Mesoscale Bulk Electronics

1
Mesoscale Bulk Electronics
2
Beyond the MOSFET

• Mesoscale
• An intermediate scale, on the order of 10 nm,
• Materials have some properties of bulk material,
• But surface effects are important,
• And more quantum phenomena become important
• Bulk
• Materials structures fabricated using bulk
  processes, w/o atomic precision
• Electronics
• Electron states are used for primary
  information-processing operations
• not photons (optical), or whole atoms (mechanical)

3
What happens _at_ mesoscale?

• MOSFET scaling hampered by quantization of
• charge
• becomes important _at_ L ? 10 nm in all materials
• energy levels
• important in semiconductors _at_ L ? 10 nm
• Can alternative device operating principles
  exploit these quantization effects rather than be
  hampered by them?
• Some approaches
• Single-electron transistors
• Quantum wells / wires / dots, quantum-dot CAs
• Resonant tunneling diodes / transistors

4
Coulomb blockade effect SET

• Based on charge quantization
• of energy levels may not be noticeably
  quantized
• Comprised of an island of (typically) metal
  surrounded by insulator.
• Narrow tunnel junctions to transistor
  source/drain - ? 5-10 nm typical
• Gate controls of electrons that may occupy
  island
• within precision of 1, out of millions

5
Quantum Wells/Wires/Dots

• Usually use semiconductor material.
• Electron position is narrowly confined in 1, 2,
  or 3 dimensions, respectively.
• ?E between distinct momentum states becomes large
• In quantum dots, total of mobile electrons may
  be as small as 1!
• Non-transistorlike quantum-dot logics
• Most notably, quantum dot cellular automata by
  Notre Dame group

6
Quantized Energy Levels

• The narrower the space,
• the smaller the ?? gap between normalmodes n
  and n1
• the larger the frequency energy gapbetween
  those modes
• More confinedspaces have widerenergy gaps
  betweentheir distinctmomentum states.

?/3, 3E
?/2, 2E
?, E
7
Resonant Tunneling Diodes

• Usually based on quantum wells or wires
• 1-2 effectively classical degrees of freedom

Source
Drain
Island (narrow bandgap)
Tunnel barriers (wide bandgap)
Electron tunnelsthrough barrier
Quantized momentum state
Electron flow
Unoccupied states
Occupied states inconduction band
Energy
8
Resonant Tunneling Transistors

• Like RTDs, but an adjacent gate electrode helps
  adjust the energy levels in the island

Gate
Source
Drain
9
Future Semiconductor Structures

• SOI (Silicon-on-Insulator)
• Band-engineered transistors
• Vertical transistors
• FinFETs (Chenming Hu group _at_ Berkeley)
• Double-gate transistors (e.g. Philip Wong IBM)
• Multi-layer chips (Lee _at_ Stanford)
• Quantum FET analysis (Merkle 93)
• atom-width wires (need ref)

Go through ITRS presentation
10
Nanoelectronics Technologies

• Scaled MOSFET structures - prev. slide
• Quantum wells/wires/dots - covered last time
• quantum dot cellular automata - go thru website
• Various single-electron devices - today
• Spintronics - electron (/or nuclear?) spin
  based electronics- today
• Molecular electronics - today or Friday

11
Quantum Dot Cellular Automata

• Wires x vs , fan-out, wire-crossing
• Speed 2 ps/cell
• compare light can go 0.6 mm in 2 ps
• ordinary electronic signals 0.3 mm
• MOSFET gate delay according to ITRS 99
• 11 ps in 99, 5.7 in 05, 2.4 in 14
• Gates inverters, majority gates, full adder
• Paradigms
• ground state computing
• clocked QCA pipelining (adiabatic, reversible)
• Molecular version 20 fs/cell (100x smaller)

12
Spintronics
UF contacts Arthur Hebard,Jeff Krause

• Cf. Das Sarma group at UF
• Info written into spin orientation of electrons
• persists for nanoseconds in conduction es
• compare 10 fs lifetime for momentum decay
• Spin control, propagation along wires, selection,
  detection
• Datta-Das and Johnson spin-based transistors
• Potential medium for quantum computation

13
Molecular Electronics

• Tour wires
• Molecular switches
• Carbon nanotube devices

14
Helical Logic
See plastictransparencies,readingsfor details

• Proposal by Merkle Drexler 96
• Do w. conductors insulators only!
• no fancy semiconductors, superconductors, or
  tunnel junctions needed...
• The wires are the devices!
• Uses simple Coulombic repulsion between
  electrons to do logic
• Scalable to single electrons atom-wide wires!
• Externally clocked...
• by rotation of CPU within a fixed electrostatic
  field
• Can be used reversibly 10-27 J, 1K, 10 GHz!

15
HL Overall Physical Structure

• Consider a cylinder of sparse (high-permissivity)
  insulating material (e.g., air), containing
  embedded helical coils of cold conductive or
  semiconductive wire, rotating on its axis in a
  static, flat electric field (or, unmoving in a
  rotating field).
• An excess of conduction electronswill be
  attracted to regions on wire closest to
  fielddirection.
• These electron packetsfollow the field along as
  itrotates relative to thecylinder.
• Next slide Logic!

16
Switch gate operation 1 of 3
Datawire
Conditionwire
17
Switch gate operation 2 of 3
Datawire
Coulombicrepulsion
Conditionwire
18
Switch gate operation 3 of 3
Datawire
Conditionwire
19
Nano-mechanical logics
See plastictransparencies,readingsfor details

• First proposed by Drexler, 1992 ( earlier)
• Typically, very low leakage!
• due to high energy barriers (mechanical rigidity)
  in interactions involving bonded atoms, vs. just
  electrons
• Pretty fast due to small size, but probably...
• 1000s slower than molecular electronics might
  be
• basically, because atoms are 1000s heavier
  than electrons
• Drexlers logic of rods, cams, springs
• Molecular scale components
• Covalently bonded, atomically precise
• Merkles (1993) buckling logic
• No sliding-contact interfaces
• Scalable from macroscale to mesoscale

Also seeSmithsplanarmechanicallogics
20
Molecular Electronics
See plastictransparencies,readingsfor details

• Tour wires
• Various molecular switches
• Various carbon nanotube devices
• Potential problem areas
• High resistance of existing molecular devices.
• Maintaining thermal reliability in face of low
  node capacitances and voltages.
• High leakage currents, due to tunneling or
  thermal excitation over small, narrow barriers.

21
Biochemical computing

• Selected points on DNA computing
• Adlemans experiment
• Cyclic Mixture Mutagenesis
• Reversible DNA Turing Machines
• Seemans self-assembling structures
• Winfrees tile self-assembly logics
• DNA computing has many disadvantages
• High cost of materials
• Slowness of diffusive molecular interactions
• Slowness/cost/unreliability of lab steps
• Prob. wont ever be a cost-effective computing
  paradigm (except maybe for in vivo apps)

Seereadingsfor details
22
Optical computing

• Not viable at the nanoscale anytime soon!
• Due to entropy density issues mentioned earlier
• High enough info. flux requires extremely
  energetic photons, with too-high effective
  temperatures
• Or, waveguides considerably smaller than photon
  wavelengths - EMF theory suggests Impossible!
• All-optical computing requires nonlinear
  interactions, between photons materials.
• Optics (or more generally, EMF waves) will remain
  useful for communications, but only
• in contexts where extreme bandwidth density is
  not required (or extreme temperatures can be
  tolerated)