1D/2D Victory Process tutorial

vpex01.in : 1D/2D Victory Process tutorial


Requires: Victory Process : Core Simulator, 2D Diffusion & Implantation, 2D Oxidation
Minimum Version: Victory Process 6.8.0

This example demonstrates 2D process simulation for 28nm NMOS transistor with STI (Shallow Trench Isolation). The simulation deck consists of about 20 actual process steps - etching, deposition, ion implantation, diffusion/oxidation, and more then 50 auxiliary statements, including initialization, mesh generation, exporting structure for visualization,and parameter extraction.

The goal of this example is to provide a basic tutorial for Victory Process in 2D mode. It also demonstrates generic compatibility with Athena syntax as well as some advantages (including better internal meshing and convergence) over Athena.

Most of time consuming numerical algorithms of Victory Process are efficiently parallelized. By default Victory Process runs on just one processor. To ensure the best performance on your computer the following simulation condition simflags="-P all" is specified in the go line starting Victory Process. This means that all processors available will be used in parallel section of Victory Process. If you want to use a smaller number of processors you can substitute "all" with a desired number, e.g. simflags="-P 4".

Simulation starts with Init statement, which specifies

  • wafer properties: material, orientation, initial doping,
  • simulation domain: lateral (x-direction) interval [0,0.45], substrate depth, and height of gas region above substrate. This domain corresponds to right half of the NMOS device with STI. The complete structure suitable for device simulation will be obtain by "mirroring" command in the end of simulation.
    Note: always make sure that the total thickness of all layers deposited above substrate will not be larger than GASHEIGHT.
  • the size of mesh cells (0.01 micron) of the base grid which is used for geometry representation.In fact Victory Process uses Nested Cartesian Grid approach to represent geometry and material boundary movements. By default meshdepth = 2 ,which means that only one nested mesh level will be used during this process. As the result the minimum resolution of the layer boundaries representation will be 0.0025 microns. The resolution can be further increased by either decreasing of base grid cell size or specifying meshdepth > 2.

Victory Process uses non-uniform Cartesian grid to resolve volumetric data, e.g. doping or defect concentration, stress, etc. In 2D mode this grid is defined by a set of line x and line z statements. This grid is similar to initial grid generated by Athena. The differences are: the "depth" coordinate is "z" (not "y"), the lines apply to the whole simulation domain including gas region, and triangle elements are not generated in Victory Process.

This example uses automatic selection of simulation dimensionality. This means that simulation will start in 1D and will switch to 2D when required by the process flow. This is a default mode and, as in Athena, this capability saves simulation time without loosing any accuracy. If you want to force Victory Process to run in 2D from the very beginning you can specify flow.dim=2d .

The first process sequence is N-well formation consisted from boron implant and diffusion cycle with temperature ramp-up, constant, and ramp-down steps. Default analytical implant model and default diffusion model are used in this sequence. In the end of this process sequence the structure is still 1D. It is saved for visualization using export command. The well profile is visualized using TonyPlot.

The next process sequence is STI formation. A half of 0.3 microns shallow trench with 85 degrees side wall is formed by geometrical etch and then oxidized in dry ambient. The active area is masked with nitride layer. Note that Victory Process allows to specify the deposition interval, using parameters left.to, right.to, between . Full physical oxidation model should be used for this non-planar structure. The oxidation simulation is slightly slower then Athena for similar structures. In the same time Victory Process is usually more stable than Athena which often produces oxidation artifacts due to irregular grid around slopped trench walls. The trench then filled with oxide which is automatically planarized. This is achieved by using parameters max and thick=0.0 in the deposit statement which means that the trench will be filled up to horizontal plane at the highest point of the structure. This capability is more convenient and stable than deposition step followed by "etch above" step in Athena.

The next step is Vt implant which uses( as all subsequent implants ) Monte Carlo model. The Monte Carlo implant model is very efficiently parallelized. If you want to increase simulation speed and number of processors set in the go statement is not set to "all" you can set higher number of processors used by Monte Carlo implant using parameter parallel .
The number of simulated ion trajectories ( parameter n.ion ) is very important because it determines a trade-off between accuracy and simulation time. For 2D simulation this number should be preferably in interval between 200000 and 1000000. You always can start with lower number of trajectories during first stages of simulation flow design and increase it for final runs.

Following removal of sacrificial oxide, thin (~7nm) gate oxide layer is grown using full physical oxidation model. Extract capability of Deckbuild is used to measure gate oxide thickness.

After that, the poly-silicon gate is formed using geometrical deposition and then re-oxidized in a dry ambient.

The HALO implant with 25 degrees tilt is performed using Monte Carlo model. Note that Victory Process allows to specify several rotation angles in one implant statement. In this case rotation=0 and rotation=180 are used to assure implant symmetry. The low energy Arsenic implant is used to dope extension before spacer is formed using geometrical deposition and etching of nitride. The shape of the spacer is shown by TonyPlot.

The final process sequence of this example is S/D formation. The plus.one model is specified in the high dose arsenic implant statement. The model estimates interstitial profile generated after implant by scaling the impurity profile with factor dam.fac. The advanced single-pair diffusion model is used for final S/D anneal. This model should be used when there are substantial amount of point defects generated during preceding process step. In this specific case high dose S/D implant is the source of interstitial which tend to pair with impurities and therefore affect final distribution. The single_pair model takes into account only dominant or primary impurity-defect pairs, e.g. Boron-Interstitial pairs. The scaling factor dam.fac usually serve as calibration parameter.

In the end of simulation, two parameters of the simulated device - S/D junction depth and estimated (1D) threshold voltage - are extracted. Finally, the contact metal is added using geometrical etch and deposition and the full structure ready for device simulation is obtained using (bold} mirror capability in the {export}statement.

To load and run this example, select the Load example button in DeckBuild Examples window. This will copy the input file and any support files to your current working directory. Select the run button to execute the example.

Several intermediate structures will be created. The following plots demonstrate different steps of the process flow:

vpex01_well.png shows the 1D well profile.

vpex01_cmp.png shows the 2D structure after trench etch, oxidation and planarization.

vpex01_spacer.png shows the shape of spacer area obtained using geometrical deposition and etch commands

vpex01_final.png shows the final full structure with absolute net doping which is ready for device simulation. During the export the structure is also mirrored to obtain a full MOS device with source and drain side.