Noise and Reliability in Advanced Bipolar Technologies

UTA Noise and Reliability Laboratory 1

Md Mazhar Ul Hoque

Advisor: Prof. Zeynep Celik-Butler

Department of Electrical EngineeringUniversity of Texas at Arlington

Noise and Reliability in Advanced Bipolar Technologies


Outline Introduction:

Limitations in existing noise models. Importance and motivation of research.

Noise in polysilicon emitter bipolar transistors: Experimental setup. SIB modeling:

noise mechanisms, effect of bias, geometry and IFO. SIC modeling:

noise mechanisms, effect of bias and IFO. SVr modeling:

Collector-emitter measurement setup, effect of bias and IFO. Computer codes.

Noise in SiGe heterojunction bipolar transistors: Dominant noise source. Effect of selective collector implant. Effect of higher extrinsic base implant. Effect of SiGe epi-SiGe poly interface. Effect of emitter-poly overlap. Physical origin and modeling of SIB.

Conclusion


Current BJT low-frequency noise models

Gummel-Poon:

VBIC – Flicker noise due to IBE, IBEP

MEXTRAM:

MODELLA:

ff

IKFI EF

AFB

BN

2

fMULTI

KFMULTI

KFNf

MULTiN BB

B

21

222

ff

IIMULTKFiN

AFLERE

AF

B

1

2


Importance and motivation

Only one single noise source in base current: SIB=KF.IBAF.

Noise in base current dominant only for higher base series resistance.

Does not include any geometry, temperature or process parameter.

Noise in collector current and internal resistances are neglected. Noise in collector current contributes at lower base series

resistance. Noise from internal resistance contributes at higher bias current.

To model all possible noise sources in advanced bipolar transistors: Noise in base current, collector current and internal resistances.

Developing physics based scalable equations for the noise sources: Incorporating geometry, temperature and process dependant

parameters. Writing computer source code for device and circuit analysis CAD

tools to incorporate all possible noise sources in BJT.

Research goal:

Limitations in existing noise models:


Polysilicon emitter bipolar transistors: 2nd generation BiCMOS technology, Texas

Instruments Inc. NPN and PNP transistors. Variable size, variable IFO thickness.

SiGe heterojunction bipolar transistors: 1st generation BiCMOS technology, National

Semiconductor Corporation. NPN transistors. Variable size, variable design rules. Variable doping in base and collector.

Advanced bipolar technologies under investigation









Conclusion


100K 100K4.8V RS

RL 12V

br

brVS

BISr

bi bi

CISor crVS

cr

LRVS

LR

2BVS

2BR

erVS

er

SRVS

SR

1BVS

1BR

CVSBVS

(a)

(b)

Collector-base measurement setup


CBrB IIVV SwSvSuS

CBrC IIVV SzSySxS

rVS If has the dominant contribution, ux

S

S

B

C

V

V

vy

S

S

B

C

V

V BIS If has the dominant contribution,

wz

S

S

B

C

V

V CIS If has the dominant contribution,

Collector-base measurement

for unity coherence:

CB

BC

VV

V

SS

S

22

brerr VVV SSS


Dominant noise source 0.7x100 μm2 NPN, thickest IFO,

RS=1MΩ0.7x100 μm2 PNP, thickest IFO, RS=1MΩ

100

101

102

103

104

105

106

0

1

10-8 10-7 10-6

measured at 1 HzS

IB contribution dominant

SIC

contribution dominantS

Vr contribution dominant

S VC/S

VB

Coh

eren

ce (a

t 1 H

z)

IB (A)

0.5

10-1

100

101

102

103

104

105

106

0

1

10-7 10-6 10-5

measured at 1 HzS


SIC



S VC/S

VB

Coh

eren

ce (a

t 1 H

z)

IB (A)

0.5

Calculated SVC/SVB considering SIB contribution dominant closely matches experimental SVC/SVB; SIB contribution dominant.









Conclusion


Area dependence of SIB in NPN transistors

10-23

10-22

10-21

10-20

1 10

IB= 0.1mA

IB= 0.2mA

IB= 0.3mA

IB= 0.4mA

IB= 0.5mA

IB= 0.6mA

S IB (A

2 /Hz)

Emitter Area (m2)

~ 1/A

Noise source homogeneously

distributed underneath the emitter.

[ Darby Lan ]


Internal emitter resistance and ideality factor of base current

1

1.1

1.2

1.3

1.4

1.5

1.6

1.0E-03 5.0E-03 9.0E-03 1.3E-02 1.7E-02

gmi (ohm-1)

m

re=27.8 ohmnb=1.07

~ m = 27.8 gmi + 1.07

0.7x100 μm2 NPN, thickest IFO, RS=1MΩ

C

B

V

V

i

LSS

rRm

imi r

g

Bi qI

kTr

bmie ngrm

m

gmi (ohm-1)


Base current dependence diffusion fluctuation in emitter.

tunneling fluctuation in IFO.

recombination fluctuation in emitter- base space charge region.

recombination fluctuation at spacer oxide interface.

Emitter perimeter dependence Noise source distributed homogeneously around the emitter. Recombination fluctuation at the spacer oxide interface.

Emitter area dependence Noise source distributed homogeneously underneath the

emitter. Diffusion fluctuation in emitter or tunneling fluctuation in IFO.

diffdiffB BI IS

2tuntunB BI IS

2recrecB BI IS

Physical origin of SIB


recBtunBdiffBB IIII SSSS

tunBdiffBB III SSS

Unity ideality factor of negligible .

negligible .

BI

recBIS

recBI

Physical origin of SIB (cont.)


Diffusion and tunneling fluctuation component of SIB

1.00E-17

2.10E-16

4.10E-16

6.10E-16

8.10E-16

1.E-08 2.E-07 3.E-07 5.E-07 6.E-07 8.E-07

IB(A)

S IB/I B

(A/H

z)

~ SIB/IB = 1.13x10-9 IB + 7.85x10-17

SIBT = 1.13x10-9

SIBD = 7.85x10-17

0.7x100 μm2 NPN, medium IFO, RS=1MΩ

TBDBB III SSS

2BTBD IKFIKF

BTDB

I IKFKFI

SB


Area dependence of tunneling fluctuation of SIB in PNP transistors

Tunneling-fluctuation source homogeneously distributed underneath

the emitter.

Thickest oxide

Smaller device-dimension severely affected by

lateral diffusion and mask undercut; effective

emitter area considered.10-10

10-9

10-8

10-7

10-1 100 101 102

KF T

Emitter area: AE (m2)

~1/AE

(0.847)

Thickest IFO


Wm WpL

0 x1 x2 x3

base mono-silicon

oxide Poly-silicon

metal contact

c

arri

er c

once

ntra

tion

p(

x) sox

sm

p(x)

p(x)

αm, Dm αp, Dp

p

pmDWs

xpxp

132

1

1

1110

p

p

moxm

mDW

ssDW

xpp

232

1

11

)()(ln

)()0(ln

moxp

p

m

m

p

p

m

m

B

I

ssDW

DW

xpxp

fDq

xpp

fDq

I

SDB

Diffusion fluctuation in SIB

αm : Hooge parameter in mono-Si

αp : Hooge parameter in poly-Si

Assumption: αm = αp =α


Tunneling fluctuation in SIB

tan

ox

oV

oxide permittivity barrier height for the minority carriers modified Planck’s constantarea of IFOloss tangent of the oxide

A

2

2

211

mn

t

p

p

m

m

mox

B

I

t

SDW

DW

ss

I

SmnTB

AfVqkTLm

t

S

oxomn

tmn

2

3*

2 3tan

mnt

mntS

*m

q

k

L

tunneling probability of minority carriers fluctuation in effective mass of minority carrierselectronic chargeBoltzmann constant IFO thickness

mnt


Scaling effect on SIB

Relative AF

IFO thickness0.7x100( μm2 )

0.7x0.7( μm2 )

thickest 1.12 1.89medium 1.23 2.09thinnest 1.01

Diffusion fluctuation dominant in large (0.7x100μm2) devices.

Tunneling fluctuation dominant in small (0.7x0.7μm2) devices.

AFBI IKFS

B

AF for SIB in PNP transistors

tunBdiffBB III SSS

2BTBD IKFIKF


Scaling effect on SIB in PNP transistors with thickest IFO

Total SIB

Tunneling fluctuation component of SIB

Diffusion fluctuation component of SIB

SIB

SIBT

SIBD

10-24

10-23

10-22

10-21

10-20

10-19

S I(A2 /H

z)

A = 0.7x100m2

AF=1.12

Diffusion fluctuation dominant

10-24

10-23

10-22

10-21

10-20

10-19

10-18

A = 0.7x10m2

AF=1.79

10-24

10-23

10-22

10-21

10-20

10-19

10-8 10-7 10-6 10-5

S I(A2 /H

z)

A = 0.7x0.7m2

AF=1.89

Tunneling fluctuation dominant

10-24

10-23

10-22

10-21

10-20

10-19

10-8 10-7 10-6 10-5

IB(A)

A = 0.7x2.8m2

AF=1.66


Poly-Si emitter

oxide dl

ds

mono-Si emitter

dl < ds

Large device small device

base

Thicker polysilicon in smaller transistors. Lower dopant concentration; shallower junction.

higher chance for minority carriers from base to reach and tunnel through the oxide interface; tunneling fluctuation dominant.

FLUORINE EFFECT: fluorine enhances oxide break-up. lower fluorine (from BF2) concentration causes less oxide breakup in smaller PNP devices; increased

tunneling: (no fluorine in the transistors studied here)

Emitter plug effect in smaller device


Poly-Si emitter

oxide d1

d1 < d2

base

d2

mono-Si emitter

Polysilicon surface almost perpendicular to wafer surface at emitter window sidewall.

Reduced doping concentration in the perimeter region.

Shallower junction close to emitter window perimeter. An overlap of the emitter-base space charge region

with poly-mono silicon interface might occur close to perimeter.

Perimeter depletion effect


Perimeter/area vs. emitter area

emitter area (μm2)

Peri

met

er/a

rea

For smaller transistors perimeter/area ratio increases sharply. non ideal peripheral component of base current

increases. fluctuation in non-ideal base current might become

significant.

Perimeter depletion effect in smaller transistors

1.00

2.00

3.00

4.00

5.00

0 20 40 60 80 100 120


Effect of IFO on DC characteristics of 0.7x100μm2 NPN transistors

0

100

200

300

400

500

600thickest IFOmedium IFOthinnest IFO

0.3 0.4 0.5 0.6 0.6 0.7 0.8 0.9 110-14

10-12

10-10

10-8

10-6

10-4

10-2

100

102

IC (thickest IFO)

IB (thickest IFO)

IC (medium IFO)

IB (medium IFO)

IC (thinnest IFO)

IB (thinnest IFO)

0.3 0.4 0.5 0.6 0.6 0.7 0.8 0.9 1

IC

IB

Cur

rent

(A)

Cur

rent

gai

n

VBE

(V)

IFO increases the current gain significantly.


Effect of IFO on DC characteristics of 0.7x100μm2

PNP transistors

10-12

10-10

10-8

10-6

10-4

10-2

100

102

0.3 0.4 0.5 0.6 0.6 0.7 0.8 0.9 1

IC (thickest IFO)

IB (thickest IFO)

IC (medium IFO)

IB (medium IFO)

IC (thinnest IFO)

IB (thinnest IFO)

IC

IB

0

40

80

120

160

0.3 0.4 0.5 0.6 0.6 0.7 0.8 0.9 1

thickest IFOmedium IFOthinnest IFO

VBE

(V)

Cur

rent

(A)

Cur

rent

gai

n

No significant improvement in current gain with increasing IFO thickness.


Effect of interfacial oxide on SIB in NPN and PNP transistors

10-24

10-23

10-22

10-21

10-20

10-19

10-8 10-7 10-6 10-5


S IB (A

2 /Hz)

IB (A)

10-24

10-23

10-22

10-21

10-20

10-19

10-8 10-7 10-6 10-5


S IB (A

2 /Hz)

IB (A)

0.7x100 μm2 NPN 0.7x100 μm2 PNP

IFO increases SIB both in NPN and PNP transistors.


In PNP, diffusion fluctuation in mono and poly-silicon emitter dominates; less effect of IFO.

In NPN, both diffusion fluctuation in mono and poly-silicon emitter, and tunneling fluctuation through IFO contribute.

tundiffB IBIBI SSS 2IBKFIBKF TD

AFBI IKFS

B

Difference in the effect of IFO in NPN and PNP transistors

0.7x100μm2 transistors

Device type

AF

PNP ~ 1NPN ~ 1.5


Physics behind difference in NPN and PNP transistors

Hole barrier height larger than electron barrier height at interfacial oxide: minority carriers (holes) severely suppressed in NPN; significant

current gain improvement. minority carriers (electrons) not suppressed significantly in

PNP; little current gain improvement. Effect of fluorine:

fluorine accelerates oxide break-up. fluorine from emitter dopant BF2 in PNP creates more oxide

breakup; reduced tunneling and increased diffusion through broken oxide.

Increased oxide breakup and faster diffusion of boron makes the emitter deeper in PNP: increased recombination-base current, reduced current gain

improvement. higher diffusion fluctuation in larger monosilicon emitter region.

Different oxidation rate of the base material could create different IFO thickness for NPN and PNP transistors on the same wafer.









Conclusion


10-21

10-20

10-19

10-18

10-17

10-16

10-15

10-5 10-4 10-3

thickest IFO: AFC=1.72medium IFO: AFC=1.71

S IC (A

2 /Hz)

IC (A)

10-21

10-20

10-19

10-18

10-17

10-16

10-15

10-6 10-5 10-4 10-3

thickest IFO: AFC=1.54medium IFO: AFC=1.74thinnest IFO: AFC=2.01

S IC (A

2 /Hz)

IC (A)

0.7x100 μm2 NPN 0.7x100 μm2 PNP

Effect of interfacial oxide on SIC

IFO increases SIC both in NPN and PNP transistors.


Some researchers assign the same origin of SIB to SIC.

is merely an amplified SIB.

SIC modeling

DCNCC III SSS

CDCN IKFIKF 22L

VI

R

SS CC

Number fluctuation in SIC requires further investigation.

carrier transport determined by electric field in base collector

junction.

Hooge type fluctuation at the collector side of base-

collector junction.

)12( KCI IS

C

KCIh 22

1K

1.E-07

1.E-06

1.E-05

1.E-05 1.E-04 1.E-03IC (A)

outp

ut c

ondu

ctan

ce h 22

(ohm

-1)

~ IC(0.98)

K=0.98

0.7x100μm2 PNP, thickest IFO


n(x)

α, D

Diffusion fluctuation in SIC

)()0(ln2 BBC

I

Wnn

fWqD

I

SDC B

Bs

B WDWD

Wnn

//

)()0(

s saturated drift velocity

WB

collector

0

base mono-silicon

oxide Poly-silicon

metal contact

c

arri

er c

once

ntra

tion

n(

x)









Conclusion


Effect of bias on coherence in NPN and PNP transistors

0

0.2

0.4

0.6

0.8

1

1.2

100 101 102 103 104 105C

oher

ence

Frequency (Hz)

a: IB= 301 nA

b: IB=493 nA

c: IB=1.09

d: IB=1.48

e: IB=1.99

0.7x100 μm2 NPN, thickest IFO, RS=1MΩ 0.7x100 μm2 PNP, thickest IFO, RS=1MΩ

0

0.2

0.4

0.6

0.8

1

100 101 102 103 104 105

Coh

eren

ce

Frequency (Hz)

a: IB= 72 nA

b: IB=264 nA

c: IB=384 nA

d: IB=546 nA

e: IB=760 nA

'a' to 'e'(increasing bias current)

Coherence decreases with increasing bias in NPN transistor.


10-17

10-16

10-15

10-14

10-13

10-12

100 101 102 103 104 105

RS=1MRS=100KRS=50KRS=10KBackground noise

S VB (V

2 /Hz)

Frequency (Hz)

decreasing RS

10-15

10-14

10-13

10-12

10-11

10-10

10-9

100 101 102 103 104 105

RS=1MRS=100KRS=50KRS=10KRS=1KRS=120

S VC (V

2 /Hz)

Frequency (Hz)

decreasing RS

Effect of varying base bias resistance (RS) on SV

0.7x100 μm2 NPN, thickest IFO, IB=384nA

SVC does not decrease any more for RS smaller than 1 kΩ. SVB becomes comparable to system background noise for

RS smaller than 10 kΩ.


4.8V 12V

1K100K

RS

RL

SVE SVC

RE

Collector-emitter measurement setup


0.7x100 μm2 NPN, thickest IFO, RS=100Ω

Dominant noise source in collector-emitter measurement

Unity coherence.One single

dominant noise source:

SIC or SVr?

101

102

103

104

105

10-7 10-6 10-5

measured

SIC

contribution dominant

SIB


SVr


S VC/S

VE

cohe

renc

e at

1 H

z

IB (A)

0.5

1

0


10-18

10-17

10-16

10-15

10-14

10-4 10-3 10-2

thickest IFO : AFC=3.21

medium IFO : AFC=3.58

S IC (A

2 /Hz)

IC (A)

SIC in collector-emitter measurement

0.7x100 μm2 NPN, RS=100Ω

no physical explanation for

such high current dependence:SPURIOUS?

AFCCI IKFCS

C

58.3~05.3AFC


10-15

10-14

10-13

10-12

10-11

10-4 10-3 10-2

thickest IFO

medium IFO

S Vr (V

2 /Hz)

IE (A)

~ IE

2

err VV SS

erE SI 2

brerr VVV SSS

0.7x100 μm2 NPN, RS=100Ω

Effect of interfacial oxide on SVr


Effect of internal resistance in collector-emitter measurement

0.7x100 μm2 NPN, thickest IFO

Collector-base measurement, RS=1MΩ

Collector-emitter measurement, RS=100Ω

1.E-12

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

0.4 0.5 0.6 0.7 0.8 0.9 1VBE (V)

Bas

e cu

rren

t (A

)

noise measurement

region

exponential fitted line( η=1.12 )

1.E-12

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

0.4 0.5 0.6 0.7 0.8 0.9 1VBE (V)

Bas

e cu

rren

t (A

)

noise measurement

region

exponential fitted line( η=1.1 )


Tunneling fluctuation through interfacial oxide

AfVqkTLm

t

S

r

S

oxomj

t

e

r mje

2

3*

22 3tan

mjt tunneling probability of the majority carriers through IFO

mjtS fluctuation in mjt

base mono-silicon

oxide Poly-silicon

metal contact

fluctuation in minority carrier tunneling

fluctuation in majority carrier tunneling erS

TBIS


Interfacial oxide thickness

Relative Emitter area interfacial oxide thickness (Å) fromIFO thickness (μm2) minority carrier majority carrier

fluctuation (SIBT) fluctuation (SVre)

PNP NPN NPN 0.7x100 14.50 4.08

thickest 0.7x10 16.90

0.7x2.8 14.50

0.7x0.7 11.00

medium 0.7x100 4.13 11.30 2.26

0.7x0.7 8.78

thinnest 0.7x100 2.70

0.7x0.7 2.81

VVh 363.0 VVe 097.0

oh mm 81.0* oe mm 08.1*

tan

Inconsistency:uncertainty in loss tangent, mass and potential barrier of the carriers at oxide interface.

obtained

from Kleinpenning et. al, IEEE Trans. on Elec. Dev., vol. 42, 1995.

as high as 1.8V and as high 1 V found in literature.

unique for each sample.tanhV eV









Conclusion


Experimental and Simulated data

Simulated Noise vs Lab

0.1

1.0

10.0

100.0

1000.0

1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07

Freq (Hz)

Nois

e (V

/sqr

t(Hz)

AU)

Production Models THS4271 Lab Spot Models

Noise data measured at TI

Simulation results vs. Measured

Experimental data

Simulated with existing model

Simulated with modified model


Computer source code

Existing BERKELEY SPICE source code has been modified. Source code written at Texas Instruments Inc. by

Douglas Weiser. and have been added to the

existing noise model in addition to . An area scaling factor has been added to the

equations to make the noise models scalable.

The modified BERKELEY SPICE source code is available to the SRC (Semiconductor Research Corporation) member companies (TI, Intel, IBM, Motorola, National Semiconductor Corporation, AMD, etc.)

CIS rVS

BIS









Conclusion


xyz

shallow trench shallow trench

n+ poly Si emitter SiGe epi-SiGe poly interface dielectric isolation

n+ Si emitter

p+ SiGe poly p+ SiGe poly

DTI(deep trench

isolation)

p+ SiGe epi

p+ SiGe epi

DTI(deep Trench

isolation)

p SiGe epi

DTI(deep trench

isolation)

Device structure under investigation

x = emitter-poly overlapy = composite enclosure of polyz = DTI enclosure of composite

Selectively implanted

collector

Transistors described as x-y-z; e.g. SIC:25-10-25.


100

101

102

103

104

105

10-6 10-5 10-4

experimentalS


SIC



S VC/S

VB

IB (A)

SIC:20-20-40

103

104

105

106

0

0.2

0.4

0.6

0.8

1

10-6 10-5 10-4

base series resistanceinput resistancecoherence at 10 Hz

Res

ista

nce

(

Coh

eren

ce a

t 10

Hz

IB (A)

SIC:20-20-40

Dominant noise source

Calculated SVC/SVB with SIB contribution dominant closely matches experimental SVC/SVB ; SIB contribution dominant.

IB increases rπ decreases coherence increases.









Conclusion


Selectively implanted collector (SIC)

Higher collector doping

Higher collector-base capacitance

Lower maxf

Retarded Kirk-effect Higher Tf

Selectively implanted collector

Higher and reduced degradation of

Reduced base-width by compensation of boron tail

Tf

maxf


Effect of selectively implanted collector

10-14

10-12

10-10

10-8

10-6

10-4

10-2

0

50

100

150

200

250

300

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

SIC:25-10-2525-10-25

noise measurementregion

I B, I

C (A

)

curr

ent g

ain:

VBE

(V)

IC

IB

10-21

10-20

10-19

10-18

10-7 10-6 10-5

SIC:25-10-2525-10-25

S IB.f

(A2 ) a

t 10

Hz

IB (A)

~ IB

2

SIC retards Kirk-effect higher β for wider range of bias.

No degradation in noise.









Conclusion


Higher extrinsic base implant (HEBI)

High dose implantation in extrinsic base

Smaller base resistance

Higher maxf

Lower RF noise figure


Effect of higher extrinsic base implant

10-14

10-12

10-10

10-8

10-6

10-4

10-2

0

100

200

300

400

500

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

SIC:25-10-25HEBI,SIC:25-10-25

noise measurementregion

I B, I

C (A

)

curr

ent g

ain:

V

BE (V)

IC I

B

10-21

10-20

10-19

10-18

10-7 10-6 10-5

SIC:25-10-25HEBI,SIC:25-10-25

S IB.f

(A2 ) a

t 10

Hz

IB (A)

~ IB

2

Higher gain for higher extrinsic base implant is possibly due to relative changes in Ge profile.

No degradation in noise.









Conclusion


SiGe epi-SiGe poly interface

Non-selective epitaxial growth of SiGe base

SiGe epi on top of exposed Si

SiGe poly on top of shallow trench

Interface of epitaxial and polycrystalline SiGe

smaller distance between SiGe epi-SiGe poly interface & intrinsic base

Smaller base-collector capacitance

Higher maxf

TfHigher


Effect of SiGe epi-SiGe poly interface

10-14

10-12

10-10

10-8

10-6

10-4

10-2

0

50

100

150

200

250

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

SIC:20-10-25SIC:20-20-40

noise measurement region

I B, I

C (A

)

curr

ent g

ain:

V

BE (V)

IC

IB

10-22

10-21

10-20

10-19

10-18

10-17

10-7 10-6 10-5 10-4

SIC:20-20-40SIC:20-10-25

S IB.f

(A2 ) a

t 10

Hz

IB (A)

~ IB

2

No significant impact on DC characteristics. No degradation in noise; boron implantation in

extrinsic base passivates the interface-defects.









Conclusion


Effect of emitter-poly overlap

10-15

10-14

10-13

10-12

10-11

10-10

IB = 0.5 A

IB = 1 A

IB = 4 A

100 101 102 103

~ 1/f

S VC (V

2 /Hz)

SIC:20-10-25

Frequency (Hz)

10-15

10-14

10-13

10-12

10-11

10-10

100 101 102 103

IB = 0.5uA

IB = 1uA

IB = 4uA

IB = 10uA

S VC (V

2 /Hz)

Frequency (Hz)

SIC:10-10-25~ 1/f

Significant g-r noise for smaller emitter-poly overlap. g-r noise diminished at higher currents in comparison

to increased 1/f noise.


Effect of emitter-poly overlap (cont.)

10-22

10-21

10-20

10-19

10-18

10-17

10-7 10-6 10-5 10-4

SIC:25-10-25SIC:20-10-25SIC:10-10-25

S IB .f

at 1

0 H

z (A

2 )IB(A)

~ IB

2

10-14

10-12

10-10

10-8

10-6

10-4

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

SIC:25-10-25SIC:20-10-25SIC:10-10-25

I B (A

)

VBE

(V)

r = 3.08

Higher non-ideal base current for smaller emitter-poly overlap.

SIB deviates from ~IB2 dependence for smaller emitter-

poly overlap.









Conclusion


10-22

10-21

10-20

10-19

10-18

10-17

10-7 10-6 10-5 10-4

20-20-40SIC:20-20-4010-20-40SIC:10-20-4025-10-25

SIC:25-10-25SIC:20-10-25SIC:10-10-25HEBI:25-10-25HEBI,SIC:25-10-25

S IB .f

at 1

0 H

z (A

2 )

IB(A)

~ IB

2

1

1.2

1.4

1.6

1.8

2

1 1.5 2 2.5 3 3.5 4

AF

r

smaller emitter-poly overlap (x = 0.1 m)

Physical origin of SIB

All except smaller emitter-poly overlap ηr≤1.6, AF≈2.

Smaller emitter-poly overlap ηr>3, AF≈1.


SIB model

21 BI IKFS B

22

1)2(1

1

i

iN

iB

fIKF

1N i

2AF

Total number of traps in emitter- base junction.

Characteristics time constant of ith trap.

SRH g-r centers in E-B depletion region

Capture and emission of carriers

Perturbation in free carrier density

Continuous distribution of time constants

1/f noiseSuperpositionof Lorentzians

: :


Noise from Surface recombination current

Smaller emitter-poly overlap

Extrinsic base moves closer to emitter perimeter

Two parallel BJTs

n+ emitter, p SiGe intrinsic base, collector

n+ emitter, p+ SiGe extrinsic base, collector

Thin depletion region & high electric field

Recombination through trap assisted tunneling

n+ emitter - p+ extrinsic base diode

n+ poly Si emitter

p SiGe epi(intrinsic base)

p+ SiGe epi(extrinsic base)

n+ Si emitter

dielectric isolation


SIB model for smaller emitter-poly overlap

1AF

fKTAC

NqKF

sc22

42

2N

A scC

Oxide slow state volume density.

Attenuation tunneling distance.

Area of surface region.

Surface capacitance per unit area.

: :

: :

AFBI IKFS B

From intrinsic E-B junction

From surface of E-B junction

22

21 svB BBI IKFIKFS


1.E-15

1.E-14

1.E-13

1.E-12

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

0 0.2 0.4 0.6 0.8 1

ηi = 1.09

ηr = 1.47 IBs

IBv = IB - IBs

I B(A

)

VBE(V)

22

2

12KF

I

IKF

I

S

s

v

s

B

B

B

B

I

2N = 9.65x1015 ~ 3.08x1016 /eV/cm3.

Published values = 1017 ~ 1018 /ev/cm3.

Uncertainties in and . scC)exp(KT

qVII

r

BEBB sos

22

21 svB BBI IKFIKFS

Separation of SIB components









Conclusion


Conclusion

The base current fluctuations dominate at relatively high external base resistance values.

Collector current fluctuations contribute when the external base resistance becomes comparable to or less than the input resistance.

Fluctuations caused by the internal emitter resistance take over for high currents with small base resistance.

From the dependence of noise magnitude on the geometry and temperature, fluctuations in the non-ideal base current was neglected.

Polysilicon emitter bipolar transistors:


Conclusion (cont.)

SIB originates from diffusion fluctuations of minority carriers in poly and monosilicon emitter, and tunneling fluctuations of minority carriers at IFO; SIC from diffusion and number fluctuations of minority carriers in base, and SVre from tunneling fluctuations of majority carrier at IFO.

Diffusion fluctuations dominate over the tunneling fluctuations for larger devices; tunneling fluctuations dominate over diffusion fluctuations for the smaller ones.

IFO increases the current gain significantly in NPN; both tunneling and diffusion fluctuations contribute.

No significant improvement with increasing IFO thickness in PNP; diffusion fluctuations dominant.


Conclusion (cont.)

Selectively implanted collector causes higher current gain for wider range of bias; no noise degradation.

No significant effect of SiGe epi-SiGe poly interface and higher extrinsic base implant.

Severe impact of smaller emitter-poly overlap: increases non-ideality of the base current;

trap assisted tunneling current due to parasitic BJT.

g-r noise at lower bias currents. At higher bias currents, fluctuations from trap

assisted tunneling current contribute. For all transistors except smaller emitter poly

overlap, noise originates from intrinsic E-B junction.

SiGe heterojunction bipolar transistors:


Acknowledgements

Zeynep Celik-ButlerSupervising Professor:

UTA Noise & Microsensor Group:

Semiconductor Research Corporation:

Texas Higher Education Board, Advanced Technology Prog.

Texas Instruments Inc. National Semiconductor Corporation

Bigang Min, Siva P. Devireddy, Shadi Dayeh, Enhai Zhao

Noise and Reliability in Advanced Bipolar Technologies

Documents

Transcript of Noise and Reliability in Advanced Bipolar Technologies