Chihak Ahn

Extended Point Defects in Crystalline Materials: Ge and Si

B diffusion measurements are used to probe the basic nature of self-interstitial point defects in Ge. We find two distinct self-interstitial forms--a simple one with low entropy and a complex one with entropy ∼30  k at the migration saddle point. The latter dominates diffusion at high temperature. We propose that its structure is similar to that of an amorphous pocket--we name it a morph. Computational modeling suggests that morphs exist in both self-interstitial and vacancylike forms, and are crucial for diffusion and defect dynamics in Ge, Si, and probably many other crystalline solids.

Calculations of effect of anisotropic stress/strain on dopant diffusion in silicon under equilibrium and nonequilibrium conditions

Understanding changes in dopant diffusion under strain is critical for controlling junction profiles in current and future very large scale integrated technology due to expanding use of large strains to enhance channel mobility. We use density functional theory calculations to investigate the stress dependence of boron (B) and arsenic (As) diffusion including vacancy (V) and interstitial (I) mechanisms under arbitrary stress/strain states. We have also analyzed the effects of stress on I and V diffusion with resulting impact on transient enhanced diffusion and coupled diffusion. For B diffusion, which is primarily mediated by I, we find greatly enhanced diffusion under tensile stress. Due to low symmetry of calculated transition state, we predict strongly anisotropic diffusion under anisotropic strain, with the strongest effects in direction of strain. This has a major impact on control of lateral junction abruptness as seen in two-dimensional simulations. The predicted behavior is consistent with combine...

First principles calculations of dopant solubility based on strain compensation and direct binding between dopants and group IV impurities

We investigated binding between dopant atoms such as boron and arsenic and various elements in group IV (e.g., C, Ge, Sn, and Pb) to explore opportunities for increasing dopant solubility, which is becoming critical for nanoscale semiconductor technology. Using first principles calculations, we find the dominant component of binding to be global strain compensation. We find negligible direct local binding between B and Ge, in contrast to some suggestions in the literature. Considering strain compensation and negative deviation from Vegard’s law of lattice parameter for SiGe, we predict the enhancement of boron segregation ratio across epitaxial Si∕SiGe interfaces, which agrees well with previous experimental observations. Due to nearest neighbor binding plus substantial strain compensation, Sn may have some promise for enhancing B solubility. For C∕As, the first nearest neighbor interaction is repulsive. However, the large negative induced strain due to carbon overcompensates this effect in the solubility...

Transfer of physically-based models from process to device simulations: Application to advanced Strained Si/SiGe MOSFETs

Integrated process and device simulations were used to predict sub-45nm Strained-Si/Si0.8Ge0.2 device performance. Physically-based process models, generalized from Si to strained-Si and SiGe, describe dopant implantation and diffusion, including amorphization, defect interactions and evolution, as well as dopant-defect interactions. The models are used within a unique simulation tool to reproduce the electrical characteristics of Si/SiGe devices.

Charge carrier induced lattice strain and stress effects on As activation in Si

We studied lattice expansion coefficient due to As using density functional theory with particular attention to separating the impact of electrons and ions. Based on As deactivation mechanism under equilibrium conditions, the effect of stress on As activation is predicted. We find that biaxial stress results in minimal impact on As activation, which is consistent with experimental observations by Sugii et al. [J. Appl. Phys. 96, 261 (2004)] and Bennett et al. [J. Vac. Sci. Technol. B 26, 391 (2008)].

Modeling of Defect Evolution and TED under Stress based on DFT Calculations

The incorporation of strain in order to improve mobility has become an important element in CMOS device scaling. In this work, we have developed a new moment-based model of extended defect kinetics and further studied the impact due to stress on the energies of impurities, point defects and particularly extended defects. We specifically look at point defect clusters which control transient enhanced diffusion (TED). The results enable comprehensive models for dependence of nanoscale device structures on stress which can be used for process optimization

Review of Stress Effects on Dopant Solubility in Silicon and Silicon-Germanium Layers

We present a review of both theoretical and experimental studies of stress effects on the solubility of dopants in silicon and silicon-germanium materials. Critical errors and limitations in early theory are discussed, and a recent treatment incorporating charge carrier induced lattice strain and correct statistics is presented. Considering all contributing effects, the strain compensation energy is the primary contribution to solubility enhancement in both silicon and silicon-germanium for dopants of technological interest. An exception is the case of low-solubility dopants, where a Fermi level contribution is also found. Explicit calculations for a range of dopant impurities in Si are presented that agree closely with experimental findings for As, Sb and B in strained Si. The theoretical treatment is also applied to account for stress effects in strained SiGe structures, which also show close correlation with recently derived experimental results for B-doped strained SiGe which are presented here for the first time.

Predictive Model for B Diffusion in Strained SiGe Based on Atomistic Calculations

Using an extensive series of first principles calculations, we have developed general models for the change in energy of boron migration state via interstitial mechanism as a function of local alloy configuration. The model is based on consideration of both global strain compensation as well as local effects due to nearby arrangement of Ge atoms. We have performed KLMC (Kinetic Lattice Monte Carlo) simulations based on change in migration energy to explain the reduced B diffusion in strained SiGe and compared our results to experimental observations. These models include significant effects due to both global stress and local chemical effects, and accurately predict the B diffusivity measured experimentally in strained SiGe on Si as a function of Ge content.

Comment on "Diffusion of n-type dopants in germanium" (Appl. Phys. Rev. 1, 011301 (2014))

The authors of the above paper call into question recent evidence on the properties of self-interstitials, I, in Ge [Cowern et al., Phys. Rev. Lett. 110, 155501 (2013)]. We show that this judgment stems from invalid model assumptions during analysis of data on B marker-layer diffusion during proton irradiation, and that a corrected analysis fully supports the reported evidence. As previously stated, I-mediated self-diffusion in Ge exhibits two distinct regimes of temperature, T: high-T, dominated by amorphous-like mono-interstitial clusters—i-morphs—with self-diffusion entropy ≈30 k, and low-T, where transport is dominated by simple self-interstitials. In a transitional range centered on 475 °C both mechanisms contribute. The experimental I migration energy of 1.84 ± 0.26 eV reported by the Munster group based on measurements of self-diffusion during irradiation at 550 °C < T < 680 °C further establishes our proposed i-morph mechanism.

Atomistic modeling of dopant diffusion and segregation in strained SiGeC

In this work, density functional theory calculations are used to calculate the separate effects of stress/strain and chemical binding on diffusion, segregation and solubility of dopants in group IV alloy materials. Kinetic lattice Monte Carlo calculations are used to extract the effects of anisotropic stress and random alloy distributions. We find that segregation and solubility is dominated by stress effects, but that chemical interactions of Ge and C with point defects have a significant effect on diffusivity in SiGeC alloys.