T. Y. Tan

Point defects, diffusion processes, and swirl defect formation in silicon

The paper consists of three parts. In the first part we review the basic experimental and theoretical results which shaped our present knowledge on point defects and diffusion processes in silicon. These results concern on one side oxidation effects which established that silicon self-interstitials and vacancies coexist in silicon and on the other side diffusion of gold into dislocation-free silicon which allowed to determine the self-interstitial contribution to silicon self-diffusion and to estimate the corresponding vacancy contribution. In the second part we discuss topics for which an understanding is just emerging within the framework of coexisting self-interstitials and vacancies: reaching of local dynamical equilibrium between self-interstitials and vacancies; rough estimates of the thermal equilibrium concentrations of self-interstitials and vacancies and their respective diffusivities, and finally, various possibilities to generate an undersaturation of self-interstitials. In the third part we examine swirl defect formation in silicon in terms of vacancies and self-interstitials.

Intrinsic gettering by oxide precipitate induced dislocations in Czochralski Si

Conditions for effective intrinsic gettering by oxide precipitate induced dislocations, which we suggest as an important mechanism in explaining device leakage limited yield enhancement due to oxygen in Czochralski‐grown Si wafers, are examined. The effectiveness of this mechanism is demonstrated.

Silicon nanowhiskers grown on 〈111〉Si substrates by molecular-beam epitaxy

Silicon nanowhiskers in the diameter range of 70 to 200 nm were grown on 〈111〉-oriented silicon substrates by molecular-beam epitaxy. Assuming the so-called “vapor–liquid–solid” (VLS) growth process to operate, we initiated the growth by using small clusters of gold at the silicon interface as seeds. The in situ generation of the Au clusters as well as the growth parameters of the whiskers are discussed. The experimentally observed radius dependence of the growth velocity of the nanowhiskers is opposite to what is known for VLS growth based on chemical vapor deposition and can be explained by an ad-atom diffusion on the surface of the whiskers.

A Model for the Silicon Wafer Bonding Process

The bonding speed (or contact wave velocity) of silicon and fused quartz wafers has been measured as a function of temperature. The results show that the bonding process stops to operate at temperatures above 90°C and 320°C for fused quartz and bare silicon wafers, respectively. By comparing our results to infrared spectra obtained from silica gel we develop a tentative model of the bonding process. This model is based on the assumption that the initial wafer bonding process occurs via hydrogen bonds of adsorbed water. This model explains why the bonding strength increases in two distinct steps during high temperature annealing. By introducing a phenomenological time constant τ we can also account for the fact that in an intermediate temperature range the bonding strength does not depend on annealing time as it has been reported in the literature.

Oxygen diffusion and thermal donor formation in silicon

The information available on the diffusion of oxygen and on the formation of thermal donors in silicon is critically reviewed. In this context the effects of intrinsic point defects on the diffusion-controlled growth of oxygen precipitates is investigated in some detail. Seemingly contradictory experimental results on the diffusivity of oxygen in silicon at temperatures around 400° C are explained in terms offast-diffusing gas-like molecular oxygen in silicon. The concept of molecular oxygen is also invoked in a newly suggested model of thermal donor formation in silicon. The diffusivity of molecular oxygen in silicon is estimated to be around 10−9cm2s−1 at 450° C, almost nine orders of magnitude higher than the diffusivity of atomic oxygen in interstitial position.

A “smarter-cut” approach to low temperature silicon layer transfer

Silicon wafers were first implanted at room temperature by B+ with 5.0×1012 to 5.0×1015 ions/ cm2 at 180 keV, and subsequently implanted by H2+ with 5.0×1016 ions/cm2 at an energy which locates the H-peak concentration in the silicon wafers at the same position as that of the implanted boron peak. Compared to the H-only implanted samples, the temperature for a B+H coimplanted silicon layer to split from its substrate after wafer bonding during a heat treatment for a given time is reduced significantly. Further reduction of the splitting temperature is accomplished by appropriate prebonding annealing of the B+H coimplanted wafers. Combination of these two effects allows the transfer of a silicon layer from a silicon wafer onto a severely thermally mismatched substrate such as quartz at a temperature as low as 200 °C.

Ion-induced defects in semiconductors

Abstract The status of our knowledge of ion-induced defects in semiconductors will be reviewed, including the charge-state dependence of defects, novel defect migration mechanism and enhanced damage production mechanisms. The main emphasis will be on defects in silicon where a panorama of defects is emerging which encompasses the evolution of damage from vacancies and interstitials and their aggregates to stacking faults and dislocations to disordered zones and the development of an amorphous layer.

Mechanisms of doping‐enhanced superlattice disordering and of gallium self‐diffusion in GaAs

Recently available Ga‐Al interdiffusion results in GaAs/AlAs superlattices allow us to conclude that Ga self‐diffusion in GaAs is carried by triply negatively charged Ga vacancies under intrinsic and n‐doping conditions. The mechanism of the Si‐enhanced superlattice disordering is the Fermi‐level effect which increases the concentrations of the charged point defect species. For the effect of the p dopants Be and Zn, the Fermi‐level effect has to be considered together with dopant diffusion induced Ga self‐interstitial supersaturation or undersaturation. Self‐diffusion of Ga in GaAs under heavy p‐doping conditions is governed by positively charged Ga self‐interstitials.

Oxygen precipitation and the generation of dislocations in silicon

Abstract Oxygen-rich precipitates in silicon, and the generation of dislocations at the interfaces between the precipitates and the matrix were observed. The precipitates are square-shaped plates with 〈110〉 sides and {100} habit planes. Prismatic dislocation loops are generated in silicon by the mechanism of prismatic punching. The loops are identified as interstitial loops with ½ 〈110〉 Burgers vectors and 〈110〉 axis. An ideal loop is rhombus-shaped with line senses in 〈112〉 directions. Nucleation of loops follows the mechanism of Ashby and Johnson (1969): a shear loop is nucleated on an initial slip plane and completes a rhombus-shaped prismatic loop by repeated cross-slips. To our knowledge, the loop nucleation process is identified for the first time for a system completely under internal stress.

Diffusion mechanism of zinc and beryllium in gallium arsenide

The outstanding features associated with Zn and Be diffusion in GaAs substrates and GaAs/AlGaAs superlattices are explained either quantitatively or semiquantitatively using the kick‐out mechanism, in which it is assumed that the doubly positively charged Ga self‐interstitial governs Ga self‐diffusion. These features include (i) the dependence of the Zn solubility upon the pressures of the As and Zn vapor phases, (ii) the square power‐law dependence of the Zn diffusivity on its own background concentrations under Zn isoconcentration diffusion conditions, (iii) the different shapes of the Zn in‐diffusion profiles, (iv) the much lower diffusivities of Zn and Be under out‐diffusion conditions than under in‐diffusion conditions, and (v) the tremendous enhancement effect of Zn in‐diffusion on GaAs/AlGaAs superlattice disordering and the undetectable effect of Be under out‐diffusion conditions. Some useful quantitative information has been obtained. Strictly on a qualitative basis, we have found that the Longin...