Using ATHENA Monte Carlo Module for Ion Implantation Simulation in Silicon Carbides

Introduction

The Monte Carlo Implantation Module of ATHENA has proved to be a very accurate tool for simulation of various implantation processes. In this paper we demonstrate that the module can be successfully used not only for classical silicon-based technologies but also for other materials used in semiconductor industry. Silicon carbides were selected for this demonstration not only because they are widely used in power and high frequency electronics but also because they are most challenging objects for simulation due to their complicated lattice structures and electronic stopping models.

The large number of silicon carbide (SiC) polymorphic types (more than 150) presents a variety of physical properties, some of them critically important for fabrication of power devices. For example, the 4H polytype of SiC with a band gap of 3.27 eV and electron mobility almost twice that of 6H-SiC is a material of immense interest for production of power and high frequency devices. Ion implantation as well as related effects including damage formation, amorphization, dopant activation and annealing play an important role in fabrication of planar devices in semiconductors. These components of device processing are even more critical for SiC-based technologies because extremely low impurity diffusivities are typical in SiC and therefore ion implantation becomes the only practical selective-area doping method.

The implementation of ion implantation processing is one of the key challenges in silicon carbide device fabrication. The crystal structure of the polymorphic 4H type has wide openings in certain crystallographic directions, e.g. [11-23] and [11-20], potentially giving rise to deep ion channeling effects. For example, implantation of aluminium in the [11-23] direction results in a channeled profile, which is three times deeper compared to implantation with normal incidence and five times deeper compared to an implant off major axis (or “random”) direction. Besides, to facilitate epitaxial growth, most commercial (0001) wafers are offered with 3.5° to 8.5° miscut, therefore ion implantation performed in the normal direction to such wafers corresponds to a “random” direction. Given the directional complexity of 4H-SiC and 6H-SiC structures, it is extremely difficult to minimize or accurately predict the channeling effects because of the proximity of open channels ([0001], [11-23] and their equivalents).

Modeling of ranges and stopping is a powerful research technique for predicting dopant profiles in crystalline materials. As channeling plays an important role in the propagation of ions in crystalline SiC, modelling is an indispensable tool for optimizing initial implant conditions and avoiding the long tails in the implanted profiles. The ion implantation simulator could also be used in optimizing the implant conditions to obtain profiles of a desired shape. For example, it is typical for SiC technology that deep box-like doping profiles are formed using multiple implant process steps with different energies and doses.

In this work we describe a modelling technique based on the Binary Collision approximation along with the physical model used in order to achieve highly predictive simulation of ion implantation in crystalline SiC targets.

 

Simulation Model

The Monte Carlo ion implant module uses the Binary Collision (BC) approximation a detailed description of which has been presented elsewhere, [1-3]. The principal assumption in this approximation is that the interaction of energetic particles may be separated into a series of two-body collisions. Although not exact, this close approximation replaces the real trajectory of the moving particle by straight paths between the deflection points. The important benefit of this approach is the moderate speed of calculation combined with the possibility to accurately predict effects characteristic to single crystal structures. The slowing down of energetic particles is a result of nuclear and electronic stopping. The description of target atoms uses the translational symmetry of the crystal. A list is made of the positions of target atoms in the crystal, using one of the lattice sites as the origin of co-ordinates. The crystal search procedure finds all target atoms ahead of the ion’s motion with impact parameters less than pmax, a threshold value above which scattering is negligible. Crystal temperature is modelled by means of normally distributed displacements of the lattice atoms from their equilibrium positions. The amplitude of these displacements is evaluated with the Debye model.

Of critical importance for modelling ion implantation in crystalline targets is the electronic stopping model. Moving particles interact in a complex manner with the electrons in the target. The net effect of these interactions is velocity and position dependent retarding force acting on the moving ion. The electronic stopping in solids could be separated into two essentially different components - an electron excitation part and a continuous drag force acting on the moving particle from the surrounding electron gas. In our simulator, the value of the quasielastic electron excitation part is estimated by Firsov’s model for electronic energy losses with a correction for high energies when energy transfer diminishes as described in, [4]. This is essentially a local inelastic energy loss because it is localized at points of closest approach in the ion-atom collisions. Thus it affects the scattering angles and the remaining energy after each collision. The other part of the electronic stopping, the so-called non-local inelastic energy loss, depends on the electronic density in the crystal. This electronic stopping in SiC is difficult to model due to a complex electron density distribution, see for example [5], and a complex crystal structure with many open channels and planes. Our approach is to use the stopping model of Wang, Ma and Cui, [6], with corrections for the highly anisotropic valence electron density distribution.

 

Discussion

The validation of the implemented model has been performed with published experimental as-implanted profiles. The compiled data were selected from articles with well explained experimental conditions. In Figure 1 are shown simulated test results for 60 keV Al implants into 4H-SiC compared to experimental profiles taken from, [7]. Wafer orientation is (0001) and the major flat has been chosen to be (1100). Implants are performed with normal incidence to the surface, Figure 1(a), or oriented along the [11-23] channel, Figure 1(b). Both profile sets show deep tails due to channeling, the profiles of Al implanted along [11-23] channel being much deeper. On the offsets in Figure 1(a)-(b) are shown the crystal structures as seen in the direction of ion implantation. The [11-23] channel is 17° off the normal in the (1-100) plane. Two factors contribute to the deep tails of the profiles in this direction, a) the well formed open channel in 4H-SiC and, b) the Z1 dependence of the electronic stopping, Se (see [5,8] and references therein). Under channeling conditions, Se(Z1) has well defined oscillatory dependence. It turns out that Z1=13 (aluminum) lies very close to a minimum on the Se(Z1) curve. Therefore, aluminum ions travel much deeper along the channel than phosphorus (Z1=15), [9]. Any optimization of the initial implant conditions would require correct prediction of channeled profiles. Once a satisfactory agreement of the aluminium ion distributions implanted under channeling conditions, Figure 1(a),(b), has been achieved, the final validation of the model is implantation in a ‘random’ direction, i.e. away from any open channel in the 4H-SiC lattice. Figure 1(c) shows such one set of profiles obtained by 9° tilt off the normal, [0001]. The model correctly predicts the peaks due to the “random” portion of ion’s stopping and the tails due to the “channeling” fraction by aluminum ions occasionally entering open channels in the 4H-SiC lattice.

Figure 1. Experimental (SIMS) and calculated (BCA simulation) profiles of 60 keV Al implantation into 4H-SiC at different doses (shown next to the profiles) for a) on-axis direction, b) direction tilted 17° of the normal in the (1-100) plane, i.e. channel [11-23], and c) a “random” direction - 9° tilt in the (1-100) plane. Experimental data are taken from [7].

 

 

In Figure 2 is shown more practical application of the model for multiple Al implants into 6H-SiC. Once again, the level of agreement with the experiment can be achieved only if the range and the shape of each individual profile agrees well with the corresponding experimental one.

Figure 2. Box profile obtained by multiple Al implantation into 6H-SiC at energies 180, 100 and 50 keV and doses 2.7 x 1015, 1.4 x 1015 and 9 x 1014 cm-2 respectively. The accumulated dose is 5 x 1015cm-2. Experimental profile is taken from [10].

 

 

Conclusions

Using an appropriate electronic stopping model for SiC, one can obtain highly predictive simulation results of ion implantation within the binary collision approximation formalism. Accounting for the anisotropy of the electronic density distribution in the 4H-SiC and 6H-SiC lattices is critical for simulation of predictive implant distributions not only along open channel directions, but along “random” direction as well. The described model is implemented in the Monte Carlo Implant Module and successfully used by several ATHENA customers. These results were originally presented at 14th International Conference on Ion Beam Modification of Material (IBMM-2004) and will be published in the Nuclear Instruments and Methods in Physics Research B.

 

References

  1. M. T. Robinson, Radiation Effects and Defects in Solids 130-131, 3 (1994).
  2. M. T. Robinson, and I. M. Torrens, Physical Review B 9, 5008 (1974).
  3. I. R. Chakarov, and R. P. Webb, Radiation Effects and Defects in Solids 130-131, 447 (1994).
  4. M. J. van Dort, P. H. Woerlee, and A. J. Walker, Solid-State Electronics 37, 411 (1994).
  5. C. H. Park, B.-H. Cheong, K.-H. Lee, and K. J. Chang, Physical Review B 49, 4485 (1994).
  6. Y.-N. Wang, and T.-C. Ma, Physical Review A 44, 1768 (1991).
  7. J. Wong-Leung, M. S. Janson, and B. G. Svensson, Journal of Applied Physics 93, 8914 (2003).
  8. R. Smith, M. Jakas, D. Ashworth, B. Owen, M. Bowyer, I. Chakarov, and R. Webb, Atomic and Ion Collisions in Solids and at Surfaces, R. Smith Ed., Cambridge University Press, 1997.
  9. R. G. Wilson, D. M. Jamba, P. K. Chu, C. G. Hopkins, and C. J. Hitzman, Journal of Applied Physics 60, 2806 (1986).
  10. T. Kimoto, A. Itoh, H. Matsunami, T. Nakata, and M. Watanabe, Journal of Electronic Meterials 25, 879 (1996).

 

Download pdf version of this article