J Low Temp Phys (2011) 164:179236

DOI 10.1007/s10909-011-0373-x
Basic Operation of Cryocoolers and Related Thermal
Machines
A.T.A.M. de Waele
Received: 3 March 2011 / Accepted: 15 May 2011 / Published online: 10 June 2011
The Author(s) 2011. This article is published with open access at Springerlink.com
Abstract This paper deals with the basics of cryocoolers and related thermodynamic
systems. The treatment is based on the rst and second law of thermodynamics for
inhomogeneous, open systems using enthalpy ow, entropy ow, and entropy produc-
tion. Various types of machines, which use an oscillating gas ow, are discussed such
as: Stirling refrigerators, GM coolers, pulse-tube refrigerators, and thermoacoustic
coolers and engines. Furthermore the paper deals with Joule-Thomson and dilution
refrigerators which use a constant ow of the working medium.
Keywords Thermodynamics Cryocoolers Thermoacoustics Dilution
refrigerators
1 Introduction
This paper deals with the basics of cryocoolers and related thermodynamic systems.
A cryocooler is a standalone cooler of table-top size which is used to cool some
particular application. Reference [1] is a recent review. In this paper the essence of
the system operation will be discussed. Whenever possible mathematical complica-
tions will be avoided. Distracting practical effects, such as high-order disturbances,
streaming, heat leaks, transient effects, etc. will be neglected. The starting point will
be the rst and second law of thermodynamics for inhomogeneous, open systems [2].
Important concepts as enthalpy ow, entropy ow, and entropy production will be
introduced. Thoroughly understanding thermal machines (coolers and heat engines)
without these concepts is impossible.
After the introduction and a short discussion of ideal regenerators and heat ex-
changers, various types of coolers are discussed which operate with oscillating ows.
A.T.A.M. de Waele ()
Eindhoven University of Technology, PO Box 513, 5600MB Eindhoven, The Netherlands
e-mail: [email protected]
180 J Low Temp Phys (2011) 164:179236
Fig. 1 (Color online) General representation of a system that consists of a number of subsystems. The
interaction with the surroundings of the system can be in the form of exchange of heat and other forms of
energy, exchange of matter, and change of shape. The interactions between the subsystems are of a similar
nature and lead to entropy production. In this gure the

V
k
stand for dV
k
/dt
Next, properties are discussed of less ideal regenerators, followed by a treatment of
thermoacoustic coolers and engines. The paper ends with the discussion of Joule-
Thomson coolers and dilution refrigerators which operate with a steady ow of the
working uid. Appendix A gives some useful thermodynamic formulae, Appendix B
is a derivation of the so-called volume-ow equation, and Appendix C is about the
harmonic model.
2 The First and Second Law of Thermodynamics
The laws of thermodynamics apply to well-dened systems. Figure 1 is a general rep-
resentation of a thermodynamic system. We consider systems which can be inhomo-
geneous. We allow heat and mass transfer across the boundary (nonadiabatic, open
systems), and we allow the boundary to move. In our formulation we will assume
that heat and mass transfer and volume changes take place only at some well-dened
regions of the system boundary. Equations (1) and (4) are not the most general formu-
lations of the rst and second law. E.g. kinetic energy terms are missing and exchange
of matter by diffusion is excluded.
2.1 First Law
The rst law reads
dU
dt
=

Q
k
+

H
k

k
p
k
dV
k
dt
+P. (1)
J Low Temp Phys (2011) 164:179236 181
In (1) t is time,
U is a function of state, called the internal energy of the system.

Q
k
are the heat ows into the system at the various regions of the boundary which
are labeled with k.

H
k
are the enthalpy ows into the system due to the matter that ows into the
system. It is dened as

H
k
=

n
k
H
mk
=

m
k
h
k
, (2)
where

n
k
is the molar ow of matter owing into the system and H
mk
its molar
enthalpy,
1
h
k
the specic enthalpy (i.e. enthalpy per unit mass), and

m
k
=M
k

n
k
(3)
the mass ow with M
k
the molar mass.
dV
k
/dt are the rates of change of the volume of the system due to the various
moving boundaries, p
k
is the pressure behind boundary k.
P takes into account all other forms of power applied to the system by its environ-
ment (such as electrical, shaft power, etc.).
We use the notation

Y for the ow of a thermodynamic state function Y and dY/dt

for the rate of change of Y. Even though the dimensions of

Y and dY/dt are the

same their physical meaning is distinctly different. For the heat ow we use the dot
notation.
2.2 Second Law
The second law reads
dS
dt
=

Q
k
T
k
+

S
k
+

S
ik
with

S
ik
0. (4)
Here
S is a function of state, called the entropy of the system.
T
k
represent the temperatures at which the heat ows

Q
k
enter the system.

S
k
represent the entropy ows into the system, due to matter owing into the sys-
tem, given by

S
k
=

n
k
S
mk
=

m
k
s
k
. (5)
Here S
mk
and s
k
are the molar and specic entropy of the matter, owing into the
system, respectively.
1
We will use the lower index m to indicate molar quantities. In many thermoacoustic papers the lower
index m is used to indicate the mean value. For the latter we will use the lower index 0.
182 J Low Temp Phys (2011) 164:179236
Fig. 2 Schematic diagram of a heat engine. A heating power

Q
H
enters the system at some high tempera-
ture T
H
, and

Q
a
is released at ambient temperature T
a
. A power P is produced and the entropy production
rate is

S
i

S
ik
represent the entropy production rates due to internal processes. Each of the
entropy production rates is always positive or zero. This is an essential aspect of
the second law. The summation is over all processes in the system. The most im-
portant irreversible processes, for our application, are heat ow over a temperature
difference and mass ow over a pressure difference.
In many cases

Q/T is also considered as an entropy ow which is associated with
the heat ow. In this case the second law is a conservation law with ow and source
terms.
2.3 Consequences of the First and Second Law
2.3.1 Heat Engines
Figure 2 is a schematic diagram of a heat engine. The machine is cyclic and in the
steady state. A heating power

Q
H
enters the engine at a temperature T
H
and a heat
ow

Q
a
leaves it at ambient temperature T
a
. A power P is produced. The sign con-
ventions of P and the heat ows are chosen in such a way that their values are positive
under normal system operation. They differ from the signs dened in Fig. 1.
Due to the irreversible processes inside the engine entropy is produced at a rate

S
i
.
After one cycle, the state of the engine is the same as at the beginning of the cycle,
thus, on average, dU/dt = 0 and dS/dt = 0. The system is closed so

H = 0 and

S =0. The boundaries of the system are xed so dV

k
/dt =0. As a result the rst law
for the engine reduces to

Q
H

Q
a
=P (6)
and the second law to
0 =

Q
H
T
H

Q
a
T
a
+

S
i
with

S
i
0 (7)
J Low Temp Phys (2011) 164:179236 183
or

S
i
=

Q
a
T
a

Q
H
T
H
0. (8)
If

Q
a
would be zero condition (8) would reduce to

S
i
=

Q
H
T
H
0. (9)
As T
H
> 0 and

Q
H
0 condition (9) would not be satised. So

Q
a
cannot be zero.
This means that a heat engine can operate only if heat is released at a lowtemperature.
This is the well-known Kelvin formulation of the second law.
Eliminating

Q
a
from (7), with (6), gives
P =
_
1
T
a
T
H
_

Q
H
T
a

S
i
. (10)
As

S
i
0 we must require
P
_
1
T
a
T
H
_

Q
H
. (11)
The efciency of a heat engine is dened as
=
P

Q
H
. (12)
With (11) it follows
1
T
a
T
H
. (13)
This famous relation shows that the efciency of all thermal engines has a maximum
given by the Carnot efciency dened as

C
=1
T
a
T
H
. (14)
From (10) it can be seen that the efciency of all heat engines is reduced due to the
term T
a

S
i
. This important quantity is called the dissipated energy.
2.3.2 Refrigerators
Refrigerators, as depicted in Fig. 3, can be treated in a similar way as heat engines.
The rst law gives

Q
a
=P +

Q
L
(15)
and the second law
0 =

Q
L
T
L

Q
a
T
a
+

S
i
with

S
i
0 (16)
184 J Low Temp Phys (2011) 164:179236
Fig. 3 Schematic diagram of a refrigerator.

Q
L
is the cooling power at some low temperature T
L
, and

Q
a
is released at ambient temperature T
a
. A power P is supplied to the system and

S
i
is the entropy production
rate
or

S
i
=

Q
a
T
a

Q
L
T
L
0. (17)
Eliminating

Q
a
from (17) by (15) gives

S
i
=
P +

Q
L
T
a

Q
L
T
L
0. (18)
If P would be zero then (18) would reduce to

S
i
=
_
1
T
a

1
T
L
_

Q
L
0. (19)
This condition is not satised since both

Q
L
0 and T
a
> T
L
. This means that heat
can not ow from a low temperature to a high temperature without doing work. This
is Clausius formulation of the second law.
Turning back to (18), we see that
P =
T
a
T
L
T
L

Q
L
+T
a

S
i
. (20)
As

S
i
0 we must require
P
T
a
T
L
T
L

Q
L
. (21)
The Coefcient Of Performance (COP) of coolers is dened as
=

Q
L
P
. (22)
J Low Temp Phys (2011) 164:179236 185
With (21) follows

T
L
T
a
T
L
. (23)
This relation shows that the COP of all refrigerators has a maximum given by

C
=
T
L
T
a
T
L
. (24)
This quantity is called the Carnot COP of refrigerators. From (20) it can be seen that
the COP of the cooler is reduced by the dissipated energy T
a

S
i
.
2.3.3 The Volume-Flow Equation
From the rst law a very useful expression can be derived which we will call the
volume-ow equation. In case of a periodic pressure change, with angular frequency
, the characteristic length scale for temperature variations of the working uid is
given by the thermal penetration depth

given by

=
2V
m
C
p
. (25)
Here is the thermal conductivity, V
m
the molar volume, and C
p
the molar heat
capacity at constant pressure. Viscous effect are characterized by the viscous pene-
tration depth

given by

=
2

(26)
with the viscosity and the density. The two penetration depths are related by the
Prandtl number
P
r
=
C
p
M
, (27)
with M the molar mass, as follows

=P
r

. (28)
For many working uids P
r
is practically constant and of order 1, so the two penetra-
tion depths are of the same order of magnitude. For helium gas P
r
0.66.
For helium, at typical working conditions of cryocoolers, the thermal penetration
depth is in the 0.10.5 mm range. Many spaces in thermal machines, such as the
compression and expansion spaces in Stirling coolers, the pulse tube in a pulse-tube
refrigerator, etc. are much larger than this, so the bulk of the gas is thermally isolated
from the surroundings. In other words, in many cases, the pressure changes in the
bulk of the gas are adiabatic. Furthermore the pressure p in these spaces is practically
homogeneous. Finally, at not too low temperatures and not too high pressures, the gas
can be considered to be an ideal gas. Under these conditions a very useful relation
186 J Low Temp Phys (2011) 164:179236
Fig. 4 Gas ows with varying
pressure: (a) inlet and outlet;
(b) inlet and piston; (c) xed
volume with valve with ow
conductance C
can be derived (Appendix B) which reads as follows
V
p
dp
dt
=

V
k

dV
k
dt
. (29)
Here V is the volume of the system,

V
k
are the volume ows at various positions k,
given by

V
k
=

n
k
V
mk
, (30)
the dV
k
/dt are the rates of change of the volume due to moving boundaries at position
k (usually a piston), and dp/dt is the rate of change of the pressure.
If, in addition to the conditions mentioned above, the system is closed all

V
k
=0
so, with (181), Equation (29) reduces to
V
p
dp
dt
=
dV
dt
(adiabatic, closed). (31)
From this relation we nd
pV

=constant (adiabatic, closed). (32)

Equation (32) has the same form as the well-known Poisson relation but it has a
much wider validity since it includes systems with an inhomogeneous temperature
distribution. The Poisson relation is just a special case of (32).
In Fig. 4 three situations are depicted. For the case of Fig. 4a, in which gas ows
into and out of a control volume with xed volume V and with a varying pressure
p(t ), (29) reads

V
1
=

V
2
+
V
p
dp
dt
. (33)
This relation holds, in particular, for the pulse tube in a pulse-tube refrigerator (PTR).
For the case of Fig. 4b gas ows into a control volume with, on the other side, a
J Low Temp Phys (2011) 164:179236 187
Fig. 5 Heat conduction. The
dotted line represents the
system. A heat ow

Q
1
enters at
T
1
and

Q
2
leaves at T
2
moving piston with area A and velocity v. Now (29) reads

V
1
=vA+
V
p
dp
dt
. (34)
For the case of Fig. 4c, where gas ows out of a container through a valve with ow
conductance C (as in the buffer volume of a PTR),
0 =C(p p
0
) +
V
p
dp
dt
. (35)
2.4 Entropy Production Rates
Now we will derive the expressions for the entropy production rates due to ows of
heat and matter.
2.4.1 Heat Conduction
Consider the heat conduction through an adiabatic bar in the steady state (see Fig. 5).
It is a closed system with xed boundaries and no external work is done, so the rst
law reduces to

Q
1
=

Q
2
=

Q. (36)
The second law gives
0 =

Q
1
T
1

Q
2
T
2
+

S
i
with

S
i
0 (37)
or, with (36),

S
i
=

Q
_
1
T
2

1
T
1
_
0. (38)
Relation (38) is the expression of the entropy production rate due to transport of heat
over a temperature difference. As

Q > 0, Equation (38) demands that T
1
T
2
. In
other words: heat ows from a high temperature to a low temperature. This is again
the Clausius formulation of the second law.
The heat ow in a bar of length L and cross-sectional area A can, for small tem-
perature differences, be written as

Q=
A
L
(T
1
T
2
) (39)
188 J Low Temp Phys (2011) 164:179236
with the thermal conductivity. The entropy production rate is now

S
i
=
A
L
(T
1
T
2
)
2
T
1
T
2
. (40)
The dependence of

S
i
on the driving force (T
1
T
2
) is quadratic, which is charac-
teristic for expressions of the entropy production rates in general [2].
2.4.2 Throttling
Now we derive an expression for the entropy production in the throttling process as
shown in Fig. 6. The process is adiabatic by denition. In the steady state

n
1
=

n
2
=

n (41)
The rst law demands that
0 =

n
1
H
m1

n
2
H
m2
(42)
so, in a throttling process, the molar enthalpy is constant
H
m1
=H
m2
. (43)
The second law gives
0 =

n
1
S
m1

n
2
S
m2
+

S
i
. (44)
With (43) and (163) we get
_
S
m
p
_
H
m
=
V
m
T
(45)
so, with (44),

S
i
=

n
_
2
1
V
m
T
dp. (46)
Introducing the volume ow rate

V =

nV
m
(47)
Fig. 6 Schematic diagram of the throttling process. The dotted line represents the (adiabatic) system.
A molar ow

n
1
enters at position 1 and

n
2
leaves at 2
J Low Temp Phys (2011) 164:179236 189
this may also be written as

S
i
=
_
2
1

V
T
dp. (48)
For small pressure drops, or in cases where

V/T is practically constant (as in liquid

ow through a valve),

S
i
=

V
T
(p
1
p
2
). (49)
With the ow conductance C

V =C(p
1
p
2
) (50)
we get

S
i
=
C(p
1
p
2
)
2
T
. (51)
3 Ideal Regenerators and Heat Exchangers
In this section we will discuss ideal regenerators and ideal heat exchangers.
3.1 Ideal Regenerator
An important component of refrigerators, operating with oscillatory ows, is the re-
generator. A regenerator consists of a matrix of a solid porous material, such as spher-
ical particles or metal sieves, through which gas ows, as shown in Fig. 7. The matrix
must have a high heat capacity and a good heat contact with the gas. At the same time
it should have a low ow resistance. These are conicting requirements.
The thermodynamic and hydrodynamic properties of regenerators are extremely
complicated, so one must make simplifying assumptions. The degree of idealization
differs from case to case. In its most extreme form an ideal regenerator has the fol-
lowing properties:
1. the heat capacity of the matrix per unit volume is much larger than of the gas;
2. the heat contact between the gas and the matrix is perfect;
3. the ow resistance of the matrix is zero;
Fig. 7 Schematic picture of a
regenerator
190 J Low Temp Phys (2011) 164:179236
4. the porosity g, this is the fraction of the volume taken by the matrix material, is
zero;
5. the thermal conductivity in the ow direction is zero;
6. the gas in the regenerator is ideal.
In an ideal regenerator the entropy production rate is zero. The average enthalpy
ow in the regenerator is

H
r
=
1
t
c
_
t
c
0

n
r
H
m
(T )dt =

n
r
H
m
, (52)
where t
c
is the cycle time. The bar indicates time average. If the working uid is an
ideal gas then, with (175),

H
r
=C
p

n
r
T . (53)
If conditions 1 and 2 are satised then the gas temperature varies with position but at
a certain point it is constant in time so

H
r
=C
p
T

n
r
. (54)
In the steady state the time-average molar ow

n
r
in the regenerator is zero (in fact
everywhere in the system) so

n
r
=0. (55)
As a result of (54) and (55) the average enthalpy ow in an ideal regenerator is zero

H
r
=0. (56)
The fact that the average enthalpy owin an ideal regenerator with an ideal gas is zero
implies that it has no cooling power. Any heat load on the regenerator has nowhere
to go and can only lead to an increase of the local temperature. As we will see later,
this differs for an ideal regenerator with a nonideal gas.
Depending on the situation one or more assumptions, which model the ideal re-
generator, may be dropped. Usually they are replaced by other assumptions of a less
rigorous nature. They will lead to nonzero enthalpy ow and entropy generation in
the regenerator. This will be the topic of later sections.
The development of regenerator materials with a large heat capacity per m
3
around
10 K [3] is one of the main reasons why the performance of cryocoolers is improved
so much in recent times. A collection of specic heat data is found in Fig. 8 [4].
3.2 Ideal Heat Exchanger
An ideal heat exchanger has zero ow resistance and the temperature of the gas,
leaving the heat exchanger, is equal to the (xed) body temperature T
X
of the heat
exchanger. However, even a perfect heat exchanger cannot affect the temperature T
i
J Low Temp Phys (2011) 164:179236 191
Fig. 8 Heat capacities per unit volume, c
V
, of various important regenerator materials as functions of
temperature. (GAP is Gd
2
AlO
3
; SS is stainless steel)
of the incoming gas. This leads to a fundamental form of entropy production, which
we now will calculate.
The heat transfer rate from the gas, with ow rate

n > 0 and with gas tempera-
ture T , to the heat-exchanger body, over a small section of the exchanger, is

Q=

nC
p
T, (57)
with the associated entropy production rate, according to (38),

S
X
=

Q
_
1
T
X

1
T
_
. (58)
The total entropy production rate when the gas is cooled from T
i
to T
X
is

S
X
=

n
_
i
X
_
1
T
X

1
T
_
C
p
dT. (59)
If T
i
T
X
this reduces to

S
X
=

nC
p
(T
i
T
X
)
2
2T
2
X
. (60)
Since both T
i
T
X
and

n are rst-order terms the entropy production rate in the heat
exchanger

S
X
is of third-order and can be neglected in low-amplitude considerations.
192 J Low Temp Phys (2011) 164:179236
Fig. 9 Schematic diagram of a Stirling cooler. The system has one piston at ambient temperature T
a
and
one piston at low temperature T
L
. The upper half shows the entropy ows and the lower half the energy
ows
4 Stirling Refrigerators
We will now turn to the treatment of the most important types of cryocoolers and
related thermal machines. We will start with the Stirling-type refrigerators. The basic
type of Stirling-type cooler is depicted in Fig. 9. At the left it consists of a piston,
compression space, and heat exchanger, all at ambient temperature T
a
. Next comes a
regenerator. At the right there are a heat exchanger, the expansion space, and a piston,
all at the low temperature T
L
. The gas is compressed at ambient temperature and ex-
panded at low temperature. The thermal contact with the surroundings, left and right,
at the temperatures T
a
and T
L
is supposed to be so good that the compression and
expansion are isothermal. The work, performed during the expansion, is recovered.
The working uid is helium.
The cooling cycle is divided in four steps as depicted in Fig. 10. We start a cycle
when the two pistons are in their most left positions, the cold piston touches the cold
heat exchanger:
1. From a to b. The warm piston moves to the right over a certain distance while the
position of the cold piston is xed. The compression at the hot end is isothermal
by denition, so a certain amount of heat Q
a
is given off to the surroundings at
temperature T
a
.
2. From b to c. Both pistons move to the right so that the volume between the two
pistons remains constant. The gas enters the regenerator at the left with temper-
ature T
a
and leaves it at the right with temperature T
L
. During this part of the
cycle heat is given off by the gas to the regenerator material. During this process
the pressure drops and heat has to be supplied to the compression and expansion
spaces to keep the temperatures constant.
3. From c to d. The cold piston moves to the right while the position of the warm
piston is xed. The expansion is isothermal so heat Q
L
is taken up from the appli-
cation.
J Low Temp Phys (2011) 164:179236 193
Fig. 10 (Color online) Four
states in the Stirling cycle
Fig. 11 pV -diagram of the
ideal Stirling cycle
4. From d to a. Both pistons move to the left so that the total volume remains con-
stant. The gas enters the regenerator at the right with temperature T
L
and leaves
it at the left with T
a
so heat is taken up from the regenerator material. During this
process the pressure increases and heat has to be extracted from the compression
and expansion spaces to keep the temperatures constant. In the end of this step the
state of the cooler is the same as at the start.
In the pV diagram (Fig. 11) the cycle consists of the well-known form of two
isotherms and two isochores. The volume V in this diagram is the volume between
the two pistons. At each point of the cycle the pressure in the system and the volume
are well-dened. However, during the steps 2 and 4 the temperature of part of the gas
is T
a
and of the other part it is T
L
.
In practice the cycle is not divided in discrete steps as described above. Usually
the motion of both pistons are driven by a common rotary axes which makes the
motions of the two pistons harmonic. It is typical for the Stirling cycle that the phase
difference between the motion of the two pistons is about 90

.
The cycle is reversible and, with the surroundings of the cooler, heat is exchanged
only at two xed temperatures, so the COP is the Carnot COP given by (24). This
194 J Low Temp Phys (2011) 164:179236
can also be seen as follows. In Fig. 9 the entropy and energy ows are indicated.
Based on the rst law the power P
H
, supplied to the warm piston, is equal to the
heat ow

Q
a
to the surroundings, so P
H
=

Q
a
. The power P
L
, recovered at the cold
piston, is equal to the cooling power

Q
L
, so P
L
=

Q
L
. The second law, applied to the
regenerator and the two heat exchangers, gives

Q
a
T
a
=

Q
L
T
L
. (61)
If the power P
L
, released by the expansion, is used to reduce the net power P, sup-
plied to the system, so P =P
H
P
L
, this results in a COP of
=

Q
L
P
H
P
L
=
T
L
T
a
T
L
. (62)
The cold piston, as described above, is rather impractical, so, in many cases, the
cold expander is avoided by using a displacer. A displacer is a solid body which
moves back and forth and drives the gas back and forth between the warm and the
cold end of the system through the regenerator. Ideally the pressure over the displacer
is zero, so no work is required to move it. Its motion is synchronized with the mo-
tion of the piston. Typically it is ninety degrees out of phase. Also in this case the
cycle is reversible and heat is exchanged, with the surroundings, only at two xed
temperatures, so the efciency is also the Carnot efciency given by (24).
Another type of Stirling cooler is the split-pair type [5] as shown in Fig. 12. It con-
sists of a compressor, a split pipe, and a cold nger. Usually the compressor is a linear
compressor where the piston is driven by an AC magnetic eld as in loudspeakers.
There are often two pistons which move in opposite directions to reduce mechanical
vibrations. The pistons are suspended by so-called exure bearings which provide
Fig. 12 Schematic diagram of a split-pair Stirling refrigerator. The cooling power is supplied to the heat
exchanger of the cold nger. Usually the heat ows are so small that there is no need for physical heat
exchangers around the split pipe
J Low Temp Phys (2011) 164:179236 195
stiffness in the radial direction and exibility in the axial directions. The piston and
the compressor casing dont touch so no lubricants are needed.
In the cold nger the moving part is the regenerator which works as the displacer
at the same time. It is suspended by a spring which can also be a exure bearing.
The motion of the displacer/regenerator is driven by the pressure drop between the
upper space (expansion space) and the lower space and by the pressure difference
between the bouncing volume and the cold-nger volume. Hence, the area of the
cross section of the guiding rod is an important design parameter. The cooler operates
at a frequency near the resonance frequency of the mass-spring system inside the
cold nger. The motions of the piston and the displacer/regenerator are similar to the
motions of the displacer-type Stirling cooler.
5 GM-Refrigerators
We will now describe the workhorse of many low-temperature systems: the Gifford-
McMahon (GM) cooler [6]. GM coolers are robust machines that nd wide-spread
application e.g. in MRI and cryopumping. Figure 13 is a schematic diagram. The
working uid is helium at pressures in the 10 to 30 bar range. The cold head contains
a regenerator and a displacer which usually are combined in one body. The varying
pressure is obtained by connecting the cold head periodically to the high- and low-
pressure sides of a compressor through a rotating valve which is synchronized with
the motion of the displacer.
During the opening and closing of the valves irreversible processes take place, so
GM-coolers are intrinsically irreversible. This is a clear disadvantage of this type of
machine. On the other hand the cycle frequencies of the compressor and the displacer
are uncoupled. E.g. the compressor runs at power-line frequency while the cycle of
the cold head is 1 Hz. This means that the swept volume of the compressor can be
50(60) times smaller than of the cooler. Basically cheap compressors of domestic
refrigerators can be used, but special precautions have to be taken to prevent over-
heating of the compressor since they are not designed for helium. In addition very
high quality purication traps have to be installed to prevent oil vapor from entering
the regenerator.
The cycle can be divided in four steps, with Fig. 14, as follows: The starting po-
sition is with the low-pressure valve closed, the high-pressure valve open, and the
Fig. 13 Schematic diagram of a
GM-cooler. V
l
and V
h
are buffer
volumes of the compressor. The
compression heat is removed by
the cooling water of the
compressor via a heat
exchanger. The rotary valves
alternatingly connect the cooler
to the high- and the low-pressure
sides of the compressor and runs
synchronous with the displacer
196 J Low Temp Phys (2011) 164:179236
Fig. 14 The four stages in the
cooling cycle of the GM cooler
displacer all the way to the right (so in the cold region). All the gas is at room tem-
perature.
1. From a to b. The displacer moves to the left while the cold head is connected to
the high-pressure side of the compressor. The gas is forced to pass the regenerator.
It enters the regenerator at ambient temperature T
a
and leaves it with temperature
T
L
so heat is given off by the gas to the regenerator material. Due to the high
density of the low-temperature gas some additional gas will ow from the high-
pressure side of the compressor through the regenerator to the low-temperature
space.
2. From b to c. The cold head is connected to the low-pressure side of the com-
pressor with xed position of the displacer. Part of the gas ows through the re-
generator to the low-pressure side of the compressor. Expansion of the gas takes
place. The expansion in the cold space is isothermal so heat is taken up from
the application. During this phase of the cycle the useful cooling power is pro-
duced.
3. From c to d. The displacer moves to the right while the cold head is still con-
nected to the low-pressure side of the compressor forcing the cold gas to pass the
regenerator, while taking up heat from the regenerator.
4. From d to a. The cold head is connected to the high-pressure side of the com-
pressor with xed position of the displacer. In the end of this step the cycle is
closed.
The ideal cooling power can be obtained immediately from the rst law, applied
to the expansion space with volume V
e
and pressure p
e

Q
L
=p
e
dV
e
dt
. (63)
We have seen before that the average enthalpy ow, in an ideal regenerator, is zero.
The energy

Q
L
leaves the cold end of the system as enthalpy transported by the
displacer.
The ow distribution in regenerators is homogeneous. As a result the cooling
power tends to be proportional to the area of the cross section of the regenerator.
This is illustrated in Fig. 15 where the length of GM coolers and the external di-
ameter of their heat exchangers are plotted as functions of the square root of the
J Low Temp Phys (2011) 164:179236 197
Fig. 15 Length L and diameter
D of the GM-coolers of
Cryomech plotted as functions
of
_

Q at 80 K [7]
cooling power at 80 K,

Q
1/2
[7]. The length is practically constant and the di-
ameter (minus 13 mm) is nicely proportional to,

Q
1/2
. So, increasing the cooling
power is simply a matter of increasing the diameters of the cooler components.
However, instabilities were found in regenerators with a large diameter/length ra-
tio [8].
6 Pulse-Tube Refrigerators
6.1 Components of PTRs
This section gives a description of the basic operation of a Pulse-Tube Refrigerator
(PTR) in the steady state.
2
This type of cooler was invented by Mikulin [11]. At the
moment there are many variants of PTRs. A so-called Stirling-type single-orice
PTR is represented schematically in Fig. 16. From left to right it consists of:
1. A piston which moves back and forth.
2. A heat exchanger X
1
(after cooler) where heat is released at room temperature
(T
a
) to cooling water or to the surroundings.
3. A regenerator.
4. A heat exchanger X
L
at low temperature (T
L
) where heat is absorbed from the
application.
5. A tube, often called the pulse tube.
6. A heat exchanger X
3
to room temperature (T
a
).
7. A ow resistance (orice).
8. A buffer volume, in which the pressure p
B
is practically constant.
In this section all ow resistances are neglected except from the orice. The ow
conductance C of the orice is adjusted for optimum performance. Typically the
2
The operation of this system should not be confused with the operation of the so-called basic pulse tube
whose operation is based on a heat shuttle interaction with the tube wall [9, 10].
198 J Low Temp Phys (2011) 164:179236
Fig. 16 (Color online) Schematic diagram of a Stirling-type single-orice PTR. From left to right: piston,
after cooler (X
1
), regenerator, low-temperature heat exchanger (X
L
), tube (pulse tube), second room-tem-
perature heat exchanger (X
3
), orice (O), buffer. The dotted rectangle represents thermal (vacuum) insu-
lation
dimensionless ow resistance of the orice , dened by
=
V
t
Cp
0
, (64)
is of order one. Here is the angular frequency and p
0
the average pressure. The
system is lled with helium at p
0
of typically 20 bar. The part in between the heat
exchangers X
1
and X
3
is below room temperature.
6.2 Cooling Principle
Driven by the piston, the gas moves back and forth and the pressure p
t
in the pulse
tube varies. The operation frequency is typically 1 to 50 Hz and the working uid
is helium at, say, 10 to 30 bar. The pressure varies smoothly. Acoustic effects, such
as traveling pressure waves, or fast pressure changes, are absent. In the regenerator
and in heat exchangers the gas is in good thermal contact with its surroundings while
in the pulse tube the gas is thermally isolated. In the pulse tube compression of gas
leads to heating and expansion to cooling.
In and around the pulse tube we can distinguish three types of gas parcels. There
are gas parcels that, during a cycle, move in and out the pulse tube via the cold heat
exchanger X
L
, gas parcels that move in and out via the hot heat exchanger X
3
, and
gas parcels that never leave the pulse tube. Together the latter form the so-called gas
piston. Schematic drawings of the temperature-position curves of these three types of
gas parcels are given in Fig. 17. At the hot end gas ows from the buffer via the orice
into the tube with a temperature T
a
if p
t
<p
B
. If p
t
=p
B
the gas at the hot end stops
and if p
t
> p
B
the gas moves back towards the hot end of the tube and, eventually,
through the heat exchanger X
3
and the orice into the buffer. So gas elements enter
the pulse tube if p
t
< p
B
and leave it if p
t
> p
B
. Consequently, at the hot end, the
gas leaves the tube with a temperature higher than the inlet temperature T
a
and heat
is released via the heat exchanger X
3
to the surroundings.
At the cold end the analysis is a bit more complicated due to the fact that the
velocity v
L
at the cold end is determined by the velocity v
H
of the gas at the hot end
J Low Temp Phys (2011) 164:179236 199
Fig. 17 Left side: a gas element
enters the tube at temperature
T
L
and leaves it at a lower
temperature hence producing
cooling. Right side: a gas
element enters the tube at
temperature T
a
and leaves it at a
higher temperature hence
producing heating. Middle: this
gas element is in the gas piston
and never leaves the pulse tube.
It moves to the right with high T
and to the left with low T
and by the elasticity of the gas column in the tube, according to (33),
v
L
=v
H
+
V
t
A
t
p
t
dp
t
dt
(65)
with V
t
and A
t
the volume and area of the pulse tube respectively. Still the situation
at the cold end is basically the same as at the hot end. At the cold end the gas enters
the tube with temperature T
L
when the pressure is high. It returns to X
L
when the
pressure is low and the temperature is below T
L
. Hence producing cooling.
Also gas parcels in the gas piston move to the right with a high temperature and
back with a low temperature. As we will see in the next section this is reason why
there is a net enthalpy ow in the pulse tube.
6.3 Thermodynamics of PTRs
In this Subsection we will analyze the PTR based on the rst and second law of ther-
modynamics. In the ideal case entropy is produced only in the orice. In all the other
subsystems

S
i
=0. The heat ows

Q with the surroundings are nonzero only in the
heat exchangers. Flows to the right are counted positive. The signs of the power and
the heat ows are dened in Fig. 18. We split up the PTR in subsystems going from
right to left. First we consider the simple system represented in Fig. 19a, containing
only the orice. At the reservoir side the pressure p =p
B
and the temperature T =T
a
are constant. As a result the molar entropy S
m
and the molar enthalpy H
m
are con-
stant as well. Similar to (55) it follows that the average enthalpy and entropy ows in
the tube, connecting the orice with the buffer, are zero. In the tube, connecting the
orice and the heat exchanger X
3
, the temperature is constant as well so the average
enthalpy ow here is also zero. The rst law for the orice shows that the heat ow
into the orice, needed to keep its temperature constant, is zero as well.
200 J Low Temp Phys (2011) 164:179236
Fig. 18 Schematic diagram of a single-orice PTR. In the upper half of the gure the entropy ows are
indicated. The lower half gives the energy ows. If a particular ow is zero in a certain region this is
indicated by 0. It is assumed that the PTR is ideal so that dissipation only takes place in the orice
Fig. 19 Thermodynamic systems containing the orice (a), the heat the exchanger X
3
(b), the pulse tube
and its heat exchangers (c), and the regenerator and its heat exchangers (d)
The average entropy owin the tube between the heat exchanger X
3
and the orice

S
3
is nonzero due to the pressure variations. With (173) we may write it as

S
3
=R

nln
p
t
p
0
. (66)
Since the heat ow to the orice is zero the entropy production rate in the orice is

S
O
=

S
3
. (67)
J Low Temp Phys (2011) 164:179236 201
So the entropy, which is produced in the orice, ows towards X
3
as shown in
Fig. 19a.
Next we consider the exchanger X
3
as a thermodynamic system (Fig. 19b). In the
pulse tube the gas moves back and forth isentropically, so the average entropy ow
here is zero
3

S
t
=0. (68)
The second law with (67) gives that the average heat ow, extracted at X
3
, is given
by

Q
H
=T
a

S
O
. (69)
So heat has to be extracted at X
3
. From the rst law this must be equal to the average
enthalpy ow in the pulse tube

H
t
=

Q
H
. (70)
It is interesting to note that the entropy ow to system b comes from the right and the
enthalpy ow from the left.
Now we consider the system containing the tube and its two heat exchangers
(Fig. 19c). The average enthalpy ow in the pulse tube is given by

H
t
=C
p

nT . (71)
The gas moves to the right with a high temperature and to the left with a low tem-
perature (see the gas parcel in the gas piston in Fig. 17). As a result the net enthalpy
transport in the pulse tube

H
t
is nonzero. However, the enthalpy ows at the left and
right of system c are zero (see (56)). The rst law then gives that

Q
L
=

Q
H
. (72)
This important relation shows that the cooling power is equal to the heat released
at the warm heat exchanger X
3
. It seems to contradict the Clausius formulation of
the second law applied to the system Fig. 19c. However, the Clausius principle only
applies to closed systems.
The second law, applied to X
L
, with (68) gives

Q
L
=T
L

S
r
, (73)
where

S
r
is the entropy ow in the regenerator. The negative sign means that the
average entropy ow in the regenerator is directed from the cold to the warm end.
3
That the entropy ow is zero is not obvious for surfaces in the neighborhood of one of the heat exchangers
since gas parcels can exchange heat with the heat exchanger when they move back and forth. This problem
can be solved by starting to make up the balance of the entropy, transported through the surface in a cycle,
when the rst parcel moves into the pulse tube. All parcels passing the surface will return with the same
entropy.
202 J Low Temp Phys (2011) 164:179236
This is logical since the gas moves to the right when the pressure is high (low en-
tropy, see (173)) and to the left when the pressure is low (entropy high). In the
ideal case there is no entropy production in the regenerator so the average entropy
ow

S
r
=constant. (74)
Now consider a system consisting of the regenerator and its two adjacent heat
exchangers X
1
and X
L
as in Fig. 19d. The gas in the (adiabatic) compression space
moves back and forth isentropically, so the average entropy ow left of the after
cooler X
1
is zero. Combining (73) and (74) gives

Q
L
T
L
=

Q
c
T
a
. (75)
The rst law requires that the average heat release at the aftercooler

Q
c
=P. (76)
Here P is the power input in the compressor. The COP (see (22)) is found by com-
bining (75) and (76) which gives the well known [12] result
=
T
L
T
a
. (77)
Comparing this value with the Carnot COP, given by (24), shows that the efciency
of an ideal PTR is less than of an ideal cooler. This is due to the dissipation in the
orice. Close to room temperature the difference between and
C
is big, but at
temperatures in the range of liquid nitrogen (77 K) and lower the difference in COP
is insignicant for most applications.
Combining (69) and (72) shows that

Q
L
=T
a

S
O
. (78)
So the cooling power is determined by the dissipation at the hot end of the tube which,
in turn, is determined by the component of the oscillating ow which is in phase with
the pressure. The ow at the cold end contains a term which is not in phase with the
pressure variations (see (65)). This component is bad for the efciency as it adds to
the dissipation in the regenerator but not to the cooling power. Furthermore the out-
of-phase component of the gas ow is a load on the regenerator. Most of the research
and development work in PTSs aimed at suppressing the out-of-phase ow at the
cold end of the pulse tube. The phase difference between the pressure and the ow
at the cold end can be controlled with proper devices such as: double inlet, inertance,
four-valve, active buffer, and warm expander [1316].
6.4 The Various Types of PTR
So far we have discussed the Stirling-type single-orice PTR (Fig. 16). The pres-
sure variations are generated by a compressor which is directly connected to the
J Low Temp Phys (2011) 164:179236 203
Fig. 20 Schematic diagram of a GM-type PTR. The compressor delivers constant pressure levels p
l
and p
h
. The varying pressure in the cooler unit is obtained through a rotating valve which alternatingly
connects the regenerator to the low- and high-pressure sides of the compressor
cold head. Typical operation frequencies are 20 to 50 Hz and the temperature range
50 K and higher. However, for getting cooling, the source of the pressure varia-
tions is unimportant. PTRs for lower temperatures (20 K and below) usually oper-
ate at low frequencies (1 to 2 Hz) and with pressure variations from 10 to 25 bar
(pressure ratio 2.5). At room temperature the swept volume per cycle would be
very high (up to one liter and more). Therefore the compressor is uncoupled from
the cooler just like in the GM-coolers. A schematic diagram is given in Fig. 20.
A system of valves is needed, which alternatingly connects the high pressure and
the low pressure to the hot end of the regenerator. Usually this is a rotating valve.
The high-temperature part at the compressor side is the same as in GM-coolers
(see Fig. 13). Therefore, this type of PTR is called a GM-type PTR. The gas ows
through the valves are accompanied by losses which are absent in the Stirling-type
PTR.
PTRs can be classied according to their shape. If the regenerator and the tube
are in line (as in Figs. 16 and 20) we talk about a linear PTR. The disadvantage of the
linear PTR is that the cold spot is in the middle of the cooler. For many applications it
is preferable that the cooling is produced at the end of the cooler. By bending the PTR
we get a U-shaped cooler as shown in Fig. 21. Both hot ends can be mounted on the
ange of the vacuum chamber at room temperature. This is the most common shape
of PTRs. For some applications it is preferable to have a cylindrical geometry. In that
case the PTR can be constructed in a coaxial way so that the regenerator becomes a
ring-shaped space surrounding the tube.
The lowest temperature, reached with single stage PTRs, is just above 10 K [17].
However, one PTR can be used to precool the other as in Fig. 21. Note that the hot
end of the second tube is connected to room temperature and not to the cold end of the
rst stage. In this clever way it is avoided that the heat, released at the hot end of the
second tube, is a load on the rst stage. In applications the rst stage also operates as
a temperature-anchoring platform for e.g. shield cooling of superconducting-magnet
cryostats. Matsubara and Gao were the rst to cool below 4 K with a three-stage
PTR [18]. With two-stage PTRs temperatures of 2.1 K, so close to the -point of
helium, have been obtained. With a three-stage PTR 1.73 K has been reached using
3
He as the working uid [19].
It is also possible to couple two independent PTRs where one precools the other.
In these systems there can be no uncontrolled internal circulations (DC-ow) of the
204 J Low Temp Phys (2011) 164:179236
Fig. 21 Two-stage, U-shaped,
GM-type, double-orice PTR.
The rst stage precools the
second one. Note that the hot
end of the second stage is
connected to room temperature
and not to the cold end of the
rst stage. The so-called minor
orices, which suppress a
possible DC ow, are not shown
gas and each of the PTRs can be optimized separately. In this kind of system the
world record of low temperature in PTRs (1.27 K) is reached while the second sys-
tem was operated with
3
He as the working uid [20]. With a superuid vortex cooler,
precooled by this system, the temperature has been lowered to 1.19 K with the po-
tential of obtaining 0.7 K in this way [21]. For lower temperatures one needs
3
He
cryostats or dilution refrigerators which will be discussed later.
7 Real Regenerators
Cryocoolers can reach temperatures below 20 K where helium is not an ideal gas.
Furthermore ideal regenerators, as described above, do not exist in practice. For the
most general case numerical models must be used [2224]. In this Section we will
discuss rather realistic regenerators, but there will be still some idealizations. We will
limit the discussion to the linear approximation, using the harmonic model. In the
harmonic approximation regenerators can be mathematically described by the trans-
fer matrix. This takes into account the viscosity of the working uid, the porosity of
the matrix, the ratio of the two end temperatures, the hydraulic diameter of the pores,
the length and cross section of the regenerator, the frequency of the oscillations, and
the gas pressure as in (211) and following [25].
7.1 Temperature Variations
An expression for the temperature variation of the gas is derived in Ref. [26]. Ne-
glecting the heat conduction term (this can be treated separately) Equation (31) in
Ref. [26] reads
gC
p
V
m
T
g
t
=

n
r
A
r
_
C
p
T
g
l
+H
p
p
l
_
+gT
g

V
p
t
+(T
r
T
g
). (79)
J Low Temp Phys (2011) 164:179236 205
Here g is the porosity (void fraction) of the matrix, H
p
is dened by (165), is the
heat-exchange coefcient per unit volume, and
V
the volumetric expansion coef-
cient (see (167)). The rst term in the right-hand side is the convective contribution,
the second term is due to the compression of the gas (see (162)), and the last term due
to heat ow from the matrix to the gas. In complex form, using that the p/l-term
would give a second-order contribution,
i
gC
p
V
m0

T
g
=
n
r
C
p
A
r
dT
0
dl
+igT
0

V0
p
_

T
g

T
r
_
. (80)
The index 0 indicates time-averaged values.
For the matrix temperature, again neglecting heat conduction,
(1 g)c
V
T
r
t
=(T
g
T
r
) (81)
where c
V
the heat capacity per unit volume of the bulk regenerator matrix material as
given in Fig. 8. Equation (81), in complex form, gives

T
r
=

i(1 g)c
V0
+

T
g
. (82)
Equations (80) and (82) can be used to calculate the enthalpy ow in the regenerator.
Together with the heat conduction, and requiring that the total average energy ow is
constant, the temperature prole of the regenerator can be calculated.
7.2 Nonideal-Gas Effects in PTRs
7.2.1 Introduction
In Sect. 3 it was shown that an ideal regenerator, with an ideal gas, has no cooling
power i.e. heat, applied somewhere to the regenerator, will lead to an unlimited in-
crease of the local temperature. This is different if the working uid is a real gas. In
this Section, we will consider the situation that heat

Q
r
(l) is supplied externally to
the regenerator as shown in Fig. 22.
In the nonideal-gas effect the pressure dependent contribution of the enthalpy, H
p
in (164), plays a key role. The H
p
values for
4
He can be obtained from Ref. [27]
and for
3
He from Refs. [28, 29]. Examples at 15 bar are given in Fig. 23. For a Van
Fig. 22 Schematic drawing of
the regenerator with a heat load
d

Q
r
/dl per unit length,
distributed along the matrix, and
a cooling power

Q
L
at the cold
end
206 J Low Temp Phys (2011) 164:179236
Fig. 23 Plots of H
p
at 15 bar
for
4
He and
3
He as functions of
temperature
der Waals gas the values of H
p
at T = 0 and T = are the same and equal to the
parameter b, the total volume of the atoms. Figure 23 also shows that helium is not
an ideal gas (H
p
=0) even at the high temperatures.
Since nonideal-gas effects are most relevant in the second stage of a PTR we will
consider the regenerator of the second stage of a PTR. It will be treated as ideal.
However, it cannot be assumed that the thermal conductivity in the ow direction
is zero since without heat ow energy conservation cannot be satised. But we will
assume that the ow resistance and the void volume are zero, that the thermal contact
between the gas and the matrix is perfect, and that the heat capacity of the matrix is
very big. Furthermore we assume that the pressure variations are harmonic
p =p
0
+p
A
cos t (83)
with p
A
p
0
. As the ow resistance is neglected p
A
is constant. Finally we assume
that the molar ow is in phase with the pressure and given by

n
r
=n
A
cos t. (84)
As the void volume of the regenerator is zero n
A
is constant. That the ow is in phase
with the pressure can be achieved by proper ow-controlling devices at the hot end
of the second-stage pulse tube. The expressions, in the following subsections, will be
treated to lowest relevant order. This means that pressure and temperature changes
will be treated in rst order, and energy-related quantities such as enthalpy ow and
cooling powers to second order. The discussion in this section regards PTRs, but can
also be applied to other types of coolers.
7.2.2 Cooling Power
The index L will be used to indicate values at the cold end of the regenerator. The
cooling power of the second stage can be calculated from the rst law, applied to X
L
,
0 =

Q
cL
+

Q
L
+

H
rL

H
t
. (85)
J Low Temp Phys (2011) 164:179236 207
Here

Q
cL
is the heat ow on X
L
by conduction via the regenerator,

Q
L
the applied
heating power,

H
rL
the average enthalpy ow from the regenerator, and

H
t
the aver-
age enthalpy ow in the pulse tube. At the regenerator side of X
L
the temperature is
constant so, using (164),

H
rL
=H
pL

np. (86)
At the pulse-tube side the entropy is constant so (163) gives

H
t
=V
mL

np. (87)
Combining (85), (86), and (87) gives

Q
L
=(V
mL
H
pL
)

np

Q
cL
. (88)
With (83), (84), and (166) we get

Q
L
=
1
2
n
A
p
A
V
mL
T
L

VL

Q
cL
. (89)
With the volume-ow amplitude
U
A
=V
mL
n
A
(90)
we get

Q
L
=
1
2
U
A
p
A
T
L

VL

Q
cL
. (91)
So the cooling power depends on
VL
. Even if

Q
cL
= 0 the cooling power

Q
L
is
zero if
VL
= 0. So the lowest temperature, that can be reached with PTRs, is the
temperature where
V
=0. For
4
He this is just above the lambda point and for
3
He
at 15 bar it is 1.04 K [29]. The lowest temperature, reached in experiment, is 1.27 K,
[20] so very close to the theoretical minimum.
In Ref. [30] it is derived that the COP of an ideal PTR, for T
L
below about 7 K, is
given by
=
V
mL

VL
V
mH
T
L
. (92)
Equation (92) shows that the COP is determined by V
mL

VL
, which is larger for
3
He
than for
4
He (see Fig. 24). Also the difference in thermal conductivity and viscosity
of
3
He and
4
He plays a role.
So far we focused on the regenerator. One may wonder how the real-gas effect
affects the energy ow in the pulse tube. The answer is in (87) which can also be
written as

H
t
=

V
L
p. (93)
208 J Low Temp Phys (2011) 164:179236
Fig. 24 The product V
m

V
at
15 bar for
3
He and
4
He as
functions of T
The values of

Vp at the cold and the hot end of the pulse tube are the same, so the
enthalpy ow in the pulse tube is unaffected by the fact that we are dealing with a
nonideal gas.
7.2.3 Temperature Prole in the Regenerator
In the previous subsection we have seen that the cooling power of PTRs is reduced
due to real-gas effects. Fortunately some of this cooling power can be recovered by
using the cooling power of the regenerator which is nonzero in the case of real gases.
This can be understood as follows: the average total energy ow E
r
in the regenerator
is the sum of the heat ow

Q
c
=A
r

r
dT
dl
, (94)
(with
r
the thermal conductivity of the regenerator) and the average enthalpy ow,
so
E
r
=

Q
c
+

H
r
. (95)
If

Q
r
(l) is the total amount of heat, supplied externally to the regenerator between 0
and l, energy conservation requires that
E
r
(l) =E
r
(0) +

Q
r
(l) . (96)
With (164) isothermal enthalpy changes, as in our ideal regenerator, are equal to
H
m
=H
p
p, so, with (83) and (84) the time average enthalpy ow in the regenerator
is given by

H
r
=
1
2
n
A
p
A
H
p
. (97)
J Low Temp Phys (2011) 164:179236 209
Fig. 25 Temperature proles in
case heat is supplied at the
middle of the regenerator. The
heating powers are indicated in
watt at the corresponding curves
Due to our assumptions of zero ow resistance and zero void volume, n
A
p
A
does not
depend on l. These relations result in
A
r

r
dT
dl
=
1
2
n
A
p
A
H
p
(T ) E
r
(0)

Q
r
(l). (98)
The heat input per unit length d

Q
r
(l)/dl can be distributed evenly e.g. if the re-
generator is used to precool an incoming ow of
4
He (as in Ref. [31]) or
3
He (as
in Fig. 36). The heat input can also be discrete (as in Refs. [32, 33]). We will con-
sider here the case where heat is supplied only in the middle of the regenerator (so at
l =L
r
/2). Integration of (98) is performed numerically. The values of T
L
and T
H
are
xed (here at 4 and 50 K respectively) and, by iteration, the value of E
r
(0), which sat-
ises this boundary condition, is determined. Next the temperature prole is obtained
from (98). In our calculations we take L
r
=0.15 m, A
r
=20 cm
2
, and the operating
frequency =2 Hz in a PTR with p
0
=15 bar. The effective thermal conductivity is
taken as
r
=2.5 W/Km and
1
2
n
A
p
A
=0.13 mol MPa/s.
Figure 25 gives the temperature proles in the regenerator for four values of the
heating power

Q
r
supplied in the middle of the regenerator. For zero

Q
r
the tem-
perature prole is rather at at the low-temperature end. So

Q
cL
is practically zero
and the load on the X
L
from the regenerator side is mainly due to enthalpy ow. The
temperature T
i
of the point where the heat is supplied increases with

Q
r
and a kink in
the T -prole appears. Figure 26 gives plots of

Q
L
and T
i
as functions of

Q
r
. At low

Q
r
values the

Q
L
remains fairly constant (hence free cooling power) at a value of
about 0.42 W. For

Q
r
> 2.7 W

Q
L
is negative, which means that the temperature of
4 K cannot be maintained without external cooling.
8 Thermoacoustics
The eld of thermoacoustics is pioneered by Rott [34] and by Swift [35] and his co-
workers. See Ref. [36] and the references therein. In order to thoroughly understand
thermoacoustic machines it is important to understand sound in terms of temperature-
position variations rather than the usual pressure-velocity variations. Therefore this
Section starts with a description of sound in these terms. In ordinary speech the sound
intensity is 65 dB which corresponds with pressure variations of 51 mPa, displace-
ments of 0.2 m, and temperature variations of 43 K. In thermoacoustic systems,
210 J Low Temp Phys (2011) 164:179236
Fig. 26 The cooling power

Q
L
and the temperature T
i
as
functions of

Q
r
. The dotted
lines represent the temperatures
of the hot and the cold ends
with sound levels of 180 dB, the pressure variations are 0.3 bar, displacements more
than 10 cm, and the temperature variations 24 K. Although a pressure ratio of 30%
is very big for acoustic systems, this is still much smaller than the pressure ratios
obtained e.g. in GM-type refrigerators, where pressure variations of a factor 3 are
common.
8.1 Sound
The wave equation in one dimension reads

2
v
x
2


2
v
c
2
t
2
=0 (99)
with v the gas velocity and c the sound velocity satisfying
c
2
=
p
0

0
. (100)
For an ideal gas
c
2
=
RT
0
M
, (101)
with M the molar mass. In these expressions p
0
, T
0
, and
0
are the average pressure,
temperature, and density respectively. Important special cases of solutions of (99) are
monochromatic plane waves which are superpositions of traveling waves to the right
and to the left
v =v
Ar
cos(
r
k(x ct )) +v
Al
cos(
l
+k(x +ct )). (102)
By proper shifts in the t - and x-axis and with
=kc (103)
this can be put in a simpler form
v =v
Ar
cos(t kx) +v
Al
cos(t +kx). (104)
J Low Temp Phys (2011) 164:179236 211
The pressure variations are given by
p =c
0
[v
Ar
cos(t kx) v
Al
cos(t +kx)]. (105)
The deviation x of the gas-particle position from its equilibrium position x can be
obtained by integration of (104) and is given by
x =
v
Ar

sin(t kx) +
v
Al

sin(t +kx). (106)

With (105) and (169) the temperature variations are
T =
cM
C
p
[v
Ar
cos(t kx) v
Al
cos(t +kx)]. (107)
Equations (106) and (107) form a parametric representation of a tilted ellipse in
the T x plane with t as the parameter. If v
Ar
=v
Al
we have a pure standing wave
and the ellipse of the T x plot is reduced to a straight line with slope
dT
dx
=
cM
C
p
tan(kx). (108)
Figure 27a is a graph of the position and the velocity amplitude of a pure standing
wave in a half-wavelength tube together with the plot of the pressure and temperature
amplitudes. At the tube ends the displacement is zero while the temperature variation
is maximal, so the T x plot is a vertical line here. All points, representing the T
x in time, move back and forth over the lines, as shown in Fig. 27b, so the average
enthalpy transport, by a standing wave, is zero.
If we would put a thin horizontal metal plate (with negligible longitudinal heat
conduction) in the sound eld the thermal interaction of the moving gas with the
plate leads to the thermoacoustic effects. The interaction with the gas would lead to
a temperature prole in the plate which exactly matches the lines in Fig. 27b which
have the so-called critical temperature gradients. If we would force (by heat exchang-
ers) the actual temperature gradient in the plate to be smaller than the local slopes of
Fig. 27 a: Plot of the
amplitudes of the velocity and
displacements, and the pressure
and temperature variations in a
half-wavelength tube of a pure
standing wave. b: corresponding
T x plots of a standing wave.
c: T x plots of a pure
traveling wave
212 J Low Temp Phys (2011) 164:179236
the T x plots, due to heat exchange with the surroundings, we have a cooler (or
heat pump). If the temperature gradient in the plate is forced to be larger then we have
an engine.
If v
Al
=0 or v
Ar
=0 we have pure traveling waves. With the proper scales of the
axis, (106) and (107) represent circles in the T x diagram. Figure 27c represents
the T x plots of a pure traveling wave to the right. The gas elements move to
the right with high temperature and back with a low temperature, so there is a net
transport of enthalpy. It is interesting to compare the T x plots with similar plots
in the pulse tube of a PTR (Fig. 17).
8.2 Energy Transport by Sound
Sound can transport energy. In general the enthalpy ow is given by (2). With the
molar ow

n =

V
V
m0
=
p
0
RT
0

V (109)
and the volume ow

V =Av (110)
with A the tube cross section, and (175) we get for the average enthalpy ow

H =
p
0
C
p
RT
0
AvT . (111)
As the expansion and compression are isentropic we use (169) to get the well-known
relation of the enthalpy ow, which is also the acoustic energy ow

H =Avp. (112)
Note that this relation only holds for small amplitudes. Therefore, in general, (112)
cannot be used to calculate the energy transport in the pulse tube of a PTR.
The velocity is given by (104) and the temperature by (107) so substitution in
(111) gives

H =
p
0
2c
A(v
2
Ar
v
2
Al
). (113)
This elegant relation shows that the enthalpy ow is independent of the x-coordinate
(which is a consequence of energy conservation) and proportional to the difference
in the velocity amplitudes squared of the positive and the negative sound wave.
8.3 Thermoacoustic Systems
Perhaps the simplest type of thermoacoustic device (from the point of view of con-
struction) is depicted in Fig. 28a. It represents a so-called thermoacoustic prime
mover. Typically it is 20 cm long with two cm diameter. Another important com-
ponent is the so-called stack. Usually this is a stack of metal plates (stainless steel) at
J Low Temp Phys (2011) 164:179236 213
Fig. 28 a: Schematic diagram
of a thermoacoustic prime
mover; b: schematic diagram of
a thermoacoustic refrigerator
a mutual distance 2y
0
which is several times the thermal penetration depth

of the
gas. The stack can also be a plug of loosely packed stainless steel wool or screens.
There is also a heat exchanger to ambient temperature and a resonator tube. It is
heated at the left e.g. by a propane ame. If the temperature at the hot side is high
enough the system produces a loud sound.
Figure 28b represents a thermoacoustic refrigerator. It is similar to the prime
mover, but it has a driver (loudspeaker) at the left. The systems of Fig. 28a and b
are so-called quarter-wavelength tubes which means that they resonate at a frequency
so that the length of the tube is equal to a quarter of the wavelength of the sound. The
quarter wavelength corresponds with the left half of Fig. 28a,b where the slopes of
the T x lines of the standing wave are negative. In other words: the left side of a
horizontal plate in the sound eld is hotter than the right side. Although the conus of
the loudspeaker in Fig. 28b is moving it is close to a velocity node, so its amplitude
is small.
The operation of standing-wave systems relies on the time delay between gas
transport and heat transport. They are in essence irreversible machines, so they have
an inherently reduced efciency. The mathematical description of the operation of
these so-called standing-wave systems is more complicated than of traveling-wave
systems. Therefore we will start with the discussion of traveling-wave systems.
8.3.1 Traveling-Wave Systems
This subsection treats the basics of so-called traveling-wave thermoacoustic engines
in general terms. The engine, described in the paper by Backhaus and Swift [38], can
serve as a model system. Figure 29a is a schematic drawing. It consists of a resonator
tube and a loop which contains a regenerator, several heat exchangers, a compliance,
a connecting tube, a pulse tube (also called thermal buffer tube), and a section, with a
smaller diameter, called the inertance. As the energy ow in the regenerator is small
(ideally it is zero), the main energy ow in the loop is from the hot heat exchanger via
the pulse tube, the inertance, and to the main heat exchanger. This can be considered
as energy transport via a traveling wave as in Fig. 27c. Hence the name traveling-
wave systems.
One of the fascinating properties of thermoacoustic systems is that they produce
sound if the hot side is hot enough. Reference [39] describes transient effects theo-
retically. The set of rst-order differential equations, which describes the dynamics
214 J Low Temp Phys (2011) 164:179236
Fig. 29 (a) Active end of the thermoacoustic engine. The symbols are explained in the text. (b) Model
system. a, b, c, d, e, and f are position indicators. The arrows dene the directions of positive ow
of the individual components, is combined to a single high-order (fourth-order) dif-
ferential equation which determines the time dependences of all dynamic variables.
It is subsequently solved analytically. Unfortunately the full mathematical treatment
is rather elaborate, even for a simplied model, and beyond the scope of this paper.
Here only a description will be given of the basic principles.
The gas in the system starts to oscillate spontaneously with a certain angular
frequency if the temperature T
t
of the hot heat exchanger is above an onset tem-
perature T
O
. If T
t
<T
O
oscillations in the system die out with a certain decay time. If
T
t
is increased the decay time becomes longer and longer. If T
t
>T
O
the oscillations
grow. Only if T
t
=T
O
there are steady oscillations. The amplitude is determined by
the heating power

Q
t
applied to the hot heat exchanger (minus the heat loss due to
conduction to the surroundings). In principle T
t
= T
O
, independent of

Q
t
if

Q
t
is
high enough. On the way to the steady state spectacular transient effects are possi-
ble, with temperature overshoots and bursts of high-intensity sound, which are not
systematically studied so far.
As thermoacoustic systems tend to oscillate at their resonance frequency they have
to be big to keep the frequency low. There are several interesting ideas to reduce the
frequency by introducing solid [40] or liquid masses [41]. By doing so thermoacous-
tic systems merge with the more classical type of Stirling systems as the free-piston
Stirling machines discussed above.
Now we turn to the steady-state operation and apply the harmonic model to de-
rive the onset temperature and operation frequency. Our model system is depicted in
Fig. 29b and consists of an open resonator tube R and a loop containing a bypass tube
B, a regenerator r, and a pulse-tube section t. It is assumed that all tubes have the same
cross sectional area A. So there is no section in the loop with a smaller diameter (in-
ertance) nor sections with a large diameter (compliance) [42]. The temperatures are
assumed to be close to ambient temperature T
a
except for inside and near the regen-
erator. Dissipative effects are neglected everywhere except in the regenerator.
J Low Temp Phys (2011) 164:179236 215
For the bypass channel B, with length L
B
, holds (225)
_
p
c

U
c
_
=T
B
_
p
d

U
d
_
(114)
with
T
B
=
_
c
B
i
c
0
A
s
B
i
A
c
0
s
B
c
B
_
(115)
with c the sound velocity and
0
the gas density at T
a
. In order to get compact ex-
pressions c
B
is used for cos(kL
B
) and s
B
for sin(kL
B
) (with k given by (103)) and
a similar notation for the pulse tube (label t) and the resonator tube (label R). By
replacing the index B by t or R the transfer matrices of the pulse tube T
t
and the
resonator tube T
R
are obtained respectively. For the regenerator, with assumed (real)
ow resistance R
r
and zero porosity, holds (207)
_
p
d

U
d
_
=T
r
_
p
e

U
e
_
(116)
with
T
r
=
_
1 R
r
0
T
a
T
t
_
. (117)
For the loop c, d, e, f we write
_
p
c

U
c
_
=T
lp
_
p
f

U
f
_
(118)
with T
lp
given by
T
lp
=T
B
T
r
T
t
. (119)
Matrix multiplication with (115), a similar relations for T
t
, and (117) results in
T
lp
=
_
_
c
B
c
t

T
a
T
t
s
B
s
t
+
iA
c
0
c
B
R
r
s
t
c
B
R
r
c
t
+i
c
0
A
c
B
s
t
+i
T
a
T
t
c
0
A
s
B
c
t
iA
c
0
s
B
c
t

A
2
c
2

2
0
R
r
s
B
s
t
+
iA
c
0
T
a
T
t
c
B
s
t
T
a
T
t
c
B
c
t
s
B
s
t
+
iA
c
0
R
r
s
B
c
t
_
_
.
(120)
Furthermore
det T
lp
=det T
B
det T
r
det T
t
=
T
a
T
t
. (121)
At the intersection point
p
b
= p
c
= p
f
p (122)
gives, with (118),
p = T
11lp
p +T
12lp

U
f
(123)
216 J Low Temp Phys (2011) 164:179236

U
c
= T
21lp
p +T
22lp

U
f
. (124)
So
p =
T
12lp
1 T
11lp

U
f
(125)
and, with (121),

U
c
=

T
a
T
t
+T
22lp
1 T
11lp

U
f
. (126)
For the resonator tube
p
b
= c
R
p
a
+i
c
0
A
s
R

U
a
, (127)

U
b
= i
A
c
0
s
R
p
a
+c
R

U
a
. (128)
For an open resonator tube, as discussed here,
p
a
=0. (129)
With (127), (128), and (125) this gives

U
b
=
c
R
ic
0
s
R
T
12lp
1 T
11lp

U
f
. (130)
Mass conservation at the intersection gives

U
f
=

U
b
+

U
c
. (131)
With (126) and (130) we see that nonzero solutions can be found only if
0 =1 +T
11lp
+
Ac
R
ic
0
s
R
T
12lp

T
a
T
t
+T
22lp
. (132)
Substituting the matrix elements from (120) gives
0 = c
B
c
t
s
B
s
t
+c
B
c
R
s
R
s
t
1 +
T
a
T
t
_
c
B
c
t
s
B
s
t
+c
R
s
B
s
R
c
t
1
_
+
iAR
r
c
0
_
c
B
s
t
+s
B
c
t
c
B
c
R
s
R
c
t
_
. (133)
This relation leads to rather simple expressions for the oscillation frequency and the
onset temperature as follows: from the real part we get
T
a
T
t
=
c
B
c
t
s
B
s
t
+
c
R
s
R
c
B
s
t
1
1 c
B
c
t
+s
B
s
t

c
R
s
R
s
B
c
t
(134)
J Low Temp Phys (2011) 164:179236 217
and from the imaginary part
s
t
c
t
+
s
B
c
B
=
c
R
s
R
. (135)
So the oscillation frequency is given by
tan(kL
R
)[tan(kL
t
) +tan(kL
B
)] =1. (136)
Substitution of (135) in (134) results in
T
a
T
t
=
c
B
c
t
. (137)
In other words the onset temperature is given by
T
t
=T
a
cos(kL
t
)
cos(kL
B
)
. (138)
Equation (138) shows how T
t
depends on the position of the regenerator in the loop.
For the system of [38] with L
t
= 24 cm, L
B
= 93 cm, and L
R
= 2 m (136) gives
= 82 Hz and (138) gives and onset temperature of 334 K. In practice the onset
temperature will be higher due to dissipative effects, which have been neglected in
this treatment.
8.3.2 Standing-Wave Systems
In this section we will derive expressions for the onset temperature and the oscilla-
tion frequency of the prime mover represented in Fig. 28a. Figure 30 represents the
thermoacoustic prime mover, divided in three subsystems. The transfer matrix T
ad
,
relating the pressure variations and the volume ows at the ends, a and d, is equal to
the product of the transfer matrices of the three subsystems
T
ad
=T
ab
T
bc
T
cd
. (139)
Here T
ab
, T
bc
, and T
cd
are the transfer matrices for the back side (system ab), of the
stack (system bc), and of the resonator tube (system cd) respectively. In particular

U
a
=T
21ad
p
d
+T
22ad

U
d
. (140)
Fig. 30 The thermoacoustic
prime mover, divided in three
subsystems: ab, bc, and cd
218 J Low Temp Phys (2011) 164:179236
The volume ow at the left and the pressure variation at the tube exit are zero, so

U
a
=0 and p
d
=0. Nonzero oscillations are possible only if
T
22ad
=0. (141)
With (139) Equation (141) gives
0 =T
21ab
(T
11bc
T
12cd
+T
12bc
T
22cd
) +T
22ab
(T
21bc
T
12cd
+T
22bc
T
22cd
). (142)
The transfer matrices T
ab
and T
cd
are given by (225), and T
bc
by (200). Note that T
a
and T
b
(the temperatures at position a and b) of (200) in this case have to be identied
with T
H
(the hot-end temperature) and T
a
(ambient temperature) respectively, so that
here =(T
a
T
H
) /T
a
. For simplicity it is assumed that the density and the speed
of sound at the hot side are equal to the values at ambient temperature. Substituting
the expressions for the matrix elements in (142), using (100) and (103), results in
0 = tan(kL
H
)
_
A
H
A
R
tan(kL
R
) +
A
H
A
S
kL
S
1 f

_
+[1 +( 1)f

]
A
S
A
R
kL
S
tan(kL
R
)
_
1 +g
S
T
H
T
a
T
a
_
. (143)
Here L
H
, A
H
are the length and area of the hot side, L
S
, A
S
are the length and the
free-ow area of the stack, and L
R
, A
R
are the length and area of the resonator. The
equations for the real and imaginary parts of (143) result in expressions for and
the onset temperature. They depend only on the ratios of the areas. Equation (143) is
quite complicated but, with algebraic software, it is quite easy to evaluate. A typical
result is given in Fig. 31 where the frequency = /2 and the onset temperature
are given as functions of the relative half plate distance
r =
y
0

. (144)
Equation (143) also has solutions with higher harmonics at higher onset temperatures.
These are not shown in Fig. 31. The pressure amplitude is determined by the heat
input at the hot end.
Fig. 31 Onset temperature and
oscillation frequency as
functions of the relative plate
distance r. In this particular case
nitrogen is used as the working
uid and A
H
=A
S
=A
R
,
L
H
=L
S
=5 cm, L
R
=20 cm,
p
0
=1 bar, T
a
=300 K
J Low Temp Phys (2011) 164:179236 219
8.3.3 The Thermoacoustic Cooler
An example of a thermoacoustic cooler is represented in Fig. 28b. The cooling results
from heat shuttling by the gas parcels moving back and forth in the stack as long as
the temperature gradient in the stack is smaller than the critical temperature gradient.
The enthalpy ow in the stack is essentially nonzero. For the cooling power

Q
L
and
the power P, applied to the speaker, Equation (20) holds. The power P ows from
the speaker to the left heat exchanger, transmitted by the sound wave. So the system
operates due to a traveling wave at the left of the stack. Although these systems are
often called standing-wave systems, a traveling waves play an essential role in the
operation of the cooler. The enthalpy and entropy ows in the stack are negative
(so from the cold to the warm side). This type of cooler is studied extensively by
Tijani [37]. In particular he has shown that the COP can be improved by using gas
mixtures.
9 JT Refrigerators
So far we have treated coolers which use oscillating ows and a regenerator or a
stack. In the remaining part of this paper we will treat two important coolers which
use a steady ow of the working uid and counterow heat exchangers to produce
cooling. The rst is the Joule-Thomson cooler.
9.1 System Description
The Joule-Thomson (JT) cooler is invented by Carl von Linde and William Hampson
so this cooler is also called the Linde-Hampson cooler. Basically it is a very simple
type of cooler which is widely applied as the (nal stage) of liquefaction machines
and cryocoolers. It can easily be miniaturized, but it is also used on a very large scale
in the liquefaction of natural gas. A schematic diagram of a JT liqueer, is given
in Fig. 32. Basically it consists of a compressor, a counterow heat exchanger, a JT
valve, and a reservoir. In this discussion we assume that the heat exchanger is ideal.
This means that it has no ow resistance and that the gas at the low-pressure side
leaves the heat exchanger with room temperature. In Fig. 32 the pressures and tem-
peratures refer to the case of a nitrogen liqueer. At the inlet of the compressor the gas
is at room temperature (300 K) and a pressure of 1 bar (point a). After compression
it is at 300 K and a pressure of 200 bar (point b). Next it enters the heat exchanger
where it is precooled. It leaves the exchanger at point c. After the JT expansion, at
point d, it has a temperature of 77.36 K and a pressure of 1 bar. The liquid fraction
is x. The liquid leaves the system at the bottom of the reservoir (point e) and the
remaining gas ows into the counterow heat exchanger at the cold side (point f). As
said before, it leaves the heat exchanger with a temperature which is equal to room
temperature (point a).
220 J Low Temp Phys (2011) 164:179236
Fig. 32 Schematic diagram of a
JT liqueer. A fraction x of the
compressed gas is removed as
liquid. At room temperature it is
supplied as gas at 1 bar, so that
the system is in the steady state.
The symbols af refer to points
in the Ts-diagram of Fig. 33
9.2 Thermodynamic Analysis
Now we will calculate the liquid fraction x. For this we need the thermodynamic
properties of nitrogen which can be obtained from the Ts-diagram in Fig. 33 [43].
The calculation of x becomes surprisingly simple if we consider the system indicated
by the dotted rectangle in Fig. 32. This is an adiabatic system (no heat exchange with
its surroundings) with rigid walls and it is in the steady state. For such a system the
rst law (see (1)) reduces to a conservation law for the enthalpy. This reads as follows
h
b
=xh
e
+(1 x)h
a
(145)
so
x =
h
a
h
b
h
a
h
e
. (146)
The labels a, b, c, d, e, and f correspond with the points in Figs. 32 and 33. Clearly
there can only be liquefaction if x >0. As h
a
>h
e
this means
h
a
>h
b
. (147)
This is true for nitrogen at room temperature. For every substance there is a certain
temperature T
i
, the so-called inversion temperature, where (h/p)
T
changes sign.
For a starting temperature below the inversion temperature liquefaction is possible
via a JT expansion.
With the enthalpy values, obtained from Fig. 33, we get with (146) x =0.07. Now
the value of the enthalpy in point d can be calculated with h
d
= xh
e
+(1 x)h
f
=
J Low Temp Phys (2011) 164:179236 221
Fig. 33 (Color online) Ts-diagram of nitrogen with isobars, isenthalps, and the lines of coexistence. The
pressures are given in bar, the specic enthalpy in J/g. The dots, labeled af, correspond with the points in
Fig. 32
307 J/g. Similarly s
d
=xs
e
+(1x)s
f
=5.2 J/g K. The enthalpy of point d is equal to
the enthalpy in point c. Following the isenthalp of 307 J/g (which runs parallel with
the isenthalp of 300 J/g) we nd the starting temperature of the expansion at point c.
It is about T
c
= 165 K. The corresponding entropy can be read from the diagram
s
c
=4.2 J/g K.
There are several irreversible processes in this cooler. The entropy per gram, pro-
duced at the JT expansion, is
s
iJT
=s
d
s
c
=5.2 4.2 =1.0 J/g K. (148)
The total entropy production, per gram, is
s
itot
=xs
e
+(1 x)s
a
s
b
=1.33 J/g K. (149)
The difference between s
itot
and s
iJT
is due to the irreversible processes in the heat
exchanger. On the low-temperature side of the heat exchanger the high-pressure gas
leaves the heat exchanger with temperature T
c
= 165 K while the temperature of
the gas entering the heat exchanger is T
e
=77.36 K. This difference in temperatures
between the two sides is fundamental and cannot be avoided even in the case of an
ideal heat exchanger.
If the JT-cooler is used as a cooler (and not as a liqueer) irreversible processes
can be reduced by using gas mixtures instead of pure uids. Also the high pressures,
which are 200 bar in the case of pure nitrogen, can be reduced signicantly [44].
222 J Low Temp Phys (2011) 164:179236
10 Dilution Refrigerators
Dilution refrigerators are fantastic machines: starting from 4.2 K they provide con-
tinuous cooling to temperatures as low as 2 mK without moving parts in the low-
temperature region. Also from the thermodynamic and hydrodynamic point of view
they are interesting since they are based on a unique combination of a Fermi liquid
(
3
He) and a superuid (
4
He). The
4
He component is at rest, but it is superuid so it
allows easy ow of
3
He through the
4
He (up to a certain critical velocity). The ow
of
3
He in the dilute side is driven by a pressure gradient, balancing the viscous forces,
just like any other normal uid. In the superuid
4
He, which is at rest, the pressure
gradient is balanced by an osmotic pressure gradient. The latter is possible thanks
to the fact that, at T = 0, a mixture of
3
He and
4
He separates in one phase, which
is pure
3
He (the concentrated phase), and another phase (the dilute phase) which is
not pure
4
He but contains only 93.4%
4
He and the rest (6.6%) is
3
He. Due to the
Fermi character of the
3
He the concentration of 6.6% results in an osmotic pressure
(which is a property of the superuid
4
He) of
0
= 2209 Pa even at absolute zero.
This allows a pressure drop, to a maximum of
0
, for driving the
3
He through the
dilute channel.
In this section the essence of the operation of dilution refrigerators will be given
from a thermodynamical point of view. For details we refer to Refs. [45, 46] and the
references therein.
10.1 Main Components
Figure 34 is a schematic diagram of the dilution unit of a dilution refrigerator. The
working uid is
3
He which is circulated by pumps at room temperature. These pumps
(not shown in Fig. 34) bring the pressure of the
3
He to a value p
c
which usually is a
few hundred millibar. The
3
He enters the cryostat and is precooled by a helium bath
Fig. 34 (Color online)
Schematic diagram of the
low-temperature part of a
dilution refrigerator (the dilution
unit). The components are
described in the text
J Low Temp Phys (2011) 164:179236 223
at 4.2 K. Next, the
3
He enters the vacuum chamber where it is further cooled down to
a temperature of 1.21.5 K by the so-called 1 K bath, which is a pumped
4
He bath.
The pressure of the incoming
3
He, p
c
, is larger than the
3
He vapor pressure p
v
at
the temperature of the 1 K bath, so the
3
He condenses (goes to the liquid phase).
The heat of condensation is removed by the 1 K bath. Next the
3
He enters the main
impedance. This is a capillary with a very large ow resistance that ensures that p
c
is large enough to guarantee condensation on the 1 K bath for the design
3
He ow
rate

n
3
. Next the
3
He exchanges heat with the still (to be explained later) which is
at a temperature T
S
of around 500 to 700 mK. At 700 mK the
3
He vapor pressure
is about 100 Pa. This is a low pressure, but yet large enough that there is a risk that
the liquid
3
He starts to boil again. So, the
3
He enters a secondary impedance which
has a ow resistance so that the pressure in the still heat exchanger is larger than
the vapor pressure at T
S
. Next the
3
He enters a set of counterow heat exchangers
owing down where it is cooled by a cold ow of
3
He in the other side which ows
up. Usually the heat exchangers in the high-temperature range of above 50 mK are
of the tube-in-tube type. In the colder regions they are more complicated since they
need a large surface area to reduce the Kapitza resistance. After leaving the coldest
heat exchanger the
3
He enters the mixing chamber.
In the mixing chamber the
3
He passes the so-called phase boundary. The phase
boundary separates the concentrated phase (practically 100%
3
He) and the dilute
phase (6.6%
3
He and 93.6% superuid
4
He) which are in equilibrium. In passing
the phase boundary the
3
He is diluted. The heat, needed for the dilution, will be
calculated later on and is the cooling power of the refrigerator. The
3
He leaves the
mixing chamber in the dilute phase. On its way up the cold
3
He in the dilute phase
cools the warm ow of
3
He in the concentrated phase owing down through the heat
exchangers until it enters the still. On its way up the
3
He concentration is gradually
reduced from 6.6% in the mixing chamber to only 0.7% in the still. Yet the vapor in
the still is practically (96%) pure
3
He. A heating power

Q
S
(to be calculated below) is
supplied to the still to maintain a steady ow of
3
He through the system. The pressure
in the still is reduced by pumps at room temperature to pressures of about 10 Pa and
pressurized again to a pressure p
c
so the cycle is closed.
In the next sections the basic thermodynamics of the dilution unit will discussed.
For a more detailed treatment the reader is referred to Ref. [46].
10.2 The 1 K Bath
We can understand the function of the 1 K bath and the still surprisingly well from the
thermodynamic properties of pure
3
He. Therefore we give here the H
m
T diagram
of
3
He (Fig. 35) [47]. In this section we will rst show why a precooling stage, such
as the 1 K bath, is necessary. In order to do this we look at what would happen if
it would not be there. Consider the system inside the dashed contour indicated as
system 1 in Fig. 34. The essence of this system is that only pure
3
He passes the
system boundaries. In the steady state the rst law reads
0 =

Q+

n
3
(H
m1
H
m2
) (150)
224 J Low Temp Phys (2011) 164:179236
Fig. 35 (Color online) H
m
T
diagram of
3
He. Lines of
constant pressure are given
together with the phase diagram.
The blue dots refer to points
discussed in the text
where the index 1 (2) applies to the entrance (exit) of the
3
He ow at a molar ow rate
of

n
3
and

Q is a heating power, supplied somewhere to the system. If there would be
no 1 K bath the temperature of the incoming
3
He would be 4.2 K. With a pressure of
0.5 bar the molar enthalpy can be obtained from Fig. 35 to be H
m1
(0.5 bar, 4.2 K) =
99 J/mol. If the vapor from the still leaves the system at a temperature of 0.7 K H
m2
(0 bar, 0.7 K)= 35 J/mol. With these numbers (150) gives

Q=

n
3
(35 99) =64
J
mol

n
3
. (151)
Equation (151) shows that a positive

n
3
is only possible if the applied heat is negative,
in other words: if there is some form of external cooling. This is provided by the 1 K
bath.
Due to the presence of the 1 K bath
3
He enters system 1 as liquid at 1.2 K instead
of as gas at 4.2 K. Now Fig. 35 shows that H
m1
=6 J/mol. Substituting this value in
(150) gives

Q
S
=

n
3
(35 6) =29
J
mol

n
3
. (152)
This heat usually is supplied at the still, hence the label S.
Note that the heating power at the still (see (152)) is calculated without using
the mixture properties. The reason is that only pure
3
He is passing the boundaries
of system 1. However, if we consider system 2 in Fig. 34 pure
3
He enters at the
left, but
3
He leaves the system in the dilute phase. In this case the enthalpy has to
be replaced by the so-called osmotic enthalpy. In this way it can be shown that the
temperature at the dilute side is typically half the value of the concentrated side [45].
This illustrates the high cooling power of the dilute phase which can be used e.g. for
thermal grounding of electrical wiring and for cooling radiations shields.
Figure 35 shows that the enthalpy of
3
He at 4.2 K can be reduced by increasing the
pressure from 0.5 bar to 2 to 3 bar. At 4.2 K the minimum H
m
42 J/mol, which is
just above the enthalpy of the vapor at 0.7 K (35 J/mol). Hence it is possible to reach
J Low Temp Phys (2011) 164:179236 225
Fig. 36 Schematic drawing of a
dry dilution refrigerator. In the
left corner is the two-stage PTR
cooling the radiation shields and
the incoming
3
He (also via the
regenerator of the PTR)
a steady state, with a nonzero
3
He circulation, if the incoming pressure is a few bar
and a fairly simple heat exchanger is used between the incoming and outgoing
3
He
[47, 48].
The so-called dry systems, shown in Fig. 36, use a cryocooler instead of liquid
helium and liquid nitrogen to precool the
3
He and cool the radiation shields. The
precooling temperature of the
3
He is around 3 K [49]. In that case the enthalpy of the
incoming
3
He is already low enough and hardly any heat exchanger between vapor
and gas is needed at the warm end of the dilution unit.
10.3 The Mixing Chamber
In this Subsection the origin of the cooling power of the mixing chamber will be ex-
plained. The left part of Fig. 37 represents the mixing chamber. The dark-gray area
is the concentrated phase and the light-gray area the dilute phase. The right part rep-
resents the temperature prole. The
3
He enters the mixing chamber with temperature
T
i
and the temperature of the phase boundary, which separates the concentrated and
the dilute phase, is T
M
. On its way from the inlet (at l =L
c
) to the phase boundary
(at l =0) the temperature of the
3
He is reduced from T
i
to T
M
. In this section we will
neglect viscous heating. In the concentrated phase the total energy ow
E =

Q
c
(l) +

n
3
H
3
(153)
226 J Low Temp Phys (2011) 164:179236
Fig. 37 Left: schematic
drawing of the mixing chamber.
The dark area is the
concentrated phase and the
lighter area the dilute phase.
The systems a and b are
explained in the text. The right
part represents the temperature
prole inside the mixing
chamber
is constant as can be derived from applying the rst law to system a of Fig. 37. In
(153)

Q
c
is the heat ow due to heat conduction and H
3
is the molar enthalpy of
3
He in the concentrated phase (at very low temperatures this is pure
3
He, x
c
= 1).
In a well-designed dilution refrigerator the heat ow at the inlet is practically zero

Q
c
(L
c
) 0. In that case the heat ow to the phase boundary (l = 0, system a in
Fig. 37) is given by

Q
c
(0) =

n
3
[H
3
(T
i
) H
3
(T
M
)]. (154)
The total heat ow

Q
t
to the phase boundary is the sum of

Q
c
(0) and the
externally-applied heating power

Q
M

Q
t
=

Q
c
(0) +

Q
M
. (155)
In order to express

Q
t
in terms of the thermodynamic
3
He properties we consider
system b. Its upper border is just above the phase boundary and the lower system
border just below the phase boundary. The mixing is reversible so

Q
t
can be derived
from the second law of thermodynamics, applied to system b, with

S
i
=0
0 =

Q
t
T
M
+

n
3
(S
3
S
d
). (156)
In (156) S
3
(T
M
) is the molar entropy of pure
3
He and S
d
(T
M
, x
s
) is the entropy of one
mol
3
He in the dilute phase, and x
s
is the saturated
3
He concentration of the dilute
phase (which is also a function of T ).
In the concentrated and the dilute phase the
3
He behaves as an ideal Fermi gas
with the molar entropy given by
S
F
=

2
2
R
T
T
F
(x)
(157)
with T
F
the Fermi temperature. For the concentrated phase
S
3
(T ) =

2
2
R
T
T
F
(1)
(158)
J Low Temp Phys (2011) 164:179236 227
where T
F
(1) is the Fermi temperature of pure
3
He. For the entropy of the dilute phase
we can write
S
d
(T, x
s
) =S
F
(T, x
s
) =

2
2
R
T
T
F
(x
s
)
(159)
where T
F
(x
s
) is the Fermi temperature of the dilute phase. So with (156)

Q
t
=

n
3

2
2
RT
2
M
_
1
T
F
(x
s
)

1
T
F
(1)
_
. (160)
Equation (160) shows that the cooling power of the mixing chamber is due to the
difference in the Fermi temperatures of the dilute and the concentrated phase. Now
the Fermi temperature is given by
T
F
=
1
8
_
3

_
2
h
2
m

3
k
B
_
N
A
V
m
(x)
_
2/3
(161)
with k
B
Boltzmanns constant, V
m
the volume (of the mixture) containing one mol
of
3
He, and m

3
an effective mass. In the dilute phase m

3
= 2.46m
3
and in the con-
centrated phase m

3
= 2.8m
3
, with m
3
the mass of the bare
3
He atom. The factors
2.46 and 2.8 are about the same, so the main difference between T
F
(x
s
) and T
F
(1) is
due to the difference in molar volume. At low temperatures V
m
(x
s
) = 426 cm
3
and
V
m
(1) = 37 cm
3
, leading to T
F
(x
s
) = 393 mK and T
F
(1) = 1.8 K respectively. So
the cooling power of a dilution refrigerator is really based on the increase of molar
volume (dilution) of the
3
He!
11 Conclusion
The combination of the rst and second law of thermodynamics for open, inhomoge-
neous, systems with the concepts enthalpy ow, entropy ow, and entropy production,
provides a powerful tool for analyzing a large variety of thermal machines such as
cryocoolers and heat engines. A deep understanding of the system operation can be
obtained, usually with simple mathematics, which leads to elegant expressions with
general validity.
Acknowledgements M.E.H. van Dongen and M.E.H. Tijani are acknowledged for their stimulating in-
terest and carefully reading the manuscript. The staff and students of the Institute of Refrigeration and
Cryogenics of the Zhejiang University in Hangzhou (China) are acknowledged for their stimulating inter-
est.
Open Access This article is distributed under the terms of the Creative Commons Attribution Noncom-
mercial License which permits any noncommercial use, distribution, and reproduction in any medium,
provided the original author(s) and source are credited.
228 J Low Temp Phys (2011) 164:179236
Appendix A: Some Useful Formulae
A.1 General Relations
The molar entropy satises
dS
m
=
C
p
T
dT
_
V
m
T
_
p
dp. (162)
The molar enthalpy
dH
m
=T dS
m
+V
m
dp (163)
and also
dH
m
=C
p
dT +H
p
dp (164)
with C
p
the molar heat capacity at constant pressure and
H
p
=
_
H
m
p
_
T
=V
m

_
V
m
T
_
p
T, (165)
where V
m
the molar volume. We can also write
H
p
=V
m
(1 T
V
) (166)
with the volumetric expansion coefcient

V
=
1
V
m
_
V
m
T
_
p
. (167)
A.2 Ideal Gas Relations
In most cryocoolers helium is used as the working uid. At temperatures above 20 K
and pressures below 30 bar helium behaves as an ideal gas. In that case
pV
m
=RT. (168)
In (168) R is the molar ideal gas constant.
4
The relation for the entropy change be-
comes
dS
m
=C
p
dT
T
R
dp
p
. (169)
Other useful expression are
dS
m
=C
V
dT
T
+R
dV
m
V
m
(170)
4
Here R =8.314510 J/mol K is the molar ideal gas constant, not to be confused with the specic ideal gas
constant which depends on the molar mass of the gas and is expressed in J/kg K.
J Low Temp Phys (2011) 164:179236 229
and
dS
m
= C
p
dV
m
V
m
+C
V
dp
p
, (171)
=
C
p
C
V
. (172)
An analytical expression for S
m
of an ideal gas is
S
m
(p, T ) =C
p
ln
T
T
0
Rln
p
p
0
. (173)
For the enthalpy
dH
m
=C
p
dT. (174)
As C
p
is constant for an ideal gas we may write
H
m
(p, T ) =C
p
T. (175)
Similarly we have for the molar internal energy
U
m
(p, T ) =C
V
T. (176)
Appendix B: The Volume-Flow Equation
In this appendix the so-called volume-ow equation is derived. The internal energy
of the gas in a volume V can be expressed as
U =
_
V
U
m
V
m
dV. (177)
If the pressure is homogeneous and the gas is ideal then U
m
=C
V
T and V
m
=RT/p
so, (177) gives
U =
C
V
R
pV. (178)
Note that (178) holds even though the temperature in V is not homogeneous. Since
H
m
= C
p
T the enthalpy ows into the control volume are given by

H
k
=

n
k
C
p
T
k
.
With

n
k
=p

V
k
/RT
k
we get

H
k
=
C
p
R
p

V
k
. (179)
The system is adiabatic so


Q
k
=0 and also P =0, so (1), with (178) and (179),
reads
C
V
R
V
dp
dt
+
C
V
R
p
dV
dt
=
C
p
R
p

V
k
p

dV
k
dt
. (180)
230 J Low Temp Phys (2011) 164:179236
With
dV
dt
=

dV
k
dt
(181)
and C
V
+R =C
p
and =C
p
/C
V
we get our nal result
V
p
dp
dt
=

V
k

dV
k
dt
. (182)
Appendix C: The Harmonic Model
In many thermal machines the variation of the system parameters, such as pressure,
temperature, and ow, is periodic with angular frequency . For small amplitudes,
the variations can be well approximated in the linear regime. In these cases the vari-
ations can be described in terms of sint and cos t which is called the harmonic
approximation. The harmonic approximation treats systems in the steady state as the
amplitudes are considered time independent. Before discussing the harmonic model
in more detail we have to pay attention to some delicate aspects of temperature vari-
ations and the volume ow (or velocity).
Consider the temperature T of the gas, just left of the heat exchanger X
3
of the
PTR in Fig. 16. As long as gas ows to the left (

n <0) the temperature at this point is

constant and equal to T
a
. However, if the gas ows to the right, back into X
3
(

n >0),
the temperature is higher (see Fig. 17). The T t dependence is given by the curve,
labeled T
real
in Fig. 38. If the molar ow rate at the hot end is given by

n =n
A
cos t (183)
then the time dependence of the temperature is of the form
T =T
a
+T
A
(cos t | cos t |). (184)
This is clearly not harmonic. In the harmonic model a temperaturetime dependence
is used which is represented by the dotted line, labeled T
harm
in Fig. 38, and is of the
form
T =T
a
+T
A
cos t. (185)
Yet, using (185) instead of (184) gives the correct value for the enthalpy ow in the
pulse tube. The reason is that the average enthalpy ow is given by

H =C
p

nT (186)
and that | cos x| cos x =0, so (184) and (185), with (183), give the same result. So the
harmonic expression for the temperature (185) may be used even though the actual
T t dependence deviates from the harmonic dependence.
In the steady state the average velocity of all gas particles is zero: i.e. after one
cycle they are at the same position as before. However, the volume ow through a
J Low Temp Phys (2011) 164:179236 231
Fig. 38 The full line, labeled
T
real
, represents the actual
temperature-time dependence of
the gas in the pulse tube close to
the hot heat exchanger. The
dotted curve, labeled T
harm
,
gives the temperature
dependence as used in the
harmonic model. The curve,
labeled

n, represents the ow
xed surface, or velocity of the gas particles passing a xed surface, contains a DC
contribution even if the mass (molar) ow through that surface is zero. Starting from

n =

V/V
m
=0 with V
m
=V
m0
+V
m
gives, for small V
m
,

V =

VV
m
V
m0
. (187)
Usually

V and V
m
both are of rst order (e.g. in the pressure variation), so the
average volume ow is of second-order. This means that it plays an important role
e.g. in the enthalpy ow since this is also of second-order. With the volume ow
given by

V =vA, with v the velocity and A the surface area, we see that the average
velocity is also nonzero in second order.
Instead of using sint and cos t it is advantages to use the complex formalism
since this reduces the number of variables and equations by a factor of two. In the
complex formalism of e.g. a pressure variation p the complex amplitude p is intro-
duced dened by
p =Re( pe
it
). (188)
The time average of second-order quantities, such as enthalpy ow,

H =

nH
m
(189)
become

H =

nH
m
=Re( ne
it
)Re(

H
m
e
it
) (190)
or

H =
1
2
(Re nRe

H
m
+Im nIm

H
m
) =
1
2
Re ( n

H

m
) (191)
where

H

m
is the complex conjugate of

H
m
.
232 J Low Temp Phys (2011) 164:179236
The complex formalism allows the use of transfer functions which relate the pres-
sure variation p
a
and volume ow

U
a
at position a with p
b
and

U
b
at position b of a
certain control volume via the linear relations
p
a
= T
11
p
b
+T
12

U
b
, (192)

U
a
= T
21
p
b
+T
22

U
b
. (193)
The transfer relations (192) and (193) can also be written in terms of the transfer
matrix
T =
_
T
11
T
12
T
21
T
22
_
(194)
so that
_
p

U
_
a
=T
_
p

U
_
b
. (195)
The transfer-matrix elements have different dimensions. The T
11
and T
22
are dimen-
sionless, while T
12
is in Pa s/m
3
and T
21
in m
3
/Pa s.
5
Components is series can be expressed in products of transfer matrices. A use-
ful property is that the determinant of the product matrix is equal to the product of
the determinants of each of the individual matrices. Below expressions will be given
for some important components of thermal machines. The derivations are beyond the
scope of this paper. The expressions can be very complicated, containing unfamil-
iar functions like Bessel functions and tanh-functions of complex parameters. This
makes it hard to grasp quickly the essence of what is going on. But we start with a
simple example.
For an adiabatic volume V, in which gas ows in at position a and out at position b
(Fig. 4a), we get from (33) that the transfer matrix is given by
T
V
=
_
1 0
i
V
p
0
1
_
(196)
and
det T
V
=1. (197)
The transfer matrix for stacks can be derived from expressions in Ref. [35]. The
relationships tend to be very complicated. Here we give the relations for an ideal gas
as the working uid, and stack plates with very large heat capacity. The differential
equations for the pressure and volume ow variations are as follows
d p
dx
=
i
0
1 f

1
A
S

U, (198)
5
In some important cases simpler relations between the dynamic properties at the two ends of the system
can be obtained if not the pressure and volume ow are related, but the pressure and molar ux (or mass
ux).
J Low Temp Phys (2011) 164:179236 233
d

U
dx
= [1 +( 1)f

]
iA
S
p
0
p +
(f

)
(1 f

)(1 P
r
)T
0
dT
0
dx

U. (199)
For a short stack of length L
S
and small temperature difference these correspond with
a transfer matrix
T
S
=
_
1
i
0
1f

L
S
A
S
[1 +( 1)f

]i
V
S
p
0
1 g
S
T
b
T
a
T
b
_
. (200)
Here A
S
is the free-ow surface area of the stack and
V
S
=L
S
A
S
. (201)
The function g
S
is dened by
g
S
=
f

(1 f

)(1 P
r
)
, (202)
where f

and f

are the so-called Rott functions given by

f
,
=
tanhz
,
z
,
with z
,
=(1 +i)
y
0

,
(203)
with 2y
0
the plate distance. The functions f (z) are even functions of z so they are
functions of
z
2
=i
2y
2
0

2
. (204)
Due to (25) and (26) the Rott functions are not functions of

but of . For
y
0
/ 1 Equation (200) reduces to
T
S
=
_
_
1 3

y
2
0
L
S
A
S
i
V
S
p
0
T
a
T
b
_
_
. (205)
This is the same as for a slit with Poiseuille ow, isothermal compression, and zero
porosity. For y
0
/ 1 (200) becomes
T
S
=
_
1 i
0
L
S
A
S
i
V
S
p
0
1
_
. (206)
This is the same as from a sound duct for L 0 (see (225)).
Equation (200) holds for small temperature differences. For large temperature dif-
ferences, assuming a linear temperature prole, and taking effective values for the
Rott functions, (198) and (199) can be integrated. This results in expressions con-
taining Bessel functions, similar to the expressions for the regenerator, which we will
describe now.
234 J Low Temp Phys (2011) 164:179236
For a very simple regenerator with ow conductance R
r
and zero porosity the
transfer matrix is
T
r
=
_
1 R
r
0 T
a
/T
b
_
(207)
and
det T
r
=
T
a
T
b
. (208)
For a regenerator with nonzero ow resistance but with constant specic ow
resistance z
r
, dened by
dp
dl
=z
r
v, (209)
with the viscosity given by
=
a
_
T
T
a
(210)
(with
a
the viscosity at the temperature T
a
), nonzero porosity, an ideal gas, and a
linear temperature prole with slope dT
0
/dx, the derivation is given in Ref. [25]. The
result is
T
r
=
_
T
11
T
12
T
21
T
22
_
r
(211)
with elements
T
11r
= F
2
(u
b
)F
3
(u
a
) +F
1
(u
b
)F
4
(u
a
), (212)
T
12r
= [F
4
(u
b
)F
3
(u
a
) +F
3
(u
b
)F
4
(u
a
)]
1
k
r
T
b
, (213)
T
21r
= [F
1
(u
a
)F
2
(u
b
) +F
1
(u
b
)F
2
(u
a
)]k
r
T
a
, (214)
T
22r
= [F
4
(u
b
)F
1
(u
a
) +F
3
(u
b
)F
2
(u
a
)]
T
a
T
b
(215)
the parameter k
r
given by
k
r
=
igA
r
p
0
dT
0
/dx
(216)
and the functions F
1
to F
4
dened as
F
1
(u) = BesselI(0, u), (217)
F
2
(u) = BesselK(0, u), (218)
F
3
(u) = uBesselI(1, u), (219)
F
4
(u) = uBesselK(1, u). (220)
J Low Temp Phys (2011) 164:179236 235
BesselI and BesselK are the Bessel functions of the rst and second kind respectively.
The four functions (217)(220) satisfy
F
1
F
4
+F
2
F
3
=1. (221)
By substitution of (212)(215) it can be found that, due to (221), also in this case
holds
det T
r
=
T
a
T
b
. (222)
The transfer matrix elements are functions of u
2
a
and u
2
b
with
u
2
a
=
16
25
ig
0
z
r
(dT
0
/dx)
2
p
0
T
5/2
a
(223)
and
u
2
b
=
5/2
u
2
a
(224)
with
=
T
b
T
a
.
The transfer matrix for a traveling-wave duct with length L and cross sectional
area A with a monochromatic plane wave with wave number k can be derived from
(104) and (105). The result is
T
D
=
_
cos(kL) i
c
0
A
sin(kL)
i
A
c
0
sin(kL) cos(kL)
_
. (225)
Note that
det T
D
=1. (226)
For a short (kL 1) tube with volume V =AL Equation (225) reduces to (196).
References
1. R. Radebaugh, J. Phys., Condens. Matter 21, 164219 (2009)
2. S.R. de Groot, P. Mazur, Non-Equilibrium Thermodynamics (North-Holland, Amsterdam, 1969)
3. T. Kuriyama, R. Hakamada, H. Nakagome, Y. Tokai, M. Sahashi, R. Li, O. Yoshida, K. Matsumoto,
T. Hashimoto, Adv. Cryog. Eng. 35B, 1261 (1990)
4. Volumetric Heat Capacities for Regenerator Materials. NIST, Physical and Chemical Properties Divi-
sion
5. A.T.A.M. de Waele, W. Liang, Cryogenics 48, 417 (2008)
6. H.O. McMahon, W.E. Gifford, Advances in Cryogenic. Engineering 5, 354 (1960)
7. http://www.cryomech.com/
8. J.H. So, G.W. Swift, S. Backhaus, J. Acoust. Soc. Am. 120, 1898 (2006)
9. W.E. Gifford, R.C. Longsworth, Adv. Cryog. Eng. 11, 171 (1966)
10. W.E. Gifford, R.C. Longsworth, Trans. ASME 264 (1964)
11. E.I. Mikulin, A.A. Tarasov, M.P. Shkrebyonock, Adv. Cryog. Eng. 31, 629 (1984)
236 J Low Temp Phys (2011) 164:179236
12. P. Kittel, Adv. Cryog. Eng. 43, 1927 (1998)
13. J. Zhu, P. Wu, Z. Chen, Cryogenics 30, 514 (1990)
14. S.W. Zhu, S.L. Zhou, N. Yoshimura, Y. Matsubara, Cryocoolers 9, 269 (1997)
15. Y. Matsubara, Proc. 17th Int. Cryogenic Eng. Conf., Inst. Phys. (1998), p. 11
16. S.W. Zhu, Y. Kakimi, K. Fujioka, Y. Matsubara, Cryogenics 37, 461 (1997)
17. Z.H. Gan, W.Q. Dong, L.M. Qiu, X.B. Zhang, H. Sun, Y.L. He, R. Radebaugh, Cryogenics 49, 198
(2009)
18. Y. Matsubara, J.L. Gao, Cryogenics 34, 259 (1994)
19. M.Y. Xu, A.T.A.M. de Waele, Y.L. Ju, Cryogenics 39, 865 (1999)
20. N. Jiang, U. Lindemann, F. Giebeler, G. Thummes, Cryogenics 44, 809 (2004)
21. I.A. Tanaeva, U. Lindemann, N. Jiang, A.T.A.M. de Waele, G. Thummes, Adv. Cryog. Eng. 49B,
1906 (2004)
22. D. Gedeon, Sage: object orientated software for Stirling-type machine design, in Proc. of the 29th In-
tersociety Energy Conversion and Engineering Conference, vol. 4 (American Institute for Aeronautics
and Astronautics, Monterey, 1994), pp. 19021907
23. J. Gary, A. OGallagher, R. Radebaugh, E. Marquardt, REGEN3.2 Regenerator Model: User Manual.
NIST Technical Note (2001)
24. G.W. Swift, Thermoacoustics: A unifying perspective for some engines and refrigerators. The Acous-
tical Society of America. ISBN 0-7354-0065-2 (2002)
25. A.T.A.M. de Waele, H.W.G. Hooijkaas, P.P. Steijaert, A.A.J. Benschop, Cryogenics 38, 995 (1997)
26. A.T.A.M. de Waele, P.P. Steijaert, J. Gijzen, Cryogenics 37, 313 (1997)
27. V.D. Arp, R.D. McCarty, Thermophysical properties of helium-4 from 0.8 to 1500 K with pressures
to 1500 MPa. NBS Technical Note 1334 (1989)
28. Y. Huang, C. Guobang Chen, V. Arp, J. Chem. Phys. 125, 054505 (2006)
29. R.H. Sherman, F.J. Edeskuty, Ann. Phys. 9, 522 (1960)
30. M.E. Will, A.T.A.M. de Waele, J. Appl. Phys. 98, 044911 (2005)
31. G. Thummes, C. Wang, C. Heiden, Cryogenics 38, 337 (1998)
32. A. Ravex, T. Trollier, J. Tanchon, T. Prouv, Cryocoolers 14, 157 (2006)
33. R. Radebaugh, E.D. Marquardt, J. Gary, A. OGallagher, Cryocoolers 11, 409 (2001)
34. N. Rott, Adv. Appl. Mech. 20, 135 (1980)
35. G.W. Swift, J. Acoust. Soc. Am. 84, 1145 (1988)
36. S.L. Garrett, Resource letter: TA-1. Am. J. Phys. 72, 11 (2004)
37. M.E.H. Tijani, Loudspeaker-driven thermo-acoustic refrigeration. Ph.D. dissertation, Eindhoven Uni-
versity of Technology (2001)
38. S. Backhaus, G.W. Swift, J. Acoust. Soc. Am. 107, 3148 (2000)
39. A.T.A.M. de Waele, J. Sound Vib. 325, 974 (2009)
40. M. Poese, R. Smith, S. Garrett, R. van Gerwen, P. Gosselin, International Institute of Refrigeration,
6th Gustav Lorentzen ConferenceNatural Working Fluids, Glasgow UK (2004)
41. K. Tang, T. Lei, T. Jin, G.G. Lin, Z.Z. Xu, Appl. Phys. Lett. 94, 254101 (2009)
42. Y. Li, PhD thesis, Eindhoven University of Technology (2011)
43. NBS Tech Note 648 (1973)
44. G. Venkatarathnam, in Cryogenic Mixed Refrigerant Processes, ed. by K.D. Timmerhaus C. Rizzuto.
Int. Cryogenic Monograph Series (Springer, Berlin, 2008)
45. O.V. Lounasmaa, Experimental Principles and Methods Below 1 K (Academic Press, London, 1974)
46. F. Pobell, Matter and Methods at Low Temperatures (Springer, Berlin, 2007)
47. J. Kraus, Cryogenics 17, 173 (1977)
48. A.Th.A.M. de Waele, A.B. Reekers, H.M. Gijsman, Cryogenics 17, 175 (1977)
49. K. Uhlig, Cryogenics 48, 138 (2008)