The Boltzmann transport equation has been used to calculate range anddamage distributions in multilayer targets of general interest for semiconductor fabrication. A comprehensive review of the calculations will be presented, with particular emphasis on how large angle scattering events and channeling phenomena may be included. Examples of the quality of fit between the theory and experiment show that difficult phenomena (such as residual channeling) can be reasonable modelled.

High-quality, epitaxial layers regrown from ion-implanted, amorphous Si are an important element in integrated circuit fabrication. Two regimes are well-characterized and understood: furnace annealing in the solid phase from 500° to 1000°C and pulsed-laser annealing in the liquid-phase. Growth rates have been measured in the range from 1 Å/sec in the solid phase to 10 m/sec in the liquid phase. Impurity redistribution is in hand and only the strong enhancement of solid phase rates due to the presence of 0.1 at. % or greater of dopants remains as a controversial issue.

The emergence of rapid thermal annealing (10 to 20 sec at 1000 to 1100°C) as an attractive method of restoring crystal order has introduced a new area of controversy; in this case, growth rates and impurity redistribution. The experimental data from several groups will be presented in order to highlight the different mechanisms that have been invoked. The redistribution of implanted B and As will also be discussed.

Rapid Thermal Processing (RTP) is used to study shallow junction formation for high dose implanted silicon. The residual damage from As damage is efficiently removed using high temperature-short time anneals (1100°C - few seconds), while limited arsenic atom diffusion is obtained. The diffusion properties are also characterized by concentration enhanced diffusion at higher doping. The higher doping is metastable, with reversible changes in resistivity observed for sequential 1100°C-800°C-1100°C-800°C thermal cycles. RTP gives shallower defect free As junctions than standard long time anneals. Boron junctions are limited by the depth extension of a large ion-channeling-tail which is shown to undergo local enhanced diffusion. The approaches to form shallow p+ junctions without channel tails are discussed. A summary of ion damage studies is made. Some generalizations for determining an RTP advantage or disadvantage are made, based on activation energy differences of effects.

This paper reviews the status of implantation and annealing in III-V multilayer heterojunctions. The limitations on diffusion doping of multilayers are noted. Current heterojunction device applications for implantation are summarized. Issues relevant to the annealing of multilayers are discussed, including Si migration in modulation doped structures, the effectiveness of rapid annealing in reducing SI diffusion, and impurity enhanced diffusion in superlattices. The formation of damage in multilayers and the unique properties of strained multilayers are also discussed. Finally, implant activation studies are summarized.

The first experimental mapping of the ion beam channels near the Si 〈001〉 axis is reported. Remarkably fine structure is observed, with over thirty microchannels identified within 20 degrees of the Si 〈001〉 axis. Each map is made from about 16000 channeling spectra with each channeling probability becoming a pixel of information in the final plot. Contour maps have been constructed for ions of He and N with energies from 50 to 2000 keV. All features are identified by theoretical plots of silicon planar channels and axial channels.

We have studied the effects of ion-channeling in B+ and As+ implanted Si which has been preamorphized to different depths by 28 Si+ irradiation at 60 keV and 180 keV. It is shown that the junctions for incompletely amorphized substrates are only slightly deeper than predicted by LSS theory. For preamorphized layers extending beyofnd-the predicted junction depths the implantations of 25 keV B+ and 120 keV 75As+ with a dose of 5-1015 cm-2 yield junction depths in the order of 2200 - 2300 Å after furnace annealing at 700°C for 30 minutes.

The work presents the main electro-optical characteristics of very shallow silicon n+-p and P+-n junctions fabricated through masked ion implantation of phosphorus (31P+) and boron (11B+), respectively. The dependence of electro-optical characteristics on technological parameters (implant energy E and dose Q, oxide thickness Xox, impurity redistributions, etc.) are outlined in detail.

The implant step was performed through a thermally-grown silicon dioxide (SiO2) mask of variable thickness (Xox = 100-1200 Å) at energies ranging from E=10 keV to 50 keV and with a dose level Q = 5 x 1014 - 2 x 1015 cm-2. As a rule, after the implant step the oxide layer was etched down. In several samples, the Si0 2 layer was preserved in the finished devices.

The characterization of fabricated junctions was performed using a Karl Zeiss quartz prism monochromator coupled to a Tektronix 31/4661 data acquisition system. The main peculiar results could be summarized as follows: (a) all fabricated devices possess a relatively large bandwidth (up to 580 nm) of the relative spectral response; (b) for oxide thickness Xox Rp (projected range), the low energy implant yields devices with good blue response; (c) for Xox equal or larger than Rp a pronounced shift of the peak responsivity wavelength toward the visible spectrum of the light occurs; (d) a distinctive feature of devices consists in the high peak values of the short-circuit current density under AM1 insolation (in excess of 33 mA/cm2); (e) the preservation of SiO2 layer in the finished devices improves their electro-optical characteristics.

A new model which combines heterogenous nucleation and growth process to explain the mechanism of the crystal regrowth of amorphous silicon, in contact with a Si single crystal substrate, is presented. This model explains the effects of substrate orientation and impurity concentration on the kinetics of crystal regrowth. A comparison of experimental results with the predictions of the model is also included.

Investigations were carried out on the damage of SiO2 and the Si-SiO2 interface induced by boron implantation through polysilicon/SiO2 /p-Si structures with doses up to 1014cm−2 and annealed at 950°C. Using the constant voltage stressing technique, both capacitance-voltage and thin-oxide tunneling current measurements showed that both electron trapping and hole trapping are increased, and that ion-induced electron trapping overcompetes hole trapping for boron doses higher than 5×1013cm−2.

Silica glass platelets have been implanted with Fe(+) ions, 125 KeV and 10⋀-5 A to doses of 10⋀14 cm⋀-2, 10⋀15 cm⋀-2, 10⋀16 cm⋀-2 and 5 x 10⋀16 cm⋀-2. Ion scattering measurements show that the peak of the Fe ion distribution is ~95 nm below the sample surface, approximately Gaussian, with a width at half maximum amplitude of 100 nm. The intensity of a component in the EPR spectra of implanted samples with a width of 300 gauss and an approximately isotropic shape increases with increasing dose. The increase, proportional to dose for doses <⋀16 cm⋀-2, is larger for doses ≥10⋀16 cm⋀-2. In addition for doses ≥10⋀16 cm⋀-2, the component is orientation dependent. Subsequent heat treatments in air at 700° and 800° C alter the shape, intensity, and orientation dependence of the component. The spectral component in as-implanted samples is attributed to paramagnetic states of Fe ions. The threshold for magnetic exchange interactions and consequently long range magnetic ordering at ~300K occurs at a dose ≥10⋀16 cm⋀-2. Data on effects of thermal treatments on intensities and shapes are interpreted in terms of changing chemical composition of precipitating-particles.

A model for formation of amorphous silicon by light ion implantation is proposed. It is suggested that accumulation of point defects and/or complexes is required at the initial stage of the amorphization process. Amorphous zones can only form at the end of incoming light ion tracks when the pre-accumulated concentration of point defects reaches a critical value. Depending on the uniformity of the point defect distribution, two possibilities for the second stage of amorphization are suggested when ion implantation is performed at different temperatures.

Silicon wafers implanted with boron ions below and above the critical amorphization dose at various temperatures have been investigated using cross section specimens in high resolution TEM. Complementary analyses of these specimens by Electron Paramagnetic Resonance have revealed the presence of dangling bonds in amorphous zones and point defect clusters. Extrinsic stacking faults with 1/3 <111> displacements and other smaller distortions with 1/x<111> displacements were also found to result from the amorphization process. Liquid nitrogen temperature was found to be necessary to cause complete amorphization of silicon by boron ion implantation.

A procedure to be used in ion implantation calculations has been developed to determine the stopping power of an ion at low energy as a function of its effective charge. The ion effective charge accounts for screening of the ion and has been found to have considerable effect on the stopping power through its dependence on the target electron density. Steps in the procedure include: the calculation of the Fermi momentum of the target, calculation of the relative velocity between the projectile and target electron cloud, determination of the screening distance for the ion, and calculation of the proton stopping power Sp according to the density-functional formalism. The ion stopping power is then is the ion effective charge. The procedure can be applied to semiconductors and metals. Comparisons are reported with the predictions of the Firsov and Lindhard methods which do not include any effective charge or shell structure considerations. The computer program MARLOWE has been modified to include this method for calculating the stopping power. Results in the form of implanted boron profiles in silicon will be presented.

This paper reviews recent detailed investigations into the crystal growth kinetics and the microstructure of ion-beam-stimulated epitaxial crystallisation of silicon. Beam-induced crystallisation at temperatures between 200-400°C is found to be characterised by an activation energy of 0.24eV. Furthermore, in this temperature regime, crystal growth on (100) silicon is found to be free of extended defects except for a sharp hand of dislocation loops centred about the range of the ions employed to stimulate crystallisation. A higher temperature regime (>400°C) is observed in which the growth kinetics are less well defined but appear to be associated with an apparent activation energy of >0.5eV. In this regime, extended defects are observed to extend from the ion range to the surface. Results are presented which strongly suggest that nuclear energy deposition precisely at the amorphous-crystalline interface is responsible for crystallisation under ion irradiation. It is argued that the major fraction (2.4eV) of the thermal-only activation energy for epitaxial crystallisation of silicon is likely to be associated with the formation of nucleation sites for growth, a step which is achieved athermally under ion irradiation. In addition, the growth rate per unit ion fluence is found to be independent of substrate orientation at temperatures <450°C and independent of doping concentration for temperatures <400°C. These results are consistent with our proposed model for beam-induced crystallisation.

Various models for predicting high fluence ion collection profiles are reviewed. Recent calculations based on the diffusion approximation are described. The solute and defect probability distributions are calculated by a MONTECARLO code, TRIM. The method takes into account the effects of sputtering, including preferential sputtering of one of the components, and lattice dilation. In addition, the effects of radiation enhanced diffusion and radiation induced segregation are also considered. The calculations include the coupling of solute and defect fluxes. The described formalism can account for observed maximum attainable concentrations and distributions in high fluence implantation conditions of practical interest.

Some renarks about the mechanism for dynamic annealing during high dose rate P implantation of Si are reported. TEM observations and RBS channeling measurements show that the ion bombardment enhances the amorphous to crystalline transformation in the temperature range 200 ≤ T ≤ 600°C. It is found that the ratio between the observed recrystalli-zation velocity and the thermal SPE velocity decreases with increasing temperature. This indicates that a transition temperature must exist between the ion-assisted recrystallization regime and the ?lermal SPE regime. For the energy (100 keV) and the dose rate (60,uA/cm2) used in our experiments the transition temperature is about 700°C.

Epitaxial Ge-Si alloy films were grown on Si(100) by molecular beam epitaxy, subsequently given a shallow P implant, and subjected to rapid thermal processing. Heat treatment causes solid-phase epitaxial regrowth of the amorphized implanted layer similar to the case of pure Ge. Phosphorus redistribution, loss, and trapping at the Ge-Si/Si interface are also observed. Anomalous electrical activation is observed for P concentrations below 1 at.%, where the-carriers are either trapped or compensated at room temperature, but not below 100K. Analyses were carried out by Rutherford backscattering and channeling, secondary ion mass spectrometry, and temperature-dependent electrical transport.

Depth profiles of 1-2000 keV boron in amorphous and crystalline silicon have been calculated using the computer simulation code COSIPO. Clear inconsistencies have been found in the experimentally measured range parameters. The optimum selection in channeling reduction is found to be a 9° tilt from the ˂001˃-axis combined with a 7° rotation from the ˂100˃- axis. A smaller offset angle yields noticeable channeling at low energies.

The need for low temperature processes in VLSI CMOS technology has led to increasing interest in fully implanted wells. In comparison with diffused wells the advantages are good control of the doping profile, low lateral distribution and self-immunity to latch-up.

The technique using a medium energy machine with multiple charged ions is not viable in a production context and only high energy machines can be improved to meet the high throughputs required.

We have performed boron and phosphorus in the 1 to 3 MeV range on tandem and Van de Graaff machines to prove their effectiveness. Spreading resistance for boron and phosphorus and SIMS for boron are the characterizatton methods uped. The agreement with expected profiles for doses in the 1013 to 1014 /cm2 range is good. We checked the compatibility with a N-well CMOS process: - low doping level at the surface to ensure low capitances and no disturbance of the channel region - high doping level at one micrometer under the surface to lower punchthrough and latch-up effects. The efficiency of masking materials such as Si02 and photoresists is experimentally measured detecting residual doping underneath various thicknesses of the masking pattern. The application to C-MOS technology is discussed.

Cross-sections were determined for displacement of arsenic from substitutional sites in silicon by 1.5MeV He+ ions. Results are presented for channelled and random incidence on (111) and (100) silicon, doped by arsenic implantation and annealed by scanning electron beam irradiation. Layers doped during MBE growth were also studied for comparison.

Inner L-shell ionisation of host atoms is proposed to explain the observed displacement cross-section and its variation with arsenic concentration. The results are used to indicate the limits for using Rutherford backscattering and channelling techniques to measure the substitutionality of dopants in single crystals.

We report on an investigation of ‘buried’ oxygen implants formed by 0+ implantation at 400 KeV and 3.5 MeV into p-type CZ (100) wafers with a dopant density NA 1015 cm−3. Peak concentrations of 1 × 1019 cm−3 to 2 × 1020 cm−3. were investigated. Test devices were fabricated on implanted and annealed wafers using conventional wafer processing. For the 400 keV implants, a 4 µm epitaxial buffer layer was grown subsequent to the 0+ implant. For a dose of 3 × 1015 cm−2 the lifetime reduction ratio for the effective generation lifetime τg at the implant peak is greater than 103 relative to an unimplanted region where τg = 150 µS. C-V and SRP profiles show evidence for oxygen donor compensation. TEM analysis reveals a well defined layer at 1 µm with respect to the original implant surface containing a relatively high density of small precipitates and dislocation loops. DLTS measurements on diodes reveal 2 electron traps designated as E1 and E2. The trap energy ∆E and capture cross section σ are (EC-ET)1 = 0.41 ± 0.020 eV and (EC-ET)2 = 0.22 ± 0.030 eV with σ 1019 cm-3. The estimated trap density NT for the dominant trap is 8.2 × 1013 cm−3 for a calculated peak 0+ implant concentration of 6.8 × 1019 cm−3. The values of ∆E are in good agreement with values for unimplanted CZ wafers subjected to 2-step precipitation anneals. The experimental results provide direct evidence that ion implantation provides an effective method of introducing atomic oxygen in silicon at concentration exceeding its solid solubility during processing to produce a buried low lifetime region.