No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
A Comprehensive Model for the Diffusion of Boron in Silicon in Presence of Fluorine
Abstract
We present a fully calibrated and parameterized model for the diffusion of boron in silicon in the presence of fluorine. This model reproduces experiments for a wide range of different experimental conditions including non-amorphizing, partially amorphizing, and fully amorphizing ion implantation. It is physical in the sense that fluorine interacts with boron indirectly via point defects. The model extends existing work consistently and is computationally more efficient. It should be suitable for the description of a large variety of experimental conditions from the emerging field of BF3 plasma implantation in photovoltaics to the field of ultra-shallow junctions.
... Defining an efficient coanneal process is therefore a non-trivial task that can be optimized with the help of simulation. Simulation of B annealing with current tools [6] is predictive for standard implantation conditions and since recently even for fluorine containing plasma implantation [7]. The situation is very different for P, for which postimplantation annealing has not been as thoroughly investigated due to the lower relevance of P for the formation of ultra shallow junctions, and for which modeling is particularily complicated as both interstitials and vacancies contribute to diffusion. ...
... The ratios of transport capacities (7) are shown in their dependences on the P concentration in Fig. 3 The free (active) P concentration C free P is obtained from the total P concentration C P via Eq. (11). ...
We present a fully calibrated model for the diffusion, segregation, and activation of phosphorus for typical annealing conditions of implanted silicon solar cells. In contrast to existing process simulation software, this model allows the quantitative prediction of doping profile distributions, and, thereby, sheet resistances, surface concentrations, and junction depths. The model also provides an intuitive understanding of the dependence of these quantities on the parameters of the annealing process. A post-print version of the article can be found at publica.fraunhofer.de/documents/N-322023.html
... It should be suitable for the description of a large variety of experimental conditions from the emerging field of BF 3 plasma implantation in photovoltaics to the field of ultra-shallow junctions. The first four sections of this chapter have been published in Solid-State Electronics ( Wolf et al., 2013). ...
Ion implantation technology has become economically competitive for solar cell doping in about 2012. The decisive reduction in production costs was achieved by using plasma implantation
tools, which allow a throughput of more than 2000 wafers per hour. Implantation technology now opens the door for sophisticated cell concepts with high conversion efficiencies. The present thesis clarifies key questions in the simulation of the associated annealing processes. This is achieved by developing models for the reaction and diffusion kinetics of implanted phosphorus, implanted boron and implantation-induced dislocation loops. These models are based on conceptual ideas that originate from the field of microelectronics but have hitherto not been applied to the considerably different fabrication conditions of solar cells. Compared to standard solar cell doping using POCl3 diffusion, ion implantation allows a much better control of the doping process. Besides its immediate advantages though, it comes with the drawback of inevitable implantation damage. This work investigates the question of how implantation
damage, in particular dislocation loops, has to be annealed to obtain high-performing solar cells. To address this question, a model for the reaction kinetics of dislocation loops has been
developed and shown to be viable for characteristic situations of solar-cell processing. This model is the first one to capture the late stages of Ostwald ripening and the transition from faulted to
perfect dislocation loops. The two primary dopant species of interest for solar-cell doping by implantation are phosphorus
and boron. Due to the complexity of the diffusion and segregation behavior of phosphorus, its annealing behavior is still under debate today and no generally predictive models exist. In
this thesis, a predictive model for phosphorus annealing suitable for the particular requirements of solar cell processing has been developed, calibrated and applied. As part of the model development, a general strategy for an effcient calibration of models describing impurity diffusion via two species has been found. The annealing behavior of boron, on the other hand, is well
understood and highly predictive models exist. The plasma implantation tools of the solar cell industry though usually implant BF3 instead of atomic boron. The diffusion behavior of boron in
presence of uorine differs fundamentally from that without
uorine and no generally valid model for this situation has been available previously. In the present thesis, a comprehensive model for this problem has been implemented and successfully tested. Finally, the models for the annealing behavior of implantation damage, phosphorus and boron were studied with respect to their relevance for the electronic properties of solar cells. In particular, simulated predictions of the dislocation-line density and its associated recombination activity have been compared with recombination-current measurements from a large-scale experimental study performed at the University of Hannover and the ISFH Hameln. The quantitative agreement obtained provides, for the first time, strong evidence that dislocation loops are responsible for implantation damage-related degradation of solar cells. Combining futhermore process
simulation with device simulation, main factors that in
uence solar-cell performance could be investigated. This allowed to verify assertions from the literature as well as suggesting processing options.
... The reason for the reduction of TED lies in the ability of the co-implanted species to capture Int, which otherwise would contribute to Int-mediated dopant diffusion. Many simple and complex models have been developed to describe the interaction of C [13] [14], F [14]- [17], and N [14][18] on point defects. ...
Technology Computer-Aided Design is widely used for the development and optimization of advanced device formation. Ion implantation and thermal annealing are the main focus of dopant profile simulation for process technologies. In this paper, we review the current continuum modeling capabilities and calibration for ion implantation and thermal processes, including Monte Carlo ion implantation, co-implantation, amorphization, recrystallization, damage evolution, and dopant diffusion and activation. In addition, modeling of alternative doping techniques such as thermal implantation, plasma doping, and melt laser annealing will be addressed. Continuum front-end process simulation of Si-based devices including advanced CMOS, memory, power, and optoelectronic devices is considered to be mature. With the introduction of new channel materials, the models and calibration of alternative materials such as SiGe, Ge, and III–V are also required and their current status is discussed.
The effect of fluorine (F) on diffusion of boron (B) in silicon (Si) is investigated by secondary ion mass spectrometry of Si, B, and F diffusion using pre-amorphized natSi/²⁸Si isotope multilayers that are co-implanted with B and F. By the presence of F, diffusion of B is suppressed while that of Si is enhanced. A quantitative analysis of the experimental results based on our diffusion model shows that the suppression of B diffusion is due to (1) Si interstitial undersaturation caused by the time-dependent formation and dissolution of F-vacancy (FV) clusters and (2) direct interaction between B and FV clusters. The model developed in this study enables an accurate simulation of B and Si diffusion in the presence of F in Si.
Ion implantation is a technique that has been demonstrated to improve solar cell efficiency and eliminate process steps in standard and advanced cell designs. Intevac has developed a high productivity, continuous flux ion implantation tool for solar cells. We demonstrate improved n-type emitters over POCl3 diffused emitters, and selective patterning capabilities. Additionally, it is shown that non-mass analyzed implantation provides similar performance as mass-analyzed implantation, yet at a much lower capital cost.
A potentially cost-effective ion implanter for solar cells has become commercially available very recently. As the emitter dopant profiles differ from the standard diffusions, a combination of process simulation and device simulation is used to predict possible applications as front emitter. The simulations show that ion energies of 10 to 30 keV and doses in the range of 5x10^14 to 7x10^15 cm^-2 are sufficient for reducing the phosphorus peak density and, hence, obtaining cell efficiency levels above 20%, if appropriate surface passivation and wafer materials are used. The simulations strongly indicate, however, that cell efficiency improves only marginally if the cell has a fully metallized rear Al-BSF and a boron-doped Cz base in the degraded state. Simulated cells with a local rear Al-BSF show an efficiency improvement of more than 0.3% absolute in the degraded state.
Ion implants of 1 keV B-11(+) and 5 keV BF2+, to a dose of 1 x 10(15)/cm(2) at a tilt angle of 0 degrees, were implanted into preamorphized (Si+, 70 keV, 1 x 10(15)/cm(2)) wafers. These samples were rapid thermal annealed in an ambient of 33 ppm of oxygen in N-2 at very short times (<0.1 s spike anneals) at 1000 and 1050 degrees C to investigate the effects of the fluorine in BF2 implants on transient enhanced diffusion (TED). By using a relatively deep preamorphization of 1450 Angstrom, any difference in damage between the typically amorphizing BF2 implants and the nonamorphizing B implants is eliminated because the entire profile (<800 Angstrom after annealing) is well contained within the amorphous layer. Upon annealing, the backflow of interstitials from the end-of-range damage from the preamorphization implant produces TED of the B in the regrown layer. This allows the chemical effect of the fluorine on the TED of the B in the regrown Si to be studied independent of the damage. The secondary ion mass spectroscopy results show that upon annealing, the presence of fluorine reduces the amount of B diffusion by 30% for the 1000 degrees C spike anneal, and by 44% for the 1050 degrees C spike anneal. This clearly illustrates there is a dramatic effect of F on TED of B independent of the effects of implant damage. Analysis of the temperature dependence of the enhancement factors point to transient enhanced diffusion not boridation as the source of the interstitials. (C) 1998 American Institute of Physics. [S0003-6951(98)03035-6].
The transient enhanced diffusion of acceptor impurities severely affects the realization of ultrahigh doping regions in miniaturized Si-based devices. Fluorine codoping has been found to suppress this transient diffusion, but the mechanism underlying this effect is not understood. It has been proposed that fluorine-impurity or fluorine–native-defect interactions may be responsible. Here we clarify this mechanism combining first-principles theoretical studies of fluorine in Si and purposely designed experiments on Si structures containing boron and fluorine. The central interaction mechanism is the preferential binding of fluorine to Si-vacancy dangling bonds and the consequent formation of vacancy-fluorine complexes. The latter effectively act as traps for the excess self-interstitials that would normally cause boron transient enhanced diffusion. Instead, fluorine-boron interactions are marginal and do not play any significant role. Our results are also consistent with other observations such as native-defect trapping and bubble formation.
The evaporation of {311} self‐interstitial clusters has recently been linked to the phenomenon of transient enhanced diffusion in silicon. A theory of cluster evaporation is described, based on first‐order kinetic equations. It is shown to give a good account of the data over a range of temperatures. The theory simultaneously explains several of the unexpected features of transient enhanced diffusion, including the apparently steady level of the enhancement during its duration, and the dependence of the duration on implant energy and dose. The binding energy used to match the theory to data is in good agreement with molecular dynamics calculations of cluster stability in silicon. © 1996 American Institute of Physics.
We have investigated the diffusion enhancement mechanism of boron-enhanced diffusion (BED), wherein boron diffusivity is enhanced four to five times over the equilibrium diffusivity at 1050 °C in the proximity of a silicon layer containing a high boron concentration. It is demonstrated that BED is driven by excess interstitials injected from the high boron concentration layer during annealing. For evaporated layers, BED is observed above a threshold boron concentration between 1% and 10%, though it appears to be closer to 1% for B-implanted layers. For sub-keV B implants above the threshold, BED dominates over the contribution from transient-enhanced diffusion to junction depth. For 0.5 keV B, this threshold implantation dose lies between 3×1014 and 1×1015 cm−2. It is proposed that the excess interstitials responsible for BED are produced during the formation of a silicon boride phase in the high B concentration layers. © 1999 American Institute of Physics.
The local structure of fluorine incorporated in crystalline silicon following solid phase epitaxial regrowth was investigated by means of x-ray absorption spectroscopy at the F K-edge. We clearly demonstrate that most F is found in SiF4 molecules in the crystalline matrix. A kinetic pathway, which explains our observation and which is also able to rationalize previous results in a common and coherent framework, is proposed.
We report on the F incorporation into Si during solid-phase epitaxy (SPE) at 580 °C and with the presence of B and/or As, clarifying the F incorporation mechanism into Si. A strong segregation of F at the moving amorphous-crystalline interface has been characterized, leading to a SPE rate retardation and to a significant loss of F atoms through the surface. In B- or As-doped samples, an enhanced, local F incorporation is observed, whereas in the case of B and As co-implantation (leading to compensating dopant effect), a much lower F incorporation is achieved at the dopant peak. The F enhanced incorporation with the presence of B or As is shown to be a kinetic effect related to the SPE rate modification by doping, whereas the hypothesis of a F-B or F-As chemical bonding is refused. These results shed new light on the application of F in the fabrication of ultrashallow junctions in future generation devices.
We investigated the effect of F on the electrical activity of B-doped junctions in preamorphized Si. It is shown that while the carrier dose introduced by B is reduced in the presence of F, no indication of B-F complexes formation can be found and B maintains its full substitutionality. Investigations on F-enriched crystalline Si demonstrated and quantified the n-type doping of F. These results clarify that the loss of holes in junctions coimplanted with B and F is not due to a chemical interaction between B and F, but simply to a dopant compensation effect.
The formation and evolution of F-induced nanobubbles in Si was investigated. Si samples were preamorphized, implanted with F, and partially regrown by solid phase epitaxy (SPE). It is shown that nanobubbles are formed already in the amorphous side of partially regrown samples and are then incorporated in crystalline Si during SPE. The bubbles are interpreted as the result of the diffusion and coalescence of F atoms and dangling bonds already in the amorphous matrix. During high temperature annealing after SPE, F outdiffuses; correspondingly, the bubbles partially dissolve and transform from spherical- to cylinder-shaped bubbles.
The redistribution of fluorine during solid phase epitaxial regrowth (SPER) of preamorphized Si has been experimentally investigated, explained, and simulated, for different F concentrations and temperatures. We demonstrate, by a detailed analysis and modeling of F secondary ion mass spectrometry chemical-concentration profiles, that F segregates in amorphous Si during SPER by splitting in three possible states: (i) a diffusive one that migrates in amorphous Si; (ii) an interface segregated state evidenced by the presence of a F accumulation peak at the amorphous-crystal interface; (iii) a clustered F state. The interplay among these states and their roles in the F incorporation into crystalline Si are fully described. It is shown that diffusive F migrates by a trap limited diffusion mechanism and also interacts with the advancing interface by a sticking-release dynamics that regulates the amount of F segregated at the interface. We demonstrate that this last quantity determines the regrowth rate through an exponential law. On the other hand we show that neither the diffusive F nor the one segregated at the interface can directly incorporate into the crystal but F has to cluster in the amorphous phase before being incorporated in the crystal, in agreement with recent experimental observations. The trends of the model parameters as a function of the temperature are shown and discussed obtaining a clear energetic scheme of the F redistribution and incorporation in preamorphized Si. The above physical understanding and the model could have a strong impact on the use of F as a tool for optimizing the doping profiles in the fabrication of ultrashallow junctions.
The simulation of deep-submicron silicon-device manufacturing processes relies on predictive models for extended defect clusters. For submicroscopic interstitial clusters and {311} defects, an efficient and highly accurate model for process simulation has been developed and calibrated recently [1]. This model combines equations for three small interstitial clusters and two moments for {311} defects. In this work, we extend this model to include dislocation loops and to reproduce a greatly increased range of experimental data, including thermal annealing of end-of-range defects after amorphizing implants.
Thermal annealing after preamorphization and solid-phase epitaxy of ultrashallow B implants leads to deactivation and diffusion driven by interstitials released from end-of-range defects. F inhibits these processes by forming small clusters that trap interstitials. A competing B-F interaction causes deactivation when F and B profiles overlap. Both pathways suppress B transient enhanced diffusion.
We show direct evidence, obtained by positron annihilation spectroscopy, for the complexing of fluorine with vacancies in silicon. Both float zone and Czochralski silicon wafers were implanted with 30 keV fluorine ions to a fluence of 2×1014 ions / cm 2 , and studied in the as-implanted condition, and after annealing to 650 ° C for 10 and for 30 min . The “2-detector” background reduction technique for positron annihilation was applied. The spectra reveal a significant concentration of fluorine-vacancy complexes after annealing, for both Czochralski and float zone material, supporting the results of computer simulations of the implantation and annealing process.
We have explained the role of fluorine in the reduction of the self-interstitial population in a preamorphized Si layer under thermal treatment. For this purpose, we have employed a B spike layer grown by molecular-beam epitaxy as a marker for the self-interstitial local concentration. The amorphized samples were implanted with 7×1012, 7×1013, or 4×1014 F/cm 2 at 100 keV, and afterwards recrystallized by solid phase epitaxy. Thermal anneals at 750 or 850 ° C were performed in order to induce the release of self-interstitials from the end-of-range (EOR) defects and thus provoke the transient enhanced diffusion of B atoms. We have shown that the incorporation of F reduces the B enhanced diffusion in a controlled way, up to its complete suppression. It is seen that no direct interaction between B and F occurs, whereas the suppression of B enhanced diffusion is related to the F ability in reducing the excess of silicon self-interstitials emitted by the EOR source. These results are reported and discussed. © 2004 American Institute of Physics.
An electron paramagnetic resonance (EPR) study on fluorine-vacancy defects (FnVm) in fluorine-implanted silicon is demonstrated. Fluorine implantation is an important technology for Si microdevices and EPR measurements showed that this process created a variety of FnVm defects of different sizes (V2, V4, and V5). In FnVm, a Si–F bond exhibited a different chemical nature compared to a Si–H bond in hydrogen-vacancy complexes. The most primitive defect was FV2 (F0 center) and the final types were FnV5 (F1 center) and FnV2 (F2 center) which increased in annealing processes as low temperature as 200 °C.
A systematic and physically based method for extracting of a unified parameter set for a point-defect diffusion model is proposed. A sensitivity matrix analysis is used to construct the sequence of the extraction and to select the data sot to be fitted.
The International Roadmap for Semiconductors requires ultrashallow, highly activated, abrupt dopant profiles in the source/drain extension regions, for technology nodes beyond 45 nm. The authors contrast B and BF2 implants in Si and silicon on insulator (SOI) substrates with and without a preamorphizing implant (PAI). The objective of the study is to compare between Si and SOI substrates, PAI and non-PAI condition, and B and BF2 implants. The results show the absence of the “reverse annealing effect” in BF2 implants, which is observed in B implants. The presence of F appears to impede the formation of boron interstitial clusters, which is shown in the case of B implant. The BF2 implants follow a similar trend for SOI and Si with and without PAI.
The lattice location of F atoms in Si was experimentally studied. Si single crystals were amorphized, implanted with F, and afterwards the amorphous layer was recrystallized. Some of the samples prepared in this way were also annealed at 750 °C for 60 min. The 19F(p,αγ)16O resonant nuclear reaction at 340.5 keV was employed to measure the probability of a close encounter between protons and F nuclei as a function of the incident angle with respect to six major crystalline directions. The predictions of several ab initio calculations proved to be incompatible with the present experimental findings.
In this study, we examined the diffusion of boron near the projected
ranges of high-dose 11B and BF2 ions implanted in
silicon. A buried layer of 10B was formed and the
redistribution of 10B was used to monitor the diffusion
behavior. Boron precipitation at 850 °C produces the pileup of
10B in the peak region of 11B. The formation of a
band of boron clusters was indicated by the accumulation of
10B between the projected range and the diffusion tail of the
11B profiles during annealing at 750 °C, following
11B+ implantation. The 10B
redistribution at 750 °C also demonstrated the retardation of
transient enhanced diffusion (TED) in the peak region. Under
BF2+ implantation, TED retardation near the
projected range of 11B is suppressed by the decrease in
implantation energy, because the surface proximity effect reduces the
density of end-of-range (EOR) defects at low implantation energies.
Boron clustering is suppressed in front of the diffusion tail of
11B because only a peak of 10B at the EOR
dislocations exists at 750 °C following BF2+
implantation.
Effects of annealing on the properties of P- and B-implanted Si for interdigitated back contact (IBC) solar cells were investigated with annealing temperature of from 950 to 1050 °C. P-implanted samples annealed at 950 °C were enough to activate dopants and recover the damage by implantation. As the annealing temperature was increased, the diode properties of P-implanted samples were degraded, while that of B-implanted samples were improved. However, in order to activate an implanted B ion, B-implanted samples needed an annealing of above 1000 °C. The implied Voc of lifetime samples by quasi-steady-state photoconductance decay followed the trend of diode properties on annealing temperature. Finally, IBC cell was fabricated with a two-step annealing at 1050 °C for B of the emitter and 950 °C for P of the front and back surface fields. The IBC cell had Voc of 618 mV, Jsc of 35.1 mA/cm2, FF of 78.8%, and the efficiency of 17.1% without surface texturing.
We have investigated and modelled the diffusion of boron implanted into crystalline silicon in the form of boron difluoride BF2+. Low-energy BF2+ (2 keV) and dose of 1 × 1015 cm−2 have been used. Diffusion profiles have been measured using secondary ion mass spectrometry (SIMS). RTA were carried out at 950, 1000, 1050 and 1100 °C. The results show that concentration profiles for BF2+ implant are slighty shallower than those for a direct B+ ion implantation. This could be attributed to the presence of fluorine, which can trap interstitial silicon so that Si supersaturation is low near the surface. Following Uematsu's works, the simulations satisfactory reproduce the SIMS experimental profiles.
We have simulated the transient enhanced diffusion (TED) of boron (B) after
amorphizing BF2 ion implantation in silicon. A unified simulation is done based on
the models for B diffusion, for TED by self-interstitial clusters, for B clustering and B
precipitation, and for end-of-range (EOR) defects. The simulation overestimates the
diffusion using the normal values for the efficiency of EOR defects as a source of
self-interstitials. The simulation well reproduces the experimental profiles when the
efficiency is reduced so that the defects maintain self-interstitial concentration at
thermal equilibrium values. This reduction is attributable to the presence of fluorine at
EOR defect sites, which may prevent the release of self-interstitials. In addition, the
second peak near the amorphous/crystalline (a/c) interface observed in experimental
profiles is reproduced, and the peak is attributed to B precipitates.
We have simulated the transient enhanced diffusion (TED) of high-concentration phosphorus (P) in silicon during post-implantation annealing. Based on the models for P diffusion, for TED by self-interstitial clusters, and for end-of-range (EOR) dislocations as both a sink for and source of self-interstitials, a unified simulation is done, taking into account P clustering and P pile-up. P clustering is taken into account only beneath EOR dislocations, and P pile-up is estimated by a diffusion-segregation term in the diffusion equations. We have satisfactorily fitted P depth profiles at high doses (˜1015 cm-2) in a wide range of annealing conditions (700 1000°C).
Plasma immersion ion implantation is an alternative doping technique for the formation of Ultra Shallow Junctions in semiconductor. In this study, we present the PIII technology developed by the company Ion Beam Services and called PULSION®. We explain the advantages of PIII for the conception of thin emitter solar cells and the use of N type silicon in the fabrication of photodiode. Electrical characterisations of solar cells prepared by immersion of silicon wafer in BF3
plasma are presented, showing a satisfying photovoltaic behaviour and more specially an increase of internal quantum efficiency in the short wavelength range, due to the thickness of the emitter.
We have simulated the transient enhanced diffusion (TED) of boron fluoride (BF2+) implanted in crystalline and germanium amorphized silicon. Based on recently published models, the effect of fluorine on boron diffusion in silicon has been introduced and a modelling has been suggested. In order to simulate the boron experimental profiles, we have assumed that fluorine forms clusters involving interstitial boron which reduces the junction depth. Experimental results indicate that fluorine behaviour depends on amorphization energy. Moreover, even no germanium preamorphization is performed, silicon is still amorphized by fluorine species. Hence, BF2+ implantation leads to an amorphous/crystalline (a/c) interface near the surface. An improvement of published models is suggested taking into account fluorine effects. The simulations satisfactory reproduce the SIMS experimental profiles for a large scale of experimental conditions.
The authors performed systematic ab initio calculations of fluorine clustering in silicon. The calculated formation energies were used to implement a new kinetic Monte Carlo (KMC) model. They present the ab initio results, discuss the new KMC model, and compare the resulting simulated profiles to experimental profiles. The calculated formation energies show clear trends with the number of missing silicon atoms and the number of fluorine atoms. The deduced KMC model based on the ab initio energetics is able to reproduce the reduction in boron transient enhanced diffusion in the presence of fluorine.
The ab initio pseudopotential method was used to study the boron diffusion and pairing process in crystalline silicon. The results show that substitutional B attracts interstitial Si with a binding energy of 1.1 ± 0.1 eV. We show that B diffusion is significantly enhanced in the presence of the Si interstitial due to the substantial lowering of the migrational barrier through most likely a kick-out mechanism. The resulting mobile boron can also be trapped by another substitutional boron with a binding energy of 1.8 ± 0.1 eV, forming an immobile and electrically inactive two-boron pair along a 〈001〉 direction. It is also found that the pairing of these two boron atoms involves the trapping of a Si interstitial. Alternatively, two B pairs that do not trap the Si interstitial were found to be energetically unfavorable. All of these findings are consistent with experimental results. © 1996 The American Physical Society.
The transient‐enhanced diffusion of phosphorus in silicon has been investigated for doses and energies below the threshold
for amorphization of the substrate. Transient‐enhanced diffusion occurs both for furnace and RTA annealing, and can increase
the junction depth by 0.1–0.2 μm over that predicted by standard diffusion models. The magnitude of the transient is shown
to depend on the implantation dose and energy, the anneal time and temperature, and the background doping level. A simple
model of damage annealing is proposed which quantitatively describes the point defect enhancement of dopant diffusion and
so allows each of these dependencies to be understood. In addition, these experiments provide a new means to characterize
the diffusion and charge states of the point defects themselves.
Based on ab initio density-functional energetics for saturated (n = 2m+2) fluorine-vacancy clusters FnVm for m up to 4, the authors set up a model showing that (a) fluorine-vacancy (F–V) aggregates in Si can form in any size and concentration for sufficient concentrations of incorporated (e.g., implanted) F, and (b) the F to V ratio in F–V complexes (i.e., the inverse capture efficiency of self-interstitials) is an ensemble average over many cluster sizes. It ranges between 4 and 2, with typical values of 2.2–2.5, consistent with recent experimental estimates.
In this article, the postimplanted fluorine effect on boron transient-enhanced diffusion (TED) and dose loss during a 750°C annealing is shown. 19F implants at 2 keV, after 11B implant at an energy of 1 keV,3×1014∕cm2, have been investigated in the dose range between 1×1013∕cm2 and 6×1014∕cm2 without a preamorphizing implant. Below a F-implant dose of 1×1014∕cm2, the reduction or non-enhancement of boron-TED effect is observed. In the case of a F-implant dose of 6×1014∕cm2, the enhanced boron TED (∼2.6×) in crystalline Si and the increased dose loss(∼2×) than that of a normal boron TED is shown, and this anomalous diffusivity enhancement persists for 120 min at 750°C. The B+F 6×1014∕cm2 consecutive implant damage is smaller than that of the BF2 5 keV implant. In the case of the B+F 6×1014∕cm2, a high content of fluorine is retained around the end-of-range (EOR) damage region within 120 min. These results indicate that the fluorine retained around the EOR region may affect the enhanced boron TED in crystalline Si at 750°C. Boron-diffusion model, which describes both the native interstitial fluorine and the boron-fluorine chemical effect, explains that the suppression of boron TED with a low F-implant dose is due to the reduction of the interstitial supersaturation.
We have compared the electrical characteristics and the depth profile of ultrashallow junctions formed by boron implantation at 0.5 keV and BF2 implantation at 2.2 keV. The modeling of the boron profile was performed using the Monte Carlo method for an as-implanted profile and the computationally efficient method for transient-enhanced diffusion. A junction depth of BF2 is shallower than that of boron after annealing. For an ultrashallow junction, HF dipping prior to rapid thermal annealing causes a significant loss of dopant and high sheet resistance. Considering the 0.1 μm metal–oxide–semiconductor field-effect transistor application, the optimizations of implantation and annealing conditions are necessary to satisfy the requirement of junction depth and sheet resistance. © 1999 American Institute of Physics.
The beneficial effects of F implantation on the modification of extended defects in Si have been studied. Preamorphized Si samples were implanted with F (75 keV, 6×1015 F/cm2) and regrown by solid phase epitaxy (SPE) at 700 °C. The formation, just after SPE, of a band of bubbles overlapping the F enriched region has been evidenced, clearly demonstrating the formation of F-vacancy (V) complexes with determined stoichiometry. Moreover, the authors demonstrate that these F-V complexes inhibit the formation of extended defects, acting as efficient traps for Si interstitials. These results represent a promising route toward point defects engineering in microelectronic application.
While it is known that F modifies dopant diffusion in crystalline Si, the physical mechanisms behind this process are still unclear. In this work we report experimental studies about the F control of the point defect density in preamorphized Si layers. These studies put the basis for the understanding of the F behavior and for the realization of ultra-shallow junctions. We first investigated the F incorporation process during the solid phase epitaxy (SPE) of amorphous Si layers. We elucidated the role of the SPE temperature on the F incorporation and suggested a new route towards a F profile engineering. Moreover, we explained the role of F in modifying the point defect population (self-interstitials, Is, and vacancies, Vs), employing B and Sb spike layers as markers for Is and Vs, respectively. We clearly showed that F decreases the B diffusion while enhances the Sb one, pointing out the capacity to induce an Is undersaturation or a Vs supersaturation. These data rule out the hypothesis of a chemical bonding between F and the dopants. Such F ability in modifying the Is/Vs density resulted to be a transient effect, because strictly correlated with the presence of F in the Si samples, which decreases with the annealing time. In addition, we evidenced that even if F is spatially separated from B, i.e., localized between shallow-implanted B and the end-of-range (EOR) region, it still suppresses the enhancement of B diffusivity, due to the EOR defects dissolution. These studies, besides improving the current understanding of the physical mechanisms by which F influences the dopant diffusion in Si, could be helpful for the realization of ultra-shallow junctions for the future metal-oxide-semiconductor devices.
Low-thermal-budget annealing of ion-implanted BF
2+, P, and As in Si was studied for shallow-junction formation. Implant doses were sufficient to amorphize the silicon surface region. Low-temperature furnace annealing and rapid-thermal annealing of ionimplanted boron, phosphorus and arsenic in silicon exhibit a transient enhanced diffusion regime resulting injunction depths considerably deeper than expected. The origin of this transient enhanced diffusion is the annealing of ion-implantation damage in the silicon substrate. We have found that point-defect generation during the annealing of either shallow end-of-range damage or small clusters of point defects dominates the transient enhanced diffusion process depending upon the annealing temperature and time. The net effect of damage annealing is to reduce the activation energy for dopant diffusion by an amount equal to the activation energy of the supersaturation of point defects in silicon. Models which can describe the transient enhancement characteristics in dopant diffusion during both furnace and rapid-thermal annealing of these implants are discussed.
The effects of low‐dose ion implants with Si+, Ne+, and F+ on the transient enhanced diffusion of B in silicon after annealing at 900 °C for 30 min have been investigated. Processing conditions such as implant dose (3.5×1013 cm-2) and energy (30–60 keV) were chosen to simulate the lightly doped drain implant in a 0.35 μm complementary metal‐oxide‐semiconductor technology. An epitaxially grown B‐doping superlattice is used to extract directly depth profiles of average Si self‐interstitial concentration after processing. For Si+ the transient enhanced diffusion of B increases with the energy of the implanted ion. Ne+ implanted with the same energy as Si+ causes more transient enhanced diffusion, while Ne+ implanted with the same range as Si+ causes slightly less. Implantation of F+ enhances the B diffusivity considerably less than Si or Ne implantation. These effects were modeled using simulations of defect diffusion in the presence of traps. A trap concentration of (2.4±0.5)×1016 cm-3 gave good agreement in all situations except F+ implantation, where (6.6±0.6)×1016 cm-3 traps were necessary. It is proposed that this is caused by additional traps for Si interstitials that are related to F+. © 1995 American Institute of Physics.
The diffusion of B implanted in Si has been investigated at different concentrations in a wide range of experimental conditions (temperature from 800 to 1000 °C and time from 10 s to 8 h) by using furnace and rapid thermal treatments. In particular, the transient enhanced diffusion induced by the implantation damage in the early phase of the annealing and the precipitation occurring in concomitance with the diffusion for dopant concentration exceeding the solid solubility have been extensively analyzed. A simulation program taking these phenomena into account has been developed by modifying the supreme iii code. A satisfactory agreement with experimental data has been obtained for all the investigated conditions. The model represents a significative improvement of the diffusion simulation of B implanted in crystalline Si. In fact, the more commonly used codes of process simulation do not evaluate adequately the effects of the above considered phenomena.
Characteristics of rapid thermal and pulsed laser annealing have been investigated in boron fluoride‐ (BF+ 2 and BF+ 3 ) ‐implanted silicon using cross‐section and plan‐view electron microscopy. The amorphous layers recrystallize by the solid‐phase‐epitaxial growth process, while the dislocation loops below the amorphous layers coarsen and evolve into a network of dislocations. The dislocations in this band getter fluorine and fluorine bubbles associated with dislocations are frequently observed. The secondary‐ion mass spectrometry techniques were used to study concomitant boron and fluorine redistributions. The as‐implanted Gaussian boron profile broadens as a function of time and temperature of annealing. However, the fluorine concentration peak is observed to be associated with dislocation band, and the peak grows with increasing time and temperature of annealing. The electrical properties were investigated using van der Pauw measurements. The electrical activation of better than 90% and good Hall mobility were observed in specimens with less than 500‐Å dopant‐profile broadening. In pulsed laser‐annealed specimens, the boron profile broadens both toward the surface and into the deeper regions of the crystal. However, the fluorine concentration profile exhibits a decrease in peak concentration with only a limited broadening.
Depth distributions, measured by secondary ion mass spectrometry (SIMS),and carrier profiles, measured by differential capacitance-voltage (C-V) profiling, are reported for boron and fluorine implanted as B, F, BF, or BF//2 ions into random and channeling orientations of crystalline silicon, and into silicon amorphized by silicon ion implantation. Low boron energies of 8 and 10 keV and the corresponding energies of 36 and 45 keV for BF//2 ions are emphasized because of their use for high resolution device and circuit applications in silicon and silion-on-sapphire. Amorphizing crystalline silicon prior to boron implantation eliminates the significant channeling tails on 8- or 10-keV boron profiles. Fluorine penetrates more deeply into crystalline silicon than boron does. Both boron and fluorine redistribute during annealing, but with quite different characteristics, and depending on the implantation fluence. The fluorine redistribution profiles are strongly influenced by the magnitude and distribution of damage that remains after annealing. Fair agreement is shown between boron atom depth distributions measured by SIMS and C-V electrical profiles. Pearson IV moments are given for the low energy boron profiles. The use of these profiles for modeling calculations is discussed. The SUPREM model of an exponential for the channeling tail of boron implants in crystalline silicon is also evaluated.
Electrical properties of recrystallized amorphous silicon layers, formed by BF+ 2 implants or Si++B+ implants, have been studied by differential resistivity and Hall‐effect measurements. Electrical carrier distribution profiles show that boron atoms inside the amorphized Si layers can be fully activated during recrystallization at 550 °C. The mobility is also recovered. However, the tail of the B distribution, located inside a damaged region near the original amorphous‐crystalline interface, remains inactive. This inactive tail has been observed for all samples implanted with BF+ 2 . Only in a thicker amorphous layer, formed for example by Si+ predamage implants, can the entire B profile be activated. The etch rate of amorphous silicon in HF and the effect of fluorine on the recrystallization rate are also reported.
The time evolution of B diffusion and electrical activation after ion implantation and annealing at 800 and 900 °C is studied using secondary‐ion mass spectrometry and spreading‐resistance profiling. The time evolution at 800 °C is observed in both crystalline and post‐amorphized samples. Amorphized samples show near‐normal concentration enhanced diffusion. Crystalline samples show anomalous transient diffusion, with a rapidly diffusing low‐concentration region and a static peak region above a critical concentration C enh =3.5×1018 cm-3. The peak region above C enh is shown to be electrically inactive. The static, inactive B is released over a period of many hours, compared with the transient diffusion enhancement which relaxes to near‐normal within 30 min. The time evolution of B diffusion at 900 °C is observed as a function of implantation dose. A critical concentration for transient diffusion, C enh =8×1018 cm-2, independent of dose, is observed at this temperature. The transient diffusion enhancement in the diffusing part of the B profile increases with dose, up to a dose of ∼5×1014 cm-3, and saturates at higher doses. A comparison with published data shows that C enh ∼n i within a factor 2 over the temperature range 550–900 °C. We interpret our observations in terms of a nonequilibrium point‐defect model of diffusion and intermediate defect formation.
In this work the authors study the interaction of F with point defects and the influence of F on B diffusion in crystalline Si. The authors perform 25 and 100 keV F + implants and combine them with a 40 keV Si + implant. The appearance of peaks in the F profile during annealing supports the idea of the formation of F complexes with vacancies and Si interstitials. In all samples implanted with F + analyzed in this work, B diffusion is higher than in equilibrium conditions indicating that F + implants in crystalline Si produce a Si interstitial supersaturation. However, B diffusion is reduced when F + is coimplanted with Si, compared to only Si implants. This effect is more evident when B is located in the region where the F + implant generates an excess of vacancies, but it also appears in the Si interstitial-rich region. The results indicate that the effect of F on B diffusion in crystalline Si is time dependent.
BF+ 2 ions were implanted in (100) silicon at room temperature with an energy of 40 keV through a 140‐Å‐thick SiO 2 layer. Boron profiling by secondary‐ion mass spectrometry indicates that subsequent annealing in a conventional furnace in the 650–850 °C range for 30–240 min results in a pronounced secondary peak in the B and F profiles, in addition to the near‐surface primary peak located in the vicinity of the projected range of the implanted species. This phenomenon was also observed in BF+ 2 ‐implanted samples which were rapid thermal annealed at 900 °C for 15–60 s. The depths of the secondary peaks in the B and F profiles correspond to the depths of a damaged layer observed by cross‐sectional transmission electron microscopy. Isochronal furnace annealing revealed that there is no chemical interaction between B and F atoms during annealing. This is also supported by the observation of F atoms not affecting the B segregation coefficient during oxidation of the BF+ 2 ‐implanted samples. The end‐of‐range extended dislocations appear to be responsible for the gettering of B and F atoms during annealing.
An analytical model for high-concentration diffusion profiles was previously developed, where the diffusion coefficients depend on the concentration of impurities. By conducting experiments, we have obtained data on the diffusion of low-energy ion-implanted B, As, and BF2 over a wide temperature range. It is possible to describe the diffusion profiles using a constant surface concentration source with an offset period related to the transient enhanced diffusion. The surface concentration, Ns, was dependent on the temperature for B but was less dependent for that of As. Ns was lower in samples implanted with BF2+ than with B+. Therefore, the sheet resistance of BF2+ ion-implanted layers will always be lower than that of B+ implanted layers for a given junction depth. The offset time is independent of the impurities, and depends on temperature.
We have investigated and modelled the diffusion of boron implanted into crystalline silicon in the form of boron difluoride BF2+. We have used published data for BF2+ implanted with an energy of 2.2 keV in crystalline silicon. Fluorine effects are considered by using vacancy-fluorine pairs which are responsible for the suppression of boron diffusion in crystalline silicon. Following Uematsu's works, the simulations satisfactory reproduce the SIMS experimental profiles in the 800–1000 °C temperature range. The boron diffusion model in silicon of Uematsu has been improved taking into account the last experimental data.