When John Poate joined Bell Labs in 1971 after a graduate career in nuclear physics he declared: “No more nuclear physics for me, I am going to exploit the field of ion beams and solids.” Indeed he did and that’s why we are here today. This talk will trace a little of John’s career at Bell Labs as observed by a close colleague over a span of 26 years. Erbium ion implantation: from planar amplifiers to nanophotonics.

As CMOS devices continue to scale to ever-smaller dimensions the energy of the implants continue to decrease while the doses increase. This coupled with increasing wafer size has driven the throughput of implant tools to an unacceptable level. Molecular implantation of large boron containing molecules offers an interesting solution to this ever-worsening problem. A new implant material coupled with a new style ion source has recently proven to be a promising answer to this dilemma. An electron impact ionization source used to ionize B₁₈H₂₂ has been shown to effectively deliver more than 20 mA of B to the wafer at sub-keV energies without the use of deceleration. The quality of the beam from this plasma-free ion source results in tighter control of the angular divergence of the beam while delivering an order of magnitude more beam current than conventional ion sources. Auto amorphization with these large clusters can eliminate the need for pre-amorphizing implants, thus saving a process step and reducing leakage due to the end-of-range defects created by the pre-amorphizing implants. Device performance characteristics have been measured utilizing split-lot testing with B⁺ and B₁₈H₂₂⁺ implants for source drain extension doping and show no deviation.

Ion beams of kilovolt and megavolt energies have ranges in solids reaching up to tens of microns, yet such ion beams can be used to make, modify, and study materials on a nanometer length scale. Examples are the formation of nano-cavities in Si following He implantation and how the thermodynamics of curved surfaces influences impurity gettering, and the transformation of spherical gold nano-particles into an aligned array of nano-rods which exhibit optical properties dependent on the relative orientation of the deformed particles and the direction and polarisation of the incident light.

The optical properties of erbium-doped silicon-rich oxides were examined to assess their potential for fabricating an integrated optical amplifier. Samples were prepared by co-implantation of Si and Er into SiO2 followed by a thermal anneal at temperatures in the range from 800 to 1100 oC. The active fraction of erbium, the effective erbium excitation cross-section and the 1.5 um luminescence lifetime were studied as a function of annealing temperature and before and after hydrogen passivation. Optical pump-probe measurements were also performed by prism-coupling a 1.54 μm probe beam into a waveguide formed by the silicon-rich oxide and monitoring its intensity and temporal response as the waveguide was optically pumped from above with a chopped 477 nm excitation source. Induced-absorption (losses) of the 1.54 μm probe beam in erbium-doped and un-doped silicon-rich silicon oxide waveguides was observed in all cases. For the samples containing only well defined nanocrystals, a fast (~60 μs) induced absorption component associated with free carriers within the silicon nanocrystals is reported, whilst for samples containing defective nanocrystals or nanoclusters, a much slower (> 10 minutes) component is observed. The free carrier absorption is shown to be reduced by delaying the probe beam relative to the pump beam in cases where it dominates.

This talk will review some of the most recent advances in the control of defects and impurities for the fabrication of novel silicon devices. Two main examples will be given. In the first one the interaction of self-interstitials with boron atoms will be investigated. It is shown that a strong boron diffusion, even at room temperature, is observed. The role of F in the interaction with point defects for the suppression of B diffusion and the production of ultra-shallow junctions will be elucidated. In the second example the role of rare earth ions in the fabrication of room temperature operating light emitting devices is presented. The limiting problems towards practical applications and possible routes to solve them will be discussed.

Work in the early 1990 s quantitatively coupled the plus-one interstitial, extended defects visible in the TEM, and the Transient Enhanced Diffusion of Boron. A series of experiments demonstrated that the initial interstitial content seen in defect size was consistent with the plus-one model, and the interstitial evaporation rate was consistent with the supersaturation seen in TED. This simple story became more complicated with the inclusion of impurities (C, O, F), and at very high or very low energies, with defect clustering, the bad dose regime, near-surface recombination, and vacancy clusters, not to mention BICs (and I won t). This talk will show how the original underlying physics connects the basic results all the way through to current commercial implantation techniques including cocktail impurity implants and TED following laser-surface anneal activation.

The fundamental physics of pulsed laser melting were explored extensively in the 1980's, driven by Si junction annealing requirements. While RTA, spike and flash annealing have so far addressed these requirements for industry, the continued push for shallow and abrupt junctions leads once again to consideration of melt annealing. The issues today include many of the same fundamental questions, but must be addressed in an inhomogeneous space including Si-Ge and Si-C alloys, strain-field interactions, high-K gate dielectrics, poly gates and, potentially, metal gates. In addition, technically critical issues such as pattern dependencies must be resolved. In this talk, I will discuss our current work on these new materials and annealing methods, and contrasting the interactions with the pioneering work of those early days.

The early research on laser annealing of ion-implanted silicon was unprecedented in its fervor and intensity (at least until the days of high Tc superconductors). The physical understanding gained from this work led to the development of Rapid Thermal Annealing (RTA), and accelerated the development of Silicon on Insulator (SOI) in its various forms. Laser Annealing Symposia helped transform the MRS into a major scientific organization. John Poate was at the forefront of all these activities. As I was fortunate to interact with John for over 25 years, many of my projects were influenced by his insight and vision. Some examples will be given.

Multilayer interconnects for silicon devices require thin films with uniform thickness and full density. Unfortunately, most deposition procedures produce films containing voids, pinholes, three dimensional islands, facets, breadloafing at the edges of vias and trenches, or other undesirable features. We describe kinetic Monte Carlo models that exhibit most of these features, and can therefore help in the selection of conditions that mitigate these effects.

The successful growth of single-crystal epitaxial silicide-silicon interfaces two decades ago(1) provided an unprecedented opportunity to compare, on atomicallycontrolled and homogeneously structured interfaces, the predictions of theories on the formation mechanism of Schottky barrier height with experiment. The surprisingobservation of a strong dependence of the Schottky barrier height on the interface structure and the subsequent demonstration of the prevalence of Schottky barrierheight inhomogeneity fundamentally reshaped our views on the formation of interface dipole at material interfaces. Here we briefly examine our present understanding ofthe formation of the Schottky barrier height, and apply the same concepts to describe band alignment at other interfaces of interest, such as the high-K dielectric and the metal-organic interfaces. (1). R. T. Tung, J. M. Gibson and J. M. Poate, Phys. Rev. Lett. 50, 429 (1983).

Islanding for epitaxial systems with similar materials and relatively high lattice mismatch usually occurs by the two stage Stranski-Krastanow (S-K) growth transition. As is well known, a flat 'wetting layer' forms first but, at a particular point, islands are produced and dominate the subsequent growth. In this presentation, the way in which surface segregation in the initial 'wetting layer' can straight-forwardly account for the island nucleation point will be described. Quantitative calculations for the InGaAs/GaAs system will be presented and these demonstrate the close agreement with experiment that can be achieved. Extension of the model to the islanding transition in the SiGe/Si system will be discussed. Overall, the surface changes which occur during the S-K transition will be considered and the likely behaviour of adatom number density at the transition point will be highlighted.