Добавил:
Upload Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:

Книги2 / 1993 Dutton , Yu -Technology CAD_Computer Simulation

.pdf
Скачиваний:
66
Добавлен:
11.03.2016
Размер:
17.91 Mб
Скачать

C.l. ID BJT

359

Title Stanford' BiCMOS-2um Process (npn)

 

Comment

Start with <100> Silicon. p-doped to 20

ohm Resistivity.

Initialize

<100> Silicon Boron Concentration-9.0E14 Thickness-6

+Spaces-300

Comment

Changing Default Coefficients to Prevent Excessive Outdiffusion

Wet02 <100> Lin.H.0=2.428e6 Lin.L.0=2.793e4

Segregation Phosph Si

lOx Mui.O=O.O Model. 1

Segregation Phosph Si lAir Trans.0=1.66e-7

Comment

Initial Oxidation to 2370 Angstroms

Diffusion

Time=35

Temp=BOO

T.Rate=5.714

Diffusion

Time=10

Temp=1000 Dry02

Diffusion

Time=32

Temp=1000 Wet02

Diffusion

Time=10

Temp=1000 Dry02

Diffusion

Time=30

Temp=1000 T.Rate=-6.66

Comment

n-Well Drive-in

 

Diffusion

Time=35

Temp=800

T.Rate=5.7l4

Diffusion

Time=10

Temp=1000 Dry02

Diffusion

Time=30

Temp=1000 Wet02

Diffusion

Time-l0

Temp=1000 Dry02

Diffusion

Time-20

Temp=1000 T.Rate=7.5

Diffusion

Time=960

Temp-l1S0

Comment

Stop Simulation at

BOO C

Diffusion

Time=65.6 Temp-1150 T.Rate--5.333

Comment

Masking Oxide Etch after Collector Litho.

Etch

Oxide

 

 

Comment

Collector Implant.

 

Grid

Layer. 1 XdX=0.12 dX=0.005

Implant

Phosphorus Pearson Dose=3e13 Energy=lOO

Comment

Collector Drive-in

 

Diffusion

Time=35

Temp=800

T.Rate=5.7l4

Diffusion

Time=10

Temp=lOOO Dry02

Diffusion

Time=30

Temp=lOOO Wet02

Diffusion

Time=10

Temp=1000 Dry02

Diffusion

Time=20

Temp=1000 T.Rate=5.0

Diffusion

Time=200

Temp=ll00

Comment

Stop Simulation at 800 C

Diffusion

Time=55.7 Temp=llOO T.Rate=-5.385

Comment

Oxide Etch.

 

Etch

Oxide

 

 

360 APPENDIX C. TEMPLATES FOR PISCES SIMULATION

COlIIIDent

Groll Pad

Oxide

 

 

Grid

Layer.1 dX=O.OOS XdX=0.08

Diffusion

Time~30

Temp=800

T.Rate=S.O

Diffusion

Time=42

Temp=9S0

Dry02

Comment

Stop Simulation at 800 C

Diffusion

Time~22.S Temp-9S0 T.Rate=-6.66

COlIIIDent

Nitride

Deposition l

Densification

Deposit

Nitride Thickness=0.08 dX=O.OOS

Comment

Field Oxidation

 

 

Diffusion

Time=3S

Temp·800

T.Rate=S.714

Diffusion

Time=10

Temp-1000 Dry02

Diffusion

Time=190 Temp=1000 Wet02

Diffusion

Time=10

Temp=1000 Dry02

COlIIIDent

Stop Simulation at 800 C

Diffusion

Time=24

Temp=1000 T.Rate=-8.333

Comment

Strip Oxidized Nitride

Etch

Oxide

 

 

 

Etch

Nitride

 

 

 

Etch

Oxide

 

 

 

Comment

Groll Clean-up Oxide

 

Grid

Layer.1 Dx-0.001 Xdx=0.0125

Diffusion

Time=5

Temp=8S0

Dry02

Diffusion

Time=14.15 Temp=850

Wet02

Diffusion

Time-5

Temp-8S0

Dry02

COlIIIDent

Strip Clean-up Oxide

 

Etch

Oxide

 

 

 

Comment

Gate Oxidation for KOS

Grid

Layer. 1 Dx=0.001 Xdx=0.0125

Diffusion

Time=20

Temp=850

 

Diffusion

Time=5

Temp=8S0

Dry02

Diffusion

Time=14.15 Temp=8S0

Wet02

Diffusion

Time=5

Temp=850

Dry02

Comment

Gate Oxide Etching

 

 

Etch

Oxide

 

 

 

Comment

Active Base Implant

 

Implant

Boron Pearson Dose=4.5e13 Energy=35

COlIIIDent

Poly Deposition

 

 

C.l.

lD BJT

 

361

Deposit

Polysilicon Thickness=O.15 Temp=620 Pressure=O.0004

Comment

Nitride Deposition

 

Deposit

Nitride Thickness=O.03 dX=O.005

Comment

Oxide Deposition

 

Deposit

Oxide Thickness=O.5

Comment

Etch LTO Used for the Extrinsic Base Masking

Etch

 

Oxide

 

 

Comment

Polysilicon Oxidation

Grid

 

Layer.2 Dx=O.OOl Xdx=O.015

Diffusion

Time~20

Temp=800

T.Rate=5.0

Diffusion

Time-5

Temp=900

Dry02

Diffusion

Time=200 Temp=900

Wet02

Diffusion

Time-5

Temp=900

Dry02

Diffusion

Time=13.3 Temp=900

T.Rate=-7.5

Comment

Sidewall Oxidation

 

Diffusion

Time-5

Temp-850

Dry02

Diffusion

Time=6

Temp=850

Wet02

Diffusion

Time=5

Temp=850

Dry02

Comment

n-Channel SID and n+ Poly Implant

Implant

Arsenic Pearson Dose-le16 Energy-80

Comment

n-Channel and Emitter Drive-In

Diffusion

Time=25

Temp=800

T.rate=8.O

Diffusion

Time=10

Temp=1000 Dry02

Diffusion

Time-24

Temp=1000 T.rate--8.333

Comment

Poly Resistor Anneal/Oxidation

Diffusion

Time=25

Temp=800

T.rate=4.0

Diffusion

Time=40

Temp=900

Dry02

Diffusion

Time=tO Temp=900 T.rate=-10.0

Etch

 

Oxide

 

 

Etch

 

Nitride

 

 

Etch

 

Polysilicon

 

Print

Layer

 

 

 

Plot

 

Plotdev=xterm Net

Chemical Xmax=3

Savefile

Struct

File=npn.str

Savefile

Export

File=npn.exp

Stop

 

 

 

 

362 APPENDIX C. TEMPLATES FOR PISCES SIMULATION

The output net doping profile from SUPREM III is shown in

Fig. C.2(a), and the actual doping profile used by PISCES is shown in Fig. C.2 (b) in order to put the back contact directly to the collector region. Note that after stripping the polysilicon layer, the depth of the emitter-base junction is about 0.18 j.lm, and that of the base-collector junction is around 0.33 j.lm. In doing PISCES simulation the polysilicon layer can be modeled using the finite surface recombination velocity for the minority carriers (holes in npn transistors) at the emitter contact. In the following we will treat the heavily doped polysilicon layer as a metal contact. The input file for PISCES in order to simulate the Gummel plot and IT vs. VBE for a fixed base-collector bias and at the same time to perform zero-frequency analysis is included below. The so-called "zerofrequency" is actually an asymptotic extreme for low-frequency analysis (J -;. 0). Users can use standard features of ac analysis in PISCES by specifying a low value offrequency (such as freq=l Hz) to get the same, zero-frequency analysis result.

Title npn from Stanford BiCMDS process

option plotdev-xterm

mesh rect nx-2 ny=40l

x.m n=l

1=0

r=1

x.m n=2

1=1.0

r=l

y.m n=l

1=0

r=1.0

y.m n=40l

1=2.2 r=1.0

region num=l ix.l=l ix.h=2 iy.I=1 iy.h=40l silicon

elec num=1 ix.l=l ix.h=2 iy.l=l iy.h=l elec num=2 ix.l=l ix.h=l iy.l=37 iy.h=39 elec num=3 ix.l=1 ix.h=2 iy.l=401 iy.h=40l

dop

sup boron

infil=npn.exp x.l=O

x.r=l

dop

sup

arsenic

infil=npn.exp x.l=O

x.r=1

dop sup

phosphorus

infil=npn.exp x.l=O x.r=l

contact

num=2 surf.rec vsurfn=leO vsurfp=le7

symb newton carr=2

C.l. lD BJT

363

c: 18 a

~

."

[] 17

L

."

c:

~ 16

c: a

u 15

[))

a

...I

o

0.5

1.5

2

2.5

3

Dlstance fram eurface (mlcrons)

(a) Net chemical doping profile from SUPREMo

PISCES-IIS009!

(Tl

20

 

 

 

 

 

E

 

 

 

 

 

 

U

 

 

 

 

 

 

 

19

 

 

 

 

 

c

 

 

 

 

 

 

0

 

 

 

 

 

 

.,..,

18

 

 

 

 

 

0

 

 

 

 

 

 

.,c.....,

17

 

 

 

 

 

C

 

 

 

 

 

Q)

 

 

 

 

 

 

u

 

 

 

 

 

 

c

16

 

 

 

 

 

0

 

 

 

 

 

U

 

 

 

 

 

 

 

IS 1Il.1Il0

0.50

1.lIlril

1.50

2.1Il1ll

2.51

 

 

 

01 s tones

(urn)

 

 

(b) Net doping profile used in PISCES.

Figure C.2: SUPREM simulated net doping profile for Stanford's BiCMOS npn.

364 APPENDIX C. TEMPLATES FOR PISCES SIMULATION

 

-e--

Zero.freq Analysis

10'

--6- Freq·dep AnBtyai8

10'

10' '- ...'-~ o-J'--'-~"""""~~

.........

J.~~----"-~~----"-~~-'--'

0.4

0.5

0.6

0.7

0.8

VB. M

Figure C.3: Comparison between the computed (zero-frequency) and simulated (frequency-dependent ac analysis) iT VS. VEE.

model temp=300 srh auger conmob fldmob bgn method itlimit=15 trap p.tol=1.e-8 c.tol=1.e-8

plot.ld dop x.s=l x.e=1 y.s=O y.e=2.2 log abs

solve ini

log ivfil=npn.iv acfile=npn.ac

solve v2=O.4 v3=1.4 vstep=O.025 nstep=24 elect=23 proj lowf term=2

plot.1d x.a=v2 y.a=i3 log abs min=-12 plot.ld x.a=v2 y.a=i2 log abs unch

end

The computed iT VS. VEE based on the formula

iT =

gm

= _g_3_2_

(C.l)

 

211"(CEE + CEe)

21l"C22

 

where g32 and C22 are obtained from the zero or low frequency analysis (lowf in solve card), are compared to the result from the real, frequency dependent ac analysis. It can be seen that before the high-level injection (HLI) occurs, the agreement is perfect, and after the transistor enters into the HLI operation regime, the computed c's can no longer

C.2. MOS CAPACITORS

365

be considered as the accurate representation of the capacitance as explained in [C.1], and the results from analytical formulation are no more correct. For this particular structure (npn from the triple diffusion), the dip at VBE = 0.9 V is abnormal and is most likely caused by the low doping level at the collector contact.

C.2 M 0 S Capacitors

The special features to pay attention to with 1D MOS template are (1) the way to put the source region in the simulation region and (2) how to evaluate the integral charge after each solution is found and to assemble the results of charge vs. bias in one data file. The following example shows the process. The purpose of the simulation is to compute the channel charge vs. VGS with VBS as the parameter.

For n-channel capacitor at VBS = -1 V, VGS is ramped from 0 to 1.5 volts with voltage step smaller around where the threshold voltage is expected.

title Simulation of nMOS Capacitor

options plotdev=xterm

mesh rect nx=4 ny=Sl

x.m n=l

1=0

r=l

 

x.m n=2

1=0.1 r=l

 

x.m n=3

l=S.O

r=l

 

x.m n=4

1=6.0 r=l

 

y.m n=l

1=0

 

r=1.0

y.m n=2

1=0.0418

r=1.0

y.m n=3

1=0.05

r=1.0

y.m n=Sl

1=4.0

 

r=1.0S

region num=l ix.l=l ix.h=4 iy.l=l iy.h=2 oxide region num=2 ix.l=l ix.h=4 iy.l=2 iy.h=Sl silicon

$ Electrodes: 1- Gate, 2- Substrate, 3- Source elec num=l ix.l=l ix.h=4 iy.l=l iy.h=l

elec num=2 iX.l-l ix.h=4 iy.l=Sl iy.h=Sl elec num=3 ix.l=l ix.h=l iy.l=2 iy.h=3

dop suprem3 infil=nchcap.exp boron $ n-type source region

366 APPENDIX C. TEMPLATES FOR PISCES SIMULATION

dop uniform x.left=O x.right=O.l y.top=0.0418 y.bot-O.OS n.typ concz l.e20

$ Specify the gate material to set the work function right contact num=l n.poly

$ For initial solution, solve Poisson's equation only symb newton carr-O

model temp=300 srh

method itlimit=30 p.tol=1.e-8 c.tol=1.e-8 trap

plot.2d grid no.fil bound pause

plot.ld dop x.s=S. x.e=S. y.s=O. y.e=4. log abs pause

solve ini

$ Switch to solving the complete Semiconductor equations symb newton carr=2

$ Using projection scheme for the guess to the next solution

solve v2=-1.0 proj

$ Integral charge of the channel carriers in the substrate

extract ele x.min=S. x.max=6. y.min=0.0418 y.max=4 outfil=nchcap.qn

solve vl=0.2 proj

extract ele x.min=S. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn solve vl=O.4 proj

extract ele x.min=S. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn

solve vl=O.S proj

extract ele x.min=5. x.max=6. y.min=0.0418 y.max=4 outfil=nchcap.qn solve vl=O.S5 proj

extract ele x.min=5. x.max=6. y.min=0.0418 y.max=4 outfil=nchcap.qn solve vl-0.6 proj

extract ele x.min=5. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn solve vl=O.65 proj

extract ele x.min=5. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn solve vl=O.7 proj

extract ele x.min=S. x.max=6. y.min=0.0418 y.max=4 outfil=nchcap.qn solve vl=O.7S proj

extract ele x.min=S. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn solve v1=O.8 proj

extract ele x.min=S. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn solve v1=O.8S proj

extract ele x.min=S. x.max=6. y.min=0.0418 y.max=4 outfil=nchcap.qn

solve vl=O.9 proj

extract ele x.min=5. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn solve vl=1.0 proj

C.2. MOS CAPACITORS

367

extract ele x.min=S. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn solve vl=1.1 proj

extract ele x.min=S. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn solve vl=1.2 proj

extract ele x.min=S. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn solve vl=1.3 proj

extract ele x.min=S. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn solve vl=1.4 proj

extract ele x.min-S. x.max=6. y.mina O.0418 y.max=4 outfil=nchcap.qn solve vl=l.S proj

extract ele x.min=S. x.max=6. y.min=O.0418 y.max=4 outfil=nchcap.qn end

Figure C.4: Template for simulation of n-channel MOS capacitor.

The output of the mesh is shown in Fig. C.5 in which there are four grids in the lateral (x -) direction. The first two grids from the left are only 0.1 JLm apart, and this spacing is served to form a source region. Only one grid spacing (y = 0.0418 '" 0.05 JLm) is assigned to the source region for contacting to the channel after it is formed. The second and third grids should be separated far enough (in this example, it is almost 5 JLm) to buffer the depletion region between the source and substrate from affecting the right-most part of the simulation region. The last spacing between grids 3 and 4 is set to be 1 JLm to ease the computation of the channel charge per square surface area.

The doping profile is shown in Fig. C.6 (a). The curve of Qn vs. Vas can be obtained by manipulating the data file nChcap. qn, and is plotted in Fig. C.6 (b).

368 APPENDIX C. TEMPLATES FOR PISCES SIMULATION

Figure C.5: Mesh for the MaS template.

j::~

 

5.C

10~

 

2,Q10'"

"L~",Dis tones ",.(urn)

<4.010"

,

"oe(V}

 

 

1.0 H)'"

 

 

 

'~,~~~~~~,~~--~~~

1Il.lIla 0.S!II

1.1i"11 1.502.00 e.sli!I 3.'"11 3.50 ~.IIH!1 "t.S

 

(a) Net

doping profile for

n-channel

(b) Qn vs. Vas for VSB = 1 V.

MOS.

 

 

 

 

 

Figure C.6: Plots from running the template for n-channel MOS.