Abstract: Interaction of ion beams with solids

Publisher Website
Google Scholar
Abstract
Ion implantation is now a widely used technique for modifying the chemical and physical properties of a shallow surface region of a solid. In principle, almost any combination of ion (Z1) and target (Z2) can be used over a wide concentration range, and with independent and precise control of the number and the depth distribution of the implanted ions. Furthermore, with modern vacuum techniques, ion implantation is an ideal ’’clean room’’ process, an advantage which has not yet been fully exploited. We will review briefly the major physical processes involved in implantation: viz., (i) the number and depth distribution (range) of the implanted ions, with particular emphasis on some of the complicating factors, such as ion reflection, charge exchange (in the beam), channeling, enhanced diffusion effects, and trapping at interfaces, which must be either eliminated or taken into account; (ii) the depth and nature of the radiation damage produced by the implantation; and (iii) the ultimate lattice location of the implanted ions within the crystal structure of the host lattice. The radiation damage problem will then be considered in more detail, as this is generally the most important and also the most complicated aspect of ion bombardment. In particular, some recent work on the role of molecular ion beams will be presented in which equal atom−dose implants of various diatomic and monatomic ions were injected into Si and Ge at room temperature. In each case, the molecular ion implant had the same energy per atom and the same atomic flux and fluence as the corresponding monatomic implant. Yet, the molecular beam (As+2, Sb+2, Bi+2) produced ∠50% more damage than the monatomic beam, indicating that damage production depends not only on the total amount but also on the localized concentration of deposited energy. For the highest density cascades studied, the observed amount (Nd) of damage per ion corresponds to almost 100% of the predicted cascade volume; furthermore, there is no evidence of the molecular effect levelling off as the damage approaches this 100% limit. Also, the Nd values in semiconductor crystals are all considerably greater than the upper limit of 0.8 ν (E)/2Ed predicted by the Kinchin−Pease equation, 1 and the discrepancy becomes even worse for similar implants at low (30 K) temperature. These results provide strong experimental evidence for a breakdown of the generally accepted linear−cascade model, and suggest that at fairly low energies some sort of energy−spike model (as recently proposed by Sigmund 2) may indeed be more appropriate.

This publication has 0 references indexed in Scilit: