International School ‘Quantum Metrology and Fundamental Constants’,

Les Houches, October 2007

Fundamental metrology in the

future: measuring the single
quantum?

John Gallop
NPL & Imperial College
 Outline of Talk

Now more than 50 years of progress in redefining the base units of the
SI in terms of fundamental constants. The job is almost complete…?

BUT new physics and technologies are always emerging and these
will require developments and modifications to the SI.

Metrology is evolving from averages over ensembles (e.g. pressure of
a gas, temperature of a sample of water, magnetisation of a ferromagnet,
power in an electromagnetic wave) to include measurement on single
entities.

Existing examples: single atoms, ions, photons

Future examples involve condensed matter: single spins, single
microwave photons, phonons? Temperature of a single atom?
Temperature of a single mechanical vibrational mode?

New physical quantities will be needed. What are the units of spin
current?
 Functional Metrology at the Nanoscale

Josephson Flux quantisation

um junctions Superconductivity
qubits
nt

Super- Magnetic
ua

spintronics
paramagnetism
Q

Meta- Optics THz Electromagnetism

materials
Coulomb blockade
Nanoscale electrical Continuous Electric
conductance current
Nanoscale thermal Heat capacity Thermal
conductance
Single phonons
Molecular NEMS MEMS oscillators
al Mechanical
vibrations
sic
Rapidly increasing s Biological
proteins DNA complexity
Cells la
C

0.1 1.0 10 100 103 104 105

Length scale (nm)
Indicates functionality where traceable metrology is
required and not currently available
Note the ‘gap’ (white strip) between quantum and classical
behaviour occurs at different length scales, depending on the
functionality. For ‘harder’ functions the gap exists at larger
scales (e.g. superconductivity), whereas for ‘softer’ functionality
(e.g. biology) the gap is close to the atomic scale.
Iconic Developments

NEMS
resonator
Quantum GHz
mechanics QIP resonators
applications
Low D
nanoSQUIDs
Nano- carbon
Single spin thermometry
detection Spin Hall effect
Spintronics
Quantum Metrology: the Cambrian
explosion (~2000-2015…?)
The fossil record of
the Burgess shale SNS/NSN Phase slip
(500My bp) shows a source
turnstile
sudden explosion of
new & exotic types JAWS
SET 2e
of flora and fauna. pump pump
Most did not survive. Sluice

But a few became

very successful….

There may be similarities with the situation in condensed matter QM.

Lots of exotic new devices.
Not all will survive. But a few will become very successful……..
SQUID: Superconducting Quantum
Interference Device A SQUID is a highly sensitive
superconducting electronic device.
Superconducting
ring
It combines:-
Josephson junction

Flux Quantisation

Josephson junction

The maximum response frequency of a

SQUID is set by the Superconducting
SQUID ring
Energy Gap ∆

2∆ ∼ 4kBTc
f ~ 2∆/h ~ 1-10 THz
It allows very small changes in
magnetic field to be measured
dBmin~ 1-10 fT/(Hz)1/2 δB~ 1-10 fT /Hz1/2
Nanoscale Metrology with SQUIDs

SQUIDs have been called ‘the most sensitive measuring

devices for any physical parameter’
This may be too good to be true….but SQUIDs are certainly the most
sensitive measuring devices for a wide range of physical parameters:
Magnetic Flux Flux density
Rotation Voltage
Magnetisation Magnetic Resonance
Temperature Energy
Quantum Limits & QIP Displacement

SQUID are evolving: SQUIDs at the nanoscale

Energy Sensitivity of SQUIDs
Thermal noise in a SQUID limits energy sensitivity, due
to voltage and current noise in the shunt resistances of
the Josephson junctions.
The limiting flux noise spectral density Sφ may be written:

S φ2
εn = = 16k B T ( LC ) 1/ 2

2L
where εn is minimum detectable energy change, C and L
are the capacitance and inductance of the SQUID and T is
temperature.
So εn can be improved by reducing C or L (as well as the
more obvious T). That means NanoSQUIDs !!
Approach to Optical-lithographically
fabrication
nanoscale- SQUIDs
Conventional SQUIDs fabrication:

•Tri-layer junctions
•Optical lithographically fabrication 3 micron
µm
•Size limited to > 2µ loop

•Nano SQUID loops require small Josephson junctions.

•These are hard to make in trilayer (C high, Jc low)
•The low tunnel current density requires large areas.
• Microbridges based on single layer can provide sub-
micron and nanoscale junctions with low C and high Jc
Approach to Nanometre sized FIB fabrication

nanoscale- SQUIDs Junction: 65 nm x 80 nm

NanoSQUIDs fabrication
based on microbridge
junctions:
Loop:dia=371nm

E-beam direct write

lithography using SEM E-beam: Nanometre SQUIDS

FIB (focussed ion beam)

milling

SPM anodisation

Loop: 200nm
Noise performance of Nb nanoSQUID
• Measured magnetic flux noise spectral
density, expressed in units of the flux
quantum Φ0, as a function of frequency
from 0.1Hz to 100kHz.
• Note there is a region at low frequency
where the noise spectrum has a 1/f form.
• Above 1 Hz there is a much weaker
frequency dependence.
• Even at 1Hz the spectral density is as
low as 0.8 µΦ0/Hz1/2
• In the white noise region around 1kHz
the flux noise is as low as 0.2 µΦ0/Hz1/2.

• The frequency roll-off at higher

frequencies represents the flux-locked Flux noise spectral density at T=6.8K,
loop bandwidth of the measurements. I = 80 µA, RI = 90 mΩ

• The SQUID is operated at 6.8K

Nano-SQUIDs & Quantum Metrology

As SQUID dimensions reduce towards nano-scale the
energy sensitivity improves

The magnetisation sensitivity improves (see later)

Various hard limits apply

These often relate to quantum limits, where the
continuous nature of the variable breaks down

Many synergies with quantum measurement and
quantum computation

Beyond these limits new devices, entangled hybrids of
SQUIDs and other nanoscale systems may emerge.

Condensed matter physics, quantum metrology &
measurement in general should benefit.
 Quantum Limited SQUIDs
The Josephson junctions are also characterised by a plasma frequency
ωp where ω = 1
p
Φ 0C / ic

At low temperatures if hω p ≥ k BT
this classical treatment breaks down and the back reaction noise
arising from circulating noise currents in the SQUID loop must be taken
into account. In this case it is found that
h
εn ≥
2
(agrees with the Uncertainty Principle!)
Experimentally the best SQUIDs have a minimum flux sensitivity of
around 1x10-7 Φ0/(Hz)1/2, limited by room temperature amplifier noise.
SQUIDs as Quantum Objects
Solid-state circuit Quantum
Electrodynamics (QED) is now
being demonstrated and
developed, using a number of
different technologies.
Atom QED techniques are M A. Sillanpää, J I. Park and R W. Simmonds, Nature v449 (2007) pp.438-42
exploited for quantum state
manipulation
A SQUID in the quantum limit
can itself act as a qubit, a
quantum object.
Recent results have
demonstrated 2 SQUID qubits
coupled via a superconducting
transmission line.
Further developments can be
expected…..
Nanomechanical Resonators

There is great interest in nano-mechanical oscillators: several types:-
 Diving board (cantilever)
 Guitar string
 Torsion bar with inertial mass

The frequency of vibration depends on length, stiffness and tension.

AFM cantilevers vibrate at a few kHz to a few MHz

Short stiff nanoscale tuning forks can vibrate at 10 GHz
(These modes are the lowest phonon modes of a beam)

How could we make use of them?

1/ 2
E t where E is Youngs modulus,
Mode frequency scales as f ~   ρ is density, t is thickness and
ρ
2
L L length
Euler-Bernoulli equation for beam
deflection (18th century)
The deflection u(x,t) of any point on the neutral axis of a beam under
the influence of a distributed force f(x,t) is described by the following
4th order differential equation where E is Young’s modulus, ρ the
density and I the moment of inertia.

∂ 2 u( x , t ) ∂ 4 u( x , t )
ρ + EI + f ( x ,t ) = 0
∂t 2
∂x 4

Four boundary conditions are required.

The solutions for the time independent eigenmodes form an
orthonormal set and any beam deflection can be written as a linear
superposition of these modes.

A measurement of the cantilever shape can tell us about the force
distribution.
Properties of a mechanical resonator
1/ 2
E t where E is Youngs modulus,
Mode frequency scales as f ~   ρ is density, t is thickness and
ρ
2
L L length

Another important parameter is the spring constant k. This is the force F

required for transverse displacement (assumes spring is in linear region)

dF
k=
dx
For a rectangular beam resonator, width w, the spring constant is
3
Ew  t 
k=  
4  L

As cantilevers reduce in size to the nanoscale it becomes more and
more difficult to accurately measure the dimensions.

Also E and ρ become less certain due to surface and edge effects.
Force Measurement at the Nanoscale

There are real problems with measuring forces accurately
at the nanonewton level and below.

Generally an Atomic Force Microscope (AFM) is used in
these situations.

A sharp tip on the end of a cantilever is functionalised and
brought close to a functionalised substrate to investigate the
force between the two functionalised molecules.

Typical forces between single biological molecules (such
as nanomotors) are only at the 1-100pN level.

The cantilever deflection is measured and that allows the
force to be calculated IF the cantilever spring constant is
known.
Cantilever spring constant by Radiation
Pressure
F
Static case: F = k Z (c.f. F = -kx)
k= F/Z

Dynamic case: F=kZ/Q

x (at resonance) k=FQ/Z
z

To measure spring constant, k, using a dynamic technique;

• apply a time-dependent force, F, at resonance frequency, ω0
• measure time-dependent displacement, Z, and quality factor, Q

Optical radiation pressure force is traceable to a fundamental constant, c;

2 PR where c = speed of light
F= PR = intensity of reflected light
c
Dynamic Method
• Intensity-modulated laser exerts a radiation pressure force on AFM
cantilever
• The time-dependent force causes the cantilever to bend to and fro
• Change in cantilever slope displaces CW laser spot on detector
• Signal from quadrant detector allows cantilever motion to be monitored

Modulated laser off Modulated laser on Quad

det
AFM CW AFM
Cantilever Cantilever
Mod
CW Mod
laser laser
Quadrant Quadrant
detector detector

CW
Frequency-dependence of
cantilever motion
Calculated frequency-dependence of cantilever motion
1.0 90
The cantilever’s complex
— amplitude, Z(ω)
motion close to resonance 0.8

Amplitude (arb units)

— phase, Φ(ω) 0
frequency, ω0, is given by:

Phase (degrees)
0.6
 ω0 2 / k  -90
z (ω ) =  2  F (ω )
ω
 0 - ω 2
+ i ωω 0 / Q 
0.4

-180
0.2

Amplitude is given by;

0.0 -270
ω0 / k
2

Z (ω) = F (ω ) 0 5000 10000 15000 20000

[(ω 0
2
-ω )
2 2 2 2
+ ω ω0 /Q 2
]
1
2 Frequency (Hz)

Phase is given by; if ω < ω0 then Φ(ω) = 0˚

Amplitude at resonance:  − ωω 0  if ω = ω0 then Φ(ω) = -90˚
Φ (ω ) = tan 
−1
2 
Z (ω 0 ) =
QF (ω 0 )  Q (ω 0
2
− ω ) if ω > ω0 then Φ(ω) = -180˚
k
when measured with respect to the driving force
Mechanical versus optical
excitation
Piezo excitation (i.e. mechanical) Modulated laser excitation (i.e. optical)
0.0012 180 0.0012 0

— amplitude — amplitude
0.0010 0.0010

— phase
0
— phase -90

Amplitude (Vrms)
Amplitude (Vrms)

Phase (degrees)
0.0008 0.0008

Phase (degrees)
-180

0.0006 0.0006 -180

-360
0.0004 0.0004
-270
-540
0.0002 0.0002

0.0000 -720 0.0000 -360

0 5000 10000 15000 20000 0 5000 10000 15000 20000
Frequency (Hz) Frequency (Hz)

• Piezo excitation is dominated by cantilever resonance, however other resonances

are also observed in amplitude and phase. Messy!
• Optical excitation produces a single resonance peak and 180º phase change,
however thermal effects are observed at low frequency
• For optical excitation, magnitude and position of applied force can be changed
 Calculation of cantilever spring
constant
FQ
At ω = ω0, spring constant, k, is given by k =
Z
2 PR PA where P = 2.00 mW, P = 0.11 mW
Radiation pressure force, F = + R A
c c
F = 11.9 pN
Quality factor, Q, is ratio of resonance frequency to peak width
Q = 10688 ÷ 158 = 67.6

Amplitude of cantilever oscillations, Z, is calculated from quadrant detector

signal and calibration data : Z = 2.66 nm

Hence spring constant k = 0.30 N/m

Estimated total error budget ≈ 10%

 Exquisitely Sensitive Mass Balance

How to (possibly) weigh a single molecule:

Molecule of H2O has mass of ~ 3x10-26kg

Best conventional mass balance ~ 10-11kg

But a mechanical resonator can be used
as a mass balance

For a fundamental vibration at frequency f,

adding a mass m at the end of a diving board
cantilever will shift the resonant frequency by
an amount δf/f α m/M
where M is the cantilever mass

But as the frequency increases the

amplitude of oscillation goes down….
 Quantum Mechanics of Mechanical
Systems
We are used to thinking about single
photons of light or X-rays. Or of single
phonons propagating heat.
But what about a quantised guitar
string?
This should be possible (and very
interesting)
From Peng et al., PRL (2006)
At a temperature T single quanta of
energy E = hf can only be seen against For a frequency of 440Hz this would
the background of thermal noise if require T < 1nK…… (very hard)
But if f ~ 1010 Hz you only need T~ 0.01K
k B T << E = hf How to make a very high pitched string?
Choose a very small, stiff string…..
Nanomechanical resonators as quantum
oscillators

Micro- and nano-mechanical resonators when
operated in high vacuum and at low temperatures
can have high resonant frequencies (ω/2π ~ GHz)
and high quality factors (Q ~104 –105 in vacuum).

Several recent reports have approached the
standard quantum limit in such resonator systems.

H = hω c a + a + hω m b + b − hga + a (b + b + )

The Hamiltonian couples an etalon (frequency ωc) with a MEMS resonator

forming one mirror of the etalon with resonant frequency ωm, coupling strength g

Many reports in the literature of optical fields being used to cool a MEMS by up
to a factor of 40.

Radiation pressure of an intense optical field impinging on a perfectly reflecting
vibrating mirror is able to entangle in a robust way the first two optical sideband
modes.

Under appropriate conditions, the generated entangled state is of EPR type
Marshall, W., Simon, C., Penrose, R. & Bouwmeester D, Phys. Rev. Lett. Vol. 91 pp.130401 (2003)

S Pirandola, S Mancini, D Vitali, P Tombesi, J. Optics B Vol. 5 pp.523-9 (2003)

Nanomechanical quantum resonators

First non-classical observation of mechanical resonator at T=110mK

and f = 1.4GHz Alexei Gaidarzhy, Guiti Zolfagharkhani, Robert L. Badzey, and Pritiraj Mohanty,
‘Evidence for Quantised Displacement in Macroscopic Nanomechanical
Oscillators’, Phys. Rev. Lett. Vol. 94 pp.030402 (2005)
Possible experiments:

Influence of gravity and gravity gradient fields on decoherence of
nanomechanical resonators.

Coupling superconducting qubits to quantum mechanical resonators

Entanglement of different eigenmodes of a quantum mechanical
resonator
How to excite and read-out smaller
cantilevers?
To make the resonant frequency higher (and the QM limit easier to
reach) the resonator linear dimension L must be reduced and the
stiffness parameter k increased.
This reduces the amplitude of the oscillation for a particular mode
So we need a particularly sensitive detector to measure this amplitude
A nanoSQUID perhaps ???……

Bdc
Irf
force

nanoSQUID cantilever
Schematic of Lorentz force excitation with SQUID readout
Simple analytic model of inductance
change with paddle angle

SQUID loop

axis of rotation

The hatched & shaded area represents the effective area of the SQUID
loop inductance. This varies with the angle φ between paddle and SQUID.

A SQUID with 300nm diameter, operated at 4K should be easily capable
of measuring the thermal noise in a torsional resonator fitting inside it.
Cantilever coupled to nanoSQUID

So now we could try to
combine a NEMS system with
the most sensitive detector, Junction
beam

a nanoSQUID!

The trick is to optimise the paddle
coupling between the two parts SQUID loop

Inductive coupling has proved

to be the lowest noise method.

The upper figure shows a Si3N4 torsional

oscillator coupled to a Nb microSQUID

The lower figure shows a 10nm diameter
carbon nanotube welded across an air bridge
200nm wide, between Si3N4 supports.
CNT across 200nm air-
bridge
Higher Order Mechanical Modes
Even for a simpler bar resonator there are
much higher order modes than the
fundamental.
For more complex shapes such has this
paddle the higher order modes can have
much higher frequencies.

How might this be used?

If a bio-molecule (protein or DNA) fixed to
the paddle changes shape the amplitude of
each mode will change.
In principle if enough modes are analysed
the shape change may be deconvolved.

What is sensitive enough to read out these very

small changes? nanoSQUID, rfSET, etc.
What is small enough and stiff enough to vibrate
at high frequencies, with low mass?…..
 Carbon Nanotubes
A single layer hexagonal lattice of carbon
atoms is ‘rolled-up’ to give a tube.
The chirality of the tube affects its properties.
Unit cell

Nanotube
axis

A
T = t1a1 + t 2 a 2 C
a1
O
C = na1 + ma 2 ≡ (n, m ) a2 h

(n,m are integers, 0<|m|<n)

One Dimensional Electron Transport

1D

Depends on θ and φ

Metallic to semiconductor

C=na1+ma2

“Zigzag”(n,0)
–Metallic if n/3 integer

“Armchair”(n,n)
–All metallic

“Chiral” (n,m)
–Metallic if (2n+m)/3
integer
Carbon Nanotubes

STM image of the structure of several parallel SWNT.

AFM scan of dispersed MWNTs on HOPG.
1.0

0.8
Normalised Frequency

0.6

0.4

0.2

0
0 1 2 3 4

Diameter (nm)

Distribution of MWNTs diameters. Solid line is a fit to a

AFM image of a SWNT rope imaged in non
normal distribution with mean of 1.9± 0.48 nm. contact mode on HOPG
Carbon Nanotubes
in Cryogenic AFM

AFM on
cold stage

Conductance scan of CNT ring. Dark

Cold
Finger areas have low resistance.

AFM at room and cryogenic temperature

with different environments (high vacuum,
variable gas atmospheres etc.).

Carbon nanotubes show

ballistic transport at RT Topography scan shows carbon
nanotube ring.
Quantised Electrical Conductance in a
CNT at Room Temperature
Assume a 1-D wire with SEM picture of Multiwall
adiabatically tapered non- CNT, protruding from a
reflecting reservoir bundle.
connections: Below: quantised
Schematic of 1-D wire: electrical conductance
plateaux
Reservoir 1 Reservoir 2
µ1 µ2 8

Since (µ1 - µ2 ) = -e(V1 - V2 ) 6

the resulting conductance G GT/Go 5

4
is:
2e 2
3
i
G= = 2

V1 − V2 h 1

0
-800 -600 -400 -200 0 200 400
z (nm)
 Phonon Spectrum of a CNT
• Low temperature thermal transport through a
nanotube Is due to four acoustic phonon modes
(one longitudinal, two degenerate transverse and
one twisting mode)
∞
hω ( k )ν ( k )(η hot ( k ) − η cold ( k ))T ( k )
I therm = ∫ .dk
0 4π 2

• It is possible to simplify this integral by

noting that the group velocity dispersion
relationship dω/dk between angular frequency
and wave number exactly cancels with the density
of states function dk/dω.

• Thus the integral may be transformed to being

over the frequency domain:
∞
k B2 x 2e x  T
G thrm =
h ∫(
0 e −1
x 2
Tm
)
xk B dx
h here x=hω/kBT.
Predicted Quantised Thermal
Conductance
In the limit of linear response, ∆T << T, the resulting expression
for thermal conductance depends only upon Tm(q), the mode
thresholds at q ‹ 0 (labelled qm), and fundamental constants:

π 2 k B2 (Thot Tcold )1 / 2
G th = x no. of populated modes
3h
opt
The lowest energy optical phonon mode has a frequency fminof
~ 3THz for a (10,10) SWNT. Thus for
opt
hf min
T << ~ 250 K
kB

only acoustic (massless) phonon modes are populated and the

onset of quantised thermal conductance could be observed.
Thermal measurements on CNTs
Laser
Mirror

Photodiode

Z-piezo

Resistance
themomete
Cantilever r
MWNT fibre Experimental set-
up: the z-piezo of
V
MWNTs the AFM is used to
lower the MWNT
RS fibre onto the
HOPG HOPG substrate.
V
 Ballistic Phonons & Electrons in CNTs at RT
Correlation between 1200
thermal and electrical 299 K
264 K
228 K
conductance steps for 299 K
264 K
MWNTs on HOPG for a 800 228 K
192 K

number of different CNTs

Gth
with a range of
temperature differences. 400

Evidence for ballistic

phonon conduction…. 0
0 1 2 3 4
Quantised thermal G0
conductance expected
below 30K. Modelling:
The trend lines show calculated ratios of Gth to G0 for both
armchair (long-short dash line) and zig-zag tubes (dash line) at the
As we go to lower T will true extremes of the diameter range observed for our tubes (0.9 and
quantum effects emerge? 3.3 nm respectively). Solid line is linear fit to all the measured
data, yielding a slope of 250.
Ballistic Phonon Conduction
• A CNT may be treated as a phonon waveguide.
• A simple calculation of the cross over temperature Tco
We have shown that for the onset of thermal conductance quantisation in a
crystalline structure with lateral dimension w and sound
CNTs do exhibit very high
velocity v occurs when:
thermal conductivity which
kBTco π
is believed to be due to ≈ ∆k ≈
ballistic phonon transfer. hν w
SWNT have diameters as small as 2 nm and phonon
We have predicted that velocities are up to ~20 km/s, hence T would be ~ 240K
the thermal conductance
should be quantised at
A more realistic calculation for a (10,10) SWNT the
reasonably high lowest energy optical phonon mode has a frequency
temperatures (< 30K) foptm of ~ 0.5 THz. Thus for
hf optm
π k (Thot Tcold )
2 2 1/ 2
π k T
2 2 Tco ≈ ~ 25 K
Gth = B
= B kB
3h 3h
only acoustic (massless) phonon modes are populated
and the onset of quantised thermal conductance should
be observed.
Current Breakdown in
Nanotubes Icnt (uA) Icnt (uA)

160 300

140
250
120
200
100

80 150

60

If the current through a nanotube is

100
40

50
20

ramped slowly the electrical 0

2160 2165 2170 2175 2180

G (Go)
2185 2190 2195 2200
0
1690 1695 1700 1705

G (Go)
1710 1715

breakdown is not initially total. 1

0.9

0.8
1.4

1.2

0.7 1

0.6

Steps in the current are seen as the

0.8
0.5
0.6
0.4

0.3 0.4

resistance increases 0.2

0.1

0
0.2

0
2160 2165 2170 2175 2180 2185 2190 2195 2200 1690 1695 1700 1705 1710 1715

Piezo V (V) Piezo V (V)

Finally the tube breaks. 87

86.8
89.5

89.4

86.6 89.3

86.4 89.2

The breakdown currents are of the 86.2

86

85.8
89.1

89

88.9

order of up to 1mA 85.6

85.4

85.2
88.8

88.7

88.6

85 88.5
2160 2165 2170 2175 2180 2185 2190 2195 2200 1690 1695 1700 1705 1710 1715

This represents a current density of

1014 A/m2
Carbon nanotubes could provide a
Even superconducting wires cannot solution to electrical interconnects in
match this high density electronic chips.
 Graphene: single sheets of carbon
atoms
Very recently a new form of carbon has been
produced: large sheets of graphite, just one
atom thick. This has been named graphene

It is generally
made by using
sellotape to peel
layers off platelets
of graphite until
only one layer of
atoms is left

It is not strong but is an ideal

2D electrical conductor which
demonstrates quantum Hall
effect at room temperature
Geim et al., Nature (2004)
Since they are planar, unlike CNTs, they can be
patterned into 2D circuits
Since it is not chiral there are no yield issues …..
QHE in novel systems: graphene
1 µm

Recent experiments reveal unusual
QHE in graphene samples.

Unlike conventional 2D systems, the
observed quantisation is described by half
integer rather than integer values.

This can be attributed to the peculiar
topology of the graphene band structure
with a linear dispersion relation and
vanishing mass.
Yuanbo Zhang, Yan-Wen Tan, Horst L. Stormer & Philip Kim,
Nature vol. 438 pp. 201-4 (2005)

K. S. Novoselov et al. Nature Physics vol. 2 pp.177-80 (2006)

Spintronics (2007 Nobel physics prize)

Spin has a number of potential advantages over charge:
 faster
 lower energy
 less subject to external perturbations

Manipulation of magnetic spin by electric fields

Semiconductors (with much lower carrier densities) are potentially
more useful than metals and easier to integrate with existing
systems.

Electronic spins may be manipulated by incident optical fields.

Analogously to the anomalous Hall effect but applicable to
nonmagnetic systems, the Spin Hall Effect opens new ways to
induce and control spin currents in semiconductors without applying
magnetic fields or introducing ferromagnetic elements.
Macroscopic Tunnelling of Spin in
Mn12 (Mn 0 (CH COO) (H 0) )
12 12 3 16 2 4

High spin state molecules, such as

Mn12, total spin S = 10, have been
shown to undergo macroscopic
tunnelling between degenerate states
with different Sz values.

These molecules have been
proposed as possible qubit realisations
in view of the uniform properties of all J. R. Friedman, M. P. Sarachik, J. Tejada and
R. Z, Phys. Rev. Lett. v76 (1996) pp.3830-3
molecules, their high magnetic
anisotropy and the weak coupling A proposal to carry out qubit
between neighbours. SWAP operation between
endo-fullerene spin and a

But the magnetic properties of an single molecular magnet.
isolated single molecule have yet to be M. Feng & J. Twamley, Europ. Phys. Jnl. v.69
studied. (2005) pp.699-705
Spin Hall Effect

This can be understood as the spin analogue of the
conventional Hall effect

A pure spin current is generated transverse to an applied
electric field. No magnetic field is required.

A pure spin current is one in which there is a net spin
transport in a specified direction but there is no charge
transport.

Spin accumulation is expected across a sample edges.

An intrinsic effect is predicted, arising from the band structure
and the nature of the spin-orbit coupling in the material.

Others propose the dominant, or possibly the only, effect
arises from strain in the lattice of the magnetic semiconductor
sample.
Observation of Spin Hall Effect in
Metals

First electrical measurements of the spin
Hall effect. These are made in a diffusive
metallic conductor. A spin polarised current is
injected from a FM tunnel electrode.

An induced voltage is measured, from the
conversion of the injected spin current into
charge imbalance through the spin Hall
effect.

The voltage is proportional to the
component of the injected spins that is
perpendicular to the plane defined by the spin
current direction and the voltage probes.
Pσ SH sin θ exp(− LSH / λ SF )
RSH =
2tσ c2
S. O. Valenzuela and M. Tinkham, Direct electronic measurement of the spin Hall effect,
Nature v.442 (2006) pp.176-9
Miniature dc SQUIDs for Detection of
Single Atomic Spin-Flips
We analyse progress towards:

A SQUID based system capable of detecting single atomic
spin-flips.

Scaling of flux sensitivity with SQUID loop dimension

Experimental verification from a series of miniature
niobium and HTS dc SQUID devices

Smallest SQUID (loop size 0.25 µm × 0.25 µm) should be
capable of detecting a few spins in a 1 Hz bandwidth.

Future work to deposit a spin sample, of nanoscale
dimension, within the SQUID loop.
Minimum Detectable Spin Flip
Detection
The minimum detectable flux change for a SQUID
with loop inductance L and Josephson junction
capacitance C is:

φ ns = (2εL) = (32k BT (LC ) L)1 / 2

1 1
2 2

Taking a signal-to-noise ratio of 1 as setting the

limit for spin flip sensitivity:-

N min =
( 3/ 2
16a k B TL C )
1/ 2 1/ 2

3µ
 Spin Sensitivity as a Function of
SQUID Loop Dimension
We can calculate what is the minimum number N of
electronic spin flips (µB) which can be detected, as a
function of the SQUID washer size a.

1 .10
6
Assume

1 .10
5 1. that the inductance is in
the form of a square
Minimum spin flip detection

1 .10
4
washer with square hole
of side a. Then
1 .10
3

Ls = 1.25µ 0 a
100
2. Optimum coupling
10 between SQUID and
spins
1
1 .10 1 .10 1 .10 1 .10
6 5 4 3
SQUID loop size (m)
 SQUID System Noise as a Function
of Loop Size
30
• Noise measurements show a
×10-7 φ0/√
√Hz

φ ns (10 φ 0/Hz )
white floor level of 5×

1/2
×3µ
for the 3× µm SQUID. 20

• Limited by head amplifier noise

-7
√Hz of our SQUID
~ 1 nV/√
electronics.
10

• φns exhibits the expected linear

scaling with a, except in the limit 0
→0 where it approaches noise
a→ 0 20 40
plateau. a (µ m)
Variation of measured flux noise density with
SQUID loop dimension.
 Predicted Spin Sensitivity

10000

Scaling of spin sensitivity with
SQUID loop dimension as 1000

S n (µ B /Hz )
1/2
predicted
100

Our 3×3 µm SQUID has a
predicted sensitivity of ~ 40
10
electron spins/√Hz 1

predicted limit of just 0.1
2.5 spins/√Hz in the thermal 0 20 40
noise limit (0.8 K)
a (µ m)

How to do better? Must reduce a, optimise coupling …

Where to put the magnetic
particle or spin?

Particle

Junction: 65 nm x 80 nm

Loop:dia=371nm
Nanoscale SThM

Thermometry
• An important requirement is to develop a
thermometer with minimum size and thermal
capacity for nanoscale measurement
• Wollaston wire probe (commercially available)
Cantilever: Pd-Rh coated by Ag (r~75 µm)
• Other possible thermometers: Tip(Pd-Rh) r~5 µm)
– Nanostructured probe (Glasgow
University e-beam lithography)
– CNT as resistance thermometer (must
not be ballistic)
– Resonant oscillation of cantilever
(Young’s modulus and thermal expansion are T
dependent)
– Noise thermometry of cantilever
(measure Q and noise amplitude of cantilever in
contact with surface) Cantilever (PECVD Si3N4 150µmx60µmx500nm
Tip(40nm Pd) r~100nm)
Limits on Temperature Measurement
Measurement of very small temperature changes (pK)
generally use thermometers with SQUID readout
Two distinct forms based on SQUIDs:
High Resolution Thermometer (HRT) measures the
temperature change of a paramagnetic salt, coupled to a
SQUID. Curie-Weiss law susceptibility plus SQUID
senses very small changes in T.
Transition Edge Sensor (TES) relies on highly non-
linear R(T) of a superconductor around Tc. SQUID
voltage readout provides extreme sensitivity (pK).
But is there a limit on T measurement sensitivity?
Thermodynamic Fluctuation Limit
It was pointed out long ago (see e.g. Landau & Lifshitz; ‘Statistical
Physics’) that for a system with heat capacity C there is a
Gaussian probability for temperature fluctuations ∆T given by
2
k BT
(∆T ) 2
=
C
Power spectral density ST(f) of temperature fluctuations is:
4 k BT 2
ST ( f ) =
G
where G is the thermal conductance to the heat sink controlled at <T>

Readout sensitivity of a HRT/SQUID is a few pK/(Hz)1/2, easily capable of

measuring ∆T for real systems (Day et al. JLTP 60 p359 (1997)) which is more
typically 10-10K /(Hz)1/2
Realisation of Temperature on Nanoscale
• There is a growing discussion
about temperature measurement
as the length scale reduces….
• Can temperature be
meaningfully defined on
nanometer length scales?
• Quantum description becomes
essential at the nanoscale.
• Experimental data is also
becoming necessary.
Michael Hartmann, Günter Mahler & Ortwin Hess, ‘Existence of Temperature on the Nanoscale’, Phys. Rev. Lett. 93
080402 (2004)
P. S. Dobson, G. Mills & J. M. R. Weaver, ‘Microfabricated temperature standard based on Johnson noise
measurement for the calibration of nano-thermometers’, Rev. Sci. Instrum. 76 054901 (2005)
Quantum Roulette Noise Thermometer

Aim: to develop radically new form of primary
thermometer, based on evolution of a time series of states
of a single particle.

Here the single particle is the quantised flux state of a
superconducting ring.

Measure trajectory of this quantum ‘particle’ in flux space,
then calculate from first principles its temperature.
• QRNT ring contains a small d.c. SQUID
whose effective ic can be modulated by an
r.f field (Φxr.f) applied to junction coil.
• The larger ring coil can be used to apply
flux to the main QRNT ring (Φxd.c.).
• A separate d.c. SQUID magnetometer
measures the flux state (Φi) of the QRNT
ring.
Quantum Roulette Noise Thermometer

BUT beware the use of d-wave superconductors.

The simple predicted temperature dependence of
P(Φ,T) breaks down since I(θ,T) is complicated….
 Cantilever cooling by laser
Au-coated Si cantilever forms one mirror of
a miniature Fabry.Perot resonator
The cantilever mechanics is strongly
coupled to the light field
This change is delayed, leading to a force
on the lever, proportional to its velocity.
This light-induced viscous damping causes
cavity cooling of the lever.

Cantilever temperature is measured by its

noise amplitude (optically measured) and
cooling from 300K to 18K has been
demonstrated.
What is the cantilever lattice temperature?
NIST: Cantilever Cooling by Microwaves

K. R. Brown, J. Britton, R. J. Epstein, J. Chiaverini, D. Leibfried

& D. J. Wineland, Phys. Rev. Lett. v.99 (2007)
 The end of classical metrology?

Metrology based on
Information, Flux, Charge & Spin,
rather than
Temperature,Voltage, Current &
Resistance
The beginning of true quantum
metrology and of a deeper
understanding?
Classical tunnelling at room T……
Acknowledgements
Thanks to the following colleagues at NPL:

Ling Hao
John Macfarlane
Sasha Tzalenchuk
Julia Davies and to David Cox

Olga Kazakova (Surrey University )

Lizzie Brown
Patrick Josephs-Franks
Carol Webster