Transcript SiliconQuantumComput..

Silicon-based
Quantum Computation
C191 Final Project Presentation
Nov 30, 2005
Cheuk Chi Lo
Kinyip Phoa
Dept. of EECS, UC Berkeley
Silicon-based Quantum Computation:
Presentation Outline
I.
II.
III.
IV.
Introduction
Proposals for Silicon Quantum
Computers
Physical Realization: Technology and
Challenges
Summary and Conclusions
Introduction: Why Silicon?

We know silicon from years of
building classical computers

Donor nuclear spins are wellisolated from environment
low error rates and long
decoherence time

Integration of quantum
computer with conventional
electronics

Scalability advantages?
Introduction: DiVincenzo’s Criteria
1.
Well-defined qubits
2.
Ability to initialize the qubits
3.
Long decoherence time
4.
Manipulation of qubit states
5.
Read-out of qubit states
6.
Scalability (~105 qubits)
II. Overview of Silicon Quantum
Computation Architectures
Silicon Quantum Computer Proposals
Shallow Donor Qubits
Deep Donor Qubits
Electron Shuttling
Exchange Coupling
Magnetic Dipolar Coupling
Silicon-29 Qubits
Silicon Shallow Donor Qubits:
Qubit Definition and State Manipulation
Spin
Resonance
BAC
BDC
J-Gate
(Exchange
Coupling)
A-Gate
(Hyperfine Interaction)
Control gate
barrier
Silicon-28
Qubit
magnetic
dipolar coupling
BE Kane, Nature, 393 14 (1998)
AJ Skinner et al, PRL, 90 8 (2003)
R de Sousa et al, Phys Rev A, 70 052304 (2004)
S-Gates
(Electron shuttling)
Summary of Silicon Shallow Donor Qubits



Qubit: donor nuclear spin or hydrogenic qubit (nucleus +
electron spins)
Initialization: Recycling of nuclear state read-out + nuclear
spin-state flip via interaction with donor electron
Decoherence time: e.g. at 1.5K



Qubit Manipulation



nucleus spin T1 > 10 hours
electron spin T1 > 0.3hours
Single Qubit Manipulation: hyperfine interaction + spin
resonance
Multi-qubit Interaction: Exchange coupling, Magnetic
dipolar coupling or Electron shuttling
Read-out: Transfer of nucleus spin state to donor electron
via hyperfine interaction, then read-out of electron spin state
Physical Realization of a Si QC
Some common features that must be realized in
a shallow donor Si QC are:
Array of single, activated 31P atoms:
 Single-spin state read-out:
 Integrated control gates
 Process Variations

Formation of Ordered Donor Arrays
“Top-down”  single ion implantation
T Schenkel et al, APR, 94(11) 7017 (2003)
“Bottom up”  STM based Hydrogen Lithography
JL O’Brien et al, Smart Mater. Struct., 11 741 (2002)
Spin-State Read-out with SET’s &
Fabrication of Control Gates
Read-out: Spin state  Charge state (e.g. measurement by SET)
Read-out Challenges:
i. SET’s are susceptible to 1/f and telegraphic
noises (from the random charging and
discharging of defect/trap states in the silicon
host)
ii. alignment and thermal budget of SET’s with
the donor atom sites also present as a
fabrication challenge.
(UNSW)
Control Gate Challenges:

Qubit-qubit spacing requirements for different
coupling mechanisms:
 Exchange Coupling: 10-20nm
 Magnetic Dipolar Coupling: 30nm
 Electron Shuttling: >1m

State-of the art electron beam lithography:
 can do ~10nm, but not dense patterns
Qubit interaction control gates extremely
challenging!
(L Chang, PhD Thesis, EECS)
Process Variations
Process Variations may arise from:
i.
substrate temperature gradient,
ii. uneven reagent use during fabrication,
iii. differences in material thermal expansion
iv. strain induced by the patterning of the substrate
(leads to uncertainty in ground state donor
electron wavefunction, due to incomplete mixing
of states)
Consequences:
i.
Need careful tuning and initialization of
qubits
ii. Limit of scalability?
iii. Introduce strain in silicon intentionally?
•
lifts degeneracy of electronic state 
less vulnerable to process variations
(IBM)
Silicon Deep Donors Proposal
Excited State
Bi
Er
Bi
Optical
Excitation
Ground State
Bi
Bi
Er
Bi
AM Stoneham et al, J. Phys.: Condens. Matter, 15 (2003), L447
Er
Bi
Initialization, Manipulation
and Readout?

Initialization by polarized light or injection of
polarized electron


Manipulation with microwave pulses


like the work by Charnock et. al. on N-V centers in diamond
Readout optically



both are not very possible under room temperature
detection of photons emitted
potentially require detection of single photon
Disorderness of donor ion

Irreproducibility and difficult to address qubits
Decoherence Time and
Thermal Ionization
Summary of Silicon Deep Donor Qubits





Qubit: deep donor (e.g. Bismuth) nuclear spin, proposed to
work at room temperature.
Initialization: Optical pumping or injection of polarized
electron, questionable in feasibility.
Decoherence time: fraction of nanosecond at room
temperature
Qubit Manipulation: via applying intense microwave pulse,
like N-V centers in diamond
Read-out: optical readout of photon emitted from transition
between two states
Silicon-29 Quantum Computer Overview
Manipulating qubits
by Dysprosium (Dy)
magnet
Readout using
MRFM CAI
Initialize with circularly
polarized light
NMR-type quantum
computer
TD Ladd et. al. , PRL, 89(1) 017901, 2002
Decoherence Times

Long decoherence time (T1 and T2)
Reported T1 as large as 200 hours,
measured in dark
 Experimentally find T2 as long as 25
seconds
 T2 is reduced by the presence of 1/f
noise due to the traps at lattice
defects and impurities

Summary of Silicon NMR quantum
computer





Qubit: Chains of silicon-29 isotope for ensemble
measurement
Initialization: Optical pumping with circularly polarized light
Decoherence time: measured as long as 200 hours in dark
at 77K for T1 but only 25 seconds for T2
Qubit Manipulation: combination of static magnetic field
and RF magnetic field
Read-out: with cantilever, performing MRFM CAI
Problem:
RF Coil, Dy Magnet & MRFM
The deposition method of Dy
magnet is not outlined! It won’t
be trivial to incorporate
The cantilever tip for MRFM is
not included in the schematic.
How to insert it?
TD Ladd et. al. , PRL, 89(1) 017901, 2002
Summary and Conclusions





Several proposals for implementing quantum computer in silicon
 Shallow donor (phosphorus) qubit
 Deep donor (bismuth) qubit
 Silicon-29 NMR quantum computer
Difficulties faced in each proposals
Arguments on the feasibility
Most experimental efforts are on shallow donor qubits
Convergence with conventional electronics processing
requirements:
 Currently: 90nm technology node (~45nm features)
 22nm technology node in 2016!
 Strained-silicon: hot topic of research in semiconductor industry
 Narrower transistor performance window with ordered dopants
 Single-electron transistors and other nanoelectronics
(http://www.ITRS.net)
Thank You
Thank You!