New Model for High Energy Implants in ATHENA

Introduction

In our earlier paper [1] a new Binary Collision Approximation (BCA) module was described. It was shown that models and algorithms implemented into the BCA model allow to accurately predict implant profiles with significant channeling as well as the dose and screen oxide thickness dependencies for relatively low implant energies.

Recently, this module was extended into 2D and completely integrated with ATHENA. Implementation of the BCA module will allow ATHENA users simulate 2D implant profiles with accuracy which could not be achieved even with the best experimentally verified analytical models (e.g. SVDP). The main advantage of the BCA approach is its ability to simulate profiles for those conditions for which reliable experimental results do not exist. Also, experimental 2D profiling techniques are still unique, very expensive, and yet to become widely available. Therefore, there are no experimentally verified tables of lateral moments which could be used for 2D analytical simulations.

This paper is focused on simulation of medium and high energy implants. Since upper energy limits for SVDP model are 100 keV for Boron and 180 keVfor Phosphorus and Arsenic, it is reasonable to consider medium energies starting from upper SVDP limits up to ~1 MeV and high energies above 1 MeV.These implant energies are widely used in different technologies. Double and triple well formation in CMOS technology using multiple medium and high energy implants is a very important application.

 

BCA Model Description

In the BCA, the deflection of the trajectories of moving particles is calculated in a strictly binary way - between the moving atom and the closest atom in the lattice. The ion implantation is simulated by following the fate of a large number of sequentially generated pseudo-projectiles, each of which carries an equivalent dose corresponding to a fluence increment obtained by dividing the total fluence by the number of pseudo-projectiles and scaled for the topography under interest.

The atomic collisions are considered to be composed of a quasi elastic part and of an essentially separate electron excitation part. The barycentric scattering angle is evaluated with the MAGIC formula [2]. The program uses the screened Coulomb potential with the universal screening function described in [2]. The screening length in the calculation is the universal one proposed in [2]. In order to properly account for the channeling, the BCA module uses non local and local inelastic energy loss models. The local impact dependent inelastic energy loss model is the exponential model suggested by Oen and Robinson [3]. The damage accumulation is based on the displacement model of Kinchin and Pease [4].

 

Simulation Results and Comparison with Experiments

Boron
Zero tilt angle, medium energy boron implant simulation is the most challenging problem. From one side, the experimental profiles are very sensitive to even slight changes in the experimental setup including precise surface orientation and ion beam direction and beam width. From the other side, simulation results are very sensitive to details of local electronic stopping models. Figure 1 shows the effect of beamwidth on low dose zero tilt, 200 keV boron implant into <100> direction. The BCA simulation profiles demonstrate two distinctive peaks: random scatter and channeling. The height ratio between the peaks strongly depends on ion beam width or divergence. In the ideal case of absolutely collimated beam the crystalline peak is ~3 times higher than the amorphous peak. In the case of beamwidth = 1 degree (incident ion directions are uniformly distributed between -1 and 1 degree) the scatter peak becomes higher. The experimental profiles do not always show these distinctive peaks but rather have a wide plateau as in Figure 2. The difference could be attributed both to possible uncertanties in experimental setup and to inaccuracy of local electronic stopping model. Currently implemented generic local electronic stopping model does not take into account details of the electron distribution in the <100> channel.

 

Figure 1. Effect of beamwidth on zero tilt 200 keV low dose boron implant.
1: beamwidth = 0;
2: beamwidth = 0.5 degrees;
3: beamwidth = 1.0 degree.

 

 

Figure 2 shows that the above analysis is valid for a higher energy of 380keV.

 

Figure 2. Comparison of BCA simulation () and measurement for
a 380keV Boron implant at zero degrees. The simulation is within
the resolution of the measurement technique.

 

 

Phosphorus
Figure 3 compares simulated and experimental [6] profiles for 500 keV and 1000 keV , 1.5x10 ions/cm Phosphorus implants at zero degree tilt along <100> axis. As expected the second peak is much lower than for comparable boron energies. In these cases agreement is even better than for boron implants because local electronic stopping for channeling phosphorus is less dependent on peculiarities of electron distributions.

Figure 3. Zero degree phosphorus implants simulated
using the BCA model () compared to measured data.

 

Arsenic
Figure 4 shows simulated and experimental [7] 300 keV As profiles. They do not have the second peak but rather characteristic long tails.

 

Figure 4. 700 keV Arsenic implant simulated by the BLA
model () and compared to measured curve.

 

Conclusion
It is shown that the BCA module recently integrated into ATHENA under the MC Implant module could be used for quite accurate simulation of medium and high energy (up to 1 MeV) implants in <100> direction of crystalline Silicon. In separate simulations the validity of implemented stopping power models is confirmed for off axis boron implants with energies up to 3 MeV. Simulated results are close to experimental data when limits of the experimental accuracy are considered. Our BCA implementation is 5 to 10 times faster than UT-MARLOWE [8] which allows practical use for high energy implant simulations. The BCA module can be used for extraction of spatial moments for our computationally efficient 2D model described in [9].

References

  1. "Accurate Simulation of Channeling Implants
    Using Monte Carlo Techniques."
    TCAD Driven CAD Vol.8, No.8, 1997.
  2. J.F. Ziegler, J.P. Biersack, and U.Littmark
    The Stopping and Range of Ions in Solids
    Pergamon Press, Oxford, Vol.1 (1985)
  3. O.S. Oen and M.T. Robinson, Nuclear Instruments and Methods
    The Stopping and Range of Ions in Solids
    Pergamon Press, Oxford, Vol.1 (1985) Vol. 132, 647 (1976)
  4. G.H. Kinchin and R.S. Pease, Rep. Prog. Phys.
    Vol.18 (1955) p.1.
  5. R.J. Schreutelkamp et.al. Nuclear Instruments and Methods
    Vol. B55, 615 (1991).
  6. V. Raineri, Nuclear Instruments and Methods in Physics research
    Vol. B62, p. 331 (1992)
  7. R.G.Wilson, J. Appl. Phys.,
    Vol.52, p. 3985 (1981)
  8. S. Tian et. al. UT-MARLOWE, Version 4.1
    University of Texas at Austin, 1998.
  9. M.Temkin and I. Chakarov
    Semiconductor Process and Device Performance Modeling
    (MRS Symposium Proceedings) Vol.490, p.27 (1998).