Download George Pickett - JRA1 - European Microkelvin Collaboration

Joint Research Activity 1
Opening the Microkelvin Regime
to Nanoscience
This specific joint research activity is central to the whole project:
“Opening the microkelvin regime to nanoscience”
It is this activity which is going to make the whole thing happen.
Let us consider the tasks one by one.
Task 1
Developing the new technology needed to cool nanosamples and circuits to around or below 1
mK
To integrate nanoscale experiments into sub-millikelvin cryostats will require new technology. The
difficulties are largely those of making thermal contact to the electron gases in the nanostructures. This is
especially true with semiconductor nanostructures. At ultralow temperatures such substrates become
effective thermal vacua and thermal contact is often restricted to the pathways via the metallic leads to
the circuits.
The only quantity which matters in cooling such circuits is the ratio of the heat leak into the circuit
material to the thermal contact to the refrigerant. Both quantities have to be attacked in parallel. First we
can make great efforts to reduce the external heat leak.
With the best current refrigerators we can create enclosures which are so well insulated that the heat
leak into an isolated non-conductor is already at the level set by the background radioactive heating
(largely from nearby constructional concrete).
Metallic samples experience additional heating from eddy currents generated by motion in stray
magnetic fields. However, these can also be reduced to a level below 4 pW per mole which translates to
~10-24 watts into a micron cube sample.
The real difficulties come when we attach leads, as this immediately connects the outside world. We
have to take this problem very seriously and start with the best electrical filtering possible, which
fortunately is being pursued with in JRA2. Secondly we must enhance the thermal contact to the
refrigerator. In a semiconductor 2DEG, for example, the substrate makes virtually no contribution to
thermal contact at the lowest temperatures which runs entirely via the leads. Using ideas from BASEL
and ULANC we can thermally anchor each lead directly in the mixing chamber liquid with sintered silver
pads and then furnish each lead with its own mini nuclear stage to absorb any final incoming energy in
the nuclear bath.
Just look for a moment at some of our best technical setups for cooling helium (and then copper).
This is simply to give an idea of what we can do now.
This is what the ult community brings to the table. I.e. largely the input from TKK, CNRS and
ULANC.
Let’s start with cooling superfluid 3He.
Since we only need a small volume of
copper to cool liquid 3He, let’s get it as
close to the specimen as possible, that
is immerse it in the liquid.
So we start with a thin-walled paperepoxy box (to put our liquid 3He and
refrigerant in).
We have added a sapphire tube (as in
this experiment we want to make
NMR measurements in the tower soproduced).
We add the refrigerant, a stack of Cu
plates coated on one side with a ~
1mm layer of sintered silver powder
to make thermal contact with the
liquid.
We add the refrigerant, a stack of Cu
plates coated on one side with a ~
1mm layer of sintered silver powder
to make thermal contact with the
liquid.
We add a silver sinter pad to make
contact for precooling and a filling
tube.
To cut down the heat leak we add a
second stage, also furnished with a
precooling link, and filling tube.
We put the inner cell inside.
This allows the inner cell to have a
very thin wall (und thus low slowrelease heat leak) because the
pressure is supported by the outer cell
wall.
The outer-cell copper refrigerant pads
are connected by high conductivity Ag
wires (rr~103) to a single crystal Al
s/conducting heat switch.
Further silver wires lead to
The
heat switch
is connected
to the
precooling
pads
which will
sitAg
cooling
sinters tochamber
sit in the mixing
in the mixing
of the
chamber.
dilution refrigerator (at ~ 2
(The
mK).cone is the mixing chamber
base.)
The whole structure is supported by a
thin-walled epoxy cylinder.
We insert the cell into the mixing
chamber.
We push the cone joint together and
screw it up.
Done!
This system will cool superfluid helium-3 to around or below 80
mK.
Now let us use a similar system just to cool the copper refrigerant.
To do this we put a multiple coper stage in the inner volume as in the previous
setup..
We start with an epoxy box
which we will immerse in the
outer cell
(the box being filled with
vacuum).
We attach three high-purity Agwire supports, two at the bottom
and one which also acts as
thermal link to the heat switch.
(Remember the epoxy is acting at
these temperatures almost as a
thermal vacuum.)
We attach three high-purity Agwire supports, two at the bottom
and one which also acts as
thermal link to the heat switch.
Ag thermal link
(Remember the epoxy is acting at
these temperatures almost as a
thermal vacuum.)
Ag Mechanical
supports
The supports are spot-welded to
the first Cu refrigerant plate.
First Cu plate
A superconducting heat switch
(Al or Sn) is melted/spot-welded
to the copper plate.
S/c heat switch
A second Cu refrigerant plate is
added.
Cu plate No 2
Then a second heat switch
Finally the third and final Cu plate
is added.
A Pt NMR thermometer is added
– this measured the susceptibility
of the Pt nuclei and gives us the
temperature simply from Curie’s
law
The thermometer is a bundle of
fine uninsulated Pt wires
soldered with pure silver and on
which we will do NMR with a set
of coils immersed in the outer cell
– not touching the inner parts).
We glue on to the final plate a
pure Ag wire heater to calibrate
the thermometer (using a
microscopic amount of epoxy to
stick it to the plate.
Finally we connect the heater
with pure tin leads to a thermal
anchor on the outer plate, (and
put it all back in the box).
The box, is put in turn inside a
Lancaster outer stage. (Only
contact to final Cu plate is via the
heat switch and the Sn leads.)
(Ignore the jumps. That’s a problem of the heat switch)!
Note temperature 5 mK, that’s 8 orders of
magnitude colder than room temperature
(centre of Sun only 5 orders warmer).
(Ignore the jumps. That’s a problem of the heat switch)!
Thus from our experience with working with quantum fluids we can cool superfluid 3He to ~ 80
mK and the electron system in copper to ~5 mK.
That’s state of the art.
How do we translate that into a system for cooling nanoscience samples?
The following setup was used for a nano Andreev interferometer experiment in the mixing
chamber of one of our machines using the simplest possible methods.
Just to get us started.
We start with a bundle of highconductivity silver lead wires f1
mm each with a sintered Ag pad
to act as a thermal ground.
The silicon substrate was glued
with black Stycast directly on to
another similar thermal ground
wire
The silicon substrate was glued
with black Stycast directly on to
another similar thermal ground
wire and the circuit connections
were bonded straight on to the
silver lead wires.
That circuit immersed fully in the helium in the mixing chamber cooled to at least 4 mK.
The nano community think that 30 mK is about the limit for dilution refrigerators so do not think
in these terms.
Now this was without any particular clever filtering on the leads.
We of course would do that but that is the job for JRA2 which we will be hearing about.
To enhance the thermal contact to the refrigerator we use ideas from BASEL and ULANC. We thermally
anchor each lead directly in the mixing chamber liquid with sintered silver pads as above and then
furnish each lead with its own mini nuclear stage to absorb any final incoming energy in the nuclear
bath.
Finally we can envisage completely new tailor-made nanoscale structures independent of conventional
semiconductors. Ideal candidates for microkelvin cooling are carbon-nanotubes and graphene
structures which can be directly immersed in superfluid 3He where there is a dense 3He quasiparticle gas
making orders of magnitude better contact directly to the structures.
Finally we can envisage completely new tailor-made nanoscale structures independent of conventional
semiconductors. Ideal candidates for microkelvin cooling are carbon-nanotubes and graphene
structures which can be directly immersed in superfluid 3He where there is a dense 3He quasiparticle gas
making orders of magnitude better contact directly to the structures.
Task 2
Building with our SME partner BlueFors a self-standing dilution refrigerator plus nuclear cooling stage
with nanoscample capability which can be used in any lab in the world without the need for refrigerants.
This builds on task 1 and is our direct contribution to European and other workers outside the
consortium who have no access to refrigerant technology. (Coord CNRS TKK)
This opens up nanoscience to everybody.
Task 3
The next-generation microkelvin facility (ULANC, SAS, TKK, CNRS, BASEL, RHUL)
Using the combined knowledge and expertise of the applicants we are also planning an entirely new
advanced microkelvin refrigerator facility intended exclusively for condensed-matter and nanoscale
experiments at milli- and microkelvin temperatures. This will be sited at ULANC in a purpose-built
90+m2 laboratory hall with 7 m clearance and a 3 m dewar pit dedicated to this project, which is
supported by €k400 from the UK Science Research Investment Fund. The access-giving laboratories in
this consortium have a very large fraction of the world expertise and capability in carrying out
experiments at sub-millikelvin temperatures. We propose to build on this unique European resource by
pooling our existing knowledge along with the technology developed in tasks 1 and 2 above to make
this the most advanced sub-microkelvin facility for nanokelvin studies that current knowledge will
allow. (coord ULANC).