Assignment II

 

1. From EXAMPLES open example #mos1ex02. It simulates NMOS fabrication using Athena and Id(Vds) using Atlas.

a) In Atlas, save files after sweeping Vd for each gate voltage. Use Tonyplot to plot each of these structures. These structures are just like structures obtained in fabrication but they contain information on electrical parameters and they will be used to analyze transistor operation.

b) Modify the program by adding DIBL analyzes for Vd=0.1V, 1V and 3.3V as shown below for 0.1V:

log off
solve init
method newton
# Bias the drain
solve vdrain=0.1
save outf=mos1ex02_4.str
# Ramp the gate
log outf=mos1ex02_5.log master
solve vgate=0 vstep=0.25 vfinal=3.0 name=gate
save outf=mos1ex02_5.str

Then plot all I-V in one figure:

# plot results
tonyplot -overlay -st mos1ex02_5.log mos1ex02_7.log mos1ex02_9.log -set mos1ex02_9.set

 

The log files are your results from simulations that can be ploted in a specific format stored in a .set file ex: mos1ex02_9.set. You will have to make this set file by opening your first *.log file (here I had mos1ex02_7.log ) in Tonyplot, then adding (=overlaying) the next one and the last one (for Vd=3.3V); finaly save this file as a set file.

Analyze the DIBL effect in this transistor by ploting Id(Vg) for various Vd values.

c) Observe various electrical parameters and their distributions in your structures such as: electric field, potential, recombination rates, impact ionization, electron/hole concentrations etc. Selection can be made from Display/Definition/Contour in Tony Plots. See the role of Vg and Vd on electron conentration in the channel. To see electrical parameters w/o any bias you have to save the structure after solve init command (set all voltages to 0).

d) Now scale down your transistor to make 0.1 µm channel length. From Athena and also from Atlas it will be clear that the structure has to be changed: junctions' depths (mainly by changing implantation conditions), substrate concentration (i.e. in the well) and oxide thickness (please notice the oxide thickness diffrence in a big and small device related to the poly-Si size - LOCOS-like effect at the poly-Si edges). This example illustrates selected degradation that occurs while scaling MOSFETs. Analyze the results and make comments about correlations between fabication of the structures and their operation.

2. Run example #bjtex04.in "NPN Gummel plot and Ic/Vce". Determine if models (in the menu) of the BJT operation include important parameters (ex. bandgap narrowing, F-D statisticts, Auger and SRH recombination etc.). If not you may want to include them to see their influence on devive operation. Change dopant concentration at the surface of the emitter to 5e20 cm-3. See the effect in the Gummel characteristics Ic(Vbe) and Ib(Vbe). Change the base width from 0.1 µm to 0.05 µm. Adjust other parameters of fabrication that are needed to accomodate this thinner base. Run the program again and compare the results.

 

Each project has to be written individually (one person - one unique report) even though you are encouraged to work in groups on the simulations.

Return to ECE 7366 General or Home 

Return to Silvaco

 

Important comments regarding Atlas simulator

Input files contain 1) Structure Specification (mesh, region, electrode, doping), 2) Material Model Specification (material, models, contact, interface), 3) Numerical Method Specification (method), 4) Solution Specification (log, solve, load, save), and 5) Extract (Extract, Tonyplot).

Atlas can run (go atlas) either using a device structure built in a preceding Athena ex. STRUCTURE OUTF=bjt01.str or using a structure created earlier by DevEdit, Athena, or Atlas (MESH INFILE=nmos.in). Defining the initial mesh of the structure is important for correct simulations. Ex. X.MESH LOCATION=1 SPACING=0.2 sets the grid points dense enough (here: in x direction) for optimized accuracy and time of calculation. Atlas can use Regrid command to add fine grid in regions with fast varying parameters (steep profiles, small radius etc.).

Electrodes names and Contacts have to be specified. Contacts are assumed to be ohmic or else, if you specify the workfunction (or just list a material that have workfunction stored) they will be treated as Schottky contacts.

Models for devices to be used in your simulations: 1) Carrier Statistics (Fermi-Dirac, Bandgap Narrowing - mainly for BJT), 2) Mobility Models ( CVT - for MOSFET, Klaassen - for BJT = separates minority and majority carriers' mobility), 3) Recombination Models ( CONSRH - concentration dependent lifetime Shockley Read-Hall, Auger - important in BJT, S.N and S.P. - Surface recombination in Si at insulator interfaces, in the INTERFACE statement), 4) Impact Ionization (IMPACT SELB), Tunneling Models and Carrier Injection Models (FNORD - Fowler-Nordheim - tunneling through insulators, BBT.STD - band-to-band direct transition at high fields).

DC solutions are obtained for specified conditions such as a gate and a drain voltage SOLVE VGATE=2.0 SOLVE VDRAIN=1.0 or sweeping the bias SOLVE VCOLLECTOR+2.0 SOLVE VBASE=0.0 VSTEP=0.05 VFINAL=1.0 NAME=base. Results (SOLVE VGATE=1.0) are stored OUTF=solve_vgate1 and can be used to produce a family of curves after loading the file LOAD INFILE=solve_vgate1 that would be saved in LOG OUTFILE=MOS_DRAIN_SWEEP1 after solving SOLVE NAME=drain VDRAIN=0 VFINAL=3.3 VSTEP=0.3.

 

Return to ECE 7366 General or Home 

Return to Silvaco