An Improved Approach to Accurately Model Shallow B and  BF 2 Implants in Silicon

Abstract

In this paper, an improved modeling approach is described for simulating as-implanted boron impurity profiles for B + and BF2 ~ implants into single*crystal silicon. This method uses the sum of two Pearson distribution functions to account for the nonchanneling and channeling components of the implant distribution. The ratio of the two Pearson functions varies with dose, which accounts for the change in the degree of channeling with dose. This modeling approach has been compared with experimentally measured SIMS profiles for a wide range of energies and doses for shallow B § and BF2 § implants. The excellent agreement indicates that this method offers a large improvement in simulation capability for B § and BF2 § implants. In addition, this method should be applicable to accurately model other impurities which have channeling tendencies.In the development of submicron-integrated circuit technologies, much attention is currently focused on the achievement of reproducible and uniform (over the wafer) shallow, compact impurity profiles for both shallow junctions and the adjustment of various threshold voltages. Also, the entire impurity profile is of interest, in particular for the shallow source-drain junctions in CMOS. This is because of the need to simultaneously have low sheet resist~ ance, very high surface concentration for low contact resistance, a shallow junction, and profile control in order to minimize high electric fields which can cause hot carrier reliability problems. This has been a particularly difficult task for the case of boron due to the strong channeling tendency of this light atom in silicon. Indeed, the degree of channeling has been shown to depend considerably on both the tilt and rotation (or twist) angles during ion implantation, even for those ranges of angles for which channeling is generally considered to be minimal (1-3).Extensive use of process modeling is mandatory for timely, efficient technology development and in order to understand the process control issues in manufacturing. Thus it is necessary to be able to accurately predict impurity profiles beginning with the as-implanted distribution and continuing with the evolution (diffusion) of the profile in subsequent thermal treatments. In addition, the models used to describe the profile evolution in furnace and rapid thermal processing can only be valid over a wide range of conditions if they begin with the correct initial as-implanted impurity distribution. Otherwise, erroneous assumptions must be made in the model in order to relate its predicted diffused profile with the initial implanted profile. This is especially the case when rapid thermal annealing is used.Historically it has been difficult to accurately simulate as-implanted boron distributions for both B+ and BF2 + implants. This is mainly due to the ease with which boron is able to channel in the silicon lattice. Also, at the lower energies of interest for shallow profiles, channeling occurs more easily, further aggravating the simulation difficulty. The original LSS th...

Enhanced phenomenological models of ion channeling contribution to doping profiles in crystals

“…The "channeled" fraction forms a long-range edge of a profile (z > 250 nm) due to the ions captured into the stable channeling at the crystal surface. For these fractions, the shapes of partial doping profiles are qualitatively similar to those of the asymmetric Pearson type IV distributions that are generally accepted in the physics of implantation and are used as basis functions w r,ch of the standard "dual Pearson" model [20].…”

“…Legendre [18]) polynomials, the heuristic empirical models [19][20][21] are widely used for practical calculations. To some degree, they reflect the physics of ion transport and implantation into crystals.…”

“…The models differ by a choice of w-functions. For example, within the scope of the most accepted "dual Pearson" model [20] both w r and w ch are Pearson-type distributions [22] having independent sets of parameters p r and p ch . The existing empirical models manifest a reasonable capability of fitting the doping profiles for a limited set of implantation setups (ion beam energy and direction, implantation dose).…”

One-Dimensional Models for Diffusion and Segregation of Boron and Ion Implantation of Aluminum in 4H-Silicon Carbide

“…However, according to a multiday BCA simulation using a large number of ion trajectories, values of the simulated aluminum concentration do not monotonically decrease when the aluminum concentration becomes comparable to an n-type drift-layerdoping level (in the order of 10 15 cm -3 ). A continuous-function approximation, just like the dual-Pearson approach established for ion implantation into silicon (Tasch et al, 1989), is thus needed. The historic development and basic concepts of boron diffusion in SiC are reviewed as follows.…”

Parameter Extraction Procedure for Ion Implantation Profiles to Establish Robust Database based on Tail Function

“…The accuracy of the simulation depends on the function used. Functions to express ion implantation profiles have also been developed by using Gauss, joined half Gauss [1], Pearson [2,3], dual Pearson [4,5], and tail functions [6][7][8][9]. It has been demonstrated that the dual Pearson accurately covers the whole ion implantation profile, and it is the standard function implemented in commercial simulators [10].…”