An introduction to cryogenics
Adam L. Woodcraft
Home page:http://woodcraft.lowtemp.org
SUPA, Institute for Astronomy, Edinburgh University, Blackford Hill, Edinburgh EH9 3HJ, UK
(Dated: October 25, 2007. Last major change: October 23, 2007)
Contents
1. Introduction
1
2. Providing a cryogenic environment
1
A. Cooling methods
1
1. Liquid cryogens
2
2. Mechanical
3
3. Cooling below 4 K
4
B. Types of cryostat
5
3. Getting and keeping your sample cold
5
A. Thermal conductivity
5
B. Contact conductance
7
C. Heat capacity and thermal expansion
7
D. Thermal radiation
7
E. Other heat transfer mechanisms
8
1. Gas conduction
8
2. Heat switches
8
3. Film creep
8
4. Some other aspects of experimental design
8
A. Wiring
8
B. Choosing materials
8
C. Grounding and RF
9
5. When things go wrong
9
A. Leaks
9
B. Blocked exhaust ports
9
6. Warming up!
9
References
10
1. INTRODUCTION
These notes aim to give a somewhat basic introduction to the
hows (and occasionally whys) of cryogenics. Cryogenics is
not a very well defined term; here I have taken it to mean sys-
tems that operate at temperatures of 77 K and below, with the
emphasis on systems operating below 4.2 K and above 1 K.
The first two temperatures are, of course, the boiling points of
nitrogen and helium respectively.
A cryogenic experiment or system is normally dominated by
the need to get something cold and keep it cold, with other
elements of the design subservient to that. Generally, the need
to operate at cryogenic temperatures makes even an otherwise
simple experiment complicated, and the colder you need to go,
the harder life generally gets. Since it is so time consuming to
get things cold, the ability to experiment with new techniques
is somewhat limited, and most low temperature labs cling to
a set of tried-and-trusted methods. Often these differ from lab
to lab, and in the most interesting cases are contradictory.
First, a few basic definitions. In these notes, I shall use the
following words in the way defined here (these are not the
only definitions)
–
cryostat: the system that a low temperature experiment
goes in
–
dewar: something for storing and (possibly) transporting
liquid cryogens
Helium comes in two isotopes; the common helium-4 (4He),
which is obtained from oil and gas wells (it starts life as alpha
particles) and the much more exotic helium-3 (3He) which
does not occur in nature, but is produced by the decay of tri-
tium.
There are several books to which you can turn to expand on
these notes, and after some years of neglect, old books are be-
ing reprinted in new editions, and new books appearing. The
following lists the ones that I refer to the most; there are other
books which are also well worth reading.
– Ekin (Ref. [1]); this is a new book, and gives lots of useful,
detailed, down-to-earth advice on many aspects of cryogen-
ics, though it is generally limited to temperatures above 1 K,
and has an emphasis on making measurements on supercon-
ductors (you may or may not think this is a good thing).
– White and Meeson (Ref. [2]) is also a very worthwhile read,
and is up to date.
– Pobell (Ref. [3]) is another standard textbook, and a new
edition has recently appeared.
– Richardson and Smith (Ref. [4]) is a collection of semi-
nars from the Low Temperature Group at Cornell Univer-
sity, and although somewhat out of date in places gives a lot
of hands-on advice.
– Wigley (Ref. [5]) gives useful information on mechanical
properties of materials.
2. PROVIDING A CRYOGENIC ENVIRONMENT
A. Cooling methods
Every cryogenic experiment needs some way of reaching low
temperatures. There are two ways of doing this; using cryo-
gens or using mechanical coolers.
Liquid cryogen cooled cryostats use the same principle as the
ancient technique of using blocks of ice to keep food cold in
storage and transport – incoming heat melts the ice, maintain-
ing a constant temperature until no ice remains. These days,
2
of course, we have replaced ice with mechanical refrigeration.
With a liquid cryogen, evaporation of the liquid maintains the
temperature at the boiling point, and this is how the major-
ity of cryogenic experiments are cooled and kept cold, though
mechanical cryogenic coolers have become more popular in
recent years.
1. Liquid cryogens
a. Cooling with cryogens. Cryogens provide cooling in
two ways:
– Heat is used in
evaporating the liquid - it absorbs
latent
heat.
– Heat is then used in warming the gas from the boiling point
to room temperature; the heat required to do this is given by
the
enthalpy change.
The two main cryogens in use in cryogenics are helium and
nitrogen; some properties are given below (B.P. is the boiling
point at 1 atmosphere, and the costs are of course approxi-
mate):
Cryogen B.P. Latentheat Cost/litre
Nitrogen 77K 161kJl−1
£0.1
Helium-4 4.2 K 2.6 kJ l−1
£3
If you are only going to cool an experiment to 77 K, then
liquid nitrogen is the sensible solution. If you need to cool
your system down to 4 K, then it makes sense to first cool to
77 K with liquid nitrogen, and then introduce helium, since
nitrogen is about a thousand times cheaper than helium per
Joule of latent heat. This works out particularly well because
the heat capacity of components in a cryostat drops sharply
with temperature, so by the time it has reached 77 K, most of
the heat has already been removed from the cryostat.
We can see how useful this approach is from the following
figures, which compares the quantity of cryogens required to
cool 1 kg of copper [3] with and without the use of nitrogen
to pre-cool the system:
Cryogen Temperature range Amount evaporated
Helium-4 300 K to 4.2 K
32 l
Nitrogen
300 K to 77 K
0.46 l
Helium-4
77 K to 4.2 K
2.2 l
While the latent heat of helium is rather small, the enthalpy
between 4.2 K and 300 K is much larger; about 200 kJ/l. We
can therefore reduce our consumption of cryogens even fur-
ther by using the enthalpy. If all the enthalpy were utilised,
we would have the following consumption [3]:
Cryogen Temperature range Amount evaporated
Helium-4 300 K to 4.2 K
0.8 l
Nitrogen
300 K to 77 K
0.28 l
Helium-4
77 K to 4.2 K
0.16 l
We can see that while the saving in nitrogen is not spectacular,
if we can use all the enthalpy, we can save a large amount of
helium. In practise, this is not possible. While you are guaran-
teed to use all the latent heat of a cryogen (it will all boil away
eventually), there is no such guarantee for the enthalpy; the
gas must leave the cryostat at room temperature in order for
all the enthalpy to be used. This is hard to achieve. However,
you can certainly arrange to maximise the gain from enthalpy.
For example, you can design your cryostat so that the gas is
slowed down as it leaves the cryostat, by passing through baf-
fles or a medium such as cotton wool.
Helium storage dewars generally have a set of radiation
shields which are thermally anchored to the neck, enabling
them to be cooled by the enthalpy of the boil-off gas. Some
cryostats have a liquid nitrogen bath used to cool radiation
shields to 77 K to reduce the amount of heat reaching the 4 K
stage. This is particularly common when optical access is re-
quired (for example for an astronomical instrument) and there
is a filter to be cooled at 77 K.
Finally, it is important to note that nitrogen has a rather large
latent heat of melting1. It is therefore very important to re-
move all the nitrogen before starting to fill with helium, oth-
erwise you will waste a large amount of helium to solidify the
nitrogen.
Hydrogen and neon are occasionally used as cryogens (boiling
points at 1 atmosphere of 20 K and 27 K respectively). How-
ever, neon is ten times as expensive as helium, and hydrogen
is explosive when mixed with air.
b. Handling, storing and transferring cryogens. Cryogens
are usually stored in a transport dewar, either filled locally
(from a nitrogen tank or helium liquifier), or delivered along
with their contents (which will then boil off at one or two
litres/day). Transferring nitrogen is fairly straightforward; it
can be transferred using flexible tubing, or poured (carefully)
from one container to another. The high latent heat means that
a polystyrene bucket can hold nitrogen for several hours.
Transferring helium is much harder work. The low latent
heat means that even transfer tubes must be vacuum jack-
eted. These vary in complexity from simple rigid affairs to
long flexible tubes (much more expensive, but well worth the
money). The basic principle is that a transfer tube is sealed
into the storage dewar and the cryostat, and the storage dewar
is pressurised (either using helium gas from a cylinder or by
squeezing a bladder attached to the dewar), thus causing he-
lium to be transferred. In practise getting a transfer right can
be tricky, with some cryostats being much easier than others.
As discussed above, pre-cooling with liquid nitrogen is al-
most universal. Make sure you get all the nitrogen out after
pre-cooling; this is usually done by pressurising the bath and
forcing it out through a tube which reaches the bottom; make
sure it really
does reach the bottom! To get the most from
the enthalpy, it’s usually best to transfer helium into the bot-
tom of the cryostat for the initial cool-down, so that the cold
gas passes up past the rest of the cryostat, providing cooling.
This is particularly important when there is a magnet to be
cooled down. If necessary the cryostat should have some sys-
tem for permitting this. Subsequent transfers should be done
into the top of the helium reservoir, otherwise warm gas pass-
1 More accurately known as the latent heat of fusion
3
ing through the transfer tube at the start will boil off liquid
helium that is still present.
A very good set of instructions is given in reference [1], but
the best way to learn is to do the first few transfers with some-
one who knows what they are doing until you (and they) are
confident that you know what you’re doing. It is disturbingly
easy to use up a whole dewar without getting any helium into
your cryostat.
c. Safety. There are three main dangers associated with
handling liquid cryogens:
– Their low temperature, which can cause severe skin dam-
age in the same way as extreme heat.
Burns can occur from
contact with liquid cryogens, cold gas produced by evapora-
tion of liquid cryogens, and with pipework containing cold
gas (e.g. helium boil-off from a cryostat). In extreme cases
frostbite may occur.
– The large difference between volume in the liquid and gas
phase (about a factor of 700) which can lead to an
explosion
if cryogenic liquids are unable to escape from a volume.
–
Asphyxiation due to displacement of oxygen by evaporated
cryogens (see point above).
Therefore, when handling cryogens, use the appropriate pro-
tective equipment (gloves, face shield, etc.) to prevent burns;
you should bear in mind that skin can stick to cold metals.
Eye protection from cold gases is particularly important. An-
other good reason for eye protection is that some solids cooled
to cryogenic temperatures become brittle, a common example
being flexible tubing used to carry liquid nitrogen, which can
shatter at unexpected moments.
The most serious accidents generally occur due to the explo-
sive bursting of a dewar when evaporated gases have nowhere
to go, often caused by build up of ice and frozen air in the
outlet. Cryostats should all have safety valves designed to pre-
vent this from happening, but nevertheless caution is required;
make sure dewars and cryostats have a small positive pressure
to avoid ingress of air
(and don’t leave them open to the air
for any longer than necessary), keep a close eye on them, and
make sure the correct valves are open. A burst cryostat is an
impressive sight but one best viewed after the occasion, and
ideally when it belongs to someone else.
Precautions against asphyxiation include making sure that
labs are well ventilated, routing exhaust gases outside the
building, and having an oxygen depletion alarm in the lab. For
obvious reasons, don’t travel with a nitrogen or helium dewar
in a lift (and make sure that nobody else gets into the lift with
it while you’re running up or down the stairs.)
Another fact to bear in mind is that nitrogen will evaporate
first from liquid air, leaving an oxygen-enriched liquid and
a potential
combustion hazard. Precautions against this in-
clude keeping the lid on buckets of nitrogen, and keeping cold
surfaces which may condense air (such as pipes carrying he-
lium boil-off gas) clean (no grease, oil, or anything else that is
readily combustible).
The above is not intended as a comprehensive guide to safety
hazards with cryogens; your lab should have a code of practise
for handling liquid cryogens – use it!
2. Mechanical
a. Cooling with mechanical refrigerators An alternative to
using liquid cryogens to cool an experiment is to use a me-
chanical cooler. There are two types generally in use:
– Gifford-McMahon (
GM) coolers
– Pulse tube refrigerators (
PTRs)
Both consist of two components, linked by plumbing: a room
temperature compressor (which may be water cooled), and an
expansion chamber inside the cryostat. Both work on the prin-
ciple of compressing helium gas at room temperature (or a
little above) in the compressor, then sending it into the cryo-
stat where it is expanded to produce cooling. The gas passes
through a high heat capacity “regenerator” which cools the
gas on its way into the cryostat, and is in turn cooled by the
expanded gas as it leaves.
A G-M cooler uses a piston in the expansion chamber,
whereas a pulse tube refrigerator avoids the use of cold mov-
ing parts by using acoustic pulses in the helium gas to replace
the moving piston. This has the advantage of reduced vibra-
tion, and reduced maintenance costs.
Both types of cooler are commercially available, and can pay
for themselves in a year or so by removing the need to buy liq-
uid helium (they are incredibly inefficient, however, and use a
large amount of electrical power; for example 11 kW to pro-
duce 1.5 W of cooling at 4.2 K). Typical cooling powers at
4 K vary from 0.5 to 1.5 W. Of course, avoiding liquid cryo-
gens has other advantages; it makes operation much safer and
more straightforward, and for labs without a liquefier avoids
the need to plan ahead by ordering sufficient helium, and pos-
sibly ending up with it all boiling away if a cool-down has to
be aborted. It also avoids being tied to a cryostat by having
to be available to do helium transfers every couple of days...
To cool to 4-K with a mechanical cooler requires a two stage
system, with the first stage operating at a temperature of 40 to
80 K. The first stage typically has a large cooling power (e.g.
40 W at 45 K, compared with 1.5 W at 4 K), and can thus be
used to cool radiation shields in a similar manner to a helium
dewar with a liquid nitrogen bath.
The main downside to mechanical coolers (apart from high
initial cost, and the possibility of repair bills) is the vibrations
they cause. This can interfere with or even prevent experi-
ments from being carried out, although with careful design
the effects of vibration can often be mitigated. If you buy a
cooler from the USA, it will generally be optimised to operate
from a 60 Hz electricity supply, and will have worse perfor-
mance when operated in Europe with a 50 Hz supply (moral:
make sure they quote you the performance at 50 Hz).
Mechanical coolers are particularly attractive when cryostats
are run in remote locations, such as on an astronomical in-
strument sitting half way up a large telescope on the top of
a mountain. Sometimes they can be used along with liquid
cryogens, in applications where the cooler cannot reach a suf-
ficiently low base temperature to be used alone, but instead
reduces boil-off of the cryogens (for example a rather exotic
system used in an X-ray astronomy satellite where a mechan-
ical cooler reduced boil-off from a solid neon dewar, which
4
itself reduced boil-off from a liquid helium dewar).
b. Safety Most of the points above on safety with liquid
cryogens don’t apply to systems using mechanical coolers, but
some care is still required (for example it is still possible to
burn yourself with a mechanically cooled cryostat by opening
it up before it is warm).
3. Cooling below 4 K
So far I have discussed methods of cooling to 4 K. Cooling
to lower temperatures involves further complexity, and I will
give a very brief summary here.
a. Pumped helium-4 systems Helium-4 boils at 4.2 K at
one atmosphere, but like any gas, the boiling point will re-
duce as the pressure is reduced. Therefore, by pumping on
the liquid we can reach a temperature of about 1 K (below this
the vapour pressure become too small). This can be done by
pumping on the main bath of a liquid helium dewar, or, more
commonly, on a small container (
1-K pot) which is continu-
ally replenished with helium from the main bath. The latter
scheme is popular since it means the cryostat can be re-filled
without warming to 4 K, and is somewhat more efficient, since
you don’t waste cooling power cooling the helium in the main
bath (helium has a very high heat capacity, and around half
the liquid is lost when cooling from 4.2 K to 1 K).
Currently, commercial mechanical coolers will not reach 1 K.
However, it is possible to use a self-contained fridge contain-
ing helium-4, in which the helium-4 is pumped using a
sorp-
tion pump; this relies on the fact that charcoal adsorbs helium
at low temperatures, but the helium can be driven off by heat-
ing the charcoal to above 20 K. In operation, the charcoal is
heated and the helium driven into the 1 K pot. The charcoal is
then allowed to cool, pumping the helium back out of the 1 K
pot and providing cooling. The disadvantage of such a sys-
tem is that it does not provide continuous cooling, although a
system of two separate fridges can be used to provide quasi-
continuous cooling. Commercial mechnically cooled systems
are also available in which Joule-Thompson expansion of he-
lium is used to reach approximately 1.5 K.
b. Pumped helium-3 systems Helium-3 maintains a use-
ful vapour pressure to lower temperatures than helium-4, and
pumping on helium-3 can produce temperatures lower than
300 mK. Helium-3 does not occur in nature and is conse-
quently very expensive (more than 10 000 times as expensive
as helium-4). It must therefore be handled in a closed system,
unlike helium-4 which is commonly pumped away into the at-
mosphere.
Helium-3 refrigerators can be constructed using
an external gas handling system to pump on the helium-3 pot
and return it to the cryostat, or they can be self contained using
sorption pumps. It is usual to combine sorption systems with
a helium-4 sorption refrigerator to produce a self-contained
system which can be bolted onto the 4 K plate of a helium or
mechanically cooled cryostat to produce temperatures below
300 mK, albeit with a rather low cooling power. Such systems
are popular for cooling detector systems in astronomy, fusion
diagnostics and other fields.
c. Dilution refrigerators Cooling below 300 mK requires
more sophisticated techniques. The most common is a
di-
lution refrigerator. These are commercially available from
several companies in Europe, the US and Japan. I will not go
into the details of how they operate here (copious information
is available in many books (for example Refs. [2, 3]) other
than to say that cooling is produced by mixing helium-3 and
helium-4 at low temperatures. This can be used to achieve
temperatures of a few mK.
To operate a dilution fridge requires a helium-3 circulation
system similar to that used for a helium-3 fridge. Usually
a 1-K pot is used to pre-cool the incoming helium-3. How-
ever, a number of companies now offer dilution refrigerators
operating with a pulse tube cooler, rather than a liquid he-
lium bath. These have to operate without a 1 K pot since
there is no helium available to run one from. It turns out that
this is possible with a change in design along with modifica-
tions to the gas handling system at room temperature, but it
makes operation of the fridge somewhat more complicated.
Recently, self-contained dilution refrigerators using sorption
pumps have been developed.
d. ADRs An alternative to the use of a dilution fridge is
to use magnetic cooling, with an
adiabatic demagnetisation
refrigerator, or
ADR. The general principle is that a param-
agnetic salt is magnetised in a field of, say, 6 Tesla. This will
produce heating, which is taken away by contact with a helium
bath2 at a temperature of 4 K. This contact is arranged to be
via a thermal link which can be opened (either by a solenoid
or rod emerging from the cryostat). Reducing the field will
then cause the salt pill to cool, to temperatures below 100 mK.
Variations on this scheme are possible; to reduce the base tem-
perature and improve performance they can be operated from
an upper temperature of 1 K, or two or more salt pills can be
operated in series. The main disadvantages are low cooling
power and lack of continuous operation, although systems are
currently in development that use multiple salt pills to achieve
continuous operation. The magnetic fields required can also
interfere with experiments.
However, they have several advantages over dilution fridges.
Firstly, they are less prone to failing, as they do not have the
intricate system of pumping lines and capillaries that are prone
to plugging up, leaking, or both. At room temperature, all that
is needed is a high current power supply, rather than a sys-
tem of pipework, pumps and valves. Secondly, it is extremely
easy to change temperature. With a dilution fridge, it is easy
enough to warm a sample with a heater, but time-consuming
to cool it down again. With an ADR, any temperature can
be reached within a few seconds just by adjusting the mag-
netic field. For some experiments, this control is extremely
useful. Although until recently popular only for astronomical
2 Mechanical coolers can also be used, but are less ideal because they have a
fixed cooling rate, and a heat load larger than this will cause them to warm
up. A cryogen bath, on the other hand, has an essentially unlimited cooling
rate without warming above the “base” temperature – until the cryogen
runs out, of course.
5
instrumentation, commercial systems for materials character-
isation have appeared in the last few years, some coupled with
mechanical coolers to produce near-automatic operation from
room temperature to below 100 mK.
Adiabatic demagnetisation of metal nuclei can be used to pro-
vide cooling to below 1 mK, but that is outside the scope of
this section.
B. Types of cryostat
There are various types of cryostat that are used. There are
two basic possibilities: the sample to be measured can be im-
mersed in liquid helium, or it can be placed in an evacuated
chamber (usually called the
vacuum can). The simplest type
of cryostat for both types of measurements is a small device
which can be inserted into the neck of a helium transport de-
war. However, most experiments require something more so-
phisticated.
The first cryostats used glass dewars. These are limited in size,
and somewhat fragile, and are not in common use nowadays,
having been replaced by metal dewars.
There are two common configurations. In perhaps the most
popular one, the sample is located in a vacuum can surrounded
by liquid helium. This ensures that the outside of the vac-
uum can is at 4 K (or lower), but has the disadvantage that a
helium-tight seal must be made which works at low temper-
atures. This is usually achieved by making an O-ring out of
indium (a very soft metal); lead (e.g. fuse wire) can also be
used, but is harder to get right. Some people have success-
fully replaced indium O-rings with rings of Kapton R
©, a poly-
mer. In vacuum can systems, an
exchange gas is usually put
into the vacuum can during cool-down in order to speed ther-
malisation of the internal components. Helium is commonly
used, but this interferes with leak-checking if a leak from the
surrounding helium is suspected. Using helium-3 avoids this
problem, but is a somewhat expensive solution. Hydrogen can
be used down to 20 K, but has to be handled with care because
of the explosion risk when mixed with air, and of course ni-
trogen can be used as far as 77 K.
An alternative cryostat design has the experimental area out-
side the helium reservoir. (Of necessity, this approach is
adopted with a cryocooled system). This has the advantage
that the vacuum seal is made at room temperature, and can
thus use a conventional rubber O ring. Such joints are less
likely to fail, and are a lot easier and faster to make than a low
temperature seal. The main disadvantage is that since the vac-
uum space reaches the outside of the dewar, it is not possible
to speed up the cool-down of internal components by using an
exchange gas.
3. GETTING AND KEEPING YOUR SAMPLE COLD
So far I have discussed how to produce and maintain a cryo-
genic environment. In an experiment, there will generally be
something (I’ll call it the “sample”) that needs to be cooled. If
it can be placed in the cryogenic liquid, then successful cool-
ing is almost assured. In general, however, you will want to
put it in vacuum, in which case making good thermal contact
becomes important. In addition, the sample will be connected
to the outside world, almost certainly by wiring, and possibly
in other ways (such as via fill lines for a sample involving he-
lium). These connections are of course routes for unwanted
heat to enter the sample, and this has to be taken into con-
sideration. While heat transfer is usually the most important
consideration, heat capacity and thermal expansion can also
be important. Getting all these things right involves a knowl-
edge of material properties at low temperatures.
The good news is that there are many places in which material
property values can be looked up, including the popular cryo-
genics text books, some of which I listed in the introduction,
and in particular Ref. [1], which has an extensive selection
of tables and graphs. The bad news is that there is no sin-
gle comprehensive source of information, so for a relatively
obscure material you may have to search in several places,
and in the end go looking for individual papers. Ref. [6] lists
of lot of sources of information. The closest thing to single
comprehensive source is the set of books usually known as
Touloukian (e.g. Ref. [7]), but they are over 35 years out of
date, and not very good for temperatures below 1 K3. Even
worse, especially for thermal conductivity, there has been lit-
tle attempt to critically analyse measurements from the litera-
ture to produce recommended values.
There is a rather limited selection of information on the world
wide web; see Refs. [9] and [10]. The latter should be used
with some care, as for some materials, insufficient detail is
given on what the material actually is4. Useful information
on the general behaviour of material properties are given in
various textbooks, in particular Refs. [2] and [3].
A. Thermal conductivity
The material property which is usually of greatest interest is
thermal conductivity. Variation between materials at cryo-
genic temperatures is much greater than at room temperature,
with ranges of many orders of magnitude. For materials with
high conductivity, as the temperature is reduced from room
temperature, the conductivity will increase, reach a peak, and
then decrease. For other materials, the trend is generally
downwards with decreasing temperature. Values for a few
materials are shown in Fig. 1 as a guide to general behaviour;
more detailed information can be found in the sources men-
tioned above.
There are two main obstacles to looking up the conductivity
of a material:
3 A slightly more up to date version will be found in Ref. [8], but this only
covers elemental materials.
4 And note that you have to get the capital letters in the right place in the
URL!
6
Figure 1: Approximate thermal conductivity values for some com-
monly used materials (brass, constantan, copper-nickel, stainless
steel, Pyrex, G10 and Torlon 4203). Pure metals are not shown as
their conductivity varies hugely with conductivity (see Fig. 2). From
[3, 5, 11, 12].
– Many materials have just not been measured, or if they have
the results were either not reported, or appeared in a suffi-
ciently obscure place that you can’t find it.
– There can be considerable variation between samples of
nominally the same material, so the values you look up may
not be correct for the material you have. (And sometimes,
measurements in the literature are just wrong).
The latter point is particularly important in the case of pure
metals, where the conductivity can vary by many orders of
magnitude depending on purity. To take an example, the con-
ductivity of copper at 1 K can vary from below 50 W/m/K
to nearly 105 W/m/K depending on purity. It is true if one
has some idea of the purity it is possible to reduce the likely
range somewhat, but the uncertainty can still be considerable,
as can be seen from Figure 2. One solution to this problem is
to make use of the
Wiedemann-Franz law, which relates the
Figure 2: Recommended ranges for the conductivity of annealed alu-
minium with 4N, 5N and 6N purity (solid lines). Note that the lower
limits for 5N and 6N purity are the same. Recommended values
for vacuum annealed 5N purity copper are also shown, along with
the higher conductivity values possible following oxygen annealing
(dashed lines). From Ref. [13].
electrical and thermal conductivity of a metal:
κ =
LT
ρ
,
(1)
where L is the
Lorenz number, and T is temperature. For
sufficiently low temperatures, L is expected to take the the-
oretical value of L0 = 2.45 × 10−8 WΩK−2. Measuring
the resistivity of a sample is relatively straightforward, espe-
cially if you have a helium transport dewar around to dip the
sample into. However, very pure metals can require a rather
large current in order to get a measurable voltage drop. Mag-
net power supplies are very useful here, as they are designed
to put a large and well controlled current through a very low
resistance.
For alloys, the resistance is much easier to measure, and a
resistance bridge as used for reading resistance thermome-
ters can be used. However, for low conductivity alloys the
Wiedemann-Franz law will underestimate the thermal con-
ductivity, because the heat transported by electrons becomes
sufficiently small that conduction through the crystal lattice
will also be significant, and this has no electrical counterpart.
Once you have a value for conductivity, κ(T), calculating the
heat leak P for a material with area A and length L, between
temperature T1 and T2, is straightforward:
P =
A
L ∫
T2
T1
κ(T)dT.
(2)
Often we can represent conductivity as a power-law:
κ(T) = κ0Tβ,
(3)
and we can then write
P =
A
L
κ0
β + 1 (
Tβ+1
2
− T
β+1
1
) .
(4)
7
At sufficiently low temperatures (and 4 K is almost always
sufficiently low), metals will have a value of β = 1. Crys-
talline dielectrics will have a value around β = 3, and amor-
phous materials (such as glass) have a more complex be-
haviour, generally having an approximate value of β = 2 be-
low 1 K.
Finally, please remember the difference between
conductivity,
κ, measured in W/m/K, and an intrinsic property of a material,
and
conductance, G, measured in W/K, which is the property
of a particular thermal link. The two are related by
G =
A
L
κ.
(5)
It is easy to remember which is which if you compare to resis-
tance and resistivity, which people rarely mix up. (I don’t
know why conductivity and conductance are harder to get
right, but they are, and I still get it wrong myself on occasion.)
B. Contact conductance
Usually, there will be one or more joints between your sample
and the source of cooling. In the simplest case, the sample is
directly bolted to the bottom of the helium bath, or the cold
stage of a mechanical cooler. In more complex systems, there
may be a system of thermal links stretching for a metre or
more.
Making good thermal contact between two surfaces is not
straightforward, and at very low temperatures the conductance
across different types of joint can vary by over 9 orders of
magnitude! The general principle for good thermal contact is
to join clean surfaces with as high a pressure as possible. The
most common material for thermal contact is copper, which
forms an insulating oxide layer. This can be removed every
time a joint is made by filing the surface, but plating with gold
is a much more satisfactory solution. Pressure is normally ap-
plied by bolts or screws, and in general it is the total force that
is applied, not the area of the joint, that matters. Well prepared
joints can have a conductance that approaches that of the bulk
material. If it is not possible to apply large forces, or to have
clean contacts, a small amount of vacuum grease can help by
increasing the contact area. However, although it can make a
bad joint better, it is likely to make a good joint worse, and
the lower the temperature, the less effective it will be (since
the conductivity of the grease drops much more sharply with
temperature than for a metal).
C. Heat capacity and thermal expansion
Heat capacity and thermal expansion are also important,
though usually not as critical as thermal conductivity. Unlike
conductivity, they are not very sensitive to small variations in
material composition, so using values from the literature is
more accurate.
The heat capacity of most materials is approximately de-
scribed by the
Debye model, which characterises them by the
Debye temperature, ΘD, which is around room temperature
for many metals. Above the Debye temperature, heat capac-
ity approaches a temperature-independent value. Well below
this (say T < ΘD/25), the heat capacity varies as T3. This
is not the full story, since this only describes the heat capac-
ity due to lattice vibrations. Electrons will also contribute a
linear component, though this usually only becomes signifi-
cant at temperatures below 10 K. It is also important to know
that some non-metals, in particular those with an amorphous
structure, have very high heat capacities at low temperatures.
The thermal expansion coefficient reduces with decreasing
temperature, and very little takes place below 77 K. Further-
more, the temperature variation of thermal expansion is sim-
ilar for almost all materials. Reference [14] gives a criti-
cal analysis of thermal expansion data from the literature for
many materials, along with recommended values.
D. Thermal radiation
Everything emits thermal radiation. A body with
emissivity ǫ
and temperature T will emit energy at a rate of
E = ǫσT4,
(6)
where σ = 5.67 × 10−8Wm−2K−4 (Stephan’s constant). A
blackbody corresponds to ǫ = 1.
To reduce heat on our sample from thermal radiation, we can
do two things:
– Reduce the temperature of the surroundings.
– Reduce the emissivity of the surroundings.
We achieve both of the above by surrounding the sample with
a
radiation shield at as low a temperature as practical, and by
reducing the emissivity of the radiation shield as much as we
can. Some emissivity figures are given in the table [2, 15]:
Material
Emissivity
Polished copper
0.02-0.04
Lightly oxidised copper
0.1
Highly oxidised copper
0.6
Aluminium (pure and alloys) 0.01-0.06
Highly oxidised aluminium 0.2
Gold
0.015-0.03
Ice
a
0.9
aGreater than about 10µm in thickness
From this we can see that a polished copper or aluminium
shield would do nicely. However, such a shield will not stay
polished, particularly if handled, and in practise a gold plated
copper shield is the best solution. (Aluminium is a bit tricky to
make good thermal contact to, and is best avoided for shields
unless it is really necessary to save weight5).
It should be noted that the peak wavelength of thermal ra-
diation changes with temperature, and thus measurements of
5 It also superconducts below around 1 K, depending on alloy type, resulting
in a poor thermal conductivity.
8
emissivity at room temperature should be used with some cau-
tion to apply to cryogenic systems.
E. Other heat transfer mechanisms
1. Gas conduction
Conduction through gases can be a blessing and a curse. On
the one hand, introducing an exchange gas is a very useful
way of getting your experiment cold in the first place, when
different parts of the experiment will operate at different tem-
peratures and you have therefore carefully designed a poor
thermal contact between them. On the other, even a small
leak can add sufficient helium gas to what is supposed to be a
vacuum space to stop your experiment from working.
The conductivity through a gas does not depend on pressure,
unless the pressure is so low that it enters the ballistic regime,
where the mean free path becomes larger than the size of the
container. Therefore there is no need to get carried away with
adding exchange gas - a pressure of around 0.1 mbar at 300 K
is sufficient, and no matter how impatient you are getting,
adding more gas isn’t going to speed things up.
2. Heat switches
Sometimes it is not possible to use an exchange gas. There are
other tricks for making thermal contact between parts of the
experiment that must be thermally isolated when cold. You
can use a
heat switch, in which thermal contact can be turned
on and off. This can be done with a movable mechanical con-
tact, or with a tube full of gas which freezes out at the operat-
ing temperature. Other tricks include materials which go su-
perconducting at the operating temperature (superconductors
are very good conductors of electricity but not heat), some-
times referred to as a passive heat switch.
3. Film creep
Below 2.17 K, helium-4 goes superfluid. Superfluid helium
(sometimes referred to as helium-II), has many weird and
wonderful properties. One of them is that a superfluid film
will creep up the sides of a container containing helium-II. At
some point it reaches a temperature where it evaporates, be-
ing replaced by more helium flowing up through the film. This
means that if you pump on superfluid helium, the film will try
to escape up the pumping line. You stop this happening with
a restriction in the pumping line.
4. SOME OTHER ASPECTS OF EXPERIMENTAL DESIGN
Obviously the details of cryogenic experiments vary widely,
but there some aspects that are common to almost all low tem-
perature experiments.
A. Wiring
There are two main ways you can go wrong with wiring.
Firstly, you can end up dumping too much heat into your ex-
periment. This might stop it from doing what you want alto-
gether, or more subtly it may appear to work but you get the
wrong result (for example, heat leaking down thermometer
wires can cause the thermometer to read a higher tempera-
ture than whatever it is supposed to be measuring). You avoid
this by proper
heat sinking, also known as thermal anchoring,
generally achieved by wrapping the wire several times round
a copper bobbin. Depending on how cold your experiment
is, you may need to heat sink at several different temperature
stages.
The Wiedemann-Franz law says that electrical and thermal
conductivity are proportional to each other, so in principle
it doesn’t matter what material you use for wiring; you just
choose the cross-section to get an acceptable heat leak. In
practise this would mean ludicrously small cross-sections if
you used a pure metal such as copper, so alloys such as con-
stantan and manganin are used. These also have the advan-
tage that the change in resistivity as a function of temperature
is much smaller than for a pure metal. At low temperatures,
superconducting wire such as NbTi can be useful since it gets
you round the Wiedemann-Franz law, offering high electrical
conductivity and low thermal conductivity.
The second way you can go wrong is if the wiring goes open
circuit when the experiment is cold. When this happens, the
failure is usually at a soldered joint, though if you aren’t care-
ful thermal contraction could also cause a wire to snap. Since
there is no way of getting inside a cold cryostat, if it was a vital
wire, you have to warm up. Extreme care is therefore required
when soldering wires in a cryogenic experiment, and using an
eyeglass to inspect joints thoroughly and re-doing them if you
have any doubt whatsoever is not being too paranoid.
B. Choosing materials
The mechanical properties of materials used in cryostats are
important. Some metals go
brittle when cooled down. You
are safe with anything with an fcc (face centred cubic) struc-
ture, but in practise, low temperature labs usually stick to a
small number of well known materials for the majority of
uses. For metals, these are generally stainless steel, copper,
brass and – sometimes – aluminium). There is a larger number
of commonly used non-metals, including Teflon R
©, Mylar R
©,
Kapton R
©, G10 (the green stuff that printed circuit boards are
made from) and various epoxies.
Brass is a very useful material; it has a reasonable thermal
conductivity and is easy to machine. On the other hand, it
is not particularly well characterised, and sometimes recycled
which occasionally results in the inclusion of unwanted mate-
rials. Folklore says that extruded bar can contain longitudinal
cracks that could cause leaks, so anything that has to be leak-
light is best constructed from plate.
Copper comes in many forms. Oxygen Free High Conductiv-
9
ity (
OFHC) copper is popular in low temperature labs. OFHC
copper is useful if it is to be hydrogen brazed, as oxygen reacts
with the hydrogen resulting in porous joints. However, despite
a common feeling that if it has “High Conductivity” in the
name it must be good, the thermal (and electrical) conductiv-
ity is no better than regular commercial copper (electrolytic
tough pitch), and possibly worse. Beware of copper tubing;
although it is a tempting material to make radiation shields
from, it is likely to be phosphorous-deoxydised copper, which
has a very poor thermal conductivity.
Sometimes you want a good thermal but poor electrical con-
ductor. Sapphire does this quite nicely around 4 K, but is
not useful at ultra-low-temperatures. Since the conductivity
of a dielectric varies with temperature as T3, the conductivity
drops6 by a factor of over 60 000 between 4 K and 100 mK,
whereas that of a metal drops by a mere factor of 40.
Another time where it pays to get the right material is the
choice of tubing for transferring liquid nitrogen (and boil-off
helium gas). Cord-reinforced PVC tubing is easily available
and will do the job - for a while, but is rather prone to shat-
tering when cold. Silicone rubber is the answer; readily and
cheaply available (it’s used in the food industry), and so long
as you’re careful not to stress it when it’s cold, it rarely shat-
ters.
C. Grounding and RF
An area that is very important, but hard to get right, is elec-
trical grounding and the reduction of RF (radio frequency) in-
terference. RF interference can affect your experiment, and in
particular can cause thermometers to warm up and thus give
incorrect readings. It can be reduced by filtering wires where
they enter a cryostat, and in extreme cases putting the entire
cryostat in an electrically shielded room. Further filtering at
low temperatures can help. Getting grounding right is a com-
plex task and one where it really pays to find an expert. If none
are to hand, some advice on grounding (as applied to making
the film “Jaws”) can be found in Ref. [4].
5. WHEN THINGS GO WRONG
Many things can go wrong with a cryogenic experiment, and
here are two of the more popular ones.
A. Leaks
Leaks can be the bane of the low temperature physicists life,
especially with complex systems such as dilution fridges. If
6 The moral of this is to beware of people with experience of cryogenics
at 4 K “helping” you with your ultra-low-temperature experiment. It’s a
different world...
you are lucky(?) the leak is present at room temperature; it is
much harder to find a leak if it only appears when the cryostat
is cold. Unfortunately, this is all too common an occurrence.
Even worse is one which only lets superfluid helium through...
The main weapon in the hunt for leaks is the
leak detector; a
mass spectrometer which is designed to sense helium (fortu-
nately for us, helium is the standard gas used for leak detection
in industry). Searching for a leak requires experience and it is
a good idea to get someone who has done it before to help
you. A leak detector with a graphical display is very useful
for convincing yourself you really have found a leak.
There are various methods of leak-hunting, depending on
where the leak is. If it is possible to pump on the system con-
taining the leak (and the helium background inside it is not too
high), then helium from a gas cylinder can be squirted around
the various candidate locations for the leak through a narrow
tube while the leak detector is used to pump on the offending
system. Start at the top and work your way down, since he-
lium is lighter than air. Sometimes it can be useful to increase
the signal by putting a bag around a suspected leak and filling
it with helium.
If the leak only appears when the cryostat is cold you have
more of a problem. Usually (but not always) cooling to 77 K
is sufficient to find the leak, in which case it is sometimes
possible to find it by dunking the cryostat in liquid nitrogen,
then pulling it out sufficiently to squirt helium round. This is
not terribly gentle on the cryostat, though.
B. Blocked exhaust ports
A more urgent problem is a dewar or cryostat with an ice plug.
This is another time where your first reaction should be to get
experienced help. An ice plug can sometimes be freed by ju-
dicious prodding with a hollow metal pipe. If you do clear
the blockage, you are likely to get a large stream of high pres-
sure, cold gas, so wear appropriate protective equipment. It is
a good idea to make sure you have the appropriate equipment
ready before you cool down, as you don’t want to be running
round finding copper tubing etc. when you’ve just discovered
that your cryostat has plugged.
Clearing a blockage is potentially dangerous, and if you have
any doubts about what you’re doing, don’t do it. And it’s
much better to make sure you don’t get a blockage in the first
place.
6. WARMING UP!
Finally, at some point (perhaps days, weeks or even months)
after you cooled your cryostat down, it’s time to warm up.
This is easy compared with everything else, but sometimes it
can be worth putting heaters in suitable places to help speed
things up. This is obviously something best thought of before
you cool down.
And when you’ve got it warm, it is a good idea to make
sure that everything has warmed above 0◦C before cracking
10
the vacuum and opening up, otherwise everything inside will
cover itself with water frozen out of the air. If you
must open it
up early, then a heat gun can be used to drive off the moisture
and finish warming everything up.
[1] Jack W. Ekin.
Experimental techniques for low-temperature
measurements. Oxford University Press, Oxford, 2006.
[2] G. K. White and P. J. Meeson.
Experimental techniques in low-
temperature physics. Oxford: Clarendon Press, 4th edition,
2002.
[3] Frank Pobell.
Matter and Methods at Low Temperatures.
Springer, 1992.
[4] Robert C. Richardson and Eric N. Smith.
Experimental Tech-
niques in Condensed Matter Physics at Low Temperatures.
Addison-Wesley„ 1988.
[5] D. A. Wigley.
Materials for low-temperature use. Oxford Uni-
versity Press, 1978.
[6] J. G. Weisand II, V. Flynn, E. Thompson, and R. P. Reed. A
reference guide for cryogenic properties of materials. Tech-
nical Report SLAC-TN-03-023, SLAC, 2003. Available at
www.slac.stanford.edu/cgi-wrap/getdoc/slac-tn-03-023.pdf.
[7] Y. S. Touloukian, P. E. Liley, and S. C. Saxena.
Thermophysi-
cal properties of matter. Vol. 1. Thermal conductivity: metallic
elements and alloys. Plenum, 1970.
[8] C. Y. Ho, R. W. Powell, and P. E. Liley. Thermal conductivity
of the elements.
J. Phys. Chem. Ref. Data, 3 Suppl. 1, 1974.
[9] http://researchmeasurements.com/about.html.
[10] http://cryogenics.nist.gov/MPropsMAY/material%20properties.htm.
[11] O. V. Lounasmaa.
Experimental principles and methods below
1K. Academic, London, 1974.
[12] G. K. White.
Experimental techniques in low-temperature
physics. Clarendon Press, Oxford, 1979.
[13] A. L. Woodcraft. Recommended values for the thermal con-
ductivity of aluminium of different purities in the cryogenic to
room temperature range, and a comparison with copper.
Cryo-
genics, 45(9):626–636, 2005.
[14] R. J. Corruccini and J. J. Gniewek. Thermal expansion of tech-
nical solids at low temperatures.
NBS monograph number 29.
[15] Ed. B. Hands.
Cryogenic Engineering. Academic Press, Lon-
don, 1984.
Copyright cO 2007 Adam L. Woodcraft
