- National Physical Laboratory

advertisement
Development of ultra-low noise nanoSQUIDs
using FIB for quantum measurement
Ling Hao, John Gallop and David Cox
Quantum Detection Group, National Physical Laboratory, Teddington, TW11 0LW, UK
Contact: E-mail: ling.hao@npl.co.uk
Introduction
•Q
uantum metrology and nanoscience are driving needs for single
particle detection across physics (e.g. QIP, single photon detection,
NEMS, nanomagnetism and spintronics).
2. The ion beam current is kept as low as possible (<5 nA) to
avoid poisoning.
3 During milling the sample is imaged with the e-beam so
that milling is stopped as soon as all the Nb is removed (live
Imaging is possible provided the e- beam current exceeds the
ion beam current).
•N
anoSQUIDs are a new manifestation of an old but exciting
technology which addresses some of these requirements.
• We have fabricated various Nb nanoscale SQUIDs with loop
size down to 100 nm with nanobridge weak-links (~50 nm) as
Josephson junctions, using a simple FIB-based approach.
• These devices show non-hysteretic IVCs and very low noise at
operating temperatures > 4.2 K.
• This thin W layer also provides a shunt resistor for
junctions.
Fabrication: Nb nanoscale
devices using FIB
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.
SEM image of
superconducting
Nb nano-bridge
structures ~80
nm wide.
SEM image of tri-loop
gradiometer nanoSQUID
prepared by FIB. Note the
magnetic Co nanoparticle
place inside one loop.
Ultimate goal:
Can we make a SQUID based system
capable of detecting single electron
(atomic) spin-flips?
Required:
• The magnetisation sensitivity
improves as the SQUID loop
dimension gets smaller (so make the
SQUID loop as small as possible).
Preparation of
Nb junctions
using FIB.
• S ingle nanomagnetic particles (e.g. PtFe bead) and NEMS
resonator have been incorporated into the nanoSQUID system
using micro/nano manipulation and FIB techniques.
NanoSQUIDs for detection
of Magnetic Nanoparticle
• Smallest SQUID (loop size 250 nm x 250 nm) should be capable of
detecting a few spins in a 1 Hz bandwidth.
• Deposit or manipulate a spin sample, of nanoscale dimension,
within the SQUID loop.
• Optimise coupling between Spin sample and NanoSQUID
No particle, no hysteresis:
150nm Fe-Pt particle, hysteresis:
The limiting flux noise spectral density Sf may be written:
where en is minimum detectable energy change, C and L are the
capacitance and inductance of the SQUID and T is temperature.
NanoSQUID characterisation and
measurement
So en can be improved by reducing C or L (as well as the more
obvious T). That means NanoSQUIDs !!
Quantum Limited SQUIDs
The Josephson junctions are also characterised by a plasma
frequency ωp where:
Our results demonstrate a significant phase shift on the ‘SSA signal
vs flux’ plot due to the hysteresis of the magnetic nanoparticle.
Cryogenic NEMS resonators readout
with nanoSQUID
I-V curve for a SQUID by FIB at
T=8.75 K with Ic=240 µA. Note the
absence of hysteresis.
• We were the first group
to propose a SQUID to
read out NEMS
At low temperatures if:
• Prototype paddle
resonator/SQUID tested
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:
• The trick is to optimise
the coupling between
resonator & SQUID
Voltage vs. applied magnetic
fields. Operating temperature is
T = 6.01 K & Ib=280 µA.
(agrees with the Uncertainty Principle!)
Experimentally the best SQUIDs have a minimum flux sensitivity
of around 1x10-7 F0/(Hz)1/2, limited by room temperature amplifier
noise.
Approach to nanoscale- SQUIDs
Voltage amplitude
~195 µV, dV/dΦ is ~ 2.5 mV/ Φ0
NanoSQUID characterization in 2stage
SQUID configuration
•O
ptical lithographically
fabrication.
Nano SQUID loops require small
Josephson junctions:
Nanometre sized FIB fabrication
• The low tunnel current density
requires large areas.
•M
icrobridges based on single
layer can provide sub-micron and
nanoscale junctions with low C
and high Jc.
Model using finite element software.
• New NEMS-SQUID mask
design underway
New SQUID slot
structure (1 µm x 100
nm) for optimal beam
coupling
The SEM picture shows a prototype
ISTED, absorber has been deposited
in the nano-SQUID square loop
(~200 nm x 200 nm).
Schematic of 2stage setup using
a SQUID series array (SSA) to read
out the nanoSQUID
• 16-SQUID series array current sensor.
• Input inductance < 3 nH.
E-beam: Nanometre SQUIDS
NanoSQUIDs fabrication based on
microbridge junctions:
• Current noise√ SI <10 pA/√Hz @ 4.2 K.
• Magnetically unshielded operation.
Noise performance of Nb nanoSQUID
• E -beam direct write lithography
using SEM.
• F IB (focussed ion beam) milling.
• Lorentz force excitation
is effective (Our SQUIDs
operate in field > 1 T)
d.c. SQUID readout
of first 3 modes of
paddle
Inductive Superconducting
Transition-edge Detector (ISTED)
• S ize limited to > 2 µm.
• These are hard to make in trilayer
(C high, Jc low).
SQUID
loop
• Inductive coupling has
proved to be an effective
method.
• Fabricate smaller NEMS
beam/SQUID devices to
maximise frequency &
optimise coupling
Optical-lithographically
fabrication
Si3N4 torsional
oscillator
coupled to a Nb
microSQUID
Loop: 200nm
Fabrication: nanoscale
junctions using FIB
FIB milling of nanoscale microbridge junctions down to 50 nm (FEI
Nova Nanolab 600 dual beam FIB)
But for superconducting device: Ga ion implantation can poison
superconducting thin film (Tc is lower or superconductivity
destroyed).
Our method:
1. Deposit W(CO)6 using an e-beam over the Nb tracks, giving a
pad ~ 150 nm thick.
• This thickness provides a protective layer to prevent Ga
ion implantation in the Nb junction during milling.
Absorber
hv
• We have demonstrated a new
technique (ISTED) where the
Squid
sensitive element is an isolated,
passive absorber of low thermal
mass, kept just below its transition
temperature Tc and inductively coupled to a SQUID sensor.
• Incoming particles are sensed in terms of a transient change in the
inductive coupling, rather than a change in resistance.
Spectrum of photon counts as a
function of energy at T=8.15 K.
We measured flux noise spectral density, expressed in units of the
flux quantum Φ0, as a function of frequency from 0.1 Hz to 100 kHz.
• 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.
Flux noise spectral density at T=6.8K, bias
current 80 µA.
• 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 fluxlocked loop bandwidth
of the measurements.
FIB ISTED: 50 µm device with
nanojunctions ~80 nm.
The pulse height output from SQUID NPL6-4E_Sc with 50 µm
absorber, operated at 8.15 K was analysed for 3 different situations:
This is lowest reported NanoSQUIDs
noise above 4 K.
633 nm laser weakly coupled
Laser off - fibre warm
Laser off - fibre cold
Note that the energy resolution is ~0.2 eV or ~10% (~100 nm
at 633 nm), even at this high operating temperature, energy
sensitivity ~3.5 x10-25J/Hz. NEP~1x10-17 W/Hz1/2 at 8 K.
Published papers:
L.Hao et al, APL, Vol. 98, 092504 ,2011.
L. Hao, Journal of Physics: Conference Series 286 (2011) 012013
Ed Romans et al, IEEE Transactions on Applied Superconductivity,, 2011
L.Hao et al, JPD Vol.43 p474004, 2010.
L.Hao et al, IEEE Transactions on Applied Superconductivity, Vol. 19, pp.693-6, 2009
L.Hao et al, SUST, Vol.22 p064011 2009.
L.Hao et al, APL 92 192507, 2008.
L.Hao et al, IEEE Transactions on Applied Superconductivity, Vol. 17, pp.742-5, 2007
L. Hao et al, J. Appl. Phys. 99, 123916, 2006.
D. Drung et al., IEEE Transactions on Applied Superconductivity, Vol. 17, pp.699 – 704, 2007
Hao et al., IEEE Trans. Instrum. Meas. 56 pp.392-5, 2007
L. Hao et al., IEEE Trans. Appl. Supercond., v 15, p.514-517 2005.
D. Drung et al., Supercond. Sci. Technol. vol. 19 pp.S235-41, 2006.
J. C. Gallop, Supercond. Sci. & Technol., vol. 16 pp. 1565-72, 2003.
L. . Hao et al. IEEE Trans. Instrum. & Meas. Vol. 54, No.2, 2005
L. Hao, J. C Gallop and J. C. Macfarlane, IEEE Trans. Instrum. Meas., Vol. 52, No. 2, pp. 328-332,
April, 2003.
L Hao, J C Gallop, C. H. Gardiner, P. Josephs-Franks, J C Macfarlane, S K H Lam and C Foley,
Supercond. Sci. & Technol., vol. 16 No. 2 pp.1479-1482, 2003.
L. Hao et al, IEEE Trans. Appl. Supercond., Vol.13. No2. pp.622-625., 2003.
9686/0711
• Tri-layer junctions.
Temperature dependence of critical
current. The curve is a fit to the
points: Ic~(1-T/Tc)2.28, and Tc = 8.5 K.
paddle
junction
© Queen’s Printer and Controller of HMSO, 2011.
Conventional SQUIDs fabrication:
beam
Download