Prospects and challenges

(1)

Proprietary

& Confidential

2D Materials for Electronics:

Prospects and challenges

Dr. Yu-Ming Lin 林佑明

Deputy Director

Advanced Transistor Research Division

Taiwan Semiconductor Manufacturing Company Ltd.

(TSMC)

ASIAA/CCMS/IAMS/LeCosPA/NTU-Phys/NTNU-Phys

Joint Colloquium

(2)

Proprietary

& Confidential

P. 2

Outline

 Introduction – the role of 2D materials for future electronics

 Structure and properties – graphene, transition metal dichalcogenides, and black phosphorus

 Synthesis

 2D Devices – transistor and tunnel FET

 Conclusion

(3)

Proprietary

& Confidential

Family of 2-D Crystals

Material Structure Bandgap Mobility

(cm ² /Vs)

Graphene 0 1000-100,000

h-BN ~7.2 eV (indirect) -

TMD (MX ₂ )

M: W, Mo, Hf, Zr, Ti, Cr, Ta

X: S, Se, Te

0.6 – 2.3 eV, and depending on layer #

10~500

Black

Phosphorus (BP)

~2 eV (Monolayer)

~0.3eV (few layer)

100~1000

• More than 140 two-dimensional (2-D) materials, and the number is still growing

• Covering insulators, semiconductors, and metals

(4)

Proprietary

& Confidential

P. 4

Promises of 2D Materials

(5)

Proprietary

& Confidential

Source: Intel

2.3K Transistors 12 mm

²

1.3B Transistors 122 mm

²

50 Years of Moore’s Law

Imagine fitting the entire China population in 10 original music halls –

That’s the scale of Moore’s Law!

(6)

Proprietary

& Confidential

P. 6

50 Years of Moore’s Law:

Can it be continued?

IEEE Spectrum Nov. 2011, p50

FinFET (2011) High-k Metal Gate

(2008) Strained Si (2003)

Need innovation for each generation

below 10 nm!

(7)

& Confidential

The Major Issue is Leakage

The transistor becomes more difficult to

“Turn Off” as the gate dimension shrinks.

No leakage control, no scaling!

Electrostatic Length Body thickness

Haensch et al., IBM J. R&D 50, pp. 339-361 (2006)

(8)

Proprietary

& Confidential

P. 8

 Electrostatic length

 L ~ 5  to suppress short-channel effect (SCE)

 Severe mobility degradation for body thickness < 5nm

Scaling Issue for 3D Materials

Materials Dielectric Constant

Body Thickness (nm)

Si 11.7 0.82

Ge 16.2 0.58

GaAs 12.9 0.74

* Double gate with EOT = 0.8 nm, L = 5nm

L > 5

L < 5

Less than 1nm!!

DIBL: Drain-Induced Barrier Lowering

S D

t _S

t _OX

G

(9)

Proprietary

& Confidential

Amir et al. (2015 VLSI)

2D: Ultra-thin Channel with Good Mobility

2D materials could exhibit high mobility values for sub-nm thickness!

2D Channel FET

(10)

Proprietary

& Confidential

P. 10

Fundamental Properties of (selected) 2D Materials

(11)

Proprietary

& Confidential

Family of 2-D Crystals

Material Structure Bandgap Mobility

(cm ² /Vs)

Graphene 0 1000-100,000

h-BN ~7.2 eV (indirect) -

TMD (MX ₂ )

M: W, Mo, Hf, Zr, Ti, Cr, Ta

X: S, Se, Te

0.6 – 2.3 eV, and depending on layer #

10~500

Black

Phosphorus (BP)

~2 eV (Monolayer)

~0.3eV (few layer)

100~1000

(12)

Proprietary

& Confidential

P. 12

Graphene: Lattice and Band Structure

• Two carbon atoms per unit cell

• Massless fermion

• Linear E(k) dispersion

• Zero bandgap due to inversion symmetry Lattice structure

Band structure

In-plane sp2 covalent bond

(13)

Proprietary

& Confidential

2-D Transition Metal Dichalcogenide

• Lattice structure (1H)

Top view ^X |

M | X

M = Transition metal (Mo, W, etc.) X = Chalcogen (S, Se, Te)

Graphene/BN MX

₂

4 Layer 3 Layer 2 Layer 1 Layer

• Direct gap at monolayer

• Indirect gap with smaller E _G at few layers

• Weak interlayer bonding by van der Waals forces

• Band structure of MoS ₂

Splendiani et al.,

Nano Lett. 10, 1271 (2010)

(p _z ,d _z )

(14)

Proprietary

& Confidential

P. 14

A. Splendiani et al., Nano Lett. 10, 1271 (2010) K. F. Mak et al., PRL 105, 135805(2010)

1 Layer 4 Layer

Direct to Indirect Gap in TMD - Layer dependence

10 ⁴ times stronger photoluminescence (PL) in 1 layer MoS ₂

more layers

A I

(15)

Proprietary

& Confidential

Black Phosphorus - Phosphorene

• Lattice structure – puckered honeycomb

a

₁

a

₂

Conduction band Valence band

Castellanos-Gomez et al., 2D Mater. 1, 025001 (2014)

• Direct gap (ML to bulk)

• E _G = 0.3eV for bulk, and 2.0eV for ML

• Highly anisotropic effective mass in x (light) and y (heavy) for both electron and holes

Lam et al., EDL 35, 963 (2014)

• Band structure

sp

³

hybridization

(16)

Proprietary

& Confidential

P. 16

Synthesis of 2D Materials

(17)

Proprietary

& Confidential

Carbon in All Dimensionalities

Graphite (3D) Graphene (2D)

Nanotube (1D) Fullerene (0D)

(18)

Proprietary

& Confidential

P. 18

How to Make Graphene?

(19)

Proprietary

& Confidential

Or, Try This at Home …

You may win one of these ten years ago….

Also works for other 2D Materials!

(20)

Proprietary

& Confidential

P. 20

Large-Area Graphene Growth by CVD

20 Metal (Ni, Cu, Ru, etc)

C

C C

C

~ 600°C

C

C C

Cool down C

C Metal (Ni, Cu, Ru, etc)

C

C C

CH ₄ CH ₄

SEM of CVD graphene on Cu

(21)

& Confidential

Copper

Clean copper and reuse

Electrochemical delamination CVD growth

Low-Cost Graphene Production

(22)

Proprietary

& Confidential

P. 22

Large-Area Graphene Production

2013 ~

~ 10m

60 inch 2012

~1 m 2005

~5 m

2009 - 2010

~50 cm

(23)

Security C – TSMC Secret

23

W531 R&D / ATRD

Scalable Synthesis of MX ₂

van der Zande et al., Nat. Mater. 12, 554 (2013) MoS

₂

by MoO

₃

/S CVD at 700C

K. K. Liu et al., Nano Lett. 12, 1538 (2012) MoS

₂

by thermolysis at 1000C

Challenge – how to maintain Mo:S = 1:2?

Sulfurization of sputtered metal

Gatensby et al., Appl. Surf. Sci. 297, 139 (2014)

(24)

Synthesis of Black Phosphorus

M. Kopf et al., J. Cry. Gro. 405, 6 (2014) BP by transport reaction at 650C

Challenges –

Breakthrough to achieve direct synthesis of thin-film BP on substrate!

Raman Spectra of Synthesized BP

(25)

25

W531 R&D / ATRD

Fiori et al., Nat. Nano. 9, 768 (2014)

Graphene

MoS ₂

Status of 2D Materials Synthesis

 A number of growth methods to produce large-area TMD films on various substrates (oxide, metal) are available

 Top-down growth of phospherene is yet to be demonstrated

(26)

26

W531 R&D / ATRD

2D Electronics

(27)

Proprietary

& Confidential

Graphene Technology Roadmap

K.S.Novoselov, et.al; A roadmap for graphene; Nature, Vol 490,(2012),192

Flexible and Rigid

Graphene touch panel Rollable Display Graphene FET and

Circuits

(28)

Proprietary

& Confidential

P. 28

Graphene vs. bulk Semiconductors

Si GaAs In

_.53

Ga

_.47

As InAs InSb Graphene

Electron mobility (cm

²

/Vs) at n = 10

¹²

cm

^-2

600 4,600 7,800 20,000 30,000

25,000 (flake)

~3000 (Epieaxial)

~2000 (CVD)

Electron saturation velocity

(10

⁷

cm/s) ¹ ^1.2 ^0.8 ^3.5 ⁵ ⁸

Ballistic mean free path (nm) 28 80 106 194 226 400

Band-gap (eV) 1.12 1.42 0.72 0.36 0.18 0

Key features

– A true two-dimensional system – Ambipolar transport

– Optically transparent

– Flexible with high mechanical strength

– Zero bandgap

(29)

& Confidential

Graphene Field-Effect Transistor (FET)

25 20 15 10 5 0

Current [uA]

-50 0 50

Vg [V]

V

_d

= 0.1 V

D.O.S

E

hole electron

Dirac point

S D S D

graphene

Back gate

source drain

Poor on/off ratio

Bad for digital switches

(30)

Proprietary

& Confidential

P. 30

Graphene for RF electronics

0.6

0.5 D rain C urrent [m A/  m] 0.4

-3 -2 -1 0 1 2 3

Vg [V]

Vd = 1V

i _g

i _d D

G S

g

m

V

g I



 

g m

T C

f g



 2

f _T : cut-off frequency where current gain is 1

f i

h i

g

d 1

21  ~

High on/off ratio is not required for RF applications.

(31)

Proprietary

& Confidential

GHz Graphene Device and Circuit

Lin et al., Science 327, p. 662 (2009)

Gate length: 240 nm f _T : 60 – 100 GHz

Gate length: 550 nm

f _T : 25 – 50 GHz

(32)

Proprietary

& Confidential

P. 32

GHz Graphene Device and Circuit

Output (200 MHz)

f

_LO

(4 GHz) f

_RF

(3.8 GHz)

Output (7.8 GHz)

Lin et al., Science 332, 1294 (2011)

(33)

& Confidential

Beyond 400 GHz Operation

Lg = 40nm 350 GHz (IBM)

Nanowire Gate ~ 40 nm 427 GHz (UCLA)

PNAS 109, 11588 (2012)

Nature 472, 74 (2011)

(34)

Proprietary

& Confidential

P. 36

Creating a Bandgap in Graphene

Edge roughness

Width variation

E _g  1.4eV/W, (Width in nm)

Han et al, Phys. Rev. Lett. 98, 206805

W

Graphene Nanoribbon

Challenge-

Atomically smooth edge with a precise width control

(35)

& Confidential

Choice of 2D Material for CMOS

Ling et al., PNAS 112, 4523 (2014)

(36)

Proprietary

& Confidential

P. 38

Nourbakhsh et al. (2015 VLSI Symp. )

, N=2 for double gate

Short-Channel MoS ₂ FET

- Good switching down to 11nm

(37)

& Confidential

Phospherene p-FET

Xia et al., Nature Comm. 5, 4458 (2014)

(38)

Proprietary

& Confidential

P. 40

Stability of Phosphorene in Air

10 mins 24 hours

BP BP/Gr

B. Ozyilmaz et al.,

Nature Comm. 6, 6647 (2015)

BP disintegrates in ambient.

Chen et al., Nature Comm. 6, 7315 (2015)

Passivation with BN

(39)

Proprietary

& Confidential

Issues of Contacts in 2D Materials

 Schottky barrier and Fermi-level pinning

 Currently achievable Rc for TMDs and BP are still two orders of magnitude higher than Si and ITRS requirement

F. Schwierz et al., Nanoscale 7, 8261 (2015)

ITRS 2024 requirement: Rc = 0.065 .mm

(40)

Proprietary

& Confidential

P. 43

Tunnel FET: Low voltage operation

n+ n+

G

S D

MOSFET with n/i/n junction Tunnel FET with n/i/p junction

Thermal tail is filtered by bandgap and thus is not limited to 60mV/dec as in MOSFET

e ^-

V

E _C

E _V

I _A /I _B ~ exp (-V/kT)

V = 60mV, I _A /I _B = 1/10

A B

(41)

Proprietary

& Confidential

Tunnel FET vs. MOSFET

60 mV/dec

Lower operating voltage  lower power comsumption!

(42)

Proprietary

& Confidential

P. 45

Tunnel FET based on TMDs

Appenzeller et al., IEEE XCDC (2015)

From tunneling consideration:

1. Direct bandgap

2. Small mass in transport direction

3. Bandgap ~ 0.6eV

(43)

& Confidential

Comparison of TFET Performance

A. Seabaugh et al., J. EDS 2, 44(2014)

(44)

Proprietary

& Confidential

P. 47

2D Candidates for TFET

MoS 2 MoSe 2 MoTe 2 WS 2 WSe 2 WTe 2 BP 1L BP 3L BP 5L

m hx /m o

0.64 0.72 0.75 0.43 0.5 0.3

0.15 0.15 0.14

m hy /m o 6.35 1.12 0.89

Eg (eV) 1.68 1.51 1.085 1.93 1.56 0.75 1.53 0.73 0.52



K’

K



x

y

x Multilayer BP is also promising:

• Bandgap < 1eV

• Highly anisotropic carrier mass (m _x << m _y )

• Direct bangdap

(45)

& Confidential

Vertical Tunnel FET with 2-D Materials

 Stacking of n- and p-type 2D crystals

 Enhanced tunnel area

 Ultra-short junction

In-plane TFET

Vertical TFET

3D 2D

Gong et al., Appl. Phys. Lett. 103, 053513 (2013)

n-type p-type

(46)

49

W531 R&D / ATRD

2D Heterojunction Vertical TFET

T. Roy et al., ACS Nano 9, 2071 (2015)

(47)

& Confidential

D. E. Nikonov and I. A. Young IEEE J. XCDC 1 (2015)

Tunnel FETs are promising for better EDP than Si MOSFETs.

(48)

Proprietary

& Confidential

P. 51

Prospects and challenges

2D Materials for Electronics:

Prospects and challenges

Deputy Director

Advanced Transistor Research Division

Taiwan Semiconductor Manufacturing Company Ltd.

(TSMC)

ASIAA/CCMS/IAMS/LeCosPA/NTU-Phys/NTNU-Phys

Joint Colloquium

Outline

 Introduction – the role of 2D materials for future electronics

 Structure and properties – graphene, transition metal dichalcogenides, and black phosphorus

 Synthesis

 2D Devices – transistor and tunnel FET

 Conclusion

Family of 2-D Crystals

Material Structure Bandgap Mobility

(cm 2 /Vs)

Graphene 0 1000-100,000

h-BN ~7.2 eV (indirect) -

TMD (MX 2 )

M: W, Mo, Hf, Zr, Ti, Cr, Ta

X: S, Se, Te

0.6 – 2.3 eV, and depending on layer #

10~500

Black

Phosphorus (BP)

~2 eV (Monolayer)

~0.3eV (few layer)

100~1000

• More than 140 two-dimensional (2-D) materials, and the number is still growing

• Covering insulators, semiconductors, and metals

Promises of 2D Materials

Source: Intel

2.3K Transistors 12 mm

1.3B Transistors 122 mm

50 Years of Moore’s Law

Imagine fitting the entire China population in 10 original music halls –

That’s the scale of Moore’s Law!

50 Years of Moore’s Law:

Can it be continued?

IEEE Spectrum Nov. 2011, p50

FinFET (2011) High-k Metal Gate

(2008) Strained Si (2003)

Need innovation for each generation

below 10 nm!

The Major Issue is Leakage

The transistor becomes more difficult to

“Turn Off” as the gate dimension shrinks.

No leakage control, no scaling!

Electrostatic Length Body thickness

Haensch et al., IBM J. R&D 50, pp. 339-361 (2006)

 Electrostatic length

 L ~ 5  to suppress short-channel effect (SCE)

 Severe mobility degradation for body thickness < 5nm

Scaling Issue for 3D Materials

Materials Dielectric Constant

Body Thickness (nm)

Si 11.7 0.82

Ge 16.2 0.58

GaAs 12.9 0.74

* Double gate with EOT = 0.8 nm, L = 5nm

L > 5

L < 5

Less than 1nm!!

DIBL: Drain-Induced Barrier Lowering

S D

t S

t OX

G

Amir et al. (2015 VLSI)

2D: Ultra-thin Channel with Good Mobility

2D materials could exhibit high mobility values for sub-nm thickness!

2D Channel FET

Family of 2-D Crystals

Material Structure Bandgap Mobility

(cm 2 /Vs)

Graphene 0 1000-100,000

h-BN ~7.2 eV (indirect) -

TMD (MX 2 )

(cm ² /Vs)

TMD (MX ₂ )

t _S

t _OX

(cm ² /Vs)

TMD (MX ₂ )

Top view ^X |

• Indirect gap with smaller E _G at few layers

• Band structure of MoS ₂

(p _z ,d _z )

10 ⁴ times stronger photoluminescence (PL) in 1 layer MoS ₂

• E _G = 0.3eV for bulk, and 2.0eV for ML