ESD与TCADSilvaco仿真(浙江大学)

1、esd与tcad仿真报告人：浙大微电子 崔强email: 共163页2welcome! 热烈欢迎各位参加本次讲座的学员。由于本人水平有限，在座的各位如果有什么问题，请立刻打断我。共163页3welcome!本次讲座分3个小节，中途有两次休息，欢迎大家和我探讨。1. 单元1：tcad简介2. 单元2：esd的tcad仿真简介3. 单元3：esd的仿真评价体系共163页4单元1 1.1：tcad总体简介 1.2：tsuprem4/medici 1.3：athena/atlas 1.4：dios/ dessis （ise-tcad）共163页51.1：tcad总体简介tcadtechnology c

2、omputer aided designtsuprem/medici（avanti，被synopsys收购）athena/atlas（silvaco公司）dios/dessis（ise公司，被synopsys 收购）共163页61.2：tsuprem4/medici tsuprem4/medici是avanti公司的二维工艺、器件仿真集成软件包。tsuprem4是对应的工艺仿真软件，medici是器件仿真软件。在实践中，可以将tsuprem4的工艺仿真的结果导入到medici中，从而进行较为精确的仿真。共163页71.2 .1：tsuprem4capabilitiestsuprem-4 sim

3、ulates silicon ic process fabrication1.ion implantation2.epitaxial growth3.diffusion4.oxidation of silicon and polysilicon5.etching and deposition6.silicidation of silicon and polysilicon共163页8 specifications1.two-dimensional2.supports up to 40,000 nodes3.written in c共163页9 command input language th

4、e input language is made up of commands and corresponding parameters1.there is only one command per line2.the line can be up to 80 characters long. if it s longer, we can continue in the next line, but the last character in the previous line must be a “+” character.共163页10types of commandsthere are

5、two types of commands1. declaration (used to set parameters)2. action (execution used to perform a process step)共163页11parameters there are three types of parameters1. numerical (e.g. temp=1000)2. logical (e.g. clear)3. character (e.g. “nmos3a”)共163页12files associated with running tsuprem-4 simulati

6、ons1. input file (*.inp)2. output file (*.out appended automatically by program)3. tsuprem-4 structure file (*.str)4. medici/davinci file (*.dev)5. universal (tif) format file (*.tif)共163页13create a simulation1. setting up the initial grid2. models and coefficients (method)3. process statements (dep

7、osition, expose, develop, etch, implant, diffusion, epitaxy)4. electrical calculations in tsuprem-45. extracting results (non-electrical) from tsuprem-4共163页141.2 .2：medici features：medici solves poissons equation and the current continuity of electrons and holes in two dimensionsthese equations can

8、 be extended to include the heat equation and the energy-balance equationsthe following modes of analysis can be considered: dc, ac, transient共163页15 doping and structure information can either be input from a process simulator e.g. tsuprem4 or generated within medici a wide range of mobility and re

9、combination/generation models available output to parameter extraction programs such as aurora possible共163页16 full post processing capabilities including plotting internal quantities, terminal characteristics extract capability makes calculations with a wide range of parameters possible for pre and

10、 post processing solutions. a target parameter can be identified for optimization共163页17advanced application modules1. lattice temperature aam solves the heat equation2. optical device aam enhanced radiation effects, ray tracing3. heterojunction device aam conduction across a material boundary with

11、discontinuous energy gap共163页184.programmable device aam allows a charge boundary condition on a floating electrode5.circuit analysis aam allows devices to be treated as circuit elements in a spice type circuit6.anisotropic device aam allows anisotropic material parameters useful in the treatment of

12、 sic type applications共163页19meshinitiates a mesh and must appear first when defining a structure.x.meshy.mesheliminateused to specify exact locations of mesh lines produces a rectangular grid which can be reduced in density by using eliminate to remove excess nodes away from area of interestcreate

13、a simulation共163页20spreadboundaryolder statements whose function is no longer so necessary. spread allows the creation of a locos shaped structure on a rectangular mesh and boundary allows a set of coordinates to be input to define a regions topography.共163页21tsuprem4used to transfer surface feature

14、s and doping profiles from tsuprem4 onto an existing medici meshstitchnew command to allow multiple tsuprem4 files to be usedregionused to define regional properties where no material data already exists共163页22electrodeadds location of electrodes to structurerenamerenames electrodes or regionsprofil

15、eallows addition of doping information either by creating simple profiles or inputting from a process simulatorregridallows regridding of mesh based on some internalquantities共163页23 some statementsregrid statement the regrid statement can be used to refine a grid for greater accuracy.1. specify qua

16、ntity to be used e.g potential, electric field, min.carr2. refinement criterium e.g ratio=2 “regrid doping log ratio=2 in.file=test.dop + smooth=1”共163页24 rename electrodes when electrodes are transferred from tsuprem-4 to medici, they will be numbered sequentially. it is often convenient to rename

17、them with names easier to remember. “rename electrode oldname=1 newname=source rename electrode oldname=2 newname=drain”共163页25 models the physics is selected on the model statement. this is supported by a mobility statement and a material statement which allow default parameters for the mobility an

18、d other models to be altered within specified regions or materials.共163页26 the models that can be selected can broadly be divided into the following categories:1.recombination and generation models2.mobility modeling3.models affecting relation between carrier density and electric field. i.e bandgap

19、narrowing, fermi-dirac or boltzmann statistics, quantum mechanical4.energy balance modeling共163页27modeldesciptionsrhshockley read hallconsrhsrh + concentration dependant lifetimesaugerauger recombinationr.tunnelsrh including tunnelling in presence of strong electric filedsimpact.iclassic chynoweth e

20、xpressionii.tempinvokes a temperature based version of the impact ionization model for use with the energy balance modelrecombination and generation models共163页28modellow fieldtransverse fieldparallel fieldcommentsccsmobcarrier-carrier scatteringconmobconcentration dependence from tables 300kanalyti

21、canalytic alternative to conmob with temp dependencephumobcarrier-carrier scattering, different donor and accetor scattering, screening, useful for bipolarsmobility models共163页29modellow fieldtransverse fieldparallel fieldcommentslsmmostreats surface scattering and bulk effectsgmcmobmodified lsmmob

22、to include screened and unscreened impurity scatteringmobility models共163页30modellow fieldtransverse fieldparallel fieldcommentssrfmobbasic and enhanced model for surface scattering.requires vertical grid spacing inversion layersrfmob2unimobneeds rectangular grid in inversion layer models surface sc

23、atteringprpmobgeneral model for degradation of mobility with transverse electric field mobility models共163页31modellow fieldtransverse fieldparallel fieldcommentstfldmobuniv texas mobility modelfldmobcarrier heating and velocity saturation effectshpmobaccounts for both parallel and perpendicular fiel

24、d dependencemobility models共163页32modeldescriptionfermidirfermi dirac statistics instead of boltzmann.incompleincomplete ionization of impuritiesbgnbandgap narrowing modeling especially important for bipolarsqm.philiaccounts for quantum mechanical effects in mosfet inversion layers using van dorts b

25、andgap widening model.other models共163页33 boundary conditions the standard boundary conditions are that the normal component of the electric field and the current densities disappear at the boundaries (neumann boundary conditions) except at the electrodes where fixed conditions can be ascribed (diri

26、chlet conditions). variations in those conditions can be applied using the contact statement and the interface statement can be used to input discontinuities at material boundaries.共163页34 contact statement the default conditions are that ohmic conditions applied and the electrostatic potential equa

27、ls the applied voltage. the carrier densities are then calculated from the space charge neutrality relation. “ contact name=gate workfunction=4.35 comment (defaul unit: ev) contact name=base current”共163页35 solution technique in order to obtain the solution we essentially have to decide on two thing

28、s.1. select equations to be solved2. decide how to bias our structure statements which fall into this group are symbolic, method, solve.共163页36symbolic statement1. poissons equation2. electron current-continuity equation3. hole current continuity equation4. lattice temperature (heat) equation5. elec

29、tron energy-balance equation6. hole energy-balance equation共163页37共163页38 what to do when the folloing message appears on the screen?“ error number 166 detected in line number xx. more than 4 solutions failed to converge. execution terminated! ”共163页39 the primary causes of non convergence are:1. po

30、or initial guess bias step too large2. lack of necessary physical models3. poor simulation grid4. depletion layer touching the electrode共163页401.3：athena/atlas tsuprem4/medici是avanti公司的二维工艺、器件仿真集成软件包。tsuprem4是对应的工艺仿真软件，medici是器件仿真软件。在实践中，可以将tsuprem4的工艺仿真的结果导入到medici中，从而进行较为精确的仿真。共163页411.3.1：athenan

31、登录界面共163页42ngui 方式设定网格共163页43n编程方式设定网格共163页44n定义初始衬底共163页45n栅极氧化共163页46共163页47n离子注入共163页48n多晶硅栅的淀积共163页49共163页50n几何刻蚀共163页51共163页52 n多晶硅氧化 “method fermi compress”共163页53共163页54#polysilicon dopingimplant phosphor dose=3e13 energy=20 crystaln多晶硅掺杂共163页55共163页56n氧化层淀积和侧墙氧化隔离共163页57“#source/drain implan

32、timplant arsenic dose=5e15 energy=50 crytal#source/drain annealingmethod fermidiffus time=1 temp=900 nitro press=1.00”n源/漏极注入和退火共163页58共163页59“#open contact window etch oxide left p1.x=0.2”n氧化物的刻蚀和金属的淀积刻蚀共163页60共163页61“#aluminum deposition deposit aluminum thick=0.03 divisions=2；”共163页62共163页63“#etc

33、h aluminum etch aluminum right p1.x=0.18”共163页64共163页65n半个nmos结构的镜像共163页66n电极的确定 和保存athena结构文件共163页67 “electrode name=source x=0.1 electrode name=drain x=1.1 electrode name=gate x=0.6”共163页681.3.2：atlas共163页69 n登录界面共163页70n导入athena结构共163页71n模型命令组共163页72ncategory栏中选择recombination选项 共163页73n数字求解方法命令组共

34、163页74n解决方案命令共163页75ntonyplot绘出idvds特性曲线族 共163页761.4：dios/ dessis 工艺及器件仿真工具ise-tcad（tcad：technology computer aided design）是瑞士 ise ( integrated systems engineering ) 公司开发的dfm（design for manufacturing）软件，是一种建立在物理基础上的数值仿真工具，它既可以进行工艺流程的仿真、器件的描述，也可以进行器件仿真、电路性能仿真以及电缺陷仿真等。共163页771.4.1： dios dios简介简介 dios输入

35、文件是由一系列连续执行的命令构成。dios输入文件的后缀及扩展名为：“_dio.cmd”。dios的输入语言并不区分字母的大小写。不过，文件名和电极触点名是区分大小写的。共163页78 一个典型的dios文件一般以初始化的命令开始，并且初始化命令不可以省略。例如： title(.)grid(.)substrate(.)共163页79 之后，可以根据需要选择性地添加仿真命令语句，如：mask(.)implant(.)diffusion(.)deposit(.)etching(.)共163页80 在完成了这些仿真语句之后，可以用：1d(.)save(.)命令对仿真结果进行保存。用“end”命令作为

36、整个文件的结束。共163页81title(.)” 命令命令该命令总是出现在dios输入文件的最开始的地方，用来对仿真进行初始化。例如：title(simple nmos example)这条指令对仿真进行了初始化，并且把图形窗口命名为“simple nmos example”。title(test, sidiff=off, newdiff=1)该命令同样也是对仿真进行初始化，并把图形窗口命名为“test”，同时，sidiff=off表示仅在除硅以外的层次扩散，比如氧化层和多晶硅，以节约仿真时间。newdiff=1表示所有层次都定义网格和掺杂，n各种命令说明共163页82 grid(.)”命令命

37、令网格命令一般跟在“title”命令之后，它是用来定义器件结构初始化网格的，同时也包括了器件的横向和纵向范围。在默认的情况下，dios在每一步仿真之后都会对网格进行重新编制，这样可以解决在制做工艺中几何尺寸和掺杂浓度改变而引起的问题。如果说没有明确指定网格调整参数，那么dios将会通过自己默认的调整标准对网格进行调整。例如： grid (x (0.0, 0.4), y (-10.0, 0.0), nx=2)在该命令中没有对网格的调整标准。它对器件横向范围从0um到0.4um，纵向范围从-10.0um到0um的网格进行了初始化的指定。参数nx=2定义了所包含三角形为2,即网格x方向是由2个三角形

38、构成。共163页83 “substrate(.)”命令命令定义硅衬底的晶向和掺杂。例如：substrate(element=b, concentration=5.0e15, orientation=100)该命令定义了硅衬底的晶向是（100），掺杂浓度为5.0*1015 atoms/cm3的硼。共163页84 “mask(.)”命令命令在dios中，这条命令是用来对仿真中所要用到的掩膜板进行仿真，以及完成掩膜板形成图案的沉积。例如：mask (material=resist, thickness=800nm, x (0.1, 0.3)该命令定义了一块厚度为800nm的光刻胶（resist表示p

39、hotoresist的意思，即光刻胶），其覆盖的范围是横向位置从0.1um到0.3ummask (material=po, element=p, concentration=3e19, thickness=180nm, xleft=0.2, xright=0.4)该命令沉积了一层厚度为180nm的掺入杂质磷的多晶硅层，其范围为从0.2um到0.4um。共163页85“implant(.)”命令命令这条命令是用来对离子注入进行仿真的。其中的“function”参数允许用户选择使用“分析注入”还是“monte carlo注入”。如果用户选择前一种，则注入参数来自于默认图表。如果需要使用其它注入参数

40、，可以另外创建注入图表，并在仿真中使用。例如：implant(element=bf2, dose=5.0e12, energy=25kev, tilt=7)该命令以倾斜角度为7，能量为25kev，注入剂量为5.0*1012 atoms/cm2的bf2离子。implant(element=as,dose=1.0e14,energy=300kev,tilt=0,rotation=-90,function=crystaltrim)该命令用monte carlo方式仿真，用crystaltrim函数注入砷离子。共163页86“diffusion(.)”命令命令在dios中，“diffusion”是用来

41、对器件制做工艺中所有高温步骤进行仿真的命令。包括：热退火、氧化、外延层的生长和硅化物的生长。可以选择的扩散模型中既有简单的常量扩散模型，也有将杂质和点缺陷配对等都包括的复杂完整模型。平衡态和瞬态聚集模型允许考虑杂质激活效应的精确仿真。另外，还支持杂质和点缺陷等参数的自定义。例如：diffusion(temperature=1050, time=10s)该命令仿真了温度为1050度，时间为10秒的高温环境。diffusion(temperature=1000, time=20min, atmosphere=o2)该命令仿真了干氧氧化，温度为1000度，时间为20分钟，气体为o2。diffusio

42、n(atmosphere=epitaxy, time=1.0s, temperature=1050, growthrate=1000 nm /s, element=ge, concentration=1.0e20）该命令仿真了一个sige外延层的生长,在生长外延的环境中，加入ge，浓度为1.0e20，就形成了sige的外延层，时间为1.0秒，温度为1050度，生长速率为1000 nm/s。共163页87 “deposit(.)”命令命令在dios中，该命令是用来沉积物质层的。用于各向同性或异性沉积、表面平整化、选择性沉积以及化学机械抛光。例如：deposit(material=po, thic

43、kness=0.2um, element=p, conc=3.0e19)该命令进行多晶硅层的沉积仿真，厚度为0.2um，掺杂浓度为3.0*1019 atoms/cm3的磷原子。deposit (material=ox, dtype=fill，yfill=2.0um )该命令用以仿真化学机械抛光。“fill”表示平整化，yfill=2.0um表示机械抛光的纵向距离。共163页88 “ecthing(.)”命令命令该命令用来仿真刻蚀。该命令包含多个选项，可以在仿真中灵活的定义刻蚀形状。例如：etching (material=ox, time=5.0min, rate(iso=100nm/min)

44、该命令仿真了一个刻蚀时间为5 分钟，刻蚀速率为100 nm/min的各向同性的氧化物的刻蚀。共163页89 “1d(.)”命令命令这是一个保存命令，进行过仿真的器件，任何x-y分布的dios变量都可以通过该命令来保存。例如：1d (file=channel, xsection(0.0), species (btotal, ptotal), fac= -1.0, append=off)这是对在x= 0.0处的，硼和磷的总浓度作为深度的函数进行保存。fac= -1.0为坐标比例缩小因子共163页90 “save(.)”命令命令这条命令用来保存器件的最终结构，并且文件可以载入重新进行仿真。在“sav

45、e”命令执行之后，文件可以由dessis载入进行器件仿真。例如：save(file=tst)把器件保存为文件“tst.dmp.gz”。save（file=nmos, type=mdraw）保存为mdraw格式，提供dessis作为器件仿真文件共163页91dessisnise-tcad的仿真结构流程共163页92file * 输入文件：grid = nmos_mdr.grddoping = nmos_mdr.dat* 输出文件plot = n3_des.datcurrent = n3_des.pltoutput = n3_des.log 共163页931.“file”部分主要定义器件结构的输入

46、文件和输出文件的名称；2. “ * ”引导注释行；3.“grid”和“doping”语句分别指定器件结构的网格文件和掺杂文件；4.“plot”语句定义仿真时计算的变量，扩展名为“_des.dat”；5.“current”语句定义最后输出的电学数据（比如电流、电压、电极上电荷），扩展名为“_des.plt”；6.“output”语句定义输出日志文件，记录dessis运行情况，扩展名为“_des.log”。共163页94electrode name=source voltage=0.0 name=drain voltage=0.1 name=gate voltage=0.0 barrier=-0.

47、55 name=substrate voltage=0.0 共163页951.“electrode”部分定义器件的电极相关信息。2. 值得注意的是，在多晶硅“gate”上，接触定义必须是欧姆接触。3.“name=”语句定义每个电极，这个电极名称必须和grid文件定义一致；4.“voltage=0.0”语句定义电极的电压初始值；5.“barrier=-0.55”语句定义金属半导体功函数差，这样多晶硅电极才能把它当成金属。共163页96physics mobility (dopingdep highfieldsat enormal)effectiveintrinsicdensity (bandga

48、pnarrowing (oldslotboom)共163页971.“physics”部分定义器件仿真过程中使用的物理模型。2.“mobility (dopingdep highfieldsat enormal)”语句定义三个模型：掺杂依赖（doping dependence）模型、高电场饱和模型、横向电场依赖（transverse field dependence）模型；3.“effectiveintrinsicdensity (bandgapnarrowing (oldslotboom)”语句定义硅能隙窄化模型，它决定载流子的浓度。共163页98plot edensity hdensity

49、ecurrent hcurrentpotential spacecharge electricfieldemobility hmobility evelocity hvelocitydoping donorconcentration acceptorconcentration1.plot”部分定义所有的计算变量，dessis要仿真的变量都将被存入plot文件。共163页99math extrapolaterelerrcontrol1.“math”部分定义dessis仿真时算法的设置，包括仿真器类型、仿真误差标准的设置。2.“extrapolate”语句定义仿真时采用外推法定义迭代下一步的数值；

50、3.“relerrcontrol”语句定义迭代反复计算时加入误差控制。共163页100solve #初始解决方案：poissoncoupled poisson electron quasistationary (maxstep=0.05goal name=gate voltage=2 ) coupled poisson electron 共163页1011. “solve”部分定义一系列的仿真，包括仿真所需要的一些参数。2. “poisson”语句定义初始化采用非线性泊松方程；3. “coupled poisson electron ”语句定义在初始偏置下电子的连续性方程；4. 这条语句定义仿

51、真的一些设定，包括最大步长0.05，栅压仿真到2v，采用泊松方程仿真。共163页102nmos的输出特性共163页103互动时间 any question?共163页104休息时间共163页105单元2 本单元讲述了几个用tsuprem4/medici仿真esd的几个例子：共163页106单元2 2.1：直流仿真 2.2：混合电路仿真 2.3：温度仿真 2.4：极值功率密度 2.5：功率分布仿真共163页1072.1：直流仿真n直流仿真ggnmos结构图共163页108n直流仿真ggnmos的i-v图共163页109n直流仿真ggnmos结构图共163页110n直流仿真scr的i-v图共163页1112.2：混合电路仿真nhbm放电模式的等效电路图共163页112n混合电路仿真的器件共163页113n5kv的esd情况下的i-t，v-t图。共163页114n5kv的esd情况下的i-t，v