Advanced Analytical Ion Implantation Models in ATHENA

The down scaling of VLSI technologies down to very deep sub-micron feature sizes leads to higher requirements for quality prediction of implants. The most important trend is the continued reduction of thermal budgets. Final dopant profiles are determined by implant profiles to a much higher degree of accuracy than ever before. Also, modern technologies exploit a much wider variety of implant conditions: low energies (<20 keV) for very shallow junction formations, very high energies for low thermal budget well formations, high angled implant for LDD engineering, zero-angled implants or continuous rotation of the wafer to avoid non-uniformity of device characteristics over the wafer.

Accurate and computationally effective ion implantation simulation methods should be implemented inside general purpose process simulators in order to facilitate rapid technology development and better control of manufacturing processes. These simulators already include advanced diffusion models based on implant damage enhancement of diffusion rates. Without accurate prediction of implant profiles as well as the implant-induced defect distributions, all efforts to improve diffusion models would be wasteful. Until recently, the available ion implantation models had been either computationally inefficient such as Monte Carlo with channeling or limited to narrow ranges of energies, doses, and tilt angles as in most of analytical methods and look-up tables. A secondary issue is inadequate modeling of the lateral implant distributions.

During 8 years development at SILVACO, the process simulator ATHENA/SSUPREM4 has evolved to include many new models and features for ion implantation simulation. These include increased energy intervals in analytical tables, implementation of double Pearson method for limited energy range, ability to take into account implant tilt and rotation angles and predict shadowing effects for arbitrary geometries, a fast Monte Carlo method for amorphous targets, reasonable Monte Carlo model for crystalline targets, and the ability to follow development of ion-implant cascades.

Since modern technology demands even better quality of ion implant simulation, the ion implantation models in ATHENA are currently going through serious revision. In this article we will present some results of new analytical models. We will describe new 2D analytical profile method as well as developments in ATHENA's Monte Carlo implant module in subsequent issues of the Simulation Standard.

 

Additional Models from the University of Texas, Austin

The most difficult challenge for ion implantation theory is to accurately account for channeling dependence on energy, implant angles and dose, and overlaying non-crystalline material thickness (e.g. screen oxide). Only finely tuned advanced Monte Carlo programs could address this problem from the first principles.

The Dual-Pearson approach [1] is the only computationally efficient, semi-empirical model capable of simulating 1D-profiles with channeling effects taken into account. Al Tasch and his co-workers at U. of Texas, Austin, Eaton Corp, and Charles Evans and Associates showed [1-3] that the linear combination of two Pearson functions could describe very wide variety of profiles even with pronounced channeling effects. They also performed a very large amount of implants and SIMS measurements from which nine Dual-Pearson parameters were extracted and gathered into a look-up tables [4-6].

These look-up tables are implemented into ATHENA. ATHENA follows a four-step interpolation algorithm devised by Al Tasch et al. Table 1 shows parameter ranges currently available in ATHENA.


Table 1. Range of validity for University of Texas models in ATHENA.

Notes:
(1)Monte Carlo estimations for low energies (0-15keV) will be added
(2)Only for 15 - 80 keV
(3)Numerical extrapolation to 200 keV and Monte Carlo estimations for low energies (0-15keV) will be added


The excellent agreement between SIMS profiles and the Dual-Pearson model has been demonstrated in [6]. Here we give several examples of predictive capabilities of the model implemented into ATHENA.

Figure 1 shows tilt angle dependence of boron ion implants into (100)silicon. As can be seen the boron distribution is very sensitive even for a small variation in the tilt angle. This is critically important for all boron implants because for energies as low as 2 keV the channeling is still well pronounced.

 


Figure 1. Tilt angle dependence of boron implant profiles in {100} Si; energy = 35keV, dose =1x1013 cm-2 native surface oxide.

Figure 2 shows the depth distribution of boron implanted into (100)silicon as a function of the surface screen oxide. The effect of the presence of such an oxide is similar to the one introduced by variation of the tilt angle. The surface screen oxide partially randomizes the ion flux, which leads to less channeling with increasing oxide thickness.


Figure 2. Boron implantation through surface screen oxide. Tilt angle is 0o, energy and dose are as in Figure 1.

Figure 3 shows the energy dependence of implanted boron profiles. For comparison, the experimental data [7] is also given.


Figure 3. Comparison of boron implants at different energies: tilt=0, dose=1013cm-2 . Comparison to experiments from [7] are shown.

Figure 4 shows profiles of implanted phosphorus at 100keV for doses of 1e13, 5e13, 2e14, 5e14 and 1e15. The experimental profiles are also from [7]. As the implant dose increases more damage is created which results in additional dechanneling of phosphorus ions. Therefore the profile tail shortens with increasing dose.


Figure 4. Dose dependence of phosphorus implants along <100> channel in Si: asterisks - simulations with Tasch model in ATHENA, squares - experimental profiles from [7].

Conclusion

The last two figures clearly show that the implemented model agrees very well with not only the experimental profiles measured by U. of Texas, Austin, but also with independent experiments.

Implementation of ion implantation models developed by U. of Texas, Austin, allows ATHENA users to very accurately and quickly simulate critical implantation steps in modern technologies.

References

  1. A.F. Tasch, H. Shin, C.Park, J. Alvis, and S.Novak, "An Improved Approach to Accurately Model B and BF2 Implants in Silicon", J. Electrochem. Soc., 136, p.810, 1989.
  2. K.M. Klein, C. Park, A.F. Tasch, R.B. Simonton, and S. Novak, "Analysis of Implanted Boron Distribution Dependence on Tilt and Rotation Angle", J. Electrochem. Soc., 138, p.2102, 1991.
  3. C.Park, K.M. Klein, A.F. Tasch, R.B. Simonton, S. Novak, and G. Lux, "A Comprehansive and Computationally Efficient Modeling Strategy for Simulation of Boron Ion Implantation into Single-Crystal Silicon with Explicit Dose and Implant Angle Dependence", COMPEL, 10, p.331 (1991)
  4. C.Park, K.M. Klein, and A.F. Tasch, "Efficient Modeling Parameter Extraction for Dual Pearson Approach to Simulation of Implanted Impurity Profiles in Silicon", Solid State Electronics, vol. 33, p.650, 1990.
  5. S.J.Morris, Vu Do, P. Gupta, S.H. Yang, C. Park, K.M. Klein, A.F. Tasch, R.B. Simonton, and G.E. Lux "Accurate and Computationally Efficient Modeling of BF2 Implants into Single-Crystal Silicon", in Extended Abstracts of Spring. Mtg. of the Electrochem Soc., 92-1, p.450 (1992)
  6. R.Simonton and Al.Tasch, " Channeling Effect in Ion Imp7lantation", in Handbook of Ion Implantation Technology, ed. J.F.Ziegler, p.119, 1992.
  7. R.J. Schreutelkamp, V. Raineri, F.W. Saris, R.E. Kaim, J.F.M. Westendorp, P.F.H.M. van der Meulen, K.T.F. Janssen, Channeling implants if B and P in silicon, Nucl. Instr. Meth. B55, p.615, 1991