F-91405 ORSAY cedex FRANCE

J. SAINT MARTIN, V. AUBRY-FORTUNA, A. BOURNEL, P. DOLLFUS,

S. GALDIN, C. CHASSAT

stmartin@ief.u-psud.fr

Influence Of Ballistic Effects Influence Of Ballistic Effects In Ultra-Small MOSFETsIn Ultra-Small MOSFETs

INSTITUT D'ÉLECTRONIQUE FONDAMENTALE

IEF, UMR CNRS 8622, Université Paris Sud,

Centre scientifique d'Orsay - Bât. 220

ConteContentsnts

Introduction

Channel length and ballistic transport

Bias dependence on ballisticity

Ballisticity and long-channel effective mobility

ConteContentsnts

Introduction

Channel length and ballistic transport

Bias dependence on ballisticity

Ballisticity and long-channel effective mobility

Transport in decanano Transport in decanano MOSFETMOSFET

• Bint = % of ballistic electrons at the drain-end

• Beff =

Entrance Exit

y0

Scatteringcountingregion

Source Drain

Front gate

Back gate

on

on_with_a_ballistic_channel

I

I

Monte Carlosimulation

Quasi-ballistic transport: electron mean free path is similar to Lch

ConteContentsnts

Introduction

Channel length and ballistic transport

Bias dependence on ballisticity

Ballisticity and long-channel effective mobility

Studied Studied devicesdevices

m = 4.46 eVVDD = 0.7 V

Tox = 1.2 nm

TSi = 5 nm

Lch

LG

Tox = 1.2 nm

Tspacer = 5 nm Ts pacer = 5 nm

SiO2

SiO2

Undoped

5×1019 cm-3 5×1019 cm-3

Toverlap = 5 nm Toverlap = 5 nm

yx

0

From quasi-stationary to From quasi-stationary to quasi-ballisticquasi-ballistic

0

500

1000

1500

2000

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Dra

in c

urre

nt I

D (

µA

/µm

)

Source-drain voltage VDS (V)

VGS = 0.7 V

Lch = 15 nm

Lch = 25 nm

Lch = 50 nm

Lch = 100 nm

0.1

1

10

100

0 10 20 30 40 50 60

Frac

tion

of

elec

tron

s (%

)Number of scattering events

VGS = VDS = 0.7 V

100 nm50 nm25 nm

15 nm

Lch = 200 nm

Drain-end of the channel

Transition for Lch ≈ 50 nm in undoped channels

0.1

1

10

100

0 10 20 30 40 50 60

Frac

tion

of

elec

tron

s (%

)Number of scattering events

VGS = VDS = 0.7 V

100 nm50 nm25 nm

15 nm

Lch = 200 nm

Drain-end of the channel

Ballisticity and Ballisticity and channel lengthchannel length

0

20

40

60

80

100

0 50 100 150 200

Intr

insi

c ba

llis

ticit

y B

int (

%)

Channel length Lch (nm)

Bint

Bint strongly increases for Lch < 100 nm

0

20

40

60

80

100

0 50 100 150 200

Bal

list

icit

y (%

)Channel length Lch (nm)

Bint

Beff

Beff - Bint

Ballisticity and Ballisticity and currentcurrent

500

1000

1500

2000

2500

0 50 100 150 200

Dra

in c

urre

nt I on

(A

/m)

Channel length Lch (nm)

Standard Si channel

Ballistic Si channel

Gap between Bint and Beff no more constant for Lch < 30 nm

BBeffeff(B(Bintint) in undoped ) in undoped channelschannels

Strong impact of Bint on Beff for Bint < 20 %

Decreasing impact of Bint on Beff

Beff(Bint): quasi linear correlation for Bint > 20%

0

20

40

60

80

100

0 20 40 60 80 100Eff

ecti

ve b

alli

stic

ity

Bef

f (%

)

Intrinsic ballisticity Bint (%)

100 50 35 25 20 15Channel length Lch (nm)

ConteContentsnts

Introduction

Channel length and ballistic transport

Bias dependence on ballisticity

Ballisticity and long-channel effective mobility

BBintint as a function of V as a function of VDS DS for for different Vdifferent VGSGS

Bint decreases when VDS increases

Bint decreases when VGS increases

30

35

40

45

50

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Bal

list

icit

y B

int (

%)

Source-drain voltage VDS (V)

VGS = 0.3 V

VGS = 0.5 V

VGS = 0.7 VLch = 25 nm

‘‘Sta-bal’ and ‘bal-sta’ Sta-bal’ and ‘bal-sta’ architectures architectures

staorbal

staorbal

balsta

‘‘sta-bal’’ ‘‘bal-sta’’

BBintint (V (VDSDS): ‘sta-bal’ vs. ): ‘sta-bal’ vs. ‘bal-sta’‘bal-sta’

25

30

35

40

45

50

55

60

65

0 0.2 0.4 0.6 0.8 1 1.2

Bal

listic

ity B

int (

%)

Source-Drain voltage VDS (V)

bal-sta

standard

sta-bal

VGS = 0.7 V (Same VT)

Bint decrease is due to scatterings in the 2nd half of the channel

Higher phonon scattering rate > driving field

ConteContentsnts

Introduction

Channel length and ballistic transport

Bias dependence on ballisticity

Ballisticity and long-channel effective mobility

Studied strained Studied strained SGMOSSGMOS

0

20

40

60

80

100

120

0 0.05 0.1 0.15 0.2

Mob

ility

enh

ance

men

t (%

)Ge content (x)

Long channel

1.2×1019 cm-3

Tox = 0.7 nm

TSi = 10 nm

LG = 25 nm

SiO2

Strained Si

1.2×1019 cm-3

2×1020

cm-3

2×1020

cm-3

Si1-xGex (P) Pseudo substrate

LG = 18 nm

V. Aubry-Fortuna et al.,to be published

‘x’ increase Strain increasemt population increase eff enhancement

eff saturation for x > 0.15

2600

2800

3000

3200

3400

3600

14

16

18

20

22

24

26

28

0 0.05 0.1 0.15 0.2Ge content (x)

Ballisticity B

int (%)C

urre

nt Ion (

A/m

) VGS - VT = 0.8 V

VDS = 1 V

Ion and Bint increase similarly

IIonon(B(Bintint) vs. ) vs. IIonon((effeff))

1

1.2

1.4

1.6

1.8

2

2.2

1 1.1 1.2 1.3 1.4

Bint/Bint(Si)

µeff/µeff(Si)

Ion/ Ion(Si)

x = 0.10 0.15 0.20

Bint more relevant than µeff to account for Ion

(Bint(Ion) linear for Bint [ 15%, 30%])

ConclusiConclusionsons

Connections between Bint and:

• channel length

• strain

• bias

High quasi ballistic influence for Bint < ≈ 20 %

Bint more relevant than µeff to account for Ion

Role of MOS architecture (cf. paper)