Fundamentals of the Ion Implantation Process

Figure 1 shows a schematic view of the path of an individual ion as it loses energy in a material, thereby forming a shallow surface-modified region. As indicated in the figure, the ion does not travel in a straight path to its resting place, due to collisions with the target atoms. Target atoms are displaced from their lattice sites with sufficient energy that they can themselves displace additional target atoms, resulting in a collision cascade. These individual collisions with lattice atoms within a single collision cascade are shown in the insert at the bottom of Fig. 1.

Fig. 1 Schematic view of ion implantation process (top) and depiction of the interactions with substrate atoms in a single collision cascade (bottom). Source: Ref 6

The actual integrated distance traveled by the ion is called the range, R. The net penetration of the ion into the material, measured as projected onto the original trajectory, is called the projected range. Ion implantation is a random process, normally producing a Gaussian-shaped impurity depth profile whose centroid is defined as the average projected range, Rp. The statistical nature of the scattering process gives rise to a distribution of ions around the projected range depth. The average fluctuation (standard deviation) in the projected range is called the range straggling, ARp.

The ion energy directly affects both the range and the distribution of the implanted ions in a given substrate. At higher energies a greater spread in the ion distribution (a greater ARp) is realized for a given dose, which is the number of implanted ions per unit area of the surface of the material (areal density), usually expressed as ions/cm2. The areal dose (also called fluence) is a convenient unit of ion implantation because the actual volume concentration associated with an implantation is a function of ion species, ion energy, and substrate material. For metals, beneficial doses can span the range of 1015 to 1018 ions/cm2, depending on the application (see the section "Applications" in this article). The determination of the implanted atomic concentration (atoms/cm3) requires relating the areal density of implanted atoms to their spatial extent. At low ion doses, the implant profile can normally be approximated as a Gaussian distribution centered about the projected range. Accordingly, the width of the distribution can be expressed by the standard deviation of the Gaussian distribution, ARp. The expression for the peak concentration (Np) of a Gaussian distribution is given in terms of the applied dose (NA) and range straggling as:

Figure 2 shows the projected range and range straggling for nitrogen ions implanted into iron versus the initial ion energy. The range distribution is shown as a Gaussian distribution with Rp and ARp characteristic of the ion energy. For a given energy, a lighter ion such as nitrogen will penetrate farther and will undergo more large-angle scattering (leading to a broader distribution) than would a heavier ion such as chromium. Each implanted ion can displace hundreds to thousands of lattice atoms as it travels into the substrate surface. This results in a net damage distribution that is also normally Gaussian-shaped but is situated closer to the surface than the range distribution, because each ion creates damage between the surface and its final resting place in the lattice. This lattice damage can render ionic or covalently bonded lattices amorphous during implantation, whereas a nondirectionally bonded (metal) substrate can either self-anneal, with no residual damage, or else can retain point defects or extended defects (e.g., dislocations) resembling those of a heavily work-hardened metal or alloy. Ion implantation of certain species (e.g., phosphorus in iron) can stabilize amorphous structures in metals in a manner analogous to bulk rapid quench techniques such as splat cooling.

Fig. 2 The projected range, Rp, and range straggling, ARp, of nitrogen ions implanted into iron vs. the initial ion energy. Source: Ref 6

As an ion penetrates a material, there is a certain probability that a surface atom will be ejected from its lattice site, because there is some momentum directed back toward the surface in a collision cascade that allows an atom to receive sufficient energy to overcome its surface-binding energy. This phenomenon is called sputtering and is analogous to the erosion of materials by the impact of high-velocity particles. The ratio of the number of substrate atoms ejected per incident ion is commonly termed the sputtering coefficient, S. As a general rule, for a given substrate material, S will increase with increasing ion mass and will increase sharply at more oblique angles of incidence. For a given ion at normal incidence, S depends principally on the surface binding energy of the material, which can be related to its heat of sublimation. Therefore, for a given ion, S will decrease with increasing heat of sublimation for the substrate material. Values for S can range from less than 1 for the case of a light ion incident on a heavy substrate (e.g., nitrogen implanted into iron) to greater than 10 for very heavy ions in a lighter substrate (e.g., tantalum ions implanted into iron).

Figure 3(a) shows a schematic view of the evolution of the concentration depth profile of an implanted element for the case where S is greater than unity. In this particular example, the projected range of ions is shown as being 600 A. This closely represents the implantation of 200 keV chromium ions into steel (or iron). At a low dose, the distribution is essentially Gaussian, as discussed earlier, and there is a low concentration of implanted atoms at the surface. However, as the dose increases and the surface is eroded, an increasing number of previously implanted atoms are exposed on the surface, and these near-surface atoms are subject to sputtering just as the target atoms are.

Fig. 3 (a) Schematic view of the development of implanted impurity profiles from low to high doses. Source: Ref 6. (b) Computer simulation of four consecutive 5 x 1016 Cr/cm2 implantations at 200 keV energy into iron, accounting for sputter erosion of the surface for each implantation. The sum of the four individual profiles (the curve of dark squares) yields a distribution with a maximum near the receded surface. Source: Ref 7

Fig. 3 (a) Schematic view of the development of implanted impurity profiles from low to high doses. Source: Ref 6. (b) Computer simulation of four consecutive 5 x 1016 Cr/cm2 implantations at 200 keV energy into iron, accounting for sputter erosion of the surface for each implantation. The sum of the four individual profiles (the curve of dark squares) yields a distribution with a maximum near the receded surface. Source: Ref 7

At a particular dose level, called the saturation dose, a steady-state situation is established whereby the rate of removal of previously implanted atoms is equal to the arrival rate of energetic ions. The peak concentration of the implanted element distribution approaches the surface, as shown in Fig. 3(b), as the implanted dose increases and approaches the saturation dose. Generally, the maximum surface concentration of the surface element is equal to 1/(S + 1). For the example of 200 keV chromium ions in iron, S has a value of about 5 atoms/ion, so the maximum surface concentration of chromium would be about A or 16 at.%. In this case, the steady-state dose would be approximately 2 * 1017 ions/cm2. This behavior is quantitatively predicted by analytic models, as shown in Fig. 3(b), which illustrates the sputter removal of the surface and the surface enrichment of the implanted species resulting from successive (lower-dose) implantations.

For certain ion-substrate combinations and concentrations, chemical reactions with molecular species present in the vacuum chamber can be induced. An excellent example of this is titanium implantation into steel. At high titanium doses, highly reactive titanium atoms become exposed on the surface and react with the residual CO and CO2 hydrocarbon molecules normally present in the vacuum chamber. When the partial pressure of carbon-containing molecules is 1 * 10-6 torr or higher, the implantation results in the formation of an amorphous Fe-Ti-C surface layer that has been shown to reduce friction and wear greatly (see the section "Applications" in this article). When the background pressure in the chamber is less than 10-6 torr, there are insufficient carbon-containing molecules to produce the carburized surface layer. A consequence of these surface chemical reactions is a considerably reduced sputtering yield because of the increase in the surface binding energy due to compound formation. This phenomenon has been exploited by deliberately introducing gases into the vacuum chamber during implantation. For example, for tantalum implantation into steel, where normally S is approximately 10, the introduction of oxygen into the chamber to 1 * 10-5 torr (partial pressure) reduces the sputtering by more than a factor of three, due to the formation of an almost pure Ta2O5 layer at the surface for a tantalum dose of 2 * 1017 ions/cm2.

0 0

Post a comment