Roland A. - PVD for microelectronics (779636), страница 69
Текст из файла (страница 69)
This is consistent with experimental observations (see, for example, Fig. 6.20) of up-tilted columns of significantlylower density than on the planar areas. From a functional point of view,368R. POWELL AND S. M. ROSSNAGELFIG. 10.13 Microstructure ( a ) and tilm density (b) for deposition of a thin film over a step with adepositing llux which is So from nornlal incidence IlO. 161.this sort of low-density coverage is not desirable for diffusion barrier orseed layer applications. The model can also be extended to the asymmetryintrinsic to long-throw deposition at the wafer edge. Figures 10.14a and10.14b show the observed film structures at the wafer center (Fig. 10.14a)or perpendicular to the wafer edge at the outer edge of the wafer (Fig.10.14b).
Experimentally, similar results were seen in Fig. 6.8.A related characteristic of the molecular dynamics approach is the development of a grain structure to the deposited film. This would typically occur in an elevated-temperature deposition process where therePROCESS MODELING FOR MAGNETRON DEPOSITION369FIG. 10.14 Simulation using SIMBAD of deposition (a) near the wafer center and (b) near the waferedge for long-throw deposition [ 10.16].was significant surface diffusion and mobility of the deposited species(see Section 7.3). The modeling allows predictions to be made of thenumber of grains formed as well as related features such as the dependence of the reflow time on the number of grains present.
In effect, withmore grains present, the surface diffusion is slowed down somewhat because of the inhibiting presence of extra grain boundaries, which aresinks for depositing atoms [10.24].10.3.3MONTE CARLO MODELSMonte Carlo modeling techniques can also be used to help describe thephysical properties of the deposited film, including the density and thestress. As a result, for example, it is possible to correlate the film stresswith the depositing kinetic energy of the film atoms [ 10.24] (Fig. 10.15).This shows trends similar to those observed experimentally: At low sputtering pressures and presumably high kinetic energies for the nonthermalized depositing atoms, the films tend to be in compressive stress. This is370R.
POWELL AND S. M. ROSSNAGEL1.000.04* Stress9Density*II --...t~E..Qi iiLLOo00.95,...0.020.00_E0.90f-"0.85r-0.020123~C.._z-456n0.80Incident Energy of Ni ( e V )(a)0.04?~0<13I", \0.03.....9S t r e s s.\9" Impurity-0.02Z0.02EOI0.,0.00~.*""-'-(./3i0.01-"O9--0.02,0l1,I2E--c -i,._<.,!3~t4,I5,0.006Incident Energy of Ni ( e V )(b)FIG. i0.15(a) The combined effect of ion bombardment with 5 0 - e V Ar and gas impurities on thestress and density of a condensing Ni film as a function of the incident energy of the Ni. (b) T h e stressand impurity levels tor the same case. The ion-to-atom ratio was 0.76 and the gas-to-atom tlux ratiowas 0.134 ! 10.241.partly due to the defects introduced by the more energetic deposition(higher kinetic energy, more energetic bombardment by reflected neutrals)and may also be related to incorporated inert gas impurities in the film.Another option available in this type of modeling is the examination ofthe chemical nature of the film.
One example of this is the case of alloysputtering described in Chapter 6, in which the transport of one of thespecies was different from the other. In general, the heavier species wasPROCESS MODELING FOR MAGNETRON DEPOSITION371scattered less during transport and was more directional than the lighter,more easily scattered species. As a result, the composition in a deep feature can be shown to be dependent on the depth into the feature, becomingenriched in the heavier, more directional species at the bottom [10.26].It is also possible to track the reaction level during reactive sputteringwith a Monte Carlo model. Figure 10.16 shows the relative nitrogen levelin a deep trench for the reactive deposition of TiN from a nonnitrided Ti target.
In this figure, the darker regions are close to 1:1 stoichiometry (TIN),but the lighter areas at the bottom of the trench are clearly under-nitrided,which may have implications on the quality of the diffusion barrier [10.16].10.4 ConclusionProcess modeling can be a versatile way to perform a virtual experiment,modeling a sputtering, transport, or deposition process without everturning on a sputtering system.
Commercial versions of several of theseFIG. 10.16 The SIMBAD-predicted chemical composition for the case of reactive deposition of TiNinto a deep trench. The target in this case was nonnitrided. The legend at the right side of the figureindicated the relative nitrogen concentration.372R. POWELL AND S. M. ROSSNAGELmodels also exist, either for sale to the end user or as a service, so thatthe user can gain access to the prediction capabilities without generatinga new model. The brief description and examples in this chapter are atiny fraction of the available, published studies in this technology. Forexample, a very similar modeling approach can be used to describechemical transport and reactions at a wafer surface, modeling a CVD orPECVD deposition process. Also, close examination of some of the subtle features in the molecular dynamics approach can lead to insight intophysical characteristics of the deposit such as stress or defect density[10.24-10.26].At the time of this chapter (1998), progress is also underway with aneven more physical approach to film modeling, one that tracks the arrivaland behavior of individual atoms rather than ensembles of atoms.
The lowenergy physics of these atoms is relatively well understood, but it has simply not been practical using available computers to develop single-atombased, film models. However, with the much wider availability of parallelcomputing coupled with the increasingly smaller dimensions of relevantfilm structures, a full-atomic three-dimensional model should be developed soon.
As described earlier, as trench widths move down toward 1000angstroms (0.1 micron) - - characteristic of the 4 or 16 gigabit DRAM gene r a t i o n s - the trench is now only 400 atoms or so wide. Therefore, it ismuch more conceivable that a computer will be able to model the development of films in three-dimensional features of this size range with available computing resources.References10.1.
S. M. Rossnagel, Modeling of Film Deposition fi~r Microelectronic Applications, AcademicPress, San Diego, 1996.10.2. J. P. Biersack and L. G. Haggmark, "A Monte Carlo computer program for the transport of energetic ions in amorphous targets," Nucl. Instrum. & Meth. 174:157 (1980).10.3. J. P. Biersack and W. Eckstein, "Sputtering studies with the Monte Carlo program TRIM.SP,"Appl. Phys.
A 34:73 (1984).10.4. W. Eckstein, "Energy distributions of sputtered particles," Nucl. lnstrum. & Meth. in Phys.Res. B18:344 (1987).10.5. D. Ruzic, "Fundamentals of Sputtering and Reflection," in Handbook of Plasma ProcessingTechnology, pp. 70-87 S.
M. Rossnagel, J. J. Cuomo, and W. D. Westwood, Eds., NoyesPublications, Park Ridge, N J, 1990.10.6. D. Ruzic, "The effects of surface roughness characterized by a fractal geometry on sputtering," Nucl. Instrum. & Meth. in Phys. Res. B47:118 (1990).10.7. W. D. Westwood, "Calculation of deposition rates in diode sputtering systems," J.
Vac. Sci. &Tech. 1 5 : I - 9 (1978).PROCESS MODELING FOR MAGNETRON DEPOSITION37310.8. I. Abril, A. Gras-Marti, and J. A. Valles-Abarca, "Energy transfer processes in glow discharges," J. Vac. Sci. & Tech. A4:1773-1778 (1986).10.9. J. A. Valles-Abarca and A.
Gras-Marti, "Evolution towards thermalization and diffusion ofsputtered particle fluxes: Spatial profiles," J. Appl. Phys. 55:1370-1378 (1984), and J. A.Valles-Abarca and A. Gras-Marti, "Slowing down and thermalization of sputtered particlefluxes: Energy distributions," J. Appl. Phys. 54:1071-1075 (1983).10.10. G. M. Turner, I. S. Falconer, B.
W. James, and D. R. McKenzie, "Monte Carlo calculation ofthe thermalization of atoms sputtered from the cathode of a sputtering discharge," J. Appl.Phys. 65:720 (1986).10.11. L.T. Ball, I. S. Falconer, D. R. McKenzie, and J. M. Smelt, "An interferometric investigationof the thermalization of copper atoms in a magnetron sputtering discharge," J. Appl. Phys. 59:720 (1986).10.12.
R. E. Somekh, "The thermalization of energetic atoms," J. Vac. Sci. & Tech. A2:1285-1291(1984).10.13. T. Motohiro and Y. Taga, "Monte Carlo simulation of the particle transport process in sputterdeposition," Thin Solid Films 112:161-173 (1984).10.14. D. Liu, S. K. Dew, M. J. Brett, T. Janacek, T. Smy, and W. Tsai, "Experimental study and computer simulation of collimated sputtering of Ti thin films over topographical features," J. Appl.Phys.
74:1339-1344 (1993).10.15. C. Sorlie, M. J. Brett, S. K. Dew, and T. Smy, "Advanced process simulation of metal film deposition," Solid State Tech., 101 (June 1995).10.16. M. J. Brett, S. K. Dew, and T. J. Smy, "Thin Film Microstructure and Process SimulationUsing SIMBAD," in Modeling ~?["Fihn Deposition for Microelectronic Applications, S. M.Rossnagei, Ed., Academic Press, San Diego, 1996.10.17. S. M.
Rossnagel and R. S. Robinson, "Monte Carlo model of topography development duringsputtering," J. Vac. Sci. & Tech. 21: 790(1982).10.18. S. Hamaguchi, M. Dalvie, R. T. Farouki, and S. Sethuraman, "A shock-tracking algorithm forsurface evolution under reactive ion etching," J. Appl. Phys. 74:5172 (1993).10.19. S. Hamaguchi and S. M. Rossnagel, "Simulations of trench-filling profiles under ionized magnetron sputter metal deposition," J.
Vac. Sci. & Tech. BI3:183 (1995).10.20. S. Hamaguchi and S. M. Rossnagel, "'Liner conlk)rmality in ionized metal sputter depositionprocesses," J. Vac. Sci. & Tech. B14 (1996).10.21. S. Hamaguchi, unpublished work, IBM Research, 1995.10.22.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
