Semiconductor doping is a critical step in microelectronic device fabrication. Particularly, ultra-shallow junction formation for the CMOS 45-nm node is today intensively studied. Laser thermal processing (LTP) has already shown potentiality to achieve abrupt and ultra-shallow junctions, with a very low resistivity. However, the laser process has to be integrated in the conventional process flow of a real CMOS device fabrication. Therefore, the laser treatment needs to preserve the integrity of the different irradiated structures like transistor gates. Optical coatings, including reflective and anti-reflective coatings, can be used to protect the structures and to control the lateral diffusion of the dopants.In this work, we have studied different optical coatings (different materials and thicknesses) irradiated by a long pulse SOPRA VEL 15 excimer laser (200 ns-15 J).Junctions have been characterized by 4-point probe, in situ reflectivity, UV photometry and secondary ion mass spectroscopy. The efficiency and the integrity of the different coatings have been studied for different laser irradiation conditions in solid and molten phases. The results show that a proper optical coating optimizes the coupling of the deposited laser energy and is promising for improving the integration of the laser activation process of future CMOS junctions. (C) 2004 Elsevier B.V. All rights reserved.

According to the International Technology Roadmap for Semiconductors (ITRS), source and drain extensions thickness for 65 nm and below technology nodes MOSFET lead to a major challenge. Rapid thermal processing (RTP) tools reach the limit of their physical abilities in term of temperature cycle and pulsed laser thermal processing (LTP) tool arise as a major potential candidates to solve the fundamental problem of ultra-shallow junction (USJ) activation. LTP experiments have been performed with two different XeCl excimer lasers (λ=308 nm) with different pulse characteristics (20 and 200 ns). We examine the influence of the pulse duration on LTP of B+ (with and without Ge+ pre-amorphization) implanted silicon samples on the basis of real-time optical monitoring of the laser induced process, four-point probe resistivity measurements and secondary ion mass spectroscopy (SIMS) depth profiles. Experimental results are compared to model calculations for both laser pulses. The activated dopant dose, junction depth and sheet resistance, as a function of the laser fluence and shot number for both lasers, confirm the relevance and efficiency of laser processing to realize ultra-shallow and highly doped junctions required for the future CMOS generations. Influence of the pulse duration on the USJ formation process is also discussed. In particular, we show the capabilities of the 200 ns pulse to activate B in Si without melting the Si doped layer.

According to the International Technology Roadmap for Semiconductors (ITRS), the doping technology requirements for the MOSFET source and drain regions of the future CMOS generations lead to a major challenge. A critical point of this evolution is the formation of ultra-shallow junctions(USJ) for which present technologies, based on ion implantation and rapid thermal annealing, will hardly meet the ITRS specifications. Laser Thermal Processing (LTP) has been shown to be a potential candidate to solve this fundamental problem. In the present paper, LTP experiments have been performed with two XeCl excimer lasers (λ= 308 nm) with different pulse characteristics. The first laser (Lambda Physik, Compex 102) delivers 200 mJ laser pulses with a duration of ∼25 ns. The second laser is an industrial tool (SOPRA, VEL 15) that delivers 16 J laser pulses with a duration of ∼200 ns and allows to anneal a few cm die in a single laser shot. Here we examine the influence of the pulse duration on LTP of B + (with and without Ge + pre-amorphization) and BF 2 implanted silicon samples on the basis of real-time optical monitoring of the laser induced melting/recrystallisation process, four-point probe resistivity measurements, secondary ion mass spectrometry (SIMS) depth profiles. Experimental results are compared to finite element modelisation (FIDAP Fluent Software) that takes into account both laser pulses. The activated dopant dose, junction depth and sheet resistance, as a function of the laser fluence and shot number for both lasers, confirm the efficiency of laser processing to realize ultra-shallow and highly doped junctions as required by the future CMOS generations. Influence of the pulse duration on the USJ formation process is also discussed.

In the last few years, laser thermal processing (LTP) has become a potential solution for sub-0.1 mum technology requirements, as focused by the international technology roadmap for semiconductors (ITRS).This paper presents a numerical simulation of the propagation of the melting front and regrowth of Si-based structures during excimer (XeCl-308 nm) laser irradiation. The influence of the pulse duration is highlighted in the simulation and compared with experiments. Two different types of XeCl lasers with different pulse durations have been used for the experiments in order to validate the model (SOPRA VEL 15, 200 ns, 15 J and Compex Lambda Physik, 20 ns, 200 mJ). The comparison between the simulation and the experimental results has been carried out using results from in situ reflectance measurements (transient reflectivity), secondary ion mass spectroscopy, spreading resistance profiles and four-point probes. The different implantation conditions used for this study were B+ or BF2+ implanted Si wafers, with or without Ge+ pre-amorphization. Both experimental and simulation results show the potentiality of the LTP annealing technique for realization of ultra shallow junction under 0.1 mum. (C) 2002 Elsevier Science B.V All rights reserved.

Silver-copper alloys are condensed in a liquid-solid domain of the phase diagram on a tilted molybdenum substrate regulated in temperature. After a droplets regime, a film which contains a monolayer of nodular solid crystals forms. The size distribution and density of the particles in the film are measured after different condensation times. Results show that in our experimental conditions a ripening process occurs which is evidenced by a decrease of the number of particles with time, and a broad particle size distribution. However, the decrease rate is smaller than expected without a condensation flux. A model is then developed to interpret and generalize these results. This model results from the modification of the Lifshitz-Slyosov model [1] to take into account the supply from the vapour phase. It is shown that the higher the flux of material to solidify from the vapour phase is, the more the growth from the vapour phase overcomes the ripening process. Once the particle density reaches a characteristic value which is simply proportional to the incoming flux, no more particles are dissolved. The system then tends towards a monomodal distribution with a radius which grows in t1/3.

An experimental study of silver–copper alloys condensation with a composition entering a liquid–solid domain of the phase diagram is presented. The vapour beams were produced by two effusion cells heated by tungsten resistors and condensed on a tilted molybdenum substrate, regulated in temperature. Condensation durations were chosen to study the successive stages of the film formation. The microstructures of the quenched condensates showed that the mechanisms were similar in the case of copper rich and silver rich domains. At the beginning, droplets of condensation formed. The surface covered with the alloy increased with time. Later on, coalescences of droplets were observed and a film was established. Both droplets and film contained solid particles whose morphology was nodular. The average nodule radius and the number density of particles in the film were determined by image analysis. It was demonstrated that the cube of the average particle radius grew with time and that the number density of crystals was inversely proportional to time. These results were in good agreement with the Ostwald ripening law.

The influence of segregation due to thermal convection on the microstructure of Sn-Cu6Sn5 fibrous eutectic alloys is studied in a Bridgman type configuration. The eutectic microstructure is characterized by means of image analysis, X-ray diffraction and scanning and transmission electron microscopy. In the absence of segregation, the eutectic is regular and its growth controlled by that of the Cu6Sn5 fibers. The effect of interphases on eutectic spacing, through orientation relationships between fibers and matrix, is also evidenced. The influence of segregation can be summed up by the following effects. At first, in agreement with the Jackson and Hunt model, it leads to a variation of the eutectic spacing which results from a variation of the fiber volume fraction. Then, the spacing is much greater than the one obtained in the absence of segregation, due to a different tin growth plane and non-optimized filer/matrix orientation relationships. Finally, the absence of steady state leads to a large dispersion of the spacing associated with a microstructural disorder. Copyright (C) 1996 Acta Metallurgica Inc.

The use of synchrotron radiation in the study of ordering transitions is presented. The peak-profile measurements can be carried out with a resolution one order of magnitude better than with classical sources in all directions of the reciprocal lattice. In the case of Fe–Al alloys, the small-angle scattering techniques can be transposed in the vicinity of a Bragg superstructure peak, which provides information on the antiphase configuration of the ordered alloys. Moreover, the high brilliance of the undulator of the ESRF ID10 beamline enables coherent scattering experiments. The conditions necessary to record holograms from an inhomogeneous sample are discussed and some results are presented.

We consider transient behaviour in antiphase coarsening in simple binary alloys, and corrections to late-stage asymptotic scaling with an exponent of one half. (The problem might be of some importance to the metallurgical properties of composite microstructures.) Segments of antiphase boundaries in three dimensions, with initially small total curvature, evolve rapidly, resulting in an effective growth exponent n(eff)(t) = 1/2[1 + xi'/R(t)BAR] where xi' is a microscopic length and R(t)BAR is the characteristic antiphase-domain size at time t.

Natural convective effects on dendritic primary spacing and morphology, in directional dendritic solidification of Al-Cu binary alloys, have been studied experimentally as a function of density gradient orientation in the interdendritic liquid relative to the gravity vector. Diffusion controlled experiments in a microgravity environment have been used as a reference. In a previous paper [Dupouy et al., Acta Met. 37 (1989) 1143], the convective effects in terms of segregations were predicted and analysed. In the present paper, morphological aspects are discussed: it is found that the development of lateral dendritic branches is strongly affected by the flow configuration.

This paper is concerned with the effect of natural convective patterns evidenced in a previous paper [Acta metall.37, 1143 (1989)], on the dendritic primary spacings. Primary spacings in normal gravity environment have been found to be much smaller than those measured in a microgravity environment, the latter being in good agreement with the diffusion controlled theoretical predictions. The scaling analysis of convective effects developed in the first paper allows us to propose a relationship, which gives the primary spacing as a function of the experimental parameters, in convective transport conditions of the solute in the liquid. This correlation is in good agreement with our experimental results, and those of the literature.

We consider a free-surface liquid layer convecting under an applied horizontal temperature gradient. The surface gets slightly tilted by an angle which depends on layer thickness through competing capillary and gravity effects. This result might be of some interest in Marangoni solidification set ups. It also opens up a new way of accurately measuring the surface-tension temperature coefficient.

FeAl25 and FeAl25B(0.05–1) (wt.%) thin ribbons were produced by the planar-flow-casting process, either under vacuum or under helium pressure, on a steel wheel spinning at 40 m s−1. Grain morphology, grain size and the presence of segregations were correlated with boron content and quench atmosphere. FeAl25 alloys, rapidly solidified under vacuum, exhibit large polygonal grains (15 μm), generally crossing the whole thickness of the ribbon, while under 100 mbar of helium pressure, the grain size is about 5–6 μm. We think this difference is probably due to grain growth during secondary cooling under vacuum, when the ribbon has left the wheel. The addition of boron in FeAl25 alloys first improves the quality of the quench. The grain shape changes from polygonal to columnar with boron content, and a segregated structure is observed, appearing at a few microns from the wheel side and tending to be more pronounced towards the free surface. A finite difference heat flow model and calculations undertaken to quantify secondary cooling of the ribbons out of the wheel are used to interpret the microstructure of the ribbons.

Image analysis characterization of fibrous eutectics microstructures In this work, we study the compacity and the regularity of the network defined by the centers of the fibres in a fibrous eutectic. Various methods are used to quantify these two properties in structures whose regularity is intermediate between perfectly hexagonal and random. Structures of Al-Al3Ni, Sr modified Al-Si and InSb-NiSb eutectics are examined. A random structure generated from a hard core model is also studied.

Macroscopic natural convective effects in directional dendritic solidification of Al-Cu alloys are studied experimentally as a function of density gradient orientation in the interdendritic and bulk liquid relative to the gravity vector. Diffusion controlled experiments in a microgravity environment are used as a reference. Three different types of macrosegregation correspond to three different convective modes. A scaling analysis is performed in order to give a general criterion of the transition from diffusive to convective transport conditions. Our work confirms that a large number of the experimental studies previously performed to analyse primary dendrite spacings or structural transitions, have been affected by convection.

A comparative experimental study of thermal and thermosolutal convection is carried out by solidifying a concentrated Ge-Si binary system and a dilute Ge-Ga system in a new Bridgman furnace where radial thermal gradients have been minimized and well characterized. Very different solutal boundary layer extents (δGeSi > 3 cm, δGega ≦ 2 mm) are found for the two systems solidified under the same thermal conditions. Simple analytical hydrodynamic models are then used to explain these results. It is demonstrated that the larger boundary layer extent observed for Ge-Si and the corresponding pure diffusive transport regime are the result of the stabilizing effect of the longitudinal solutal gradient in this system. It is also shown that Hart's analytical model [J. Fluid Mech, 49 (1971) 279] can be adapted to describe the thermal/solutal coupling of the fluid flow in crystal growth configurations and therefore to predict, in a first approximation, the segregation behaviour in the crystals through a boundary layer model [Favier, Acta Met. 29 (1981) 197, 205].

The different types of convective phenomena which may occur during the dendritic solidification of metallic alloys are discussed from an order of magnitude analysis. Bulk thermal convection and/or interdendritic solutal convection have to be considered according to the values of the experimental data. Scaling laws for the solute boundary layer resulting from bulk thermal convection have already been derived. It is shown here that the interdendritic flow depends on a solutal Grashof number Gr based on the horizontal density gradient and a characteristic length Ls which is of the order of the liquid channels width. For Gr < 1, which is generally verified in practical cases, the interdendritic flow velocity Ur is proportional to the Grashof number. This a priori law compares favorably with the results of horizontal solidification experiments where the mean interdendritic flow velocity has been estimated from the resulting measured macrosegregation. In these experiments, as well as for most horizontal dendritic solidifications of metallic alloys at 1 g, the ratio UrR (R is the growth rate) is of order one. In order to cancel the interdendritic flow effects, this ratio has to be lowered by one order of magnitude. According to our analysis, this can be obtained by performing the experiments either at a slightly reduced g level (∼10−1 g), or at 1 g in a vertical stable configuration with a sufficiently low residual horizontal thermal gradient.

In Part I, a typical situation of crystal growth from a doped melt in the presence of thermal buoyancy driven convection has been considered and an order of magnitude (OM) analysis of the dopant (or solute) transport in the liquid has been performed. From this analysis, the appropriate dimensionless parameters of the problem, the different solute transport regimes and the corresponding scaling laws for the extent of the solute boundary layer, have been derived; accordingly, the different situations concerning the resulting longitudinal macrosegregation in the crystal have been distinguished. In this Part, a precise determination of the length scale Ls to be considered for the flow is performed with the help of the available literature data on the momentum transport in this configuration. From the knowledge of Ls, the OM results of Part I are converted into practical laws expressed in terms of classical Gr, Re, Gr Sc and Pe numbers. The a priori laws thus obtained are shown to be coherent with the available literature data on segregation in such a configuration. At last, these laws are applied to discuss the experimental conditions (especially g-level) to be achieved in order to obtain a pure diffusion solute transport in the melt and no macrosegregation in the crystal.