ATLAS常见问题解答Q1 The ATLAS User’s manual, Vol. II said that the conversion from resistivity to carrier concentration is base on table, which is derived from Arora mobility model. Could you inform me in more detail about thosetables and the conversion tool (if there is one)?A1 The conversion of resistivity to carrier concentration in silicon using the RESISTI parameter in the DOPING statement is base on the followingequation:Carrier Concentration = 1 / (R * q * u) (in cm-3)where R is the resistivity (in ohm-cm); q is the charge; and u is themobility (in cm^2/V.s).In ATLAS, Arora mobility is used for computing the carrier concentration.The computation of the resistivity to carrier concentration in Silicon forboth N-type and p-type carrier are as shown in the figure below.Top Q2Part I: How can I simulate the short circuit current, Isc and open voltage, Voc of a Solar Cell in ATLAS?A2 The short circuit current, Isc can be extracted from the IV curve when thevoltage is 0 while the open circuit voltage can be extracted from the IV curve when the current 0. The figure below shows the location of the Isc and Voc values. The commands used to extract both values are as follows:extract init inf="Solar_IV.log"# Extract Short Circuit Current extract name="short_circuit_current" y.val from curve(v."cathode",i."cathode") where x.val=0# Extract Open Circuit Voltage extract name="open_circuit_voltage"x.val from curve(v."cathode",i."cathode") where y.val=0TopQ3 Part II: How can I include a variable load resistance (external load) into the simulation program and simulate the IV characteristics of a Solar Cell?A3 There are 2 methods of adding an external resistance to a device:i. The first method is to add a lump element; andii. The second method is to add a distributed contact resistance.The first method is to specify the value of the resistance in the "RESISTANCE" option of the CONTACT statement. However, this method is only applicable for planar metal-semiconductor surface.The second method is to add a distributed contact resistance, fornon-uniform metal-semiconductor surface. For this method, theCON.RESIST option was used instead in the CONTACT statement. Below give an example of how to compute the CON.RESIST value.Assuming an external load of 50ohm at the Cathode contact, the CON.RESIST (in ohm-cm^2) is calculated as follow:CON.RESIST = 50 x Cathode_length x cathode_width = 50 x 140 um x 1um = 7e-5 ohm-cm^2Then, you need to specify the CONTACT statement as below:CONTACT NAME=CATHODE CON.RESIST=7E-5Below shows the simulated IV characteristics of a Solar Cell with and without a 50 ohm external load.To extract the channel length, use the following EXTRACT statement as follows:extract name="junc_source" xj silicon mat.occno=1 y.val=0.22junc.occno=1 extract name="junc_drain" xj silicon mat.occno=1y.val=0.22 junc.occno=2 extract name="Channel Length"$"junc_drain"-$"junc_source"Top Q4 How to extract Ioff in SILVACO TCAD tools?A4 Firstly you should know the definition of Idsat and Ioff. When Ioff is defined as the drain current at gate voltage equals to 0v, you can usefollowing statement:Extract name=”Ioff” y.val from curve(v.”gate”, i.”drain”) where x.val=0After run this statement, you will get a result which is stored in file"results.final".TopQ5 I have finished the simulation of BJT and get the IV curve, but I could find how to display the curve beta vs. base voltage. Is that possible to display such curve?A5 Yes, it can. Firstly launch TonyPlot and open the log file which include IV curve. Then, click “tools”-> “display” in the menu, select “Function 1” in the “Y Quantities”After that, click button “Functions…” which is in the left-bottom. A window will pop up like following:In this window, type in “Collector Current / Base Current” which is also showed in above picture ( because beta is defined as beta=Ic/Ib )Then click “Ok” to finish. Then you will get a curve which x axis is base voltage, and y axis is beta. See follows:TopQ6 How to obtain the Capacitance-Voltage (CV) curve of a MOS capacitor at low frequency?A6 In the simulation of CV curve, SRH model needs to be added to the input deck file. This is because when reverse biasing the MOS capacitor,GENERATION in the depletion region actually dominate overRECOMBINATION, as the carrier concentrations are less than theirthermal equilibrium values; i.e. pn < ni2.The generated carriers in the depletion region can contribute to the flow of an external current, I. This I = Integral of C w.r.t. V. Therefore, whenthe generated carriers increases -> I increases -> C increases.The generation rate in the depletion region, G = ni/2T, where T is thelifetime of the effective lifetime, i.e. T = (Tn + Tp)/2. Therefore, G isinversely proportional to T => The shorter the lifetime T, the greater the generation rate G => greater I => greater C.Hence it is necessary to add the SRH model and also reduce theeffective lifetime of the carriers in the bulk for simulating the CV curve.This can be done by adding the following statements:material taun0=1e-11 taup0=1e-11 model srhBelow shows the simulated Capacitance-Voltage (CV) curve of a MOS capacitor at high and low frequency.TopQ7 The MOS capacitor being a two terminal device, the two capacitances should be identical in magnitude. Simulation however shows them to be a few orders of magnitude different. Our theoretical estimation tell us that Csubstrate-gate between 8e-16 to 1.5e-15 Farad is correct.Cgate-substrate is between 3.9e-19 and 4.8e-19 F is confusing to us.A7 Atlas performs a small signal ac analysis as described in the paper by Laux Ref. 74 of Atlas vol II Bibliography.According to ref. 46, the admittance matrix is defined as:Y(ij) = I(i) / V(j)where I(i) and V(j) are the phasor terminal current and voltagerespectively. The real part and imaginary part of the admittance matrix determine the conductance matrix G and the capacitance matrix Crespectively.In Atlas, it is these conductance and capacitance that are being output.For instance, the capacitance matrix C is calculated as follows:Imag Y21 = -1*(2*3.14*frequency*Cgate>substrate) [input port=Gate,output port=Substrate]Imag Y12= -1*(2*3.14*frequency*Csubstrate>gate) [inputport=Substrate, output port=Gate]Therefore, the value of Cgate>substrate does not equal toCsubstrate>gate.Top Q8 In general, how to solve convergence problem?A8 To solve convergence problem,1) Try the refine your mesh especially at areas where carriers activity aremost prominent such as areas near and at the junction or the Schottkycontact,2) Use a smaller step size (e.g. vstep = 0.1 or even vstep = 0.01) ratherthan a large one such as 2V.3) By default, only Newton method is used. Try to use both the Gummeland Newton Methods in the Method statement. This will cause the solver to start with Gummel iterations and then switch to Newton, ifconvergence is not achieved. This is a very robust, although more time consuming way of obtaining solutions for any device.4) Refer to ATLAS User's Manual on "Choosing Numerical Method" pp.2-27 to decide on the most appropriate methods to be used for yourdevice.TopQ9Can you tell me how to retrieve the default parameters of SiC in theATLAS?A9The default value of all material can be found in ATLAS User's manualvol. II Appendix B (You should have a pdf format of ATLAS manual in your silvaco directory: /doc). You may also view it by adding the "print" option in the Model statement.TopQ10I notice in Appendix B, Page 21, there lists some optical properties of several semiconductors such as Si, AlAs, GaAs... How about ternary compound semiconductor and quaternary semiconductors, such as InGaAs or InGaAsP? A10All the default complex index of refraction for Si, AlAs, etc are listed inB-21. For those materials which are lack of complex index of refraction, you need to specify the index indicated in pp. 8-11 of ATLAS User's Manual Vol. 1.i. To set single values of Refractive Index regardless of wavele ngth, you may use the REAL.INDEX and IMAG.INDEX parameters of theMATERIAL statement to set the real and imaginary indices of the ternary and quaternary material, region or regions;ii. To set a Wavelength Dependent Refractive Index for multispectral simulations, you need to specify index versus wavelength:a.In a file known as the INDEX file. Note that if you use this method, onlylinear interpolation from the table in the INDEX file will be used.b. Use the C-INTERPRETER function in which it can returns wavelength dependent real and imaginary indices.Also note that you can add the INDEX.CHECK parameter to any SOLVE statement to print out the refractive indices being used for that bias step.TopQ11 How to define a new material such as InGaN by using the C-Interpreter? A11 It is possible to simulate the user-defined material such as InGaN by using the C-Interpreter. The template file for the C-Interpreter is asattached. Please save this file in your working directory i.e.C:\SILVACO\work. The procedure involves in specifying e.g. the bandcomposition, for user-defined material i.e. InGaN using the C-Interpreter are listed below:1. In your MATERIAL statement, you need to define: materialmaterial=InGaN f.bandcomp="xxx.lib"where F.BANDCOMP specifies the name of a file (in this case is xxx.lib) containing a C-INTERPRETER function for the specification oftemperature and composition dependent band parameter models.2. Next, open the template file i.e. C_template, using WORDPADApplication. Go to the F.BANDCOMP function and you will seesomething as shown below:/** Temperature and composition dependent band parameters* Statement: MATERIAL* Parameter: F.BANDCOMP* Note: This function can only be used with BLAZE.* Arguments:* xcomp composition fraction "X"* ycomp composition fraction "Y"* temp temperature (K)* *eg return: band gap (eV)* *chi return: affinity (eV)* *nc return: conduction band density of states* *nv return: valence band density of states* *degdt return: derivative of Eg with respect to T*/int bandcomp(double xcomp,double ycomp,double temp,double*eg,double *chi,double *nc,double *nv,double *degdt){return(0); /* 0 - ok */}This function accepts the X composition, Y composition value and the temperature. It can return the bandgap, electron affinity, conduction and valence band density of states and also derivative of Eg with respect to temperature. ATLAS will then solves the electrical characteristics self consistently with the returned values.You can add your equations after the parenthese {, to express the eg, chi, nc and nv in term of the xcomp and/or ycomp, etc ... For example, if the bandgap equation for In(x)Ga(1-x)N is 1.89x + 3.42(1-x) - 3.8x(1-x), the command is as follows:*eg = 1.89*xcomp + 3.42*(1-xcomp) - 3.8*xcomp*(1-xcomp);(Refer to Appendix A: C-Interpreter Functions in ATLAS User's Manual Vol II for more detail)3. After you have added the equations, save this file as "xxx.lib" and make sure that this file is in the same working directory as the input deck file.4. You can then run the input deck file.TopQ12 In Atlas, I can only define CVT transverse field dependent mobility model. But how to define parallel field dependent CVT MODEL?A12 To define parallel field dependent CVT MODEL, go to Deckbuild Commands -> Models. Select "CVT" and "Field Dependent" under theMOBILITY section. Then you can define their parameters by clicking on the "Define Parameters" -> "Field Dependent ..." or "CVT ...".TopQ13 How to get low temperature characteristics of MOSFETs devices in ATLAS for 2D simulation?A13 You can simulate the low temperature characteristics by setting the "Temperature" parameter in the MODEL statement. For example, to set the temperature to 100K, type: model temperature=100TopQ14 I would like to use different dielectric constant value of gate materials. How to incorporate it?A14 You can define the dielectric constant of the Gate using the MATERIAL statement (refer to pp. 15-53 of ATLAS User's Manual Volume 2) asfollows:MATERIAL material=X permittivity=Ywhere X is the material from the Table B-1 in Appendix B (ATLAS User's Manual Volume 2) that the MATERIAL statement should apply. All theregions defined as being composed of X material will be affected. Y isthe dielectric permittivity of the material X.TopQ15 What are the parameters that can be adjusted so that the bandgap between Ec and Ef will be altered to that of strained-Si?A15 The correct parameters to be adjusted are "NC300" and "NV300" in the MATERIAL statement. This is because (referring to ATLAS User Manual pp. 3-5 "Effective Density of States") from equation 3-27 and 3-28, Ncand Nv is a function of (Ef - Ec) and (Ev - Ef) respectively. Therefore, by decreasing the value of, says Nc, and with a given doping concentration, the difference between (Ec - Ef) will also decrease. Note that the default values for NC300 and NV300 is 2.8e19 and 1.04e19 respectively.TopQ16 How can I plot the energy band diagram in TONYPLOT?A16 To plot the energy band level in TONYPLOT, in your input deck file, you need type the statements:output qfn qfp con.band val.bandstruct outfile=test1.strThe OUTPUT statement specifies the eletron quasi-fermi level, holequasi-fermi level, conduction band and valence band (refer to the ATLASUser's Manual Vol. 2 pp. 15-106 for the OUTPUT statement) to beincluded into the structure file (second statement). Without thisstatement, you will not have the fermi energy level, conduction band, etc in your structure display. Then plot the structure "test1.str" and do aCutline (refer to VWF Interactive Tools Vol. 1 Tonyplot Cutline pp. 6-32).Top Q17Why the capacitances extracted by ATLAS has two pair of values for e.g.the capacitance between base and emitter, ATLAS calculateC(base>emitter) and C (emitter>base)? Furthermore, why are these two capacitances i.e. C(base>collector) and C(collector>base) not equal?A17 This is because you are performing DC analysis together with AC analysis using the SOLVE statement:# 2-base 3-emittersolve v2=xxx v3=xxx vstep=xxx electr=23 nstep=x ac freq=1e6 directIn this case, ATLAS will actually pump-in the small signal at the Emitter electrode, Base Electrode and Collector Electrode one at a time andcalculate the Conductance and Capacitance. If the small-signal is input at the Base Electrode, then the capacitance displayed in TONYPLOT will be Cemitter>base, Cbase>base and Ccollector>base. Since in morecases, we are only interested in the AC analysis with small-signal at the Base Electrode, therefore, Cemitter>base, Cbase>base andCcollector>base are the correct capacitances to look for.The capacitance values must be extracted when the transistor is in theOFF state.Above shows the capacitances plot of an HBT. As can be seen from the graph, C(base_collector) and C(collector_base) are exactly the same below V=0.65V (threshold voltage of the HBT is about 0.65V). Beyond this threshold voltage, the capacitance values become different.Below shows the capacitance extracted when the HBT is in the OFF state i.e. Vbase and Vcollector=0.1V at frequency=1MHz:Small Signal Electrode: emitterElectrode Conductance(siemens/micron)Capacitance(farads/micron)emitter -1.05E-14 -1.02E-15 base 1.19E-14 1.02E-15 collector -9.27E-16 3.50E-26 Small Signal Electrode: baseElectrode Conductance(siemens/micron)Capacitance(farads/micron)emitter 1.22E-14 1.02E-15 base -1.03E-12 -1.89E-15 collector 1.02E-12 8.68E-16 Small Signal Electrode:collectorElectrode Conductance(siemens/micron)Capacitance(farads/micron)emitter -1.16E-15 3.50E-26base 1.02E-12 8.68E-16collector -1.02E-12 -8.68E-16As can be seen, the simulatedC(emitter_base)=C(base_emitter)=1.02E-15,C(base_collector)=C(collector_base)=8.68E-16 ... etcIn short, the capacitance should be simulated without turning ON thetransistor.TopQ18When simulating the spectrum response of an InGaN/GaN QuantumWell LED, the simulated spectrum is not symmetrical on both sides. The spectrum simply falls to zero intensity abruptly after reaching the peakintensity as shown below. May I know how to obtain a symmetricalspectrum plot?A18 You need to include the Lorentz line broadening model to the quantum well model. Line broadening is enabled by specifying LORENTZ on the MODEL statement. The extent of the broadening can be specified byeither the WELL.GAMMA0 or the WELL.TAUIN parameters of the MATERIAL statement. By adding the LORENTZ model and changing the WELL.GAMMA0 to 2e-2, below shows the simulated LED Spectrum plot.。