Inhomogeneous current distribution at oxide interface

Isao H. Inoue

Inhomogeneous Current Distribution at Oxide Interface

National Institute of Advanced Industrial Science & Technology (AIST) (Tsukuba, Japan)

2

Quantum critical phenomena

Electrostatic carrier doping

Mott transistor Exotic phonomena

集中講義「モットトランジスタは実現できるのか？」

[email protected] http://staff.aist.go.jp/i.inoue/Seminar @ TITech, 16 July 2014

本来は金属であるはずの物質が、強い電子相関（電子どうしに働くクーロン斥力）のために絶縁体となっている物質を「モット絶縁体」と呼ぶ。モット絶縁体にキャリアをドープすると、局在していた全てのキャリアがいっせいに動き出して金属になる。いわゆるモット転移である。このモット転移を利用して、新概念のトランジスタを作れないだろうか。そのためには何が必要なのか？本講義では、「強相関」「電界効果ドーピング」という二つのキーワードに関する基本的な物理を簡単に解説し、それに伴う物理現象を紹介する。そして、現状の半導体デバイスの問題点と「モットトランジスタ」の可能性について議論したい。

3

日時：2014年7月15日(火) 10:30-12:00 (90分), 13:15-14:45 (90分), 15:00-16:30 (90分) 同　　16日(水) 10:30-12:00 (90分), 13:15-14:45 (90分), 15:00-16:30 (90分) 　　　　　　　　　　　　　　　　　　　　　　　(講演会) 場所：東京工業大学大岡山キャンパス

4

Quantum critical phenomena

Electrostatic carrier doping

Mott transistor Exotic phonomena ✔on the horizon!


Tem

p

Ordered State

Classical Critical Point

Quantum criticality

Quantum Critical Point

Physical Parameters(Pressure, Magnetic field,

Carrier number, etc.)

[email protected] http://staff.aist.go.jp/i.inoue/

Seminar @ TITech, 16 July 2014

Tem

p

Ordered State

Classical Critical Point

Pressure

Super

UGe2

Super

CeCu2Si2

Quantum Critical Point

Physical Parameters

Tem

p

Tem

p

Ferro Antiferro

What happens at QCP?

Pressure

6


Continuous and reversible control of electronic states on the verge

of the quantum critical point

Randomness-free method: quantum critical phenomena is

so vulnerable to disorders

7

For QCP study, we need …


Phys. Parameters are alwayseither Pressure or Mag. Field

UGe2

Saxena et al., Nature 406, 587 (2000) [ Coleman, Nature 406, 580 (2000) ]

Ferro

Super

Pressure

Tem

p

Yuan et al., Science 302, 2104 (2003)

CeCu2Si2

AFSuper

Pressure

Tem

p

8

[email protected] http://staff.aist.go.jp/i.inoue/Seminar @ TITech, 16 July 2014 9

Electrostatic Carrier DopingWhy is electrostatic carrier

doping so difficult?


TiO2-x Co1-xO, Fe1-xO, Ni1-xO valence electron

1st electron ionisation

2nd electron ionisation

neutral composite

neutral composite

valence electron

Defects in transition-metal oxides

11

Transition-Metal Oxides ≈ ionic crystals (because of the strong electron correlations)

!

!

!

→ Defects form easily under large electric field.

Why is electrostatic carrier doping so difficult?


10-2

10-3

10-4Cha

nnel

Res

istiv

ity (Ω

cm)

300220Temperature (K)

260

Nd0.5Sm0.5NiO3

VG = 0VVG = -2.5V

S. Asanuma et al., APL. 97, 142110 (2010)

VO2

10 2

10 6

Cha

nnel

Res

ista

nce

(Ω) 1010

300150 200 250100Temperature (K)

M. Nakano et al., Nature. 487, 459 (2012)

Good examples of electrostatic carrier doping?


Electrochemical reaction…?

Science 339, 1402 (2013)

VO2VO2


Electron correlation



Electrolytic colouration

16



M. M. Abraham et al., J. Solid State Chem. 51, 1 (1984)

Mn impurity of ~100ppm

MgO single crystal (transparent)

2.5mm thick

Pt cathod

Pt anode+ + + + + + + + + + + + +

apply 1.1kV for 2hrs at 1050ºC

17

-


€

MnO+ 2e− →Mn(metal)+ 12 O2

Electroreduction!


M. M. Abraham et al., J. Solid State Chem. 51, 1 (1984)

Mn impurity of ~100ppm

MgO single crystal (transparent)

2.5mm thick

Pt cathod

Pt anode+ + + + + + + + + + + + +

apply 1.1kV for 2hrs at 1050ºC

18

-


nanocracks at grain boundary

T. Schmidt et al., Appl. Phys. Lett. 73, 2173 (1998)

instantaneous destruction of Ti film

atomic rearrangement occurs!

108

107

106

J [A/cm2]

J ~ 107A/cm2

oxygen is provided from air: no barrier is formed in vacuum

I. H. Inoue et al., Phys. Rev. B77, 035105 (2008)

19

Current induced oxidation

Electroxidation!


"Redox memory" RRAM, ReRAM, Memristor, and so on

20


Resistance switching of redox memory I. H. Inoue et al., Phys. Rev. B77, 035105 (2008)


Seminar @ TITech, 16 July 2014 22

“Lichtenberg Figure”http://www.CapturedLightning.com/

An electron accelerator of three million volts blasts electrons through the acrylic sheet. It traps the electrons inside. The electrons will stay trapped for hours, but a knock with a sharp point opens a path for them to make a quick escape.

Dielectric breakdown


“Lichtenberg Figure”http://www.CapturedLightning.com/

e

Electrons gather from all parts of the block, joining up to form larger and larger streams of electric current on their way toward the exit point. As the charge leaves, it heats up and damages the plastic along the branching trails it follows, leaving a permanent trace of its path

River of electrons

e

e

e

e


H. J. Wiesmann and H. R. Zeller, J. Appl. Phys. 60, 1770 (1986)

Breakdown forms a bush

filamentation (rapid flow of space charge)

amplification, propagation and multiplication (branching)

a point of high local field (rough electrode, conducting

inclusions, etc.)

24


H. A. Fowler, J. E. Devaney, and J. G. Hagedorn, IEEE Transactions on Dielectrics and Electrical Insulation, 10, 73 (2003)

25

Breakdown forms a bush


cathode (-)

25th June 2008 26

anode (+)

cathode (-)

anode (+)

cathode (-)

anode (+)

e O2

O2

O2

O2

electro-reduction

€

CoO→12O2(gas) + Co2+ + 2e−

Forming of RRAM

e ee e

e

ee

e e e

e

[email protected] http://staff.aist.go.jp/i.inoue/Seminar @ TITech, 16 July 2014Nanoelectronics Days 2008 @ Aachen 15 May 2008

Open Faucet Closed Faucet

Reset

Jon ~ 1×107 A/cm2

Current density at a faucet of 7x10-10cm2 (φ150nm)

27

Switching of RRAM

Transition-Metal Oxides ≈ ionic crystals (because of the strong electron correlations)

!

!

!

→ Defects form easily under large electric field.

Can we apply large electric field to TMO

without creating oxygen defects?

Isao H. Inoue Neeraj Kumar Ai Kitou

Yes! Use Parylene to suppress

the defects formation

National Institute of Advanced Industrial Science & Technology (AIST) (Tsukuba, Japan)


"Biocompatible glass is coated with protective substances for anti-migration and insulating properties and this is where the Parylene C coating comes. !It also protects the microchip from natural substances in the body, that may penetrate through micro-cracks caused by mechanical damages."

From "moving a pet to Australia" website

by ParyleneProtect surface

Parylene coated rotors and stators are used to control the Canadian arm for NASA Space Shuttle.

Parylene coated circuit boards provides excellent resistance to moisture, chemicals, and mold. Circuit boards for medical equipment can be steam and gamma sterilised. Parylene can also prevent dendrite and tin whisker growth.

From "Paratronix Inc." website

Parylene coating of paper documents, autographs, and photos retards the aging process and protects from moisture, mold, and chemicals.

30


Creation of oxygen vacancies is suppressed. !Channel is kept clean.

P.-J. Chen et al., Lab on a Chip 6, 803 (2006)

conformal coating

by ParyleneProtect oxide surface

oxides



Parylene/SrTiO3 FET

Using Parylene for the gate insulator, mobility is drastically enhanced.

But carrier density is not large...

SrTiO3

32

High-k/Parylene bilayer to accumulate more carriers

must be very thin


"Biocompatible glass is coated with protective substances for anti-migration and insulating properties and this is where the Parylene C coating comes. !It also protects the microchip from natural substances in the body, that may penetrate through micro-cracks caused by mechanical damages."

From "moving a pet to Australia" website

In General, Parylene film is very thick

Parylene coated rotors and stators are used to control the Canadian arm for NASA Space Shuttle.

Parylene coated circuit boards provides excellent resistance to moisture, chemicals, and mold. Circuit boards for medical equipment can be steam and gamma sterilised. Parylene can also prevent dendrite and tin whisker growth.

From "Paratronix Inc." website

Parylene coating of paper documents, autographs, and photos retards the aging process and protects from moisture, mold, and chemicals.

Parylene film in most of the literatures are more than ~1µm thick.

34


High-k (HfO2, Ta2O5, etc.)/Parylene bilayer

Hybrid gate insulator ! high-k materials (~15 < ε < ~25) + Parylene-C (ε=3.2)

Isao Inoue and Hisashi Shima, Japan Patent Number: 5522688, Date of Patent: 18th April, 2014



SrTiO3

AlHfO2

Au

Ti

parylene

BF-TEM image

36


SrTiO3

Al

HfO2

Au

Ti

parylene

BF-TEM image

37 [email protected] http://staff.aist.go.jp/i.inoue/Seminar @ TITech, 16 July 2014

SrTiO3

Al

HfO2

parylene

BF-TEM image

38


SrTiO3

Al

HfO2

Au Ti

parylene

STEM-EDS mapping

39

We are preparing FET devices using a conventional

photolithography

“Intel 4004 IC” the original microprocessor or “computer on a chip.”


Preliminary data of 20µm devices

41

High-k/Parylene/SrTiO3

cleaner interface

continuous doping control

National Institute of Advanced Industrial Science & Technology (AIST) (Japan)

Christos Panagopoulos

Nanyang Technological University (Singapore)

Azar B. Eyvazov*

Isao H. Inoue

CNRS & Université Paris Sud (France)

Pablo Stoliar** Marcelo J. Rozenberg***

*also AIST (now a PhD student in Cornell University, US)

**also Universidad Nacional de San Martin, Argentina, and Université de Nantes, France

***also Universidad de Buenos Aires, Argentina


Unusual I-V curves

Hardly seen in Al2O3/SrTiO3 Al2O3/SrTiO3 has some amount of carriers from the first

K. Ueno et al., App. Phys. Lett. 83, 1755 (2003)

10-6

44

A. B. Eyvazov et al., Sci. Rep. 3, 1721 (2013)


When increasing VSD

10-6

ΔV/VSD

I SD

ΔV/VSD

I SD

0.3mm/0.8mm = 0.375

increasing VSD for large fixed VG

0.3mm/0.8mm = 0.375

normal

abnormal !

increasing VSD for small fixed VG

45



10-6

When increasing VG

ΔV/VSD

I SD

ΔV/VSD

I SD

0.3mm/0.8mm = 0.375

0.3mm/0.8mm = 0.375

increasing VG for any fixed VSD

normal

abnormal !

not observed

46



Proc. Phys. Soc. 82, 954 (1963)

“Nonlinear…” by E. Scholl, Cambridge Univ. Press (2001)

Negative Differential Resistance



ΔV/VSD

I SD

0.3mm/0.8mm = 0.375

increasing VG for any fixed VSD

current path!

ΔV/VSD

I SD

0.3mm/0.8mm = 0.375

increasing VSD for small fixed VG

field domain!

Negative Differential Resistance

48



Numerical simulation: results

49



Comparison of Exp & Calc

10-6

50


Experiment Calculation


Simulation of path formation

51



Schematic picture of channel

52


High-k/Parylene/SrTiO3 FET

cleaner interface

filamentation53


same S-shape I-V curves were observed

Hardly seen in Al2O3/SrTiO3 Al2O3/SrTiO3 has some amount of carriers from the first

0 0 . 0 0 5 0 . 0 1 0 . 0 1 50

0 . 2

0 . 4

0 . 6

V 12 ( V )

290 K

280 K

I D (

nA

)

0 0 . 0 1 0 . 0 2 0 . 0 3 0 . 0 40

0 . 0 5

0 . 1

0 . 1 5

V 1 2 ( V )

180 K

220 K

210 K

200 K ID

( n A

)

€

ΔV ≡V1 −V2

€

ΔV ≡V1 −V2

55

Parylene/SrTiO3 FET


1011 1012

H. Nakamura et al., Appl. Phys. Lett. 89, 133504 (2006)

h/e2=25.8kΩ

Filamentation at 7K

Filamentation Occurs even at 7K!!

Then, what happens at ultra-low T


Seminar @ TITech, 16 July 2014 5834

SC at LaAlO3/SrTiO3 interface

S. Thiel et al., Science 313, 1942 (2006)

N. Reyren et al., Science 317, 1196 (2007)

TC ~ 200mK HC2 ~ 0.1T JC ~ 100µA/cmnonvolatile


34

SC at electrolyte/SrTiO3 interface

K. Ueno et al., Nature Materials 7, 855 (2008)

TC ~ 400mK HC2 ~ 0.1T JC ~ 300µA/cm


33

SC of bulk SrTiO3-δ

M. Jourdan et al., Eur. Phys. J. B 33, 25 (2003)

TC ~ 140mK HC2 ~ 0.3T JC ~ 100A/cm2 ~ 100µA/cm (for t10nm)

Good agreement in the orders with ! 1) SC at LaAlO3/SrTiO3 interface, 2) SC at electrolyte/SrTiO3 interface, !and ! 3) our gate-annealed SC (next slide).

31

Sample A


VG threshold is lowered. Nonvolatile metallic state.

Bulk-like superconductivity due to oxygen vacancies?

TC ~ 350mK HC2 ~ 0.1T

prolonged (one-day) applicationof large VG

SC was seen only after

Al electrode

Volta

ge (m

V)

0

0.2

Temperature (K)1.8 2.21.4

H. Nakamura et al., J. Phys. Soc. Jpn. 78, 083713 (2009).

SC in "gate-annealed" parylene/SrTiO3


32

Oxygen vacancy creation on SrTiO3

M. Janousch et al., Adv. Mat. 19, 2232 (2007)

0.2 mol% Cr-doped SrTiO3

By applying 105V/cm for about 30 min

Pt

Pt

Oxygen vacancies are created, and distributed in the channel, and form a metallic path.

All the superconductivity of SrTiO3 interface shown here might be caused by oxygen defects…

Is this the conclusion? Is this the conclusion?

No. we observed another

All the superconductivity of SrTiO3 interface shown here might be caused by oxygen defects…

39

1011 1012

Domain formation of doped carrier and percolation transition


Another “ ” State


TC ~ 400mK HC2 ~ 0.1T JC ~ 300µA/cm

·SC at electrolyte/SrTiO3 interface ·gate-annealed SC ·SC at LaAlO3/SrTiO3 interface ·bulk superconductivity

is Fragile


Summary

zzz

high-k/Parylene to protect surface

Filamentation