SIMULATION OF THE DEPOSITION OF BCC METALS INCLUDING ANISTROPIC EFFECTS USING 3D-Films

Transcription

1 SIMULATION OF THE DEPOSITION OF BCC METALS INCLUDING ANISTROPIC EFFECTS USING 3D-Films T. Smy and R.V. Joshi 1 Dept. of Electronics, Carleton University, Ottawa, ON, Canada K1S 5B6, ph: , fax: : tjs@doe.carleton.ca 1 IBM, T.J. Watson Research Center, Yorktown Heights, NY 10598, USA 1

2 Abstract This paper presents both experimental results and a simulation study of anisotropic growth for deposited thin films of BCC metals. Experimental results present W films deposited using CVD over flat substrates, high aspect ratio trenches and overhang structures 3D-Films a three dimensional Monte Carlo program microstructural growth simulator was used to perform the simulations. The capability of modeling faceted grain growth due anisotropic adatom mobilities and surface energies was added to 3D-Films. Results show faceted grain growth does produce a rougher surface, however, the film is dense and mostly columnar in nature. The surface roughness is 2-3 times larger for the faceted films

3 Introduction PVD and CVD deposited metal films can be strongly influenced by anisotropic factors including adatom mobility and surface energies. Both microstructure and surface roughness of the film is determined largely by a combination of transport characteristics and anisotropic surface energies Three dimensional (3D) microstructural modeling of the deposition of thin film metal layers for ULSI applications has previously ignored anisotropic growth. The paper will present as an example CVD of W onto flat substrates, 0.5 µm wide trenches and overhang structures. 3D-Films simulations will be compared to SEMs of the films and surface roughness information will be extracted from the simulations

4 3D-Films: Basic model 3D-Films is Monte Carlo simulation tool that represents the film as aggregation of a large blocks, with each blocks representing the average behavior aproximately 1000 atoms. Particles are serially launched from just above the growing film and follow straight line trajectories until they strike the substrate or the film. The particles aggregate into the film after a short surface diffusion to minimize a surface curvature-dependent chemical potential. The combination of ballistic shadowing and short range surface diffusion successfully accounts for the formation of the columnar thin film microstructure characteristic of refractory metals at typical deposition temperatures.

5 3D-Films: Simulation Framework Particle Transport Substrate

6 3D-Films: Effecient CVD Simulation For Simulation of CVD films 3D-Films uses a modification of the standard Monte Carlo method which substantially decreases the execution time for low sticking coefficient simulations The method separates the transport processes and deposition process. This allows for a normalization of the incident flux magnitude before deposition. Material is then deposited stoichasitcally over the entire surface. This produces a substantial improvement in execution (many orders of magnitude speed-up).

7 3D-Films: Anisotropic Growth Modeling To model anisotropic effects in the growth of the film two enhancements to the code were made. As the film nucleates a crystal orientation is assigned to the forming grain. Anisotropic adatom mobilities and anisotropic bonding energies were assigned to each grain.

8 3D-Films: Anisotropic Growth Modeling To simplify the implementation in 3D-Films the number of grain orientations used limited. For the work in this paper two film textures were investigated. In-plane Faceting The anisotropic grain growth was limited to being in the plane of the substrate. 8 different grain orientations were used for the 3D-Films simulations. Random Faceting A random grain orientation was used allowing for 24 orienations in the 3D-Films simulations For comparison Isotropic or unfaceted films were also simulated.

9 Results: Flat films (S t = 0.001) SEM In-plane Random

10 Results: Flat films (S t = 0.001) 370C 430C 3D-Films depiction showing grain microstructure for random faceting.

11 Results: Deposition on to flat substrates For the deposition of a W film deposited in a LPCVD reactor at 370 K and 30 mtorr using SiH 4 and W F 6 precursors anisotropic growth is present with the top edges of elongated grains clearly seen on the surface. The 3D-Films simulation deposited with a low mobility and in-plane faceting is obviously very anisotropic. The randomly orientated film is more chaotic in structure and the anisotropic nature of the grains is more difficult to see. However, the rough surface of this film is like that of the SEM. The real film appears to show a structure with features of each simulation indicating a texture that is partially random. Increasing the deposition temperature to 430 K produces a longer surface diffusion length and a more pronounced anisotropic structure.

12 Results: Flat films (S t = 0.05 and S t = 1.0) S t = 1.0 S t = 0.05 Random Isotropic

13 Results: Flat films (S t = 0.05 and S t = 1.0) For comparison simulations of isotropic and random faceted films were deposited over a wide range of S t values. For S t = 1.0 appropriate for PVD films a very rough surface of the unity sticking is produced by a faceted film. The unfaceted film, however, displays little variation in surface roughness with variation in S t.

14 Results: Deposition over topography W CVD films deposited over a 0.5 micron trench and an overhang struture.

15 Results: Deposition over topography S t = 1.0 S t = 0.05 S t = 0.001

16 Results: Deposition over topography S t = 1.0 S t = 0.05 S t = 0.001

17 Results: Deposition over topography S t = 0.05 S t = 0.001

18 Results: Deposition over topography An analysis with respect to the trench and overhang structures was also undertaken. Films with a variety of S t values and faceting modes were compared to the SEM cross-sections. Step coverage and microstructure were compared leading to an evalution of the S t value appropriate for the CVD deposition

19 Results: Deposition over topography Both types faceting and isotropic growth produce tight grain structures. Random faceted growth appears less ordered and uniform, but is still basically columnar. Using these results and those for the overhang structure a sticking coefficient of was found to appropriate.

20 Results: Simulated AFM for flat films St = 1.0 St = 0.05 Random Faceting Isotropic St = 0.001

21 Results: Simulated AFM for flat films Both the simulations and experimental results indicate that the faceted growth appears to produce a rougher film surface. To investigate this simulated AFM images were generated. The effect of the faceted growth can clearly be seen in these figures and increase in surface roughness for faceted films at low S t observed. The random orientated film with S t = 1.0 had the largest variation in surface height and all other film images were normalized to this film.

22 Results: Simulated AFM for flat films It can be clearly seen that the isotropic films exhibit little variation in surface roughness and have the expected grain structure. The in-plane film also has a smaller amount of surface roughness but does, of course, clearly show the anisotropic grain growth. From this AFM data simulated RMS surface roughness values can be calculated. The plot clearly illustrates the larger surface roughness of the random faceted film and the increase in roughnes as S t approaches unity.

23 Results: Simulated AFM for flat films In-plane Faceting (S t = 0.05)

24 Results: RMS roughness 0.08 RMS Roughness (microns) Faceted films Isotropic Films S t