See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/301294539

Ullmann's Encyclopedia of Industrial Chemistry: Cryogenic Technology

Chapter · September 2013

DOI: 10.1002/14356007.b03_20.pub2

CITATIONS READS

3 1,620

All content following this page was uploaded by Christoph Windmeier on 13 February 2019.

The user has requested enhancement of the downloaded file.

Cryogenic Technology
CHRISTOPH WINDMEIER, Linde AG, Engineering Division, Pullach (M€
unchen),
Germany
RANDALL F. BARRON, Louisiana Tech University, Ruston, United States

1. Introduction . . . . . . . . . . . . . . . . . 2 5.1.1. Types of Cryogenic Heat Exchangers 35

2. Uses. . . . . . . . . . . . . . . . . . . . . . . 3 5.1.2. Heat Exchangers for Dilution
3. Liquefaction Systems . . . . . . . . . . . 6 Refrigerators . . . . . . . . . . . . . . . . 38
3.1. Joule–Thomson Effect . . . . . . . . . . 7 5.1.3. Heat Exchanger Design Problems. . 39
3.2. Basic Linde–Hampson System. . . . . 8 5.1.4. Regenerators . . . . . . . . . . . . . . . . 42
3.3. Precooled Linde–Hampson System. . 11 5.2. Compressors . . . . . . . . . . . . . . . . . 43
3.4. Linde Dual-Pressure System . . . . . . 12 5.3. Expanders . . . . . . . . . . . . . . . . . . 45
3.5. Claude Liquefier . . . . . . . . . . . . . . 13 5.4. Pumps . . . . . . . . . . . . . . . . . . . . . 47
3.6. Liquefaction Systems for Natural Gas 14 5.5. Insulation . . . . . . . . . . . . . . . . . . . 49
3.7. Collins Helium-Liquefaction System. 15 6. Low-Temperature Separation
3.8. Hydrogen Liquefaction . . . . . . . . . . 16 Systems . . . . . . . . . . . . . . . . . . . . 50
3.9. Miniature Liquefaction Systems . . . 17 6.1. Minimum Work of Separation. . . . . 50
3.10. Comparison of Liquefaction Systems 18 6.2. Properties of Mixtures . . . . . . . . . . 51
4. Cryogenic Refrigeration Systems . . . 18 6.3. Flash Calculation. . . . . . . . . . . . . . 53
4.1. Thermodynamically Ideal 6.4. Principles of Cryogenic Rectification 54
Refrigerator . . . . . . . . . . . . . . . . . 18 6.5. Air Separation . . . . . . . . . . . . . . . 55
4.2. Refrigerator Cascades . . . . . . . . . . 20 6.5.1. Linde Single-Column System . . . . . 56
4.3. Joule–Thomson Refrigerators . . . . . 20 6.5.2. Linde Double-Column System . . . . 57
4.4. CVI Cold-Gas Refrigerator. . . . . . . 22 6.5.3. Heylandt System . . . . . . . . . . . . . . 58
4.5. Stirling Refrigerator . . . . . . . . . . . 23 6.5.4. Argon Separation . . . . . . . . . . . . . 59
4.6. Vuilleumier Refrigerator . . . . . . . . . 25 6.5.5. Helium and Neon Separation. . . . . 60
4.7. Solvay Refrigerator . . . . . . . . . . . . 26 6.5.6. Krypton and Xenon Separation . . . 61
4.8. Gifford–McMahon Refrigerator . . . 27 6.6. Conditioning of Natural Gas . . . . . . 62
4.9. Pulse Tube Refrigerators . . . . . . . . 29 6.6.1. Removal of Nitrogen and Helium . 63
4.10. Magnetic Refrigerators. . . . . . . . . 30 6.7. Separation of Hydrogen-rich Gases . 64
4.11. Dilution Refrigerator . . . . . . . . . . 32 6.7.1. Partial Condensation Process . . . . . 64
5. Components of Liquefiers and 6.7.2. Liquid Methane Wash. . . . . . . . . . 65
Cryocoolers . . . . . . . . . . . . . . . . . 34 6.7.3. Liquid Nitrogen Scrubbing . . . . . . 66
5.1. Cryogenic Heat Exchangers . . . . . . 34 6.8. Physical Adsorption . . . . . . . . . . . . 67

Abbreviations J–T Joule–Thomson

LIN liquid nitrogen
BET Brunauer–Emmett–Teller LMTD log-mean temperature difference
COP coefficient of performance LNG liquefied natural gas
CWHE coil wound heat exchanger LOX liquid oxygen
DR dilution refrigerator MR magnetic refrigerator
FOM figure of merit MRC mixed refrigerant cascade
FUG Fenske–Underwood–Gilliland NPSH net positive suction height
GAN gaseous nitrogen NTU number of transfer units
GOX gaseous oxygen OPTR orifice pulse tube refrigerator
GM Gifford–McMahon refrigerator PFHE plate-fin heat exchanger

# 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

10.1002/14356007.b03_20.pub2
2 Cryogenic Technology

PPHE perforated-plate heat exchanger ad adiabatic

PTR pulse tube refrigerator B bottom product
STP standard temperature and pressure b bubble point
conditions c cold; compressor
VM Vuilleumier refrigerator D distillate product
d dew point
Symbols e expander
g vapor or gas phase
A heat exchanger surface, m2 h hot or high, e.g., high temperature
a constant in van der Waals equation, Pa i intermediate pressure stream; inversion,
m6/kg2 component index
b constant in van der Waals equation, m3/kg j component index
C number of components; parameter in the id ideal
BET equation l liquid phase; low, e.g., low temperature
c specific heat, J kg
1 K
1 m molar; mixture; number of plates in the
e heat exchanger effectiveness lower section of the column
F number of independent thermostatic min minimum
properties n number of plates in the upper section of
h specific enthalpy, J/kg or J/mol the column
K distribution coefficient o outside; ambient
L length, m p constant pressure
l length coordinate, m r refrigerant
P number of phases rev reversible
p pressure, Pa R rejected, e.g., heat rejected
pi partial pressure sat saturated conditions
Q heat flow rate, J/s w warm
q specific heat of phase change, J/mol
qm mass flow rate, kg/s
qv molar flow rate, mol/s
R reflux ratio; universal gas constant, J 1. Introduction
mol
1 K
1
r refrigerant mass flow rate ratio Cryogenic technology is most commonly
s specific entropy, J kg
1K
1 or J mol
1 described as the “science and art of producing
K
1 cold temperatures”. Cryogenic technology thus
T temperature, K deals with the development and utilization of
U heat transfer coefficient, W m
2 K
1 low-temperature processes and equipment. The
v specific volume, m3/kg term “cryogenic” usually refers to temperature
W work, J levels below
150 C (123 K) [1]. This dividing
w work flow rate, J/s line between the cryogenic temperature range
x expander mass flow rate ratio; mole frac- and the field of ordinary refrigeration is
tion in the liquid phase selected by the normal boiling points of the
y liquid yield; mole fraction in the vapor so-called permanent gases: Helium, hydrogen,
phase neon, nitrogen, oxygen, carbon monoxide,
a relative volatility argon, oxygen, methane, and krypton cannot
g isentropic exponent be liquefied by isothermal compression at
D difference ambient temperatures. In contrast substances
h efficiency such as freons, hydrogen sulfide, ammonia,
mJT Joule–Thomson coefficient, K/Pa propane, and other conventional refrigerants
boil above
150 C at atmospheric pressure
Subscripts and possess a sufficiently high critical point
A absorbed or added, e.g., heat absorbed that they can be liquefied by ambient tempera-
act actual, e.g., actual work ture isothermal compression. In terms of
 Cryogenic Technology 3

above’s definition, the first production of cryo- Rocket Propulsion Systems. Many of the
genic temperatures is attributed to LOUIS PAUL large launch vehicles used in the United States
CAILLETET and RAOUL PICTET who succeeded in space program use liquid oxygen as the oxidiz-
1877 shortly after one another to produce liquid ing agent [3]. The first flight test of a liquid-fuel
oxygen by isothermal compression coupled rocket using liquid oxygen was made by ROBERT
with precooling. Cryogenic technology took GODDARD in 1926. The NASA Saturn V Apollo
the next great leap about two decades later vehicle used liquid oxygen as the oxidizing
when in 1895 CARL VON LINDE and WILLIAM agent for all three stages and liquid hydrogen
HAMPSON independently developed a quite sim- as the fuel for the upper two stages. The Space
ilar process for continuously liquefying air, Shuttle propulsion system uses liquid hydrogen
relying on internal, recuperative heat exchange. as the fuel and liquid oxygen as the oxidizing
Following milestones were the first liquefaction agent.
of hydrogen by SIR JAMES DEWAR in 1898 and Cryogenic fluids are used because of the
the first liquefaction of helium by KAMMERLINGH high specific impulse (thrust per unit mass
ONNES in 1908, followed by the discovery of flow rate) that can be achieved with the hydro-
superconduction in 1911 also by ONNES. A gen – oxygen combination. The fluids must be
detailed historical overview on the develop- in the liquid phase so that a sufficient mass may
ment of cryogenic research up to modern times be stored on board within a reasonable volume.
is given in [2]. The use of liquid hydrogen as a fuel in space
vehicles introduces several special problems in
addition to the problem of its comparatively
2. Uses low liquid density. Because the normal boiling
point of liquid hydrogen is only 20.3 K, ambi-
Applications of cryogenic technology vary ent air liquefies on the outside of an exposed,
widely, both in scope and size (Fig. 1). Some uninsulated liquid hydrogen storage tank. The
of the areas involving cryogenic technology are condensed air on the tank poses safety problems
discussed briefly in this chapter. and results in excessive heat transfer to the cold

Figure 1. Comparison of temperature requirements of cryogenic applications with achievable temperatures of various
cryocoolers
a) Liquefied natural gas (LNG); b) Air separation; c) Cryosurgery; d) Hydrogen liquefaction; e) Sensors and superconductors;
f) Joule–Thomson; g) Stirling, reversed Brayton; h) Gifford–McMahon; i) Pulse tube; j) Collins helium liquefier; k) Dilution
and magnetic refrigeration
4 Cryogenic Technology

liquid hydrogen. To avoid this problem, liquid etc., has been increased by a factor of 2–4 by
hydrogen tanks must be insulated. this technique [7].
Liquid hydrogen has the second lowest vis-
cosity of all liquids (after helium), which makes Ground Freezing. The ground freezing tech-
ordinary sealing difficult. In many cases, the nology is a special discipline in the field of civil
only solution is to use a welded construction, and tunnel engineering [8]. It exploits the
which makes component disassembly and properties of artificially frozen soil, namely
inspection both complicated and expensive. the temporary increase in stability of the soil
An advantage of liquid hydrogen is that it and/or the ability of the frozen soil to prevent
can be used as the coolant for rocket nozzles. groundwater seepage. For this purpose a multi-
Liquid hydrogen is usually at a pressure of ca. 7 tude of vertical pipes are inserted into the
MPa before it is injected into the combustion ground which are then continuously flooded
chamber. Because the critical pressure of with a cryogenic liquid, usually liquid nitrogen.
hydrogen is ca. 1.3 MPa, no problems of
two-phase flow occur in the nozzle coolant Recycling of Rubber. One of the more dif-
passages with regenerative cooling. ficult materials to recycle is automobile or truck
tire. Each year, ca. 13.5106 t of tires are
Steel Industry. Oxygen is required in the scrapped worldwide [9, 10]. By a conservative
steel-making process, and the economical pro- estimate, these tires contain ca. 4.7106 t of
duction of oxygen involves separation of air in a rubber that could be recycled. The conventional
cryogenic system [4]. In the open-hearth pro- reclamation method involves chemical devul-
cess, the conventional use of air as the gas canization of the rubber and mechanical grind-
above the molten charge in the furnace requires ing of the resulting material. The widespread
additional heating by an auxiliary fuel. The use use of steel-belted tires makes this process
of oxygen allows the operation of open-hearth uneconomical today.
steel furnaces at 130 % of the capacity when When rubber is cooled to a temperature
air is used. below 210 – 175 K, it passes through a glass
transition and becomes brittle. In this brittle
High-strength pressure vessels may be con-
condition, the rubber can be crushed easily into
structed by cryogenic forming, in which the small particles. The cord and metal components
cylinder is stretch-formed at liquid nitrogen
in the original tire can be separated readily by
temperatures. This process (called Ardeform-
magnetic and air blast techniques.
ing) involves placing the fabricated vessel in a
The freezing process is usually carried out in
bath of liquid nitrogen (77.4 K) and introducing
a freezing tunnel, in which the rubber is carried
high-pressure gaseous nitrogen into the vessel.
along by a conveyor belt and liquid nitrogen is
The vessel is allowed to expand by ca. 15 % into
sprayed onto the surface of the scrap tires. The
a stainless steel die of the appropriate shape.
vapor that is formed when the liquid nitrogen
After this stretching process, the vessel is
evaporates recirculates through the tunnel by a
allowed to warm slowly to room temperature.
series of fans which increase the velocity of the
An improvement in tensile strength as high as a nitrogen gas. The high velocity of the gas
factor of three has been observed.
enhances the heat-transfer process and
Tool steels used in manufacturing processes improves the effectiveness of the freezing pro-
have been cryogenically treated to improve cess. After passing through the freezing tunnel,
wear resistance of tool. The tool steel is cooled the material is transferred immediately to a
slowly (cooling rate ca. 3  C/min) to liquid size-reduction mill, where the cord and metal
nitrogen temperatures and is then allowed to components in the original tire are separated
soak at the cryogenic temperature for ca. 20 h. from the rubber particles in a single operation.
During this process, essentially all of the The crumb rubber particles are virtually free
retained austenite in the tool steel is trans- of fiber and steel contaminants. This property
formed to martensite, which is a more wear allows the rubber to be used in a variety of
resistant phase [5, 6]. The useful lifetime of applications, including highway repairs using
parts, such as slitter knives, high-speed drills, tire – asphalt mixes, soil conditioning, and
 Cryogenic Technology 5

mixing with coal for use as a fuel extender [11]. vibrations of the atoms of the amplifier element
Currently, the cryogenic process is the only do not seriously degrade the performance of the
economical method to recover rubber from system. Cryogenically cooled masers are used
steel belted tires. in radio astronomy applications and in satellite
communication systems [4, pp. 157–170].
Food Processing and Transportation.
These systems use a compact, continuously
Freezing as a technique for preserving food
operating, closed-cycle helium refrigerator to
was used as early as 1840. Currently, frozen
maintain a temperature as low as 4.2 K.
foods are prepared by moving the food pack-
Tiny superconducting elements, called
ages on a conveyor belt through a liquid nitro-
SQUIDs (superconducting quantum interfer-
gen spray in a specially insulated tunnel, called
ence devices), are used in extremely sensitive
tunnel freezer or spray freezer [12]. Initial
digital magnetometers and voltmeters [15]. In
contact with the liquid nitrogen freezes all
addition to SQUIDs, other electronic devices
exposed surfaces and seals in both the flavor
that use superconductivity in their operation
and aroma of the food product. The food pack-
include superconducting amplifiers, transform-
ages are then passed through an “equilibrium
ers, and magnets.
zone” in the tunnel that is cooled by the cold
Superconducting magnets (solenoids) are
nitrogen vapor.
used as levitation systems for trains that operate
The cryogenic food freezing process has
at speeds up to 500 km/h [16]. These trains are
several advantages over the conventional food
used for commuter service between Tokyo and
freezing process that uses cold air. Cryogenic
Osaka in Japan. The levitation system consists of
freezing requires ca. 7 min compared with ca.
a normal track conductor coil on the ground and
30 min required by conventional freezing meth-
a superconducting coil on board the of train. The
ods. In conventional freezing, ca. 3–15 mass%
superconducting coils are maintained at liquid
(primarily water) of the product is removed by
helium temperature (4.2 K) by means of a
the air blast. In contrast, cryogenic freezing
cryostat filled with liquid helium. The propul-
results in only 0.2–0.3 mass% dehydration
sion unit for the train is a linear induction motor,
[13]. The reduced in moisture loss results in
with the primary coil located on the ground and
improved flavor, texture, and appearance of the
the secondary coil placed on the train.
frozen food when compared with the resulting
product of conventional freezing. The cryo- Biological and Medical Applications.
genic freezing system is fairly simple and Containers cooled by liquid nitrogen are used
requires ca. 10 % of the maintenance expendi- to preserve whole blood, tissue, bone marrow,
tures of a conventional system. and semen for extended periods of time [17].
Liquid nitrogen has also been used as the Glycerol is mixed with such material as bull
refrigerant in frozen-food transport trucks and semen to protect the cells during freezing. The
railway cars [14]. The system consists of an semen is slowly cooled to ca. 5  C and is
insulated container for the liquid nitrogen and a
maintained at this temperature for at least 5 h
series of nozzles that are activated by a ther-
to allow the glycerol to penetrate the cells
mostatically controlled switch to spray liquid
adequately. The cells are then cooled at 1  C/
nitrogen into the refrigerated compartment. The
min to
15  C and then at 5  C/min to
79  C.
system is quite versatile in that the foods may
The frozen material can then be stored indefi-
be maintained at any prescribed temperature.
nitely at
196  C in liquid nitrogen.
Temperatures lower than those economically
Whole blood cells have been preserved by
achieved by mechanical refrigeration may be placing the blood in a flat metal container and
maintained with the cryogenic system. Mois-
immersing the container in liquid nitrogen. The
ture retention of the cargo is good because the
blood is thawed for use by abrupt immersion in
inert atmosphere inside the storage space pre- an agitated bath of warm water. Whole blood
vents oxidation and drying.
has also been frozen by spraying it directly into
Electronics. Sensitive microwave amplifiers liquid nitrogen.
(masers) are cooled to liquid nitrogen or liquid Cryogenic surgery (cryosurgery) is mainly
helium temperatures to assure that thermal used to achieve the destruction of pathologically
6 Cryogenic Technology

altered tissue. It has been used in eye surgery and able to compare their efficiencies, a suitable
for the treatment of tumors, Parkinson’s disease, reference system has to be defined [22]. The
and various lesions [18]. This surgical technique physical limitations of a process are described
has several advantages over conventional sur- by the thermodynamically ideal system. It is
gery, including reduced problems with bleeding, thus usually chosen as reference. The perfect
greater control of the area of influence, and less refrigeration cycle is the Carnot cycle. How-
trauma to the system [19]. ever, liquefaction is essentially an open-system
Several special-purpose probes have been arrangement. Therefore, the ideal liquefaction
developed for cryosurgery. Generally, liquid system only requires the first two processes in
nitrogen (or sometimes liquid argon) flows the Carnot cycle, which are a reversible iso-
through a vacuum-insulated cannula to the thermal compression followed by a reversible
exposed tip on the probe and freezes the tissue adiabatic (isentropic) expansion process to the
to be destroyed. The size of the frozen region is saturated liquid state. The ideal process is
controlled by adjusting the mass flow rate of the shown in Figure 2.
cryogenic liquid through the cannula. The gas to be liquefied is compressed
reversibly and isothermally from ambient con-
Vacuum Technology. The vapor pressure of ditions (point 1) to some high pressure (point 2).
all substances exhibits an exponential depen- This high pressure is selected such that the gas
dency in regard to the absolute temperature. will reach the saturated liquid state (point f) after
Thus very low values of absolute (and partial) expansion in an adiabatic, reversible (isentropic)
pressures can be reached by trapping molecules expander.
on cold surfaces, usually cooled by liquid helium By applying the first law of thermodynamics
(4.2 K), liquid hydrogen (20.3 K), or liquid to the system consisting of the compressor,
nitrogen (77.4 K). A respective vacuum pump expander, and liquid receiver, the following
is generally referred to as cryopump. A cryo- equation is obtained:
pump is operated in batch or continuous mode by
either regularly refilling the cryogenic liquid or Qrev
W id ¼ qm ðhf
h1 Þ
continuously recondensing the boil-off vapor in
a cryocooler. Additionally the active surface of where Qrev is the heat rejected from the iso-
the cryopump has to be regenerated (cleaned thermal compressor, Wid ¼ Wc
We is the net
from trapped molecules) from time to time.
Depending on the prevailing capture mecha-
nism, cryopumping is classified as cryoconden-
sation, cryosorption, or cryotrapping. Latter
term either refers to the condensation of low
volatile components from a noncondensable gas
stream (e.g., water or oil vapor), or the temporal
delay of component migration in contact with a
cold surface without actually condensing it.
Depending on the nature of the condensed
species, vacuum levels between 0.1 and 10
8 Pa
can be reached. Cryopumping is also suitable
for realizing high volume-flow rates. Typical
applications include space-simulation cham-
bers, low-temperature physics, and industrial
production applications as, e.g., vacuum-
Figure 2. Thermodynamically ideal system for gas
coating systems [20, 21]. liquefaction 
A) Schematic: a) Isothermal compressor; b) Isentropic
3. Liquefaction Systems expander
B) Liquefaction thermodynamics

Numbers on the temperature–entropy diagram correspond
Several different systems are used to liquefy to similarly numbered points on the schematic diagram;
gases at cryogenic temperatures. In order to be f: fluid state
 Cryogenic Technology 7

ideal work input, qm is the mass flow-rate a pressure drop. Since outer heat transfer and
through the system, hf is the specific enthalpy changes in kinetic energy are usually negligi-
of the liquid, and h1 is the specific enthalpy of ble, and because there is no transfer of work
the gas at ambient conditions. Applying the from and to the fluid flow, the resulting expan-
second law of thermodynamics around the sion can be considered to be isenthalpic. A
whole system the following expression is nonideal fluid will hence experience a change
obtained: in temperature. This fundamental phenomenon
is called the Joule–Thomson (J–T) effect and it
Qrev ¼ qm T1 ðs2
s1 Þ ¼ qm T1 ðsf
s1 Þ enables the production of cold temperature
levels by isenthalpic expansion of high-
where T1 is the ambient temperature, sf is the pressure fluids.
specific entropy of the liquid, and s1 is the The origin of the J–T effect lies within the
specific entropy of the gas at ambient occurrence of intermolecular forces in nonideal
conditions. fluids: A pressure reduction is accompanied by
Substitution for the reversible heat with- an increase of specific volume. Consequently
drawal in the first expression gives the expres- the average intermolecular distances also
sion for the ideal work being equal to the increase. The internal work done on attractive
minimum work requirement for liquefaction or repulsive interactions will result to a respec-
of a gas: tive change of system temperature. The con-
stant enthalpy lines on the temperature–

W id =qm ¼ T1 ðs1
sf Þ
ðh1
hf Þ ð1Þ pressure plane are shown in Figure 3. The effect
of temperature change for an isenthalpic
The negative sign in Equation (1) originates change in pressure is represented by the
from the thermodynamic sign convention: work Joule–Thomson coefficient, mJT, defined by
done by the system is positive; work done on
the system is negative. The minimum (ideal)

@T
mJT ¼
work of liquefaction for several common cryo- @p h
genic fluids is given in Table 1. The minimum
work of liquefaction is always equal to the
exergy of liquefaction obtained by setting T1 The Joule–Thomson coefficient is equal to
of Equation (1) equal to ambient temperature. the slope of the isenthalpic lines in Figure 3. In
the region where mJT < 0, isenthalpic expansion
results into an increase in temperature, whereas
3.1. Joule–Thomson Effect in the region where mJT > 0, respective expan-
sion results into a temperature decrease. The
A fluid flow encountering an obstruction like,
curve that separates the two regions is called the
e.g., an expansion valve, an orifice plate, a
inversion curve. The J–T coefficient is zero
porous plug, or a capillary tube, will experience
along the inversion curve, because the slope
of the isenthalpic line is zero.
Table 1. Ideal work requirements for liquefaction of gases Thermodynamics show that the J–T
beginning at 300 K and 101.3 kPa coefficient may be expressed in terms of volu-
Gas Normal bp, K Ideal work of
metric and caloric properties as follows [23]:
liquefaction, "
 #
kJ/kg 1 @v
mJT ¼ T
v ð2Þ
cp @T p
Helium [4He] 4.2 6 772
Hydrogen 20.4 11 915
Neon 27.1 1 136 where v is the specific volume and cp the
Nitrogen 77.4 760
Air 78.7 735
specific heat at constant pressure of the
Carbon monoxide 81.7 745 material.
Argon 87.3 747 Since an ideal gas does not possess any
Oxygen 90.2 629 intermolecular forces one can expect a value
Methane 111.6 1 080
of zero for its J–T coefficient. Substituting the
8 Cryogenic Technology

This expression shows that the J–T coefficient

is positive when T < 2a/bR and negative when
T > 2a/bR.
The inversion curve is formed by all points at
which the J–T coefficient is zero. The inversion
temperature Ti from Equation (3) is given by:

2a
Ti ¼ ð1
b=vÞ2
bR

The maximum inversion temperature for a van

Figure 3. General temperature–pressure diagram for a der Waals fluid is the temperature on the inver-
real gas sion curve where p ¼ 0 (or b/v ¼ 0) or Ti, max ¼
2a/bR. The maximum inversion temperature
(in K) for several gases is as follows:
equation of state of an ideal gas v ¼ (RT)/p into
Equation (2) gives:
Helium [4He] 45

 Carbon monoxide 652
@v R v Hydrogen 205
¼ ¼
@T p p T Neon 250
Argon 794
Air 603
where R is the gas constant and p is the pressure. Oxygen 761
Substitution into Equation (2) gives mJT ¼ 0 for Nitrogen 621
an ideal gas. For a real gas, the J–T coefficient Methane 939
may be negative, zero, or positive, depending
on the initial temperature and pressure of the The maximum inversion temperatures for
material. helium, hydrogen, and neon are below ambient
The simplest equation of state for real gases temperature. When starting out from ambient
is the van der Waals equation of state [24]: temperature, liquefaction systems that use J–T

expansion alone to produce low temperature
a
pþ ðv
bÞ ¼ RT cannot be used to liquefy these gases: Addi-
v2
tional means such as external sources of refrig-
where the constant a gives a measure of the eration or expansion engines must be used in
intermolecular forces between the molecules, order to liquefy these gases.
and the constant b provides a measure of the
finite size of the gas molecules. Constants a and 3.2. Basic Linde–Hampson System
b are thus dependent upon the species and can
be determined using the first and second deriv- Historically, the Linde–Hampson system was
ative of the isothermal line at the critical point the second system used to liquefy gases at
[25]. cryogenic temperatures (the cascade system
The expression for the J–T coefficient for a was used first) [26]. The principle flow scheme
van der Waals gas is given by Equation (3): of Linde–Hampson liquefaction system is
shown in Figure 4.
ð2a=RTÞð1
b=vÞ2
b The working gas (here air) is compressed (a)
mJT ¼ ð3Þ
cp ½1
ð2=vRTÞð1
b=vÞ2  from ambient conditions to a pressure of ca. 20
MPa. The high-pressure gas then passes
For large values of the specific volume, as through an internal heat exchanger (recupera-
typically occurring in gas phases, Equation (3) tor, b), in which it is precooled by the return
can be approximated by: stream of cold, low-pressure gas. At point 3, the
cold high-pressure gas is expanded in a Joule–


mJT ¼
1 2a

b
Thomson valve (c) to a pressure of the order
cp RT of ambient pressure. At the exit of the valve
 Cryogenic Technology 9

of thermodynamics:

ðqm Þf ðh1
h2 Þ
ðQo =qm Þ

y¼ ¼
qm h1
hf

where (qm)f ¼ mass flow rate of liquid; qm ¼

total mass flow rate; Qo ¼ heat inleak from
ambient; h1 ¼ enthalpy at the outlet at the warm
end of the heat exchanger; h2 ¼ enthalpy at the
inlet at the warm end of the heat exchanger; and
hf ¼ enthalpy of the saturated liquid. (The
numerical subscripts refer to the corresponding
points in Figure 4.) The liquid yield may also be
written in terms of the effectiveness of the heat
exchanger, defined by:

h1
hg
e¼
h01
hg

with h01 ¼ enthalpy of the low-pressure stream

at pressure p1 and at temperature T10 ¼ T2; hg ¼
enthalpy of the saturated vapor. If the heat
exchanger was ideal, the cold gas would leave
the warm end of the heat exchanger at the same
temperature as that of the inlet temperature of
the warm stream. The actual exit temperature of
the cold stream is:

T1 ¼ T01
ð1
eÞðT01
Tg Þ

The term (1
e) is called the heat-exchanger

ineffectiveness [27].
The liquid yield for the Linde–Hampson
system may be written in the alternative form:

ðh01
h2 Þ
ð1
eÞðh01
hg Þ
ðQo =qm Þ

Figure 4. Linde–Hampson liquefaction system  y¼ ð4Þ
ðh01
hf Þ
ð1
eÞðh01
hg Þ
A) Schematic: a) Compressor; b) Heat exchanger; c) J–T
valve; d) Liquid receiver
B) Liquefaction thermodynamics The heat-exchanger effectiveness must be

Numbers on the temperature–entropy diagram correspond
close to unity for the system to produce liquid.
to similarly numbered points on the schematic; f: fluid state;
g: vapor state The minimum required heat-exchanger effec-
tiveness for barely successful liquefaction may
be determined from Equation (4) by setting the
liquid yield equal to zero. The result is:
(point 4), the stream is in the two-phase (liquid–
vapor) region. The liquid phase is collected in ðh01
h2 Þ
ðQo =qm Þ
emin ¼ ð5Þ
the liquid receiver (d) and withdrawn. The ðh01
hg Þ
saturated vapor is returned through the heat
exchanger to furnish cooling for the incoming A typical value of emin for a nitrogen liquefier
warm gas. operating between ambient pressure and tem-
The liquid yield, or fraction of the warm gas perature (101.3 kPa and 300 K) and 20.3 MPa is
stream that is liquefied, is given by the first law ca. 0.86 (while neglecting the heat inleak).
10 Cryogenic Technology

The power requirement for the compressor is inversion temperature lower than the ambient
given by Equation (6): temperature, such as helium, hydrogen, and
neon (Fig. 5). For these gases, the enthalpy


W T2 ðs01
s2 Þ
ðh01
h2 Þ
¼ ð6Þ
difference of isothermal compression ðh01

qm hc h2 Þ is negative at ambient temperature; there-
fore, the calculated value of the liquid yield
where hc ¼ isothermal efficiency of the from Equation (4) is negative. The system can
compressor. The negative sign is included in never get started from the warm condition,
Equation (5) because work is done on the gas by because the gas would experience an increase
the compressor. in temperature when it expands through the J–T
In accordance to the definition of the figure valve at ambient temperature. Even if the sys-
of merit (FOM), one obtains for a liquefaction tem could be cooled somewhat initially, the gas
system: would expand through the J–T valve into the
vapor region and completely miss the two-
ð
W id =qm Þ yð
W id =qm Þ phase region [22, pp. 71–73].
FOM ¼ ¼
½
W act =ðqm Þf  ð
W act =qm Þ
The designer of a liquefaction system has
little freedom in choosing the conditions at
where Wact ¼ actual work requirement of the point 1, which is usually near ambient temper-
compressor. Performance values for the ideal ature and pressure. However, the designer can
and actual Linde–Hampson systems using var- select the maximum pressure (at point 2) in the
ious gases are given in Table 2. Linde–Hampson liquefaction system. To maxi-
Intrinsic inefficiencies (exergy loss exclud- mize the liquid yield for the system, the opti-
ing compression) of the Linde–Hampson liq- mum pressure at point 2 must be the pressure
uefier originates from two sources: The largest that minimizes the enthalpy h2. Mathemati-
share (ca. 75%) is caused by the irreversibility cally, for minimum h2, the following must
of the isenthalpic expansion process. The apply:
remaining portion (ca. 25%) can be attributed
to losses related to the inevitable temperature

@h
spread between cold end inlet (low pressure, ¼0
@p T¼T1
saturated vapor) and outlet (high pressure) of
the heat exchanger. The reason for the latter lies From the chain rule of calculus,
within the decreasing values of isobaric heat
capacities with decreasing pressure and the



@h @h @T
¼
¼
cp mJT
inequality of flow rates through the heat @p T @T p @p h
exchanger caused by the removal of liquid
product at the cold end. Therefore, the pressure that minimizes h2 is the
The basic Linde–Hampson cycle cannot be pressure for which mJT ¼ 0; i.e., point 2 should
used to liquefy gases which have a maximum lie on the inversion curve. For air at 300 K, the

Table 2. Performance of the Linde–Hampson liquefaction system for different fluidsa

Fluid Normal bp, K Liquid yield y ¼ (qm)f/qm Work per unit mass liquefied, kJ/kg FOMb

Idealc Actuald Idealc Actuald Idealc Typicald

N2 77.4 0.0708 0.0581 6673 11 620 0.1151 0.0661

Air 78.8 0.0808 0.0672 5621 9 653 0.1313 0.0765
CO 81.6 0.0871 0.00749 5381 8 934 0.1428 0.0860
Ar 87.3 0.1183 0.1090 2750 4 265 0.1741 0.1123
O2 90.2 0.1065 0.0957 3804 6 048 0.1671 0.1051
CH4 111.7 0.1977 0.1888 3957 5 919 0.2758 0.1844
a
State points are p1 is the 101.3 kPa, p2 is the 20.265 MPa, T2 ¼ 300 K.
b
Figure of merit.
c
Ideal system operates with e ¼ 1.00 and hc ¼ 1.00, heat transfer from ambient is neglected.
d
Typical system operates, e ¼ 0.975 and hc ¼ 0.7, heat inleak from ambient is neglected.
 Cryogenic Technology 11

Figure 5. J–T inversion curves of common cryogenic fluids

a) Methane; b) Air; c) Neon; d) Hydrogen; e) Helium
Cooling by isenthalpic expansion is only possible left
(inside) of the respective inversion curve

Figure 6. Yield of liquid nitrogen  [y ¼ (qm)f/qm] for the

corresponding pressure is ca. 400 MPa. How- Linde–Hampson liquefaction system as a function of the
ever, practical systems are generally designed temperature of the gas at the warm end of the heat
to minimize the work per unit mass liquefied. exchanger (see Fig. 4)

Yield is mass fraction compared to ideal system.
The optimum pressure to achieve this condition
can be found by numerical methods. For air, the
optimum pressure at point 2 to minimize the point d. This factor also places an upper limit
power input per unit mass liquefied is ca. 20 on the liquid yield for this system.
MPa [26, p. 71]. The liquid yield for the precooled Linde–
Hampson system is given by Equation (7):

3.3. Precooled Linde–Hampson System ðh1
h2 Þ þ rðha
hc Þ
ðQo =qm Þ

y¼ ð7Þ
ðh1
hf Þ
A plot of the liquid yield y as a function of the
temperature at the inlet of the heat exchanger where r ¼ (qm)r/qm ¼ refrigerant mass flow rate
(point 2) for the basic Linde–Hampson system is ratio; ha ¼ enthalpy of the auxiliary refrigerant
shown in Figure 6 for liquefaction of nitrogen. at point a; and hc ¼ enthalpy at point c. The
The performance of the liquefier can be second term in Equation (7) represents the
improved if it is modified such that the gas enters improvement in liquid yield over the basic
the heat exchanger at a temperature lower than Linde–Hampson system because of the use of
ambient temperature. This is the basic principle precooling.
of the precooled system shown in Figure 7. The power requirement per unit mass flow
A separate refrigeration system, using a fluid rate through the main compressor is given by
such as ammonia, propane, or a freon com- Equation (8):
pound, is used to precool the main gas stream.
The critical temperature of the auxiliary coolant
W=qm ¼
T1 ðs1
s2 Þ
ðh1
h2 Þ
þ rðhb
ha Þ ð8Þ
must be greater than ambient temperature so hc
that the precoolant can be condensed against
cooling water or ambient air. The last term in Equation (8) represents the
The temperature at point 3 is limited by the additional work required by the auxiliary refrig-
minimum temperature of the precoolant at eration system.
12 Cryogenic Technology

Figure 7. Precooled Linde–Hampson liquefaction system

a) Main compressor; b) Heat exchanger; c) J–T valve;
d) Liquid receiver; e) Refrigerant compressor; f) Refriger-
ant condenser Figure 8. Linde dual-pressure liquefaction system
a) Low-pressure compressor; b) High-pressure compressor;
c) Heat exchanger; d) J–T valve; e) Liquid receiver

Another limiting factor in the design of

the precooled system is that the refrigerant
mass flow rate (qm)r must be small enough so has been added. The high-pressure gas is passed
that the liquid entering the heat exchanger through a three-channel heat exchanger (c) and
at point d can be completely vaporized. If expanded to an intermediate pressure, where
the auxiliary refrigerant is not completely some of the gas is liquefied. The vapor is
vaporized at point a, the liquid enters the returned through the three-channel heat
refrigerant compressor and damage to the exchanger to the high-pressure compressor.
compressor results. The liquid stream is finally expanded to the
low pressure level of the cycle.
The liquid yield for the dual-pressure system
3.4. Linde Dual-Pressure System is given by:

The basic Linde–Hampson system may be ðh1
h3 Þ
½ðqm Þi =qm ðh1
h2 Þ

y¼
modified, as shown in Figure 8, by compressing ðh1
hf Þ
only a portion of the main gas stream through
the entire pressure range [28]. where (qm)i ¼ mass flow rate of the intermedi-
In the Linde dual-pressure system, the gas to ate pressure stream at point 8, and qm ¼ mass
be liquefied is compressed from near ambient flow rate through the high-pressure compressor.
pressure to an intermediate pressure (a) and The work requirement per unit mass com-
then to a high pressure (b) after a return stream pressed in the high-pressure compressor is
 Cryogenic Technology 13

given by: replacing the isenthalpic pressure reduction

with an isentropic process: Energy removal

W=qm ¼ ½T1 ðs1
s3 Þ
ðh1
h3 Þ=hc by expansion through a turbine or expander

½ðqm Þi =qm ½T1 ðs1
s2 Þ
ðh1
h2 Þ=hc performing work [29]. This step is equivalent
to the production of refrigeration potential
The work per unit mass liquefied may be directly at reduced temperature. The original
determined from the following: Claude system from 1902, shown in Figure 10,
uses a piston expander (c) to provide the bulk of

W=ðqm Þf ¼ ð
W=qm Þ=y the cooling effect in the system. If a rotary
expander is utilized the cycle is sometimes also
The work requirement for liquefaction of air referred to as Kapitza cycle [30].
in the Linde dual-pressure system is shown in The liquid yield for the Claude system is
Figure 9. At an optimum value of the interme- given by Equation (9):
diate pressure, the system work requirement is
minimized for a given mass flow rate ratio, ðh1
h2 Þ þ xhad ðh3
h0e Þ
y¼ ð9Þ
(qm)i/qm. In addition, the work requirement for ðh1
hf Þ
a given intermediate pressure is decreased by
increasing the mass ratio. The mass ratio is where x ¼ (qm)e/qm ¼ expander mass flow
thermodynamically limited, because (qm)I þ ratio; (qm)e ¼ mass flow rate through the
(qm)f must be less than the mass flow rate qm expander; had ¼ adiabatic efficiency of the
through the high-pressure compressor. expander; and he0 ¼ enthalpy at the expander
outlet for an isentropic expansion. If the
expander work is used to aid the compression
3.5. Claude Liquefier of the gas, the work requirement per unit mass
compressed is given by Equation (10):
Expansion through a J–T valve is an irreversible
process; therefore, the performance of a lique-

W=qm ¼ ½T1 ðs1
s2 Þ
ðh1
h2 Þ=hc
xhad ðh3
h0e Þ ð10Þ
faction system can be improved significantly by

Figure 9. Work required to liquefy a unit mass of air in the

Linde dual-pressure system 

The intermediate pressure is the pressure of the gas Figure 10. Original Claude liquefaction system
between the low- and high-pressure compressor. a) Compressor; b) Heat exchangers; c) Expander; d) J–T

Values for (qm)i/qm valve; e) Liquid receiver
14 Cryogenic Technology

In addition to the improved thermodynamic the liquefaction of natural gas. They are usually
performance of the Claude system compared found to be economical for small, medium, and
with that of the basic Linde–Hampson system, large plant capacities. Mixed-refrigerant cycles
the Claude system has the advantage that it may range back to 1959 when KLEEMENKO suggested
be used without precooling for liquefaction of a J–T-based liquefaction process using a refrig-
helium, hydrogen, and neon. If precooling is erant mixture (instead of a pure fluid) and
used for these gases, however, the system per- intermediate phase separators to produce dif-
formance is improved significantly. ferent refrigerant compositions from a single
The FOM of a Claude-type liquefier is subs- circulating mixture. Such a system was later
tantially higher than for all Linde–Hampson generally called a mixed refrigerant cascade
cycles without additional work performing (MRC) [32], as shown exemplarily in Figure 11
expansion: (1) Refrigeration effect is produced (see also Section 4.3). Methane, ethane or
directly at cold temperatures by the withdrawal ethylene, and propane usually form the main
of work (enthalpy) in the expander, (2) sub- components of the working fluid.
cooling of the flow at the inlet of the J–T valve
greatly reduces entropy production during the
isenthalpic expansion, and (3) the cold end
temperature spread between the high- and
low-pressure stream is reduced by the direct
bypass of mass flow through the expander [31].

3.6. Liquefaction Systems for Natural

Gas

Natural gas occurs in form of a wide boiling

mixture with hydrocarbon components from
methane to C4þ in addition to sour gases
(carbon dioxide and hydrogen sulfide), inerts
(as helium and nitrogen), and high-boiling
impurities. In order to provide efficient cool-
ing over a wide range of temperatures special
requirements arise to the liquefaction technol-
ogy. But even with the multiplicity of process
variations used for production of liquefied
natural gas (LNG) being as high as the com-
positional ranges of the gas feed and require-
ments for plant capacities, practical cycles
can be related to three main technological
groups [31]:

1. J–T-based cascades
2. J–T-based mixed-refrigerant processes
3. Expansion-turbine-based systems

The latter are either open-loop Claude-type

or closed-loop reversed Brayton-type cycles
and they are used typically for smaller plant
capacities. Expander-based processes reach a
Figure 11. Single-staged mixed-refrigerant cascade for
FOM of below 30%. liquefaction of natural gas
Nowadays, mixed-refrigerant processes and a) Compressors; b) Heat exchangers; c) Separators; d) J–T
their derivatives are most commonly used for valves; e) Liquid receiver
 Cryogenic Technology 15

In order to avoid the formation of solids

during the liquefaction process a pretreatment
of natural gas is required. This usually implies
the removal of water, sour gases, and mercury
down to values in the lower ppm range. After
pretreatment, the natural-gas stream enters the
MRC system at ca. 4–6 MPa. The refrigerant
vapor is compressed (a1) and partially con-
densed in the compressor aftercooler. The
propane-rich liquid phase is separated from
the refrigerant mixture (c1), expanded through
a J–T valve, and used as the coolant in the warm
multistream heat exchanger, thus generally
referred to as precooler (b1). The vapor from Figure 12. Typical heat exchanger temperature profiles
the phase separator is passed through the pre- a) Pure-fluid J–T cascade (propane/ethylene/methane, three
cooler and condensed partially. The ethylene- stages each); b) Three-staged.mixed-refrigerant process
rich liquid phase is separated in a second natural gas liquefier
Cooling and liquefaction curve of natural gas (red) and
separator (c2) to provide cooling in the inter- evaporation curve of refrigerants (blue)
mediate multistream heat exchanger (b2). Due
to the high pressure of the precooled natural-
gas stream total condensation will occur in this
heat exchanger, therefore also called liquefier. cycle for liquefaction and a closed-loop
The remaining refrigerant vapor, mainly con- reversed Brayton-type cycle for subcooling of
sisting of low-boiling components (methane) the natural gas.
from the second separator is cooled down fur-
ther in the liquefier heat-exchanger, expanded
through a J–T valve, and recirculated in order to 3.7. Collins Helium-Liquefaction
provide refrigeration for subcooling the lique- System
fied feed stream in a cold heat exchanger (b3),
referred to as subcooler. An important development in cryogenic lique-
The composition of a refrigerant mixture faction technology after the original construc-
provides the necessary degrees of freedom to tion of the Linde–Hampson and Claude
closely match the natural gases cool-down and liquefiers occurred in 1946 [33]. During that
liquefaction curve to the evaporation curve of the year, a helium liquefier was developed by
refrigerant, and thereby helps to minimize the SAMUEL COLLINS of the Massachusetts Institute
temperature gap between the warm and cold of Technology. Commercial application was
streams (see Section 5.1 and Fig. 12). Well- initially handled by the A. D. Little Co.; later,
designed MRC natural-gas liquefiers will thus commercial development was assumed by an
reach a FOM of 45–50%. Thermodynamic associated company, Cryogenic Technology.
fundamentals of MRC are discussed extensively The introduction of the Collins helium liquefier
in [31]. transformed the existing situation, where
The most prominent J–T derivatives for research at liquid helium temperature was
LNG production are (1) the pure refrigerant restricted to a few laboratories that were willing
cascade (commercially, e.g., ConocoPhillips to endure the difficulties of constructing and
Optimized Cascade process), (2) mixed-refrig- operating helium liquefiers of their own design.
erant cycles with and without precooling (com- The maintenance and operation of the special-
mercially, e.g., APCI C3MR and Shell DMR), purpose liquefiers were both difficult and
and (3) cascaded mixed-refrigerant cycles expensive, whereas the Collins system was
(commercially, e.g., Linde MFC). In contrast, relatively simple to operate and required mini-
the AP-X cycle of Air Products utilizes a mum maintenance because of its robust design.
somewhat different cascade consisting of pro- Depending on the inlet pressure of the helium
pane-based precooling, a mixed-refrigerant gas, the Collins system uses two to five
16 Cryogenic Technology

expanders in order to precool the helium below liquefied. The vapor formed during the expan-
its inversion temperature in an open-system sion process is returned through the heat
arrangement. Partial liquefaction then occurs exchangers to provide cooling for the incoming
after low-temperature J–T expansion. The helium gas stream.
working principle of the Collins liquefier can The unit is enclosed in a Dewar vessel with a
thus be considered to be quite similar to a liquid-nitrogen-cooled shield to reduce ambient
Claude cycle but possess a larger number of heat transfer. The heat exchangers and expand-
degrees of freedom due to the increased number ers operate in an atmosphere of low-pressure
of expanders. As an improvement liquid nitro- helium gas, so that minor leaks from the
gen is generally used to precool the system and engines do not cause significant problems.
increase the liquid yield (Fig. 13). The liquid yield for the two-expander
Helium is compressed (a) to a pressure of ca. Collins liquefaction system is given by:
1.3 MPa, and any entrained oil is removed in an
oil trap. The compressed helium gas flows ðh1
h2 Þ þ xDhe þ x0 Dh0e
y¼
through the first heat exchanger (b1), and a h
hf
fraction (ca. 16 %) is bypassed through the first
expander (c1). The helium gas enters the where x ¼ (qm)e/qm ¼ mass flow rate ratio for
expander at a temperature of 40–45 K. The engine 1, x0 ¼ (qm)e0 /qm ¼ mass flow rate ratio
remainder of the helium gas stream flows for engine 2, D he ¼ enthalpy change through
through the second heat exchanger (b2), and engine 1, and D he0 ¼ enthalpy change through
leaves the heat exchanger at a temperature of engine 2. The expander work output in the
ca. 15 K. A fraction of the stream (ca. 56 % of Collins system is generally not used to aid in
the total inlet flow) is then bypassed to the compression, but the energy is dissipated in a
second expander (c2). The remainder of the mechanical or hydraulic brake, which is exter-
helium stream flows through the third heat nal to the system. Typical operating character-
istics of the Collins system are as follows [34]:
exchanger (b3) and expands through a J–T valve
(d), in which a fraction of the stream is
Liquid He production rate 25–32 L/h (3.1–3.4 kg/h)
Power requirement (He only) 45 kW
Helium flow rate compressed 5.2 kg/h
Liquid nitrogen consumption 25 L/h
Power requirement (including power
for liquid nitrogen) per unit volume 3.1 kW  h/L

3.8. Hydrogen Liquefaction

The liquefaction of hydrogen utilizes the prin-

cipal liquefaction technologies discussed in the
previous sections. If the hydrogen is precooled
below ca. 205 K a Linde–Hampson cycle can be
used. This process is usually applied for smaller
liquefaction capacities [35]. In case the liquid
hydrogen product needs to be generated under
pressure a Claude process derivative comes to
use for larger plant capacities. A peculiarity of
hydrogen liquefaction is the necessity of cata-
lytic ortho–para conversion during cool down
which increases the required amount of refrig-
Figure 13. Collins helium liquefaction system
eration. If the conversion occurs naturally in the
a) Compressor; b) Heat exchangers; c) Expanders; d) J–T stored product, the maximum storage time is
valve; e) Liquid receiver reduced to a few days due to the high amount of
 Cryogenic Technology 17

heat release during the exothermal configura- around a center mandrel (c). The outer jacket
tional conversion [36]. (d) around the tubing coil confines the flow of
the return gas-stream around the tubing. The
high-pressure gas (6–40 MPa) flows inside the
3.9. Miniature Liquefaction Systems heat-exchanger tube and is cooled by the return
cold gas flowing over the fins.
Miniature cryogenic liquefaction systems have The high-pressure gas expands through a
been used in two special applications: (1) cool- porous nozzle into the cold space chamber,
ing of electronic components in aerospace sys- where the liquid formed during the expansion
tems (particularly cooling of infrared detectors) process is used to cool the electronic
and (2) providing cooling for cryosurgery appli- components.
cations. In both cases, liquefaction is based on Freezing of contaminant gases, such as car-
the Linde–Hampson cycle. Since no withdrawal bon dioxide and water vapor, presents a serious
of cryogenic liquids takes place in these sys- operational problem with this type of liquefier.
tems a parallel classification as J–T cryocooler Typical maximum allowable concentrations of
(see Section 4.3) is also possible. contaminants are as follows:
Hymatic Cooler. A miniature liquefier used
to cool electronic components is shown in Water 1 mg/kg (dew point
77  C)
Figure 14 [37]. Because of the small cross- Carbon dioxide 1.2 mg/kg
Hydrocarbons 0.045 mg/kg
sectional area of the heat-exchanger tubing,
the system requires extremely pure air or nitro-
gen in its operation.
A sintered bronze filter (a) of 2 mm pore size CVI Liquefier. A similar miniature liquefier
at the inlet of the system excludes particles in unit, using gaseous argon as the working fluid
the incoming stream. The heat exchanger (b) was developed by the CVI Corp. (Fig. 15), [38].
consists of miniature finned tubing wrapped Compressed argon from a standard gas
cylinder (a) passes through a filter and drier
unit (b) to remove impurities. The heat
exchanger (e) in the CVI system consists of
miniature, spirally finned tubing wrapped
around a formed mandrel. The mandrel
grooves and spacers wrapped between the
tube and the outer casing force the return
gas stream to flow axially along the outside
of the tube, thereby improving the effective-
ness of the heat exchanger. The heat
exchanger is 16 mm in diameter and 76
mm in length. The J–T valve is replaced
with a fine capillary tube (f) having an inside
diameter of a few micrometers. The liquid
produced during the expansion process is
passed through the cryosurgery probe tip
(h), where cooling freezes the tissue on the
outside of the probe tip.
Precise temperature control is achieved by
automatic or manual control (c) of the gas flow
rate into the unit. A thermocouple in the tip
senses the temperature of the probe tip. A small
resistance heater within the probe tip is used
Figure 14. Miniature liquefier (hymatic system)
when rapid thawing is required, as in eye
a) Filter; b) Finned-tube heat exchanger; c) Center mandrel; surgery or retraction of the probe from a frozen
d) Outer jacket; e) Porous nozzle; f) Cold space tumor. The liquefier is enclosed in a vacuum
18 Cryogenic Technology

The ideal precooled system and the ideal

dual-pressure system have approximately the
same FOM; however, the ideal Claude system
has a significantly higher FOM. On the other
hand, the Claude system is more sensitive to
component inefficiencies. For example, the
FOM of the basic Linde–Hampson system is
decreased by a factor of 1.7 when component
inefficiencies are included, whereas the FOM
for the Claude system is decreased by a factor
of 2.2 when respective inefficiencies are intro-
duced. The performance of the Claude system
is particularly sensitive to the efficiency of the
expander, because a significant fraction of the
cooling effect in the system is provided by
expansion of the gas through the expander
(see Section 5.3).

4. Cryogenic Refrigeration Systems

Cryogenic refrigeration systems, called cryo-
coolers, are used in a variety of applications in
which a low temperature level is required,
including cooling of advanced electronic sys-
Figure 15. CVI liquefaction system for cryosurgery
applications tems, superconducting magnets and compo-
a) Argon cylinder; b) Filter and dryer; c) Control valve; nents, and space environment simulation
d) Temperature indicator; e) Heat exchanger; f) Capillary chambers. In addition to cryocoolers that pro-
tube; g) Vacuum jacket; h) Probe tip vide cooling by liquefying the working fluid,
several others, including the Philips (based on
the Stirling cycle), the Gifford –McMahon
jacket (g) having a 25-mm outside diameter to (GM), the Vuilleumier (VM), and the Pulse
reduce heat transfer to the internal parts of the Tube refrigerator (PTR) use entirely different
probe and to allow the probe to be hand-held. concepts. Two special cryocoolers, i.e., the
magnetic refrigerator (MR) and the dilution
refrigerator (DR), are used to achieve and
3.10. Comparison of Liquefaction maintain temperatures below 1 K.
Systems

The thermodynamic performance of the lique- 4.1. Thermodynamically Ideal

faction systems described in Sections 3.2–3.8, Refrigerator
which use nitrogen as the working fluid, is
given in Table 3 [39]. The first table entry is The process of heat withdrawal at a cold, and
the thermodynamically ideal system, which has discharge to a higher (usually ambient) temper-
a liquid yield (y) of unity and a FOM of unity. ature level is generally referred to as refrigera-
The remaining entries are presented for an ideal tion. The second law of thermodynamics states,
system (100 % effective heat exchangers and that heat will spontaneously only flow from high
100 % efficient compressors and expanders) to low temperature levels–so the inversion of this
and for a system with an isothermal compressor process requires the input of external work. The
efficiency (hc) and effectiveness (e) values minimum work required to invert the natural
commonly encountered in practical liquefier direction of heat flow can be determined by
design. balancing around any of the thermodynamically
 Cryogenic Technology 19

Table 3. Comparison of liquefaction systems

Liquefaction Conditions Liquid Work per unit Figure of

system yield  mass liquefied merit 
[y ¼ (qm)f/qm] (
W/mf), kJ/kg (FOM)

Ideal reversible system 1.000 768 1.000

Linde–Hampson
Ideal system p2 ¼ 20.3 MPa; 0.071 6673 0.115
hc ¼ 100 %; e ¼ 1.00
Practical system p2 ¼ 20.3 MPa; 0.058 11620 0.066
hc ¼ 70 %; e ¼ 0.975
Precooled Linde–Hampson
Ideal system r ¼ 0.15; p2 ¼ 20.3 MPa; 0.122 3945 0.195
T3 ¼ 240 K; hc ¼ 100 %;
e ¼ 1.00
Practical system r ¼ 0.15; p2 ¼ 20.3 MPa; 0.108 6317 0.122
T3 ¼ 240 K; hc ¼ 70 %;
e ¼ 0.975
Linde dual-pressure
Ideal system p2 ¼ 6.1 MPa; p3 ¼ 20.3 MPa; 0.049 3679 0.209
hc ¼ 100 %; (qm)i/(qm) ¼ 0.80;
e ¼ 1.00
Practical system p2 ¼ 6.1 MPa; 0.035 7346 0.105
p3 ¼ 20.3 MPa; hc ¼ 70 %;
(qm)i/(qm) ¼ 0.80; e ¼ 0.975
Claude
Ideal system p2 ¼ 6.1 MPa; x ¼ 0.60; 0.258 1025 0.750
hc ¼ had ¼ he ¼ 100 %;
e ¼ 1.00; T3 ¼ 250 K
Practical system p2 ¼ 6.1 MPa; x ¼ 0.60; 0.201 2232 0.344
hc ¼ 70 %; had ¼ 80 %;
he ¼ 72 %; e ¼ 0.975;
T3 ¼ 250 K

In all cases, the systems use nitrogen as the working fluid and are assumed to operate between 101.3 kPa at 300 K (26.8  C) and the conditions
stated; heat transfer from ambient is neglected.

Compared to ideal reversible system; yield is the ratio of the mass of liquid produced to the mass compressed in the compressor.

ideal refrigerators–two of which shall be dis- input to the refrigerator; the work input is
cussed as follows: (1) the Carnot refrigerator, for negative according to the thermodynamic
systems in which heat is absorbed at constant sign convention. The FOM is then defined by
temperature (Figure 16 a), and (2) the isobaric- Equation (11):
source refrigerator, for systems in which the
working fluid absorbs heat at constant pressure ðCOPÞact
FOM ¼ ð11Þ
over a range of temperatures (Figure 16 b). ðCOPÞid
The FOM for a refrigeration system is
defined as the ratio of the ideal (reversible) where the subscripts act and id denote actual
work requirement to the actual work require- and ideal cycle, respectively.
ment, for the same refrigeration effect. In this The COP for the Carnot refrigerator depends
definition the FOM is also equal to the exergetic only on the low and high temperature levels of
efficiency of the refrigerator. The FOM may be the cycle, and is given by Equation (12):
written in terms of the coefficient of perform-
ance (COP), generally defined by: Tl
ðCOPÞid ¼ ð12Þ
Th
Tl
QA QA =qm
COP ¼ ¼

W net
W net =qm
where Tl ¼ low temperature in the cycle, and Th
where QA ¼ energy absorbed from the low- ¼ high temperature in the cycle (usually ambi-
temperature region, and Wnet ¼ net work ent temperature).
20 Cryogenic Technology

Figure 16. Thermodynamic cycles

a) Carnot cycle; b) Reversible isobaric-source refrigeration
Figure 17. Cascade of Carnot refrigerators providing
systems
cooling
a) At a single temperature level; b) At multiple temperature
levels
The COP for the isobaric-source ideal refrig-
erator is given by Equation (12): (together with the respective work input) to
the top cycle at T2. There absorbed heat (and
h2
h1 work input) are discharged as QR to the ambient
ðCOPÞid ¼ ð13Þ
Th ðs2
s1 Þ
ðh2
h1 Þ
temperature Th. The COP of such a cascade is
the same as for a single-stage Carnot refrigera-
where points 1 and 2 are shown on Figure 16b. tor operating between T1 and Th. For the exam-
If Th is equal to the ambient temperature, the
ple on the right-hand side the situation is a little
denominator of Equation (13) resembles the different: Besides the heat absorbed from the
exergy difference of the working fluid between respective bottom stage QA,1 the top stage
hot and cold state which is equal to the mini-
additionally offers external refrigeration effect
mum work required to transfer the system
QA,2 at T2.
between the two states. For the practical application of this princi-
If the working fluid is in gaseous state with
ple, the Carnot refrigerators are replaced by
approximately constant specific heat, Equation
cryocoolers. Similar arrangements can be found
(13) may be written in terms of the tempera- throughout the history of cryogenics.
tures, as follows: The right-side arrangement is commonly
found in applications where refrigeration on
T2
T1
ðCOPÞid ¼ ð14Þ multiple temperature levels is required such
Th lnðT2 =T1 Þ
ðT2
T1 Þ
as cooling of sensor arrays at 4.2 K together
Equation (14) reduces to the expression for the with providing refrigeration at 77.4 K for radi-
Carnot COP (Eq. 12) in the limit as T2 ation shielding in a single cryocooler (see
approaches T1. Section 4.8 and Fig. 11). In particular cascades
of J–T refrigerators are used for the liquefaction
of natural gas (see Sections 3.6 and 4.3).
4.2. Refrigerator Cascades

If a single refrigerator is not able to reach the

targeted temperature level due to either (1) 4.3. Joule–Thomson Refrigerators
physical limitations or unsuitability of the cho-
sen working fluid, (2) inefficient process equip- Any of the liquefaction systems that do not use
ment, or (3) technical limitations of process an expansion engine may be classed as Joule–
equipment, all types of refrigerators can be Thomson (J–T) cryocoolers when operated as a
operated in series to form a refrigerator refrigerator instead of as a direct (open system)
cascade. Two simple examples of cascaded liquefier. A schematic of typical J–T cryo-
Carnot refrigerators are shown in Figure 17. coolers operated with pure fluids and wide
In the left process, heat QA is absorbed in the boiling mixtures is shown in Figure 18 together
lower Carnot cycle at T1, and discharged with the simplified processes in a T
h plot. In
 Cryogenic Technology 21

Figure 18. Idealized Joule–Thomson cryocoolers operated with a pure fluid (left, e.g., nitrogen) and a wide boiling mixture
(right). Numbers in the temperature–enthalpy diagram correspond to similarly numbered points on the schematic.
a) Compressor; b) Heat exchanger; c) J–T valve; d) Evaporator

case of a pure refrigerant (e.g., nitrogen) the The liquid in the evaporator of the J–T
process is operated typically between 10–20 cryocooler evaporates at a constant tempera-
MPa on the high-pressure side and 0.1–0.2 MPa ture; therefore, the FOM for the J–T system
on the low-pressure side. involves the COP for the Carnot refrigerator
In contrast, if a J–T cryocooler is operated (Eq. 12).
with a mixed refrigerant, the maximum pres- Liquid nitrogen may be used as the refrigerant
sure levels can be reduced to ca. 1–2 MPa [40]. for the temperature range of ca. 65–115 K. The
The mixture composition is often chosen to low-temperature limit is set by the triple-point
exhibit a low-temperature miscibility gap temperature for the refrigerant and by problems
where a second liquid phase will form that is in maintaining the low-vacuum pressure with
particularly rich in the low-boiling component. reasonable mass flow rates. If the evaporator
This will allow the refrigerant to partially evap- pressure is lowered below the triple-point pres-
orate at almost constant temperature which is a sure, nitrogen “snow” forms in the evaporator
common requirement of cryocoolers. and the expansion valve is likely to become
The refrigeration effect, QA/qm, for this clogged. In addition, the heat transfer in the
system is given by: evaporator is degraded. Therefore, and due to
the increased risk of contamination, subambient
QA =qm ¼ ðh01
h2 Þ
ð1
eÞðh01
h5 Þ pressure levels are usually avoided in practical
systems. The high-temperature limit for the
where e ¼ heat exchanger effectiveness, h01 ¼ evaporator is set by the critical point for the
fluid enthalpy at pressure p1 and temperature T2, refrigerant. If the evaporator pressure is
and h5 ¼ enthalpy at the evaporator outlet. The increased above the critical pressure, only com-
work requirement per unit mass for the compres- paratively small amounts of heat can be rejected
sor is given by Equation (6); therefore, the COP to ambient which drastically decreases the cycle
for the J–T cryocooler may be written as follows: efficiency. Additionally more internal precool-
ing has to be supplied for liquid to form on
hc ½ðh01
h2 Þ
ð1
eÞðh01
h5 Þ expansion through the J–T valve.
COP ¼
T2 ðs1
s2 Þ
ðh1
h2 Þ If maintenance of a constant temperature in
the evaporator is not a concern, the temperature
where hc ¼ isothermal efficiency of the range for the J–T cryocooler using nitrogen gas
compressor. can be extended up to ambient temperature;
22 Cryogenic Technology

however, cycles using other working fluids, adsorbed in a second adsorbent chamber, which
such as ammonia, propane, or freon com- is cooled. The cold compressor temperature is
pounds, are more effective at temperatures 150 K, and the hot compressor temperature is
above ca. 230 K. Appropriate means to extend 470 K. A COP of 0.22 was reported for nitrogen
the operating range of a J–T cryocooler would gas absorbing energy at 77 K, which corre-
be to operate a cascade of refrigerators such that sponds to a FOM of 0.64.
each stage’s refrigerant can be condensed From an external point of view, due to the
against the evaporating liquid refrigerant of a lack of liquid withdrawal, the miniature lique-
higher stage in a heat exchanger. Each of the faction systems of Section 3.8 can also be
refrigerants thus only would need to be com- classified as J–T cryocoolers.
pressed to a technically reasonable pressure
level. The temperature in the evaporator can
be controlled by regulating the pressure in 4.4. CVI Cold-Gas Refrigerator
the evaporator. Cascades of J–T refrigerators
are usually limited to operate above 77.4 K In many applications, a constant-temperature
(Fig. 19). There are no suitable refrigerants to source is not required; helium or neon may be
fill the “gap” in vapor pressure curves between used as the working fluid, and the gas need not
the critical temperature of neon (44.5 K) be condensed to provide refrigeration. The
and the triple-point temperature of nitrogen CVI cold-gas refrigerator (see Fig. 20) is an
(63.2 K). example of this type of cryocooler [42]. This
Miniature J–T cryocoolers have been built unit has been used to provide cooling for the
by using an adsorption compressor [41]. Large cryopanels in space environmental simulation
quantities of gas are adsorbed by an adsorbent. chambers, for recondensation of boiled-off
The material is then heated in a closed container hydrogen gas, and for other applications
to produce a high-pressure gas (thermal com- involving cooling in the 20 K temperature
pression). The gas flows through the J–T por- range. The working fluid for the CVI refriger-
tion of the system, and the return gas is ator is helium.

Figure 19. Vapor pressure curves of common fluids in cryogenic processes

 Cryogenic Technology 23

however, the refrigeration effect is reduced by a

factor of ca. 3. The mass of liquid nitrogen
precoolant required per unit mass of helium gas
is given by:

ðqm ÞN ðh2
h1 Þ þ ðh8
h4 Þ

¼
ðqm ÞHe hc
ha

The liquid nitrogen precoolant bath helps over-

come the ineffectiveness of the warm heat
exchanger in the system.

4.5. Stirling Refrigerator

The Stirling cycle was originally developed in

1816 by ROBERT STIRLING for use in a hot-air
engine [43]. The first Stirling cycle refrigerator
was described by ALEXANDER KIRK in 1864 [44].
Workers at the Philips Co. in Eindhoven pio-
neered to develop the refrigerator into a com-
Figure 20. CVI cold-gas refrigerator using helium gas as
the working fluid
mercially viable system in 1953 [45]. With
a) Helium compressor; b) Heat exchanger; c) Precooler; typical cooling capacities between a few milli-
d) Heat exchanger; e) Expander; f) Load heat exchanger watts at 10 K and up to several kilowatts at 77 K
Numbers in the schematic refer to subscripts in the respec- and above, Stirling refrigerators have found a
tive equations in the text wide field of practical applications as, e.g.,
cooling for space technology, infrared thermal
imaging systems, superconductors, and small
The helium gas is compressed in a rotary scale air liquefaction.
compressor (a), and after passing through a A schematic of the operating cycle for the
filter system to remove traces of oil from the Stirling refrigerator is shown in Figure 21. The
compressor, the gas is cooled in a three-channel Stirling cryocooler consists of a four-stroke
heat exchanger (b). The high-pressure stream is cycle executed inside a cylinder enclosing a
further cooled in a liquid-nitrogen-cooled pre- working piston (c) and a displacer piston (a).
cooler (c) and a cold heat exchanger (d). Trace The two chambers in the engine are connected
impurities in the helium, such as carbon diox- through a regenerator (b), which is a critical
ide, are removed in an adsorption purifier in the component of the cryocooler.
liquid nitrogen bath. The gas then expands The sequence of processes for the Stirling
through an expander (e) before passing through refrigerator is as follows:
the load heat exchanger (which replaces the Process 1 ! 2 (Figs. 22 and 21 A). The
evaporator in J–T systems). The CVI refrigera- working fluid (preferably a perfect gas) is
tor uses highly effective, brazed aluminum compressed by the working piston while heat
plate-fin heat exchangers. is rejected to cooling coils at the warm end of
The refrigeration effect for the CVI refrig- the cylinder. In the ideal case, this compression
erator is given by: process is isothermal.
Process 2 ! 3 (Figs. 22 and 21 B). With the
QA =qm ¼ h7
h6 ¼ ðh7
h5 Þ þ had ðh5
h06 Þ working piston approximately stationary, the
gas is forced through the regenerator by the
where had ¼ adiabatic efficiency of the motion of the displacer piston. The gas is
expander, and h06 ¼ enthalpy for an isentropic cooled by heat exchange with the cold regener-
expansion from point 5 to 6. The system can ator matrix. In the ideal case, this is a constant-
operate without the liquid nitrogen precooling; volume (isochoric) process.
24 Cryogenic Technology

working piston remains approximately station-

ary. The gas is warmed nearly to ambient
temperature by heat exchange with the regen-
erator matrix. Ideally, this heating process is an
isochoric process.
Since the paths of the isochoric lines for a
perfect gas are parallel in the T–s plane the
amounts of entropy transferred during cooling
and heating in the regenerator are equal. Thus,
the COP of the ideal Stirling refrigerator is the
same as that of the Carnot refrigerator, i.e. the
FOM is unity.
T3
COP ¼
T1
T3

In order to be able to reach good values of

the COP, ideal working fluids are often used for
cryogenic applications. These most commonly
include helium and hydrogen, and a more rare
utilization of neon and nitrogen.
Frictional energy dissipation, pressure drop
in the regenerator, and the ineffectiveness of the
regenerator all interact to reduce the FOM of
the actual refrigerator significantly. Their
impact also depends upon the size of the
Figure 21. Sequence of operations for the Stirling cycle machine and the temperature at the cold end
refrigerator of the cylinder. Although the 1960s large-scale
A) Process 1 ! 2; B) Process 2 ! 3; C) Process 3 ! 4;
D) Process 4 ! 1 Stirling coolers have reached values for the
a) Displacer piston; b) Regenerator; c) Working piston FOM of 0.45, modern cryocoolers typically
operate at values of 0.35 and below [46, 47].
An excellent analysis of the operation of the
Process 3 ! 4 (Figs. 22 and 21 C). The gas actual Stirling cryocooler has been given by
is expanded by simultaneous motion of both the [28, pp. 131–142]; a more exact numerical
working and the displacer pistons, while heat is analysis is presented in [48].
absorbed from the low-temperature region at The success of the Stirling cryocooler
the cold end of the piston. Ideally, this process depends largely on the effectiveness of the
is isothermal. regenerator (see Section 5.1.4). The regenera-
Process 4 ! 1 (Figs. 22 and 21 D). The cold tor commonly used has a matrix of fine wires
gas is forced back through the regenerator by which results in a large surface area and high
the motion of the displacer piston while the regenerator effectiveness. The fraction of the
ideal refrigeration effect that is wasted by the
regenerator ineffectiveness is given by:

DQ ð1
eÞ½ðT2 =T3 Þ
1
¼
Q0A ðg
1Þlnðv4 v3 Þ

where g ¼ isentropic exponent, v ¼ specific

volume, and Q0A ¼ ideal heat-transfer rate from
the low-temperature region; the numerical sub-
Figure 22. Thermodynamic cycle of the ideal Stirling scripts refer to the processes listed above. If the
refrigeration system following numerical values are used, the
 Cryogenic Technology 25

importance of the regenerator may be demon- common (near-ambient) part of the machine
strated. For g ¼ 1.67 (helium gas), v4/v3 ¼ through the cold regenerator (b2) into the cold
1.50, T2 ¼ 300 K, and T3 ¼ 78 K, cylinder (d). Due to the constant volume of the
system, this step is accompanied by a decrease
DQ=Q0A ¼ 10:48ð1
eÞ in pressure. The resulting expansion in the
warm cylinder (a) will cause heat flow Qh
Thus, if the regenerator effectiveness is 0.99, from the hot-temperature source at Th. In the
the cooling effect loss is 0.108, i.e., >10 % of next phase, the cold displacer remains station-
the ideal refrigeration effect is lost due to a 1% ary and the hot displacer moves to the top
decrease in the regenerator effectiveness. As a cooling the warm end gas in the warm regener-
result, the lowest reachable temperature is ator (b1). The resulting pressure decrease ena-
mainly determined by the quality of the bles the gas in the cold cylinder to take up the
regenerator. cooling load Qc at cold temperature Tc. As the
warm displacer remains at the top position, the
cold displacer moves to the right heating the
4.6. Vuilleumier Refrigerator cold cylinder’s content in the cold regenerator
to ambient temperature. As a result the gas
The Vuilleumier refrigerator (VM), first inside the common volume (e) is compressed
patented by RUDOLPH VUILLEUMIER in 1918 and heat QR is discharged to ambient. The warm
[49], is a derivative of the Stirling cryocooler, end displacer now moves downward with a
except that the VM uses a thermal compression stationary cold end displacer. The gas from
process instead of the mechanical compression the common part is heated up in the warm
used in the Stirling system. The VM utilizes regenerator. The resulting additional pressure
only displacers and no working pistons thus increase causes more heat to be discharged to
allowing a fully hermetic construction of the ambient conditions.
machine [50]. A schematic of the VM is shown Since the VM is a combination of two Stirling
in Figure 23. machines, the COP for the ideal VM cryocooler
In a VM machine the motion of the cold is given by the quotient of the COP for a Carnot
displacer piston (c2) is approximately 90 out of heat-engine and a Carnot refrigerator:
phase with the motion of the hot displacer
piston (c1). With the hot displacer remaining Tc ðTh
To Þ
in the lower position, the cold displacer moves ðCOPÞid ¼
Th ðTo
Tc Þ
to the left, shifting the working fluid from the
where the subscript c denotes the cold cylinder,
h the hot cylinder, and o ambient temperature.
The COP of the actual machine is obtained by
division of refrigeration capacity over heat input.

Qc
COP ¼
Qh

The factors that contribute to reduction of

the COP for the actual system include the
following: (1) shuttle loss, caused by the mis-
match between the displacer and the cylinder
wall, (2) pumping loss, caused by the gas
volume between the displacer and the cylinder
wall, (3) heat transfer through the displacers
from the warmer end to the cooler end, (4) heat
transfer through the cylinder wall from the
Figure 23. Vuilleumier refrigerator
a) Hot cylinder; b) Regenerators; c) Displacer pistons;
warmer end to the cooler end, (5) frictional
d) Cold cylinder; e) Common volume energy-dissipation between the cylinder wall
26 Cryogenic Technology

and the displacer seals, and (6) losses because The system consists of a piston–cylinder
of the ineffectiveness of the regenerators. A arrangement (g) and (h) a regenerator (e),
special technical challenge for VM lies in and a compressor (a) with inlet (c) and outlet
ensuring the long-term reliability of a hermetic (d) control valves. The cold part of the piston is
machine. The FOM of a VM can be expected to at the lower end, whereas the upper portion of
lie below 0.2. the cylinder operates at ambient temperature.
An advantage of the VM cryocooler is that The cylinder is constructed of thin-walled stain-
the thermal input may be provided by isotope less-steel tubing, and the piston is constructed
energy or solar energy. This makes the unit of a low thermal conductivity plastic material
attractive for cryogenic cooling in long-term so that heat conduction from the warm end to
space exploration and in applications where the cold is minimized. A sliding seal (f) is
mechanical vibration of the drive engine placed at the warm end of the piston to prevent
must be avoided [28, pp. 186–187]. leakage of the working gas from the system.
Problems with operating a seal at cryogenic
temperatures are avoided by placing the seal at
4.7. Solvay Refrigerator the warm end. The piston generally fits quite
loosely within the cylinder, with up to 0.5 mm
The Solvay refrigerator was originally devel- clearance, except for the gas-tight seal at the
oped in Germany in 1887 by ERNEST SOLVAY warm end.
[51]. It was one of the first refrigerators planned The regenerator uses punched copper disks
for use in an air liquefaction system that had an or brass screens as the matrix. The screens or
expansion engine. However, Solvay’s prototype disks are separated by stainless steel wire spac-
system was unable to achieve temperatures ers to prevent axial conduction along the
<178 K. The system was not considered for regenerator.
cryogenic refrigeration until the late 1950s, The inlet and outlet control valves are
when GIFFORD and MCMAHON of A. D. Little coupled to the piston crankshaft so that the
constructed a Solvay refrigerator for cooling a valves may be opened and closed at the appro-
miniature infrared detector system [52]. A priate time, depending on the position of the
schematic of the Solvay cryocooler is shown piston. These valves control the flow into and
in Figure 24. out of the expansion engine.
The sequence of operations for the Solvay
cycle is as follows:
Process 1 ! 2 (pressure buildup). With the
piston at the bottom of the stroke, the inlet valve
is opened, while the outlet valve remains
closed. The high-pressure gas from the com-
pressor flows through the regenerator, in which
the gas is cooled; the system pressure increases.
Process 2 ! 3 (intake stroke). With the inlet
valve open and the outlet valve closed, the
piston is raised to draw a volume of gas, which
has been cooled by passage through the regen-
erator, into the cylinder.
Process 3 ! 4 (expansion stroke). The inlet
valve is closed and the piston continues to move
toward the top dead center position. The gas
within the cylinder expands (i) while doing
Figure 24. Solvay cryocooler work against the piston, and the gas tempera-
a) Compressor; b) Surge vessels (b1 is at p1, i.e., low ture decreases to a low level.
pressure in the cycle, b2 is at p2, i.e., high pressure in
the cycle); c) Inlet valve; d) Outlet valve; e) Regenerator;
Process 4 ! 5 (exhaust process). When the
f) Seal; g) Displacer piston; h) Cylinder; i) Expansion space; piston reaches the top of its stroke, the outlet
j) Load heat exchanger valve is opened and the gas pressure falls to the
 Cryogenic Technology 27

final minimum value. The exhaust value

remains open as the piston moves back toward
the bottom position to force the cold gas out of
the cylinder through a load heat exchanger.
Energy is absorbed from the external region,
which is to be cooled during this process.
Process 5 ! 1 The cold gas passes back
through the regenerator, where it is warmed
back to ambient temperature. The outlet valve
is then closed, the inlet valve is opened, and the
cycle is ready to be repeated.
The refrigeration effect for this cryocooler is
given by:

QA =qm ¼ ðh5
h04 Þ
ð1
had Þðh3
h04 Þ

where h3 ¼ enthalpy at the expander inlet, h04 ¼

enthalpy for an isentropic expansion from pres- Figure 25. Gifford–McMahon refrigerator
a) Compressor; b) Surge vessels (b1 is at p1, b2 is at p2); c)
sure p3 to p4, h5 ¼ enthalpy of the stream leaving Inlet valve; d) Outlet valve; e) Regenerator; f) Seal; g)
the load heat exchanger, and had ¼ adiabatic Displacer piston; h) Cylinder; i) Upper expansion space; j)
efficiency for the expansion process between Lower expansion space; k) Load heat exchanger
point 3 and point 4 (see process 3 ! 4).
The miniature Solvay cryocooler described and the outlet valve (d) closed, the inlet valve
by GIFFORD and MCMAHON can maintain a (c) is opened and the pressure within the cylin-
temperature of 55 K with a single stage. The der is increased from the low pressure p1 to the
cylinder is a stainless steel tube having a 5-mm high pressure p2. The volume of the lower
outside diameter, a 0.13-mm wall thickness, expansion space (j) is practically zero during
and a length of 51 mm. The piston is con- this process.
structed of a linear base-laminated phenolic Process 2 ! 3 (intake stroke). With the
compound. The regenerator was actually con- inlet valve open and the outlet valve closed,
tained within the piston in their design. The the displacer piston is moved from the bottom
working fluid is helium gas, which varies in position to the top of the cylinder. This piston
pressure between 0.35 and 1.73 MPa. The cool- motion moves the gas that was originally in
down time for the unit is ca. 3 min. the upper (warm) expansion space (i) through
the regenerator to the lower (cold) expansion
space. Because the gas is cooled as it passes
4.8. Gifford–McMahon Refrigerator through the regenerator (e), its volume tends
to decrease and additional gas is drawn in
In 1960 GIFFORD and MCMAHON [53] of A. D. through the inlet valve to maintain a constant
Little proposed a new cryocooler of which a pressure within the system.
schematic is shown in Figure 25 [54]. The Process 3 ! 4 (pressure release and expan-
Gifford–McMahon refrigerator (GM) differs sion). With the displacer at the top of the
from the Solvay refrigerator in the way that cylinder, the inlet valve is quickly closed and
two spaces are used within the cylinder, and the outlet valve is slowly opened, thus allowing
the piston does negligible work as it moves the gas within the lower expansion space to
within the cylinder. The displacer piston simply expand from pressure p2 to the low pressure p1.
moves the gas from one volume to the other The gas that remains in the lower expansion
during the cycle. space does work in forcing out the gas that
The sequence of operations for the GM cycle leaves during this process; therefore, the tem-
is as follows: perature of the remaining gas drops. Ideally, the
Process 1 ! 2 (pressure buildup). With the expansion takes place reversibly and adiabati-
displacer (g) at the bottom of the cylinder (h) cally (isentropic expansion).
28 Cryogenic Technology

Process 4 ! 5 (exhaust stroke). With the In small-scale systems, regenerators are less
outlet valve open, the low-temperature gas is expensive than recuperative heat-exchangers,
forced from the lower expansion space by and the regenerator can be constructed with a
moving the displacer down to the bottom of larger heat-transfer surface area per unit vol-
the cylinder. The cold gas flows through a heat ume, which is important for miniaturization.
exchanger (k) to absorb energy from the low- Because of the back-and-forth motion of the gas
temperature source. through the regenerator, impurities deposited
Process 5 ! 1. The gas flows back through during the intake process are swept out of the
the regenerator, where it is warmed to ambient system during the exhaust stroke; thus, prob-
temperature, and the cycle is ready to be lems with impurity buildup are minimized with
repeated. the use of a regenerator.
The refrigeration effect for the GM system is The GM system is also well-suited for multi-
given by: staging [55]. In some cryogenic systems, refrig-
eration is required at intermediate
QA =qm ¼ had ½ðqm Þe =qm ðh5
h04 Þ temperatures, for example, to cool radiation
shields around the unit. In this case the system
where had ¼ adiabatic efficiency, for the expan- can be operated with two stages, one for cooling
sion process from point 3 to point 4 (process 3 the shields at 77 K, for example, and the other
! 4), (qm)e ¼ mass of gas within the lower stage to provide the low-temperature refrigera-
expansion space at the end of the expansion tion at 20 K. In addition, a multistage refriger-
process, qm ¼ total mass compressed, h04 ¼ ator generally has a higher thermodynamic
enthalpy for an isentropic expansion from point performance than a single-stage unit. Multi-
3 to point 4, and h5 ¼ enthalpy of the stream staged systems were able to reach temperatures
leaving the heat exchanger. Because the volume as low as 6.5 K. A typical two-stage GM
of the cold expansion space remains essentially refrigerator is shown in Figure 26.
constant during the expansion process, the mass The principal disadvantage of the GM cry-
ratio may be calculated from the density ratio: ocooler is that its COP is inherently smaller
than that of a Claude cryocooler operating
ðqm Þe =qm ¼ r04 =r3 between the same pressure limits. This due to

The GM process allows spatial separation of

the compressor and cold end. Due to this
design, GM tend to be much heavier than
similar rated Stirling coolers but allow the
utilization of efficient, high-speed compressors
together with low-speed displacers (typically
1–5 Hz). They therefore possess very low levels
of vibration at the cold end.
The GM refrigerator has several advantages
over Claude-type refrigerators, which also use
an expander. In the GM cryocooler, all valves
and seals operate at ambient temperature; there-
fore, problems with low-temperature seals and
valves are eliminated. Problems with leakage
around the piston seals are reduced because the
viscosity of a gas near ambient temperature is
much higher than at cryogenic temperatures.
The GM system thus allows the seals to be
constructed of conventional materials.
The GM refrigerator uses a regenerator
Figure 26. Two-stage Gifford–McMahon refrigerator
instead of a recuperative heat-exchanger a) Regenerators; b) Seals; c) Displacer pistons; d) Load heat
employed in the Claude and J–T refrigerators. exchanger
 Cryogenic Technology 29

the fact that only a fraction of the total mass

compressed in the GM system is expanded
during the expansion process. Practical GM
cryocoolers will reach a FOM of ca. 0.05 to
0.10 [56]. The cooling performance of com-
mercial GM cryocoolers ranges from the lower
watt range at 4.2 K, through tens of watt at 30 K
up to a several hundred watts at 80 K.

4.9. Pulse Tube Refrigerators

The major drawbacks of Stirling and GM

refrigerators originate from the moving dis-
placer in the cold section: A source of vibration,
wear, axial heat conduction as well as shuttle
heat loss. In 1963 GIFFORD and LONGSWORTH
[57] introduced the first regenerative refrigera-
tor without moving parts in the cold end: the
Pulse Tube refrigerator (PTR). However, the
initial PTR was hardly able to reach cryogenic
temperatures. In 1984 MIKULIN and co-workers
[58] added an orifice and a reservoir vessel to Figure 27. Orifice pulse tube refrigerator
the original design, thereby reaching 105 K. a) Compressor; b) Regenerator; c) Pulse tube; d) Heat
The setup is known as Orifice Pulse Tube exchangers; e) Orifice; f) Reservoir
Refrigerator (OPTR). Shortly after, RADEBAUGH
and his team [59] were able to achieve 60 K by
re-locating the orifice to the warm end. Further is pushed through the aftercooler and the regen-
improvements enabled the PTR to reach tem- erator by means of a compression shock wave,
peratures around 20 K with one stage and even where it is cooled down to cold temperature.
1.3 K with two stages [60]. An OPTR operated The gas inside the pulse tube is compressed
with a superfluid mixture of 3He–4He was able adiabatically and its temperature rises. (2)
to reach a temperature as low as 0.6 K [61]. When the flow rate through the orifice picks
A schematic drawing of an OPTR is shown up, the continuing downward motion of the
in Figure 27. Its main components are a com- piston causes the system pressure to remain
pressor (a) with an aftercooler (d1), a regenera- constant. The working gas moves upwards
tor (b), the cold-end heat exchanger (d2), the through the pulse tube. The gas inside the
pulse tube (c), a second warm-end heat reservoir is compressed adiabatically. With
exchanger (d3), and an orifice (e) coupled the beginning upward movement of the piston,
with a reservoir vessel (f). The orifice is an the system pressure starts to decrease. The cold
obstruction for the flow from and to the reser- gas inside the pulse tube expands adiabatically
voir causing a phase shift between system and its temperature drops. (4) As the flow rate
pressure and mass flow. In analogy to the through the orifice increases, the continuing
displacer of Stirling and GM, heating and cool- upward motion of the piston causes the system
ing effects are separated by the flow oscillation pressure to remain constant. The expanded gas
through the orifice. Ideally, the OPTR operates inside the pulse tube moves downwards and
with adiabatic compression and adiabatic passes through the cold heat exchanger where
expansion inside the pulse tube. the cooling load Qc is absorbed. Warm gas from
The sequence of operation for OPTR can be the reservoir is cooled in the second warm heat
described as follows: (1) As the piston moves exchanger.
down the system pressure rises. The portion of Due to its large range of ideal gas properties
the working gas located inside the compressor and low condensation temperature helium is
30 Cryogenic Technology

commonly used as the working gas for this type of isentropic manner. When a gas is compressed
refrigerator. As PTR possess no moving parts in isothermally, the entropy of the gas decreases
the cold end they show even lower levels of because the order of the molecules increases
vibration than GM with reduced cost and (the atoms occupy a smaller volume with no
increased reliability and life time. Depending increase in random molecular velocities). If
on their operating frequency, PTR are subdivided the high-pressure gas is expanded reversibly
into Stirling- (10–50 Hz) and GM-types (1–5 Hz). and adiabatically, the entropy of the gas
Commercial PTR are available from a maximum remains constant; therefore, the temperature
of 5 Wat 50 K up to 200 Wat 80 K with a FOM of of the gas decreases to offset the increase in
typically 0.03 to 0.20 [62, 63]. Prototypes of disorder produced by moving the molecules
OPTR delivering 2 kW of refrigeration at 130 farther apart.
K have recently been operated [64]. A paramagnetic material produces order
(reduces entropy) differently. In the absence
of a magnetic field, the dipoles of the magnetic
4.10. Magnetic Refrigerators material are more or less randomly arranged
because of the disturbing effect of random
When a liquid or gas is used as the working thermal vibrations. If an external magnetic field
fluid in a refrigerator, the operating temperature is applied to the material while the material
is limited to values above ca. 1 K. All materials temperature is kept constant, the entropy or
other than ordinary helium (4He) and the disorder of the material decreases, because
helium isotope 3He are in the solid phase at the dipoles tend to line up with the magnetic
temperatures below 1 K. The low temperature field. If the magnetic field is removed
that can be attained with boiling helium is reversibly and adiabatically, the temperature
determined by the vapor pressure of the helium. of the paramagnetic material decreases to main-
At 1 K, the vapor pressure of 4He is 16 Pa, and tain constant entropy. This process, which is
the vapor pressure of 3He is 1180 Pa. At 0.5 K, called adiabatic demagnetization, is the basis
the vapor pressures are 1.9 mPa (4He) and 22.2 for the magnetic refrigerator (MR) used to
Pa (3He) [65]. Clearly, a refrigerator operating reach temperatures down to 0.1 K.
at these low pressure levels in the evaporator is A schematic of the basic MR is shown in
impractical for relatively large capacities Figure 28. Lines of constant magnetic field
because of the size of the flow passages strength replace the constant pressure lines
required for even moderate mass flow rates. on the temperature–entropy diagram for the
History. GIAUQUE and DEBYE independently cycle. The MR consists of the following: a
suggested a method to break this “temperature working salt (e), such as iron ammonium
barrier” in the 1920s [66, 67]: A paramagnetic alum, which is the cooling medium for the
material could be used instead of a liquid or gas refrigerator; a reservoir salt (f), which acts as
as the working medium to reach temperatures a “thermal flywheel” to smooth out temperature
<1 K, because the entropy associated with the fluctuations in the space to be cooled; thermal
electron-spin system becomes much larger than valves (b) to control heat transfer to and from
the entropy associated with the lattice vibrations. the working salt; and solenoids (c and d), which
In 1933 GIAUQUE and MACDOUGALL (who won a are used to apply the required magnetic fields to
Nobel prize for the feat) were able to verify the the working salt and the thermal valves.
magnetic cooling concept experimentally [68]. The sequence of operations for this refriger-
They reached a temperature of 0.53 K with ator is as follows:
gadolinium sulfate after demagnetizing from a Process 1 ! 2 (corresponds to the isother-
field of 0.8 T and an initial temperature of 3.5 K. mal compression process in the Carnot cycle).
DEHAAS and coworkers attained a temperature of With the upper thermal valve open (heat trans-
0.27 K 2.5 weeks later by demagnetizing CeF3 fer is allowed) and the lower thermal valve
from 2.8 T at 1.3 K [69]. closed (no heat transfer), the magnetic field
The principle of cooling by adiabatic around the working salt is increased from
demagnetization is similar to the cooling 0.14 to 0.72 T. Heat transfer occurs between
effect produced when a gas expands in an the working salt and the liquid helium bath to
 Cryogenic Technology 31

Process 2 ! 3 (corresponds to the reversible

adiabatic expansion process in the Carnot
cycle). Both thermal valves are closed so that
the working salt is thermally isolated (adiabatic
process) and the main magnetic field is reduced
from 0.72 to 0.10 T. The temperature of the
working salt decreases to ca. 0.16 K from an
initial temperature of 1.15 K.
Process 3 ! 4 (corresponds to the isother-
mal heat addition process in the Carnot cycle).
The thermal valve between the working salt and
the reservoir salt is opened to allow heat
exchange between the two materials, and the
magnetic field around the working salt is
reduced to zero. Heat is transferred to the
working salt from the reservoir salt to maintain
constant temperature.
Process 4 ! 1 (corresponds to the reversible
adiabatic compression process in the Carnot
cycle). The working salt is again thermally
isolated by closing both thermal valves, and
the magnetic field around the working salt is
increased from 0 to 0.14 T. The temperature of
the working salt increases to the initial value,
and the cycle has been completed.
The energy absorbed as heat from the low
temperature region (the reservoir salt) in the
ideal case is given by Equation (15)

QA ¼ qm T3 ðs4
s3 Þ ð15Þ

where qm ¼ mass of the working salt and s ¼

magnetic entropy of the working salt. Ideally,
the heat rejected from the working salt to the
helium bath is given by:

QR ¼ qm T1 ðs2
s1 Þ ¼
qm T1 ðs4
s3 Þ

The net magnetic work done on the working salt

is equal to the net heat transfer for a cycle;
therefore,

W net ¼ qm ðT1
T3 Þðs4
s3 Þ ð16Þ

Figure 28. Magnetic refrigerator
A) Schematic diagram: a) Liquid helium; b) Thermal valve; Equations (15) and (16) show that the COP of
c) Valve magnet; d) Main magnet; e) Working salt; the ideal MR cycle is the same as that of a
f) Reservoir salt; g) Space to be cooled
B) Thermodynamic diagram for the cycle Carnot cycle.
Irreversibility reduces the performance of
the actual MR below that of the ideal system.
maintain a constant temperature in the working For example, at a source temperature of 0.4 K,
salt. The working salt is thermally isolated from the Daunt–Heer–Barnes refrigerator [70] has a
the reservoir salt during this process. refrigeration capacity of 27.5 mW, whereas the
32 Cryogenic Technology

refrigeration capacity of the ideal system oper-

ating under the same conditions is 34.4 mW. At
a source temperature of 0.3 K, the refrigeration
capacity is 14.3 mW, compared with 25.8 mW
for the ideal system.
One of the advantages of the MR is that its
operation is not dependent on the presence of a
gravitational force; thus, the system can operate
effectively in a zero-gravity environment. Infra-
red bolometers in space systems have been
cooled by MR.
One of the critical components of the MR is
the thermal valve or “switch.” In the Daunt– Figure 29. Schematic of the magnetic wheel refrigerator
Heer–Barnes refrigerator, a thin lead strip is a) Wheel; b) Pump; c) Heat exchangers
used as the thermal valve. At a temperature of 1
K, the thermal conductivity of lead in the where D T is the intrinsic temperature change
superconducting state is ca. 0.025 times the of the magnetic material as it moves from the
thermal conductivity of the material in a normal low-field region to the high-field region. The
state; at 0.2 K, the ratio is ca. 0.010. Thus, the helium gas is warmed in the high-field region to
valve can be “closed” (little heat transfer a temperature (Th þ DT ) by heat exchange with
through the valve) by decreasing the magnetic the magnetic material. The helium gas finally
field around the valve to drive the lead into the flows through a warm heat exchanger (c2), in
superconducting state. The valve can be which heat is rejected to the heat sink, and the
“opened” by increasing the magnetic field so gas is cooled back to temperature Th.
that the lead strip returns to the normal state. In a working model of the magnetic wheel
Changes in magnetic field around the thermal refrigerator, a 3 T magnetic field was used in
valves are accomplished with small “valve the high-field region, and the rotational speed
magnets” around the lead strips. Other types of the 150 mm diameter wheel was 6 rpm. A
of thermal switches have been tried, but they are temperature difference of 7 K was maintained
generally not as effective as the superconduct- with a 500 W thermal load on the cold side.
ing switch. For example, mechanical switches
allow zero thermal leakage when contact is
broken; however, energy is dissipated when 4.11. Dilution Refrigerator
the contact is made or broken, and this energy
dissipation cannot be tolerated in the region One of the most widely used refrigerators for
below 1 K. the temperature range between 0.005 and 1 K is
the 3He–4He dilution refrigerator (DR). This is
Magnetic wheel refrigerators have been a relatively new development in cryogenic
constructed for use in the temperature range technology.
between 2 and 10 K [71, 72]. A schematic of History. In 1951 HEINZ LONDON suggested
this type of refrigerator is shown in Figure 29. that cooling could be achieved by dilution of
Helium gas enters the wheel (a) in the low- liquid 3He with superfluid liquid 4He [73].
field region at a temperature of Th and is cooled Little interest existed for developing a practical
to Tl by heat exchange with the magnetic refrigerator at that time, because the phase
material. The wheel is constructed of granules separation of 3He–4He mixtures had not been
of gadolinium sulfate for improved heat discovered. LONDON [74] presented a practical
exchange with the helium gas. The gas leaves design for the DR in 1962 after the phase
the low-field region of the wheel and flows separation phenomenon was discovered exper-
through a heat exchanger (c1) in which the imentally by WALTERS and FAIRBANK in 1956
cooling load is absorbed by the helium gas. [75]. The first DR was constructed by DAS and
The gas enters the high-field region of the coworkers [76] at the University of Leiden in
magnetic wheel at temperature (Tc þ DT), 1965. In 1966 HALL and associates [77] and
 Cryogenic Technology 33

NEGANOV and others [78] constructed dilution

refrigerators that achieved temperatures
between 0.025 and 0.060 K. Commercial
DRs can continuously reach 0.005 K.
The entropy of pure liquid 4He below 1 K is
quite small and varies proportionally with T3.
On the other hand, the entropy of liquid 3He
below 1 K is somewhat larger and is propor-
tional to T. Thus, the 4He in a 3He–4He mixture
contributes little to the total entropy of the
mixture below 1 K.
Because 3He has a higher entropy in the
dilute phase mixture (dilute in 3He), an amount
of heat Q ¼ qmTDs is absorbed for each mole
of 3He that passes from the pure phase to the Figure 30. Low-temperature phase diagram for 3He–4He
dilute phase in a reversible manner. The main mixtures
a) Superfluid (He II); b) 2-phase region; c) Normal fluid (He
problem is how to mix the two materials in an I); d) Tricritical point; e) l-line
appropriate manner. Fortuitously, nature pro-
vides a reversible mechanism for the dilution:
the helium mixture separates into a dilute
phase and a concentrated phase at tempera-
tures below 0.86 K. The concentrated phase is
less dense than the dilute phase and, therefore,
floats on top of the dilute phase. The concen-
trated phase is a normal fluid (He I), whereas
the dilute phase, mainly consisting of 4He, is a
superfluid (He II). Because the two phases are
in thermodynamic equilibrium, 3He can move
reversibly by diffusion from the low-entropy
concentrated phase to the high-entropy dilute
phase, analogous to the evaporation of (low-
entropy) liquid molecules into the (high-
entropy) vapor phase above the liquid. In
Figure 30 a low-temperature phase diagram
for mixtures of 3He and 4He is depicted. The
mole fraction of 3He in the dilute phase at
absolute zero is not zero, but instead
approaches a finite value of xD ¼ 0.064.
This fact means that at low temperature the
high-entropy phase has a particle density that
is high enough to permit relatively high diffu-
sional flow rates for the 3He, which is impor-
tant in the design of a practical refrigerator.
Another property that permits design of a
practical DR is the fact that 3He has a much
higher vapor pressure than 4He at the same
temperature. This means that the vapor in equi-
librium with the liquid phase is practically pure
3
He for temperatures below 1 K, allowing Figure 31. 3He–4He dilution refrigerator
ample differences of the specific entropy in a) Vacuum pump; b) Heat exchangers; c) Helium baths;
d) Constriction; e) Still; f) Mixing chamber: f1 is concen-
the mixing chamber. A schematic of a trated mixture (85 mol% 3He), f2 is dilute mixture (6.4 mol
3
He–4He DR is shown in Figure 31. % 3He)
34 Cryogenic Technology

The gas (which is practically pure 3He) is is T1/Tm ¼ 2.80. For a value of Tm ¼ 0.040 K,
compressed in a vacuum pump (a) from ca. 4 Pa the largest value for T1 for a positive refrigera-
to 4 kPa. The gas is cooled in a heat exchanger tion effect is 0.112 K, which yields a maximum
(b1) and a liquid helium bath (c1) at 4.2 K, and is temperature difference at the cold end of the
then condensed in a liquid helium bath (c2) at exchanger of (T1
Tm) ¼ 0.072 K. Heat
reduced pressure and a temperature of 1.2 K. exchangers with large surface areas per unit
The liquid 3He is expanded through a capillary volume have been developed using sintered
tube constriction (d) and is cooled further in the metal elements within the flow passages (see
still (e), which operates at ca. 0.6 K. The liquid Section 5.1.2 and [80]). For further reading on
is further cooled in a heat exchanger (b2) before DR refer to [81].
it enters the mixing chamber (f), where 4He is
mixed with the 3He at temperatures between
0.050 and 0.005 K.
5. Components of Liquefiers and
The liquid in the mixing chamber separates Cryocoolers
into two liquid phases, a less dense concen-
trated 3He mixture (f1) and a diluted 3He mix- The three subassemblies of liquefaction
ture (f2) of higher density. As the dilute mixture systems and cryocoolers that influence the
is depleted of 3He by evaporation in the still, performance most are the heat exchanger or
3
He molecules diffuse from the concentrated regenerator, the compressor, and the expander.
phase to the dilute phase, with heat QA being Besides, especially cryogenic separation sys-
absorbed during this process. The cooling effect tems and downstream processes rely on pumps
is thus produced at the phase boundary inside to be operated at low temperatures. Addition-
the mixing chamber. The dilute mixture returns ally, the choice of insulation plays an important
through the heat exchanger to the still, where role in all cryogenic systems. So these compo-
heat is added to evaporate the 3He from the nents will be introduced shortly in the following
mixture. The mole fraction of 3He in the liquid chapter. Other components include, e.g., the
in the still is ca. 1 %, and the composition of the J–T valve, filters, purifiers, and the vessels.
vapor is between 95 and 98 mol% of 3He. Detailed information on a wide range of tech-
The refrigeration effect of the DR is given nological aspects of cryogenic systems is given
by: in respective text books [22, 28, 82, 83].

QA ¼ ðqv ÞHe ðhm
hi Þ 5.1. Cryogenic Heat Exchangers

where (qv)He ¼ molar flow rate of 3He, hm ¼ One of the most critical components in any
molar enthalpy leaving the mixing chamber, cryogenic liquefaction system or refrigerator
and hi ¼ molar enthalpy of the 3He entering is the heat exchanger. In conventional systems
the mixing chamber. For temperatures below using heat exchangers, such as regenerative
ca. 0.04 K, the enthalpies may be approximated gas turbine units, the system can operate sat-
by Equations (17) and (18) [79]: isfactorily with a heat exchanger having an
effectiveness (e) as low as 0.50. However, a
hm ¼ C1 T2m where C1 ¼ 94 J mol
1 K
2 ð17Þ heat exchanger effectiveness of 0.85 or higher
is required if the liquefier is to produce liquid
at all.
hi ¼ C2 T2i where C1 ¼ 12 J mol
1 K
2 ð18Þ The reason for this can be related directly to
the amount of exergy loss caused by irreversible
One of the more critical components of a DR entropy production sirr during heat transfer
is the heat exchanger between the mixing across a finite temperature difference:
chamber and the still, which must have a Q12 T1
T2
sirr ¼
high effectiveness; otherwise the system will qm T 1  T 2
not work. From Equations (17) and (18), the
maximum temperature ratio (corresponding sirr is proportional to the amount of heat trans-
to zero heat transfer in the mixing chamber) ferred (Q12) temperature difference between the
 Cryogenic Technology 35

two reservoirs (T1–T2) as well as inversely Tubular Heat Exchangers. Four basic
proportional to their absolute value (T1T2)
1. tubular exchangers are shown in Figure 32.
Looking at above equation it becomes evident The simplest form of tubular exchanger is the
that the temperature level contributes more and double-pipe or concentric-tube heat exchanger
more to irreversibility during finite temperature- (Fig. 32 A). This was the unit used by LINDE in
difference heat transfer at low absolute temper- his original air liquefier. The fluids are in
ature levels. If one consequently limits the driv- counterflow, with the high-pressure warm
ing temperature difference, large heat exchange stream (a) in the inner tube and the low-pressure
surface requirements arise which result to cold stream (b) in the annular region between
increased frictional losses inside the heat the two tubes. To conserve space, the concentric
exchanger. Thus, a careful design of heat tubes are wound in a large helix, and the unit is
exchangers is one of the keys to achieving effi- insulated to reduce heat transfer from the
cient refrigeration at low temperatures. surroundings.
For example, nitrogen has the following
enthalpy values [84]: Concentric-Tube Heat Exchanger with Wire
Spacer. The performance of the concentric-
h01 ¼ 462.15 J/g at 300 K and 0.101 MPa tube exchanger may be improved by wrapping a
h2 ¼ 431.50 J/g at 300 K and 20.3 MPa wire or plastic spacer around the inner tube
hg ¼ 228.66 J/g at 0.101 MPa and saturated (Fig. 32 B). This spacer tends to maintain a
vapor. concentric relation of the inner tube with the
outer tube and forces the low-density gas in the
Using Equation (5) with zero heat inleak from
ambient, the minimum heat exchanger effec-
tiveness (emin) for the basic Linde–Hampson
liquefier (or J–T cryocooler) is found to
be:

ð462:15
431:50Þ
emin ¼ 1
¼ 0:869
ð462:15
228:66Þ

The effectiveness of the heat exchanger must be

higher than 0.869 in this example for the lique-
fier to work. The importance of the heat
exchanger may also be demonstrated by noting
that the liquid yield for a Linde–Hampson sys-
tem with a perfect heat exchanger is 0.0708,
whereas for a heat exchanger with an effective-
ness of 0.99, the liquid yield for the Linde–
Hampson liquefier is decreased to 0.0658. A
reduction of the effectiveness by 1 % results
in a 7.1 % reduction in the liquid yield.

5.1.1. Types of Cryogenic Heat Exchangers

Many different types of heat exchangers (also

called recuperators) are used in cryogenic liq-
uefaction and refrigeration systems. The units
may be classified as follows: Figure 32. Tubular heat exchangers
A) Linde concentric-tube exchanger; B) Concentric-tube
1. Tubular heat exchangers exchanger with wire spacer; C) Multiple-tube exchanger;
D) Bundle-tube exchanger
2. Plate-fin heat exchangers a) High-pressure stream; b) Low-pressure stream;
3. Perforated-plate heat exchangers c) Intermediate-pressure stream
36 Cryogenic Technology

annular space (b) to follow a long helical path,

thereby increasing the gas velocity and improv-
ing the heat transfer coefficient.
Several tubes may be contained inside a
larger tube to form a multiple-tube exchanger
(Fig. 32 C). This unit can be used in the Linde
dual-pressure system for the three-channel
heat exchanger. In this case, the high-pressure
stream (a) flows in one of the smaller tubes,
the intermediate-pressure stream (c) flows in
the other smaller tubes, and the low-pressure
stream (b) flows in the space between the
small tubes and the larger enclosing tube.
Spiral windings, spacers, or longitudinal fins
are used to locate the small tubes within the
larger tube.
The bundle- or cluster-tube heat exchanger
is formed by brazing or soldering several tubes
around a central tube. The high-pressure stream
generally flows in the central tube (a), while the
low-pressure stream flows in the tubes (b)
clustered around the center tube.
Giauque–Hampson Heat Exchanger. The
classic heat exchanger used for medium- to
large-scale cryogenic liquefiers is the Figure 33. Giauque–Hampson heat exchanger
Giauque–Hampson type shown in Figure 33. a) Insulation; b) Close-fitting shroud; c) Spiral-wound
One or more coils of thin-walled stainless tubes; d) Punched spacer; e) Cylindrical mandrel
steel, aluminum, or copper tubes (c) are wound
onto a low thermal conductivity mandrel (e),
such as a stainless steel cylinder. A close-fitting A thin strip of cellulose acetate was placed
shroud (b) of the same low thermal conductivity between the spacer and the tube as the coil was
material is placed around the coil, and insulation being wound. After winding was complete, the
(a) is added on the outside. The high-pressure acetate was dissolved by acetone. Several lay-
stream flows inside the small tubes, while the ers of tubes were used to provide many parallel
low-pressure stream flows in a counter-cross paths for the low-pressure fluid, thereby reduc-
flow arrangement across the outside of the heli- ing the overall pressure drop on the low-
cal coil of small tubes between the inner mandrel pressure side.
and the outer shroud (inside the shell). Modern Giauque–Hampson-type heat
The fluid tends to follow the path of least exchangers are referred to as coil-wound heat
frictional resistance (through shortest tube or exchangers (CWHE). A schematic of a large-
the widest spacing), thus avoiding mal- scale CWHE is shown in Figure 34. The indi-
distribution of flows in both shell and tube sides vidual tubes are connected to headers at both
is a key challenge. For the high-pressure stream ends of the heat exchanger to allow the stream
inside the small tubes, approximately equal to be brought in and out of the exchanger in a
length flow passage is realized by adjusting single line. Heat exchange areas of CWHE
the helical pitch from layer to layer. For the range from 0.2 to more than 25 000 m2, and
shell side an uniform tube spacing is aspired. as much as five different streams may be incor-
An important contribution to solve the spac- porated [86]. The internal coiling of this heat
ing problem was made by GIAUQUE at the exchanger type allows good compensation of
University of California in introducing punched thermal stress as it typically occurs during cool-
brass spacers [85]. down, load change, and upset operation of a
 Cryogenic Technology 37

Figure 35. Collins heat exchanger

a) Concentric copper tubes; b) Edge-wound copper helix

exchangers have been manufactured with as

many as ten different flow passages.
Figure 34. Multistream coil-wound heat exchanger (Linde Plate-Fin Heat Exchangers. The plate-fin
AG)
a) Shell-side stream; b) Tube-side streams heat exchanger (PFHE, Fig. 36) consists of
stacks of alternating layers of corrugated die-
formed metal sheets, i.e., fins (g), between thin
plant. Typically CWHE are built to withstand metal sheets or plates (h). The edges of each
pressures up to 10 MPa on the shell side and 30 layer are sealed with bars (i), and the flow
MPa on the tube side. The maximum design passages formed within the corrugations are
pressures are, however, strongly dependent terminated with headers (e) equipped with a
upon the heat exchanger size and the choice distributor (f) at each end. Cross flow, counter
of material. flow, or multipass arrangements can be con-
structed with the individual sandwiches with
Collins Heat Exchanger. An extended- the possibility of handling ten and more streams
surface heat exchanger design developed by in the same unit. The stack may consist of as
SAMUEL COLLINS [87] is shown in Figure 35. many as 150 passages. A big advantage of
This was the heat exchanger used by PFHE is that individual streams may be added
COLLINS in his helium liquefaction system and withdrawn at any position along the ther-
(see Section 3.7). mal profile of the heat exchanger. The use of
The exchanger consists of several concentric optimized fins (Fig. 37) results in a high surface
copper tubes (a) with a close coiled helix of area, volume ratio, and a high effectiveness. As
edge-wound copper ribbon (b) in the annular much as 1 800 m2 of heat transfer area per cubic
space between the tubes. The helix is soft- meter of exchanger volume may be attained
soldered to both sides of the annular space in with a PFHE. Nowadays, the maximum module
which it is wound. The copper helix acts as a fin size is typically around 1.5 m  1.2 m  8.0 m.
or extended surface to increase the heat transfer The width and length of the modules are limited
surface area of the annulus. These heat by the size of the brazing oven; their height is
38 Cryogenic Technology

arrangement (Fig. 38 A) is commonly called

bath evaporator or submerged block evapora-
tor. Refrigerant circulation takes place by nat-
ural draft induced by evaporation
(thermosiphon effect). In case the liquid is
supplied from the top (Fig. 38 B) the arrange-
ment is referred to as falling film evaporator.
Perforated-Plate Heat Exchangers. With
increasing capacities of industrial-scale air sep-
aration units the necessity of attaining highly
effective and compact heat exchangers arose.
This led to the development of perforated-plate
heat exchangers (PPHE) by MCMAHON and
co-workers in 1950 [89]. This type of heat
exchanger consists of a stack of several parallel
perforated plates (b) with spacers (a) between
the plates (Fig. 39) [90]. PPHE are commonly
also referred to as matrix heat exchangers.
The plates are made of a high thermal
conductivity metal, such as copper or alumi-
num; the spacers are made of a low thermal
conductivity material such as plastic or stain-
less steel. The spacers are bonded to the plates
to provide sealing between the high- and low-
temperature streams and to minimize longitu-
Figure 36. Scheme of a brazed-plate fin heat exchanger dinal heat conduction in the exchanger. Heat
A) Overview (Linde AG); B) Detailed drawing exchange between the two fluid streams
a) Process streams; b) Header tank; c) Distribution fins; occurs by lateral conduction through the cop-
d) Corrugated heat-transfer fins; e) Partition plate; f) Side
bars
per or aluminum plates.
The diameter of the perforations can be made
limited by the high-temperature mechanical quite small so that a large heat transfer surface
strength of the fins under the gravitational area per unit volume can be attained. The peri-
load during brazing. Larger blocks are manu- odic disruption of the flow between each plate
factured by welding together multiple individ- results in high heat transfer coefficients (and
ual modules. Depending on the size, PFHE can high frictional losses); therefore, highly effective
be built for design pressure up to 12 MPa. PFHE heat exchangers of this type can be constructed at
possess lower frictional losses than perforated- relatively low cost. A comprehensive review on
plate heat exchangers and lower capital costs the history and on cryogenic applications of
and a higher degree of compactness than tradi- PPHE is given in [91].
tional Giauque–Hampson-type heat exchangers
and CWHE. Due to their wide application for 5.1.2. Heat Exchangers for Dilution
industrial purposes, the largest manufacturers Refrigerators
of PFHE are grouped in the Brazed Aluminum
Plate–Fin Heat Exchanger Manufacturers’ Heat exchangers for use in the dilution refriger-
Association (ALPEMA). Their standard [88] ator (DR) system present special thermal prob-
contains useful details on the properties and lems because of the large thermal boundary
limitations of brazed aluminum PFHE. resistance between helium and a solid surface
PFHE are used in a wide field of applications at very low temperatures. This thermal resist-
as, e.g., reboiler, condenser, superheater, and ance, called the Kapitza resistance, is inversely
subcooler. Two examples of PFHE arrange- proportional to T3, so the resistance can be quite
ments are shown in Figure 38. The left large in the 0.01 K temperature range.
 Cryogenic Technology 39

Figure 37. Principal fin types (Linde AG, [88])

A) Plain; B) Plain perforated; C) Serrated; D) Herringbone/wavy

The Frossati heat exchanger, developed for a Sintered metal powder heat exchangers rep-
DR, is shown in Figure 40 [80]. resent the present state of the art for DR heat
The heat exchanger consists of flow pas- exchangers and are used in commercial refrig-
sages separated by Cu–Ni foil (a), which is erators that produce 0.005 K continuously.
silver-plated and coated with a silver powder
(particle diameter 1 mm). The coating is placed 5.1.3. Heat Exchanger Design Problems
between two layers of sintered silver powder
(b), which are covered by rectangular channels Four problems may become acute in designing
(c). The sintered metal coating on the separator heat exchangers for cryogenic systems:
provides a large surface area per unit volume,
which reduces the Kapitza resistance. The liq- 1. The effect of variable fluid properties, in
uid helium mixture flows through the rectan- particular variation of specific heat
gular channel, instead of through the sintered 2. The effect of longitudinal heat conduction
metal, so the pressure drop through the heat
exchanger is not excessive. 3. The effect of uneven distribution of flow
The Frossati heat exchanger uses a sintered 4. The effect of multiphase flow
metal coating thickness of ca. 1 mm. The flow
passages vary in size from 12.2 m2 (dilute side) In many conventional, noncryogenic heat
and 5.0 m2 (concentrated side) at the cold end of exchanger design situations, the specific heats
the exchanger down to 4.0 m2 (dilute side) and 1.6 of the two fluids may be treated as essentially
m2 (concentrated side) at the warm end. The heat constant. For gases such as hydrogen in the
exchanger is 1.04 m long and is wound in a spiral cryogenic temperature range, the specific heat
to conserve space. A low temperature of 0.002 K may vary significantly with temperature (Fig. 41
is achieved continuously with this refrigerator. A). Respective data can be obtained from [92].
40 Cryogenic Technology

Figure 40. Frossati heat exchanger for a 3He–4He dilution

refrigerator
a) Separator; b) Sintered silver powder; c) Rectangular
channel

geometry under variation of flow rates (with

inherently coupled heat-transfer coefficients).
The two common approaches for calculating
heat exchangers are (1) the differential NTU-
effectiveness method, and (2) the integral

Figure 38. Arrangements of plate-fin heat exchanger in air

separation units
A) Thermosiphon arrangement; B) Downflow condenser/
evaporator
LOX ¼ Liquid oxygen; GOX ¼ Gaseous oxygen; LIN ¼
Liquid nitrogen; GAN ¼ Gaseous nitrogen

The effect may be emphasized if partial or total

condensation and evaporation occurs along the
heat exchanger length, as it happens, e.g., in all
liquefaction processes and during the evapora-
tion of refrigerant mixtures. This variation of
thermal properties can cause a thermal pinch
effect within the heat exchanger.
Two fundamental calculation modes for heat
exchangers exist: In case of heat exchanger
design, all temperatures and mass flow rates
are given and the necessary heat exchanger
surface area is calculated. Heat exchanger rating
in contrast involves the determination of missing
end point temperatures of a given heat exchanger

Figure 41. Specific heat and thermal conductivity relative

to the value at 300 K
A) Relative thermal conductivity; B) Relative specific heat
Figure 39. Perforated-plate heat exchanger a) Hydrogen at 2 MPa; b) Nitrogen at 1 MPa; c) Methane at
a) Plastic spacer; b) Perforated metal plate 4 MPa
 Cryogenic Technology 41

log-mean-temperature difference (LMTD) A and L describe the total heat exchange surface
method [93, 94]. area and the length along the heat exchanger
coordinate, respectively, dl is the length of the
LMTD Method. In case there is no occurrence infinitesimal balancing section.
of phase change, of approximately constant Table 4 presents data on the temperature
specific heats, and a known average heat- distribution in a gaseous helium–gaseous
transfer coefficient, the rate of heat transfer hydrogen counterflow heat exchanger operat-
Q may be expressed as: ing between 50 and 250 K for the hydrogen
Q ¼ F  U  A  DTm stream and between 60 and 260 K for the
helium stream. The mass flow rate of the
where U ¼ integral heat-transfer coefficient hydrogen stream is 1 kg/s, and the mass flow
expressed in W m
2 K
1, A ¼ heat exchange rate of the helium stream is 2.43 kg/s. The
surface area, and DTm expresses a suitable specific heat of the hydrogen gas is 10.55 kJ
mean temperature difference. F is a correction kg
1 K
1 at 50 K, and 14.04 kJ kg
1 K
1 at 250
factor allowing the treatment of different prac- K, which corresponds to a 40 % increase in the
tical heat exchanger types. In case of a parallel specific heat through the exchanger. If the
flow and counterflow arrangement F is equal to specific heats of the gases are assumed to be
unity. The LMTD is defined by: constant, the LMTD for the exchanger is 10 K.
However, because of the specific heat variation,
DTmax
DTmin
DTm ¼ ¼ LMDT the temperature difference (DT) becomes as
lnðDTmax /DTmin Þ
small as 3.1 K in the middle of the heat
DTmax and DTmin refer to the larger and smaller exchanger, and the actual LMTD is only 4.92
temperature difference between the two K. Thus, the actual, required heat transfer sur-
streams at the heat exchanger’s ends. In addi- face area determined by the NTU approach is
tion the energy balance around the heat about two times that calculated by the LMTD
exchanger must be adhered to: method (assuming constant specific heats).
The problem of “thermal pinch” may be
Q ¼ qm1  cp1 ðT1w
T1c Þ ¼ qm2  cp2 ðT2w
T2c Þ alleviated by “unbalancing” the heat exchanger
by increasing the hydrogen gas flow (if possi-
with qm ¼ mass flow rate and cp ¼ specific heat.
ble). Alternatively the helium gas flow can be
Subscripts w and c denote to the temperatures at
decreased by bleeding off a portion of the flow
the warm and cold end of the heat exchanger for
at some intermediate point in the exchanger.
the streams 1 and 2, respectively. Due to above
Longitudinal heat conduction along the walls
assumptions the LMTD approach will always
of conventional heat exchangers is usually not a
lead to a linear temperature profile within the
serious problem, because of the great length of
heat exchanger.
the heat exchanger. Miniature cryogenic
NTU Effectiveness Method. With the
occurrence of phase changes inside the heat Table 4. Temperature in a gaseous hydrogen–gaseous helium
exchanger and/or strongly temperature and heat exchanger
position dependent values for the heat-transfer Temperature of hydrogen Temperature of helium DT
coefficient, the NTU methods must be applied. (ðTH2 Þ), K (THe), K
The governing equations are the differential 50 60.0 10.0
conservation of energy, 70 76.9 6.9
90 95.6 5.6
dQ ¼
qm1  cp1  dT1 110 114.9 4.9
130 133.6 3.6
150 153.1 3.1
dQ ¼
qm2  cp2  dT2 170 173.5 3.5
190 194.5 4.5
coupled with the local kinetics of heat transfer: 210 215.9 5.9
230 237.8 7.8
250 260.0 10.0
A
dQ ¼ U   dl  ðT1
T2 Þ 
L DT ¼ THe
TH2
42 Cryogenic Technology

refrigerators can use heat exchangers that are as of heat-transfer coefficients for single and mul-
short as 50 mm. In addition, the effect of longi- ticomponent, multiphase flows. Lack of
tudinal conduction is most severe for a heat detailed knowledge on these phenomena will
exchanger that is very highly effective. inevitably cause the heat exchanger to either
One solution to the problem of longitudinal operate at decreased effectiveness or, in case of
conduction is the use of thin-walled tubes of a overdesign, be disadvantageous to system per-
relatively low-thermal-conductivity material formance by increasing frictional losses. In
(stainless steel). The tubes are coupled inter- addition criteria for the hydrodynamic stability
mittently by a plating of a high thermal con- of the fluid stream have to be considered (e.g.,
ductivity material (silver solder) between the Ledinegg instability [98]).
tubes [95]. The high-conductivity material
allows good heat transfer between the tubes, 5.1.4. Regenerators
and longitudinal conduction is reduced by the
thermal breaks between the sections of plating. In recuperative heat exchangers, the two fluids
Many cryogenic heat exchangers have multi- exchanging energy are separated by a solid
ple, parallel flow passages for one or both fluid surface (e.g., tube wall, plate). In a regenerator,
streams, instead of a single long flow passage for the two fluids occupy the same space alternately,
each stream. Ideally, the flow divides equally while the energy to be transferred is intermit-
among the parallel flow passages; in practice, tently stored and released from the regenerator
however, assurance of equal flow in each channel packing material (matrix). The temperature of
is not always possible because of differences in the fluid within the regenerator and the temper-
manufacturing tolerances. This results in a ature of the regenerator matrix are functions of
higher flow rate in the flow passages of smaller both position along the regenerator and time of
frictional resistance. The problems caused by operation. As a result, the performance of the
uneven flow distribution are more severe for regenerator is a function of the mass and the
highly effective heat exchangers, such as the specific heat of the matrix material, the fre-
cryogenic heat exchangers. quency of switching of the flow through the
The effect of uneven flow distribution in regenerator, and heat-transfer characteristics
cryogenic heat exchangers has been analyzed between the matrix and the fluid flow.
[96]. For a heat exchanger with a design effec- The regenerator in a cryogenic system con-
tiveness of 0.995, uneven distribution of flow sists of a matrix material contained within a
can reduce the effectiveness for the heat cylinder. For continuous operation, pairs of
exchanger in operation to only 0.95. regenerators are usually used so that the matrix
One technique for solving the problem of in one unit can be heated while the matrix in the
uneven flow distribution uses flow constrictions other unit is being cooled. Various materials
in the middle of the heat exchanger on the warm have been used for the matrix, including ran-
side [97]. If the temperature of the warm fluid domly packed small metal spheres, metal wires
increases because of a higher flow rate in the or strips, punched copper disks, and small-
passage, the constriction presents a higher flow diameter wire mesh screens. The specific
resistance, thereby reducing thewarm fluid stream heat of metals generally decreases rapidly
flow through that passage and balancing the flow. with temperatures in the cryogenic range. For
The variation of transport properties such as example, the specific heat of copper at 77 K is
viscosity and thermal conductivity with tem- only about half of its room temperature value.
perature causes continuously changing surface Hence special caution needs to be applied upon
requirement in order to transfer the same the choice of suitable regenerator materials for
amount of heat along the exchanger length cryogenic applications. Common materials
(Fig. 41 B). The effect is emphasized when include bronze and stainless steel screens
condensation and evaporation occur in the (300 to 30 K), lead and lead antimony spheres
streams as it inherently happens in liquefaction (30 to 10 K), and heavy rare-earth intermetallic
processes. Consequently, for cost effective and compounds (below 10 K) [99].
reliable design of heat exchangers special care The main advantages of regenerators over
has to be taken in predicting reasonable values recuperative heat exchangers are:
 Cryogenic Technology 43

1. A much larger heat transfer surface area per

unit volume can be achieved for the regen-
erator. As much as 4 000 m2 of surface area
per cubic meter of volume can be obtained
in a regenerator (compared with 1 800 m2/
m3 for a compact PFHE).
2. Regenerators tend to be self-cleaning
because of the periodic flow reversal.
Impurity buildup problems are not nearly
as severe as those in recuperative heat
exchangers.
3. Small-scale regenerators are generally sim- Figure 42. Typical operation ranges of various compres-
pler to fabricate, so the regenerator tends to sors types
cost less than a recuperator for the same heat a) Reciprocating; b) Screw; c) Centrifugal; d) Axial
transfer duty.

The main disadvantage of the regenerator is outlet temperature T2,act depends additionally
that some mixing of the warm and cold streams upon the adiabatic efficiency had of the stage,
occurs during the switching process. In addi- which for an ideal gas is given by:
tion, some means must be provided for switch-
T2;act ¼ cp ðT2
T1 Þ/had
ing the flow from one regenerator to the other in
the regenerator pair.
For high-pressure ratios intercoolers are
used to limit the gas temperature and reduce
5.2. Compressors the overall power consumption.
Compressors used in cryogenic systems gen-
The compressor is an important component of erally operate at ambient temperature; thus,
all liquefaction and refrigeration systems they are similar in design to compressors
because (1) it represents a significant fraction used for any gas compression service, with
of the total capital cost of the unit, (2) its power one important modification: Cryogenic com-
input accounts for the major portion of the pressors must be designed to reduce leakage
system operating costs, and (3) it is usually around seals and to reduce contamination of the
the component that determines the reliability of gas being compressed. Some applications
the system. Compressors used in cryogenic require the compressor inlet to be operated at
systems are classified as either positive dis- cryogenic temperatures, (e.g., compression of
placement compressors or turbo compressors. boil-off-gas from LNG storage, low-tempera-
Reciprocating compressors and screw compres- ture refrigerant compressors in J–T-cascades).
sors are positive displacement compressors, In this case special attention has to be paid
centrifugal and axial compressors form the towards choice of materials, lubrication and
group of turbo compressors. Typical opera- sealing concept, and manufacturing tolerances.
tional ranges for different compressor types
are indicated in Figure 42 [83, p. 442]. Positive Displacement Compressors.
During compression the gas temperature As shown in Figure 42, positive displacement
rises. During adiabatic, reversible compression compressors are applied for smaller suction
pressure and temperature increase are coupled volume flow rates. In positive displacement
by the isentropic exponent g of the gas: compressors gas is enclosed in a limited volume.
Subsequently the volume is reduced which is

k
1
p2 k accompanied by an increase of pressure. The gas
T2 /T1 ¼
p1 is released to the discharge side when the desired
pressure is achieved. This working principle
where subscripts 1 and 2 denote to the state prevents reverse flow through the compressor
before and after compression. The actual stage (surge) and makes the compressor insensitive to
44 Cryogenic Technology

changes in molar mass, suction pressure, and usually become too expensive compared to
suction temperature. the other machine types. Reciprocating units
are also typically used in systems requiring
Reciprocating compressors basically work high pressure (4–20 MPa).
like an air pump for the bicycle. When the piston One of the primary contaminants related to
moves back a vacuum is generated in the com- reciprocating compressors is the oil used to
pression chamber which forces the gas to enter lubricate seals in the compressor piston. In oil
through the inlet valves. When the compression lubricated compressors, provisions must be
chamber is filled with gas, the piston starts to made to separate the entrained oil from the
move forward, the volume is reduced, and the gas. Steel wool or glass wool filters are used to
pressure of the gas rises. When the pressure in remove liquid oil droplets, and refrigerated
the compression chamber exceeds the pressure adsorption units are used to remove oil vapor.
on the discharge side the gas leaves the cylinder Reciprocating compressors that are oper-
through the outlet valves. ated without oil lubrication avoid the problem
The pressure ratio of such a compressor of contamination of the gas being compressed.
stage is limited, normally by the discharge For safety reasons, this feature is required for
temperature of the gas. According to API 618 all oxygen compressors. Modern dry-running
[100], a maximum discharge temperature of compressors use either labyrinth seals or pol-
423 K is allowed. In today’s practice, higher ytetrafluoroethylene (PTFE) piston rings. In
discharge temperatures are possible due to case of labyrinth sealing the virtual absence of
improvements in valve technology. The corre- frictional wear between cylinder and piston
sponding pressure ratio is strongly dependent greatly reduces service intervals. PTFE rings
on the isentropic exponent of the gas. are usually reinforced with a hard filler of a
If the required overall pressure ratio is too metal powder, such as brass or copper in order
high for a single compressor stage, several to improve the wear. Glass-filled PTFE rings
consecutive compressor stages are used. In (known as Rulon-A) have been used with good
many cases an intercooler is installed between results.
the stages to limit gas temperature and power
consumption. Screw compressors are finding increased
The compression process in the cylinder has application in cryogenic systems. The rotary
a very high adiabatic efficiency (close to unit consists of two or more cylindrical rotors
100%). The losses mainly are produced by with multiple helical lobes that mesh as the
friction in the inlet and outlet valves. The rotors turn. A volume of gas is trapped between
valve-related losses are higher for gases with the lobes and is compressed as the volume of the
high molecular mass. The frictional losses can space decreases. The rotors are lubricated with
be reduced by choosing bigger valves (which oil sprayed onto the rotors at the low-pressure
makes the compressor more expensive). end of the compressor. On the discharge side the
Due to this, the adiabatic stage efficiency of oil is separated from the gas; only a residual oil
a reciprocating compressor varies strongly with content of typically 5 ppm mass fraction
the pressure ratio and molar mass of the gas. remains in the gas. Alternately dry running
Small pressure ratios and high frictional losses machines are used. In these machines direct
in the valves may reduce stage efficiencies contact between the two rotors is avoided by
below 50%. On the other hand, for light gases a timing gear that controls the relative move-
and high pressure ratios, piston compressor ment of the two rotors. This is necessary due to
stages can achieve similar adiabatic efficiencies the lack of lubrication. The technical parameters
as turbo compressors in the range of 85 to 90%. for the two types of screw compressors are very
As depicted in Figure 42 a wide range of different. Dry running screw compressors can
inlet volume flows exists, where the operating be used for high-suction volume flows up to ca.
ranges of reciprocating compressors, screw 80 000 m3/h (STP) while lubricated screws are
compressors, and centrifugal compressors over- limited to much smaller flows. In lubricated
lap. For suction volume flow rates above ca. 6 screw compressors, the oil absorbs part of the
000 m3/h (STP), reciprocating compressors compression heat and thus reduces the gas
 Cryogenic Technology 45

temperature. Therefore it is possible to reach efficiency of an axial compressor can exceed

relatively high pressure ratios of 1:10 and above 90%. Axial flow compressors are the most
within a single stage. In screw compressors the sensitive of all compressor types described
pressure difference between outlet and inlet here. Therefore for industrial applications in
results to a net mechanical load on the rotor most cases centrifugal compressors are pre-
and is thus limited. The maximum allowable ferred if axial and centrifugal compressors
pressure difference is dependent on the machine are both possible. In today’s large-scale refrig-
geometry. A pressure rise of 2 MPa per stage is eration applications the flow capacity of cen-
achievable in some cases. Rotary screw com- trifugal compressors is not exceeded, thus axial
pressors normally have a worse efficiency com- compressors are usually not used for these
pared to reciprocating and turbo compressors. purposes. Large air separation plants some-
times use axial compressors as the main air
Turbo compressors are generally more compressor.
compact and operate at higher speed than posi-
tive displacement units. The high operating Centrifugal flow compressors are also
speed allows the application of direct drives referred to as radial flow compressors. The
using gas or steam turbines. As shown in Figure adiabatic stage efficiency of large centrifugal
42, turbo compressors are generally used for compressors can reach 90%, values between 80
large-suction volume flows. One turbo compres- and 87% are common. Centrifugal compressors
sor can replace several positive displacement are widely spread for applications with high-
units. Turbo compressors consist of one or sev- suction volume flows. They provide high effi-
eral stages. A compression stage comprises ciencies, high reliability, and compact design.
rotating and stationary elements. In the rotor Most compressors in air separation plants are
the gas flow is accelerated and afterwards the centrifugal compressors. For the big refrigera-
kinetic energy is transformed into pressure in the tion compressors inside olefin plants also cen-
stator (diffusor). As the pressure ratio achievable trifugal compressors are used.
with a single stage is limited, several consecu-
tive stages are used for higher pressure ratios.
If the pressure ratio imposed on a turbo 5.3. Expanders
compressor exceeds the compression capability
of the compressor the gas starts to flow back- Expanders, or expansion engines, are machines
wards through the compressor. This initiates a that allow a pressure reduction of fluids under
phenomenon called surge (periodic change of the performance of work. Ideally, this takes
forward and backward flow direction) which place in an isentropic manner. As a result the
can quickly destroy the compressor. fluid, most commonly a gas, experiences a large
Due to the described working principle turbo decrease in temperature. Expanders are very
compressors are sensitive to changes in molar efficient means to produce refrigeration. Their
mass, suction temperature, pressure ratio, and main applications in cryogenic technology are
load. If such changes occur, often a displace- in air separation, liquefaction of hydrogen and
ment compressor is the better choice. helium, and conditioning and liquefaction of
Turbo compressors are not suitable for the natural gas [101].
compression of gases with low-molecular mass, The expanders used by CLAUDE (see
especially pure hydrogen. The reason is that the Section 3.5) were basically steam engines
pressure ratio achievable with a single stage is modified to operate with air instead of steam.
very small for light gases. Thus many stages are Initially, CLAUDE attempted to liquefy the gas
required to achieve the overall pressure ratio within the engine and to use the liquid air as
and the compressor becomes expensive. Hydro- lubricant [102]. This proved unsuccessful
gen compressors therefore normally are of the because of the low viscosity of liquid air and
positive displacement type. excessive heat transfer; thus, he finally used a
leather cup as the piston seal. In 1912 HEYLANDT
Axial flow compressors are used for very improved the efficiency of expanders by using a
high-suction volume flows. The adiabatic stage crowned piston concept [28, p. 325].
46 Cryogenic Technology

Reciprocating expanders with oil-lubricated where medium to large flow rates are required.
cylinders and exit temperatures as low as 90 K The maximum achievable flow rates lie below
were used in historic air liquefaction plants. ca. 100 000 m3/h (STP) with pressure ratios
The piston rings were usually made of an between 2 and 15, and at rotational speeds
aluminum–bronze alloy, and the cylinder has below 10 000 rpm [104].
a high-strength, replaceable, cast iron liner. To A significant contribution to cryogenic
reduce heat transfer down the piston, the stroke- expander technology was the development of
to-diameter ratio is ca. 2:1, and the piston is high-speed, gas-lubricated, bearings [105, 106],
constructed of stainless steel because of this which enabled the successful construction of
material’s low thermal conductivity. Hollow miniature refrigerators and liquefiers using rotor
piston rods were used to further reduce conduc- diameters as small as 8 mm operating at a speed
tion heat transfer. Expanders with oil-lubricated up to 500 000 rpm in the 1960s. Speeds of 900
cylinders were equipped with oil filters in the 000 rpm were achieved in helium liquefaction
discharge line to remove any oil particles. units in the 1980s [107]. In Figure 43 a detailed
For exit gas temperatures below ca. 90 K drawing of a small, radial flow expander includ-
or for systems in which the presence of oil ing a brake compressor is shown.
cannot be tolerated, nonlubricated PTFE pis- The following loss components have been
tons, similar to those in nonlubricated com- reported for a typical radial flow expansion
pressors, are used. The cylinder also contains turbine (adiabatic efficiency, had ¼ 84.0 %)
a replaceable liner. [103]:
Well-designed reciprocating expanders
reach an adiabatic efficiency of ca. 80–87%.
Typical values for the inlet pressures range
from ca. 4 to 20 MPa [82]. With reciprocating
expanders pressure ratios as high as 35 can be
utilized. They are generally limited to flow
rates below ca. 10 000 m3/h (STP).
In a study of the operating characteristics of
a reciprocating expander in an air separation
plant [103], the following components of the
energy dissipation were identified (adiabatic
efficiency, had ¼ 82%):

Nonideal expansion 6%
Valve losses 4%
Clearance losses 4%
Incomplete expansion 2%
Heat inleak 2%
Total losses 18 %

Although several patents were obtained for

rotary expanders (also turbo expanders) from
1898 to 1922, the first successful commercial
application of an expansion turbine was made
by LINDE in the 1930s. His turbine was an axial
flow, single-stage impulse turbine; inward
radial flow units were constructed later. With Figure 43. Cryogenic expander with brake compressor
installed shaft power as high as 3.5 MW, rotary (Sulzer AG, [108])
expanders are state-of-the-art components in a) Process connection; b) Intercooler connection; c) Brake
modern air separation units. Nowadays, rotary compressor rotors; d) Cooling water connection; e) Shaft;
f) Nozzle ring; g) Turbine rotor; h) Brake circuit filter;
expanders operate at adiabatic efficiencies as i) Axial bearing; j) Radial bearing; k) Velocity sensor;
high as 85–92% for larger units and are utilized l) Magnetic bearing
 Cryogenic Technology 47

Impeller losses 5.5 % adiabatic efficiency of had ¼ 72%, the work per
Blade (wheel) losses 3.0 %
Leakage 3.0 %
unit mass compressed is:
Disk friction 2.5 %
Exit loss 1.0 %
W=qm ¼ 363:97
ð0:72Þð0:60Þð166:44Þ ¼ 292:07 kJ=kg
Heat inleak 1.0 %
Total losses 16.0 %
The work per unit mass liquefied is:

The compressor efficiency does not influ-
W=ðqm Þf ¼ 292:07=0:2115 ¼ 1380:9 kJ=kg
ence the liquid yield in a liquefier or the
refrigeration effect in a cryocooler; however, The expander inefficiencies increase the work
the expander efficiency has a significant required to produce the same amount of liquid
effect on both quantities. For example, for by ca. 35 %. If an isothermal compressor
a Claude liquefier operating between 101.3 efficiency of 70 % is used for the calculations,
kPa at 300 K and 6.08 MPa, with nitrogen as the work per unit mass liquefied increases to 2
the working fluid, the enthalpy values are as 118 kJ/kg.
follows [84]:
h1 ¼ 462.15 kJ/kg at 300 K and 101.3 kPa 5.4. Pumps
h2 ¼ 450.52 kJ/kg at 300 K and 6.08 MPa
h3 ¼ 392.21 kJ/kg at 250 K and 6.08 MPa Various cryogenic processes rely on pumping
h0e ¼ 225.77 kJ/kg at 101.3 kPa and s3 of cryogenic liquids. Examples are the send out
hf ¼ 29.40 kJ/kg at 101.3 kPa and saturated of LNG (112 K) to the natural gas grid in LNG
liquid. regasification terminals and the internal pres-
surization of liquid oxygen (94 K) from the
For an expander mass flow rate ratio of x ¼ low-pressure column of large air separation
0.60, the liquid yield for a system with an ideal plants in order to produce a high-pressure gas-
expander is given by Equation (9): eous oxygen product. In both cases the cryo-
genic liquid is pumped to 3–10 MPa and
ð462:15
450:52Þ þ ð0:60Þð392:21
225:77Þ
y¼ ¼ 0:2576 subsequently evaporated. For scientific and
ð462:15
29:40Þ
commercial applications in the field of low-
For the same conditions but with an expander temperature physics, superconductors, and
having an adiabatic efficiency of had ¼ 80 %, space technology, there is need of small, reli-
the liquid yield is: able and highly efficient pumps capable of
circulating subcooled liquid helium [109, 110].
11:63 þ ð0:80Þð0:60Þð166:44Þ
Unlike compressors, pumps are much more
y¼ ¼ 0:2115 exposed to the possibility of performance
432:75
decrease and even damage by cavitation. Cryo-
The effect of the expander efficiency on the genic liquids often possess very low values of
work per unit mass liquefied is even more subcooling, so small heat ingress and frictional
dramatic. For an ideal compressor and ideal losses in the inlet can cause the pressure to drop
expander, the work requirement per unit mass below the liquid’s vapor pressure resulting to
compressed is found from Equation (10): partial evaporation, and consequently bubble
formation. The sudden increase in volume flow

W=qm ¼ ½ð300Þð4:421
3:169Þ
11:63 causes a decrease in suction mass flow coupled

ð0:60Þð166:44Þ ¼ 264:11 kJ=kg with risk of mechanical damage arising from
the downstream collapse of the gas bubbles
The work per unit mass liquefied is: when pressurized in the pump. The required
net positive suction height (NPSH) describes

W=ðqm Þf ¼ 264:11=0:2576 ¼ 1025:3 kJ=kg the minimum excess pressure that a liquid
needs to have in order to be pumped without
On the other hand, for a Claude liquefier with the risk of cavitation. To ensure a maximum
an ideal compressor and an expander with an degree of subcooling at the pump inlet by
48 Cryogenic Technology

means of hydrostatic height, cryogenic pumps lubricant from the cryogenic temperatures.
are usually located at the lowest possible level Additionally, a low NPSH value to avoid cavi-
of a process unit. tation and a low thermal mass to allow fast cool
down times are favorable. The main problems
Positive displacement pumps are used in associated with cryogenic centrifugal pumps
applications where very high pressures rises are are (1) cavitation, (2) pump bearings, and (3)
required with flow rates usually not exceeding pump seals [82, pp. 646–652].
ca. 50 m3/h. They can deliver differential pres- Centrifugal pumps can also be constructed
sures up to 70 MPa and more. A high output as submerged pumps (also submersible pump,
pressure is always achieved to the disadvantage Fig. 45) where the pump head, and sometimes
of flow rate. The construction of positive dis- the drive, is located below the liquid surface.
placement pumps is very similar to reciprocat- This design allows working without suction line
ing compressors. Due to the inherent contact thus maximizing the available subcooling at the
with cryogenic liquid, an external lubrication of pump inlet.
the piston is not required (and not desired). A The advantages of such pumps are [107,
sectional drawing of a cryogenic piston pump p. 253]:
used mainly for pumping liquid nitrogen,
argon, and oxygen is shown in Figure 44. 1. Shaft seals: Since both drive and pump heat
The pulsating outlet flow requires the installa- are located within a product atmosphere the
tion of a pulsation dampener. Reciprocating possibility of contamination and leakages
pumps are generally capable of tolerating cavi- through seals is eliminated
tation to some extent but will experience a
2. Mechnical design. The small temperature
strong decrease in mass throughput. Positive
gradient between drive and pump head
displacement pumps are commonly used in gas
allows the utilization of short drive shafts
filling station where a cryogenic liquid is
and simpler mechanical design of bearings
pressurized prior to evaporation.
3. Heat inleak. The positioning of the drive
Centrifugal pumps find use in applications under cold conditions greatly reduced the
where flow rates up to ca. 5 000 m3/h and lower heat ingress from ambient to the pump head
differential pressures are required. State-of-the- as well as the electrical resistance in the
art pumps have a high efficiency and a high motor windings due to the lower tempera-
thermal barrier to separate the bearings and the ture of the material

Figure 44. Piston pump for cryogenic liquids (Linde AG, [111])
a) Low-pressure inlet; b) High-pressure outlet; c) Vapor return; d) Vacuum insulated pump casing; e) Cylinder; f) Piston rod;
g) Piston rings; h) Packing; i) Suction valve (located in the tip of the piston); j) Discharge valve; k) Strainer; l) Lantern
 Cryogenic Technology 49

economic aspects that have to be considered

during cryocooler design. These are:

1. Efficiency. The useful refrigeration effect

and the lowest reachable temperature of
cryocooler are directly affected by the
amount of heat inleak. Additionally, once
a substance has been cooled down or a gas
has been liquefied it will often be subject to
transportation and storage for a period of
time ranging from hours to weeks. Thus
appropriate thermal insulation is required
to build and operate highly efficient cryo-
coolers and increases the economics of pro-
duction, transportation, and storage of
cryogenic substances.
2. Operability. Reliable steady-state operation
of cryogenic processes requires stable input
conditions. Insufficiently insulated cryo-
genic systems may be subject to a periodi-
cally fluctuating heat inleak that varies with
the ambient temperature (e.g., day–night
cycle). This issue usually becomes more
severe the smaller the system’s capacity
and the lower its temperature level.
Figure 45. Suction vessel-mounted, multistaged, submers-
ible LNG Pump (Ebara International) 3. Safety. Organisms that come into direct con-
a) Low-pressure inlet; b) Pump housing; c) Helical inducer; tact with cryogenic temperatures may suffer
d) Tail bearing; e) Wear rings; f) Axial pressure seal; from freeze burns that will require medical
g) Radial diffuser vane; h) Impellers; i) Radial diffusers; attention. Additionally, any uninsulated cry-
j) Main bearing housing; k) Stator; l) Motor spool;
m) Rotor; n) Coolant return line; o) Upper bearing; ogenic surface in contact to the ambient will
p) Electrical connection; q) High-pressure outlet; r) Multi- cause condensation of water vapor, leading to
stage orifice assembly; s) Discharge tubes; t) Submerged the formation of massive ice blocks possibly
electrical motor; u) Shaft; v) Axial thrust balancing device; beyond the mechanical strength of the struc-
w) Thrust equalization mechanism
ture. The condensation of air from liquid
hydrogen and liquid helium transfer lines
4. Maintenance. Possibility of withdrawal of exhibits a risk of freeze burns, intoxication,
the complete assembly from the vessel and even suffocation of operator personnel.

This pump type is commonly used for appli- For above reasons all low-temperature parts
cations where a high degree of subcooling has of cryogenic processes are usually insulated by
to be maintained for the pressurized liquid (e.g., suitable means. Typical values for the apparent
transfer pumps of LNG carriers and liquid mean thermal conductivity (W m
1 K
1) of
helium-circulation cryostats). A compact over- common insulation concepts for cryogenic sys-
view on submerged pump technology is given tems are given below:
in [112].
Foam glass / cellular glass 0.0350
5.5. Insulation Polyurethane/polyisocyanurate (PUR/PIR) foams 0.0330
Perlite 0.0330
Mineral fibers 0.0300
Unwanted heat ingress from the ambient into Aerogels 0.0200
cryogenic systems, i.e. heat inleak, is the cause Perlite/aerogel þ vacuum 0.0020
for a multitude of negative technical and Vacuum þ radiation shields < 0.0010
50 Cryogenic Technology

The choice of insulation concept has to be 6. Low-Temperature Separation

made by (1) weight and cold-temperature Systems
mechanical properties of the insulation
material, (2) the relative deviation of The principles of separating gas and liquid
thermal expansion properties of the insulation mixtures at cryogenic temperatures are the
material and the cryogenic system, (3) the same as those used in the design of separation
sensitivity and safety requirements of systems for near-ambient temperatures and
the process, and (4) the economics of the above; however, some special problems arise
system. in low-temperature separation systems. For
Depending on the size of the system, plant example, a cooling source for condensers in
components and pipes are either insulated a cryogenic system is not readily available;
individually using mechanical (and vacuum) whereas in petroleum separation systems, cool-
insulation or grouped together in a common ing water or ambient air may be used.
large insulating vessel, called cold box, often For several mixtures, such as air, natural gas,
filled with perlite. and coke-oven gas, cryogenic separation tech-
In case of a porous insulation material, the niques are the most economical methods of
temperature gradient across the insulation will separating the mixture components.
cause a flux of water molecules from the Cryogenic separation methods include rec-
ambient towards the cold surface by either tification, partial condensation or evaporation,
diffusion or convection. In order to avoid and adsorption.
the freezing out of water inside the insulation
material either vapor seals are installed or free
intermediate spaces are purged using dry air or 6.1. Minimum Work of Separation
nitrogen.
The cold parts of smaller cryocoolers The minimum work of separation occurs when
(e.g., helium liquefiers) are usually placed the separation process is carried out reversibly
completely into a closed vessel that is addi- and isothermally at temperature To of the sur-
tionally evacuated to reach the best insulat- roundings, where the subscript o ¼ outside. The
ing effect. Latter arrangement is referred to separation work per mole of mixture is given by
as cryostat. Equation (19):
The thermal radiation ingress from the X

W id =ðqv Þm ¼ yi ½To ðs1
s2 Þ
ðh1
h2 Þi ð19Þ
warm towards the cold walls of the cryostat i
vessel is blocked by wrapping multiple layers
of reflective foils, separated by low thermal where yi ¼ mole fraction of component i; the
conductivity spacers, around the equipment units for entropy (s) and enthalpy (h) are J
inside the vacuum chamber. Typically 10, 15 mol
1 K
1 and J/mol, respectively. Subscript
and 30 layers of radiation shielding are 1 denotes conditions before separation, and
applied for LNG, liquid nitrogen, and liquid subscript 2 denotes conditions after separation.
helium applications, respectively. The vac- Equation (19) is equivalent to the exergy dif-
uum should at least possess a level where ference between the mixture and its individual
the mean free path of the remaining mole- components. For conditions under which the
cules is significantly larger than the cryostat materials may be treated as ideal gases, the
vessel’s dimensions along the temperature entropy and enthalpy changes may be written as
gradient. follows:
The heat inleak into mechanically insulated
equipment is typically at least 10 times as large s1
s2 ¼ cp lnðT1 =T2 Þ
R lnðp1 =p2 Þ
as for vacuum insulation. Additionally, the
cryostat vessel may be placed into a bath of
circulating liquid nitrogen or helium to further s1
s2 ¼ R lnðp=pi Þ
decrease the thermal radiation ingress. For
detailed reading on cryogenic insulation refer
to [22, 82, 83]. h1
h2 ¼ cp ðT1
T2 Þ ¼ 0
 Cryogenic Technology 51

For isothermal separation, T1 ¼ T2 ¼ To. The mixture is given by the Gibbs phase rule:
initial pressure p1 is equal to the partial pressure
pi of the component in the mixture, and the final F ¼C
Pþ2
pressure p2 is equal to the mixture pressure p.
The quantity R is the gas constant. Making where C ¼ number of components and P ¼
these substitutions into Equation (19), the fol- number of phases present. For a ternary (three-
lowing expression is obtained for the minimum component) mixture in the vapor phase, four
work of separation with yi ¼ pi/p: independent properties are required to deter-
X
mine the state of the mixture. For example, if

W id =ðqv Þm ¼ RTo yi lnð1=yi Þ the pressure, temperature, and mole fractions of
i
two of the components are specified, all other
thermodynamic state properties are fixed and
Separation of Air. The ideal work of sepa- can then be determined.
ration for air (a mixture of 78.1 mol % nitrogen, Properties of mixtures may be presented in
21.0 mol % oxygen, and 0.9 mol % argon) at a various forms. One of the common representa-
temperature of 300 K is tions for binary (two-component) mixtures is the
temperature–composition diagram, as shown for

W id =ðqv Þm ¼ ð8:3143Þð300Þ
½0:781 lnð1=0:781Þ þ 0:210 lnð1=0:210Þ þ 0:009 lnð1=0:009Þ
the nitrogen–oxygen system (Fig. 46).
¼ 1405 J=mol In this figure, mole fraction is plotted against
temperature for a fixed value of pressure. The
The ideal work per mole of oxygen is upper line in each pair of curves is called the
dew point curve. It represents the temperatures,

W id =qv ðO2 Þ ¼ 1405=0:210 ¼ 6690 J=mol for a given composition, at which the vapor
phase begins to condense (formation of first
Similarly, the ideal work per unit mass of
oxygen is

W id =qm ðO2 Þ ¼ 6690=32 ¼ 209:1 J=g

If only the oxygen is separated completely from

the mixture (and the nitrogen and argon remain
unseparated), the ideal work per mole mixture is:

W id =qv ¼ ð8:3143Þð300Þ  ½0:790 lnð1=0:790Þ þ 0:210 lnð1=0:210Þ

¼ 1282 J=mol

The ideal work requirements per mole and per

mass unit of oxygen separated from air are:

W id =qv ðO2 Þ ¼ 6105 J=mol

W id =qm ðO2 Þ ¼ 190:8 J=g

Thus, the ideal work of separation is smaller if

only one of the components is isolated from the
mixture.

6.2. Properties of Mixtures

The number of independent thermostatic prop- Figure 46. Temperature–composition diagram for
erties (F) required to define the state of a nitrogen–oxygen mixtures
52 Cryogenic Technology

liquid droplet). The dew point curve corre- For real gases and solutions, the distribution
sponds to the condition of the saturated vapor coefficients must be determined either exper-
in the two-phase region. imentally or from thermodynamic relationships
The lower line in each pair of curves is called along with the equations of state for the mate-
the bubble point curve. It represents the tem- rials. A plot of the distribution coefficients for
peratures, for a given mixture composition, at nitrogen (K1 ¼ x1/y1) and for oxygen (K2) in
which the liquid phase begins to boil (formation liquid air is shown in Figure 48.
of the first vapor bubble). The bubble point The behavior of mixtures can also be repre-
curve corresponds to the condition of the satu- sented in terms of the relative volatility, aij,
rated liquid in the two-phase region. defined by:
The mixture properties may also be plotted
as shown in Figure 47. This equilibrium aij ¼ K i =K j
diagram is important in the design of rectifica-
tion columns to determine the number of As shown in Figure 49, the relative volatility for
“theoretical plates” required. cryogenic mixtures of nitrogen–oxygen, is a
The relationship between the mole fractions fairly weak function of the mixture temperature
y and x in the vapor and liquid phases, respec- for a constant pressure.
tively, may be expressed in terms of the distri- The composition of the vapor and liquid
bution coefficient K (sometimes called the phases of binary mixtures may be determined
equilibrium ratio), defined for component j from values of the distribution coefficients or
by [113]: the relative volatility. From the definition of the
distribution coefficient,
K j ¼ yj =xj
y1 ¼ K 1 x1
For an ideal-gas vapor phase and a perfect- y2 ¼ K 2 x2 ¼ K 2 ð1
x1 Þ

solution liquid phase, the distribution

coefficient is given by: For a binary mixture,

K j ¼ pj;sat =p y1 þ y2 ¼ 1 ¼ K 1 x1 þ K 2 ð1
x1 Þ

where pj,sat ¼ vapor pressure of component j at where subscript 1 refers to the more volatile
the mixture temperature and p ¼ mixture (lower boiling point) component and subscript
pressure. 2 refers to the less volatile (higher boiling

Figure 47. Equilibrium diagram for nitrogen–oxygen Figure 48. Distribution coefficients for nitrogen (— – —)
mixtures and oxygen (—)
 Cryogenic Technology 53

terms of the relative volatility a12, as follows:

1
K2
x1 ¼
K 2 ða12
1Þ
a12 ð1
K 2 Þ
y1 ¼
a12
1

Another set of thermodynamic curves that

are import in the design of separation systems is
the enthalpy–composition diagram. Values for
the enthalpy at various mixture pressures for the
nitrogen–oxygen system are given in Table 5.

6.3. Flash Calculation

In the rectification columns used in cryogenic

Figure 49. Relative volatility for nitrogen–oxygen separation systems, the feed stream is often
mixtures
introduced into the column by expansion
through a J–T valve. Determining the condition
of the feed stream after expansion through the
valve is important. This solution for a two-
point) component in the mixture. Solving for phase system is called the flash calculation.
the mole fraction of component 1 in the liquid If the feed stream enters the J–T valve at a
phase, Equation (20) is obtained: molar flow rate qv with known properties
(enthalpy hin and mole fraction xin, j), a mass
1
K2
x1 ¼ ð20Þ balance across the J–T valve yields
K1
K2
 
The mole fraction of component 1 in the vapor xin;j qv ¼ yj qv ðgÞ þ xj qv ðlÞ ¼ yj qv
qv ðlÞ þ xj qv ðlÞ
phase may be determined from:
where qv(g) ¼ molar flow rate of vapor from the
y1 ¼ K 1 x1 ¼
K 1 ð1
K 2 Þ
ð21Þ J–T valve and qv(l) ¼ molar flow rate of liquid
K1
K2
from the J–T valve. Introducing the distribution
coefficient, xj ¼ yj/Kj, the mole fraction of the
Equations (20) and (21) can also be written in lower boiling point component in the vapor

Table 5. Specific enthalpies of nitrogen–oxygen mixtures

Mole fraction nitrogen Specific enthalpies, J/mol

101.3 kPa 202.7 kPa 506.6 kPa

hb hd hb hd hb hd

0.0 419 7252 764 7362 1315 7536

0.1 461 7231 807 7329 1403 7507
0.2 519 7210 860 7295 1499 7478
0.3 599 7185 923 7259 1591 7448
0.4 682 7151 996 7222 1675 7415
0.5 779 7118 1079 7184 1758 7377
0.6 879 7084 1172 7145 1851 7339
0.7 984 7042 1275 7105 1947 7298
0.8 1084 6992 1388 7063 2052 7252
0.9 1181 6933 1511 7020 2152 7201
1.0 1273 6871 1644 6976 2248 7147

hb ¼ specific enthalpy at the bubble point; hd ¼ specific enthalpy at the dew point for the mixture.
54 Cryogenic Technology

phase leaving the J–T valve may be determined:

xin;j
yj ¼  
1 þ qqv ðlÞ 1
Kj
1
v

An energy balance around the J–T valve yields

the auxiliary relationship for the fraction of the
outlet stream, which is in the liquid phase:

qv ðlÞ Lg
Lin
¼ ð22Þ
qv Lg
Lf

The final governing equation for the flash

calculation results from the fact that the sum
of the mole fractions for the vapor phase is
equal to unity:
X X xin;j
yj ¼ 1 ¼   ð23Þ
j j 1 þ qqv ðlÞ 1
Kj
1
v

Figure 50. Schematic of a rectification column

The solution for the condition of the liquid a) Dome; b) Condenser; c) Liquid-on-plate; d) Downcomer
and vapor phases after expansion through the (seal); e) Boiler
J–T valve involves an iterative technique.
First, a trial temperature for the mixture, of the lower boiling point component xin. The
which is between the dew point and bubble top product or distillate product (the practically
point temperatures at the exit pressure of the pure lower boiling point component, e.g., nitro-
J–T valve, is chosen. The liquid fraction can gen in the air separation column) leaves the
be calculated from Equation (22) using this column at a molar flow rate (qv)D, enthalpy hD,
temperature, and finally the right side of and composition xD. Similarly, the bottom prod-
Equation (23) can then be determined. If the uct (the higher boiling point component, e.g.,
right side is not equal to unity, as it should be, oxygen in the air separation column) leaves the
a new trial temperature is chosen; the process column at a molar flow rate (qv)B, enthalpy hB,
is repeated until Equation (23) is satisfied. and composition xB. Heat QB is added to the
When the correct temperature is found for column in the reboiler (e), and heat QD is
the exit stream, then the composition of the withdrawn from the column in the condenser
stream may be determined from the distribu- (b). The liquid that forms in the condenser and
tion coefficients and from Equations (20) and flows back into the column is called reflux.
(21). Computer-aided flash calculations have The vapor in the feed stream flows up the
been described for multicomponent mixtures column through the layers of liquid (c) on plates
[114]. or trays. As the vapor flows through the liquid,
the vapor bubbles (which are at a higher tem-
perature than the liquid) evaporate some of the
6.4. Principles of Cryogenic lower boiling component from the liquid into
Rectification the vapor phase. The heat transfer from the
bubble results in some condensation of the
The rectification process involves cascading higher boiling component from the vapor bub-
several continuously operated partial evapora- ble into the liquid phase. As the vapor phase
tions and condensations (distillations) that are rises in the column towards the condenser it is
carried out in counterflow. A schematic of a thus enriched in the lower boiling component.
rectification column is shown in Figure 50. Similarly, the liquid phase flowing down the
The feed stream enters the column at a molar column is enriched in the higher boiling
flow rate (qv)f, enthalpy hin, and a mole fraction component.
 Cryogenic Technology 55

The composition of the top and the bot- Equations (24) and (26):
tom product streams depends on the number
of stages or plates in the upper and lower ðqv Þnþ1 ðlÞ hn ðgÞ
hnþ1 ðlÞ ðqv ÞD
¼1
¼1 ð27Þ
portions of the column, and the amount of ðqv Þn ðgÞ QD =ðqv ÞD þ hD
hnþ1 ðlÞ ðqv Þn ðgÞ
heat supplied and withdrawn in the reboiler
and the condenser, respectively. The number The equation for the “operating line” is
of theoretical plates for the column can be found by combining Equations (24) and (25):
determined by a direct numerical approach    
ðqv Þnþ1 ðlÞ ðqv ÞD
or through one of three basic shortcut yn ¼ xnþ1 þ xD ð28Þ
ðqv Þn ðgÞ ðqv Þn ðgÞ
methods, the Ponchon–Savarit method
[115], the McCabe–Thiele method [116],
The solution for the number of theoretical
or the Fenske–Underwood–Gilliland (FUG)
plates or “equilibrium stages” required for
method [117].
the upper portion of the column is found by
The Ponchon–Savarit method is an exact
successive application of Equations (27) and
method for binary systems, which can be
(28) and the equilibrium relationship yn ¼
applied for all materials; however, detailed
Kjxnþ1, between the feed condition and the
enthalpy data are required. Due to its limita-
desired condition at the top of the column.
tion to simple binary systems the McCabe–
Similar expressions can be written for the
Thiele method is in general somewhat less
lower portion of the column:
accurate, but it requires only equilibrium con-
centration data. In case of a multicomponent
ðqv ÞB hm ðgÞ
hmþ1 ðlÞ ðq Þ ðlÞ
mixture the semiempirical FUG needs to be ¼ ¼ v mþ1
1
ðqv Þm ðgÞ QB /ðqv ÞB
hB þ hmþ1 ðlÞ ðqv Þm ðgÞ
applied.
The three governing relationships for the The operating line for the lower section of the
analysis of rectification columns are the column is:
conservation of mass, the conservation of
components, and the conservation of energy.    
ðqv Þmþ1 ðlÞ ðqv ÞB
Applying these principles for the upper ym ¼ xmþ1
xB
ðqv Þm ðgÞ ðqv Þm ðgÞ
portion of the column above the feed inlet
gives: The plates in the lower section are numbered
from the boiler section (m ¼ l) upward to the
ðqv Þn ðgÞ ¼ ðqv Þnþ1 ðlÞ þ ðqv ÞD ð24Þ feed plate.
For the solution of the number of theoretical
plates the overall heat and mass balances need
ðyn Þ  ðqv Þn ðgÞ ¼ xnþ1  ðqv Þnþ1 ðlÞ þ xD  ðqv ÞD ð25Þ to be solved together with above equations and
the vapor liquid equilibrium relationship. The
solution can either be determined graphically
hn ðgÞ  ðqv Þn ðgÞ ¼ hnþ1 ðlÞ  ðqv Þnþ1 ðlÞ þ hD  ðqv ÞD þ QD ð26Þ by applying the Ponchon–Savarit or McCabe–
Thiele method, or by using a shortcut method as
where (qv)n(g) ¼ molar flow rate for the vapor FUG, or numerically solving the column’s
leaving plate n, (qv)nþl(l) ¼ molar flow rate for complete equation system, using a computer.
the liquid leaving plate (nþl) and entering plate For nitrogen–oxygen mixtures, and other sim-
n, yn ¼ mole fraction of the lower boiling point ple cryogenic mixtures, the molar heat of
component leaving the nth plate in the vapor vaporization is almost independent of the com-
phase, xnþl ¼ mole fraction of the lower boiling position, thus the McCabe–Thiele method may
point component leaving the (nþl)th plate in be applied.
the liquid phase, hn(g) ¼ enthalpy of the vapor
leaving plate n, and hnþ1(l) ¼ enthalpy of the
liquid leaving plate nþl. The plates are num- 6.5. Air Separation
bered from the feed plate (n ¼ l) upward.
The expression for the internal reflux ratio of Atmospheric air is not simply a binary mixture of
the column may be determined by combining nitrogen and oxygen. Rather, the composition of
56 Cryogenic Technology

dry, carbon dioxide free atmospheric air is as nitrogen waste gas having a residual oxygen-
follows (mole fraction): content of 7–10 mol%. The purity of the nitro-
gen is limited by the distribution coefficient of
Nitrogen 0.78084
oxygen between the liquid and vapor phase at
Oxygen 0.20942 the top of the column. The oxygen yield of the
Argon 0.00934 process is below 65%.
Other gases 0.00004 The air feed stream is compressed (a), water
Neon 18.18 ppm
Helium 5.24 ppm
vapor and carbon dioxide are removed (b), and
Methane 1.76 ppm the air is passed through a heat exchanger (c).
Krypton 1.14 ppm The cool air is further cooled in the reboiler (d)
Hydrogen 0.50 ppm of the column (e) and then expanded through a
Xenon 0.09 ppm
J–T valve (f). The heat exchange in the reboiler
produces the vapor required in the column
Depending on the components to be pro- without using an external heat source.
duced, together with the economics of the The liquid formed during expansion through
respective production capacity, a large variety the J–T valve is introduced at the top of the
of processes exist for separating air. The most column to provide the required liquid flow
important shall be introduced in the following down the column (reflux). The vapor at the
sections. For a more detailed discussion on top of the column is passed back through the
cryogenic air separation refer to [109, 83]. heat exchanger to provide cooling for the
incoming feed stream. By use of the required
6.5.1. Linde Single-Column System number of plates in the column, nearly pure
liquid or gaseous oxygen may be produced.
One of the simplest systems for separation of air If the product is required as a gas, the oxygen
is the Linde single-column system, shown in is usually passed back through the heat
Figure 51. It was first built by LINDE in 1902. exchanger to provide additional cooling for
The system is used to produce either gaseous or the incoming feed stream. Furthermore, the
liquid oxygen together with an impure gaseous feed stream must be compressed to ca. 3–6

Figure 51. Linde single-column air-separation system

a) Compressor; b) Carbon dioxide and water removal; c) Main heat exchanger; d) Reboiler; e) Separation column; f) J–T valve
 Cryogenic Technology 57

MPa, because of the additional cooling pro- A portion of the liquid at the bottom of the
vided by the return cold oxygen stream through lower column is withdrawn, expanded through
the heat exchanger. On the other hand, if the a J–T valve (d2), and introduced into the middle
product is desired as a liquid, then the feed of the low pressure column (e2), which operates
stream must be compressed to ca. 20 MPa for at a pressure of ca. 0.12–0.14 MPa. The com-
effective operation. The pressure within the position of the liquid in the boiler is ca. 35–40%
column is usually 0.13–0.20 MPa. A variant oxygen. At 0.55 MPa, the dew point of pure
of the single column system allows a low yield liquid nitrogen is 95.2 K, whereas the boiling
(< 50%) production of pure nitrogen [107]. point of pure liquid oxygen is 92.6 K at 0.13
MPa. Therefore, the liquid oxygen evaporating
6.5.2. Linde Double-Column System in the upper column can be used to condense (f)
the liquid nitrogen in the lower column without
If both purified nitrogen and purified oxygen requiring external refrigeration.
products are required at reasonable yields, the A portion of the liquid nitrogen from the
double-column system invented by LINDE in lower column is subcooled, expanded through a
1910 must be used (Fig. 52). The system allows J–T valve (d3), and introduced at the top of the
the production of gaseous and liquid products. upper column where pure gaseous nitrogen is
The feed stream is compressed (a) and water withdrawn. Liquid nitrogen can be withdrawn
and carbon dioxide are removed (b). The feed from the reflux of the pressurized column.
stream passes through a heat exchanger (c), is Liquid oxygen is obtained from the sump of
expanded in a J–T valve (d1), and fed into the the low-pressure column. If gaseous oxygen is
lower part of the pressurized column (e1), which to be produced, the liquid oxygen will be
operates at a pressure of ca. 0.5–0.6 MPa [118]. evaporated in the main heat exchanger.

Figure 52. Linde double-column air-separation system

a) Compressor; b) Carbon dioxide and water removal; c) Main heat exchanger; d) J–T valves; e) Separation columns; f)
Reboiler–condenser
58 Cryogenic Technology

6.5.3. Heylandt System exchanger is cooled and expanded through a

J–T valve (f1) to a pressure of 0.7 MPa. The air
One of the commonly used air separation is then introduced into the bottom of the lower
systems is the Heylandt system (Fig. 53). It column (g1).
uses an expansion engine to provide a signifi- Enriched liquid oxygen (oxygen content
cant portion of the refrigeration needed to of 36–38 mol%) is withdrawn from the sump
liquefy the air. of the lower column, expanded through a J–T
The air feed stream is compressed (a1) to a valve (f2) to 0.2 MPa, and introduced into the
pressure of ca. 10 MPa and water vapor and middle of the upper column. To provide
carbon dioxide are removed (b). The air stream reflux liquid for the upper column, liquid
is then compressed further to ca. 14 MPa in a nitrogen is withdrawn from the lower col-
final stage of compression (a2), which is driven umn below the condenser–reboiler (h), sub-
by the expander. In some cases the air is now cooled, and expanded through a J–T valve
precooled using external refrigeration (e). (f3) before being introduced at the top of the
About half of the feed stream is passed through upper column. Nitrogen gas from the upper
the main heat exchanger (d) and the remainder column is returned through the main heat
of the feed stream is expanded through an exchanger to provide the necessary cooling
expansion turbine (c) to a pressure of 0.2 in these units.
MPa. The stream leaves the expander at ca. The product liquids are removed from the
110 K and is introduced into the middle portion condenser–reboiler arrangement (oxygen from
of the upper column (g2). The portion of the the upper column and nitrogen from the lower
feed stream flowing through the main heat column).

Figure 53. Heylandt air-separation system

a) Compressor stages; b) Carbon dioxide and water removal; c) Expander; d) Main heat exchanger; e) Optional external
refrigerant stream; f) J–T valves; g) Separation columns; h) Condenser–reboiler
 Cryogenic Technology 59

6.5.4. Argon Separation equilibrium composition of the ternary mixture

is significantly different from that of the binary
The predominant component of air, other than nitrogen–oxygen mixture. This fact affects the
oxygen and nitrogen, is argon, which is present calculation of the number of theoretical plates
at a concentration of nearly 1 mol %. This argon required for the column.
concentration in the air feed stream means that Argon may be separated in the double-
to obtain pure oxygen and pure nitrogen an column system by using an additional crude
additional column must be used to separate the argon column, as shown in Figure 54. Argon
argon from the other components. tends to collect in the middle part of the upper
Argon boils at 87.3 K and 101.3 kPa; thus, its column (b2) where the fluid has a high oxygen
boiling point at atmospheric pressure lies concentration and a relatively low nitrogen
between that of nitrogen (77.4 K) and oxygen concentration. The mixture with ca. 10 mol
(90.2 K). If all of the argon were collected with % of argon is removed from the main air-
the nitrogen product, the composition would be separation column at that point and is intro-
98.8 mol % nitrogen and 1.2 mol % argon. On duced into the lower portion of the argon side
the other hand, if all of the argon were collected column (c). The vapor flows up the column and
with the oxygen stream, the composition would becomes enriched in argon, which has a lower
be 95.7 mol % oxygen and 4.3 mol % argon. boiling point than oxygen. At the top of the
Because argon is soluble in both liquid nitrogen crude argon column, refrigeration is supplied
and liquid oxygen, the argon is generally pres- by bypassing some of the liquid from the sump
ent in both streams unless it is removed during of the pressurized column (b1) through the
the separation process. In addition, the condenser of the crude argon column (d).

Figure 54. Argon recovery subsystem of the double-column air separation system
a) J–T valves; b) Separation columns; c) Crude argon column; d) Condenser heat exchanger; e) Subcooler
60 Cryogenic Technology

The vapor from the condenser is returned to the passed through a set of adsorbent traps filled
upper portion of the main column. with synthetic zeolite or other absorbent, where
The liquid that collects in the bottom of the the oxygen is removed. If the oxygen is desired
crude argon column is primarily liquid oxygen, as a liquid, then the gaseous oxygen is con-
which is taken back to the main column. The densed by using some of the liquid nitrogen
product that collects in the top of the crude product from the main column as the refrigerant
column is primarily argon and nitrogen, with a in an argon condenser.
small amount of oxygen.
In principle, it is possible to produce an 6.5.5. Helium and Neon Separation
argon intermediate product with an oxygen
content of 1 ppm by complete rectification The heat-transfer rate in the condenser–reboiler
inside the crude argon column. Due to the in an air separation system begins to decrease
very close boiling points of argon and oxygen after the double-column system has been in
this separation process typically requires continuous operation for 2–3 d. The reason
around 200 theoretical plates. To ensure that for this drop in performance is the gradual
condensation is still possible by cooling with collection of a noncondensable component
evaporating crude oxygen, the pressure loss in (mainly neon and helium) on the nitrogen
the crude argon column must sufficiently small. side of the heat exchanger. This accumulation
This can only be achieved by using random or lowers the partial pressure of the nitrogen in the
structured packing instead of sieve trays [119]. mixture to the point where it no longer con-
If the argon is to be produced with a simpler denses at the available temperature. As a result
column design, the crude material from the the active heat-transfer surface is covered with
argon column may be purified by two tech- a noble gas layer that leads to a significant
niques: (1) catalytic combustion to remove decrease in heat transfer efficiency between the
oxygen, followed by rectification to separate two columns. Additionally a higher pressure is
the argon and nitrogen, and (2) adsorption to required to continue condensation in the dome,
remove oxygen, followed by rectification for so the power input of the main air compressor
nitrogen–argon separation. will also increase.
In the catalytic combustion system, the crude If the collection of helium and neon is not
argon is mixed with an appropriate amount of required, this problem is solved by periodically
hydrogen gas and passed through a catalytic venting a small portion of the gas from the
combustion unit to combine the oxygen dome of the condenser–reboiler. The vent gas
impurity with hydrogen to form water vapor. contains ca. 2.1 mol % neon and 0.8 mol %
The bulk of the water vapor is subsequently helium, with traces of hydrogen.
removed by passing the gas through a con- The commercial production and distribu-
denser; the last traces of water are removed tion of neon on a large scale began in the early
by subsequently passing the gas through a set of 1960s in response to demand, e.g., as a refrig-
alumina driers. The excess hydrogen gas is erant and for applications in decorative neon
separated from the argon stream by partial lighting [120]. Helium and neon can be col-
condensation in the boiler of the rectification lected by using the air-separation column
column for argon–nitrogen separation. The shown in Figure 55.
final gas mixture is rectified to produce essen- The helium–neon portion of the dome vent
tially pure argon liquid or gas. gas may be separated from the nitrogen by
In the adsorption system, the crude argon is passing the gas mixture through a condenser–
first passed through a rectification column to rectifier (c) that is cooled by liquid nitrogen
separate the nitrogen from the mixture. The expanded through a J–T valve (a4) from the
column’s reboiler is heated by condensing main column. The boiling points of neon (27.1
pressurized gaseous nitrogen which is K) and helium (4.2 K) are so much lower than
expanded afterwards to provide cooling for that of nitrogen (77.4 K) that the liquid formed
the condenser. The oxygen–argon liquid is in the condenser–rectifier is practically pure
removed from the boiler of the column and nitrogen. The typical composition of the vapor
evaporated in a heat exchanger. The gas is from the condenser–rectifier unit is 46 mol %
 Cryogenic Technology 61

Figure 55. Helium and neon recovery subsystem of the double-column air separation system
a) J–T valves; b) Separation columns; c) Helium/neon condenser–rectifier; d) Subcooler

neon, 19 mol % helium, 33 mol % nitrogen, and oxygen product from the low-pressure column
2 mol % hydrogen. (b2) is circulated through a cryogenic adsorber
The hydrogen can be removed from the (d) and a concentration column (e) where kryp-
mixture by catalytic combustion followed by ton and xenon are enriched. The mixture is
a drier to remove the water vapor formed during usually reboiled (f) by means of condensing
combustion. Excess oxygen and nitrogen are a side draw from the pressurized column. The
then removed in an adsorption trap. The helium krypton–xenon concentration column operates
and neon are separated subsequently in a con- at the same pressure as the upper column of the
denser cooled by a closed-cycle helium refrig- air separation unit, usually 0.13–0.16 MPa. The
eration system. waste oxygen from the top of the column is fed
back into the air separation column on a higher
6.5.6. Krypton and Xenon Separation plate. Since methane has about the same rela-
tive volatility as krypton and cannot be captured
Due to their low content in the ambient air and in the cryogenic adsorber, it will also be
their low volatility, krypton and xenon will enriched in approx. the same order of magni-
accumulate in the oxygen product on the level tude as krypton. From economic considerations
of a trace impurity below 6 ppm. The basic flow it would be favorable to concentrate krypton
scheme for the first enrichment step of krypton and xenon as much as possible at this point, but
and xenon is shown in Figure 56. the degree of enrichment is practically limited
In order to capture hazardous high boiling by safety concerns: In order to avoid explosions
gases, as acetylene and nitrous oxide, the liquid in the air separation unit, the hydrocarbon
62 Cryogenic Technology

Figure 56. Krypton and xenon recovery subsystem of the double-column air separation system
a) J–T valves; b) Separation columns; c) Subcooler; d) Cryogenic adsorbers; e) Krypton–xenon concentration column;
f) Reboiler

content is usually restricted to the respective oxygen along with CF4, SF6, and C2F6 a next set
smaller concentration of either solubility limit cryogenic rectification columns occasionally
or the lower flammability limit plus an appro- coupled with chemical purification are utilized.
priate safety margin [82, 121]. The crude kryp-
ton–xenon mixture thus typically contains a 6.6. Conditioning of Natural Gas
total of only 3000 ppm of rare gases together
with at least the same (or higher) amount of Natural gas is generally a mixture of wide
hydrocarbons, mostly methane. The remainder boiling substances which has to be processed
is oxygen. For further processing the crude in order to meet customer specifications as
krypton–xenon mixture is pressurized, evapo- heating value and Wobbe index. Typical com-
rated, and heated up such that hydrocarbons can positional ranges for natural gas are given
be converted to water and carbon dioxide by below (mole fraction).
catalytic combustion. The latter are subse-
quently removed by adsorption. The hydro- Helium 0.00–0.05
carbon-free crude krypton–xenon mixture is Nitrogen 0.00–0.40
fed into a next cryogenic separation stage where Methane 0.30–0.99
Ethane 0.01–0.15
the remaining oxygen is removed by rectifica-
Propane 0.01–0.10
tion. At this point a mixture of ca. 93 mol% Butanes 0.00–0.05
krypton and 7 mol% xenon together typically Pentanes 0.00–0.03
100 ppm of oxygen and other trace impurities Hexanes 0.00–0.01
Heptane and higher hydrocarbons 0.00–0.01
has been produced. For the separation of kryp-
Carbon dioxide 0.00–0.45
ton and xenon and the removal of the remaining Hydrogen sulfide 0.00–0.15
 Cryogenic Technology 63

Sticking to the definition of the term methane-rich bottom product and methane-lean
“cryogenic” at the beginning of this chapter, top product. The reflux is generated by con-
only the separation of nitrogen (and helium) densing the top product against evaporating
falls into the scope of this section. methane in the low-pressure column. After
passing through a next heat exchanger (e)
6.6.1. Removal of Nitrogen and Helium both intermediate products are fed into the
low-pressure column (f) operating as low as
In analogy to air separation the classic process 0.17 MPa. The reflux composition for the low-
for removing of nitrogen from natural gas also pressure column is adjusted with the tray num-
involves a thermally coupled double-rectifica- ber and operating pressure in the pressurized
tion column as shown in Figure 57. column. As the amount of nitrogen is fixed by
Pretreated natural gas is cooled down in a the natural gas inlet, the reflux flow rate cannot
first heat exchanger (a) and partially condensed. be influenced in this basic process concept. The
The developing liquid contains only little nitro- purified methane product is withdrawn as liquid
gen. It is separated (c) and fed to the product from the bottom of the low-pressure column
stream. The vapor from the phase separation and pressurized (h) as much as the heat inte-
drum is fed into the pressurized column (d) at a gration in the heat exchanger allows. Typically,
pressure as high as possible, but below 3.4 MPa, a residual nitrogen content of below 3 mol% is
the critical pressure of nitrogen. The task of the aimed for. The purity of the rejected nitrogen
pressurized column is to separate the feed into a stream is typically 99 mol% or higher.

Figure 57. Nitrogen rejection from natural gas using a double-column process
a) Main heat exchanger; b) J–T valves; c) Separation drum; d) Pressurized column; e) Heat exchanger; f) Low-pressure
column; g) Reboiler–condenser heat exchanger; h) Methane pump
NG ¼ Natural gas
64 Cryogenic Technology

Due to the highly integrated nature of the Table 6. Typical compositions of hydrogen-rich gases of different
double-column separation process, only gases origins
with a nitrogen concentration above 25 to 30 Substance Composition, mol%
mol% can by processed in this arrangement.
For lower nitrogen contents, a preseparation of Steam reformer Partial oxidation Coke oven gas

hydrocarbons, e.g., by means of an additional H2 69–80 37–64 50–64

column located after the phase separation drum, N2 < 0.5 <1 3–12
CO 16–23 35–62 4–8
is required.
CH4 3–8 < 0.5 24–27
All of the helium contained in the natural gas
feed will either leave the plant together with the Difference to 100 mol%: H2O, CO2, C2þ, Ar, sulfur compounds.
nitrogen-rich product, or the helium can be
separated by partial condensation of the nitrogen
from the highest point of the pressurized col-
umn. Depending on the composition of the For both cases it is mandatory that the feed
natural gas feed, a helium concentration as gas to the cryogenic process is free from water
high as 10 mol % can be withdrawn from this (from amine scrubbing), methanol (from Rec-
point. Remaining nitrogen is condensed using tisol scrubbing), carbon dioxide, and other
cold nitrogen from behind J–T valve (b3) before components which could freeze at the low
it enters the low-pressure column. The conden- operation temperatures. Therefore process gas
sate is fed back to the top of the high-pressure is initially dried in a molecular sieve adsorber
column. Next, the crude helium is pressurized to station.
20 MPa. Additional nitrogen is removed by
subsequent partial condensation steps using liq- 6.7.1. Partial Condensation Process
uid nitrogen at atmospheric pressure and under
vacuum. The liquid is separated and fed back to If the feed is supplied at high pressures, high
the high-pressure column. The gaseous helium carbon monoxide and low methane content
now has a purity of more than 99.99 mol%. The the partial condensation process is preferred
remaining nitrogen is removed by cryogenic (Fig. 58). The pretreated feed gas (free of
adsorption down to ca. 5 ppm. carbon dioxide and water) enters the separa-
tion process at ambient temperatures and 2.5–
8 MPa. The feed gas is cooled down and is
6.7. Separation of Hydrogen-rich Gases partially condensed in the main heat
exchangers (a) using cold separation prod-
Cryogenic separation processes are used for the ucts. The gaseous phase, consisting of ca. 90
recovery of hydrogen and carbon monoxide, mol% of hydrogen, is separated from the
and for conditioning of feed gases from pro- carbon monoxide-rich liquid in a subsequent
cesses as coke ovens, partial oxidation or cata- separator (b) and is warmed before leaving
lytic reforming (e.g., steam reforming). The the system. In order to further remove the
typical compositions of such feed gases are dissolved hydrogen from the carbon mon-
shown in Table 6. oxide, the liquid from the separator is fed
The desired purity of carbon monoxide to a stripping column (d). Heating of the
depends on its downstream application and column is usually integrated into the main
can be adjusted into the ppm range with respect heat exchanger. The waste gas leaving the top
to the residual contents of hydrogen and meth- of the column is a mixture of hydrogen and
ane. In order to separate carbon monoxide from carbon monoxide. It is passed though the
synthesis gas there are basically two main main heat exchangers, and can be com-
cryogenic process types: (1) partial condensa- pressed (e1) and recycled. The sump product
tion and (2) liquid methane scrubbing. Both of the column has a defined residual amount
processes exist in several alternative configura- of hydrogen according to customer specifi-
tions depending on required product purity, cation. Typically a carbon monoxide content
recovery rate, presence of other impurities of ca. 98.5 mol% is reached. The evaporation
like methane, nitrogen, and argon in feed. of the carbon monoxide product on different
 Cryogenic Technology 65

Figure 58. Hydrogen separation by partial condensation

a) Main heat exchangers; b) Separation drum; c) J–T valves; d) Purification column; e) Compressors
LIN ¼ Liquid nitrogen; GAN ¼ Gaseous nitrogen

pressure levels provides the majority of and is fed to the methane scrubbing column
refrigeration duty for the main heat exchang- where hydrogen is purified down to a few ppm
ers. Additional refrigeration duty can be sup- of carbon monoxide using subcooled methane.
plied by evaporating and warming of liquid The bottom product is throttled (c1) and the
nitrogen in the process. The evaporated and forming vapor is separated (d) and can be
warmed up carbon monoxide streams are recycled via the compressor (e) to the feed
routed to a product compressor (e2). gas. Thereby refrigeration effect is produced
In case the feed contains noteworthy in an open-loop J–T cycle. If a sufficient enrich-
amounts of, e.g., methane or nitrogen, addi- ment of carbon monoxide cannot be reached by
tional stripping columns (and in some cases a flashing alone, the separator vessel is replaced
carbon monoxide expander) are required to by an additional hydrogen stripping column.
obtain a high-purity carbon monoxide product. The liquid is throttled (c2), partially heated in
Hydrogen purities of ca. 98 mol% can be the secondary heat exchanger (f), and enters the
reached if a two-staged partial condensation methane separation column (g), where methane
process is applied. and carbon monoxide are formed as products.
Part of the liquid methane from the bottom is
6.7.2. Liquid Methane Wash fed to the methane scrubbing column as scrub-
bing agent using the liquid methane pump (h).
In case of low-feed pressures, lower carbon mon- Excess liquid methane is throttled (c5) and
oxide and elevated methane contents, preferably evaporated in order to provide cooling the
liquid methane scrubbing is applied (Fig. 59). main heat exchanger. The carbon monoxide
The pretreated feed gas (free of carbon product is heated in the main heat exchanger.
dioxide, water, and methanol), entering the A part of the product carbon monoxide stream
separation process at ambient temperatures is separated and compressed (i). The open
and usually a pressure of 1.5–3.5 MPa, is carbon monoxide loop serves the purpose to
cooled down in the main heat exchanger (a), produce refrigeration duty for the cycle by
66 Cryogenic Technology

Figure 59. Hydrogen separation by liquid methane scrubbing

a) Main heat exchanger; b) Scrubbing column; c) J–T valves; d) Phase separation drum; e) Recycle compressor; f) Cold heat
exchanger; g) Methane separation column; h) Liquid methane pump; i) Carbon monoxide recycle compressor; j) Carbon
monoxide expander

expansion (j), providing reboiler heat and reflux pressure of typically ca. 3–10 MPa with a molar
for the methane separation column, and cooling fraction of typically > 95 mol%. The feed is
for the scrubbing column. cooled down in a series of heat exchangers
(a2
a3) and fed into the bottom of the scrubbing
6.7.3. Liquid Nitrogen Scrubbing column (b), where impurities are separated from
the hydrogen by means of liquid nitrogen as
The production of ammonia requires the adjust- scrubbing agent. The necessary refrigeration is
ment of the nitrogen/hydrogen ratio to a stoi- produced by precooling and expanding high-
chiometric value of 1:3 with a simultaneous pressure nitrogen in a J-T valve (c1). The top
removal of inerts such as methane, argon, and product is a mixture of nitrogen and hydrogen
especially carbon monoxide (latter to a level with a ratio of ca. 1:10. To establish the desired
below 5 ppm). The respective process is shown nitrogen: hydrogen ratio, additional high-pres-
in Figure 60. sure nitrogen is added to the product stream (c2).
The pretreated and precooled raw hydrogen Trace components are removed at the bottom of
(free of carbon dioxide and water) enters the the column and separated as fuel gas. To improve
cryogenic conditioning process at 215 K and a hydrogen recovery rate, the fuel gas can be
 Cryogenic Technology 67

Figure 60. Hydrogen separation by liquid nitrogen scrubbing

a) Heat exchangers; b) Scrubbing column; c) J–T valves

flashed and the hydrogen-rich vapor phase can where V ¼ volume of gas adsorbed at STP, ma ¼
be recycled into the feed gas. Possible additional mass of adsorbent, vm ¼ volume of gas required
refrigeration demand can be covered by injecting to form a monomolecular layer over the surface
liquid nitrogen into the fuel gas. of the adsorbent per unit mass of adsorbent, pi
¼ partial pressure of the species being
6.8. Physical Adsorption adsorbed, and pi,sat ¼ saturation vapor pressure
of the species at the adsorbent temperature. The
When a gas contacts a solid surface, part of the parameter C is the BET constant; a function of
gas is taken up by the solid. The general process the net energy of adsorption and the tempera-
is called sorption. If sorption is accompanied by ture of the adsorbent:
a chemical reaction between the gas and the
C ¼ expðDqa /RTÞ
solid, the process is called chemical adsorption
or chemisorption. If only the weak inter-
where Dqa ¼ difference between the heat of
molecular forces between the gas and solid
adsorption of the first layer (depends on the
are involved, the process is called physical
adsorbent) and the latent heat of vaporization of
adsorption or physisorption. Physical adsorp-
the gas. For the adsorption of nitrogen gas on
tion is used frequently in cryogenic separation
charcoal at 77.4 K, vm ¼ 0.1815 m3/kg, and Dqa
systems because the adsorbent material can be
¼ 2 492 J/mol. For the adsorption of argon gas
regenerated and reused.
on charcoal at 77.4 K, vm ¼ 0.2156 m3/kg and
Several theories have been proposed to cor-
Dqa ¼ 2 835 J/mol.
relate the data on physical adsorption. One of
Physical adsorption is widely used in cryo-
the most successful theories was proposed by
genic systems that separate and purify gas
BRUNAUER, EMMETT, and TELLER (the BET the-
mixtures and in systems that produce and main-
ory), which yields the following expression for
tain a vacuum. One of the advantages of the use
the volume of gas adsorbed per unit mass of
of an adsorbent as a vacuum pump is that the
adsorbent [122]:
medium in the vacuum chamber is not contam-
V vm  C  pi inated by the oil used in mechanical and diffu-
¼
ma ðpi;sat
pi Þ½1 þ ðC
1Þðpi /pi;sat Þ sion vacuum pumps. Cryogenic adsorption has
68 Cryogenic Technology

substantially extended the range of separable nitrogen. During this cycle, valves (c1),
gas mixtures. The gas mixtures that may be (c3), (c6), and (c8) are open, and valves
easily separated by using adsorption at cryo- (c2), (c4), (c5), and (c7) are closed. While
genic temperatures include, e.g., methane– adsorbent bed (b1) is active in adsorbing
helium, neon–helium, hydrogen–helium, nitro- the gas to be removed, adsorbent bed (b2)
gen–carbon dioxide. is regenerated or reactivated by warming it to
Nowadays, adsorbers are even used to a higher temperature and purging it with a
remove trace impurities from cryogenic gas. When the cold adsorbent bed (b1)
liquids. An example is the adsorption of becomes saturated with the material to be
acetylene and nitrous oxide from liquid oxy- removed, the valves are reversed, and the
gen in air separation plants (see Section 6.5.6). regenerated bed (b2) begins to adsorb the
The basic arrangement for adsorption separa- gaseous impurity.
tion of a gas mixture is shown in Figure 61. Cryogenic adsorption is also used to obtain
The gas stream is first cooled by passing it high-purity products which are required for
through a heat exchanger (a) and then through many applications using inert gases, such as
one of the adsorbent beds (b1). The compo- computer chip production [123]. Examples are
nent to be removed is adsorbed in the bed, the removal of carbon monoxide traces from
which is usually refrigerated by liquid nitrogen, hydrocarbons traces from oxygen, or
nitrogen traces from helium down to levels
typically below 1 ppb. A practical design
example of a cryogenic adsorption system
used for the purification of helium is shown
in Figure 62.

Figure 62. Cryogenic adsorption separation system for

helium purification (Linde AG)
a) Helium inlet; b) Heat exchanger; c) Activated charcoal;
Figure 61. General adsorption separation system d) Liquid nitrogen bath; e) Perlite insulation; f) Helium
a) Heat exchanger; b) Adsorbent beds; c) Control valves outlet
 Cryogenic Technology 69

26 M. Davies: The Physical Principles of Gas Liquefaction and

References Low Temperature Rectification, Longmans, Green & Co.,
1 R.B. Scott: Cryogenic Engineering, D. van Nostrand Co., London 1949, p. 37.
Princeton, N.J., 1959, p. 1. 27 F. Kreith, M.S. Bohn: Principles of Heat Transfer, 4th ed.,
2 R.G. Surlock: History and Origins of Cryogenics, Clarendon Harper & Row Publ., New York 1986, pp. 407–408.
Press, Oxford 1992. 28 G. Walker: Cryocoolers, Plenum Publishing, New York 1983,
3 C.A. Bailey: Advanced Cryogenics, Plenum Publishing, Part 1, pp. 280–282.
New York 1971, pp. 355–371. 29 J.G. Daunt: “The Production of Low Temperatures Down to
4 M. Sittig, S. Kidd: Cryogenics–Research and Applications, Hydrogen Temperatures,” Handbuch der Physik, vol. 4,
D. van Nostrand Co., Princeton, N.J., 1963, p. 190. Springer Verlag, Berlin 1956.
5 R.F. Barron: “Cryogenic Treatment of Metals to Improve Wear 30 V.M. Brodianski, A. Alexeev: “Cryogenics in Russia: Devel-
Resistance,” Cryogenics 22 (1982) 409–413. opment of Basic Cryogenic Cycles,” Proc. 20th Int. Cryogenic
6 D.N. Collins: “Deep Cryogenic Treatment of Tool Steels: A Eng. Conf. (ICEC20) 2005, Elsevier, Amsterdam pp. 11–18.
review,” Heat Treatment of Metals 23 (1996) 40–42. 31 G. Venkatarathnam: Cryogenic Mixed Refrigerant Processes,
7 R.M. Pillai, B.C. Pai, K.G. Satyanarayana: “Deep Cryo- Springer Verlag, Berlin 2008, pp. 39–45.
genic Treatment of Metals,” Tool & Alloy Steels 6 (1986), 32 D.T. Linnett, K.C. Smith: Mixed Refrigerant Cascade Liquefi-
205–208. ers for Natural Gas–Design and Optimization, Advances in
8 J.S. Harris: Ground Freezing in Practice, Thomas Telford, Cryogenic Engineering, vol. 15, Plenum Publishing, New York
London 1995. 1970, pp. 8–17.
9 Rubber Manufacturers Association: Scrap Tire Markets in the 33 S.C. Collins: “A Helium Cryostat,” Rev. Sci. Instrum. 18
United States. 9th Biennial Report, http://www.rma.org (1947) 157.
(accessed 9. January 2013). 34 S.C. Collins, R.L. Cannaday: Expansion Machines for Low
10 European Tyre & Rubber Manufacturers’ Association: End of Temperature Processes, Oxford University Press, London 1958.
life tyres. A valuable resource with growing potential, http:// 35 G.G. Haselden: “Refrigeration and Liquefaction Cycles”, in
www.etrma.org (accessed 9. January 2013). G.G. Haselden (ed.): Cryogenic Fundamentals, Academic
11 G. Ramos, F.J. Alguacil, F.A. Lopez: “The Recycling of End- Press, New York 1971, pp. 17–89.
of-life Tyres, Technological review,” Rev. Metal. 47 (2010) 36 H. Blechschmidt, K€ alte 9 (1976) 357.
273–284. 37 J. McInroy, Refrig. Sci. Technol. 1 (1967), 59–68.
12 R.M. Khadatkar, S. Kumar, S.C. Pattanayak: “Cryofreezing 38 R.F. Barron: “Cryoinstrumentation,” in H. von Leden, W.G.
and Cryofreezer,” Cryogenics 44 (2004) 661–678. Cahan (eds.): Cryogenics in Surgery, Medical Examination
13 D.C. Brown: “The Application of Cryogenic Fluids to Freezing Publ. Co., Flushing 1971, p. 115.
of Foods,” Advances in Cryogenic Engineering, vol. 12, 39 R.F. Barron: “Liquefaction Cycles for Cryogens,” Advances in
Plenum Publishing, New York 1967, pp. 11–22. Cryogenic Engineering, vol. 17, Plenum Publishing, New York
14 K.D. Timmerhaus, R.F. Kamm, J.D. Bays: “Whither Cry- 1972, pp. 20–36.
ogenics,” Advances in Cryogenic Engineering, vol. 10 B, 40 A. Alexeev, C. Haberstroh, H. Quack: “Further Development
Plenum Publishing, New York 1965, pp. 1–6. of a Mixed-gas Joule-Thomson Refrigerator”, Advances in
15 J.E. Zimmerman, R. Radebaugh: “Operation of a SQUID in a Cryogenic Engineering, vol. 43B, Plenum Publishing, New
Very Low-Power Cryocooler” NBS Special Publ. 508, 1978. York 1998, pp. 1667–1674.
16 K. Oshima, Y. Kyotani: “High-Speed Transportation Levi- 41 W.H. Hartwig: “Cryogenic Refrigeration Concepts Utilizing
tated by Superconducting Magnet,” Advances in Cryogenic Adsorption Pumping in Zeolites,” Advances in Cryogenic
Engineering, vol. 19, Plenum Publishing, New York 1974, Engineering, vol. 23, Plenum Publishing, New York 1978,
pp. 127–136. pp. 438–447.
17 A.U. Smith: Current Trends in Cryobiology, Plenum Press, 42 C.B. Hood, W.W. Vogelhuber, C.B. Barnes: “Helium Refrig-
New York 1970, pp. 153–180. erators for Operation in the 10 K to 30 K Range,” Advances in
18 J.P. Homasson: “Cryosurgery: State of the Art and Perspec- Cryogenic Engineering, vol. 9, Plenum Publishing, New York
tives in 2001,” Bulletin de l’Institut International du Froid 81 1964, pp. 406–506.
(2001) 2–11. 43 G. Walker: Stirling Engines, Oxford University Press, London
19 I.S. Cooper: “Cryogenic Surgery Introduction” in H. von 1980.
Leden, W.G. Cahan (eds.): Cryogenics in Surgery, Medical 44 A. Kirk, Proc. Inst. Civ. Eng. (London) 37 (1874) 244–315.
Examination Publ. Co., Flushing 1971, pp. 31–38. 45 J.W.L. K€ohler: “The Gas Refrigerating Machine and Its
20 C. Day: Basics and Applications of Cryopumps. CAS - CERN Position in Cryogenic Technique” in K. Mendelssohn (ed.):
Accelerator School and ALBA Synchrotron Light Facility: Progress in Cryogenics, vol. 2, Academic Press, New York
Course on Vacuum in Accelerators, Platja d’Aro, Spain, 1960, pp. 41–68.
16–24 May 2006, pp. 241–274. 46 G. Walker: “Design Guidelines for large Stirling Cryocoolers,”
21 K. Jousten: Handbook of Vacuum Technology, Wiley-VCH Cryogenics (1983) 113–114.
Verlag, Weinheim 2008, pp. 527–553. 47 N.G. Kirillov: “Cryogenic Technologies for Storage of Oil and
22 R.F. Barron: Cryogenic Systems, 2nd ed., Oxford University Oil Products,” Chem Petr. Eng. 40 (2004) 726–731.
Press, London 1985, pp. 61–64. 48 T. Finkelstein: “Computer Analysis of Stirling Engines,”
23 K. Wark: Thermodynamics, 3rd ed., McGraw-Hill, New York Advances in Cryogenic Engineering, vol. 20, Plenum Publish-
1977, p. 571. ing, New York 1975, pp. 269–282.
24 J.D. van der Waals: Over de Continuiteit van den Gas- en 49 R. Vuilleumier, US 1275507, 1918.
Vloeistoftoestand, Thesis, Leiden 1873. 50 R. White: “Vuilleumier Cycle Cryogenic Refrigeration”,
25 P. Atkins, J. de Paula: Physical Chemistry, 7th ed., Oxford AFFDL-TR-76–17, Wright-Patterson Air Force Base, Dayton,
University Press, London 2002, p. 22. Ohio, AD/A-027–055, 1976.
70 Cryogenic Technology

51 E. Solvay: DE 39280, 1887. 77 H.E. Hall, P.J. Ford, K. Thompson, Cryogenics 6 (1966) 80.
52 W.E. Gifford, H.O. McMahon: “A New Low Temperature Gas 78 B. Neganov, N. Borisov, M. Libury, Zh. Eksp. Teor. Fiz. 50
Expansion Cycle–Part II,” Advances in Cryogenic Engineer- (1966) 1445. Sov. Phys. JETP (Engl. Transl.) 25 (1966)
ing, vol. 5, Plenum Publishing, New York 1960, pp. 368–372. 959.
53 H.O. McMahon, W.E. Gifford: “A New Low-Temperature Gas 79 R. Radebaugh: Thermodynamic Properties of He3–He4 Solu-
Expansion Cycle–Part I,” Advances in Cryogenic Engineering, tions, NBS Technical Note 362, U.S. Govt. Printing Office,
vol. 5, Plenum Publishing, New York 1960, pp. 354–366. Washington, D.C., 1967.
54 W.E. Gifford: “The Gifford-McMahon Cycle,” Advances in 80 G. Frossati: “Proc. LT15,” J. Phys. (Paris) Collq. C6, 39 Suppl.
Cryogenic Engineering, vol. 11, Plenum Publishing, New York no. 8 (1978) 1578.
1961, pp. 152–159. 81 O.V. Lounasmaa: Experimental Principles and Methods Below
55 W.E. Gifford, T.E. Hoffman: “A New Refrigeration System for 1 K, Academic Press, New New 1974.
4.2 K,” Advances in Cryogenic Engineering, vol. 6, Plenum 82 T.M. Flynn: Cryogenic Engineering, 2nd ed., Marcel Dekker,
Publishing, New York 1961, pp. 82–94. New York 2005, pp. 652–665.
56 R. Radebaugh: “Pulse Tube Cryocoolers for Cooling Infrared 83 H. Hausen, H. Linde: Tieftemperaturtechnik, Springer Verlag,
Sensors,” Proc. SPIE XXVI, 4130 (2000) 363–379. Berlin 1985, pp. 281–337.
57 W.E. Gifford, R.C. Longsworth: “Pulse Tube Refrigeration,” 84 T.E. Strobridge: “The Thermodynamic Properties of Nitrogen
Trans. ASME J. Eng. for Ind., Paper No. 63-WA-290, August 1964. from 64 to 300 K between 0.1 and 200 Atmospheres”, NBS
58 E.I. Mikulin, A.A. Tarasov, M.M. Shkrebyonock: “Low Tem- Technical Note 129, 1962.
perature Expansion Pulse Tubes,” Advances in Cryogenic 85 G. Walker: Cryocoolers, Plenum Publishing, New York 1983,
Engineering vol. 29, Plenum Publishing, New York 1984, Part 2, pp. 6–8.
pp. 629–637. 86 H. Kreis: “Gewickelte W€armetauscher,” in W.F. Hess (ed.):
59 R. Radebaugh, J. Zimmermann, D.R. Smith, B. Loui: “A Apparate Handbuch: Technik, Bau, Anwendung, Vulkan
comparison of Three Types of Pulse Tube Refrigerators: Verlag, Essen 1990, pp. 263–268.
New Methods for Reaching 60 K,” Advances in Cryogenic 87 S.C. Collins, R.L. Canaday: Expansion Machines for Low
Engineering vol. 31, Plenum Publishing, New York 1986, Temperature Processes, Oxford Univeristy Press, London
pp. 779–789. 1958, p. 108.
60 N. Jiang, U. Lindemann, F. Giebeler, G. Thummes: “A 3He 88 ALPEMA: The Standards of the Brazed Aluminium Plate-Fin
Pulse Tube Cooler Operating Down to 1.3 K,” Cryogenics 44 Heat Exchanger Manufacturers’ Association. http://www.
(2004) 809. alpema.org, 2010 (accessed 10 January 2013).
61 A. Wanatabe, G.W. Swift, J.G. Brisson: “Superfluid Orifice 89 H.O. McMahon, R.J. Bowen, G.A. Bleyle: “A Perforated-Plate
Pulse Tube Refrigerator below 1 K,” Advances in Cryogenic Heat Exchanger,” Trans. of the ASME 72 (1950) 623–632.
Engineering, vol. 41, Plenum Publishing, New York 1996, pp. 90 R.B. Fleming: A Compact Perforated-Plate Heat Exchanger,
1519–1526. Advances in Cryogenic Engineering, vol. 14, Plenum Publish-
62 R. Radebaugh: Development of the Pulse Tube Refrigerator as ing, New York 1969, pp. 197–204.
an Efficient and Reliable Cryocooler, Proc. Inst. of Refrigera- 91 G. Venkatarathnam, S. Sarangi: “Matrix Heat Exchangers and
tion (London), vol. 1999–2000. their Application in Cryogenic Systems,” Cryogenics 30
63 J.H. Zia: “A Commercial Pulse Tube Cryocooler with 200 W (1999) 907–918.
Refrigeration at 80 K,” Cryocoolers 13 (2005) 165–171. 92 R.T. Jacobsen, S.G. Penoncello, E.W. Lemmon: Thermo-
64 G. Swift, J. Wollan: “Thermoacoustics for Liquefaction of dynamic Properties of Cryogenic Fluids, Plenum Publishing,
Natural Gas,” GasTIPS Fall 2002, 21–26. New York 1997.
65 W.E. Keller: Helium-3 and Helium-4, Plenum Publishing, New 93 R.F. Barron: Cryogenic Heat Transfer, Taylor & Francis,
York 1969, p. 112. Philadelphia 1999, pp. 272–280.
66 W.F. Giauque, J. Am. Chem. Soc. 49 (1927) 1846–1870. 94 F.P. Incropera, D.P. DeWitt, T.L. Bergman, A.S. Lavin:
67 P. Debye, Ann. Phys. (Leipzig) 81 (1929) 1154. Fundamentals of Heat and Mass Transfer, 5th ed., J. Wiley
68 W.F. Giauque, D.P. MacDougal, Phys. Rev. 43 (1933) 768. & Sons, New York 2006, pp. 686–688.
69 W.J. Haas, E.C. Wiersma, H.A. Kramers, Nature (London) 131 95 P.G. Kroeger: “Plated Tube Heat Exchanger: Analytical Inves-
(1933) 719. tigation of a New Surface Concept,” Advances in Cryogenic
70 C.V. Heer, C.B. Barnes, J.G. Daunt, Rev. Sci. Instrum. 25 Engineering, vol. 12, Plenum Publishing, New York 1967,
(1954) 1088. pp. 240–351.
71 J.A. Barclay, W.F. Stewart, W.C. Overton, R.J. Chandler, O.D. 96 R.B. Fleming: “The Effect of Flow Distribution in Parallel
Harkleroad: “Experimental Results on a Low-Temperature Channels of Counterflow Heat Exchangers,” Advances in
Magnetic Refrigerator,” Advances in Cryogenic Engineering, Cryogenic Engineering, vol. 12, Plenum Publishing, New
vol. 31, Plenum Publishing, New York 1985, pp. 743–752. York 1967, pp. 352–362.
72 W.A. Steyert: “Applications of Closed-Cycle Cryocoolers to 97 K.W. Cowans: “A Countercurrent Heat Exchanger That Com-
Small Superconducting Devices,” NBS Special Publ. 508, pensates Automatically for Maldistribution of Flow in Parallel
1978, p. 81. J. Appl. Phys. 49 (1978) 1227. Channels,” Advances in Cryogenic Engineering, vol. 19,
73 H. London: Proc. 1st Int. Conf. Low Temperature Physics, Plenum Publishing, New York 1974, pp. 437–444.
Clarendon Lab, Oxford 1951, p. 157. 98 M. Ledinegg: “Unstabilit€at der Str€omung bei nat€urlichem und
74 H. London, G.R. Clark, E. Mendoza: Phys. Rev. 128 (1962) Zwangsumlauf,” Die W€ arme 61 (1938) 891–898.
1992. 99 R.A. Ackermann: Cryogenic Regenerative Heat Exchangers,
75 K.G. Walters, W.M. Fairbank, Phys. Rev. 103 (1956) 262. Plenum Publishing, New York, 1997, p. 56.
76 P. Das, R. Bruyn Ouboter, K.W. Taconis: Proc. 9th Int. Conf. on 100 ANSI/API Standard 618-2008: Reciprocating Compressors for
Low Temperature Physics, Plenum Publishing, New York 1965, Petroleum, Chemical, and Gas Industry Services, 5th ed.,
p. 1253. American Petroleum Institute, Washington D.C. 2007.
 Cryogenic Technology 71

101 H.P. Bloch, J.A. Cameron, F.M. Danowski, R. James, J.S. 114 C.D. Holland: Multicomponent Distillation, Prentice Hall,
Swearingen, M.E. Weightman: Compressors and Expanders. Englewood Cliffs, NY, 1963, pp. 14–42.
Selection and Application for the Process Industry. Marcel 115 M. Ponchon: “Graphical Study of Distillation,” Tech. Mod 13
Dekker, New York 1982, pp. 121–134. (1921) 20–55.
102 G. Claude: Liquid Air, Oxygen and Nitrogen, P. Blakiston Son 116 W.L. McCabe, E.W. Thiele: “Graphical Design of Fractionat-
& Co., Philadelphia, Pa., 1913. ing Columns,” Ind. Eng. Chem. 17 (1925) 605–611.
103 M.L. Land: Expansion Turbines and Engines for Low Temper- 117 E.J. Henley; J.D. Seader: Equilibrium-Stage Separation Oper-
ature Processing, Advances in Cryogenic Engineering, vol. 5, ations in Chemical Engineering, J. Wiley & Sons, New York
Plenum Publishing, New York 1960, pp. 250–260. 1981, pp. 428–459.
104 H.-W. H€aring: Industrial Gas Processing, Wiley-VCH Verlag, 118 A. Arkharov, I. Marfenina, Y. Mikulin: Theory and Design of
Weinheim 2008, pp. 47–49. Cryogenic Systems, Mir Publ., Moscow 1981, pp. 333–337.
105 H. Sixsmith: “The Theory and Design of Gas-Lubricated 119 W. Rohde: “Verfahren zur Gewinnung von Rein-Argon aus
Bearings of High Stability,” Proc. 1st Int. Symp. on Luft,” Linde Berichte aus Technik und Wissenschaft 71
Gas-Lubricated Bearings, ACR-49, Office of Naval Research, (1991) 3–7.
Washington, D.C., 1959. 120 G.F. Hagenbach, J.H. Schiffhauer: “Commercial Production of
106 B.W. Birmingham, H. Sixsmith, W.A. Wilson: “The Applica- Neon,” Advances in Cryogenic Engineering, vol. 9, Plenum
tion of Gas-Lubricated Bearings to a Miniature Helium Expan- Publishing, New York 1964, pp. 557–560.
sion Turbine,” Advances in Cryogenic Engineering, vol. 7, 121 C. McKinley, F. Himmelberger: “Oxygen Plant Safety,” Chem.
Plenum Publishing, New York 1962, pp. 30–42. Eng. Prog. 53 (1957), 112–121.
107 W.H. Isalski: Separation of Gases, Clarendon Press, Oxford 122 S. Brunauer, P.H. Emmett, E. Teller: J. Am. Chem. Soc.
1989, pp. 246–247. 60 (1938), 309.
108 N. Eber: “25 Jahre Sulzer Tieftemperaturtechnik,” Techn. 123 R. Jain, J.K. Tseng: “Production of High-purity Gases by
Rundsch. Sulzer 4 (1981) 124–127. Cryogenic Adsorption,”. BOC Technology 8 (1998) 29–32.
109 G. Zahn, A. Hofmann, H. Bayer, H. Berndt, R. Doll, W. Herz,
M. Suesser, E. Turnwald, B. Vogeley, W. Wiedemann: “Test of
three Different Pumps for Circulating He I and He II,”
Cryogenics 32 (1992) ICEC Suppl. 45–47.
Further Reading
110 R. Pengo, S. Junker, G. Passardi, O. Pirotte, H.ten Kate: “Test R. Agrawal, D.M. Herron, H.C. Rowles, G.E. Kinard: “Cryogenic
Results of a 1.2 kg/s Centrifugal Liquid Helium Pump for the Technology,” Kirk Othmer Encyclopedia of Chemical
ATLAS Superconducting Toroid Magnet System,”. 19th Int. Technology, 5th ed., vol. 8, J. Wiley & Sons, New York
Cryogenic Eng. Conf., Grenoble, France, 22–26 Jul. 2002, 2004, p. 40–65.
pp. 71–74. C.J.F. Elenbaas: Cryogenic Freezing Manual, Trafford Publishing,
111 E. Turnwald: “Fl€ ussige Gase. Die Pumpe, die aus der K€alte Bloomington 2008.
kommt,” Chemische Industrie 115 (1992) 45–48. L. Zhang, L. Lin: Proceedings of the Twentieth International
112 S. Rush, L. Hall: “Tutorial on Cryogenic Submerged Electric Cryogenic Engineering Conference (ICEC 20), Elsevier,
Motor Pumps,” Proc. 18th Int. Pump User Symp., Houston, Amsterdam 2005.
May 2002, pp. 101–107. R.P. Reed, K.D. Timmerhaus: Cryogenic Engineering–Fifty Years
113 W.L. McCabe, J.C. Smith: Unit Operations of Chemical Engi- of Progress, Springer Verlag, Berlin 2007.
neering, 3rd ed., McGraw-Hill, New York 1976, p. 478.

View publication stats