Transient Enhanced Diffusion (TED) is one of the biggest modeling challenges present in predicting scaled technologies. Damage from implantation of dopant ions changes the diffusivities of the dopants and precipitates to form complex extended defects. Developing a quantitative model for the extended defect behavior during short time, low temperature anneals is a key to explaining TED. This paper reviews some of the modeling developments over the last several years, and discusses some of the challenges that remain to be addressed. Two examples of models compared to experimental work are presented and discussed.

Transient Enhanced Diffusion (TED) of dopants in Si is the consequence of the evolution, upon annealing, of a large supersaturation of Si self-interstitial atoms left after ion bombardment. In the case of amorphizing implants, this supersaturation is located just beneath the c/a interface and evolves through the nucleation and growth of End-Of-Range (EOR) defects.

For this reason, we discuss here the relation between TED and EOR defects. Modelling of the behavior of these defects upon annealing allows one to understand why and how they affect dopant diffusion. This is possible through the development of the Ostwald ripening theory applied to extrinsic dislocation loops. This theory is shown to be readily able to quantitatively describe the evolution of the defect population (density, size) upon annealing and gives access to the variations of the mean supersaturation of Si self-interstitial atoms between the loops and responsible for TED. This initial supersaturation is, before annealing, at least 5 decades larger than the equilibrium value and exponentially decays with time upon annealing with activation energies that are the same than the ones observed for TED. It is shown that this time decay is precisely at the origin of the transient enhancement of boron diffusivity through the interstitial component of boron diffusion. Side experiments shed light on the effect of the proximity of a free surface on the thermal behavior of EOR defects and allow us to quantitatively describe the space and time evolutions of boron diffusivity upon annealing of preamorphised Si layers.

Delta-doped boron marker layers in silicon have been used to test further the relationship between B transient enhanced diffusion (TED) and the flux of silicon interstitials released during the annealing stage following self implantation. We present new data which address a number of questions raised by the present models. We show that in our experiments bulk trapping of interstitials is significant only for low implant fluences (˜1012 cm2). The origin of the observed diffusion-like profiles for the interstitial flux is instead found to lie in local trapping within the δ-doped layers themselves. Boron trapped in immobile clusters may be associated with Si interstitials in approximately a 1:1 ratio; nevertheless this trapping contribution alone may not entirely account for the observed gradient. We suggest that some part of the observed TED response with depth is attributable to local trapping of silicon interstitials within the boron doped layers.

(001) CZ silicon wafers were implanted with arsenic (As+) at energies of 10–50keV to doses of 2x 1014 to 5x1015/cm2. All implants were amorphizing in nature. The samples were annealed at 700°C for 16hrs. The resultant defect microstructures were analyzed by XTEM and PTEM and the As profiles were analyzed by SIMS. The As profiles showed significantly enhanced diffusion in all of the annealed specimens. The diffusion enhancement was both energy and dose dependent. The lowest dose implant/annealed samples did not show As clustering which translated to a lack of defects at the projected range. At higher doses, however, projected range defects were clearly observed, presumably due to interstitials generated during As clustering. The extent of enhancement in diffusion and its relation to the defect microstructure is explained by a combination of factors including surface recombination of point defects, As precipitation, As clustering and end of range damage.

It has been observed that dislocation loops form and grow during annealing of silicon wafers implanted at doses above the amorphization threshold. Dislocation loops can act to store interstitials for prolonged periods of anneals, sustaining an interstitial super-saturation and thus causing substantial transient enhanced diffusion (TED). We have developed a comprehensive model which, in combination with a model and parameters for {311} defects from previous work, accounts for the formation and evolution of dislocation loops during ion implant annealing, as well as giving the correct TED behavior.

A study has been made of B transient enhanced diffusion (TED) in heavily P-doped Si using secondary ion mass spectroscopy (SIMS) and positron annihilation spectroscopy (PAS). The Pdoped silicon was implanted with boron ions of 40 keV energy to a dose of 3 x 1014 cm-2, and then annealed at temperatures ranging from 700–1000°C in a N2 ambient for varying durations. As P doping concentration increased from 3 x 1019 to 1 x 1020 cm-3, boron diffusivity and the immobile boron fraction decreased. Our experimental results are inconsistent with the predictions of the Fermi-level model and suggest that the clustering between B atoms and Si interstitials should be invoked in order to explain the immobile portion of the B peak during TED.

Raman spectroscopy is used to characterize silicon implanted with boron at a dose of 1014/cm2 or less and thermally annealed. The Raman scattering strengths and band shapes of the first-order optical mode at 520 cm-1 and of the second-order phonon modes are investigated to determine which modes are sensitive to the boron implant. The asimplanted samples show diminishing Raman scattering strength as the boron dose increases when the incident laser beam is 60° with respect to the sample normal. Thermal annealing restores some of the Raman scattering strength. Three excitation wavelengths are used and the shortest, 457.9 nm, yields the greatest spectral differences from unimplanted silicon. The backscattering geometry shows a variety of changes in the Raman spectrum upon boron implantation. These involve band shifts of the first-order optical mode, bandwidth variations of the first-order optical mode, and the intensity of the second-order mode at 620 cm-1.

We present a new approach for investigating the kinetics of defect migration during annealing of low-energy, ion-implanted silicon, employing a combination of computer simulations and atomic-resolution tunneling microscopy. Using atomically-clean Si(111)-7×7 as a sink for bulk point defects created by 5 keV Xe and Ar irradiation, we observe distinct, temperature-dependent surface arrival rates for vacancies and interstitials. A combination of simulation tools provides a detailed description of the processes that underdy the observed temperature-dependence of defect segregation, and the predictions of the simulations agree closely with the experimental observations.

Ab initio planewave pseudopotential method is used to study carbon diffusion and pairing in crystalline silicon. The calculation is performed with a 40 Ry planewave cutoff and 2×2×2 special k-point sampling with a supercell of 64 atoms. It is found that substitutional carbon attracts interstitial Si forming a <001> C interstitial with a large binding energy of 1.45 eV. The interstitial carbon is mobile and can migrate with a migration energy of 0.5 eV. The interstitial carbon can bind further to another substitutional carbon forming a substitutional carbon-interstitutional carbon pair with a binding energy of 1.0 eV. This model is used to understand the effect of high C concentration on the transient enhanced diffusion in Si.

A simulation model for Boron diffusion which takes into account the aggregation of the excess interstitials in clusters, and subsequently, the dissolution of these defects, is proposed. The interstitial supersaturation and generation rate are determined according to the classical theory of nucleation and growth of particles, in analogy with the precipitation of a new phase in heavily doped silicon. The clusters are considered as precipitates formed by interstitial Si atoms. The B diffusion is modelled on the basis of the dopant-interstitial pair diffusion mechanism. The clusters dissolution during annealing maintains nearly constant, for a long period, the interstitial supersaturation and the related enhancement of the boron diffusion. This gives a good account of the diffusion results over a large range of experimental conditions. Furthermore, this approach describes most of the behavior of the transient enhanced diffusion (TED), like the temperature dependence of the level of the B diffusion enhancement, the dependence of the duration of the phenomenon on implanted dose, and the scarce dependence on the damage distribution in depth. The results of the simulations are compared with experimental data on the kinetics of interstitial cluster dissolution and of B TED.

Our recent work on the room temperature migration and trapping phenomena of ion beam generated point defects in crystalline Si is reviewed. It is shown that a small fraction (˜10−6) of the defects generated at the surface by a shallow implant is injected into the bulk. These defects undergo a long range trap-limited diffusion and interact with both impurities, dopants and preexisting defects along their path. In particular, these interactions result in dopant deactivation and/or partial annihilation of pre-existing vacancy-type defect markers. It is found that in highly pure, epitaxial Si layers, these effects extend to several microns from the surface, demonstrating a long range migration of point defects at room temperature. By a detailed analysis of the experimental evidences we have identified the Si self-interstitials as the major responsible for the observed phenomena. This allowed us to give a lower limit of 6 × 10−11 cm2/s for the room temperature diffusion coefficient of the Si self-interstitials. Room temperature trap-limited migration of vacancies is also detected as a broadening in the divacancy profile of as implanted samples. In this case the room temperature diffusion coefficient of vacancies has been found to be ≥3 × 10−12 cm2/s. These data are presented and their implications discussed.

We used deep level transient spectroscopy to determine the concentration and depth profile of the defects introduced by MeV He and Si implants in n-type crystalline Si. We have found that only ˜ 16% of the Frenkel pairs generated by the ion escapes recombination and is stored into room temperature stable defects such as divacancies and oxygen vacancy complexes. For a light ion (He), the depth distribution of these defect complexes is strongly dependent on the O content of the substrate: it mirrors the initial distribution of I–V pairs, as calculated by TRIM (a Monte Carlo Code) when the O content is high (˜1018/cm3) while it can be much wider (up to 2 μm) in a highly pure (low O content) epitaxial substrate. This effect is due to a long range migration of vacancies before clustering or trapping at impurities. This migration is strongly inhibited for an ion of higher mass (such as Si) since in a denser collision cascade direct clustering is strongly favoured with respect to agglomeration of migrating defects.

Diffuse x-ray scattering (DXS) and computer simulation techniques were employed to investigate the defect structure produced in Si by low keV ion and MeV electron irradiations. DXS measurements were performed for keV Ga and He implants, demonstrating the ability of the technique to provide both bulk and near-surface measurements at defect concentrations of about 1000 ppm. A rigorous analysis of these results is complicated due to the complex nature of the ion damage in Si. A computer simulation framework is developed to aid in the analysis of this data. In this technique, defects are simulated and their strain fields are calculated by simply relaxing the atoms around the defect to their equilibrium positions. The diffuse scattering is then calculated from the strain field, and the results are compared to the experimental measurements. Computer simulations are presented here only for the case of electron irradiation damage and compared to published measurements. Application of the technique to more complicated structures is planned and should pose no serious problems in the computational framework already developed.

The effects of vacancy-type defects induced by ion implantation on the electricalactivation of implanted phosphorus by rapid thermal annealing (RTA) are investigated using the positron annihilation technique, secondary ion mass spectroscopy (SIMS) and the spreading resistance (SR) method. P÷ ions are implanted into bare Si wafers and into Si through Si02 films at 700 keV with doses of the order of l×1013 cm−2. After implantation, rapid thermal annealing (RTA) is performed at temperatures between 600 and 1100 °C for 20 sec. The result shows that vacancy-type defects compensate the electrical-activation of P implanted into Si and also recoiled- oxygen is affected on the electrical-activation of P. The species of main defects for compensating the electrical-activation is identified as a divacancy (V2) from the lifetime of positrons. Effects of recoiled oxygen on the electricalactivation are attributed to the formation of vacancy-oxygen complexes just below the SiO2/Si interface and a resultant decrease in the diffusion length of vacancy-type defects.

We present for the first time a physically based ion implantation damage model which successfully predicts both the as-implanted impurity range profiles and the damage profiles for a wide range of implant conditions for arsenic, boron, phosphorus, and BF2 implants into single-crystal (100) silicon. In addition, the amorphous layer thicknesses predicted by this damage model for high dose implants are also generally in excellent agreement with experiments. This damage model explicitly simulates the defect production and its subsequent evolution into the experimentally observable profiles for the first time. The microscopic mechanisms for damage evolution are further discussed.

An electronic stopping power model for boron, arsenic, and phosphorus ion implantation into single-crystal Si is reported over the energy range fr'om a few keV to several MeV, for both offand on-axis implant angles relative to the <100> crystallographic direction. Combined with previously developed models for damage accumulation, this model allows physically-based simulation of 3-D profiles over an extremely wide range of implant conditions. In particular, this allows modeling of MeV implants which are being used more and more frequently.

Irradiation of materials with energetic particles produces changes in the microstructure that affect mechanical properties. In previous work we studied the thermal aspects of the quenching of collision cascades that involve nanoscale phase transitions between the solid and the liquid states of the target. In this work we present a rigorous treatment of these phenomena, including a detailed description of the Stefan problem in three dimensions and diffusion in thermal gradients. This approach is oriented to give a quantitative description of the influence of the primary knockon spectrum on the microstructure short after the quenching of the heat spike.

The 2D-dopant and defect distributions resulting from 80 keV ion implantation of As+ ions into Si through a high-edge mask are presented. The distributions are obtained by means of an efficient computer procedure using the results of Monte Carlo simulation. Two versions of the computer code TRIM are used. The 2D-target atom redistribution is obtained as a result of cascade collisions. The simulation reveals the effect of near-mask-edge target atom depletion. This effect is related to the recoil phenomena and can be explained on the basis of simple model.

We combine molecular dynamics and Monte Carlo simulations to study damage accumulation and dose rate effects during irradiation of Silicon. We obtain the initial stage of the damage produced by heavy and light ions using classical molecular dynamics simulations. While heavy ions like As or Pt induce amorphization by single ion impact, light ions like B only produce point defects or small clusters of defects. The amorphous pockets generated by heavy ions are stable below room temperature and recrystallize at temperatures below the threshold for recrystallization of a planar amorphous-crystalline interface. The damage accumulation during light ion irradiation is simulated using a Monte Carlo model for defect diffusion. In this approach, we study the damage in the lattice as a function of dose and dose rate. A strong reduction in the total number of defects left in the lattice is observed for lower dose rates.

The effects of in situ interventions which alter defect interactions during implantation, and thereby affect the final damage state, have been investigated. Specifically, we examined the effects of internal electric fields and charge carrier injection on damage accumulation in silicon. In the first part of this work, we implanted H or He ions into diode structures which were either reverse or forward biased during implantation. In the second part, we implanted B or Si ions into plain silicon wafers whilst illuminating them with UV light. In each case, the overall effect is one of damage reduction. Both the electric field and charge carrier injection effects may be understood as resulting from changes in defect interactions caused in part by changes to the charge state of defects formed during implantation.