The ab initio pseudopotential code (VASP) was employed to explore indium and boron electrical activities in silicon in the presence of nitrogen. Electrical activities for the combinations B+N, In+N, and In+B+N were explored. Formation energy of a negatively charged supercell, (E−)f, and a band gap, Eg, from calculations with one k point were chosen as indicators of acceptor activity. For separate dopants the calculated (E−)f and Eg values indicate that substitutional B and In are effective acceptors and N is an extremely weak donor. When nitrogen is adjacent to, or separated 3 - 5 bonds from B or In, it suppresses acceptor activity. Binding is greater for In+N than for B+N in agreement with secondary ion mass spectroscopy (SIMS) data that demonstrates a greater retention of N by In. This should lead to a greater drop in activity for In+N combination versus B+N one, in agreement with spreading resistance profiling (SRP) experiments. Loss of activity in In+B+N combination might be due to long range interactions between dopants.

We present the results of ab-initio calculations for the structure and energetics of small boron-carbon (BCI) as well as carbon-carbon (C2I) clusters in Si, a continuum model for the nucleation, growth, and dissolution of the clusters, and experimental investigation by SIMS. The modeling results suggest that these clusters may play a role in controlling B diffusion in Si and SiGe systems and the experimental results seem to support the modeling findings.

Secondary ion mass spectrometry (SIMS) is an effective and powerful analytical technique, widely used in accurately determining dopant distributions (depth profiles). However, primary ion beam induced mass transport (ion mixing), especially the residual effect during SIMS profile measurements, greatly limits theaccuracy at nanometer depth resolutions by displacing and broadening the measured depth profile. In this paper, we present a simple deconvolution algorithm based on the general characteristics of the experimentally observed SIMS response function to reduce this broadening effect, thereby providing more accurate depth profiles. The results for several specific applications of this approach are presented and its strengths and limitations are discussed.

This paper studies the use of ion implantation and rapid thermal annealing for the fabrication of shallow junctions in sub-100 nm CMOS technology. Spike annealing recipes were optimized on the basis of delta-doping diffusion experiments and shallow junction characteristics. In addition, using GeF2 pre-amorphization implants in combination with low-energy BF2 and spike annealing, p-type junctions depths of 30 nm were obtained with sheet resistances as low as 390 Ω/sq. The combined finetuning of implantation and annealing conditions is expected to enable junction scaling into the 70-nm CMOS technology node.

In this work we investigated the diffusion and clustering of supersaturated substitutional carbon 200nm thick SiGeC layers buried under a silicon cap layer of 40nm. The samples were annealed in inert (N2) or oxidizing (O2) ambient at 850°C for times ranging from 2 to 10 hours. The silicon self-interstitial (I) flux coming from the surface under oxidation enhances the C diffusion with respect to the N2 annealed samples. In the early stages of the oxidation process, carbon escape by diffusion across the layer/cap interface dominates. This phenomenon saturates after an initial period (2-4h) which depends on the C concentration. This saturation is due to the formation and growth of C containing precipitates which are promoted by the I injection and act as a sink for mobile C atoms. The competition between clustering and diffusion is discussed for two different C concentrations.

Doping process windows are becoming very narrow as VLSI technology nodes scale to smaller and smaller dimensions. The time and cost required to develop new doping methods and the desire to re-use equipment will make it likely that current methods will be applied as long as possible. This means that existing process tools will have very tight stability and uniformity requirements, and metrology will be required to drive process control. The paper describes the state-of-the-art of both doping processes involving ion implantation and spike annealing, and new metrology based on Carrier IlluminationTM methods that will be required to implement in-line process control for these processes. CI offers depth resolution on the order of1Å, providing a level of control required to extend existing doping methods. The prospects of new methods such as Laser Thermal Annealing (LTA) are also discussed.

We have implemented an atomistic simulation of the Ostwald ripening of extrinsic defects (clusters, {113}'s and dislocation loops) which occurs during annealing of ion implanted silicon. Our model describes the concomitant time evolution of the defects and of the supersaturation of Si interstitial atoms in the region. It accounts for the capture and emission of these interstitials to and from extrinsic defects (defined by their formation energy) of sizes up to thousands of atoms and includes a loss term due to the interstitial flux to the surface. This model reproduces well the dissolution of {113} defects in Si implanted wafers. We have subsequently studied the characteristics of TED in the case of B implantation at low and ultra low energy. In such cases, the distance between the defect layer and the surface plays a crucial role in determining the TED decay time. The simulations show that defect dissolution occurs earlier and for smaller sizes in the ultra-low energy regime. Under such conditions, TED is mostly characterized by its “pulse” component which takes place at the very beginning of the anneal, probably during the ramping up. In summary, we have shown that the physical modelling of the formation and of the growth of extrinsic defects leads to a correct prediction of the “source term” of Si interstitials and at the origin of TED.

Molecular Dynamics (MD) and Activation-Relaxation Technique (ART) models of decaborane ion implantation into Si and following rapid thermal annealing (RTA) processes have been developed. The B and Si atomic positions for implantation of accelerated decaborane ions, with total energy 3.5- 15 KeV, into Si substrate were obtained by MD simulation. The main difference between monomer and decaborane ion implantation with the same doses is the formation of a large amorphized area in a subsurface region for the decaborane case. The number of displaced Si atoms shows non-linear energy dependence at low impact energies. At higher energies ofthe investigated range of the decaborane energy range, however, a linear dependence is observed in accordancewith the prediction of the Kinchin-Pease formula. A new method that incorporates Activation-Relaxation Technique (ART) with MD has been developed and used to study re-crystallization of Si amorphized in the implantation process.

Nitrogen implantation is used to retard gate oxide growth thereby making it particularly usefulfor dual- VT and System On A chip technologies. This paper discusses the diffusion behavior and the concomitant defect evolution at high doses of implanted nitrogen in silicon. This paper shows that as the nitrogen implant dose is increased, the extent of nitrogen diffusion reduces. This paper also reports based onTEM studies, that upon annealing at 750°C, 5 × 10014 N2+/cm2, 40 keV implant produces Type I extended defects. However, 2 × 1015 N2+/cm2, 40 keV implant, produces a continuous amorphous layer to a depth ofabout 800 to 900 Å from the surface. In addition, upon annealing at 750°C, the 2 × 1015 N2+/cm2, 40 keV implant produces Type V or solid solubility defects in addition to End of Range or Type II defects.

Plan-view transmission electron microscopy (PTEM) was used to characterize defect evolution upon annealing of low-to-medium energy, 5-30 keV, germanium implants into silicon. The implant dose was 1 × 1015 ions/cm2, sufficient for surface amorphization. Annealing of the samples was done at 750 °C in nitrogen ambient by both rapid thermal annealing (RTA) and conventional furnace, and the time was varied from 10 seconds to 360 minutes. Results indicate that as the energy drops from 30 keV to 5 keV, an alternate path of excess interstitials evolution may exist. For higher implant energies, the interstitials evolve from clusters to {311}'s to loops as has been previously reported. However, as the energy drops to 5 keV, the interstitials evolve from clusters to small, unstable dislocation loops which dissolve and disappear within a narrow time window, with no {311}'s forming. These results imply there is an alternate evolutionary pathway for {311} dissolution during transient enhanced diffusion (TED) for these ultra-low energy implants.

A molecular Dynamics (MD) model of B implantation into Si and of sputtering the Si surface withenergetic B10 clusters has been developed. The goal was to simulate the implantation of ions of decaborane (B10H14), which may become an important process for the formation of ultra shallow junctions in future MOS devices. The simulations, carried out for the cluster ion impact energy from 3.5 keV to 15 keV, have revealed the formation of a large amorphized region in a subsurface region. At low cluster impact energies in this range some of the B atoms were recoiled back from the surface, but at the energy of 12 keV and above almost all of the Boron atoms were successfully implanted into Si. The sputtering yield of Si has been also computed and found to increase with energy, reaching the value of 6 Si atoms per B10 cluster at 15 KeV. The number of displaced surface atoms correlates well with the sputtering yield, and between 3.5 and 10 keV it has a non-linear dependence on energy. At higher energies the number of displaced atoms increases linearly with energy, in agreement with the Khinchin-Pease formula [9]. The sputtering yield at 12 keV was also measured by the amount of Si removed by a decaborane beam from a thin Si film deposited on a carbon substrate. The predicted sputtering yield agrees well with this experiment.

Preamorphization is commonly used to form shallow junction in silicon CMOS devices. The purposeof this experiment was to study the effect of the preamorphizing species' mass on the interstitial concentration at the end-of-range (EOR). Isovalent species of Si, Ge, Sn and Pb were compared. Silicon wafers with a buried boron marker layer (4700 Å deep) were amorphized using implants of 22 keV 28Si+, 32 keV73Ge+, 40 keV 119Sn+ or 45 keV 207Pb+, which resulted in similar amorphous layer depths. All species were implanted at a dose of 5×1014 /cm2. Cross-sectional transmission electron microscopy (XTEM) was used tomeasure amorphous layer depths (approximately 400 Å). Post-implantation anneals were performed at 750 °C for 15 minutes. Plan-view transmission electron microscopy (PTEM) was used to observe and quantify the EOR defect population upon annealing. Secondary ion mass spectrometry (SIMS) was used to monitor the transient enhanced diffusion (TED) of the buried boron marker layer resulting from the EOR damage introduced by the amorphizing implants. Based upon the SIMS results Florida Object Oriented Process Simulator (FLOOPS) calculated the resulting time average diffusivity enhancements. Results showed that increasing the ion mass over a significant range (28 to 207 AMU) not only affects the quantity and type of damage that occurs at the EOR, but results in a reduced diffusivity enhancement.

We have investigated the lattice site location of B in Si using ion channeling in combination with nuclear reaction analysis (NRA). Silicon samples implanted with Boron at an energy of 10 keV and a dose of 5 × 1014 cm−2 (low dose samples) or 5 × 1015 cm−2 (high dose samples) were annealed at 1000 °C for 10 seconds (RTA) or at 800 °C for 10 minutes (FA). The activation efficiencies of these samples were estimated from the B atomic concentration and the hole concentration obtained by secondary ion mass spectrometry (SIMS) and spreading resistance profiling (SRP), respectively. We also studied the ion implantation damage of Si crystals using ion channeling combined with Rutherford backscattering spectrometry (RBS). We found that the activation efficiency is proportional to the substitutionality, meaning that substitutional B is fully activated without any carrier compensation. We also found that B atoms go to the substitutional sites and are activated up to the solubility limit in the high dose samples. However, the ion implantation damage of the crystalline Si in the high dose samples increases somewhat after annealing.

The method of ion implantation and spike annealing for preparing shallow junctions suitable for the extension regions bridging the channel and source/drain contacts of CMOS transistors are studied by annealing blanket implants. Junction depths at a given sheet resistance for low energy B implants are minimized for the combination of a fast ramp with a sharp-spike anneal. This is shown to be physically based on activation energy phenomenology. The fraction of electrically activated B is insensitive to implant dose, unlike the case of transient enhanced diffusion. Arsenic implants show higher activation fraction than comparably annealed P implants, without the large transient enhanced diffusion which is attributed to P and Si-interstitial coupled diffusion. For targeted sheet resistance and junction depth, spiking temperature trends lower with implant dose, concomitant with decreasing fraction of activated dopant.

The main drawbacks of the known models of the B diffusion in strained SiGe layers are summarized. A mechanism is suggested to self-consistently explain the main experimental features and original experimental data which considers the trapping of B atoms by Ge atoms during B diffusion in the Si lattice resulting in the retarded B diffusivity. Fluctuations of Ge atom numbers in a nearest B atom environment result in percolation mechanism of B transport through dilatation centers of random size. A new solid state transport modelis generalized by considering dispersion transport of positive and negative point dilatation defects.

It is now well established that the transient enhanced diffusion (TED) of ion implanted boron in silicon limits the formation of the ultra-shallow junctions required for the extreme deep sub- micron devices. It is also known that this TED is linked to the fate (elimination and agglomeration) of ion implantation related excess self-interstitials. Thus it can be expected that the final high temperature redistribution is at least partly governed by the effective initial point defect distribution at the onset of the high temperature plateau.

In this contribution we present the experimental evidence that low thermal pre-anneals, by affecting the initial self-interstitials distribution, affects boron redistribution during a subsequent high temperature anneal. Samples implanted with high dose boron at 3 keV were first annealed at 700°C for various durations. These samples, as well as reference samples without the pre-anneal, were then RTA annealed at various high temperatures around 1000°C. The resulting B profiles were measured by SIMS. It is found that the pre-annealed samples exhibits a clear reduction of the TED as compared with the reference ones.

Silicon self–diffusion coefficients were measured in intrinsic and extrinsic silicon from870 to 1070°C using isotopically pure 30Si layer. 30Si diffusion profiles are determined by secondary ion mass spectrometry. The temperature dependence of intrinsic diffusion coefficient in bulk Si isobtained. Comparing it in heavily As-doped or B-doped Si, it is found that Si self-diffusion is entirely mediated by interstitialcy mechanism at lower temperatures below 870°C.

Carbon doping of silicon has gained interest since in high concentrations, carbon can reduce oreven suppress undesirable diffusion of the base dopant boron in silicon-based bipolar transistors. This behavior can only be understood in taking into account the silicon point defects i.e. vacancies and self-interstitials. In this work, we observe the oversaturation of vacancies produced by a high carbon concentration duringannealing. Experiments with a vacancy diffusing dopant, Antimony, are shown and prove this effect: in a carbon rich sample, the antimony diffusion is enhanced about 8 times compared to samples with a much lower carbonconcentration. We also investigate the carbon co-precipitation with oxygen. The carbon precipitation, asSiC, is facilitated with a high oxygen concentration. We explain this affinity by an exchange of point defects and a volume compensation. Finally, we show the precipitation of oxygen in relation to the vacancy oversaturation at 900°C.

A quantitative model of the electrical activity of metallic precipitates in Si is presented. An emphasis is placed on the properties of the Schottky junction at the precipitate-Si interface, as well as the carrier diffusion and drift in the Si space charge region. Carrier recombination rate is found to be primarily determined by the thermionic emission charge transport process across the Schottky junction rather than the surface recombination process. It is shown that the precipitates can have a very large minority carrier capture cross-section.

Molecular dynamics simulations of boron monomer and small clusters (B4 and B10) impacting on Si(001) were performed in order to investigate the damage formation by monomer/cluster impact. These monomer and clusters show similar implant depth and efficiency, but different damage structures. At the impact of B monomer with 230eV of incident energy, some point-defects such as vacancy-interstitial pairs are mainly formed. On the other hand B10 produces several times larger number of vacancies and interstitials compared with B1, This damage structure is different from one by B1 implantation and due to high yield amorphization of implanted region. This characteristic damage formation process is expected to cause different annihilation process.