In this paper, we investigate the evolution of extended defects during a msec Flash anneal after a PAI implant and show that during the ultra-fast temperature ramp-up and ramp-down, the basic mechanisms that control the evolution of defects are not modified with respect to the relatively slower RTA anneals. In addition, we show that junction depths below 15 nm can be achieved using a US J fabrication scheme based on the combination of impurity co-implantation with msec Flash anneal.

This paper demonstrates the possibility to use a bottom-up front-end (FE) architecture for sub-32 nm CMOS nodes with a new 3D approach in front-end flow. This architecture based on the so called FRETCH (film replacement etching through contact hole) solution [1] is compatible to adapt a specific device before or after electrical test. To address the full 3D in FE concept of buried integration will be proposed by using e-beam and HSQ material. Finally a novel integration scheme fully compatible with conventional bulk flow with one mask will be depicted allowing a self assembly structure and reducing significantly the process variability.

Laser annealing can be used for applications that require confinement of heat on the top surface in order to preserve the integrity of buried structures. To confine heat near the silicon surface, the most appropriate tool is UV laser. Here, two excimer lasers with different characteristics were used to anneal boron in silicon. The first one has pulse duration of about 180 ns and the second one of about 20 ns in scanning mode. Regarding the crystal damage characterization, we report that room temperature photoluminescence (RT-PL) is a powerful and non-destructive technique to monitor crystal damage evolution. In order to investigate the effect of a surface oxide on laser annealing, samples with different (1, 20 and 150 nm-thick) surface oxide layers have been annealed. From RT-PL data, the crystal quality is found to improve with increasing laser power density. However the presence of the oxide layer leads to sudden degradation of the PL response for laser energy densities exceeding the melting threshold of Si.

This work combines plasma doping implantation (PLAD) with laser annealing using excimer laser, for the formation of ultra-shallow junctions. For that purpose, high dose BF3 was implanted in n-type silicon wafers using PLAD. The as implanted material was investigated by high resolution TEM, measured by SIMS and simulated by Monte-Carlo codes. Subsequently, the samples were annealed using a KrF laser source at 248 nm with a pulse duration 20 ns and different fluence values. Laser annealing completely recrystallizes the amorphous layer as monitored by TEM measurements, fully activates the dopants achieving low sheet resistance values as shown by Van der Pauw measurements and results in box-shaped dopant profiles with movement less than 10 nm at the junction depth as measured using SIMS.

Advanced devices may today require implantation and annealing steps after the metallic interconnection realization. Depending on the application, a thin p-doped layer has to be formed after wafer bonding. The issue, in such a case, is to correctly anneal the Boron implanted layer without degrading the buried devices and interconnections which lies at a depth around 3mum below the surface. Here, the authors propose to study different way to anneal this thin p-doped layer. Low energy and low dose implantations are performed without reaching the amorphisation threshold. Long thermal annealing at 400degC (RTP) and UV laser annealing are investigated through sheet resistance, thermal wave, SIMS or TEM. On one hand, a significant activation is obtained with RTP at temperature as low as 400degC and that Boron is activated with a better activation rate with B+ than with BF2 +. On the other hand, a much better activation was achieved with laser annealing as compared to RTP regardless of the implantation conditions

The fully depleted SOI devices present lateral isolation issues due to the shallow trench isolation (STI) process. We propose in this paper to study a new fabrication process for integrating local isolation trenches. Germanium (Ge) implantation is used to create SiGe (Silicon-Germanium) layer on thin SOI (silicon on insulator) that can be selectively etched. The advantage is the capability of implantation to localize the SiGe area on this substrate and to avoid STI process issues. Aggressive dimensions and geometries are studied and resulting material transformation (crystallization and Ge diffusion) are apprehending via SEM (Secondary Electron Microscopy) or AFM (Atomic Force Spectroscopy) to understand the etching kinetics. After optimization, we demonstrate the capability of fabricating localized trenches on SOI without degrading the neighboring Si layer or consuming the thin BOX (buried oxide).

Standard gate materials are compared to Ge implanted poly-Si and deposited poly-SiGe. It is demonstrated in this paper that the electrical resistance of the gate is significantly reduced via the use of poly-SiGe (from 30% to 40% decrease in resistance). Similarly, we show via specific optimization that localized Ge implantation is also suitable to reduce gate resistance. Physical characterizations are performed to determine the “root” causes at the origin of these improvements. In line with future publications showing strong benefits on CMOS device performance, grain size effects seem to be the main mechanisms explaining the measured improvement.

We present results of simulations on silicon-on-nothing (SON) devices featuring ultra-thin Si channels. The simulations are based on QUANTIX, a 2D-Finite Element solver, and consequently fully account for quantization effects. Quantum simulations demonstrate the impact of carriers confinement in very thin Si conduction channel, therefore leading on a substantial threshold voltage increase, up to 350 mV from TSi=10 to 1 nm. Experimental data from SON devices featuring ultra-thin (<2 nm) Si channel and a 50 nm gate length are also used to illustrate this study.

Les recherches toxicologiques dans les urines des nouveau-nés de mère toxicomanes peuvent être prises en défaut. Le but de ce travail est de vérifier que l'étude conjointe du méconium et des cheveux permet d'améliorer la sensibilité de ces analyses.

Urinary detection of prenatal drug exposure in the neonate may give false-negative results. We report our experience on meconium and hair testing, in addition to urine testing in order to improve diagnosis of fetal drug exposure.Thirty-one infants (aged 1-45 days) whose mothers were confirmed (n = 12) or suspected (n = 19) to be drug-addicted were included in the study. One or more specimens of urine, meconium or hair were collected in the 31 infants, two of the specimens in 17 and three in six. Drugs and their metabolites were detected by immunoenzymologic techniques and positive results were confirmed by gas-exchange chromatography. All the mothers and families were interviewed during admission and the information was compared to those provided by medical and social services; the results of laboratory analysis were not known by the investigators at this time of the study.The maternal drug addiction was confirmed after clinical investigation in 18 cases including the 12 cases detected by prenatal interview (group 1), and recused in 13 other cases (group 2). In group 1, nine infants of 12 had a positive urine test (seven opiate, one cocaine, one cannabis), 11 of 11 a positive meconium test (nine opiate, one cocaine, one cannabis), ten of 19 a positive hair test (eight opiate, one cocaine, one cannabis); all infants in this group had at least one positive result. In group 2, all tests were negative except one urine test positive for opiate after cesarean delivery performed under anesthesia including opiate analgesia.Urine, meconium and hair testing versus urine testing alone increase the sensitivity of laboratory analysis for detection of prenatal drug exposure.

Background.- Urinary detection of prenatal drug exposure in the neonate may give false-negative results. We report our experience on meconium and hair testing, in addition to urine testing in order to improve diagnosis of fetal drug exposure.Population and methods.- Thirty-one infants (aged 1-45 days) whose mothers were confirmed (n = 12) or suspected (n = 19) to be drug-addicted were included in the study. One or more specimens of urine, meconium or hair were collected in the 31 infants, two of the specimens in 17 and three in six. Drugs and their metabolites were detected by immunoenzymologic techniques and positive results were confirmed by gas-exchange chromatography. All the mothers and families were interviewed during admission and the information was compared to those provided by medical and social services; the results of laboratory analysis were not known by the investigators at this time of the study.Results.- The maternal drug addiction was confirmed after clinical investigation in 18 cases including the 12 cases detected by prenatal interview (group 1), and recused in 13 other cases (group 2). In group 1, nine infants of 12 had a positive urine test (seven opiate, one cocaine, one cannabis), 11 of 11 a positive meconium test (nine opiate, one cocaine, one cannabis), ten of 19 a positive hair test (eight opiate, one cocaine, one cannabis); all infants in this group had at least one positive result. In group 2, all tests were negative except one urine test positive for opiate after cesarean delivery performed under anesthesia including opiate analogsia.Conclusions.- Urine, meconium and hair testing versus urine testing alone increase the sensitivity of laboratory analysis for detection of prenatal drug exposure.

Biomimetic syntheses of enantiomerically pure 3-deoxy-d-muco- and d-myo-inositol-1,4,5 trisphosphate from d-glucose are described. Preliminary biological studies show a dramatic influence of the stereochemistry at C-4 on the binding to IP3 receptor.