Designing Nanoscale MaterialsLecture Series by 2004 Debye Institute Professor
Christopher B. MurrayIBM Research
Ornstein Laboratory 166Office phone 253 2227
Lecture Series: Designing Nanoscale Materials.
(1) Why smaller, is different; finite size effects & implications for Tech scaling.
(2) General nanoparticle production.
(3) Semiconductor nanocrystals (Quantum Dots)Part 1:
(4) Semiconductor nanocrystals (Quantum Dots), Part 2:
(5) Nanowires
(6) Nanostructured magnetic materials for IT
(7) Nanomagnetics for biotech & beyond.
(8) Self-assembled nanocrystal superlattices:
(9) Binary nanocrystal assembly a route to multifunctional nanomaterials.
(10) Nanoporous materials:
(11) Ethics, issues and emerging trends for nanomaterials research.
Designing Nanoscale Materials
1+2 Wed. Sept 08 10.00-12.303+4 Wed. Sept. 15 10.00-12.30 5 Mon. Sept. 20 11.00-12.306+7 Wed. Sept. 29 10.00-12.30 8+9 Wed. Oct. 06 10.00-12.30 10+11 Wed. Oct. 13 10.00-12.30
Why smaller, really is different:Finite size effects in nanomaterials their implications for scaling in conventional technology.
Christopher B. MurrayManager Nanoscale Materials & DevicesIBM Research
Balancing the investments in Nanotechnology:
Basic/Strategic Research
Extending theEstablished
Technologies
Exploring Alternative/Disruptive
Technologies
ImmediateImpact
Long-TermImpact
Three Questions
What is nanotechnology?
Why is nanotechnology the future of information technology?
How will we manufacture at the nanoscale?
Nanotechnology is …
… research and technology development at the atomic, molecular or macromolecular levels, in the
length scale of approximately 1 – 100 nm …
National Science Foundation
A Unique Period in History
after C. Ausschnitt, Microelectronic Engineering, 41/42 (1998) 41- 460.0001
0.001
0.01
0.1
1
10
100
1000
1800 1850 1900 1950 2000 2050 2100
Min
imum
mac
hine
d di
men
sion
(micr
ons)
Moore’s Law (1965)
90 nm Manufacturing (2004)
193 nm immersion
2004 commercial niche lithographies
2004 best lab practice
industry roadmapEUV
C om puting : W here D o W e G o From H ere?
i10 mi100 nm
N anoscale S cience
and Techn o logy
Decreasing Costs of Computation
Source: Kurzweil 1999 – Moravec 1998
Constant Field Scaling
RESULTS:Higher Density: α2
Higher Speed: αLower Power: 1/α2
per circuit Power Density: Constant
L xd
GATEn+ source
n+ drain
WIRINGVoltage, V
W
p substrate, doping NA
tox
SILICON WAFER
L/α xd/α
GATEn+
sourcen+ drain
WIRING
Voltage, V / α
W/α
p substrate, doping α*NA
tox /α
SILICON WAFER
Silicon Transistor - Already A NanodeviceLogic
TSi=7nm Lgate=6nm
Source Drain
Gate
• Power4 Chip
• 174 million transistors• Gate length = 6 nm
B. Doris et al., IEDM , paper 10.6, 2002. J. Warnock et al., IBM J. R&D, p. 27, 2002
Size dependent phase transformationThe transition from c49 to c54 TiSi2Result in a
Phases formed by heating of a 10nm Ti film. As-deposited, the grain size is ~10nm (annealing at low Temp small would yield TiSi. The TiSi2 C49 phase appears at 700°C while C54 forms at 850°C.
Silicon Transistor - Already A NanodeviceMemory
• 512Mb DRAM prototype for 1Gb and beyond
• 110 nm DRAM, 8F2
S. Wuensche et al., Symp. VLSI Technology, 2002 H. Akatsu et al., Symp. VLSI Technology, p. 52, 2002
The conventional silicon field-effect transistor is still rapidly advancing, with potential materials innovations such as …
channel
insulator
silicon substrate
insulator
Gate
Source Drain
Metal gate electrodes
High-dielectric constant gate insulators
Ultra-thin (2 – 20 nm)Si or Ge on insulator
High-electron-mobilitysubstrates (strain or orientation)
Si epitaxy on oxideEpitaxial Growth of Semiconductors on Crystalline Oxides
x-ray reflectivity
0.00 0.01 0.02 0.0310-6
10-5
10-4
10-3
10-2
10-1
100
qz (nm-1)
refle
ctiv
ity
simulation........ experiment
Ge epitaxy on oxide
Bojarczuk, Guha et al., Appl. Phys. Lett. V83, 5443-5, (2003)
Double-gate Transistor (FinFET)
Poly-Si
Tsi=20nm Tox=1.6nm H=65nm
BOX
TEOS
H
Tsi
TEM
Tox
current-carrying surfaces
Cross-section of 60 nm channel length FET
• Scalable to the smallest channel length• World-record double-gate FET device performance
“gate delay” = 0.92 ps
single-electron charging energy
r r2d C ~ 2πε0εr[ln(r/d)] d<<r
~ 1.3 aF
2-d hcp lattice, each nanocrystal has 6 nearest neighbors (nn):
Cnn = 7.8 aF
to charge nanocrystal with a single extra electron:
Ec = ~ 10 meV e2
2Cnn
Coulomb energy dominates below ~ kBT=Ec/2
T~ 60Ko
Chuck Black, Bob Sandstrom, Chris Murray, Shouheng Sun
Coulomb Blockade Effects:
Electronic Properties of Semiconductor and Metal Nanoparticles
ε
a
Charge not completely solvatedas in infinite solid
aC oεπε4=Nanoparticle capacitance
)(2
2
aCeEc =Charging Energy
10 nm Al NC
Courtesy of C. T. Black, Thesis, Harvard U.
Coulomb blockade atkBT<Ec
Structure from discrete electronic states of metal NC
STM Measurements on Single QDs
InAs QDs
U. Banin et al. Nature 400, 542 (1999).
Parameters in NC arrays
Transport – explained from Middleton-Wingreen(M-W) model
transport in linear and square arrays (how ideal?)
Achieving high operational T.Fabrication method affects
– Size of particles
– Monodispersity
– Number of particles responsible for transport
– Dimensionality
– Homogeneity of array
Self assembly:single layer 2-D array of Au crystals
NC size : 2.2 - 2.9 nm, Interparticle distance s1-2=0.85nm, 1.2 0.1nm.(Threshold voltage)VT ~ 10VT independence! (12, 48, 77K)
e2/Cmax>kbT Global structural disorder (topology)Local structural disorder (voids,interparticle distance)Local charge disorder (e.g substrate,..)
R. Parthasarathy et al., Phys. Rev. Lett.87, 186807
shortest current path ~ 8 nanocrystals
Chuck Black, Bob Sandstrom, Chris Murray, Shouheng Sun
-400
-200
0
200
400
I (pA
)
-0.4 -0.2 0.0 0.2 0.4V (V)
T = 70 KT = 2 K
GV=0 follows simple thermal-activation
10-3
10-2
10-1
100
101
102
GV
=0
(1/GΩ
)
80x10-3
604020
1/T (K-1
)
data fit by: ln(GV=0) = const. - Ec/kBTfrom fit to data, measure EC~ 10 meV
for all devices measured, 10 meV < EC < 14 meV
1.00
0.98
0.96
0.94
0.92
R/R
H=
0
-0.4 -0.2 0.0 0.2 0.4
applied field (T)
HH
Spin-dependent tunneling in Nanocrystal arrays
Phase separation of block copolymersto form columnar
arrays
Directed Self-assemblyExperimental Silicon Memory Device
The process is then used in fabricating
an exploratory silicon memory
deviceSource: C. Black, K. Guarini, IBM
Goal: Incorporate Nanoscale Components in IT Systems
Porous Dielectric for On-Chip Wiring
Poromer (dendritic polymer)
Ultra Low K Dielectrics
Basic Physics of Semiconductor Quantum DotsC. R. Kagan, IBM T. J. Watson Research Center, Yorktown Heights, NY
HighestOccupiedMolecularOrbital
LowestUnoccupiedMolecularOrbital
ConductionBand
ValenceBand
EnergyGap
Bulk Semiconductor Quantum DotLike a
Molecule
Quantum ConfinementLow Dimensional Structures
( )( )n
c EEE
−∝
1ρ( ) tconsEc tan=ρ ( ) ( )nc EEE −∝ δρ( ) ( )Cc EEE −∝ρ
Particle-in-a-Sphere
is a spherical harmonic( )φθ ,mlY
( ) ( ) ( )rYrkj
Crm
llnl φθφθ
,,, ,=Φ
is the lth order spherical Bessel function( )rkj lnl ,
ak ln
ln,
,α
=
2
2,
22,
2
, 22 ammk
Eo
ln
o
lnln
α==
a
0
∞
Pot
entia
l V
r
1s
2s
solutions givehydrogen-like orbitals with
quantum numbersn (1, 2, 3 …)l (s, p, d …)
m
Discrete energy levels
size-dependence
Size Dependent AbsorptionExample: CdSe
Energy (eV)1.5 2.0 2.5 3.0 3.5
Abs
orba
nce
(arb
itrar
y un
its)
Energy (eV)1.5 2.0 2.5 3.0 3.5
Absorbance (arbitrary units)
17Å
150Å
17 Å
21 Å
29 Å
33 Å
45 Å
55 Å
72 Å
90 Å
150 Å
Real Band Structure
Example: CdSe
E
Eg
k
Cd 5s orbitals2-fold degenerate at k=0
Se 4p orbitals6-fold degenerate at k=0Introduces splitting of bands
hh
lh
so
heavy hole
light hole
spin-orbit splitoff J=1/2
J=3/2∆so
∆cfcrystal field splitting
J=1/2
J = L + S where L=orbital angular momentumS=spin angular momentum
J good quantum number due to strong spin-orbit coupling
Metal Nanoparticles
MetalParticle-- -
-------
- ---
-------- -
Surface Plasmon Resonance
• dipolar, collective excitation between negatively charge free electrons and positively charged core
• energy depends on free electron density and dielectric surroundings
• resonance sharpens with increasing particle size as scattering distance to surface increases
Au nanoparticle absorption
Antiferromagnetically-Coupled media
Three-atom-thick layer of Ru sandwiched between two magnetic layersExpected to increase current areal density limits to surpass 100 gigabits/inch2
Magnetic thickness = Mr t Magnetic thickness
= (Mr t)top - (Mr t)bottom
AFC ( cont. )AFC Media
MRAM Technology
Schematic of two MRAM cells
MRAM cell cross-sectionin 0.18 µm technology
Magnetic tunnel junction device and electrical characteristic
Mag
neto
resi
stan
ce (%
)
Applied Field (Oe)
M2
MT
M1
MTJBit
Line
Write Word Line
MT
128kb test chip
MRAM potentially has attributes of a universal memory: fast, dense, nonvolatile, radiation hard
CatalysisAu nanoparticles supported on TiO2 substrates show high activity for oxidation of CO at room temperature and below.
Reaction proceeds at corner, step, and edge sites of Au
TiO2 Support
Oxygen Adsorption (on TiO2)
CO adsorption (on Au) 3.5 nm Au nanoparticle
12 Atoms in length
2-3 Atoms high
Haruta, M.; Date, M. Applied Catalysis A: General 2001, 222, 427-437.
Bimetallic Catalysis
CH2=CH-CN + H2O CH2=CH-CONH2
Reaction proceeds most favorably with Pd-Cu particles, and is 100% selective when using a 3:1 Cu:Pd ratio.
CH3-CH-CN
OH
Geometric effects lead to higher activity and selectivity for certain reactions.
Figure taken from: Toshima, N.; Yonezawa, T. New Journal of Chemistry 1998, 1179-1201 and references therein.
Two visions of nanofabrication…
“Old”
Top down
Lithography
Digital
Depend on Low Error Rates
Molecular Assemblers
“New”
Bottom up
Chemical Synthesis
Analog
Tolerate High Error Rates
Self-Assembly
Allowing a few components to approach equilibrium will produce only simple structures …
Synthesis
Reagents
“Guiding” or “directing” the assembly process:Semiconductor Nanocrystals
Size ProcessingSynthesis Film Growth:Self-Assembly
Nanocrystal Superlattice
Reagents
Multi-Component Nanocrystal Superlattices
F. X. Redl, K. S. Cho and C. B. Murray
Silicon Nanowires: In-situ Observation of Growth
112
110111
dark-fieldimage
112
Si2H6
heated substrate
110
111
viewing direction
Frances Ross, IBM Research
Si nanowire growth
showing wire and drop geometry, facet formation and tapering to termination
Frances Ross, IBM Research
Beyond the next transistor: Exploratory Memory
SL m-1
SLm+1
SLm
WLn-1
WLn+1
WLn
Everyone is looking for a dense (cheap) crosspoint memory.It is relatively easy to identify materials that show bistable hysteretic behavior (easily distinguishable, stable on/off states).
Relative Maturity of Nonvolatile Memory Technologies
02468
101214161820
ProductSampling
Development Single CellDemo
Charts, NoParts
Tec
hn
olo
gy
Ch
amp
ion
s (C
om
pan
ies)
Smaller projects are also exploring non-volatile memory based on …
Perovskites
Chalcogenides
Organic materials
Beyond the Next Transistor“Millipede” Storage
How will we manufacture at the nanoscale?
Carbon Nanotubes?
STM Image
Ti Ti Ti
Al AlAppl. Phys. Lett. 80, 3817 (2002)
dox=15nm
Carbon Nanotube Transistor
Comparison with silicon
transconductance
threshold voltage
channel length
gate oxide thickness
p-MOSFET a) p-CNFET
50nm
1.5nm ~15nm
2300mS/mm650mS/mm
-0.2V -0.5V
subthreshold slope 70mV/dec 130mV/dec
IOn/IOff ~106106 - 107
260nm
drive current 2100mA/mm650mA/mm(Vg-Vt=-1.0V)
a) R. Chau et al. Proceedings of IEDM 2001, p.621
VS<VG<VD
VS
VD
VG
Nanotube Infrared Emitter
Nanotube Technology ?
CNTAu
100µm
How do you get from here to there?
Plenty of room for improvement !
No new architecture !
Establishing a Technology
Understanding:Electrostatics, electrodynamicsScalability (ballistic? contact-dominated transport ?)Contacts, dopingGate insulator, interface traps?High yield, selective growth/synthesis of nanotubes with correct electrical properties (single-wall, diameter, chirality)
Engineering:Device structure with minimized parasitic resistance and capacitanceFabrication processes leading to high device density (e.g. size of contacts commensurate with gate length, means to connect one device to another)Demonstrate device/circuits which satisfies ALL performance metrics (not just some metrics)Manufacturing tools and infrastructure, integration with siliconReliability...
Molecules = Small ?
L >2.5 – 3 nm
Si FET Molecular Device
TSi=7nm Lgate=6nm
Source Drain
Gate
B. Doris et al., IEDM , 2002.
All devices are governed by electrostatics and eventually limited by tunneling- difficult to be much smaller than 2 - 3 nm
Building Molecular Structures to Study the Science
Si
SiO2
+-
+ + +- - -
electrolytesolution
pipette orneedle
also workingelectrode
Vg
Electrochemically GateMolecular Junction
Pt coating
TiAu
Allimit assembly to electrode
sidewall
n+ Si
SiO2
Source Drain
R R R R R R
R' R' R' R' R' R'
Gate
Lipid-like membrane– Self-assembled at air-water interface
– Langmuir-Schaeffer transferIn-situ polymerization
– conjugated chain (schematic)
– wide band conductor
– end-to-end channel Hydrophobic binding - gate insulator
Chemistry to Covalently Bind Molecules to Substrates
Choose Length of alkyl chainDepending on desired function
C
Si
C
SiO OO O O
C
Si
C
SiOO O O
C
Si
C
SiOO O O
C
Si
C
SiOO O O
OOH
OOH
OOH
OOH
OOH
OOHO
OHO
OH
SiO2
Si (intrinsic or doped)SiO2
Si (intrinsic or doped)
SiSiO OO O O
SiSiOO O O
SiSiOO O O
SiSiOO O O
SiSiO OO O O
SiSiOO O O
SiSiOO O O
SiSiOO O O
SiO2
Si (intrinsic or doped)
ORCH2
Si
CH2
SiO OO O O
CH2
Si
CH2
SiOO O O
CH2
Si
CH2
SiOO O O
CH2
Si
CH2
SiOO O O
OH OH OH OH OH OH OH OH
SiO2
Si (intrinsic or doped)
oxidation
reduction
Demonstratingnow
esterifactionwith molecule
having alcoholfunctionality
esterifactionwith molecule
having carboxylicacid functionality
SiO2
Si (intrinsic or doped)
= molecule of interest
C
Si
C
SiO OO O O
C
Si
C
SiOO O O
C
Si
C
SiOO O O
C
Si
C
SiOO O O
OO
OOO
OO
OO
OO
OO
OO
O
Water Subphase
OH
OH
OH
OOH
OOH
OOH
OOH
OC C C C C
O OH O OH O OH O OH O OH
C
O OH
UV
"backbone"
Water Subphase
SubstrateSubstrate
Substrate
(1) (2) (3)
“Layer-by-Layer” Growth of Conjugated Molecules
• Grow long conjugated molecules that would otherwise be insoluble to span gap between electrodes• Combine different molecules or oligomers for functionality
Au
SS S S S S S S S Br
O O O O
SS S S S S S S S Br
O O O O
SS S S S S S S S Br
O O O O
Electron-donating/Electron-accepting Reduction-oxidation active centers
AuN N
N NZnS SS
BrS
S
+Au
SS
S S SBr
SS
S S SBr
SS
S S SBr
Tailor end functionalityto assemble on oxide, metal, or semiconductor surface
Au
SS
SBr
SS
SBr
SS
SBr
+ SR3Sn
SAu
SS
S S S
SS
S S S
SS
S S S
NBS
“Double FET”with floating electroden+ Si
SiO2
SS
S S S
SS
S S S
SS
S S S
SS
SSS
SS
SSS
SS
SSS
growmetal
Ligand to bind to desired
substrate surface
Addmetal-metal
Addligand
Addmetal-metal
Addligand
RhRhO
NN NN
OO O
Rh RhO
N NN N
O O O
N
N
RhRhO
NN NN
OO O
Rh RhO
N NN N
O O O
N
N
N
N
N
N
N
NSi
Si
Si
RhRhO
NN NN
OO O
Rh RhO
N NN N
O O O
N
N
RhRhO
NN NN
OO O
Rh RhO
N NN N
O O O
N
N
NS
NS
Au
N
N
N
N
Layer 1 Layer 2
N N
N,N'-di(p-anisyl)formamidinate
NC
N
HMeO OMe=
• Tailor head group of ligand to bind to particular substrate surface
• Tailor end group to templatemetal-metal bonded unit
• Choose M-M bondM = V, Nb, Cr, Mo, W, Tc, Re, Fe, Ru,
Os, Co, Rh, Ir, Ni, Pd, Pt, Cu, Ag …
• Choose ligand to bridge M-M bonded units to tailor:
• Electronic coupling between dimetal units
• Electrochemistry• Solubility• Structure ….
Leq
Leq Leq
Leq
MLeq Leq
Leq Leq
M LaxLaxLayer-By-Layer Growth of Metal-Metal Bonded Compounds
Beyond the Next TransistorMolecular Cascade Logic
A.J. Heinrich, C.P. Lutz, J.A. Gupta, D.M. Eigler Science 2002
We are Just Getting Started!
Nanotechnology Definition Revised
The ability to design and control the structure of an object on all length scales – from the atomic to the
macroscopic – reliably and repeatedly in a manufacturing environment.