MBE Growth of III-V Materials and its Applications

to 2D/1D/0D Nanostructured Devices

•Constraint of carriers by heterostructure affects the physical characteristics of devices. •With it, researchers have put quantum physics in practice to enhance and modify the properties of devices. Constraint of carriers can be classified by degree of restriction of dimensions. As a result, carriers can move freely in 2 (1 and 0) dimensional space with 1 (2, and 3) dimensional constraint(s), respectively. •Naturally, not only introduction of new structures but also new materials, such as Sb-based 3-5 materials, can be resource of new idea. •In this presentation, the authors will show various MBE-grown 2D, 1D and 0D-structures. Low density large droplet quantum dots (QDs), long wavelength type-2 InSb QDs on InAs substrate, and short wavelength InP/InGaAs QDs (Dashes) will be discussed for 0 D-structures, and (In)GaAs nano-wires on (111) Si will be shown for 1D-structure. Finally, electrical properties type-2 quantum wells for p-type will be presented for 2D-structures.

•We are happy to work with co-workers in the world

Conclusion

• Riber compact 21 cluster MBE systems: 3MBEs+1E-beam evaporator + 1 Sputter [In/Ga/Al + As/P/Sb + Be/Si/GaTe]

• VG 80 cluster MBE systems: 2 MBEs [In/Ga/Al + As/P + Be/Si]

• VEECO 930 MBE [In/Ga/Al + As/P + Be/Si/C]

• Home-made MBE (under construction) [In/Ga/Al + N + Be/Mg]

3 Dimensional structures (artificial bulk, new materials) 2 Dimensional structures (type-1,-2 QWs)

In these MBE systems, all generation of 3-5 materials are grown.▲

High quality InSb (~ 70,000 cm2/Vs @ 300K, 2.6um) were grown on GaAs or Si wafers using InAs interlayers. High-quality thin InSb (0.4 um) were obtained by grading InAlSb buffers. ▲

All kind of Sb-As based ternary compound were grown. Sb-P based materials are under research. ◄

Digital alloy InGaAlAs were grown by repetition of short period superlattices of few monolayer thick-InGaAs and InAlAs.▲

This artificial bulk materials can be used for 1.3um QWs, uniform DBR, QCL lasers. ▼

InAs/InGaAs/InAlAsHEMT structure achieved the mobility of 14,000 cm2/Vs at 300 K and ~140,000 cm2/Vs at 77K. ◄

This structure was used for the implementation of Spin-FET.

Formation of InSbwell is confirmed. More in-depth study is necessary for real application. ◄

(electron mobility @ RT ~ 20,000 cm2/Vs, Ns ~ 1.5E12/cm2)

Cf) Target is > 40,000 cm2/Vs @ RT

Sb-based type-2 QW shows large mobility of 2DEG/ 2DHG which are critical for low power consumption devices. ◄

1/1000 of power consumption is expected with implementation of 3-5 CMOS with this structure.

GaAs nano rods grown by Ga-droplets (Catalyst-free Nano wires) were successfully achieved on (100) GaAs.◄

This nano rods has perfect ZB structures without contamination of WZ structures.

1 Dimensional structures

Gold nano-particles were used as a catalyst for 3-5 nano rods. Nano rods were grown on (111) Si substrates with gold nano-particles. ◄

The GaAs Nano rods shows perfect WZ without contamination of ZB structures.

The nano-rods can be used for perfect anti-reflects.

Various nano rods such as In(AsPSb) and Ga(AsPSb) nano rods are prepared or under research for single photon source or 1D electronic devices.

0 Dimensional structures

Conventional SK-mode grown In(Ga)As QDs are grown on GaAs, Si. These were used for solar cell, QDIP (IR sensor) etc. ▲

Very low density ( few QDs/um2) InAs QDs were grown for quantum physics studies. ◄

With GaAs/AlGaAs droplet QDs, we can grow QDs, Q-rings, Q-disks, and Q-pits. ▲

This structure can be used for template of new noble structures such as artificial molecules. ▼

MEE-mode grown InAs, InGaAs QDs were used for 1.3um-LD, PD etc.◄

InGaAlAs QDs were used for short-wavelength LD.

QD can be

Position controlled

3-5 devices can be transferred to

Si wafer

3-5 devices can be direct grown on Patterned Si

Band gap engineering

in thin film inorganic materials

1. Synthesis of Low-D Nanomaterials

2. Ultrafast Opto-Electronic Devices

3. Smart Human Interfaces

Low-D Nanostructured Materials & Devices Nonlinear nanomaterials Novel materials for nonlinear saturable absorbers & optical switches In situ synthesis of graphene on electronic and photonic devices Direct synthesis of nanomaterials onto integrated photonic devices High performance semiconductor materials & devices Novel materials for printed waveguides

Nanomaterial-Based Ultrafast Photonics Femtosecond fiber lasers Ultrafast nonlinear switches Optical logic gates Printed optical waveguides Fiber optic passive components High-speed optical signal management Novel modulation formats

Smart Contact Lenses Non-invasive health monitoring Diabetes monitoring via tear Biosensors with ultimate sensitivity & selectivity Microfluidic channels for reliable operation of the sensors Integrated chips for sensor control & data communication Remote power supply and body channel communication Tunable lenses

E

kx

ky

E

kx

ky

Nonlinear Saturable Absorption in Graphene

Inc

II

II

S

S

20

20

)3(

int0

0int0

int0

2)Im(3

~

/1

εχωαα

ααα

ααα

++=

−+

++

= Icn

nInnn 200

)3(

020 4)Re(3

εχ

+=+=

))()()(()( 3)3(2)2()1(0 ⋅⋅⋅+++= tttt EEEP χχχε

Absorption Refractive index

c

a

Deposited nickel layer

Ni atomC atom Grain boundaries

Amplified CW laser

t

CW laser after absorption

tWaveguide

Nickel on the waveguide

Interface Growth of Graphene in situ Synthesis of Graphene

Directly Grown CNTs onto Optical Fiber Substrate

Black Phosphorus-Based Field Effect Transistors

CVD-Grown Graphene

Graphene Growth on Ceramics (Metal-Free Growth)

Optical Deposition of Graphene Oxides

Ni (Catalyst)Graphene layer

(AFM data)Substrate

200 nm

Catalystremoval

Resultant graphene layer that does NOT require graphene transfer

Pulsed Output 10/90

Coupler EDFAIsolator

Isolator

PolarizationController

FiberFerrules

SWNTs or Graphene

Nanomaterial-Based Fiber Femtosecond Lasers

Nanomaterial-Based Ultrafast Nonlinear Kerr Switches

Four-Wave-Mixing in in situ Grown Graphene

Four-Wave-Mixing in Black Phosphorus

Integrated Optical Switches Functioned with Nanomaterials

Graphene onD-shaped fiber

Pump LD(λ: 1552.4 nm)

PC 2

High power EDFA

Band pass filter(3.2 nm )

PC 1

OSA

3 dB coupler

Analyzer (130 MHz~20 GHz) Band pass filter

(0.8 nm)

Signal LD (λ: 1559 nm)

EDFA

Nanomaterial-Based;Light Source

Photo DetectorTunable Filter

Optical ModulatorOptical Switch

Logic Gate

Electrical Layers

Optical Layers

Flexibility and transparency of materials and devices

Conventionally inefficient, painful glucose monitoring

Development of novel contact lens-

typed monitoring tool

Tear-based, painless, continuous monitoring

Highly efficient power management for sensor operation and data communication

Sensor head Microfluidic channels Secondary thin film battery Nano generator Integrated circuits

2014: Kick-off the Research 2016: Exhibition in Nano Korea

Individual devices and structure have been realized and displayed.

2017: Assembled Contact Lens Sensor Platform

Sensor head Microfluidic channels Secondary thin film battery Integrated circuits

2018: Newly Designed Contact Lens Sensor Platform

Sensor head Microfluidic

channels Chips for remote

power and data delivery

Transparent flexible devices (e.g. Omni-flex display devices)

Multi-purpose lenses

Health monitoring

Augmented reality (living info.)

Augmented reality (Navigation)

Artificial interfaceHigh-resolution display

on the contact lens

Future Researches

Hostmaterial

incorporating2D nanostructures

LP01-Y

∆φ(X-Y)

x

y

Pump

Probe

LP01-X

x

Polarizationrotatedprobe

Synthesis OF Low-Dimensional Nanomaterials and

Their Applications to Ultrafast Opto-Electronic Devices

2D/3D Imaging and Display Systems

• 2D or 3D images can be obtained with a single pixel sensor by proposed several imaging systems. • 2D/3D imaging system can be exploited to verify the imaging capability of prototype image sensors fabricated in laboratory level.• Any types of photo sensitive sensors to various wavelength (from UV to IR) can be tested by a designed imaging system.• Quality of primitive images obtained by the imaging system can be enhanced with various signal processing algorithms.

Conclusion

2 Dimensional Imaging system 3 Dimensional Imaging system

Optical Scanning Holography (OSH)

2 Dimensional image processing 3 Dimensional contents display

Chromatic confocal microscope : It is possible to use depth of focusgenerated in the microscope by using spectrometer

Single point confocal microscope + 2D linear stages → 3D imaging of thesamples

1cm

Cosine holo

50 100 150 200 250 300

50

100

150

200

250

300

Sine holo

50 100 150 200 250 300

50

100

150

200

250

300

Cosine holo Sine holo

Reconstruction

An optical wave is described by its amplitude and its phase. Theprincipal of holography is to generate an interference pattern so that theintensity captured in a given plane contains both amplitude and phaseinformation. It is then possible to reconstruct a 3D scene.

Optical scanning holography (OSH) is a technique to record complexhologram from a real object with a single pixel senor.

Chromatic Confocal Microscope

3D reconstruction result

200um

100um

SCAN

Single pixel imaging (mechanical scanner)

Compressive sensing does not require mechanical scanning.

Several measurement are taken for different binary pattern displayed by the SLM.

Single pixel imaging (Compressive sensing)

Emulation of plant’s vision by artificial photosynthesis

Artificial photosynthesis properties of TiO2 nanowire array was exploited to emulate plant’s vision

M: Number of binary pattern

N: Size of binary pattern

the images of Kalanchoe blossfeldiana captured by different sensors. Although the image takenby the proposed image scanning with the TiO2 nanowire array presents lower resolution ascompared to the images taken by the two different CCDs, the UV absorbing pattern is clearly seen.

Object Detection & Tracking

Stitching images Find features

Match features

Estimate homography matrix

Wrap images

Blend images

360 Degree Image Stitching

Multi-wavelength band fusion

Noise reduction filter

Eye tracking system

Motion capture based on IR markers

Set-up for Reconstruction of Holography

Reconstruction of Holography

In the reconstruction process, the hologram isilluminated by laser beam and this beam iscalled reconstruction beam. This beam is iden-tical to reference beam used in construction ofhologram.

The hologram acts a diffraction grating. Thisreconstruction beam will undergo phenomenonof diffraction during passage through the holo-gram. The reconstruction beam after passingthrough the hologram produces a real as well asvirtual image of the object.

Integral imaging display

Holographic display

Super High Barrier FilmsSiNx:H SiOx SiOxNy / SiOx

PECVD process

Silica sol dip coating

Thermal treatment

Chemical reaction formula

We fabricated high-performance inorganic SiNx/SiOxNy/SiOx barrier films

on PET substrates by using PECVD and simple dip coating.▲

SiOx

SiOxNy : 4.76 nm

SiNx

SiOx : 103.78 nm

328.51nm SiNx

PET (125㎛)

SiN x (PECVD 8min)SiOxNy

SiOx

Dip coating

( 1mm/s)

About

500nm

◀ The SiNx/SiOxNy/SiOx barrier

film had a transmittance of

90.37 % and the WVTR was

< 5.0×10−5 g/ m2 /day.

TEM images of SiNx/SiOxNy/SiOxSEM image of SiNx

300 400 500 600 700 800

40

60

80

100

Tran

smitt

ance

(%)

Wavelength (nm)

Bare PET : 88.79% SiNx on PET : 75.87% SiNx/SiOxNy on PET : 90.37%

Oxide Semiconductor and Polymer Hybrid

Nanomaterials for Soft Nano Electronics

• Intrinsic defects controlled ZnO and ZnO-graphene hybrid quantum dots: UV Photo-Excited White-Light Emission and Electro-Excited Purple-Blue Light Emitting Diodes

• Highly dispersible nanospring single-walled CNTs (NS SWVNTs): High dielectric PVDF and PDMS for energy harvesting with high contents of CNTs.

• SiNx/SiON/SiO2 super high barrier film: WVTR< 5x10-5 g/m2/day @1 dyad, <3x10-5 g/m2/day @2 dyad

• Coaxial PVDF-TrFE nanofiber(37μm)/metal wire(200 μm ): 0.3 V@0.3 N & ca.0.9 V@5.1N

Conclusion

(a) Inline magnetron sputter : Antireflection nano coating

(b) Inline magnetron sputter : TCO coater (ITO, WOx)

(c) UHV multi-sputter : Oxide semiconductor (ZnO, NiO, SnOx)

(d) Continuous Electrospinning : PVDF-TrFE/metal wire

(e) Glove box: thermal deposition for QD LED & OPV

(f) Glove box: PECVD for barrier film, slot die coating

ZnO-nanocarbon (Graphene, C60, CNTs), Carbon dots Nanospring Single Walled CNTs (NS SWCNTs)

Nano Energy Harvesting

• HR TEM image of ZnO-SWCNTs complex synthesized at various reaction time

• Chemical synthesis process for the buckled NS-CNTs

• The role of PVP and the schematic diagram

• Measured dielectric constant data of the P(VDF-TrFE) and NS-CNTs-PVP nanocomposite

a b

c d

e f

Bare PET SiNx on PET SiNx/SiOxNy/SiOx on PET10-5

10-4

10-3

10-2

10-1

100

101

< 5.0 x 10-5

1.4 x 10-2

WVT

R (g

/m2 /d

ay)

4.62

Continuously Electrospinning coating system on metal wire

◀

Power Supply

PVDF-TrFE

Solution

MotorCu wire

22 w/v% 24 w/v%

26 w/v% 28 w/v%

▲ PVDF-TrFE nanofibers on Cu wire was various concetration of PVDF-TrFE in solvent

20 40 60 80

Inte

nsity

(a.u

)

2 theta (2θ)10 20 30

Inte

nsity

(a.u

)

2 theta (2θ)

19.7 °

Cu (111)Cu (200)

Cu (220)

▲ XRD 2 θ spectra of PVDF-TrFE nanofibers on Cu wire at 28 w/v% PVDF-TrFE solution. Left 2 θ spectrum is range of 10 ° ~ 90 °, right one is enlarged at 10 ° ~ 30 ° range of left spectrum

200 μm 37 μm

1 cm 0 1 2 3 4 5 6-0.50

-0.25

0.00

0.25

0.50

Volta

ge (V

)

Time (s)

0.3N

-0.010

-0.005

0.000

0.005

0.010

Cur

rent

(µA)

0 1 2 3 4 5 6-0.5

0.0

0.5

1.0

Volta

ge (V

)

Time (s)

5.1N

-0.01

0.00

0.01

0.02

Cur

rent

(µA)2 x 2 cm2

0 1 2 3 4 5 6

-0.2

0.0

0.2

Volta

ge (V

)

Time (s)

-0.010

-0.005

0.000

0.005

0.0100.4N

Cur

rent

(µA)

0 1 2 3 4 5 6-0.5

0.0

0.5

1.0

1.5

2.0

Volta

ge (V

)

Time (s)

5.4N

-0.01

0.00

0.01

0.02

0.03

0.04

Cur

rent

(µA)

Φ = 1 cm 2 x 2 cm2◀ Cross sectional view of core-shell PVDF-TrFEnanofibers/metal wire

▶ Electrical properties of one and three core-shell PVDF-TrFE nanofiber /metal wire

Ligand exchange & Reduction

Carbonization

Amorphous domain

Crystalline domain

Citric acidCarbonization

Oleylamine

Synthesis of nano-carbon quantum dots (CQDs) by carbonization, ligand exchange, & reduction ▲

300 400 500 600 700200

300

400

500

600

700

390 nm

Emission Wavelength (nm)

Exci

tatio

n W

avel

engt

h (n

m)

477 nm

450 nm

QY :

6 %

300 400 500 600 700200

300

400

500

600

700

532 nm

450 nm

390 nm

442 nm

Emission Wavelength (nm)

Exci

tatio

n W

avel

engt

h (n

m)

QY :

30 %

2D PL mapping

Mixed dimensional semiconducting materials and their devices for next generation electronic/optoelectronic applications

Conclusion

Organic and Metal oxide semiconductors thin-films 0-dim Quantum dots

1-dim Nanowires 2-dim vander Walls (vdWs) nanosheets

New device platform technology with mixed-dimensional heterostructures

1-D nanowire 0-D QD

ex) PbS, CdSe ex) ZnO, InAS NW

ex) MoS2, WSe2, BP ex) InGaZnO

2-D nanosheet 3-D film

Mixed dimension

Pt

Au/Ti

Au/Ti

WSe2

Au/Ti

ZnO

BP

ass

0-D QD/3-D metal oxide hybrid

1-D Nanowire/2-D nanosheet hybrid

Equipment Electronic/Optoelectronic devices measurement system

Development of Organic and metal oxide semiconductors devices• High performance organic devices with solution process• Fundamental study on organic device reliability

• Organic/Metal oxide hybrid complementary inverter• Solution processed metal oxide thin film electronics

0-D PbS/3-D InGaZnO hybrid phototransistor for NIR imager• Gate tunable, highly sensitive, and easily integrated PbS sensitized IGZO hybrid

phototransistor for NIR detection

• PbS/IGZO device exhibits

photo detection capabilities

for NIR light up to 1400 nm.

• The photo-generated electrons

from the PbS sensitized layer

lead to significant negative

shifts of threshold voltage (Vth)

in the IGZO TFTs.

Metal shadow mask 1300 nm NIR image

z-axis(500μm)

x-axis(200μm)

VDD

VOUT

VIN1.5V

100 MΩ1300 nmLED source

• NIR (1300 nm) Imager

1-D Ag metal nanowire electrode

Ag NW Ag NW Ag NW

hv

hv

0

45ITO

PEDOT:PSS

P3HT:PC60BM

PET

PEDOT:PSS

P3HT:PC60BM

Ag NW

Al Al

Absorption density

-0.2 0.0 0.2 0.4 0.6

-10

-5

0 ITO Ag NW

Curre

nt d

ensi

ty (m

A/cm

2 )

Voltage (V)

P3HT:PC60BM

• High performance flexible organic solar cell

Light scattering and lighttrapping induced by the Agnanowire mesh: the enhancedlight absorption in the activelayer.

• IGZO TFT with Ag nanowire electrode based chemical/biological sensor

In Out

PDMS

Analyte

OH

OH

HO

HOOO

Glucose oxidase

OH

OH

OH

HO

OHO

b-D-glucose

OH

O

OH

Lactic acid

HO OHHydrogen peroxide

-1.0 -0.5 0.0 0.5 1.010-12

10-11

10-10

10-9

10-8

10-7

10-6

Ag NW mesh pH 3 pH 5 pH 7 pH 11

I DS (A

)

VGS (V)

VDS = 0.1 V

-20 -10 0 10 2010-1310-1210-1110-1010-910-810-710-610-510-410-3

Dark 1500 1400 1300 1000 700

VGS (V)

I DS (A

)

W/L = 1000 µm/ 50 µmVDS = 20 V

MixtureSingleOH

OH

OH

HO

OHO

b-D-glucose

+ O2OH

OH

HO

HOOO

D-glucono-1,5-lactone

+ HO OH

hydrogen peroxide

Sensors detect pH solution as well as biologically

relevant species such as H2O2, b-D-glucose, D-

glucono-1,5-Lactione, and Lactic acid in aqueous

media.

2D vdWs electronic/optoelectronic devices• Nonvolatile memory devices

Au top gateP(VDF-TrFE) (220 nm)

2D vDWs

GrapheneSource

GrapheneDrain

SiO2 (285 nm) / p+-Si

MoS2 and BP based ferroelectric memory transistor

Control gate (Au, 50 nm)20 nm

Al2O3 (35 nm)

Tunneling L (Al2O3, 5 nm)

Trapping L (BP, 6 nm)

Active L (BP, 7 nm)Glass substrate

0.56 nm

0.56 nm

4 nm4 nm

BP based charge injection memory transistor (flash memory)

• MoS2 image sensor

Au/Ti Au/Ti

MoS2Gr S/D Gr S/D

GreenLight

1Cm

1Cm

• 1D nanowire- 2D nanosheet heterostructures 1D ZnO – 2D BP diode and JFET 1D ZnO – 2D WSe2 photodiode

Device

LightSource(LED)

WSe2

• Organic/inorganic thin films and nanostructured materials based electronic/optoelectronic technologies have made great progress. In the thin film technology, organic/inorganic metal-oxide semiconductors are a promising alternative to amorphous or poly silicon.

• Low dimensional semiconducting materials such as colloidal QDs and van der Waals (2D vdWs) atomic crystals are an emerging class of new materials that can provide important resources for future electronics and materials sciences due to their unique physical properties.

• In this presentation, our group show various electronic and optoelectronic devices using 0D QD, 1D nanowire, 2D nanosheet, and 3D thin film semiconducting materials: Organic and metal oxide transistors, PbS QD sensitized InGaZnO phototransistor for NIR detection, Organic Solar cell and InGaZnO biosensor with 1D Ag nanowire electrode, and 2D vdWs semiconductor based electronic/optoelectronic devices (nonvolatile memory, image sensor, and 1D-2D heterostructures based devices). Our group will further develop a new device platform with mixed-dimensional heterostructures

• We are happy to work with co-workers in the world

Organic Semiconductor Engineering for

Printed & Soft-Electronics

Solution-Based Organic Electronics

Facile deposition of functional materials - 1D, 2D & 3D structure

<KIST die coating system>

Coating on Fiber

3D printing & Printing on 3D structure

Slot-die coating

Everywhere Electronics !

Large area- R2R

Multi function- Integration

Low cost- Solution-process

LightingPublic information

Wall Display

EntertainmentHealthcare/Sensor

Printed & Soft Electronics Vision Organic Semiconductors

Introduction 3-Dimensional self-organization of printed organic semiconductor

Fibriform, weavable organic transistor for E-Textile Split-second nanostructure control by intense pulsed light

Freestanding diamond plates (left) synthesized by KIST own multi-cathode direct currentplasma assisted CVD apparatus (right)

Carbon Science

• CVD synthesis of diamond: We developed KIST own apparatus, MCDC-PACVD which enables us to synthesis of freestanding diamond wafers (~1 mm thick) of 4”in diameter.

• Discovery of AA’ graphite: We discovered new crystalline structure of graphite which is metastable for AB graphite and shows nano-crystalline feature.• Define structures of CNTs: We defined structures of SWNTs and MWNTs to be a graphene helix and graphite helices, resulted from helical growth of a graphene

and graphited helices, respectively.• Graphene: We developed a new simple way to identify mono- and bi-layer graphene from ‘radial mode’ (appearing at 100-300 cm-1) on Raman spectrum (Monolayer

graphene is very seldom and nobody have shown evidence of the presence of monolayer graphene in micron size).

• We are happy to work with co-workers in the world.

Summary and comments

CVD synthesis of Diamond

~130 carat

Discovery of AA’ graphite

Evidence of GrapheneStructure of CNTs: Helix

Diamond shells (~ Φ 30 ㎛)

We explore the nature of carbon for next generation electronics

Φ 20 mm

DC plasma

~130 carats

HRTEM image revealing interface between diamond and graphene

Structure model of multi-wall carbon nanotubes

Structure model of single-wall carbon nanotubes

HRTEM images revealing the traces ofgraphite helix comprising MWNTs

Atomic structure of AA’ graphite assignedas orthorhombic

Energy landscape of graphite. AA’ stacking ofgraphene is metastable phase of graphite.

HRTEM image revealing mono- and bi-layer graphene where ends are curved (left) and lowenergy Raman radial mode (RM) signals (right).

Schematic models explaining radial mode (RM) of Eigen vectors

HRTEM images and Raman spectrum of plasmaseeded grown graphene

Atomic-resolution TEM morphology imagingoverlapped AA’ bilayer graphene.

Crystalline models of AA’ and AB graphiteand their simulated HRTEM images

(Lee et al., J. Phys. Chem. Lett. 2017)

(Lee et al., APL, 2013)

(Lee et al., Small, 2014)

(Lee et al., Scientific Reports, 2016)

(Lee et al., J. Phys. Chem. Lett. 2008)

(Lee et al., Diam. Relat. Mater., 2001)

(Lee et al., CVD, 2006)

Direct current plasma generated by KIST own equipment for synthesis of diamond

Memory Application based on Stochastic Devices: Random Number, Security, and Multi-Level Memory Cell

• Non-volatile memory application merges as alternative to conventional CMOS-based computing era. • Neuromorphic technologies and stochastic computing have attracted tremendous attention to overcome the limit of Moore’s law.• One of promising candidates is resistive change memory, which intrinsically has stochastic and multi-level characteristics applicable to those applications.• In this presentation, representative applications and results of ReRAM and ovonic threshold switch(OTS) devices are demonstrated – random number generator, synaptic device formachine learning, and physical unclonable function(PUF) for hardware security.

• Electrical investigation on the devices and computational works run complementarily such as device model simulation, output data processing, and novel algorithm development.• We are happy to work with co-workers in the world.

Random Number Generation (RNG) Security : Physical Unclonable Function (PUF)

0 20 40 60 80 100 120 140 160 180 200

0

1x10-8

2x10-8

3x10-8

4x10-8 LTP LTD

Nor

mal

ized

G

Normalized Pulses

-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.510-910-810-710-610-510-410-310-2

Curre

nt (A

)

Voltage (V)

Measure Fitting

0 25 50 75 100

S/A = 5.5uAResponse = 60b

Cou

nt (A

.U)

Randomness (%)

w/o Mix Mix

0 25 50 75 100

Cou

nt (A

.U)

Uniqueness (%)

w/o Mix Mix

S/A = 5.5uAResponse = 60b

Non-volatile Resistive Change Memory

Multi-Level Cell Characteristic DC I-V Model Simulation

Stochastic Cycle-to-Cycle Characteristic Stochastic Cell-to-Cell CharacteristicSynaptic Behavior : Classification Test

0 20 40 60 80 100 120 1400

102030405060708090

100

Accuracy

Acc

urac

y [%

]

Epoch

Source : IBM X-Force Threat Intelligence Quaterly (2014)

Hardware Authentication : Challenge-Response Pair by PUF

CRP Generation Algorithm

Multi-Level Conductance (from nl904092h)

Training Result

Multi-layer Perceptron Neural Network

From IEEE TCAD.2018.2789723

110001110001001011

= 9/18

111001110000000011

= 8/18

Response of a Cortical Neuron (J.P. Hayes, DAC 2015) RNG Circuit with OTS (from S. Lee)

Vin

0.000 0.002 0.004-100

0

100

200

300

400

500

bin size : 50mV

Va [m

V]

Time [sec]

Va [mV]

Null

0.00 0.02 0.04

0

100

200

300

400

500

600

700

800

Va [m

V]

A

Va [mV] Null

-100 0 100 200 300 400 5000

500

1000

1500

2000

2500

3000

Null

Cou

nt

Va [mV]

raw w/o Null

Log-Normal Dist.

0

50

100

150

Cou

nt

Output of 10-pulse train (input)

Time [sec]

RNG Algorithm for a Single Pulse Input

Modified Random Number Distribution

Bit-Error-Rate Improvement Binning Sensitivity Effect Bit Streaming Comparison

Sampling Profile to Population Randomness Comparison Uniqueness Comparison

300 400 500 600 700 800 900

Inte

nsity

[a. u

.]

Wavelenght [nm]

Growth Principles

Synthesis of Hybrid Two-Dimensional

Nanomaterials in Layered Structures

Vision

Conclusion

Roadmap for Optical Integration

Developing Optical Logic Module

Ensure efficiency of photonics devices due to extremely low and high nonlinearity

of two-dimensional nanomaterials

For ultra low power and ultrafast operation, two-dimensional nanomaterials are needed.

Direct Synthesis of 2D Nanomaterials on Insulating Substrates

“Metal-free” & “Transfer-free” growth

“High-quality” & “Large-area” growth

Metal catalysts are not allowed for real semiconductor industry.

For facile fabrication, transfer is not good.

Direct synthesis of 2D materials on insulating substrates is essential

What we need is the intrinsic properties.

What we want is not flakes form.

Facile process is our way.

“Patterned” & “Layered” growth

Selective etching process has not been developed for hybrid two-dimensional materials in layered structures.

For special functions, new metamaterial structures are needed.

Patterned and layered growth is the answer.

Graphene

h-BN

Precursor supplying unit

Non-metal only

Precursor

ROM

Resistorcall

Translate

Control Unit

ALU

Execution

Resistor

StackedMemory

B

U

S

Chipset

IBM next generation microarchitecture

CPU Modularization

ROM

OpticalIntegration

ElectronicModule

동기화

E/O

Optical Computing Unit

OpticalDistributor

O/E

StackedMemory

B

U

S

Mounting the electronic modules on optical logic module

Chipset

Laser

E/O

E/O

O/E

O/E

Growing Devices and Functions2D

Universal 2D Growth ModelLow temp. η + Nx ~ Nx

High temp. η + Nx ~ η (𝜼𝜼 + 𝑵𝑵𝒙𝒙)𝟐𝟐

𝒅𝒅𝑵𝑵𝒙𝒙

𝒅𝒅𝒅𝒅=𝑪𝑪𝑹𝑹𝟐𝟐𝒑𝒑𝒂𝒂𝒅𝒅𝒅𝒅,𝒔𝒔𝑵𝑵𝟎𝟎

𝒒𝒒𝟎𝟎𝟐𝟐𝒑𝒑𝒅𝒅𝒑𝒑𝒂𝒂𝒅𝒅𝒅𝒅𝟐𝟐 −𝜷𝜷𝒑𝒑𝒎𝒎𝑹𝑹

𝒒𝒒𝟎𝟎𝒑𝒑𝒂𝒂𝒅𝒅𝒅𝒅𝑵𝑵𝟎𝟎(𝜼𝜼 + 𝑵𝑵𝒙𝒙)𝑵𝑵𝒙𝒙

𝟐𝟐

Relation

Growths based on Energy Barriers

Prediction and control of growth characteristics based on extensive energy barriers

h-BN

Gr

Guided Growths

Various templates for high-quality growth

Y. Gong, et al. Nature Commun. 2014, 5, 3193.

295 290 285 280

Inte

nsity

[a. u

.]

Binding Energy [eV]

C=C sp2 : 284.6

C-O : 286.5

96.1%

3.9%205 200 195 190 185 180

Inte

nsity

[a. u

.]

Binding Energy [eV]

h-BNB1s

415 410 405 400 395 390

Inte

nsity

[a. u

.]

Binding Energy [eV]

h-BNN1sFWHM : 1.64

FWHM : 1.70C1s

High-quality growth of hybrid h-BN//graphene

Electro-optic Modulators

Development of synthesis of hybrid 2D materials directly on photonics system

~4 nm thick h-BN film is need for photonics and electrical devices.

Mono- or Bi-layered h-BN film is essential for remote epitaxial growths.

Improving quality

Integration

Studying of E-O modulation

Layer number control

h-BN, graphene direct patterning

Achieved

OngoingElectrical signal

0 10 20 300

20

40

60

80

Thic

knes

s[nm

]

Time [min]

12

0

6

Height : 1.93 nm

nm

Layer number controlled h-BN (Quality: 96~100%)on various insulating and semiconducting substrates

h-BNSubstrate

Flexible White LEDs

We can grow 2D materials on fragile structures.

Facile processes were only used for our LEDs.

Surface properties can enhance PL intensity up to 2 orders.

20 nm thick GO bubbles

x100 x100

Starting material is graphene oxide (GO)bubbles (Dr. Kwon Seokjoon)

Applications of Interest

Graphene for band-gap engineering

h-BN for implementing intrinsic properties

(organic electronics, flexible electronics)

h-BN for flexible barrier films

Growing h-BNC for band-gap engineering

Synthesis of hyperbolic metamaterials

for vertical interconnects

Y. Gong, et al.

Nature Commun. 2014, 5, 3193.

Homogeneous

Domains

We are CVD synthesis guys.

We are synthesizing two-dimensional nanomaterials directly on insulating and semiconducting substrates

without the assistance of metal catalysts.

By using various diffusions and extensive energy barriers, high-quality and patterned 2D nanomaterials

in layered structures can be directly formed on those substrates.

h-BN is the only 2D insulating nanomaterial that guarantees the intrinsic properties of atomic thick 2D nanomaterials.

Our synthesis provides very important clues to the implementation of high-performance photonics and electronics devices.

We are interested in light controls for optical computing.

We believe that our unique synthesis provides a very important clue for implementing

optical integrated systems.

Our 1st goal is implementing E-O modulators on 3D photonics devices.

We want to provide well designed hybrid 2D layered nanostructures.

If you want to discuss with our group, do not hesitate.

E-mail: JPark@kist.re.kr, Phone: 02-958-5328

Enables two-dimensionalnanomaterials

Quantum Random Number Generator

State-of-the-Art CMOS-APDs/SPADs

Advanced APD/SPAD Technologies for

LiDAR/Biomedical/Quantum Applications

CMOS-APD Technology

Conclusion

SPAD Technology

Applications – LiDAR, Bio, Quantum

Open for research collaboration!

We are looking for motivated postdocs and students! Contact: Dr. Myung-Jae Lee

(E-mail: mj.lee@kist.re.kr / Phone: 02-958-5309)

VR

IR

Geiger mode:SPAD

Linear mode:APD

Single-Photon Avalanche Diode

V

V

I

ConventionalAvalancheGeiger

Gain

VB

1VE + VB

Depletion RegionPhotodiode

EC

EV

Depletion RegionSPAD

EC

EV

VR

Cur

rent

Avalanche

APD SPAD

Photon

Off

OnQuench

Recharge

VB+VEVB

VB+VE

IA

Rq AvalanchePulse

Photon

Very high gain of Geiger-mode operation allows for

Single Photon Detection

Digital nature of SPAD output allows for

Time-of-Flight Detection&

Photon Counting

CMOS-Compatible Avalanche Photodetector

N-well

P+N+ N+STI

STI

P-substrate

P-well

P+ STI

P-well

P+STIHole

Diffusion

ElectronDiffusion

Penetration depth > 10 μm @ 850 nm

1~1.5 μm

Light Source

P+/N-well CMOS-APDs• Elimination of slow diffusion currents

from P-substrate Bandwidth enhancement (~GHz)

• Elimination of diffusion currents Reduced responsivity

Responsivity enhancement by high avalanche gain

[Impact ionization processes foravalanche multiplication]

• Responsivity [A/W]- Sufficient absorption region- Wider depletion width

• Photodetection Bandwidth- Photogenerated-carrier transit time - RC time constant- Parasitics

Important parameters

DepletionRegion

P N

E-field

V- V+

Popt

WD

hν

hν

hν

Electrondiffusion

HolediffusionDrift

s

D

AC

Wε

=

- Avalanche gain

- Inductive-peaking effect

Optimization and Improvement of Silicon APDs in Standard CMOS Technology [Invited Paper, Front Cover Article]

JSTQE’18

Equivalent Circuit Modeling Essential for CMOS-APD analysis and CMOS-Rx design

EDL’08

Optical Power Dependence Essential for variousoptical-interconnect applications

JSTQE’14

Junction– N+/P-well, (P+/N-well) World’s best gain-bandwidth (1820 GHz) OpEx’

10

Guard Ring– STI GR World’s best avalanche gain (2,500) EDL’

12

Active Area– 10×10 μm2 World’s best BW (7.6 GHz) TED’

13

Silicide Under Contacts Responsivity (40 %) &BW improvement (63 %)

EDL’16

Carrier-AccelerationTechnique BW improvement (40 %) PTL’

15

Spatially-Modulated APD (SM-APD) World’s best BW (12 GHz) PTL’

16

CMOS-Rx (SM-APD + TIA + EQ + LA) World’s best speed (12.5 Gb/s) OpEx’

14

CMOS-APD: Development, Full Characterization & Analysis, Optimization, and Performance Improvement Techniques

• Key features

‒ High responsivity over 0.3 A/W

‒ High bandwidth over 23 GHz@ 850 nm

‒ Based on standard CMOS

Photon time-of-flight with picosecond resolution

Object

Stopwatch

Speed of light ‘c’ = 3x108 m/sTime of flight (ToF) = 20 ns (20x10-9 s)

ToF2

Distance D = c x = 3 m

Direct Time-of-Flight (ToF) for LiDAR

We can get distancefrom object to receiver

From a pulse of light reflected by an object

3D Face Recognition

Space Navigation SystemTime-Resolved

Raman Spectroscopy

NASANASA

Apple

LeddarTech

Autonomous Vehicle

GE Healthcare

Time-of-Flight Positron Emission Tomography (TOF PET)

Fluorescence Lifetime Imaging Microscopy (FLIM)

AQUA

Near Infrared Imaging (NIRI)

Becker & Hickl GmbH

LiDAR(Light Detection and

Ranging)

BiomedicalApplications

Super-Resolution Microscopy

AQUA

U2OS cells stained with Alexa 647

Convallaria

20

-2-2

0

2

0

100

200

300

400

Phot

ons

O2HbHHb

Several world’s first & best SPADs– OpEx’15, IEDM’17, ISSCC’18, JSTQE’18, JSTQE’19, etc.– World’s best 3D IC SPAD (JSTQE’18) & SPAD-LiDAR Sensor (ISSCC’18)

P-epi

P-sub

3D connection

P+PW PW

NW NWN+N+

< 3 µm

Top

Bottom

3D IC SPAD

Max 430m

< 0.4% error

LiDAR Results

QuantumApplications

Dark Count Rate (DCR)Triggered by any generation

of free carriers instead of incident photons

Photon Detection Prob. (PDP)Probability of detection when a photon hits the

SPAD’s active area

6.8 kcps55.4 cps/μm2

31.8 % @600nm

Lowest DCR Highest efficiency Broadest spectrum

Best timing jitter

Timing JitterUncertainty of Δt between

a photon and a photon-generated avalanche pulse

Results- State-of-the-art

V. C. Coffey, Photonics Spectra, 2014

J. L

. O'B

rien,

Sci

ence

, 200

7

Uni

v. N

ice

Quantum Key DistributionOptical Quantum Computing

• High-performance CMOS-APDs can play an important role in optical-interconnect applicationsby enabling high-speed integrated optical receiver based on standard CMOS technology.

• High-performance linear-mode CMOS-APDs can also be exploited in various image applicationsrequiring high sensitivity and fast response.

• SPAD-based single-photon counting & time-of-flight sensors have received a great amount ofattention by scientific & industrial communities for a wide variety of applications. Strong impact on next-generation LiDAR & Bio & quantum applications with great scientific/economic/social potentials.

• Many SPAD-related researches/projects/collaborations are in progress. Ground breaking results are coming soon!

IDQ