus on these three important metals.
Two characteristics, particularly important for practical use of plasmonic nanocomposites, are: the location of the SPR in the desired spectral position and a high optical extinction. For the very small metal NPs (i.e. those much smaller than the optical wavelength) the extinction is dominated by absorption and the absorption coefficient () can be expressed as (Arnold & Borders, 1977): where Q is the volume fraction occupied by the metallic nanoparticles, n d is the refractive index of the host matrix,  1 and  2 are the real and imaginary parts of the dielectric constant of the bulk metal, respectively, and  is the optical wavelength. The value of  becomes maximum at the surface plasmon frequency ( SPR ), when ε + n = . (2) As we see, the SPR frequency of metal nanoparticles depends on both the optical constant of bulk metal and the refractive index of the host matrix. Estimating the wavelength variation of  1 from the optical constants given by Johnson and Christy (Johnson & Christy, 1972) for Au, Ag, and Cu, and refractive indices of common oxide matrices we can calculate the position of the SPR band for different nanocomposites, as presented in Table 1. Though the position of SRP in different nanocomposites depends also on the size and size distribution of the metal NPs, for which we must introduce size correction terms (Hövel et al., 1993) in equation (2), we can see that the tabulated SPR energies are in good agreement with the reported values for the nanocomposites fabricated by different methods (Ballesteros et al., 1997;De, 1997;Pal et al., 2000Pal et al., , 2001Vazquez-Cuchillo et al., 2001;Wang et al., 2003;García-Serrano & Pal, 2003;Barreca et al., 2004;Subramanian et al., 2004;Zeng et al., 2004;Hazra et al., 2004;Okumu et al., 2005;Ryasnyansky et al., 2005;Serna et al., 2006;Sonawane & Dongare, 2006;Mishra et al., 2008;Zheng et al., 2008;Kalidindi & Jagirdar, 2008;Kim et al., 2008;Manoilov, 2009;Jiang et al., 2009;Ma et al., 2009;Sangpour et al., 2010;Tohidi et al., 2010;Duhan et al., 2010). As the absorption coefficient depends on the volume fraction occupied by metallic NPs in the composites (eq. 1), the intensity of the SPR band increases on increasing the amount of incorporated metal NPs.
Several methods have been utilized to fabricate plasmonic nanocomposites like cosputtering, ion-implantation, sol-gel, laser ablation, etc. (De, 1997;Pal et al., 2000;García-Serrano & Pal, 2003;Barreca et al., 2004;Kim et al., 2008;Manoilov, 2009). However, irrespective of the fabrication method, practical application of these materials strongly depends on parameters like the size and shape of the incorporated metal NPs, volume fraction occupied by them, homogeneity of NP distribution, and their composition. In this respect, each of the utilized methods has certain constrains. Although ion-implantation is not as versatile as other techniques used for the fabrication of plasmonic nanocomposites, it still has several attractive advantages. For example, ion implantation followed by an additional thermal annealing has proven to be a very useful method to obtain large volume fractions of NPs embedded in solid matrices. Additionally, those NPs are located in a well defined depth below the surface, chosen mainly by the ion energy (Oliver et al., 2002). Some other advantages of ion implantation include controllability of depth profile and concentration, high purity, and the possibility to overcome low solubility restrictions. Particularly, deep ion implantation using energies of the order of MeV produces an ion www.intechopen.com Ion Implantation for the Fabrication of Plasmonic Nanocomposites: A Brief Review 329 depth distribution located some micrometers underneath the surface and wide enough to be convenient for the production of optical waveguides. Ion implantation has been also successfully used to create more complex structures such as alloys (Mattei, 2002), sophisticated aggregates (Mattei et al., 2003b;Pellegrini et al., 2007aPellegrini et al., , 2007b, core-shell structures (Peña et al., 2009) and nanorods (which are aligned, unlike those produced by other methods) (Rodríguez-Iglesias et al., 2010a;. In this chapter, we intend to review the current status of the synthesis of plasmonic nanocomposites by means of ion-implantation. In the year 1984, Picraux (Picraux, 1984) has published a useful review on the status of ion-implantation in metals, improving the understanding on the kinetics and equilibria in metal alloy reactions, surface modification, and defect structures. Similarly, Buchal et al. (Buchal et al., 1994) have reviewed the status of ion-implantation in optical materials in 1994, focusing mainly on the usefulness of the technique for producing linear and non-linear optical devices. However, apart from a good review on Cu implanted alumina by Stepanov and Khaibullin (Stepanov & Khaibullin, 2005) there exists no recent review summarizing the current advances made on the utilization of this technique for the fabrication and processing of plasmonic nanocomposites which have emerging applications in diverse fields.

Incorporated species
Matrix material (refractive index)

Implantation of metal ions in dielectrics
Since long ago, ion implantation has been used to modify the optical properties of materials. Dielectrics such as silica glass (SiO 2 ), alumina (Al 2 O 3 ), and quartz have been used traditionally for this purpose; but recently several other metal oxides like ZnO and TiO 2 have also been utilized. Historically, ion implantation has been used to change the optical properties of dielectric target materials like reflectivity, absorption, refractive index, luminescence, and electro-optic coefficients (Mazzoldi, 1987;Townsend, 1987). Those modifications are not only associated to the implanted species, but also to a number of other phenomena occurring during the process like radiation damage caused by the ion beam, and formation of colloidal precipitates of the implanted species with diameters in between 2 and 20 nm (Hughes, 1983). As the mechanism of formation of radiation damages and their elimination processes have been widely discussed in the literature (Agulló-López et al., 1988;Pal et al., 2000;Bianconi et al., 2009;Rivera et al., 2010), we will concentrate our discussion on the latter mechanism only.
In fact, the formation of colloidal precipitate of metals like Au and Ag has been utilized for hundreds of years to design beautiful colored glasses (Bamford, 1977). While the color of the implanted dielectrics depends mostly on the implanted species (metals) and to some extent the size of the formed colloidal particles, their intensity/contrast depends on the volume fraction occupied by the colloidal precipitates in the matrix. So, apart from the nature of implanted metal, a control over the implanted dose (amount of implanted species incorporated into the dielectric matrix) is essential to control the optical properties of the substrates (dielectrics). In contrast to the conventional melt growth process, where only about 1% of impurities (guest atoms) can be incorporated into a matrix, there is no such restriction in ion implantation, as the process occurs without thermodynamic restrictions.
For the ion implantation purpose, the ion beams of selected metals doses in the range of 10 16 -10 18 cm -2 have been utilized to expose solid target surfaces as big as a few cm 2 . In fact, the process has the ability to fabricate technological grade composite wafers with uniform distribution of metal NPs at precise depths below the target surface. The range of metal NP distribution, their occupied volume fraction, size (diameter), and composition can also be controlled by suitable adjustment of implantation parameters. In the following section, we will discuss how these parameters can be controlled and their effects on the characteristics of the final product (nanocomposites).

Controlling parameters in ion implantation process
Essentially any specie can be introduced into a solid matrix by means of ion implantation, thereby changing its physical properties. For this purpose, ions of the desired element are produced in an ionization chamber; then they are electrostatically accelerated to the desired energy and, finally, allowed to impinge on a solid target. As typically each ion is a single atom, the amount of material incorporated or implanted into the target can be precisely calculated by integrating the ion current over the exposure time. The amount incorporated is called dose or fluence. Generally the ion currents used for implantation are small, in the range of nano-to microamperes. Ion beam energies may vary from a few keV up to hundreds of MeV, depending on the requirement, though energies in the range of a few MeV are commonly used to produce PNCs. Greater ion energy causes higher structural damage to the target. However, due to broader depth distribution, the change in chemical composition at a certain point in the target would be small.
When high energy ions penetrate into a solid, they gradually lose energy due to non-elastic collisions with the free electrons of the solid and elastic collisions with the nuclei until they become immobilized. As the second process is an elastic one, most of the energy loss of the incident ion bean occurs very close to the surface of the solid, due to the first process. The penetration depth of the ions in the solid target depends on the ion energy, nature of ionic species, and the nature and composition of the target. Typically, the penetration depth or the range of the ions in a solid target varies in between a few nanometers to a few micrometers, and hence all the changes occurring during the implantation process remain near the surface of the target material ( Figure 1). Typical ion doses utilized for fabricating plasmonic nanocomposites vary in between 1.0x10 16 cm -2 and 2.0x10 17 cm -2 . While a greater dose frequently induces the formation of undesired nanovoids or nanocavities in the target matrix (Kooi et al., 2000) or inside metal precipitates (particles)  due to aggregation of vacancies driven by the heat produced during the high-dose ion irradiation, lower doses frequently incorporate metallic species in ionic state inside the dielectric matrices (Haug et al., 2009). While higher ion energy enhances the penetration depth of the implanted species, a higher ion dose shifts the maximum of the distribution of the implanted species in the target towards lower depths, due to the diffusion to the heavily-damaged region produced just before the ions are stopped ( Figure 2). Though the penetration depth of ionic species remains within a few micrometers beneath the surface of the implanted solid, it can extend to a larger depth on increasing the temperature of the solid target.
In the case of an amorphous solid target, the implanted ions get distributed randomly following a Gaussian profile inside the matrix, which can be expressed as: where N(x) is the concentration of ions at a distance x from the surface; N 0 is the concentration of ions at x = R p ,   is the total implanted dose and R p is the effective range of

Ion beam
Depth Ion concentration www.intechopen.com penetration by an ion, dependent of real range R c , where R c , the real range of penetration, i.e. distance covered by the ion in the implanted solid, and R p is the standard deviation of the R p value. The location of R p depends predominantly on the energy of ions and their atomic mass. The relative dispersion of implanted specie, i.e. R p /R p depends both on the atomic mass of the implanted species and the atomic mass of the solid target. Fig. 2. a) Variation of profile and depth of implantation of Cu + ions into Al 2 O 3 with the variation of ion energy, calculated using SRIM 2008 (Ziegler, 1985(Ziegler, , 2008; b) Implantation depth versus dose, calculated using dynamic computer code DYNA of Cu + ions of 40 keV in Al 2 O 3 . Adapted with permission from AIP Conf. Proc. 680 (2003) 601-604 (Stepanov & Popok, 2003).
On the other hand, for the case of a crystalline solid target, the range of penetration and the implant distribution strongly depend on the orientation of the structure with respect to the direction of beam incidence. Generally the penetration depth or the implantation range is about one order higher when the ion beam is projected along a given crystallographic direction of the target like [111], [100], etc. (Burakowski & Wierzchon, 1998). As can be seen from Figure 3, the defect distribution profile remains at a shallower level with respect to the implant distribution. Dose ( x10 15 cm -2 ) 1 -0.1 2 -0.3 3 -0.6 4 -0.9 5 -1.2 6 -1.5 7 -1.8 8 -2.1 9 -2.4 10 -2.7 Fig. 3. a) Schematic illustration of the range of penetration; b) variation of implantation range R p and standard deviation R p with ion energy for 2 MeV Au ions implanted in silica; c) distributions of implanted ions and defects. All simulations were performed using SRIM 2008 (Ziegler, 1985(Ziegler, , 2008.

Controlled growth of noble metal nanoparticles through ion-implantation
The noble metals like Au, Ag and Cu have been incorporated into dielectric matrices to produce several plasmonic nanocomposites with varied combinations of metals and dielectrics. The main objective of such works was to utilize the intense local field, strong linear optical absorption or SPR in applications such as optical sensing, biological and chemical sensing, and fabrication of all-optical devices. The electronic transitions responsible for plasmon absorption in metal NPs are also responsible for the generation of optical nonlinearities in the same spectral range, i.e. at the SPR position. To exploit the advantage of this SPR absorption in nanocomposites, it is desirable to have a high concentration of metal NPs (i.e., high filling factor) in the matrix (Flytzanis et al., 1991). Therefore, as has been mentioned earlier, apart from the adequate energy of implantation, a high dose of implanted ions is required. By the terms "high metal ion dose"  0 , we mean a dose exceeding the solubility limit of the metal atoms in the target (dielectric in the present case), which generally vary in between 10 15 -10 17 cm -2 , depending on the nature of ion species and target materials. The characteristics of the incorporated metal NPs also depend on the utilized ion dose. While there is a limiting value of implanted dose for each combination of metal ions and target material to obtain well dispersed metal NPs with controlled nucleation and growth (i.e. with definite geometrical shape and size), doses higher than this limiting value cause aggregation or coalescence of the formed NPs, strongly affecting their optical properties (Figure 4). For example, for 25 keV Ag + ion implantation into LiNbO 3 , the limiting dose was found to be  0 ~5.0x10 15 cm -2 (Deying et al., 1994), and ~10 16 cm -2 for 30 keV Ag + ions in solid epoxy resin (Stepanov et al., 1995). In fact, the implantation range and concentration distribution of the metal NPs strongly depend on the implantation dose (Figure 2b).
Fabrication of plasmonic nanocomposites by ion implantation dates back to 1973, when Davenas and co-workers (Davenas et al., 1973) pioneered this method to fabricate metal NPs of silver, calcium, etc. inside crystalline LiF and MgO matrices. Later, the method was extended for the fabrication of other metal NPs in several other matrices like sapphire (Al 2 O 3 ), silica (SiO 2 ) and even polymers. For practical applications, plasmonic nanocomposites containing metal NPs of desired sizes, high filling fraction, and controlled composition are the basic requirements. While the ion implantation technique can introduce virtually any element in any matrix without solubility restrictions, the typical cluster size distributions in ion implanted nanocomposites are not narrow. On the other hand, progress has been made to fabricate metal-dielectric nanocomposites with controlled composition of metal NPs like elemental metallic clusters, bimetallic alloy clusters and core-shell type bimetallic nanoclusters. Depending on the nature of the metallic element and the matrix, adequate adjustments of implantation conditions (e.g. implantation energy, dose) and post implantation treatments (thermal annealing, laser annealing, etc.) in adequate atmosphere (e.g. oxidizing atmosphere like air or oxygen, inert atmosphere like argon and reducing atmosphere like hydrogen or a mixture of hydrogen and argon) are necessary to fabricate each type of the above mentioned architectures. In the following sections we will discuss on the fabrication of each of these nanoarchitectures, their plasmonic behaviors, and application potentials.

Formation of nanocomposites
At a certain temperature, when the solute concentrations are above their solubility limits in a matrix, there occurs spontaneous condensation of solute. Within a homogeneous matrix and in conditions which may rule out coalescence, this condensation usually leads to the formation of spherical clusters. The theory of growth of spherical clusters has been developed and used by many authors (Lifshitz & Slezov, 1959;Aaron et al., 1970). According to this theory, at relatively low temperatures (in comparison with the melting temperature of bulk materials), formed clusters predominantly go through a stage of nucleation and growth. The exact analytical solution of the growth process of a spherical cluster is very difficult to obtain. For this reason, various approaches have been used to represent the real solution to a greater or lesser extent. In particular, we want to solve the field equation: where D is the volumetric coefficient of diffusion (assumed to be independent of position) and C = C(r,t) is the field of concentration in the matrix around the precipitate. Also, the independent equation of flow balance must be satisfied: C P is the concentration of the precipitate (constant), C I is the solute concentration in the matrix, at the precipitate-matrix interface and R is the value of r on that interface. For the condition (C P -C M )>>(C I -C M ), the R(t) is a slowly varying function of time and a way to simplify the field equation (4) is by using the invariant-field approximation in time (Laplace approximation), which is 0 Ct   , thus being the Laplace equation 2 0 C   . Now, for the steady state of the cluster, we can solve the equation (4) to obtain: where C M is the concentration of solute in the matrix far away from the precipitate. The problem is almost identical for the dissolution of clusters. In this case R(t = 0) = R 0 > 0, we obtain the expression: Note that k is positive during the dissolution and negative during the growth.

Fabrication of nanocomposites containing elemental noble metal clusters
Controlling the plasmonic behavior of noble metal clusters (nanoparticles), i.e. their SPR position and intensity in nanocomposites is of immense importance for their practical applications. Manipulation of the implantation variables (energy and dose) coupled with a post-implantation annealing allows the fabrication of metal nanoclusters with the desired SPR characteristics. Since the first report on ion implantation fabrication of such nanocomposites  (Kono et al., 2003;Norton et al., 2003;Stepanov et al., 2004) Table 2. List of noble metal ions used to incorporate in common oxide matrices with different implantation conditions and post implantation treatments.
by Davenas' group in 1973(Davenas et al., 1973, a vast effort has been devoted to their controlled fabrication using different combinations of metal particles and matrices. As can be seen from table 2, Au, Ag and Cu NPs have been incorporated into several dielectric matrices with different implantation conditions and the conditions of subsequent annealing. Implantation of these metal ions of moderate energy (typically up to 200 keV) and low doses (10 16 cm -2 ) generally produces metallic nanoclusters of homogeneous sizes, while a high dose and a high temperature post implantation treatment produce nanoclusters of bigger sizes with higher dispersion. However, both the size and the size dispersion of the formed nanoclusters depend on the nature of implanted metal, their solubility and diffusion behavior in the embedding matrix. While Ag nanoclusters of relatively uniform sizes and narrow size distributions can be grown in silica matrix even for a high ion dose ( Figure 5) www.intechopen.com Ion Implantation 338 without post-implantation thermal treatment, the same is not true for the Au and Cu nanoclusters in silica matrix ( Figure 6). As can be seen from Figure 6, Au nanoparticles formed with similar implantation conditions and thermal treatments have a broad size distribution.   (Buchal et al., 1994).
While a very low ion dose produces metal nanoclusters of extremely small sizes (typically below 1.0 nm) that exhibit no SPR band in absorption spectra, on increasing the ion dose, the clusters grow and the intensity of the SPR extinction band increases (Figure 7). On the other hand, for a high energy of the implanted ions, the size of the formed metallic clusters can be big enough to reveal an SPR peak in the absorption spectra of the composites ( Figure 8B). While a high energy implantation can place the metallic clusters at micrometer depth below the irradiated surface, a higher implantation dose can extend the cluster distribution towards the target surface ( Figure 8A). Such a broadening of cluster distribution towards lower depths is due to the interaction of the incident ions with the already formed metallic clusters, which effectively reduces the projected range of the metal ions that impinge at a later stage. Overall, incorporation of monometallic NPs in dielectric matrix through ion implantation has had a good success so far. At the present time, researchers can fabricate specific noble metal clusters of desired sizes and size distributions at precise depths inside most common dielectric matrices by controlling the implantation dose, energy and the conditions of post implantation treatment. However, from the application point of view the success has been limited. Though the SPR position in a nanocomposite could be 9.95x10 16 ions cm -2 1.25x10 17 ions cm -2 2.20x10 17 ions cm -2 modified by varying the nature of implanted metal ions (e.g. the typical SPR position of Au nanoclusters in SiO 2 is at about 540 nm, of Ag at about 420 nm, and of Cu at about 560 nm) and its intensity could be raised by increasing the implantation dose, it is not possible to tune the position of the SPR peak for a wider spectral range using monometallic nanoclusters. The fabricated monometallic plasmonic nanocomposites are suitable for the fabrication of nonlinear optical switches, effective waveguides, optical sensors, and devices covering a limited spectral range, but the limitation of spectral tunability of SPR peak position makes them unsuitable for most biological applications, such as medical diagnosis, immunoassay, and therapeutic treatments.
Therefore, researchers felt the need of fabricating plasmonic nanocomposites containing bimetallic nanoclusters to vary the SPR position of the composites in between the SPR positions of individual monometallic metal clusters, which we will discuss in the following section.

Fabrication of nanocomposites containing alloy-and core-shell type bimetallic clusters
Sequential ion implantation, where the ions of a particular metal are first implanted at a certain depth of the matrix and then the second type of metal ions are implanted at similar depths, has been utilized frequently to fabricate bimetallic nanocomposites. Depending on the conditions of implantation, the nature of metallic species and dielectric matrix, the process can lead to the formation of i) separated monometallic nanoclusters of individual metals, ii) alloy clusters, and iii) core-shell type bimetallic clusters. Formation of each of these clusters depends on their thermodynamic stability, and hence on the process of post-implantation thermal treatments along with the annealing atmosphere. Although the composition of alloy type clusters can be varied easily by sequential ion implantation of two elements in a matrix by adjusting their energy and doses to have maximum overlap between their implant-profiles and desired local relative concentrations, post implantation thermal treatments can frequently lead to (promote) separation of a component specie from the other or oxidation, instead of alloying. Formation of alloy type clusters also depends on the miscibility of the component elements, their chemical interaction with the matrix, and the creation of radiation induced defects, which influences the nucleation and growth of the nanoclusters.
As the criterion of bulk miscibility for two elements is not very stringent in the case of nanocrystals due to their high specific surface area (surface area/volume ratio) and characteristics closer to that of a molecule than to those of a massive (bulk) system (Yasuda & Mori, 1994), alloy type nanoclusters of several metal pairs like Au-Cu, Au-Ag, Pd-Ag, and Pd-Cu, have been successfully fabricated (Gonella et al., 1999;Battaglin et al., 2000Battaglin et al., , 2001a through sequential ion implantation. The SPR position in such noble metal alloy clusters remains in between those of pure elements, depending on the complex interplay between the modified free electrons and interband absorptions. As can be seen from Figure 9, the measured absorption spectra of pure Au, Ag and Au 0.4 Ag 0.6 alloy clusters of about 3 nm size in silica matrix reported by Mattei and co-workers (Mattei, 2002) agree well with the simulated extinction spectra of the NPs of similar compositions reported by Kreibig & Vollmer (Kreibig & Vollmer, 1995). Fig. 9. Comparison between Mie-theory based calculated extinction spectra of 3 nm clusters of Au, Ag, and Au 0.4 Ag 0.6 alloy in silica (9a) with the experimental absorption spectra of similar clusters prepared by ion implantation (9b). The notations "A" and "H" at the end of sample names indicate the samples were annealed in air and hydrogen atmosphere, respectively. Reproduced with permission from Nucl. Instrum. Meth. B 191 (2002) 323-332 (Mattei, 2002).
Using sequential implantation technique, Mazzoldi and collaborators were able to fabricate alloy nanoclusters of different compositions in fused silica glass at room temperature using 190 keV gold and 90 keV Cu ions (Gonella et al., 1999). The alloy clusters of compositions close to their nominal values were formed in the silica matrix even without any postimplantation thermal treatment ( Figure 10). However, Cu migrates towards the particle surface during thermal treatment and a structural reordering of the alloy occurs. The effect was more prominent when the implanted samples were annealed in air, and Cu migrated to the surface gets oxidized, causing a phase separation (Au core and Cu 2 O shell). A complete separation of Cu from the alloy cluster and formation of CuO shell around the Au core could be obtained after 5 h of thermal annealing in air (Battaglin et al., 2001b). The process could be reversed by annealing the sample again in a reducing atmosphere. It must be noted that for the thermal treatment of metal ion implanted composites we frequently use the value of the bulk melting point temperature, even though this quantity is substantially lower for a metallic cluster (either elemental or bimetallic), due to thermodynamic size effect (Buffat & Borel, 1976). Thermal annealing of a sequentially implanted sample helps to migrate one component elemental species towards the small clusters of the other element through diffusion, enhancing their size and improving their stoichiometry. In fact, the probability of formation of homogeneous alloy clusters in sequential implantation process strongly depends on the solubility of the second elemental specie in the nanoclusters of the first element. To increase the solubility and helping the structural rearrangement of metal ions, sometimes it is useful to perform the implantation of the second element at elevated target (substrate) temperatures.  (Gonella et al., 1999).
Utilizing the above mentioned strategy, Kluth and coworkers (Kluth et al., 2006) could fabricate core-shell type Co-Au nanoclusters of about 5.3 nm diameter in SiO 2 matrix. Utilization of elevated target temperature (400 0 C in the present case) during second metal (Co in this work) implantation helped to avoid the post-implanted thermal annealing process for the formation of bimetallic clusters. However, for the formation of Au clusters of uniform size before Co implantation, they used a thermal annealing treatment of the Au implanted sample at 1000 0 C for one hour, in air. We can see the formation of uniform nanoparticles with Co core and Au shell ( Figure 11) in their samples. As has been mentioned earlier, the formation of core-shell type bimetallic clusters depends strongly on the segregation of the component metals controlled by the minimization of surface free energy, resulting a phase separation as has been demonstrated theoretically for Co-Au (Johannessen et al., 2005), Au-Pt (Esparza et al., 2005), Au-Pd (Liu et al., 2005, Au-Cu  and many other systems. Fig. 11. Cross-sectional TEM images and size distribution of Co-Au nanoclusters as well as monometallic nanoclusters formed by sequential implantation. The high resolution TEM image of the core-shell cluster reveals a significant bending of the lattice fringes due to interfacial strain (lattice mismatch). Reproduced with permission from Appl. Phys. Lett. 89 (2006) 153118 (Kluth et al., 2006).
Formation of Au-Ag core-shell nanoclusters in silica matrix through sequential implantation and subsequent thermal annealing (in air) has been demonstrated by Peña and coworkers (Peña et al., 2009). Considering the miscibility and segregation behavior of the two metals, they first implanted 1.8 MeV Ag + ions at a 5.0x10 16 ions cm -2 dose (fluence) to place most of them at about 1.0 m depth below the target surface ( Figure 12). Then, the Au + ions of higher energy (3.7 MeV) were implanted at a relative higher dose (6.7x10 16 ion cm -2 ) to place them at greater depth (about 1.1 m) keeping a good portion of the implantation profiles overlapped. However, it has been observed that during the implantation of Au, preimplanted Ag segregates towards the sample surface (distribution maximum moves to 0.87 m depth). On thermal annealing at 900 0 C for 1 h in air, the co-implanted samples produced core-shell type bimetallic nanoclusters with Au core and Ag shell. The formation of both the monometallic nanoclusters (embedded in silica matrix) before thermal treatment and bimetallic nanoclusters (after thermal treatment) could be verified through the experimental optical absorption spectra in comparison with the Mie theory based simulated spectra of similar structures ( Figure 13). Though it was not possible to differentiate the core and shell regions from their HRTEM images, due to very similar lattice parameters of Au and Ag (4.078 Å for Au, JCPDS 4-0784; 4.077 Å for Ag, JCPDS 87-0720), the formation of core-shell structures could be demonstrated using high angle annular dark field (DAADF) images of the nanoclusters (Figure 14). Indeed, the ion implantation process can produce both, alloy and core-shell type bimetallic nanoclusters in silica and other dielectric matrices by adequate manipulation of implantation parameters and proper post-implantation treatments. However, apart from producing monometallic and bimetallic nanoclusters of common types, ion implantation can produce several special metallic structures and aggregates, which we will discuss in the next section.

Fabrication of nanocomposites containing unusual metallic nanostructures
The anisotropic plastic deformation process of amorphous materials under MeV ion irradiation has been reported by Klaumünzer and Schumacher as early as 1983 (Klaumünzer & Schumacher, 1983). The process can be understood by means of a visco-elastic model, in which the thermal expansion of the cylindrically-shaped high temperature ion track region causes shear stresses that relax by viscous flow, with the viscous strains subsequently frozen in upon rapid cooling of the ion track region (Trinkaus & Ryazanov, 1995;van Dillen et al., 2005). The deformation is induced by the electron energy loss, and thus occurs mostly with ions in the MeV energy range. Fig. 15. Bright-field TEM cross-sectional micrographs of the AuCuH (annealed in H 2 (4.0%)-N 2 atmosphere at 900 0 C, 1h) before (a), after a thermal annealing in air at 900 0 C for 15 min (b) and after irradiation at room temperature with 190 keV Ne ions, at a dose of 1.0x10 17 ions cm -2 (c). Adapted with permission from J. Non-Cryst. Sol. 322 (2003) 17-21 (Mattei et al., 2003a).
Utilizing the above mentioned concept, ion implantation technique has been utilized for the fabrication of several unusual nanostructures either inside dielectric matrices or in other dielectric-metal combinations with distinct plasmonic behaviors. Though all of these nanostructures do not fall into the class of commonly known nanocomposites (where nanoparticles of one material are embedded/dispersed into another matrix material), as they are in combination with at least one material apart from the metal in question, we can call them as composite nanostructures. For example, Mattei and co-workers (Mattei et al., 2003a(Mattei et al., , 2005 could disintegrate alloy clusters like Au-Cu and Au-Ag embedded in silica matrix by high energy Ne, He and Kr ions to form Au-rich satellite clusters around the initial bimetallic clusters (Figures 15 & 16).
While the optical absorption spectrum of the co-implanted (190 keV Au and 90 keV Cu, 3.0x10 16 ions cm -2 ) AuCu sample after annealing in inert atmosphere revealed SPR peak of AuCu alloy particles (550 nm, curve (a) of the Figure 15), after irradiation of 190 keV Ne ions (1.0x10 17 ions cm -2 ) the SPR peak shifts to 575 nm, which correspond to the SPR band of the main particle along with the satellite ones (aggregate). Similar phenomenon occurred for the AuAg nanoclusters embedded in silica matrix. It has been observed that in both cases, mainly the Au segregates out of the main bimetallic cluster to form Au-rich satellite clusters of smaller sizes. The size of the satellite clusters depends on the atomic mass of the irradiating ions. Higher atomic mass of the irradiating ions produces satellite clusters of bigger sizes ( Figure 16). However, the phenomenon of disintegration of the preformed noble metal clusters in dielectric matrix occurs when the atomic mass of the irradiating ions is lower than the atomic mass of the embedded metal nanoclusters. The process was tentatively related to the nuclear component of the energy released to the system during irradiation with different ions. Fig. 16. Bright-field TEM cross-sectional micrographs of the sample AuAgA (annealed in air atmosphere at 800 0 C, 1h) before (a) and after irradiation at room temperature with 380 keV Kr ions at a dose of 2.0x10 16 ions cm -2 (b) (Mattei et al., 2003a). Adapted with permission from J. Non-Cryst. Sol. 322 (2003) 17-21 (Mattei et al., 2003a).
On the other hand, irradiation by high energy ions of higher atomic mass (with respect to the implanted metals in a dielectric matrix) generates very different phenomena. Generally, when an embedded metallic nanocluster is irradiated by heavier ions of higher energy (in comparison to the energy of implantation of the metal clusters), the preformed nanocluster can grow in bigger size or get deformed. For the low dose/fluence of irradiation, the preformed metallic nanoclusters grow in size maintaining their initial shape. However, an increased dose of irradiation generates deformation of the nanoclusters. In fact, several researchers have reported this deformation process for the noble metal nanoclusters embedded in amorphous dielectric matrices. While D'Orléans and co-workers (D'Orléans et al., 2003) could observe this effect for the preformed Co nanoclusters embedded in silica matrix upon irradiating them by 200 MeV 127 I ions, Awazu and co-workers (Awazu et al., 2009) could observe this effect for the pre-implanted gold and silver nanoclusters embedded in silica matrix ( Figure 17). Ag and Au nanoclusters could also be deformed on irradiating them by high energy Si ions to form elongated or rod shaped aligned nanoclusters in silica matrix (Rodríguez-Iglesias et al., 2010a;. As we can see from Figure 18, the elongated nanoclusters are well aligned, and generate SPR absorption bands corresponding to longitudinal and transversal vibrational modes of prolate nanoclusters on oblique incidence. Indeed, the metallic clusters get deformed upon irradiation of high energy heavy ions along the direction of incident ions; i.e. they get elongated in this direction. On the contrary, the deformation or expansion of nonmetallic or dielectric clusters occurs in the perpendicular direction of the irradiating ion beams.  (d) silica-Au-silica core-shell-shell colloid (d SiO2 = 300 nm, t Au = 25 nm, t SiO2 = 133 nm) after 6 MeV Au irradiation (6.5x10 14 ions cm -2 ). Scale bars represent 500 nm. Reproduced with permission from Nucl. Instrum. Meth. B 242 (2006) 523-529 (Penninkhof et al., 2006).
Utilizing this deformation behavior of metal and dielectric nanoparticles, Pennikhof and coworkers (Penninkhof et al., 2006) could produce colloidal prolate core-shell type Au-SiO 2 and reverse nanoparticles as big as 1 m in size ( Figure 19) by irradiating high energy Au or Cu ions. As the ion irradiation induced deformation is more prominent for a dielectric like SiO 2 than a metal like Au, the extent of deformation of the SiO 2 -core Au-shell particles is not as drastic as of Au-core SiO 2-shell particles. In fact, the presence of a thin layer (shell) of metal imposes a mechanical constrain on the deforming silica core, reducing the net deformation compared to that of pure silica (Penninkhof et al., 2006).
Optical characteristics of all the above plasmonic structures, which we defined as 'special structures' are quite different from that of conventional nanocomposites where the shape of incorporated/embedded noble metal nanoclusters remains in its most stable or spherical form. As the technological application of plasmon-based nanocomposites remains on the ability of manipulating their optical behavior, which is possible mainly by controlling the size, shape and other geometrical parameters, rather than just changing their composition. Ion implantation technique has demonstrated its great potential for their fabrication and manipulations.

Prospects of ion-implantation for the fabrication of plasmonic nanocomposites
Significant advances have been made in the last few decades on the fabrication of noble metal plasmonic nanocomposites through ion implantation. By controlling the implantation parameters and post implantation annealing, noble metal nanocomposites with embedded elemental, alloy and core-shell type NPs could be produced with moderate control over the NP size, size distribution, composition and location in the matrix. The significant advantage of ion implantation technique is that the treated surface is an integral part of the sample and consequently does not suffer from possible adhesion problems associated with coatings. Therefore, the implanted structures remain as the integral parts of an optical device. The moderate heating associated with the process virtually eliminates any risks of distortion or oxidation effects. On the other hand, ion implantation produces no macroscopic change in the workpiece. From the point of view of applications, though several plasmonic nanocomposites with embedded noble metal clusters of tailored compositions and even of core-shell geometries could be fabricated extending their SPR peak position to a broader spectral range (in comparison with monometallic nanocomposites), tuning the relative thickness of the core and shell layer has not been sufficiently demonstrated. As has been shown recently (Peña-Rodríguez & Pal, 2010, application potential of plasmonic nanostructures depends on their enhanced SPR extinction, along with a precise control over its peak position. For this purpose, it is possible that in the near future we will need to fabricate multi-layer metallic nanoclusters with precise control over the thickness of each layer. Though the modern ion implanters are capable of introducing virtually any element in a solid matrix with a wide variation of energy and implantation dose, there has been no attempt to fabricate nanocomposites with trimetallic nanoclusters incorporated into dielectric matrix. On the other hand, fewer efforts have been devoted to control the shell-thickness in a core-shell type nanocluster or the fabrication of multilayer metallic nanoclusters. However, very recently this technique has been utilized for the fabrication of unusual composite nanostructures with distinct plasmonic behaviors, which are highly attractive for several applications from optical sensing to biological treatments. Considering the current interest on plasmonic nanostructures and nanocomposites and the demonstrated potential of available commercial ion implanters, utilization of this technique for fabricating new exotic nanostructures in the near future is very much expected.