CRYOGENICS

B.TECH. DEGREE COURSE

SCHEME AND SYLLABUS
(2002-03 ADMISSION ONWARDS)
MAHATMA GANDHI UNIVERSITY
KOTTAYAM, KERALA
CRYOGENICS (ELECTIVE - III)
M 806-2 3+1+0
Module 1
Introduction: Historical development- present areas involving cryogenic engineering.
Basic thermodynamics applied to liquefaction and refrigeration process - isothermal,
adiabatic and Joule Thomson expansion process - adiabatic demagnetization
efficiency to liquefaction and coefficient of performances irreversibility and losses.
Module 2
Low temperature properties of engineering materials: mechanical properties - thermal
properties - electrical and magnetic properties. Properties of cryogenic fluids -
materials of constructions for cryogenic applications.
Module 3
Gas liquefaction systems: production of low temperatures - general liquefaction
systems - liquefaction systems for neon, hydrogen, nitrogen and helium.
Module 4
Cryogenic refrigeration systems: ideal refrigeration systems- refrigerators using liquids
and gases as refrigerants - refrigerators using solids as working media.
Module 5
Cryogenic storage and transfer systems - Cryogenic fluid storage vessels cryogenic
fluid transfer systems. Application of cryogenics - cryo pumping - superconductivity
and super fluidity - cryogenics in space technology - cryogenics in biology and
medicine.
References
1. Cryogenic Systems - Barron R. F
2. Cryogenic Engineering - Scot R. W.
3. Cryogenic Engineering - Bell J.H.
2
Module I
INTRODUCTION TO CRYOGENIC SYSTEMS
1.1. Introduction
The word cryogenics means, literally, the production of icy cold; however,
the term is used today as a synonym for low temperatures. The point on the
temperature scale at which refrigeration in the ordinary sense of the term ends and
cryogenics begins is not sharply defined. The workers at the National Bureau of
Standards at Boulder, Colorado, have chosen to consider the field of cryogenics as that
involving temperatures below -150C (123 K) or - 240F (220
0
R). This is a logical
dividing line, because the normal boiling points of the so-called permanent gases, such
as helium, hydrogen, neon, nitrogen, oxygen, and air, lie below - 150C, while the
Freon refrigerants, hydrogen sulfide, ammonia, and other conventional refrigerants all
boil at temperatures above -150C. The position and range of the field of cryogenics
are illustrated on a logarithmic thermometer scale in Fig. 1.1.
In the field of cryogenic engineering, one is concerned with developing and
improving low-temperature techniques, processes, and equipment. As contrasted to
low-temperature physics, cryogenic engineering primarily involves the practical
utilization of low-temperature phenomena, rather than basic research, although the
dividing line between the two fields is not always clear-cut. The engineer should be
familiar with physical phenomena in order to know How to utilize them effectively;
the physicist should be familiar with engineering principles in order to design
experiments and apparatus.
A system may be defined as a collection of components united by definite
interactions or interdependencies to perform a definite function. Examples of common
engineering systems include the automobile, a petroleum refinery, and an electric
generating power plant. In many cases the distinction between a system and a
component depends upon one's point of view. For example, consider the transportation
system of a country. An automobile is a system also; however, it would be only one
part or subsystem of the entire transportation system. Going even further, one could
speak of the power system, braking system, steering system, etc., of the automobile. In
3
general, we shall use the term cryogenic system to refer to an interacting group of
components involving low temperatures. Air liquefaction plants, helium refrigerators,
and storage vessels with the associated controls are some examples of cryogenic
systems.
Fig. 1.1. The cryogenic temperature range.
1.2. Historical background
In 1726 Jonathan Swift wrote in Gulliver's account of his trip to the mythical
Academy of Lagado:
He (the universal artist) told us he had been thirty years employing his
thoughts for the improvement of human life. He had two large rooms full of wonderful
curiosities, and fifty men at work. Some were condensing air into a dry tangible
substance, by extracting the nitre, and letting the aqueous or fluid particles percolate.
4
At the time Gulliver's Travels was written, air was considered to be a
"permanent gas." Thus Swift envisioned the liquefaction of air some 150 years before
the feat was actually accomplished.
In the 1840s, in an attempt- to relieve the suffering of malaria patients, Dr.
John Gorrie, a Florida physician, developed an expansion engine for the production of
ice. Although Dr. Gorries engine was used only to cool air for air conditioning of
sickrooms and was not part of a cryogenic system, most large-scale air liquefaction
systems today use the same principle of expanding air through a work-producing
device, such as an expansion engine or expansion turbine, in order to extract energy
from the air so that the air can be liquefied.
It was not until the end of 1877 that a so-called permanent gas was first
liquefied. In this year Louis Paul Cailletet, a French mining engineer, produced a fog
of liquid-oxygen droplets by precooling a container filled with oxygen gas at
approximately 300 atm and allowing the gas to expand suddenly by opening a valve on
the apparatus. About the same time Raoul Pictet, a Swiss physicist, succeeded in
producing liquid oxygen by a cascade process.
In the early 1880s one of the first low-temperature physics laboratories, the
Cracow University Laboratory in Poland, was established by Szygmunt von
Wroblewski and K. Olszewski. They obtained liquid oxygen "boiling quietly in a test
tube" in sufficient quantity to study properties in April 1883. A few days later, they
also liquefied nitrogen. Having succeeded in obtaining oxygen and nitrogen as true
liquids (not just a fog of liquid droplets), Wroblewski and Olszewski, now working
separately at Cracow, attempted to liquefy hydrogen by Cailletet's expansion tech-
nique. By first cooling hydrogen in a capillary tube to liquid-oxygen temperatures and
expanding suddenly from 100 atm to I atm, Wroblewski obtained a fog of liquid-
hydrogen droplets in 1884, but he was not able to obtain hydrogen in the completely
liquid form.
The Polish scientists at the Cracow University Laboratory were primarily
interested in determining the physical properties of liquefied gases. The ever-present
problem of heat transfer from ambient plagued these early investigators because the
5
cryogenic fluids could be retained only for a short time before the liquids boiled away.
To improve this situation, an ingenious experimental technique was developed at
Cracow. The experimental test tube containing a cryogenic fluid was surrounded by a
series of concentric tubes, closed at one end. The cold vapor arising from the liquid
flowed through the annular spaces between the tubes and intercepted some of the heat
traveling toward the cold test tube. This concept of vapor shielding is used today in
conjunction with high-performance insulations for the long-term storage of liquid
helium in bulk quantities.
A giant step forward in preserving cryogenic liquids was made in 1892 when
James Dewar, a chemistry professor at the Royal Institution in London, developed the
vacuum-jacketed vessel for cryogenic-fluid storage. Dewar found that using a double-
walled glass vessel having the inner surfaces silvered (similar to present-day thermos
bottles) resulted in a reduction of the evaporation rate of the stored fluid by a factor of
30 over that of an uninsulated container. This simple container played a significant role
in the liquefaction of hydrogen and helium in bulk quantities. In May 1898 Dewar
produced 20 cm) of liquid hydrogen boiling quietly in a vacuum-insulated tube,
instead of a mist.
In 1895 two significant events in cryogenic technology occurred. Carl von
Linde, who had established the Linde Eismaschinen AG in 1879, was granted a basic
patent on air liquefaction in Germany. Although Linde was not the first to liquefy air,
he was one of the first to recognize the industrial implications of gas liquefaction and
to put these ideas into practice. Today the Linde Company is one of the leaders in
cryogenic engineering.
After more than 10 years of low-temperature study, Heike Kamerlingh Onnes
established the Physical Laboratory at the University of Lei den in Holland in 1895.
Onnes' first liquefaction of helium in 1908 was a tribute both to his experimental skill
and to his careful planning. He had only 360 liters of gaseous helium obtained by
heating monazite sand from India. More than 60 cm
3
of liquid helium was produced by
Onnes in his first attempt. Onnes was able to attain a temperature of 1.04 K in an
unsuccessful attempt to solidify helium by lowering the pressure above a container of
liquid helium in 1910.
6
The physicists at the Leiden laboratory were interested in investigating the
properties of materials at low temperatures and in checking natural principles known to
be valid at ambient temperatures, at cryogenic temperatures. It was in 1911, while he
was checking the various theories of electrical resistance of solids at liquid-helium
temperatures, that Onnes discovered that the electrical resistance of the mercury wire
on which he was experimenting suddenly decreased to zero. This event marked the
first observation of the phenomenon of superconductivity-the basis for many novel
devices used today.
In 1902 Georges Claude, a French engineer, developed a practical system for
air liquefaction in which a large portion of the cooling effect of the system was
obtained through the use of an expansion engine. Claude's first engines were
reciprocating engines using leather seals (actually, the engines were simply modified
steam engines). During the same year, Claude established l'Air Liquide to develop and
produce his systems.
Although cryogenic engineering is considered a relatively new field in the
U.S., it must be remembered that the use of liquefied gases in U.S. industry began in
the early 1900s. Linde installed the first air-liquefaction plant in the United States in
1907, and the first American-made air-liquefaction plant was completed in 1912. The
first commercial argon-production was put into operation in 1916 by the Linde
company in Cleveland, Ohio. In 1917 three experimental plants were built by the
Bureau of Mines, with the cooperation of the Linde Company, Air Reduction
Company, and the Jefferies-Norton Corporation, to extract helium from natural gas of
Clay County,' Texas. The helium was intended for use in airships for World War I.
Commercial production of neon began in the United States in 1922, although Claude
had produced neon in quantity in France since 1907.
On 16 March 1926, Dr. Robert H. Goddard conducted the world's first
successful flight of a rocket powered by liquid-oxygen-gasoline propellant on a farm
near Auhurn, Massachusetts. This first flight lasted only 2 seconds, and the rocket
reached a maximum speed of only 22 m/s (50 mph). Dr. Goddard continued his work
during the 1930s and by 1941 he had brought his cryogenic rockets to a fairly high
7
degree of perfection. In fact, many of the devices used in Dr. Goddard's rocket systems
were used later in the German V-2 weapons system.
During that same year (1926), William Francis Giauque and Peter Debye
independently suggested the adiabatic demagnetization method for obtaining ultralow
temperatures (less than 0.1 K). It was not until 1933 that Giauque and MacDougall at
Berkeley and De Haas, Wiersma, and Kramers at Leiden made use of the technique to
reach temperatures from 0.3 K (Giauque and MacDougall) to 0.09 K (De Haas et al.).
As early as 1898 Sir James Dewar made measurements on heat transfer
through evacuated powders. In 1910 Smoluchowski demonstrated the significant
improvement in insulating quality that could be achieved by using evacuated powders
in comparison with unevacuated insulations. In 1937 evacuated-powder insulations
were first used in the United States in bulk storage of cryogenic liquids. Two years
later, the first vacuum powder-insulated railway tank car was built for the transport of
liquid oxygen.
The world became aware of some of the military implications of cryogenic
technology in 1942 when the German V -2 weapon system was successfully test-fired
at Peenemunde under the direction of Dr. Walter Dornberger. The V-2 weapon system
was the first large, practical liquid propellant rocket. This vehicle was powered by
liquid oxygen and a mixture of 75 percent ethyl alcohol and 25 percent water.
Around 1947 Dr. Samuel C. Collins of the department of mechanical
engineering at Massachusetts Institute of Technology developed an efficient liquid-
helium laboratory facility. This event marked the beginning of the period in which
liquid-helium temperatures became feasible and fairly economical. The Collins helium
cryostat, marketed by Arthur D. Little, Inc., was a complete system for the safe,
economical liquefaction of helium and could be used also to maintain temperatures at
any level between ambient temperature and approximately 2 K.
The first buildings for the National Bureau of Standards Cryogenic
Engineering Laboratory were completed in 1952. This laboratory was established to
provide engineering data on materials of construction, to produce large quantities of
liquid hydrogen for the Atomic Energy Commission, and to develop improved
8
processes and equipment for the fast growing cryogenic field. Annual conferences in
cryogenic engineering have been sponsored by the National Bureau of Standards
(sometimes sponsored jointly with various universities) from 1954 (with the exception
of 1955) to 1973. At the 1972 conference at Georgia Tech in Atlanta, the Conference
Board voted to change to a biennial schedule alternating with the Applied
Superconductivity Conference. This schedule has been followed with meetings in
Kingston, Ontario in 1975, Boulder in 1977, Madison, Wisconsin in 1979, San Diego,
California in 1981, and Colorado Springs in 1983.
Early in 1956 work with liquid hydrogen was greatly accelerated when Pratt
and Whitney Aircraft was awarded a contract to develop a liquid hydrogen-fueled
rocket engine for the United States space program. The following year the Atlas ICBM
was successfully test-fired. The Atlas was powered by a liquid-oxygen-RP-l propellant
combination and had a sea level thrust of I. 7 MN (380,000 Ib
f
). At the Cape Kennedy
Space Center on 27 October 1961, the first flight test of the Saturn launch vehicle was
conducted. The Saturn V was the first space vehicle to use the liquid hydrogen-liquid-
oxygen propellant combination.
In 1966, Hall, Ford, and Thompson at Manchester, and Neganov, Borisov,
and Liburg at Moscow independently succeeded in achieving continuous refrigeration
below 0.1 K using a He
3
-He
4
dilution refrigerator. This new refrigeration technique
had been proposed in 1951 by H. London. The dilution refrigerator had certain
advantages over the magnetic refrigerator, which relied on the adiabatic
demagnetization principle to achieve temperatures in the 0.01 K to 0.10 K range. Thus
considerable research effort has been devoted to the study and improvement of the
dilution refrigerator. .
In 1969 a 3250-hp, 20D-rpm superconducting motor (Fawley motor) was
constructed by the International Research and Development Co., Ltd., in England. In
1972 IRD installed a superconducting motor in a ship to drive the electrical propulsion
system.
This chronology of cryogenic technology is summarized in Table 1.1.
9
We see that cryogenics has grown from an interesting curiosity in the times of
Linde and Claude to a diversified, vital field of engineering.
1.3. Present areas involving cryogenic engineering
Present-day applications of cryogenic technology are widely varied, both in
scope and in magnitude. Some of the areas involving cryogenic engineering include:
1. Rocket propulsion systems. All the large United States launch vehicles use liquid
oxygen as the oxidizer. The Space Shuttle propulsion system uses both
cryogenic fluids, liquid oxygen, and liquid hydrogen.
2. Studies in high-energy physics. The hydrogen bubble chamber uses liquid
hydrogen in the detection and study of high-energy particles produced in large
particle accelerators.
3. Electronics. Sensitive microwave amplifiers, called masers, are cooled to liquid-
nitrogen or liquid-helium temperatures so that thermal vibrations of the atoms of
the amplifier element do not seriously interfere with absorption and emission of
microwave energy. CryogenicaIly cooled masers have been used in missile
detectors, in radio astronomy to listen to faraway galaxies, and in space
communication systems.
Table 1.1. Chronology of cryogenic technology
Year Event
1877 Cailletet and Pictet liquefied oxygen (Pictet 1892).
1879 Linde founded the Linde Eismaschinen AG.
1883 Wroblewski and Olszewski completely liquefied nitrogen and oxygen at the
Cracow University Laboratory (Olszewski 1895).
1884 Wroblewski produced a mist of liquid hydrogen.
1892 Dewar developed a vacuum-insulated vessel for cryogenic-fluid storage (Dewar
1927).
1895 Onnes established the Leiden Laboratory. Linde was granted a basic patent on
air liquefaction in Germany.
1898 Dewar produced liquid hydrogen in bulk at the Royal Institute of London.
1902 Claude established l'Air Liquide and developed an air-liquefaction system using
an expansion engine.
10
1907 Linde installed the first air-liquefaction plant in America. Claude produced
neon as a by-product of an air plant.
1908 Onnes liquefied helium (Onnes 1908).
1910 Linde developed the double-column air-separation system.
1911 Onnes discovered superconductivity (Onnes 1913).
1912 First American-made air-liquefaction plant completed.
1916 First commercial production of argon in the United States.
1917 First natural-gas liquefaction plant to produce helium.
1922 First commercial production of neon in the United States.
1926 Goddard test-fired the first cryogenically propelled rocket. Cooling by adiabatic
demagnetization independently suggested by Giauque and Debye.
1933 Magnetic cooling used to attain temperatures below I K.
1934 Kapitza designed and built the first expansion engine for helium. Evacuated-
powder insulation first used on a commercial scale in cryogenic-fluid storage
vessels.
1939 First vacuum-insulated railway tank car built for transport of liquid oxygen.
1942 The V-2 weapon system was test-fired (Dornberger 1954). The Collins cryostat
developed.
1948 First 140 ton/day oxygen system built in America.
1949 First 300 ton/day on-site oxygen plant for chemical industry completed.
1952 National Bureau of Standards Cryogenic Engineering Laboratory established
(Brickwedde 1960).
1957 LOX-RP-I propelled Atlas ICBM test-fired. Fundamental theory (BCS theory)
of superconductivity presented.
1958 High-efficiency multilayer cryogenic insulation developed (Black 1960).
1959 Large NASA liquid-hydrogen plant at Torrance, California, completed.
1960 Large-scale liquid-hydrogen plant completed at West Palm Beach, Rorida.
1961 Saturn launch vehicle test-fired.
1963 60 ton/day liquid-hydrogen plant completed by Linde Co. at Sacramento,
California.
1964 Two liquid-methane tanker ships designed by Conch Methane Services. Ltd.,
entered service.
1966 Dilution refrigerator using HeJ-He' mixtures developed (Hall 1966; Neganov
1966).
1969 3250-hp de superconducting motor constructed (Appleton 1971).
11
1970 Liquid oxygen plants with capacities between 60,000 mJ/h and 70,000 mJ/h
developed.
1975 Record high superconducting transition temperature (23 K) achieved.
Tiny superconducting electronic elements, called SQUIDs (superconducting
quantum interference devices) have been used as extremely sensitive digital
magnetometers and voltmeters. These devices are based on a superconducting
phenomenon, called the Josephson effect, which involves quantum mechanical
tunneling of electrons from one superconductor to another through an insulating
barrier.
In addition to SQUIDs, other electronic devices that utilize superconductivity in
their operation include superconducting amplifiers, rectifiers, transformers, and
magnets. Superconducting magnets have been used to produce the high magnetic fields
required in MHD systems, linear accelerators, and tokamaks. Superconducting
magnets have been used to levitate high-speed trains at speeds up to 500 km/h.
4. Mechanical design. By utilizing the Meissner effect associated with
superconductivity, practically zero-friction bearings have been constructed that
use a magnetic field as the lubricant instead of oil or air. Superconducting
motors have been constructed with practically zero electrical losses for such
applications as ship propulsion systems. Superconducting gyroscopes with
extremely small drift have been developed.
5. Space simulation and high-vacuum technology. To produce a vacuum that
approaches that of outer space (from 10-
12
torr to 10-
14
torr), one of the more
effective methods involves low temperatures. Cryopumping or freezing out the
residual gases, is used to provide the ultrahigh vacuum required in space
simulation chambers and in test chambers for space propulsion systems. The
cold of free space is simulated by cooling a shroud within the environmental
chamber by means of liquid nitrogen. Dense gaseous helium at less than 20 K or
liquid helium is used to cool the cryopanels that freeze out the residual gases.
6. Biological and medical applications. The use of cryogenics in biology, or
cryobiology, has aroused much interest. Liquid-nitrogen-cooled containers are
used to preserve whole blood, tissue, bone marrow, and animal semen for long
12
periods of time. Cryogenic surgery (cryosurgery) has been used for the treatment
of Parkinson's disease, eye surgery, and treatment of various lesions. This
surgical procedure has many advantages over conventional surgery in several
applications.
7. Food processing. Freezing as a means of preserving food was used as far back as
1840. Today frozen foods are prepared by placing cartons on a conveyor belt and
moving the belt through a liquid-nitrogen bath or gaseous-nitrogen-cooled
tunnel. Initial contact with liquid nitrogen freezes all exposed surfaces and seals
in flavor and aroma. The cryogenic process requires about 7 minutes compared
with 30 to 48 minutes required by conventional methods. Liquid nitrogen has
also been used as the refrigerant in frozen-food transport trucks and railway cars.
8. Manufacturing processes. Oxygen is used to perform several important functions
in the steel manufacturing process. Cryogenic systems are used in making
ammonia. Pressure vessels have been formed by placing a preformed cylinder in
a die cooled to liquid-nitrogen temperatures. High-pressure nitrogen gas is
admitted into the vessel until the container stretches about 15 percent, and the
vessel is removed from the die and allowed to warm to room temperature.
Through the use of this method, the yield strength of the material has been
increased 400 to 500 percent.
9. Recycling of materials. One of the more difficult items to recycle is the
automobile or truck tire. By freezing the tire in liquid nitrogen, tire rubber is
made brittle and can be crushed into small particles. The tire cord and metal
components in the original tire can be separated easily from the rubber, and the
rubber particles can be used again for other items. At present the cryogenic
technique is the only effective one to recover the rubber from steel radial tires.
These are a few of the areas involving cryogenic engineering-a field in which
new developments are continually being made.
13
MODULE II
LOW-TEMPERATURE PROPERTIES OF
ENGINEERING MATERIALS
Familiarity with the properties and behavior of materials used in any system
is essential to the design engineer. At first thought, one might suppose that by
observing the variation of material properties at room temperature he could extrapolate
this information down through the relatively small temperature range involved in
cryogenics (some 300C) with fair confidence. In some cases, such as for the elastic
constants, this may be done with acceptable accuracy. On the other hand, there are sev-
eral significant effects that appear only at very low temperatures. Some examples of
these effects include the vanishing of specific heats, superconductivity, and ductile-
brittle transitions in carbon steel. None of these phenomena can be inferred from
property measurements made at near-ambient temperatures.
In this chapter, we shall investigate the physical properties of some
engineering materials commonly used in cryogenic engineering. The primary purpose
of the chapter is to examine the effect of variation of temperature on material
properties in the cryogenic temperature range and to become familiar with the
properties and behavior of materials at low temperatures.
MECHANICAL PROPERTIES
2.1. Ultimate and yield strength
For many materials, there is a definite value of stress at which the strain of the
material in a simple tensile test begins to increase quite rapidly with increase in stress.
This value of stress is defined as the yield strength Sy of the material. For other
materials that do not exhibit a sharp change in the slope of the stress-strain curve, the
yield strength is defined as the stress required to permanently deform the material in a
simple tensile test by 0.2 percent (sometimes 0.1 percent is used). The ultimate
strength Su of a material is defined as the maximum nominal stress attained during a
simple tensile test. The temperature variation of the ultimate and yield strengths of
several engineering materials is shown in Figs. 2.1 and 2.2.
14
Fig.2.1. Ultimate strength for several engineering materials: (I) 2024-T4 aluminum; (2)
beryllium copper; (3) K Monel; (4) titanium; (5) 304 stainless steel; (6) CI020 carbon
steel; (7) 9 percent Ni steel; (8) Teflon; (9) Invar-36 (Durham et al. 1962).
Many engineering materials are alloys, in which alloying materials with
atoms of different size from those of the basic material are added to the basic material;
for example, carbon is added to iron to produce carbon steel. If the alloying-element
atoms are smaller than the atoms of the basic material, the smaller atoms tend to
migrate to regions around dislocations in the metal. The presence of the smaller atoms
around the dislocation tends to "pin" the dislocation in place or make dislocation
motion more difficult (Wigley 1971). The yielding process in alloys takes place when
a stress large enough to pull many dislocations away from their "atmosphere" of
alloying atoms is applied. Plastic deformation or yielding occurs because of the gross
motion of these dislocations through the material.
As the temperature is lowered, the atoms of the material vibrate less
vigorously. Because of the decreased thermal agitation of the atoms, a larger applied
stress is required to tear dislocations from their atmosphere of alloying atoms. From
15
this line of reasoning, we should expect that the yield strength for alloys would
increase as the temperature is decreased. This has been found to be true for most
engineering materials.

Fig. 2.2. Yield strength for several engineering materials: (1) 2024- T4 aluminum: (2)
beryllium copper; (3) K Monel; (4) titanium; (5) 304 stainless steel; (6) CI020 carbon
steel; (7) 9 percent Ni steel; (8) Teflon; (9) Invar-36 (Durham et al. 1962).
2.2. Fatigue strength
There are several ways to express the resistance of a material to stresses that
vary with time, but the most common method is a simple reversed bending test The
stress required for failure after a given number of cycles is called the fatigue strength
S
f
. Some materials, such as carbon steels and aluminum-magnesium alloys, have the
property that the fatigue failure will not occur if the stress is maintained below a
certain value, called the endurance limit S
e
, no matter how many cycles have elapsed.
16
The temperature variation of the fatigue strength at 10
6
cycles for several materials is
shown in Fig. 2.3.
Because of the time involved to complete a test, fatigue data at cryogenic
temperatures are not as extensive as ultimate-strength and yield strength data;
however, for the materials that have been tested, it has been found that the fatigue
strength increases as the temperature is decreased.
Fig. 2.3. Fatigue strength at 1()6 cycles: (I) 2024- T4 aluminum; (2) beryllium copper;
(3) K Monel; (4) titanium; (5) 304 stainless steel; (6) CI020 carbon steel (Durham et
al. 1962).
Fatigue failure generally occurs in three stages for the case of more than
about 10) cycles: microcrack initiation, slow crack growth until a critical crack size is
achieved, and the final rapid failure either by ductile rupture or by cleavage.
Microcrack initiation usually occurs at the surface of the specimen as a result of
inhomogeneous shear deformation or at small flaws near the surface. The growth of
the microcracks occurs as the material fails at the high-stress region around the tip of
17
the crack. As the temperature of a material is decreased, a larger stress is required to
extend the crack; therefore, we should expect to observe that the fatigue strength
increases as the temperature is decreased.
For aluminum alloys, it has been found (De Money and Wolfer 1961) that the
ratio of fatigue strength to ultimate strength remains fairly constant as the temperature
is lowered. This fact may be used in estimating the fatigue strength for nonferrous
materials at cryogenic temperatures if no fatigue data are available at the low
temperatures.
2.3. Impact strength
The Charpy and the Izod impact tests give a measure of the resistance of a
material to impact loading. These tests indicate the energy absorbed by the material
when it is fractured by a suddenly applied force. Charpy impact strength of several
materials is shown in Fig. 2.4.
Fig. 2.4. Charpy impact strength at low temperatures: (I) 2024T4 aluminum; (2) beryl
lium copper; (3) K Monel; (4) titanium; (5) 304 stainless steel; (6) CI020 carbon steel;
(7) 9 percent Ni steel (Durham et al. 1962).
18
A ductile-brittle transition occurs in some materials, such as carbon steel, at
temperatures ranging from room temperature down to 78 K, which results in a severely
reduced impact strength at low temperatures. The impact behavior of a metal is largely
determined by its lattice structure. The face-centered-cubic (FCC) lattice has more slip
planes available for plastic deformation than does the body-centered-cubic (BCC)
lattice. In addition, interstitial impurity atoms interact only with edge dislocations to
retard slipping in the FCC structure; whereas, both edge and screw dislocations can
become pinned in the BCC structure. The metals with a FCC lattice or hexagonal
lattice tend to fail by plastic deformation in the impact test (thereby absorbing a
relatively large amount of energy before breaking) and retain their resistance to impact
as the temperature is lowered. The metals with a BCC lattice tend to reach a
temperature at which it is more energetically favorable to fracture by cleaving (thereby
absorbing a relatively small amount of energy). Thus these materials become brittle at
low temperatures.
Most plastics and rubber materials become brittle upon cooling below a
transition temperature also. Two notable exceptions are Teflon and Kel-F.
2.4. Hardness and ductility
The ductility of materials is usually indicated by the percentage elongation to
failure or the reduction in cross-sectional area of a specimen in a simple tensile test.
The accepted dividing line between a brittle material and 'a ductile one is 5 percent
elongation or a strain of 0.05 cm/cm at failure. Materials that elongate more than this
value before failure are called ductile; those with less than 5 percent elongation are
called brittle. The ductility of several materials as a function of temperature is shown
in Fig. 2.5.
For materials that do not exhibit a ductile-to-brittle transition at low
temperatures, the ductility usually increases somewhat as the temperature is lowered.
For the carbon steels, which do have a low-temperature transition, the elongation at
failure drops from 25 to 30 percent for the softer steels down to 2 or 3 percent during
the transition. Obviously, these materials should not be used at low temperatures in
any applications in which ductility is important.
19
Hardness of metals is measured by the indention made in the surface of the
material by a standard indenter. Common hardness tests include (1) Brinell (ball
indenter), (2) Vickers (diamond pyramid indenter), and (3, Rockwell (ball or diamond
indenter with various loads). In general, the hardness of metals as measured by any of
these means is directly proportional to the ultimate strength of the material; therefore,
the hardness increases as the temperature is decreased. This proportionality is to be
expected because a penetration test is essentially a miniature tensile test.
Fig. 2.5. Percent elongation for various materials: (I) 2024-T4 aluminum; (2) beryllium
copper; (3) K Monel; (4) titanium; (5) 304 stainless steel; (6) C1020 carbon steel; (7) 9
percent Ni steel (Durham et al. 1962).
2.5. Elastic moduli
There are three commonly used elastic moduli: (1) Young's modulus E. the
rate of change of tensile stress with respect to strain at constant temperature in the
elastic region; (2) shear modulus G. the rate of change of shear stress with respect to
shear strain at constant temperature in the elastic region; and (3) bulk modulus B, the
rate of change of pressure (corresponding to a uniform three-dimensional stress) with
respect to volumetric strain (change in volume per unit volume) at constant tempera-
ture. If the material is isotropic (many polycrystalline materials can be considered
20
isotropic for engineering purposes), these three moduli are related through Poisson's
ratio v, the ratio of strain in one direction due to a stress applied perpendicular to that
direction to the strain parallel to the applied stress:
E
B
3(1 2)

(2.1)
E
G
2(1 )

+
(2.2)
Fig. 2.6. Young's modulus at low temperatures: (I) 2024- T 4 aluminum; (2) beryllium
copper; (3) K Monel; (4) titanium; (5) 304 stainless steel; (6) ClO20 carbon steel; (7) 9
percent Ni steel (Durham et al. 1962).
The variation of Young's modulus with temperature for several materials is
given in Fig. 2.6.
As the temperature is decreased, interatomic and intermolecular forces tend to
increase because of the decrease in the disturbing influence of atomic and molecular
vibrations. Because elastic reaction is due to the action of these intermolecular and
interatomic forces, one would expect the elastic moduli to increase as the temperature
21
is decreased. In addition, it has been found experimentally that Poisson's ratio for
isotropic materials do not change appreciably with change in temperature in the
cryogenic range; therefore, all three of the previously mentioned elastic moduli vary in
the same manner with temperature.
THERMAL PROPERTIES
2.6. Thermal conductivity
The thermal conductivity k
t
of a material is defined as the heat-transfer rate
per unit area divided by the temperature gradient causing the heat transfer. The
variation of thermal conductivity of several solids is shown in Fig. 2.7. Values of the
thermal conductivity of cryogenic liquids and gases are given in Appendixes B through
E.
Fig. 2.7. Thermal conductivity of materials at low temperatures: (ll 2024-T4
aluminum: (3) beryllium copper; (3) K Monel; (4) titanium; (5) 304 stainless steel; (6)
C1020 carbon steel; (7) pure copper; (8) Teflon (Stewart and Johnson 1961).
22
To understand the variation of thermal conductivity at low temperatures, one
must be aware of the different mechanisms for transport of energy through materials.
There are three basic mechanisms responsible for conduction of heat through
materials: (l) electron motion, as in metallic conductors; (2) lattice vibrational energy
transport, or phonon motion, as in all solids; and (3) molecular motion, as in organic
solids and gases. In liquids, the primary mechanism for conduction heat transfer is the
transfer of molecular vibrational energy; whereas in gases, heat is conducted primarily
by transfer of translational energy (for monatomic gases) and translational and
rotational energy (for diatomic gases).
U sing principles of kinetic theory of gases (Eucken 1913), we may obtain a
theoretical expression relating the thermal conductivity to other properties of the
material:
t
k 1 8(9 5) C

where = specific heat ratio
= density of material
C
v
= specific heat at constant volume
v = average particle velocity
A = mean free path of particles, or average distance a particle travels before
it is deflected
For all gases the thermal conductivity decreases as the temperature is
lowered. Because the product of density and mean free path for a gas is practically
constant, and the specific heat is not a strong function of temperature, the thermal
conductivity of a gas should vary with temperature in the same manner as the mean
molecular speed u, as indicated by eqn. (2.3). From kinetic theory of gases (Present
1958), the mean molecular speed is given by
1 2
c
8g RT | `

. ,
23
where gc is the conversion factor in Newton's law (gc = 1 kg-m/N-s
2
in the SI system
of units; gc = 32.174 Ib
m
-ft/lb
r
sec
2
in the British system of units), R is the specific gas
constant for the particular gas (R = R
u
/M, where R
u
is the universal gas constant,
8.31434 J/mol-K or 1545 ft-lb
f
/lbmole-
o
R, and M is the molecular weight of the gas),
and T is the absolute temperature of the gas. A decrease in temperature results in a
decrease in the mean molecular speed and, consequently, a decrease in the gas thermal
conductivity.
All cryogenic liquids except hydrogen and helium have thermal con-
ductivities that increase as the temperature is decreased. Liquid hydrogen and helium
behave in a manner opposite to that of other liquids in the cryogenic temperature
range.
For heat conduction in solids, the thermal conductivity is related to other
properties by an expression similar to that for gases,
1
t 3
k C

(2.5)
Energy is transported in metals by both electronic motion and phonon motion;
however, in most pure electric conductors, the electronic contribution to energy
transport is by far the larger for temperatures above liquid-nitrogen temperatures. The
electronic specific heat is directly proportional to absolute temperature (see Sec. 2.7),
and the electron mean free path in this temperature range is inversely proportional to
absolute temperature. Because the density and mean electron speed are only weak
functions of temperature, the thermal conductivity of electric conductors above liquid-
nitrogen temperatures is almost constant with temperature, as would be predicted by
eqn. (2.5). As the absolute temperature is lowered below liquid-nitrogen temperatures,
the phonon contribution to the energy transport becomes significant. In this
temperature range, the thermal conductivity becomes approximately proportional to T-
2
for pure metals. The thermal conductivity increases to a very high maximum as the
temperature is lowered, until the mean free path of the energy carriers becomes on the
order of the dimensions of the material sample. When this condition is reached, the
boundary of the material begins to introduce a resistance to the motion of the carriers,
and the carrier mean free path becomes constant (approximately equal to the material
24
thickness). Because the specific heat decreases to zero as the absolute temperature
approaches zero, from eqn. (2.5) we see that the thermal conductivity would also
decrease with a decrease in temperature in this very low temperature region.
In disordered alloys and impure metals, the electronic contribution and tire.
phonon contribution to energy transport are of the same order of magnitude. There is
an additional scattering of the energy carriers due to the presence of impurity atoms in
impure metals. This scattering effect is directly proportional to absolute temperature.
Dislocations in the material provide a scattering that is proportional to T
2
, and grain
boundaries introduce a scattering that is proportional to T
3
at temperatures much lower
than the Debye temperature. All these effects combine to cause the thermal
conductivity of allays and impure metals to decrease as the temperature is decreased,
and the high maximum in thermal conductivity is eliminated in alloys.
2.7. Specific heats of solids
The specific heat of a substance is defined as the energy required to change
the temperature of the substance by one degree while the pressure is held constant (c

)
or while the volume is held constant (c
v
). For solids and liquids at ordinary pressures,
the difference between the two specific heats is small, while there is considerable
difference for gases. The variation of the specific heats with temperature gives an
indication of the way in which energy is being distributed among the various modes of
energy of the substance on a microscopic level.
Specific heat is a physical property that can be predicted fairly accurately by
mathematical models through statistical mechanics and quantum theory. For solids the
Debye model gives a satisfactory representation of the variation of the specific heat
with temperature. In this model, Debye assumed that the solid could be treated as a
continuous medium, except that the number of vibrational waves representing internal
energy must be limited to the total number of vibrational degrees of freedom of the
atoms making up the medium-that is, three times the total number of atoms. The
expression for the specific heat of a monatomic crystalline solid as obtained through
the Debye theory is
25
D/ T
3 4 x
3 x 2
0
D D D
9RT x e dx T T
C 3R D
(e 1)

| ` | `



. , . ,

(2.6) where
8
D
is called the Debye characteristic temperature and is a property of the material, and
D(T/
D
) is called the Debye function. A plot of the specific heat as given by eqn. (2.6)
is shown in Fig. 2.8, and the Debye specific-heat function is tabulated in Table 2.1.
Values of
D
for several substances are given in Table 2.2.
Fig. 2.8. The Debye specific heat function.
26
Table 2.1. Debye specific heat function
T/
D
C
v
/R
T/
D
C
v
/R
T/
D
C
v
/R
0.08
0.09
0.10
0.12
0.14
0.16
0.18
0.20
0.25
0.30
0.35
0.40
0.1191
0.1682
0.2275
0.3733
0.5464
0.7334
0.9228
1.1059
1.5092
1.8231
2.0597
2.2376
0.45
0.50
0.60
0.70
0.80
0.90
1.00
1.10
1.20
1.30
1.40
1.50
2.3725
2.4762
2.6214
2.7149
2.7781
2.8227
2.8552
2.8796
2.8984
2.9131
2.9248
2.9344
1.60
1.70
1.80
1.90
2.00
2.20
2.40
2.60
2.80
3.00
4.00
5.00
2.9422
2.9487
2.9542
2.9589
2.9628
2.9692
2.9741
2.9779
2.9810
2.9834
2.9844
2.9900
The theoretical expression for the Debye temperature is given by
1
3
a
D
h 3N
k 4 V
| `

. ,
(2.7)
where h = Plancks constant
va = speed of sound in the solid
k = Boltzmanns constant
N/V = number of atoms per unit volume for the solid.
2.8 Specific Heat of Liquids and Gases
In general, the specific heat C
v
of cryogenic liquids decreases in the same way
that the specific heat of crystalline solids decreases as the temperature is lowered. At
low pressures the specific heat c
p
decreases with a decrease in temperature also. At
high pressures in the neighborhood of the critical point, humps in the specific-heat
curve are observed for all cryogenic fluids (and in fact for all fluids). From
thermodynamic reasoning we know that the difference in specific heats for a pure
substance is given by
27
P
P
p
C C T
T T

| ` | `



. , . ,
(2.10)
In the vicinity of the critical point, the coefficient of thermal expansion
P
1
T

| `



. ,
becomes quite large; therefore, one would expect large increases of the specific heat c
p
in the vicinity of the critical point also.
The specific heat of liquid helium behaves in a peculiar way-it shows a high,
sharp peak in the neighborhood of 2.17 K (3.91
o
R). The behavior of liquid helium is so
different from that of other liquids that we shall devote a separate section to a
discussion of its properties.
Gases at pressures low compared with their critical pressure approach the
ideal-gas state, for which the specific heat C
u
is independent of pressure. According to
the classical equipartition theorem, the specific heat of a material is given by
C
v
= Rf (2.11)
where f is the number of degrees of freedom of a molecule making up the material. For
monatomic gases, the only significant mode of energy is translational kinetic energy of
the molecules, which involves three degrees of freedom. From eqn. (2.11) the specific
heat C
v
for such monatomic gases as neon and argon in the ideal-gas state is C
v
=
3
/
2
R.
In diatomic gases, other modes are possible. For example, if we consider a "rigid-
dumbbell" model of a diatomic molecule such as nitrogen (i.e., we neglect vibration of
the molecule), there are three translational degrees of freedom plus two rotational
degrees of freedom. We consider only two rotational degrees of freedom because the
moment of inertia of the molecule about an axis through the centers of the two atoms
making up the molecule is negligibly small compared with the momel1t of inertia
about an axis perpendicular to the interatomic axis. From eqn. (2.11) the specific heat
C
v
for the rigid-dumbbell model of an ideal diatomic gas would be C
v
=
5
/
2
R. This is
true for most diatomic gases at ambient temperatures, at which the gases obey classical
statistics. Because the molecules in a diatomic gas are not truly rigid dumbbells, we
should also expect to have vibrational modes present, in which the two molecules
28
vibrate around an equilibrium position within the molecule under the influence of the
interatomic forces holding the molecule together. In this case, two more degrees of
freedom due to the vibrational mode would be present, so the specific heat for a
vibrating-dumbbell molecule would be C
v
= ~R according to the classical theory.
In the actual case, the rotational and vibrational modes are quantized, so they
are not excited if the temperature is low enough. The variation of the specific heat of
diatomic hydrogen gas with temperature is illustrated in Fig. 2.9. At very low
temperatures, only the translational modes are excited, so the specific heat C
v
takes on
the value
3
/
2
R, the same as that of a monatomic gas. To determine whether the
rotational modes will be excited, one must compare the temperature of the gas with a
characteristic rotation temperature
r
, defined by
2
r
2
h
8 I k

(2.12)
where h = Planck's constant
I = moment of inertia of the molecule about an axis perpendicular to the
interatomic axis
k = Boltzmann's constant
If the temperature is less than about
1
/
3

r
the rotational mode is not appreciably
excited; if the temperature is greater than about 3
r
, the rotational mode is practically
completely excited. Most diatomic gases become liquids at temperatures higher than
3
r
; however, H
2
, D
2
, and HD are exceptions to this statement. Because of the small
moment of inertia of the hydrogen molecule, the characteristic rotation temperature is
quite a bit above the liquefaction temperature for hydrogen (
r
= 85.4 K or 153.7
o
R
for hydrogen, so 3 , = 256.2 K or 461.1
o
R). The specific heat C
v
of hydrogen gas
rises from
3
/
2
R at temperatures below about 30 K (54R) to
5
/
2
R at temperatures above
255 K (459R).
The vibrational modes for a diatomic gas are also quantized and become
excited at temperatures on the order of a characteristic vibration temperature
v
defined by
29
h
k

f
(2.13)
Fig. 2.9. Variation of the specific heat c, for hydrogen gas.
where f
v
is the vibrational frequency of the molecule. The characteristic vibration
temperature for gases is much higher than cryogenic temperatures, so the vibrational
mode is not excited for gases at cryogenic temperatures (
v
= 6100 K or 1O,980
o
R for
hydrogen gas).
The change in the specific heat of hydrogen between 30 K and 255 K is
important for hydrogen liquefiers and hydrogen-cooled helium liquefiers because it
affects the effectiveness of the heat exchangers, as we shall see in Chap. 3.
At pressures higher than near ambient, the specific heats of gases vary in a
more complicated manner with temperature and pressure. A complete coverage of this
effect is beyond the scope of our present discussion.
2.9. Coefficient of thermal expansion
The volumetric coefficient of thermal expansion is defined as the frac-
tional change in volume per unit change in temperature while the pressure on the
material remains constant. The linear coefficient of thermal expansion
t
is defined as
the fractional change in length (or any linear dimension) per unit change in
temperature while the stress on the material remains constant. For isotropic materials,
30
= 3
t
. The temperature variation of the linear coefficient of thermal expansion
for several materials is shown in Fig. 2.10.
The temperature variation of the coefficient of thermal expansion may be
explained through a consideration of the intermolecular forces of a material. The
intermolecular potential-energy curve, as shown in Fig. 2.11, is not symmetrical.
Therefore, as the molecule acquires more energy (or as its temperature is increased),
its mean position relative to its neighbors becomes larger; that is, the material expands.
The rate at which the mean spacing of the atoms increases with temperature increases
as the energy or temperature of the material increases; thus, the coefficient of thermal
expansion increases as temperature is increased.
Fig. 2.10. Linear coefficient of thermal expansion for several materials at low
temperature: (I) 2024-T4 aluminum; (2) beryllium copper; (3) K Monel; (4) titanium;
(5) 304 stainless steel; (6) C1020 carbon steel (NBS Monograph 29, Thermal
Expansion of Solids at Low Temperatures).
Since both the specific heat and the coefficient of thermal expansion are
associated with intermolecular energy, one might expect to find a relationship between
31
the two properties. For crystalline solids, the Gruneisen relation expresses this
interdependence:
G
C
B

(2.14)
where is the density of the material, B is the bulk modulus, and
G
is the Gruneisen
constant, which is independent of temperature as a first approximation. Values of the
Gruneisen constant for some materials are presented in Table 2.4.
Table 2.4. Values of the Gruneisen constant for selected solids
Material
G
Aluminum 2.17
Copper 1.96
Gold 2.40
Iron 1.60
Lead 2.73
Nickel 1.88
Platinum 2.54
Silver 2.40
Tantalum 1.75
Tungsten 1.62
We have seen previously that the bulk modulus B is not strongly dependent
upon temperature for solids (it increases somewhat as the temperature is decreased);
therefore, the coefficient of thermal expansion for solids should vary with temperature
in the same way that the Debye specific heat varies with temperature. This general
variation has been found true experimentally. At very low temperatures (T <
D
/12),
the coefficient of thermal expansion is proportional to T
3

ELECTRIC AND MAGNETIC PROPERTIES
2.10. Electrical conductivity
The electrical conductivity k, of a material is defined as the electric current
per unit cross-sectional area divided by the voltage gradient in the direction of current
flow. The electrical resistivity r
e
is the reciprocal of the electrical conductivity. The
variation with temperature of the electrical resistivity of several materials is shown in
Fig. 2.12.
32
Fig.2.11. Variation of the intermolecular potential energy for a pair of molecules. At
absolute zero, the molecular spacing would be r
0
.
When an external electric field is applied to an electric conductor, free
electrons in the conductor are forced to move in the direction of the applied field. This
motion is opposed by the positive ions of the metal lattice and impurity atoms present
in the material. Decreasing the temperature of the conductor decreases the vibrational
energy of the ions, which in turn results in a smaller interference with electron motion.
Therefore, the electrical conductivity increases as the temperature is lowered for
metallic conductors.
One of the first theories of electric resistance was developed by Drude (Kittel
1956), who treated the free electrons as an "electron gas." He obtained the following
expression for the electrical conductivity:
2
e
e
(N/ V)e
k
m

(2.15)
where N/V = number of free electrons per unit volume
33
e = charge of an electron
= electron mean free path
m
e
= mass of an electron

= average speed of an electron

Equation (2.15) gives the correct order of magnitude for the electrical
conductivity of a metal such as silver at room temperature if we assume that N/V is on
the order of the number of valence electrons per unit volume, is on the order of the
interatomic spacing, and the mean electron kinetic energy is given by
2
3
e 2
m kT
,
according to classical theory. In order to obtain the correct temperature dependence for
the electrical conductivity from eqn. (2.15), however, we must assume that the electron
mean free path is inversely proportional to T

because the electrical conductivity is

approximately inversely proportional to absolute temperature.
An application of quantum mechanics and band theory to the problem of
prediction of the electrical conductivity does not result in an equation different from
eqn. (2.15); however, it does allow us to predict the correct relationships for the
electron velocity and mean free path. According to the quantum-mechanical picture,
the velocity appearing in eqn. (2.15) is the average velocity of the electrons near the
so-called Fermi surface of the metal. For ordinary temperatures this velocity is
practically constant:
1
3 n
F
e e
h 3 N
2 m 2 m
| `

. ,
(2.16)
34
Fig. 2.12. Electrical resistivity ratio for several materials at low temperatures:
(1)copper; (2) silver; (3) iron; (4) aluminum (Stewart and Johnson 1961).
2.11. Superconductivity
One of the properties of certain materials that appears only at very low
temperatures is superconductivity - the simultaneous disappearance of all electric
resistance and the appearance of perfect diamagnetism. In the absence of a magnetic
field, many elements, alloys, and compounds become superconducting at a fairly well-
defined temperature, called the transition temperature in zero field T
0
.
Superconductivity can be destroyed by increasing the magnetic field around the
material to a large enough value. The magnetic field strength required to destroy
superconductivity is called the critical field H
C
. For Type I superconductors, there is a
single value of the critical field at which the transition from superconducting to normal
behavior is abrupt. For Type II superconductors (so-called "hard" superconductors),
there is a lower critical field H
C1
, at which the transition begins, and an upper critical
35
field H
C2
at which the transition is complete. Table 2.5 lists the critical field at absolute
zero H
0
and the transition temperature for several materials. Note that some alloys are
superconductors even though the pure elements making up the alloys are not
superconductors.
For Type I specimens in the shape of a long cylinder or wire placed parallel
10 the applied magnetic field, the critical field is well defined at every temperature and
is a function of temperature. The critical field may be related approximately to the
absolute temperature by
2
C 0 0
H H [1 (T/ T ) ] (2.24)
Although the parabolic relationship is not exactly true, it is adequate for many
purposes.
The phenomenon of superconductivity was discovered by Kamerlingh Onnes
in 1911 while investigating the electric resistance of mercury wire. After its discovery,
this new state of matter became the object of investigation by several theoretical and
experimental physicists to determine the properties of superconductors and to try to
explain the basic mechanism of the phenomenon. The early attempts to develop a
thermodynamic theory were unsuccessful because it was assumed that any magnetic
field present within the material in the normal state remained frozen-in when the
substance became superconducting.
Gorter (1933) applied the principles of reversible thermodynamics to the
superconducting phenomenon and obtained results that were in excellent agreement
with experimental measurements. This caused quite a bit of discussion because many
investigators thought that the transition was not thermodynamically reversible. In the
same year, this matter was cleared up somewhat by an experiment by Meissner and
Ochsenfeld (1933). They cooled a monocrystal of tin in a magnetic field until it
became superconducting and found that the magnetic field was expelled from within
the material when the sample became superconducting, as shown in Fig. 2.13. The
results of the Meissner-Ochsenfeld experiment indicated that the magnetic flux density
within a bulk superconducting material (Type I) was always zero, no matter what value
of magnetic flux density existed within the material before the transition.
36
Superconductivity was shown to be a case not only of zero electric resistance but also
of perfect magnetic insulation. The Meissner effect as the expulsion of the magnetic
field when a material becomes superconducting is called, forms the basis for the
frictionless bearing and superconducting motor.
Fig. 2.13. The Meissner effect. When a material is normal, the magnetic flux lines can
penetrate the material. When the material becomes superconducting, the magnetic field
is expelled from thin the material.
Shortly after the discovery of the Meissner effect, two "phenomenological
theories" of superconductivity were proposed. Gorter and Casimir (1934) proposed a
two-fluid model, in which two types of electrons took part in the electric current-the
normal or "uncondensed" ones and the superconducting or "condensed" ones. This
model was used to predict thermodynamic properties of superconductors with good
success. Fritz and Heinz London (1935) proposed an electromagnetic theory that, in
conjunction with the classical Maxwell equations of electromagnetism, predicted many
of the electric and magnetic properties of superconductors. The results of the
calculations made by the Londons showed that the magnetic field actually did
penetrate the surface of a superconductor for a very small distance (on the order of 0.1
m) called the penetration depth. The results also predicted that extremely thin
superconductors should have much higher critical fields than thick ones. Kropschot
and Arp (1961) suggested that this property of thin superconducting films could be
used in high-field, thin-film superconducting magnets.
Both the theory of the Londons and the theory of Gorter-Casimir predicted
many of the properties of superconductors, but the theories did not explain the "why"
37
of the phenomenon. It was not until 1950 that an acceptable fundamental theory of
superconductivity was suggested. Frolich (1950) and Bardeen (1950) independently
proposed a mechanism involving the interaction of electrons in the superconductor.
Their ideas were developed into the BCS theory in 1957 by Bardeen, Cooper, and
Schrietfer (1957). They applied the quantum theory to pairs of electrons produced by a
special electron-lattice interaction. This process may be visualized as one in which the
first electron moving through the lattice causes a slight displacement of the ions, which
results in a positive screening charge slightly greater than the charge of the electron.
The second electron is then attracted toward the net positive charge region. A corre-
lation exists between all the pairs of electrons in the superconductor, and a finite
amount of energy is required to break up the pair, corresponding to the so-called
energy gap for the superconductor. As the temperature is increased above the threshold
value, enough energy is available to uncouple the electron pair, and the material
becomes normal.
Ginzburg and Landau (1950) developed a phenomenological theory that
explained the difference between the Type I and Type II superconductors in terms of a
parameter k, which was related to the surface energy of the material. Those materials
for which k is less than
1/ 2
have a positive surface energy and are Type I
superconductors. On the other hand, those materials for which k is greater than
1/ 2

have a negative surface energy and are Type II superconductors. This theory of
superconductivity is now known as the GLAG theory after the four Russian con-
tributors: Ginzburg, Landau, Abrikosov, and Gorkov. Based on the concepts of the
theory, relationships between the parameter K and the upper and lower critical fields
were established.
C
C1
H (0.081 ln k)
H
k 2
+
(2.26)
and
C2 C
H 2 kH
38
From a practical standpoint, eqn. (2.27) indicates that materials with a large
value of K will have a high value of the upper critical field H
C2
.
For materials which have a relative permeability of approximately unity, the
magnetic induction and the magnetization M are related by
=
0
(H+M) (2.28)
For Type I superconductors, the magnetic induction is zero, and H = M. At.
The behavior of Type II superconductors is similar, for fields less than the lower
critical field H
C1
. For higher magnetic fields, the field begins to penetrate the material,
and the behavior of the material is shown in Fig. 2.14.
There are several properties that change either abruptly or gradually when a
material makes the transition from the normal to the superconducting state. Some of
these properties include:
1. Specific heat. The specific heat increases abruptly when a material becomes
superconducting.
2. Thermoelectric effects. All the thermoelectric effects (Peltier, Thomson, and
Seebeck effects) vanish when a material becomes superconducting. A
superconducting thermocouple would not work at all.
3. Thermal conductivity. In the presence of a magnetic field, the thermal
conductivity of a pure metal decreases abruptly when the metal becomes
superconducting, although for some alloys (for example, Pb-Bi in a limited
range of compositions) the opposite is true. In the absence of a magnetic
field, there is no discontinuous change in the thermal conductivity, but the
slope change is sharp on the conductivity-temperature curve.
4. Electric resistance. For Type I superconductors the decrease of resistance to
zero is quite abrupt; however, for Type II superconductors the change is
sometimes spread over a temperature range as large as 1 K.
5. Magnetic permeability. The magnetic permeability suddenly decreases to zero
for Type I superconductors (the Meissner effect); however, for Type II
39
superconductors the Meissner effect is incomplete for magnetic fields greater
than the lower critical field.
Fig. 2.14. Variation of the magnetization with applied magnetic field intensity for
Type I and Type II superconductors.
Over the years considerable experimental effort has been expended in the
study of superconducting alloys and compounds in order to develop materials that
would remain superconducting at high values of magnetic field and at temperatures
nearer to liquid-hydrogen temperature (about 20 K). The materials most used for
superconducting magnets have been either the body-centered-cubic alloys of niobium
and titanium or the cubic beta-tungsten-type (Al5) compounds, such as Nb
3
Sn.
The body-centered-cubic alloys have been used in construction of magnets
with field strengths up to 10 tesla, while the niobium-titanium alloys can reach 12
tesla. In order to achieve thermal and magnetic stability, the niobium-titanium wires
used in superconducting magnets are usually clad with high-conductivity copper.
Although the compound niobium-tin has superconducting properties somewhat
superior to niobium-titanium, Nb
3
Sn is extremely brittle, and special fabrication
methods are required for its use.
PROPERTIES OF CRYOGENIC FLUIDS
40
2.12. Fluids other than hydrogen and helium
A summary of some of the thermodynamic and transport properties of fluids
commonly used in cryogenic engineering is shown in Table 2.6. Further data on fluid
properties are contained in the appendix.
Liquid nitrogen is a dear, colorless fluid that resembles water in appearance.
At standard atmospheric pressure (101.3 kPa) liquid nitrogen beils at 77.36 K
(139.3R) and freezes at 63.2 K (113.8R). Saturated liquid nitrogen at 1 atm has a
density of 807 kg/m
3
(50.4 lb
m
/ft
3
) in comparison with water at 15.6
o
C (60F), which
has a density of 999 kg/m
3
(62.3 lb
m
/ft). One of the significant differences between the
properties of liquid nitrogen and water (apart from the difference in normal boiling
points) is that the heat of vaporization of nitrogen is more than an order of magnitude
smaller than that of water. At the normal boiling point, liquid nitrogen has a heat of
vaporization of 199.3 kJ/kg (85.7 Btu/lb
m
), while water has a heat of vaporization of
2257 kJ/kg (970.3 Btu/lb
m
).
Nitrogen with an atomic number of 14 has two stable isotopes with mass
numbers 14 and 15. The relative abundance of these two isotopes is 10,000:38. They
are relatively difficult to separate.
Because nitrogen is the major constituent of air (n.08 percent by volume or
75.45 percent by weight), it is produced commercially by distillation of liquid air.
Liquid oxygen has a characteristic blue color caused by the presence of the
polymer or long-chain molecule O
4
. At 1 atm pressure liquid oxygen boils at 90.18 K
(162.3R) and freezes at 54.4 K (98.0
0
R). Saturated liquid oxygen at 1 atm is more
dense than water at 15
o
C (liquid-oxygen density = 1141 kg/m
3
= 71.2 lb
m
/ft
3
). Oxygen
is slightly magnetic (paramagnetic) in contrast to the other cryogenic fluids, which are
nonmagnetic. By measuring the magnetic susceptibility, small amounts of oxygen may
be detected in mixtures of other gases. Because of its chemical activity, oxygen
presents a safety problem in handling. Serious explosions have resulted from the
combination of oxygen and hydrocarbon lubricants.
Oxygen with an atomic number of 16 has three stable isotopes of mass
numbers 16, 17, and 18. The relative abundance of these three isotopes is 10,000:4:20.
41
Oxygen is manufactured in large quantities by distillation of liquid air
because it is the second most abundant substance in air (20.95 percent by volume or
23.2 percent by weight).
Liquid argon is a clear, colorless fluid with properties similar to those of
liquid nitrogen. It is inert and nontoxic. At 1 atm pressure liquid argon boils at 87.3 K
(157.1
o
R) and freezes at 83.8 K (150.8R). Saturated liquid argon at 1 atm is more
dense than oxygen, as one would expect, because argon has a larger molecular weight
than oxygen (argon density = 1394 kg/m
3
= 87.0 Ib
m
/ft
3
for saturated liquid at 1 atm).
The difference between the normal boiling point and the freezing point for argon is
only 3.5 K (6.3R).
Argon has three stable isotopes of mass numbers 36, 38, and 40 that occur in
a relative abundance in the atmosphere in the ratios 338:63:100,000.
Argon is present in atmospheric air in a concentration of 0.934 percent by
volume or 1.25 percent by weight. Because the boiling point of argon lies between that
of liquid oxygen and that of liquid nitrogen (slightly closer to that of liquid oxygen), a
crude grade of argon (90 to 95 percent pure) can be obtained by adding a small
auxiliary argon-recovery column in an air-separation plant.
Neon is another gas that can be produced as a by-product of an air separation
plant. Liquid neon is a clear, colorless liquid that boils at 1 atm at 27.09 K (48.8R)
and freezes at 24.54 K (44.3R). The boiling point of neon is somewhat above that of
liquid hydrogen. But the fact that neon is inert, has a larger heat of vaporization per
unit volume, and has a higher density makes it an attractive refrigerant when compared
with hydrogen.
Neon (atomic weight = 20.183) has three stable isotopes of mass numbers 20,
21, and 22 that occur in a relative abundance in atmospheric air in the ratios
10,000:28:971.
Liquid fluorine is a light yellow liquid having a normal boiling point of 85.24
K (153.4R). At 53.5 K (96.4R) and 101.3 kPa, liquid fluorine freezes as a yellow
solid, but upon subcooling to 45.6 K (82R) it transforms to a white solid. Liquid
42
fluorine is one of the most dense cryogenic liquids (density at normal boiling point =
1507 kg/m
3
= 94.1 Ib
m
/ft
3
).
Fluorine is characterized chemically by its extreme reactivity, as indicated by
the emf of its electrochemical half cell (E = 2.85 volts). Fluorine will react with
almost all inorganic substances. If fluorine comes in contact with hydrocarbons, it will
react hypergolically with a high heat of reaction, which is sometimes sufficiently high
that the metal container for the fluorine is ignited. Such metals as low-carbon stainless
steel and Monel, which are used in fluorine systems, develop a protective surface film
when brought in contact with fluorine gas. This surface film prevents the propagation
of the fluorine-metal reaction into the bulk metal.
Fluorine is highly toxic. The fatal concentration range for animals is 200
ppm-hr (Cassutt et al. 1960). That is, for an exposure time of I hour, 200 ppm of
fluorine is fatal; for an exposure time of 15 minutes, 800 ppm is fatal; and for an
exposure time of 4 hours, 50 ppm is fatal. The maximum allowable concentration for
human exposure is usually considered to be approximately 1 ppm-hr. The presence of
fluorine in air may be detected by its sharp, pungent odor for concentrations as low as I
to 3 ppm. Because of its high toxicity, liquid fluorine is not utilized on a large scale.
Methane is the principal component of natural gas. It is a clear, colorless
liquid that boils at 1 atm at 111.7 K (201.1
o
R) and freezes at 88.7 K (159.7R). Liquid
methane has a density approximately one-half of that for liquid nitrogen (methane.
density = 424.1 kg/m
3
= 26.5 lb
m
/ft
3
). Methane forms explosive mixtures with air in
concentrations ranging from 5.8 to 13.3 percent by volume. Liquid methane has been
shipped in large quantities by tanker vessels. The Methane Pioneer made her maiden
voyage on January 28, 1959 with 5000 m
3
of LNG (liquid natural gas). Since that time,
several vessels have been commissioned for LNG transport.
2.13. Hydrogen
Liquid hydrogen has a normal boiling point of 20.3 K (36.5
o
R) and a density
at the normal boiling point of only 70.79 kg/m
3
(4.42 Lb
m
/ft
3
). The density of liquid
hydrogen is about one-fourteenth that of water; thus, liquid hydrogen is one of the
lightest of all liquids. Liquid hydrogen is an odorless, colorless liquid that alone will
43
not support combustion. In combination with oxygen or air, however, hydrogen is
quite flammable. Experimental work (Cassutt et al. 1960) has shown that hydrogen-air
mixtures are explosive in an unconfined space in the range from 18 to 59 percent
hydrogen by volume.
Natural hydrogen is a mixture of two isotopes: ordinary hydrogen (atomic
mass = 1) and deuterium (atomic mass = 2). Hydrogen gas is diatomic and is made up
of molecules of H
2
and HD (hydrogen deuteride) in the ratio of 3200:1. A third
unstable isotope of hydrogen exists, called tritium; however, it is quite rare in nature
because it is radioactive with a short half life.
One of the properties of hydrogen that sets it apart from other substances is
that it can exist in two different molecular forms: ortho-hydrogen and para-hydrogen.
The mixture of these two forms at high temperatures is called normal hydrogen which
is a mixture of 75 percent orthohydrogen and 25 percent para-hydrogen by volume.
The equilibrium (catalyzed) mixture of o-H
2
and p-H
2
at any given temperature is
called equilibrium-hydrogen (e-H
2
). The equilibrium concentration of p-H
2
in e-H
2
as a
function of temperature is given in Table 2.7. At the normal boiling point of hydrogen
(20.3 K or 36.5
o
R), equilibrium hydrogen has a composition of 0.20 percent o-H
2
and
99.80 percent p-H
2
. One could say that it is practically all para-hydrogen .
The distinction between the two forms of hydrogen is the relative spin of the
particles that make up the hydrogen molecule. The hydrogen molecule consists of two
protons and two electrons. The two protons possess spin, which gives rise to angular
momentum of the nucleus, as indicated in Fig. 2.15. When the nuclear spins are in the
same direction, the angular momentum vectors for the two protons are in the same
direction. This form of hydrogen is called ortho-hydrogen. When the nuclear spins are
in opposite directions, the angular-momentum vectors point in opposite directions.
This form of hydrogen is called para-hydrogen.
Table 2.7. Equilibrium concentration of para-hydrogen in equilibrium-hydrogen
Temperature
(K)
Mole fraction
Para-Hydrogen
20.27 0.9980
30 0.9702
40 0.8873
44
50 0.7796
60 0.6681
70 0.5588
80 0.4988
90 0.4403
100 0.3947
120 0.3296
140 0.2980
160 0.2796
180 0.2676
200 0.2597
250 0.2526
300 0.2507
Deuterium can also exist in both ortho and para forms. The nucleus of the
deuterium atom consists of one proton and one neutron, so that the high-temperature
composition (composition of normal deuterium) is two-thirds ortho-deuterium and
one-third para-deuterium. In the case of deuterium, p-D
2
converts to o-D
2
as the
temperature is decreased, in contrast to hydrogen, in which o-H
2
converts to p-H
2
upon
decrease of temperature. The hydrogen deuteride molecule does not have the
symmetry that hydrogen and deuterium possess; therefore, HD exists in only one form.
As one can see from Table 2.7, if hydrogen gas at room temperature is cooled
to the normal boiling point of hydrogen, the o-H
2
concentration decreases from 75 to
0.2 percent; that is, there is a conversion of o-H
2
to p-H
2
as the temperature is
decreased. This changeover is not instantaneous but takes place over a definite period
of time because the change is made through energy exchanges by molecular magnetic
interactions. During the transition, the original o-H
2
molecules drop to a lower molec-
ular-energy level. Thus the changeover involves the release of a quantity of energy
called the heat of conversion. The heat of conversion is related to the change of
momentum of the hydrogen nucleus when it changes direction of spin. This energy
released in the exothermic reaction is greater than the heat of vaporization of liquid
hydrogen, as one can see from the tabulated values of the heat of conversion and heat
of vaporization shown in Appendix D.2.
45
Fig. 2.15. Ortho-hydrogen and para-hydrogen.
When hydrogen is liquefied, the liquid has practically the room-temperature
composition unless some means is used to speed up the conversion process. If the
unconverted normal hydrogen is placed in a storage vessel, the heat of conversion will
be released within the container, and the boil-off of the stored liquid will be
considerably larger than one would determine from the ordinary heat in leak through
the vessel insulation. Note that the heat of conversion at the normal boiling point of
hydrogen is 703.3 kJ/kg (302.4 Btu/lb
m
) and the latent heat of vaporization is 443 kJ
/kg (190.5 Btu/lb
m
). The conversion process evolves enough energy to boil away
approximately 1 percent of the stored liquid per hour, so the reaction would eventually
result in much of the stored liquid being boiled away. For this reason, a catalyst is used
to speed up the conversion so that the energy may be removed during the liquefaction
process before the liquid is placed in the storage vessel.
In the absence of a catalyst, the ortho-para transformation is a second order
reaction (Scott et al. 1934); that is, the rate of change of the o-H
2
mole fraction is given
by
2 0
2 0
dx
C x
dt
(2.29)
where C
2
is the reaction-rate constant, 0.0114 hr
1
for hydrogen at the normal boiling
point, and x
0
= 1 x
p
is the mole fraction of o-H
2
present at any time t. If eqn. (2.29) is
integrated from the initial time when the a-H
2
composition is that in normal hydrogen
(0.750), the mole fraction of o-H
2
at any time is given by
o
2
0.75
x
1 0.75C t

+
(2.30)
46
for liquid hydrogen at the normal boiling point. In the presence of a catalyst that is
well mixed with the hydrogen, the reaction approaches a first order reaction for the gas
phase (Chapin et al. 1960), for which
2 0
1 0
dx
C x
dt
(2.31)
where C
1
is the reaction-rate constant, which depends upon the catalyst used, the
temperature, and the pressure of the gas. If the catalyst is added to the liquid phase, the
reaction is a zero-order one (Cunningham et al. 1960), for which
0
0
dx
C
dt
(2.32)
where C
0
is the reaction-rate constant for the zero-order reaction, which also depends
upon the catalyst, the fluid temperature, and the pressure.
Fig. 2.16. Specific heat c. for p-H
2
n-H
2
and o-H
2
(by permission from Fowler and
Guggenheim 1949).
47
These two forms of hydrogen have different specific heats because of the
different weighting of the energy levels of the two forms, as indicated in Fig. 2.16.
Because of this difference in specific heats, other thermal and transport properties are
also affected. For example, para-hydrogen gas has a higher thermal conductivity than
ortho-hydrogen gas because of the higher specific heat of p-H
2
.
2.14. Helium 4
Helium has two stable isotopes: He
4
, the most common one, and He
3
.
Ordinary helium gas contains about 1.3 10
4
percent He
3
, so when we speak of
helium or liquid helium, we shall be referring to He
4
, unless otherwise stated. Liquid
He
4
has a normal boiling point of 4.214 K (7.58R) and a density at the normal boiling
point of 124.8 kg/m
3
(7.79 lb
m
/ft
3
), or about one-eighth that of water. Liquid helium
has no freezing point at a pressure of 101.3 kPa (1 atm). In fact, liquid helium does not
freeze under its own vapor pressure even if the temperature is reduced to absolute zero.
At absolute zero, liquid helium 4 must be compressed to a pressure of 2529.8 kPa
(24.97 atm or 366.9 psia) before it will freeze, as indicated in Fig. 2.17. Liquid He
4
is
odorless and colorless and has an index of refraction near that of gaseous He
4
(n
r
=
1.02 for liquid He
4
). The heat of vaporization of liquid He
4
at the normal boiling point
is 20.90 kJ/kg (8.98 Btu/lb
m
), which is only
1
/
110
that of water.
Although helium is classified as a rare gas, and it is one of the most difficult
gases to liquefy, its unusual properties have excited so much interest that helium has
been the object of more experimental and theoretical research than any of the other
cryogenic fluids.
48
Fig. 2.17. Phase diagram for helium 4.
We notice from Fig. 2.17 that the phase diagram of He
4
differs in form from
that of any other substance. As mentioned previously, liquid He
4
does not freeze under
its own vapor pressure; therefore, there is no triple point for the solid-liquid-vapor
region of helium as there is for other substances. There are two different liquid phases:
Liquid helium I, the normal liquid; and liquid helium II, the superliquid. The phase
transition curve separating the two liquid phases is called the lambda line, and the
point at which the lambda line intersects the vapor-pressure curve is called the lambda
point, which occurs at a temperature of 2.171 K (3.9l
o
R) and a pressure of 5.073 kPa
(0.050 atm or 0.736 psia).
In Fig. 2.18, we see that the specific heat of liquid helium varies with
temperature in an unusual manner for liquids. At the lambda point, the liquid specific
heat increases to a large value as the temperature is decreased through this point. With
a little imagination, one could say that the form of the specific-heat curve looks
somewhat like the small Greek letter lambda; hence, the name lambda point. It was
first believed that the transition form helium I to helium II was a second-order
transition; however, later work has shown that the transition is more complicated.
49
Because the specific heat of liquid helium has such a different behavior from
that of other liquids, we should expect that the other thermal and transport properties
could also differ in behavior. Strangely enough, the thermal conductivity of helium I
decreases with a decrease in temperature, which is similar to the behavior of the
thermal conductivity of a gas.
Heat transfer in the other form, helium II, is even more spectacular. When a
container of liquid helium I is pumped upon to reduce the pressure above the liquid,
the fluid boils vigorously (depending upon the rate of pumping) as the pressure of the
liquid decreases. During the pumping operation, the temperature of the liquid
decreases as the pressure is decreased and liquid is boiled away. When the temperature
reaches the lambda point and the fluid becomes liquid helium II, all ebullition sud-
denly stops. The liquid becomes clear and quiet, although it is vaporizing quite rapidly
at the surface. The thermal conductivity of liquid helium II is so large that vapor
bubbles do not have time to form within the body of the fluid before the heat is quickly
conducted to the surface of the liquid. Liquid helium I has a thermal conductivity of
approximately 24 mW/m-K (0.014 Btu/hr-ft-
o
F) at 3.3 K (6R), whereas liquid helium
II can have an apparent thermal conductivity as large as 85 kW/m-K (49,000 Btu/hr-ft-
F) much higher than that of pure copper at room temperature! There is some
question about defining a thermal conductivity for liquid helium II because the thermal
conductivity defined in the ordinary manner depends not only on temperature, but also
on the temperature gradient and the size of the container (Keesom et al. 1940). The
reason for this behavior will illustrate the nature of this unique liquid.
One of the unusual properties of liquid helium II is that it exhibits
superliquidity; under certain conditions, it acts as if it had zero viscosity. In explaining
the behavior of1iquid helium II, it has been proposed (Landau 1941) that the liquid be
imagined to be made up of two different fluids: the ordinary fluid and the superfluid,
which possess zero entropy and can move past other fluids and solid boundaries with
zero friction. Using this model, liquid helium II has a composition of normal and
superfluid that varies with temperature, as shown in Table 2.8. At absolute zero, the
liquid composition is 100 percent superfluid; at the lambda point, the liquid
composition is 100 percent normal fluid.
50
Fig. 2.18. Specific heat of saturated liquid helium 4.
The addition of heat to liquid helium II raises the local temperature of the
liquid around the point where heat is being added, which raises the concentration of
normal molecules and lowers the concentration of superfluid ones. The superfluid then
moves to equalize the superfluid concentration throughout the body of the liquid.
Because the superfluid is frictionless, it can move rapidly; therefore, we see that the
high apparent thermal conductivity of helium II is really due to a fast convection
process rather than simply to conduction.
This type of transfer mechanism also explains the so-called "fountain effect"
that is observed in liquid helium II (Allen and Jones 1938; Keller 1969). When heat is
added to the powder in the apparatus shown in Fig. 2.19, the increase in temperature
tends to raise the concentration of normal fluid, and the superfluid rushes in to equalize
the concentration. Normal fluid, because of its viscosity, cannot leave through the
small openings between the fine powder particles very rapidly. The amount of helium
quickly builds up within the tube as a result of this inflow of superfluid, and finally
liquid squirts out the open end of the capillary tube. Fountains as high as 25 cm to 30
cm (10 in. to 12 in.) have been observed.
Table 2.8. Variation of the mass fraction (p
n
/p) of normal fluid in liquid helium II
according to the two-fluid model
51
Temperature (K) Mass Fraction of Superfluid Mass Fraction of Normal Fluid
0
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2.0
2.1
2.171
0.000
0.42710
4
2.52
9.66
0.00295
0.00752
0.0156
0.0292
0.0478
0.0754
0.11
0.17
0.24
0.32
0.43
0.56
0.74
1.00
1.000
0.999957
0.999748
0.999034
0.99705
0.99248
0.9844
0.9708
0.9522
0.9246
0.890
0.83
0.76
0.68
0.57
0.44
0.26
0.00
52
Fig. 2.19. The fountain effect in liquid helium II.
Fig. 2.20. The creeping film in liquid helium II.
Another phenomenon associated with superfluid liquid helium is the Rollin
film, as illustrated in Fig. 2.20. If a beaker is filled with liquid helium II from a larger
container, and the beaker is lifted above the surface of the helium II in the larger
container, a thin film (about 30 nm or 10
6
in. thick) creeps up the inner wall of the
53
beaker, over the side, and falls back into the liquid in the larger container. If the beaker
is emptied and placed, say, halfway into the liquid in the large container, the film of
liquid creeps up the outside of the beaker until the beaker is filled to a level identical
with that of the liquid in the larger container. The velocity of the Rollin film is
dependent upon the film temperature and somewhat on the condition of the surface
over which the film flows.
Second sound is another phenomenon that occurs in liquid helium II. Second
sound is similar to ordinary sound in that it possesses a definite velocity (which is
different from first or ordinary sound velocity), standing waves may be set up, and the
second sound waves may be reflected. Second sound is different from ordinary sound
in that it consists of temperature waves or local oscillations in temperature, rather than
pressure waves or local oscillations in pressure. Second sound was predicted the-
oretically by Tisza in 1939, some 7 years before it was measured experimentally by
Peshkov. The velocity of second sound varies from zero at the lambda point to 239 m/s
(783 ft/sec) near absolute zero.
2.15. Helium 3
Liquid He) is a clear, colorless fluid having a normal boiling point of 3.19 K
(5.75R) and a density at the normal boiling point of 58.9 kgfm) (3.68 lb
m
/ft3). The
heat of vaporization of liquid He) at the normal boiling point is only 8.49 kl/kg (3.65
Btu/lbm)-so small that there was some doubt in the minds of early investigators that
He
3
could be liquefied at atmospheric pressure. As in the case of liquid He
4
, liquid He
3
remains in the liquid state under its own vapor pressure all the way down to absolute
zero. He) must be compressed to 2930.3 kPa (29.3 atm) at 0.32 K (0.576R), whi.ch is
the minimum point on the freezing curve shown in Fig. 2.21, before it will solidify.
54
Fig. 2.21. Phase diagram for helium 3.
The properties of liquid He
3
differ considerably from those of liquid He
4
at
low temperatures because of quantum effects arising from the difference in mass and
the fact that He
3
has an odd number of particles in its nucleus, whereas He
4
has an even
number of nuclear particles. Liquid He
3
undergoes a different type of superfluid
transition at approximately 3.5 mK (0.006R).
Mixtures of liquid He
3
and liquid He
4
also display some peculiar properties.
At temperatures below 0.827 K (1.49R), He) and He
4
mixtures spontaneously
separate into two phases-one that is superfluid (and rich in He
4
) and the other that is
normal (and rich in He
3
). This phase separation phenomena forms the basis for the
helium dilution refrigerator used to achieve temperatures below 1 K.
55
MODULE III
GAS-LIQUEFACTION SYSTEMS
After the introduction to the behavior of materials at low temperatures, we are
now ready to study some of the systems that can produce low temperatures required
for liquefaction. The liquefaction of air in the production of oxygen was the first
engineering application of cryogenics. Even today, the production and sale of liquefied
gases is an important area in cryogenic engineering.
In this chapter, we shall discuss several of the systems used to liquefy the
cryogenic fluids. We shall be concerned with the performance of the various systems,
where performance is specified by the system performance parameters or payoff
functions. The critical components of liquefaction systems will be examined to
complete this study.
INTRODUCTION
3.1. System performance parameters
There are three payoff functions we might use to indicate the performance of
a liquefaction system:
1. Work required per unit mass of gas compressed, - W/m.
2. Work required per unit mass of gas liquefied, - W/m
f

3. Fraction of the total flow of gas that is liquefied, y = m
f
/m.
The last two payoff functions are related to the first one by
(W/m) = (W/m
f
)y (3.1)
In any liquefaction system, we should want to minimize the work
requirements and maximize the fraction of gas that is liquefied.
These payoff functions are different for different gases; therefore, we should
also need another performance parameter that would allow the comparison of the same
system using different fluids. The figure of merit (FOM) for a liquefaction system is
such a parameter. It is defined as the theoretical minimum work requirement divided
by the actual work requirement for the system:
56
FOM =
i i f
f
W W / m
W W/ m

(3.2)
The figure of merit is a number between 0 and 1. It gives a measure of how
closely the actual system approaches the ideal system performance.
There are several performance parameters that apply to the components of
real systems. These include:
1. Compressor and expander adiabatic efficiencies.
2. Compressor and expander mechanical efficiencies.
3. Heat-exchanger effectiveness.
4. Pressure drops through piping, heat exchangers, and so on.
5. Heat transfer to the system from ambient surroundings.
In our initial discussions of system performance, we shall not consider these
component factors but shall return to them when we discuss the major components of
the systems. Thus, we shall first assume that all efficiencies and effectivenesses are
100 percent and that irreversible pressure drops (losses) and heat inleaks are zero.
3.2. The thermodynamically ideal system
In order to have a means of comparison of liquefaction systems through the
figure of merit, we shall first analyze the thermodynamically ideal liquefaction system.
This system is ideal in the thermodynamic sense, but it is not ideal as far as a practical
system is concerned, as we shall see later. The perfect cycle in thermodynamics is the
Carnot cycle. Liquefaction is essentially an open-system process; therefore, for the
ideal liquefaction system, we shall choose the first two processes in the Carnot cycle: a
reversible isothermal compression followed by a reversible isentropic expansion. The
ideal cycle is shown on the temperature-entropy plane in Fig. 3:1 along with a
schematic of the system.
The gas to be liquefied is compressed reversibly and isothermally from
ambient conditions (point 1) to some high pressure (point 2). This high pressure is
selected so that the gas will become saturated liquid upon reversible isentropic
57
expansion through the expander (point}). The final condition at point f is taken at the
same pressure as the initial pressure at point 1.
The pressure attained at the end of the isothermal compression is extremely
high-on the order of 10 GPa or 80 GPa (10
7
psia) for nitrogen. It is highly impractical
to attain this pressure in a liquefaction system, which is the reason it is not an ideal
process for a practical system.
Fig. 3.1. The thermodynamically ideal liquefaction system
In the analysis of each of the liquefaction systems, we shall apply the First
Law of Thermodynamics for steady flow, which may be written in general as
2 2
net net
c c c c outlets inlets
gz gz
Q W m h m h
2g g 2g g
| ` | `

+ + + +


. , . ,

(3.3)
where Q
net
is the net heat transfer to or from the system (heat transferred to the system
is considered positive), W
net
is the net work done on or by the system (work done by
the system is considered positive), and the summation signs imply that we add the
enthalpy terms (h), kinetic-energy terms (
2
/2g
c
where is the fluid velocity and g
c
is the conversion factor in Newton's Second Law, g
c
= 1.00 kg-m/N-s
2
or 32.174 lb
m
-
ft/lb
f
-sec
2
), and potential-energy terms (gz/g
c
, where z is the fluid elevation above a
datum plane and g is the local acceleration due to gravity, 9.806 m/s
2
or 32.174 ft/sec
2
at sea level) for all inlets and outlets of the system. A system might consist of several
58
different streams, which would result in more than one inlet and one outlet. In all our
system analysis, we shall assume that the kinetic- and potential-energy changes are
much smaller than the enthalpy changes (not a bad assumption), and these energy
terms may be neglected. Thus, in our special case, the First Law for steady flow may
be written.
net net
outlets inlets
Q W mh mh

(3.4)
Applying the First Law to the system shown in Fig. 3.1,
Q
R
W
i
= m(h
f
h
1
) = - m(h
1
h
f
) (3.5)
The heat transfer process is reversible and isothermal in the Carnot cycle.
Thus, from the Second Law of Thermodynamics,
Q
R
= mT
1
(s
2
s
1
) = m T
1
(s
2
s
1
) (3.6)
because the process from point 2 to point f is isentropic, s
2
= s
f
, where s is the entropy
of the fluid. Substituting Q
R
from eqn. (3.6) into eqn. (3.5) we may determine the work
requirement for the ideal system:
i i
1 1 2 1 f
f
W W
T (s s ) (h h )
m m

(3.7)
In the ideal system, 100 percent of the gas compressed is liquefied, or m=m
f
,
so that y = 1. Notice that a liquefaction system is a work-absorbing system; therefore,
the net work requirement is negative (work done on the system), and the term - W
i
/m
is a positive number.
Equation (3.7) gives the minimum work requirement to liquefy a gas, so this
is the value which we should try to approach in any practical system. Because we have
set the final pressure at point f equal to the initial pressure at point 1, and point f is on
the saturated-liquid curve, the ideal work requirement depends only on the pressure
and temperature at point I and the type of gas liquefied. Ordinarily, we take point I at
ambient conditions. Table 3.1 lists ideal-work requirements for some common gases,
with point I taken at 10 1.3 kPa (14.7 psia) and 300 K (80F).
59
Example 3.1. Determine the ideal-work requirement for the liquefaction of
nitrogen, beginning at 101.3 kPa (14.7 psia) and 300 K (80F). From the T-s diagram
in Appendix F, we find the following property values:
h
i
= 462 J/g at 101.3 kPa (1 atm) and 300 K
h
f
= 29 J/g at 101.3 kPa (1 atm) and saturated liquid
Table 3.1. Ideal-work requirements for liquefaction of gases beginning at 300 K (80
o
F)
and 101.3 kPa (14.7 psia)
Normal Boiling Point
Ideal Work of Liquefaction,
W
i
/m
f
Gas K
o
R kJ/kg Btu/lb
m
Helium-3 3.19 5.74 8.178 3.516
Helium-4 4.21 7.58 6.819 2.931
Hydrogen, H
2
20.27 36.5 12.019 5.167
Neon, Ne 27.09 48.8 1.335 574
Nitrogen, N
2
77.36 139.2 768.1 330.2
Air 78.8 142 738.9 317.7
Carbon monoxide, CO 81.6 146.9 768.6 330.4
Argon, A 87.28 157.1 478.6 205.7
Oxygen, O
2
90.18 162.3 635.6 273.3
Methane, CH
4
111.7 201.1 1.091 469
Ethane, C
2
H
6
184.5 332.1 353.1 151.8
Propane, C
3
H
6
231.1 416.0 140.4 60.4
Ammonia, NH
3
239.8 431.6 359.1 154.4
s
1
= 4.42 J/g-K at 101.3 kPa (I atm) and 300 K
s
f
= 0.42 J/g-K at 101.3 kPa (I atm) and saturated liquid
60
Substituting these values in eqn. (3.7) results in
i
W / m
&
&
= (300) (4.42 0.42) (462 29) = 767 J/g (330 Btu/Ib
m
)
PRODUCTION OF LOW TEMPERATURES
3.3. Joule- Thomson effect
Most of the practical liquefaction systems utilize an expansion valve or Joule-
Thomson valve to produce low temperatures. If we apply the First Law for steady flow
to the expansion valve, for zero heat transfer (insulated valve) and zero work transfer
and for negligible kinetic- and potential-energy changes, we find that h
1
= h
2
. Although
the flow within the valve is irreversible and is not an isenthalpic process, the inlet and
outlet states do lie on the same enthalpy curve. We could plot a series of points of
outlet conditions for given inlet conditions and obtain lines of constant enthalpy. For a
real gas, such a plot is shown in Fig. 3.2. We note that there is a region in which an
expansion through the valve (decrease in pressure) produces an increase in
temperature, while in another region the expansion results in a decrease in temperature.
Obviously, we should want to operate the expansion valve in a liquefaction system in
the region where a net decrease in temperature results. The curve that separates these
two regions is called the inversion curve.
The effect of change in temperature for an isenthalpic change in pressure is
represented by the Joule- Thomson coefficient
JT
, defined by

JT
=
h
T
p
| `

. ,
(3.8)
where the derivative is interpreted as the change in temperature due to a
change in pressure at constant enthalpy. Note that the Joule-Thomson coefficient is the
slope of the isenthalpic lines in Fig. 3.2. The Joule-Thomson coefficient is zero along
the inversion curve because a point on the inversion curve is one at which the slope of
the isenthalpic line is zero. For a temperature increase during expansion, the Joule-
Thomson coefficient is negative; for a temperature decrease, the Joule-Thomson
coefficient is positive.
61
From calculus, it can be shown that

JT
=
p
h T
T T h
p h p
| ` | `
| `



. ,
. , . ,
(3.9)
Fig. 3.2. Isenthalpic expansion of a real gas.
From basic thermodynamics (Van Wylen and Sonntag 1976), it can be shown
that
p
p p
T
h h
dh dT dp c dT T dp
T p T
]
| `
| ` | `
+ +
]


. , . ,
. , ]
]
(3.10)
where is the specific volume of the material. By comparison of the
coefficients of dT and dp in eqn. (3.10), we see that
p
p
h
c
T

| `

. ,
and
p
T
h
T
p T
| `
| `



. ,
. ,
(3.11)
By combining eqns. (3.9) and (3.11) the Joule-Thomson coefficient may be
expressed in terms of other thermodynamic properties as
JT
p p
1
T
c T
]

| `

]

. ,
]
]
(3.12)
For an ideal gas, = RT/p, and
62
p
R
T p T

| `

. ,
Therefore, from eqn. (3.12), for an ideal gas
JT
p
1
T 0
c T
]
| `

]
. ,
]
An ideal gas would not experience a temperature change upon expansion
through an expansion valve. To the gratification of cryogenic engineers, gases are
imperfect at low enough temperatures and high enough pressures.
We have seen that an ideal gas always has a zero Joule-Thomson coefficient;
therefore, a positive or negative Joule-Thomson coefficient must arise from the
departure of real gases from the ideal-gas behavior. Enthalpy is defined as
h = u + p (3.13)
where u is the internal energy of the substance. Making this wbstitution in
eqn. (3.9),
( )
JT
p
T T
p
1 u
c p p

] | `
+
' ' ]

. , ]

(3.14)
The first term in eqn. (3.14) represents a departure from Joule's law, which
states that the internal energy of an ideal gas is a function of temperature alone. If
u = u(T) = c

T, for example, then this term is zero. This term is always negative for
real gases; thus, it contributes to the production of a temperature decrease (positive

JT
). As molecules are moved farther apart in reducing the pressure during the
expansion process, their microscopic potential energy is increased. No external work
or heat is added; therefore, this increase in microscopic potential energy must be offset
by a decrease in microscopic kinetic energy. Temperature is one measure of
microscopic kinetic energy, and any decrease in microscopic kinetic energy results in a
decrease in temperature.
On the other hand, the second term in eqn. (3.14) may be either positive,
negative, or zero. The second term represents a departure from Boyle's law, which
63
states that the product of pressure and volume for an ideal gas is a function of
temperature alone. If p = f(T) = RT, for example, then this term is always zero. A
sketch of p as a function of pressure is shown in Fig. 3.3. At low pressures and
temperatures near the saturated-vapor condition, gases are more compressible than
Boyle's law predicts because attractive forces are in action that try to condense the gas.
This means that the second term in eqn. (3.14) is negative and contributes to the
production ofa temperature decrease. This is the case for gases at room temperature,
with the exception of hydrogen, helium, and neon. At high pressures, the molecules are
squeezed close together and repulsive forces are brought into action; thus, the gas is
less compressible in this region than Boyle's law predicts. This behavior makes the
second term in eqn. (3.14) positive, which contributes to the production of a
temperature increase upon expansion. Whether the Joule-Thomson coefficient is
positive, negative, or zero for a real gas depends upon the relative magnitude of the
two terms we have been discussing.
Fig. 3.3. Variation of the product pI' with pressure and temperature for a real
gas. For an ideal gas, each isothermal curve would be a horizontal straight line.
One equation of state for gases that illustrates the behavior of real gases is the
van der Waals equation of state,
( )
2
a
p b RT
| `
+

. ,
(3.15)
where a is a measure of the intermolecular force and b is a measure of the
finite size of the molecules. For an ideal gas, a = b = 0 because an ideal gas has no
64
intermolecular forces, and the molecules are considered to be mass point with no
volume. A van der Waals gas has molecules that are considered to be weakly
attracting rigid spheres.
3.4. Adiabatic expansion
The second method of producing low temperatures is the adiabatic expansion
of the gas through a work-producing device, such as an expansion engine. In the ideal
case, the expansion would be reversible and adiabatic, and therefore isentropic. In this
case, we can define the isentropic expansion coefficient IL" which expresses the
temperature change due to a pressure change at constant entropy:
s
s
T
p
| `

. ,
(3.18)
The isentropic expansion coefficient can be related to other properties of a gas
in a manner similar to the ones we used with the Joule-Thomson coefficient:
s
p p p
s T
T T s T
p s p c T
| ` | `
| ` | `
+


. , . ,
. , . ,
(3.19)
The second factor in eqn. (3.19) is the volumetric coefficient of thermal
expansion multiplied by the specific volume, or , so
s
is positive (a temperature
decrease results from a pressure decrease) when the volumetric coefficient of
expansion is positive. This is the case for all gases (although some substances, such as
liquid water between 0
o
C and 4C, have negative coefficients of expansion). For an
ideal gas,
( )
p
/ T R / p / T
, so
s p
/ c
for an ideal gas (3.20)
For a van der Waals gas, one can show that
( )
( ) ( )
s
2
p
1 b /
c 1 2a / RT 1 b /


]

]
, van der Waals gas (3.21)
which is positive because > b.
We can make the observation that, for a gas, an isentropic expansion through
an expander always results in a temperature decrease, whereas an expansion through
an expansion valve mayor may not result in a temperature decrease. During the
65
isentropic process, energy is removed from the gas as external work; this method of
low-temperature production is sometimes called the external-work method. Expansion
through an expansion valve does not remove energy from the gas but moves the
molecules farther apart under the influence of intermolecular forces. This method is
called the internal-work method.
Table 3.2. Maximum inversion temperature
Maximum Inversion Temperature
Gas K
o
R
Helium-4 45 81
Hydrogen 205. 369 I
Ncon 250 450
Nitrogen 621 1118
Air 603 1085
Carbon monoxide 652 1174
Argon 794 1429
Oxygen 761 1370
Methane 939 1690
Carbon dioxide 1500 2700
Ammonia 1994 3590
It appears that expanding the gas through an expander would always be the
most effective means of lowering the temperature of a gas. This is the case as far as the
thermodynamics of the two processes is concerned. Between any two given pressures,
an isentropic expansion will always result in a lower final temperature than will an
isenthalpic expansion from the same temperature. The practical problems associated
with the expansion of a two phase mixture (liquid and vapor) in an expander make the
use of an expansion valve necessary in all liquefaction systems.
66
LIQUEFACTION SYSTEMS FOR. GASES OTHER THAN NEON,
HYDROGEN, AND HELIUM
3.5. Simple Linde-Hampson system
Historically, the Linde-Hampson system was the second used to liquefy gases
(the cascade system was the first), although it is the simplest of all the liquefaction
systems. A schematic of the Linde-Hampson system is shown in Fig. 3.4, and the cycle
is shown on the T-s plane in Fig. 3.5.
Fig3.4:Linde-hampson liquefaction system
In order to analyze the performance of the system, let us assume ideal
conditions: no irreversible pressure drops (except for the expansion valve), no heat
inleaks from ambient, and 100 percent effective heat exchanger. The gas is first
compressed from ambient conditions at point I reversibly and isothermally to point 2.
In a real system, process 1 to 2 would actually be two processes: an irreversible
adiabatic or polytropic compression followed by an after cooling to lower the gas
temperature back to within a few degrees of ambient temperature. The gas next passes
through a constant-pressure heat exchanger (ideally) in which it exchanges energy with
the outgoing low-pressure stream to point 3. From point 3 to point 4, the gas expands
through an expansion valve to p
4
= p
1
. At point 4, some of the gas stream is in the
liquid state and is withdrawn at condition g (saturated-liquid condition), and the rest of
the gas leaves the liquid receiver at condition g (saturated-vapor condition). This cold
gas is finally warmed to the initial temperature by absorbing energy at constant
pressure (ideally) from the incoming high-pressure stream.
67
Fig. 3.5. The Linde-Hampson cycle. The state-points refer to the numbered points in
Fig. 3.4.
Applying Law for steady flow to the combined heat exchanger, expansion
valve, and liquid receiver, we obtain
( )
f 1 f f 2
0 m m h m h mh + & & & &
because there is no work or heat transfer to or from the surroundings for these
components. Solving for the fraction of the gas flow that is liquefied,
f 1 2
1 f
m h h
y
m h h

&
&
(3.22)
The fraction of gas liquefied (the liquid yield) thus depends upon (1) the
pressure and temperature at ambient conditions (point I), which fix h
1
and h
f
and (2)
the pressure after the isothermal compression, which determines h
2
because the
temperature at state point 2 is specified by the temperature at point I.
68
We do not have much freedom in choosing or varying ambient conditions;
therefore, points I and f are practically prescribed. We are at liberty to vary the
performance of the system by varying the pressure at point 2 only. What would be the
best pressure to pick for p
2
from a thermodynamic viewpoint? In order to maximize the
liquid yield, the best or optimum pressure at point 2 would be the one that minimizes
h
2
according to eqn. (3.22). Mathematically, for minimum h
2
we must have
1
T T
h
0
p

| `

. ,
(3.23)
Referring to eqns. (3.9) and (3.11), we see that this is equivalent to saying
( )
1
1
JFT p
T T
T T
h
0 c
p

| `

. ,
(3.24)
or that the pressure which minimizes h
2
is the pressure for which
JT
= 0 for a
temperature T
1
. For optimum performance (maximum liquid yield y) of the Linde-
Hampson system, state 2 should lie on the inversion curve. For air at 300 K (80F), the
corresponding pressure is approximately 40 MPa (5880 psia); however, actual systems
commonly utilize pressures on the order of 20 MPa.
The simple Linde-Hampson system shown in Fig. 3.5 would not work for
gases such as neon, hydrogen, and helium for two reasons. First, the system would
never get started because the maximum inversion temperature for these gases is below
room temperature. Referring to the T-s diagram for helium, for example, we see that in
first starting the system, there would be no heat exchange and points 2 and 3 would
coincide. Expansion through the expansion valve at ambient temperature from point 3
to point 4 would result in an increase in temperature so that as the operation
progressed, the gas entering the heat exchanger would be continually warmed rather
than cooled. Therefore, we could never produce low temperatures. This is illustrated in
Fig. 3.6.
In the second place, from eqn. (3.22) we see that y is negative as long as h, is
smaller than h
2
, which is the case for helium, hydrogen, and neon at room temperature.
This means that, even if we could attain low temperatures with the Linde-Hampson
69
system, the expansion through the expansion valve at low temperatures would pass
completely into the vapor region, and no gas would be liquefied. Beginning at ambient
conditions, not enough energy can be exchanged in the heat exchanger to lower the gas
temperature to the point at which liquid could be obtained after expansion. This is
illustrated in Fig. 3.7. We see now why we have divided the liquefaction systems into
two categories.
The work requirement for the Linde-Hampson system may be determined by
applying the First Law for steady flow to the isothermal compressor shown in Fig. 3.4:
( )
R 2 1
Q W m h h
& &
&
(3.25)
Fig. 3.6. Start-up ofa Linde-Hampson system (no precooling) using helium or
hydrogen as the working fluid. The fluid expands through the expansion valve from 2
to 3 and increases in temperature, since this condition is to the right of the inversion
curve.
70
Fig. 3.7. Even if we could get the simple Linde-Hampson system using helium or
hydrogen started in the right direction, it is still physically impossible to transfer
enough energy in the heat exchanger to produce liquid.
Substituting for the reversible isothermal heat transfer from eqn. (3.6),
( ) ( )
1 1 2 1 2
W/ m T s s h h
&
&
(3.26)
The work requirement per unit mass liquefied is
( ) ( )
1 f
1 1 2 1 2
f y 1 2
h h W W
T s s h h
m m h h
| `
]

]

. ,
& &
& &
(3.27)
Performance figures for various gases are shown in Table 3.3.
3.6. Precooled Linde-Hampson system
A plot of liquid yield y as a function of the temperature at the entrance of the
heat exchanger (point 2) for the simple Linde-Hampson system is shown in Fig. 3.8. It
is apparent that the performance of a Linde-Hampson system could be improved if it
71
were modified so that the gas entered the heat exchanger at a temperature lower than
ambient temperature. Such a modified system is shown in Fig. 3.9, and the cycle is
shown on the T -s plane in Fig. 3.10. A separate refrigeration system using a fluid such
as carbon dioxide, ammonia, or a Freon compound is used to cool the main gas stream.
The critical temperature of the auxiliary refrigerant must be above ambient temperature
in order that the refrigerant can be condensed by exchanging heat with the atmosphere
or with cooling water at ambient temperature.
Table 3.3. Performance of the Linde-Hampson system using different fluids,
p
1
= 101.3, kPa (14.7 psia); p
2
= 20.265 MPa (200 atm); T
1
= T
2
= 300 K (80
o
F); heat
exchanger effectiveness = 100 percent; compressor overall efficiency = 100 percent.
Fluid
Normal Boiling
Point
Liquid
Yield y
=
f
m / m & &
Work per Unit
Mass Compressed
Work per Unit
Mass Liquefied
Figure
of Merit
FOM =
i
W / W
& &
K
o
R kJ/kg Btu/Ibm kJ/kg Btu/Ibm
N2 77.36 139.3 0.0708 472.5 203.2 6673 2869 0.1151
Air 78.8 142 0.0808 454.1 195.2 5621 2416 0.1313
CO 81.6 146.9 0.0871 468.9 201.6 5381 2313 0.1428
A 87.28 157.1 0.1183 325.3 139.8 2750 1182 0.1741
O2 90.18 162.3 0.1065 405.0 174.1 3804 1636 0.1671
CH4 111.7 201.1 0.1977 782.4 336.4 3957 1701 0.2758
C2H6 184.5 332.1 0.5257 320.9 138.0 611 262 0.5882
C3H8 231.1 416.0 0.6769 159.0 68.4 235.0 101.0 0.5976
NH3 239.8 431.6 0.8079 363.1 156.1 449.4 193.2 0.7991
72
Fig. 3.8 Liquid yield versus compressor temperature for a Linde-Hampson system
using nitrogen as the working fluid. The heat exchange are 100 percent effective.
Fig. 3.9 Precooled Linde-Hampson system.
73
Fig. 3.10 Precooled Linde-Hampson cycle. The state-points refer to the numbered
points in Fig. 3.9.
For a 100 percent effective heat exchanger, the temperature at points 3 and 6
(see Fig. 3.10) are the same. From a consideration of the Second Law of
Thermodynamics, T) and T
6
cannot be lower than the boiling point of the auxiliary
refrigerant at point d; otherwise, we would be transferring heat "uphill"-from a low to a
high temperature-without expending any work. These factors place restrictions on the
performance of the precooled Linde-Hampson system.
Applying the First Law for steady flow to the combined three-channel heat
exchanger, primary-gas heat exchanger, liquid receiver and expansion valve, and
auxiliary refrigerant expansion valve, for no work or heat transfers from the
surroundings to these components,
( )
f 1 r a f f 2 r d
0 m m h m h m h mh m h + + & & & & & &
(3.28)
If we define the refrigerant mass flow-rate ratio r as
r
m
r
m

&
&
(3.29)
74
where
r
m&
, is the mass flow rate of the auxiliary refrigerant, then eqn. (3.28) may be
written as follows, solving for the liquid yield
f
y m / m & &
:
a c 1 2
i f 1 f
h h h h
y r
h h h h

+

(3.30)
The first term on the right-hand side of eqn. (3.30) represents the liquid yield
for the simple Linde-Hampson system operating under the same conditions as the
precooled system, given by eqn. (3.22). The second term represents the improvement
in liquid yield that is obtained through the use of precooling. As mentioned before, the
temperature of the auxiliary refrigerant does limit the liquid yield. Otherwise, with a
suitable value of the refrigerant flow-rate ratio r, eqn. (3.30) could yield a value of 100
percent for the liquid yield y. The maximum liquid yield for the precooled system is
6 3
max
6 f
h h
y
h h

(3.31 )
where h
3
and h
6
are taken at the temperature of the boiling refrigerant at point d.
Another limiting factor is that if the refrigerant flow-rate ratio were too arge,
the liquid at point d would not be completely vaporized, and liquid would enter the
refrigerant compressor-not a very desirable situation. The variation of the liquid yield
and the limiting liquid yield is shown in Fig. 3.11 for nitrogen gas as the working fluid
and Freon-12 as the refrigerant, operating from 101.3 kPa (1 atm) and 21C (70F) to
585 kPa (5.77 atm).
75
Fig. 3.11. Liquid yield versus refrigerant flow-rate ratio for the precooled Linde-
Hampson system using nitrogen as the working fluid. The curves terminate at the value
of r that would result in liquid entering the auxiliary compressor.
If the main compressor is reversible and isothermal and the auxiliary
compressor is reversible and adiabatic, the work requirement per unit mass of primary
gas compressed is
( ) ( ) ( )
1 1 2 1 2 b a
W/ m T s s h h r h h +
&
&
(3.32)
The last term represents the additional work requirement for the auxiliary
compressor; therefore, the total work requirement for the precooled Linde-Hampson
system is greater than that for the simple system. The last term is usually on the order
of 10 percent of the total work. The increase in liquid yield more than offsets the
additional work requirement, however, so that the work requirement per unit mass of
gas liquefied is actually less for the precooled system than for the simple system, as
indicated in Fig. 3.12.
76
Fig. 3.12. Work required to liquefy a unit mass of nitrogen in a precooled Linde-
Hampson system. The dashed curve is the locus of limiting values of r to ensure that
nQ liquid enters the auxiliary compressor.
Example 3.2. Determine the liquid yield, the work per unit mass compressed,
and the work per unit mass liquefied for the simple Linde-Hampson system and for the
precooled Linde-Hampson system using nitrogen as the working fluid and Freon-I 2 as
the refrigerant. The nitrogen portion of the system operates between 101.3 kPa (14.7
psi a) and 300 K (80F) and 20.3 MPa (200 atm or 2940 psia) at point 2. The state
points for the refrigerant portion of the system are as follows:
h
a
= 207.94 kl/kg (89.40 Btu/Ibm) at 101.3 kPa (14.7 psia) and 300 K
h
b
= 250.20 kl/kg (107.57 Btu/Ibm) at 681.7 kPa (98.9 psia) and 99.7C
h
c
= 61.23 kl/kg (26.37 Btu/Ibm) at 300 K (80F) and saturated liquid
Point d is at 101.3 kPa (14.7 psia) and 29.8C (21.6F) in the two-phase
region. The refrigerant flow-rate ratio is r = 0.10
77
Simple Linde-Hampson system. From the T-s diagram for nitrogen. we find
the following property values:
h
1
= 462 J/g at 101.3 kPa (1 atm) and 300 K
h
2
= 432 J/g at 20.3 MPa (200 atm) and 300 K
h
f
= 29 J/g at 101.3 kPa (: atm) and saturated liquid
s
1
= 4.42 J/g-K at 101.3 kPa (I atm) and 300 K
s
2
= 2.74 J/g-K at 20.3 MPa (200 atm) and 300 K
s
f
= 0.42 J/g-K at 101.3 kPa (I atm) and saturated liquid
The liquid yield, according to eqn. (3.22), is
1 2
1 f
h h 462 432
y 0.0693
h h 462 29



The work requirement per unit mass compressed is, from eqn. (3.26),
( ) ( )
1 1 2 1 2
W/ m T s s h h
&
&
( ) ( ) ( ) W/ m 300 4.42 2.74 462 432
&
&

= 504 30 = 474 J/g (204 Btu/Ib
m
)
The work per unit mass liquefied is
( )
m
f
W W/ m 474
6840 J / g 2940 Btu / Ib
m y 0.0693


& &
&
&
From Example 3.1, the ideal work requirement was 767 J/g; therefore, the
figure of merit for this system is
i f
f
W / m 767
FOM 0.1121
W/ m 6840

&
&
&
&
Precooled Linde-Hampson system. By using the previously detem1ined prop-
erty values, from eqn. (3.30) the liquid yield for the precooled system is
( )
462 432 207.94 61.23
y 0.10
462 29 462 29

| `
+


. ,
78
= 0.0693 + 0.0339 = 0.1032
By the addition of precooling, there is an improvement of the liquid yield of
almost 50 percent. The total work requirement is, from eqn. (3.32),
W/ m
&
& = (300) (4.42 2.74) (462 432) + (0.10) (250.20 207.94)
= 504 30 + 4.2 = 470 J/g (202 Btu/Ib
m
)
The work requirement per unit mass liquefied is
( )
m
f
W 470
4554 J / g 1958 Btu / Ib
m 0.1032

&
&
For the precooled Linde-Hampson system, the figure of merit is
767
FOM 0.1684
4554

This value is about 50 percent better than the value for the simple Linde-
Hampson system previously worked out.
3.7. Linde dual-pressure system
The simple Linde-Hampson system can be modified in another way to reduce
the total work required, although this modification reduces the liquid yield somewhat.
Because only a small portion of the gas that is compressed is liquefied in the simple
system, we could modify it so that not all the gas is expanded to the lowest pressure
but some is expanded to an intermediate pressure. The work requirement for an ideal
isothermal compressor and an ideal gas would be RT
1
ln (p
2
/p
1
), so a reduction in the
compression pressure ratio would decrease the work requirement. This is
accomplished in the Linde-dual-pressure system shown schematically in Fig. 3.13. The
cycle is shown on the T-s plane in Fig. 3.14.
79
Fig. 3.13. Linde dual-pressure system.
Fig. 3.14. Linde dual-pressure cycle. The state-points refer to the numbered points in
Fig. 3.13.
80
The gas is first compressed to an intermediate pressure and then to the high
pressure of the cycle after a return system has been added. The high pressure gas is
passed through a three-channel heat exchanger and expanded to the intermediate
pressure at point 5, where some of the gas is liquefied. The saturated liquid and vapor
are separated in a liquid receiver, and the vapor is returned to the second compressor
through the three-channel heat exchanger while the liquid is expanded further to the
low pressure of the cycle.
Applying the First Law for steady flow to the heat exchanger, the two liquid
receivers, and the two expansion valves, we can determine the liquid yield for the
Linde dual-pressure system:
1 3 1 2
1 f 1 f
h h h h
y i
h h h h



(3.33)
where i is the intermediate-pressure-stream flow-rate ratio,
i
i m / m & &
(3.34)
The quantity
i
m&
is the mass flow rate of the intermediate pressure stream at
point 8, and m& is the total mass flow rate through the high-pressure compressor. The
second term represents the reduction in the liquid yield below that of the simple system
because of splitting the flow at the intermediate-pressure-liquid receiver.
Applying the First Law for steady flow to the two compressors, we find that
the work requirement per unit mass of gas compressed in the high pressure compressor
is
( ) ( )
1 1 3 1 3
W/ m T s s h h ]
]
&
&
( ) ( )
1 1 2 1 2
i T s s h h ]
]
(3.35)
From eqn. (3.35) we see that the work requirement is reduced below that of
the simple system by the amount given by the second bracketed term. Practical
liquefaction systems usually operate with i on the order of 0.7 to 0.8, so that the
reduction in work requirement more than offsets the reduction in liquid yield;
therefore, as for the precooled system, the work requirement to produce a unit mass of
liquid is less for the dual pressure system than for the simple system. Work
requirements for air are illustrated in Fig. 3.15.
81
Example 3.3. Determine the liquid yield, work requirement per unit mass
compressed in the high-pressure compressor, and work requirement per unit mass
liquefied for a Linde dual-pressure system operating with nitrogen as the working fluid
between 101.3 kPa (I atm) and 300 K (80F) and 20.3 MPa (200 atm). The
intermediate pressure is 5.07 MPa (50 atm), and the intermediate-pressure flow-rate
ratio is 0.80. From the T-s diagram for nitrogen, we find the following property values:
h
1
= 462 Jig at 101.3 kPa (I atm) and 300 K (80F)
h
2
= 452 Jig at 5.07 MPa (50 atm) and 300 K
h
3
= 432 Jig at 20.3 MPa (200 atm) and 300 K
h
f
= 29 Jig at 101.2 kPa (I atm) and saturated liquid
s
1
= 4.42 J/g-K at 101.3 kPa (I atm) and 300 K
s
2
= 3.23 J/g-K at 5.07 MPa (50 atm) and 300 K
s
3
= 2.74 J/g-K at 20.3 MPa (200 atm) and 300 K
From eqn. (3.33), we find the liquid yield
( )
462 432 462 452
y 0.80
462 29 462 29

| `



. ,
y = 0.0693 0.0185 = 0.0508
This value may be compared with the liquid yield of 0.0693 obtained for the
simple system in Example 3.2. The work requirement per unit mass compressed is
found from eqn. (3.35),
W/ m
&
& = [(300) (4.42 274) (462 432)]
(0.80) [(300) (4.42 3.23) (462 452)]
= 474 277.6 = 196.4 J/g (84.4 Btu/Ib
m
)
The work required to liquefy a. unit mass of gas is
f
W 196.4
m 0.0508

&
&
= 3866 J/g (1662 Btu/Ib
m
)
82
Fig. 3.15. Work required to liquefy a unit mass of air in the Linde dual-pressure
system.
The figure of merit for the Linde dual-pressure system is:
767
FOM 0.1984
3866

3.8. Cascade system
The cascade system is an extension of the precooled system, in which the
precooled system is precooled by other refrigeration systems. The cascade system was
the first liquefaction system used to produce liquid air. A cascade system suggested by
Keesom (1933) is shown in Fig. 3.16. This system uses ammonia to liquefy ethylene at
1925 kPa (19 atm), which is used in turn to liquefy methane at 2530 kPa (25 atm). The
methane is used finally to liquefy nitrogen gas at 1885 kPa (18.6 atm). Another pos-
sible cascade combination is the Freon compounds F-22, F-13, and F-14 used to
liquefy nitrogen, air, or oxygen. Ball (1954) has also described a partial cascade
system that uses only two Freon compounds.
83
Fig 3.16:The cascade system
From a thermodynamic point of view, the cascade system is very desirable for
liquefaction because it approaches the ideal reversible system more closely than any
other discussed thus far. As one can see from the complexity of the system shown in
Fig. 3.16, nothing is gained by trying to write down a general equation for the liquid
yield as we have done for the other systems. The fact that the irreversible expansions
through the expansion valves occur across smaller pressure ranges than in the other
systems would lead us to believe, however, that the cascade system would have
improved performance over the other systems mentioned. The lower pressure
requirements are another point in favor of this system. On the other hand, the cascade
system does have a serious practical disadvantage. Each loop of the system must be
completely leak proof in order to prevent the fluids from getting into the wrong place.
84
This imposes a maintenance hardship to make sure that leaks do not occur and intro-
duces a safety hazard when leaks do occur.
3.9. Claude system
The expansion through an expansion valve is an irreversible process,
thermodynamically speaking. Thus if we wish to approach closer to the ideal
performance, we must seek a better process to produce low temperatures. In the
Claude system, shown in Fig. 3.17, energy is removed from the gas stream by allowing
it to do some work in an expansion engine or expander. The Claude cycle is shown on
the T -s plane in Fig. 3.18. If the expansion engine is reversible and adiabatic (which
we shall assume to be true for this analysis), the expansion process is isentropic, and a
much lower temperature is attained than for an isenthalpic expansion, as we saw in
Sec. 3.4.
In the Claude system, the gas is first compressed to pressures on the order of
4 MPa (40 atm or 590 psia) and then passed through the first heat exchanger. Between
60 and 80 percent of the gas is then diverted from the mainstream, expanded through
an expander, and reunited with the return stream below the second heat exchanger. The
stream to be liquefied continues through the second and third heat exchangers and is
finally expanded through an expansion valve to the liquid receiver. The cold vapor
from the liquid receiver is returned through the heat exchangers to cool the incoming
gas.
Fig 3.17: the claude system
85
Fig. 3.18. The Claude cycle. Statepoints refer to the numbered points in Fig. 3.17. The
heat exchangers are 100 percent effective, and the expander has 100 percent adiabatic
efficiency.
An expansion valve is still necessary in the Claude system because much
liquid cannot be tolerated in the expander in an actual system. The liquid has a much
lower compressibility than the gas; therefore, if liquid were formed in the cy m
expansion engine positive displacement type), high momentary stresses would result.
Some rotary turbine expanders (axial-flow type) have been developed that can tolerate
as much as 15 percent liquid by weight without damage to the turbine blades.
In some Claude systems, the energy output of the expander is used to help
compress the gas to be liquefied. In most small-scale systems, the energy is dissipated
in a brake or in an external air blower. Whether the energy is wasted or not does not
affect the liquid yield; however, it does increase the compression work requirement
when the expander work is not used.
Applying the First Law for steady flow to the heat exchangers, the expansion
valve, and the liquid receiver as a unit, for no external heat transfer,
( )
f 1 f f e e 2 e 3
0 m m h m h m h mh m h + + & & & & & &
(3.36)
86
If we define the fraction of the total flow that passes through the expander as
x, or
e
x m / m & &
(3.37)
then the liquid yield may be obtained from eqn. (3.36) as
3 e f 1 2
1 f 1 f
h h m h h
y x
m h h h h

+

&
&
(3.38)
Again, we see that the second term represents the improvement in per-
formance over the simple Linde-Hampson system. Of course, x + y must be less than
unity in eqn. (3.38).
The work requirement per unit mass compressed is exactly the same as that
for the Linde-Hampson system if the expander work is not utilized to help in the
compression. This value is given by eqn. (3.26). On the other hand, if the expander
work is used to aid in compression, then the net work requirement is given by
c c
W/ m W / m W / m
& & &
& & &
Applying the First Law for steady flow to the expander, we obtain the
expander work expression,
( )
e e 3 e
W m h h
&
&
(3.39)
If the expander work is utilized to aid in compression, the net work is given
by
( ) ( ) ( )
1 1 2 1 2 3 e
W/ m T s s h h x h h ]
]
&
&
(3.40)
The last term in eqn. (3.40) is the reduction in energy requirements due to the
utilization of the expander work output.
For each value of high pressure (p
2
or p
3
) and each value of expansion engine
flow-rate ratio x, one can find by a series of calculations using the thermodynamic
charts for the fluid used in the system that there is a finite temperature at point 3 (the
condition at the inlet of the expander) that will yield the smallest work requirement per
unit mass liquefied. Energy requirements under this condition of astutely chosen T
J
are
87
shown in Fig. 3.19. We see that, for a given pressure level, there is also a value of x
that makes the work requirement a minimum. As the high pressure is increased, the
minimum work requirement per unit mass liquefied decreases.
Example 3.4. Determine the liquid yield, the total work per unit mass of gas
compressed, and the work to liquefy a unit mass of gas for the Claude system using
nitrogen as the working fluid. The system operates between 101.3 kPa (I atm) and 300
K (80F) and 5.066 MPa (50 atm or 735 psia). The expander flow rate ratio is 0.60,
and the expander work is utilized to aid in compression of the gas. The condition of the
gas at the inlet of the expander is 270 K (26F) and 5.066 MPa (50 atm).
From the T-s diagram for nitrogen, we find the following property values:
h
1
= 462 Jig at 101.3 kPa (I atm) and 300 K (80F)
h
2
= 452 Jig at 5.066 MPa (50 atm) and 300 K
h
3
= 418 Jig at 5.066 MPa (50 atm) and 270 K (26F)
h
e
= 238 Jig at 101.3 kPa (I atm) and s
e
= s
3
(T
e
= 86.1 K = - 305F)
h
f
= 29 Jig at 101.3 kPa and saturated liquid
s
1
= 4.42 J/g-K at 101.3 kPa (I atm) and 300 K
s
2
= 3.23 J/g-K at 5.066 MPa (50 atm) and 300 K
s
3
= 5, = 3.11 J/g-K at 5.066 MPa (50 atm) and 270 K (26F)
88
Fig. 3.19. Work required to liquefy a unit mass of air in the Claude system
(Lenz 1929).
From eqn. (3.38), we can determine the liquid yield:
( )
462 452 418 238
y 0.60
462 29 462 29

| `
+


. ,
= 0.0231 + 0.2494 = 0.2725
Since 0.40 kg/kg compressed was passed into the liquid receiver, we see that
almost 70 percent of this flow was liquefied. The total work requirement is, from eqn.
(3.40),
W/ m
&
& = (300) (4.42 3.23) (462 452) (0.60) (418 238)
= 347 108 = 239 J/g compressed (102.8 Btu/Ib
m
)
The work required to liquefy a unit mass of gas is
f
W 239
m 0.2725

&
&
= 877 J/g liquefied (377 Btu/Ib
m
)
For the Claude system operating under the conditions given In this problem,
the figure of merit is quite high:
89
i f
f
W / m 767
FOM 0.897
W/ m 877

&
&
&
&
Of course, in an actual system, irreversibilities in the heat exchangers,
expander, and compressor would reduce this figure considerably, as illustrated in Sec.
3.25. In any event, we see that the Claude system is a very effective liquefaction
system.
3.10. Kapitza system
Kapitza (1939) modified the basic Claude system by eliminating the third or
low-temperature heat exchanger, as shown in Fig. 3.20. Several notable practical
modifications were also introduced in this system. A rotary expansion engine was used
instead of a reciprocating expander. The first or high-temperature heat exchanger in
the Kapitza system was actually a set of valved regenerators, which combined the
cooling process with the purification process. The incoming warm gas was cooled in
one unit and impurities were deposited there, while the outgoing stream warmed up in
the other unit and flushed out the frozen impurities deposited in it. After a few
minutes, a valve was operated to cause the high- and low-pressure
Fig. 3.20. The Kapitza system. The warm heat exchanger is actually a switching
regenerator. This system was one of the first to use rotary expanders streams to switch
90
from one unit to the other. The Kapitza system usually operated at relatively low
pressures-on the order of 700 kPa (7 atm or 100 psia).
3.11. Heylandt system
Helandt (Davies 1949) noted that for a high pressure of approximately 20
MPa (200 atm) and an expansion-engine flow-rate ratio of approximately 0.60, the
optimum value of temperature before expansion through the expander was near
ambient temperature. Thus, one could eliminate the first heat exchanger in the Claude
system by compressing the gas to 20 MPa. Such a modified Claude system is called
the Heylandt system, after its originator, and is used extensively in high-pressure
liquefaction plants for air. The system is shown schematically in Fig. 3.21.
The advantage of the Heylandt system is that the lubrication problems in the
expander are not difficult to overcome. In the air-liquefaction system, the gas enters
the expander at ambient temperature and leaves the expander at approximately 150 K
(-190F) so that light lubricants can be used. In the Heylandt system, the expander and
the expansion valve contribute nearly equally in producing low temperatures, whereas
in the ordinary Claude system, the expander makes by far the largest contribution, as
one will note from Example 3.4.
3.12. Other liquefaction systems using expanders
There are many modifications that one could use to improve the performance
of the basic Claude system. In Fig. 3.22 is shown a dual-pressure Caude system,
similar in principle to the Linde dual-pressure system. In this system, only the gas that
is passed through the expansion valve is compressed to the high pressure. The gas that
is circulated through the expander is compressed to some intermediate pressure;
therefore, the work requirement per unit mass of gas liquefied is reduced. Optimum
performance for nitrogen gas compressed from 101.3 kPa (1 atm) to 3.5 MPa (35 atm
or 514 psia) is attained for this system when approximately 75 percent of the total flow
is diverted through the expander.
91
Fig 3.21: heylandt system
Fig. 3.22. Dual-pressure Claude system.
92
3.13. Liquefaction systems for LNG
Several LNG plants operate on a modification of the cascade system called
the mixed refrigerant cascade (MRC) system, as shown in Fig. 3.23. This system has
also been designated as the auto-refrigerated cascade (ARC) system (Linnett and
Smith 1970). The operation of this system is made possible by the fact that natural gas
is made up of several components that condense at different temperature levels. These
various components may be used to cool the feed stream in the MRC system without
using separate cooling circuits for the refrigerants by carefully controlling the
composition of the refrigeration cycle gas composition. This allows the use of a single
compressor for the recirculating gases instead of a separate compressor for each of the
different streams, as in the case of the ordinary cascade system.
Fig. 3.23. Mixed refrigerant cascade system used for liquefaction of natural gas.
The natural gas feed stream generally enters the MRC system at 3.9 MPa (565
psia) to 5.3 MPa (765 psia). If the natural gas feed stream pressure is sufficiently high,
the booster compressor would not be needed. The refrigerant cycle gas mixture is
93
compressed and partially condensed in the compressor aftercooler. The stream is
passed to a phase separator from which the propane-rich liquid phase is expanded
through a valve and mixed with the return gas stream to furnish cooling in the first
three-fluid heat exchanger. The vapor from the phase separator is partially liquefied in
the three-fluid heat exchanger and passed to a second phase separator, from which the
ethane-rich liquid is expanded through a valve and passed into the second three-fluid
heat exchanger.
The vapor from the second phase separator and the natural gas stream are
partially condensed in the final three-fluid heat exchanger. At this point, the
refrigeration cycle stream is primarily methane, so the stream is expanded through an
expansion valve and recirculated to provide cooling for the feed stream in the three-
fluid heat exchanger. The LNG is expanded down to the storage pressure in the liquid
receiver.
The liquefaction of natural gas for such applications as peak-shaving involves
several factors not encountered in other cryogenic liquefaction systems, such as air
liquefaction plants. In the case of air liquefaction, there are only two components
(nitrogen and oxygen) that are present in amounts larger than I percent, whereas
natural gas involves three or more components in amounts larger than I percent,
including methane, ethane, propane, and nitrogen. These components condense over a
wide temperature range. This fact makes the MRC system well suited for LNG plants;
however, the system is not applicable for liquefaction of pure gases or mixtures such
as air.
3.14. Comparison of liquefaction systems
The performance parameters of the systems discussed are given in Table 3.4
for air as the working fluid. The systems are assumed to operate between 101.3 kPa
(1 atm) and 20C (68F) and the conditions stated. Some measured performance
figures are given for comparison with the calculated values.
94
LIQUEFACTION SYSTEMS FOR NEON, HYDROGEN, AND HELIUM
3.15. Precooled Linde-Hampson system for neon and hydrogen
Because of its simplicity, the Linde-Hampson system is quite desirable for
small-scale liquefaction plants. We have seen, however, that the basic Linde-Hampson
system with no precooling would not work for neon, hydrogen, or helium because the
maximum inversion temperature for these gases is below ambient temperature. With
the precooled system, the temperature of the gas entering the basic Linde-Hampson
part of the liquefier can be lowered below ambient temperature by choosing the correct
fluid to precool the system.
In principle, any fluid that has a triple-point temperature below that of the
maximum inversion temperature of neon orhydro2en could be used as a precoolant.
Checking Table 2.6, we see that such fluids would include fluorine, oxygen, air,
methane, argon, and nitrogen. The first four can be ruled out from a practical
consideration because of their explosion hazard. Argon would be a possibility;
however, it is generally more expensive than liquid nitrogen. This leaves liquid
nitrogen as the choice for the precoolant for hydrogen- and neon-liquefaction systems.
A liquid-nitrogen-precooled Linde-Hampson system is shown schematically
in Fig. 3.24. For small laboratory liquefiers, the nitrogen-liquefaction subsystem would
be replaced by a small storage vessel from which liquid nitrogen could be withdrawn
and passed through the precooling bath, and the vapor discharged through the three-
channel heat exchanger to the atmosphere. For large-scale systems, an economic study
should be made to determine whether a separate nitrogen-liquefaction plant should be
used or not.
95
Fig 3.24: liquid-nitrogen precooled linde hampson system for neon or nitrogen
Because the part of the hydrogen system below the nitrogen precooling bath
is an ordinary Linde-Hampson system, from eqn. (3.22) the liquid yield is given by
7 4
7 f
h h
y
h h

(3.41)
Another parameter of interest for this system is the liquid-nitrogen
requirements. Applying the First Law for steady flow to the two heat exchangers in the
hydrogen or neon subsystem, the liquid-nitrogen bath, the liquid-hydrogen or neon
receiver and expansion valve, for no heat inleaks,
( )
2 2
N c f 1 f f N a 2
0 m h m m h m h m h mh + + & & & & & &
(3.42)
where
2
N
m&
is the mass flow rate of liquid nitrogen boiled away to precool the
incoming hydrogen or neon, m& is the mass flow rate of hydrogen or neon through the
compressor, and
f
m&
is the mass flow rate of hydrogen or neon which is liquefied. If
we define the nitrogen boil-off rate per unit mass of hydrogen or neon compressed as
z =
2
N
m / m & &
(3.43)
96
Then, solving for z from eqn. (3.42),
2 1 1 f
c a c a
h h h h
z y
h h h h

+

(3.44)
The amount of nitrogen boiled away per unit mass of hydrogen or neon
liquefied would be
2 2
N N
f f
m m / m
z
m m / m y

& & &
& & &
(3.45)
From our discussion in Sec. 3.6, we observed that the liquid yield for the
precooled Linde-Hampson system could be improved by lowering the temperature at
the entrance to the cold heat exchanger (point 4 in Fig. 3.24). This can be
accomplished easily in the hydrogen- or neon-liquefaction system by lowering the
pressure in the liquid-nitrogen bath. Because the liquid nitrogen boils in the precoolant
bath, a reduction in pressure lowers the boiling-point temperature or the bath
temperature. There is a practical limit to this process of lowering the bath temperature,
however. At 63.2 K (113.7"R) liquid nitrogen solidifies under its own vapor pressure
(this is the triple point to nitrogen), and further reduction in pressure results in solid
nitrogen in the bath instead of liquid. Good thermal contact is difficult to achieve
between solid nitrogen and the heat-exchanger walls because a layer of vapor forms
between the solid and the exchanger walls. This phenomenon limits the precoolant
bath temperature to values above 63.2 K. The precoolant boil-off parameter z/y is
shown in Fig. 3.25 as a function of temperature of the precoolant bath, assuming a
reversible system.
97
Fig. 3.25. Nitrogen boil-off per unit mass of hydrogen produced for the liquid-nitrogen
precooled Linde-Hampson system as a function of the liquid-nitrogen bath
temperature.
3.16 Claude system for hydrogen or neon
The Calude system does not depend primarily on the expansion valve to
produce low temperatures. Therefore, the system discussed in Sec. 3.9 may be used
for hydrogen or neon without modification. The performance is improved, however, if
a liquid-nitrogen precooling bath is used with the Claude system, as shown in Fig.
3.26. With the liquid-nitrogen precooling, the Claude system for hydrogen production
has a figure of merit 50 to 75 percent higher than that of the precooled Linde-Hampson
system.
98
Fig. 3.26. Precooled Claude system for hydrogen or neon.
3.17. Helium-refrigerated hydrogen-liquefaction system
An auxiliary helium-gas refrigerator can be used to condense hydrogen or
neon, as shown in Fig. 3.27. In this system, hydrogen or neon is compressed,
precooled by a liquid-nitrogen bath to reduce the helium-refrigerator work
requirement, and finally condensed by heat exchange with cold helium gas. The
helium refrigerator is a modified Claude system in which the gas is not liquefied but is
still colder than liquid hydrogen. The helium is compressed, precooled in the liquid-
nitrogen bath, and expanded in an expander to produce the low temperatures.
99
Fig. 3.27. Helium-gas-refrigerated hydrogen-liquefaction system.
An advantage of the helium-refrigerated system is that relatively low
pressures can be used. The compressor size can be reduced (although two compressors
are required) and the pipe thickness can be reduced, in comparison with that required
for higher pressures. The hydrogen or neon need be compressed only to a pressure
high enough to overcome the irreversible pressure drops through the heat exchangers
and piping in an actual system. Pressure from -300 kPa (3 atm) to 800 kPa (8 atm) is
usually adequate for the hydrogen loop. The system is relatively insensitive to the
pressure level of the helium refrigerator. For a helium-gas pressure of I MPa (10 atm),
work requirements of approximately 11,000 kl/kg liquefied (26,000 Btu/Ib
m
) can be
realized for a practical system, or a figure of merit of 0.11, which includes the work
required to produce the liquid nitrogen.
100
3.18. Ortho-para-hydrogen conversion in the liquefier
We saw in Sec. 2.13 that hydrogen can exist in two different forms-
parahydrogen and ortho-hydrogen. The ortho-para concentration in equilibrium
hydrogen depends upon the temperature of the hydrogen. Near room temperature, the
composition is practically 75 percent ortho-hydrogen and 25 percent para-hydrogen,
whereas at the normal boiling point of hydrogen, the equilibrium composition is
almost all para-hydrogen. When hydrogen gas is passed through a liquefaction system,
the gas does not remain in the heat exchangers long enough for the equilibrium
composition to be established at a particular temperature. The result is that the fresh
liquid has practically the room-temperature ortho-para composition and will, if left
alone in the liquid receiver, undergo the exothermic reaction there. The changeover
from ortho- to para-hydrogen involves a heat of conversion that is greater than the heat
of vaporization of para-hydrogen; therefore, serious boil-off losses will result unless
measures are taken to prevent it. This is a problem peculiar to hydrogen-liquefaction
systems that must be solved in any efficient system.
A catalyst may be used to speed up the conversion process, while the heat of
conversion is absorbed in the liquefaction system before the liquid is stored in the
liquid receiver. Because the heat of conversion results in an increase in liquid
evaporated, it is advantageous to carry out as much of the conversion in the liquid-
nitrogen bath as possible. The nitrogen is much less costly to produce than the liquid
hydrogen. Note from Table 2.7 that the equilibrium composition at temperature near
70 K (126R), corresponding to liquid nitrogen boiling under vacuum, is
approximately 55 to 60 percent para-hydrogen. Thus if the conversion is complete at
this temperature, the energy released in the liquid receiver is reduced by almost one-
half.
Two possible arrangements for ortho-para conversion are shown in Fig. 3.28.
In the first arrangement, the hydrogen' is passed through the catalyst in the liquid-
nitrogen bath, expanded through the expansion valve into the liquid receiver, and
drawn through a catalyst bed before passing into a storage vessel. The hydrogen that is
evaporated due to the heat of conversion flows back through the heat exchanger and
furnishes additional refrigeration to the incoming stream. The second arrangement is
101
similar to the first one, except that the high-pressure stream is divided into two parts
before the expansion valve. One part is expanded through an expansion valve and
flows through a catalyst bed immersed in a liquid-hydrogen bath; the converted
hydrogen is passed to a storage vessel.
Fig. 3.28. Ortho-para-hydrogen conversion arngements.
The other part of the high-pressure stream is expanded through another
expansion valve into the liquid receiver to furnish refrigeration for the catalyst bed; the
vapor is passed back through the heat exchanger to cool down the incoming gas. The
second arrangement allows approximately 20 percent higher liquid-hydrogen yields
compared with the first arrangement.
Some of the catalysts that have proved effective are (1) hydrous ferric oxide,
(2) chromic oxide on alumina particles, (3) charcoal and silica gel, and (4) nickel-
based catalyst. Of these, hydrous ferric oxide is the most active; that is, a relatively
small volume of catalyst is required to produce practically complete conversion to the
equilibrium composition. The conversion process is speeded up for any of the catalysts
102
if they are ground into fine pellets, which offer a larger surface area per unit volume
than do large chunks of material.
Certain impurities will "poison" the catalysts or severely reduce their
effectiveness (Scott et al. 1964). Methane, carbon monoxide, and ethylene act as
temporary poisons, whereas chlorine, hydrogen chloride, and hydrogen sulfide
permanently decrease the catalyst activity. It is important to remove these materials
from the hydrogen feed stream before they enter the liquefier.
3.19. Collins helium-liquefaction system
As for the case of hydrogen and neon, the precooled Linde-Hampson system
may be used to liquefy helium by using liquid hydrogen as the precoolant. This type of
precooled system was used in several of the earlier helium liquefiers (Mann et al.
1960).
One of the milestones in cryogenic engineering was the design and
development of a helium liquefier by Samuel C. Collins at the Massachusetts Institute
of Technology. This liquefier is an extension of the Claude system, as shown in Fig.
3.29. Depending upon the helium inlet pressure, from two to five expansion engines
are used in this system.
Applying the First Law for steady flow to the system consisting of all
components except the helium compressor and the expansion engines, we find that the
liquid yield
f
y m / m & &
is
2
e
ei 1 2
1 2
1 f 1 f 1 f
h
h h h
y x x
h h h h h h

+ +

(3.46)
where
1
1 e
x m / m & &
2
2 e
x m / m & &
1
e
h
= enthalpy change of fluid passing through expander 1
2
e
h
= enthalpy change of fluid passing through expander 2
1 2
e , e
m m & &
= mass flow rates of fluid through expanders 1 and 2,
respectively.
103
For more than two engines, an additional term similar to the second term for
each expander would be added to eqn. (3.46).
The cool-down time for the Collins liquefier may be reduced from about 4
hours to 2 hours by the use of a liquid-nitrogen precooling bath. In addition, the use of
the precooling bath increases the liquefaction performance of the system (the liquid
yield can be almost tripled); however, the precoolant bath is not required because the
system does not depend solely upon the Joule-Thompson effect for the production of
low temperatures.
3.20. Simon helium-liquefaction system
One of the methods used to liquefy small quantities of helium is the Simon
liquefaction system (Pickard and Simon 1948). This system does
Unless the mass released from the vessel is measured, the calculation
procedure is an iterative one because m6 must be known to find y. However, the mass
at point 6 can be determined only after the liquid yield is known.
Cailletet observed a thick mist when oxygen gas at - 32C ( - 26F) and high
pressure was suddently expanded in an apparatus similar to the Simon liquefier;
however, in general, the Simon liquefier does not work so well for gases other than
helium. There are two reasons for this fact: (1) The ration
g
/
f
is smaller for
helium than any other gas at the normal boiling point (
g
/
f
= 7.5 for helium; 53.3
for hydrogen; and 175 for nitrogen), which means that for a given liquid yield y, the
volume ratio V
f
/V is larger for helium than for other gases. (2) The specific heat of
metals is extremely small at helium temperatures, so only a small amount of cooling
capacity is lost into the walls of the vessel for helium liquefaction.
The Simon system is primarily a laboratory liquefier because it can produce
helium only in relatively small quantities. It is especially well suited for experiments
involving magnetic fields in which a small space is available between the poles of a
magnet. The liquid in the heavy-walled container could serve as the refrigerant to cool
a paramagnetic material, for example, on which experiments would be carried out.
104
MODULE IV
CRYOGENIC REFRIGERATION SYSTEMS
Systems that utilize cryogenic temperatures in their operation, such as
advanced electronic systems, superconducting magnets and motors, all depend upon an
effective refrigeration system to maintain the low temperatures required. Many
refrigeration systems have the same components and thermodynamic cycles as the
corresponding liquefaction systems. The difference between the refrigeration system
and the liquefaction system is that the liquid produced is evaporated in a refrigeration
system instead of being utilized in some other way external to the system, as in the
liquefaction system. In this chapter, we shall examine these refrigeration systems that
are similar to Liquefaction systems, and in addition we shall consider some unique
refrigerators based on entirely different concepts-the Philips refrigerator, based on the
old Stirling cycle, and the Gifford-McMahon refrigerator. We shall also examine
methods of obtaining and maintaining temperatures below 2 K (3.6R), such as
magnetic refrigeration and the dilution refrigerator.
IDEAL REFRIGERATION SYSTEMS
4.1. The thermodynamically ideal isothermal-source system
As in the preceding chapters, we shall first investigate the thermodynamically
ideal system in order to have a basis for comparison of the various practical
refrigeration systems. In the case of refrigerators, however, there are two types of low-
temperature source that may be used. If we evaporate a liquid to furnish the cooling,
energy is added to the refrigerant in an isothermal manner, as-far as the evaporator of
the refrigerator is concerned. On the other hand, we may use a cold gas such as helium
(which gets quite cold before it Liquefies), and energy is added at constant pressure
(for the refrigerant) without liquefying the gas. Therefore, we must differentiate
between an isothermal-source refrigerator and an isobaric source refrigerator. The sink
into which energy is rejected is usually the atmosphere; thus, we shall have an
isothermal sink in both cases. We use the term "source" to mean the source of heat for
the refrigerator-that is, the space to be cooled. The term "sink" refers to the region into
which heat is rejected from the refrigerator.
105
Fig. 4.1 Carnot refrigerator. (a) Reversible isothermal compression; (b) reversible
adiabatic expansion; (c) reversible isothermal expansion with heat adsorption from the
low temperature source; (d) reversible adiabatic compression.
The thermodynamically ideal isothermal-source refrigerator is the Carnot
refrigerator, shown schematically in Fig. 5.1. The processes involved in the Carnot
refrigerator are as follows.
Process 1-2. The working medium is compressed while energy is rejected to
the ~ink to maintain the refrigerant temperature constant.
Process 2-3 The working medium is expanded reversibly and adiabatically
from the sink temperature T
h
to the source temperature T
c
.
Process 3-4. Energy is transferred from the source (the region to be cooled) to
the refrigeration medium, while work is done by the medium to maintain the
refrigerant at constant temperature.
Process 4-1. The refrigerant-is then compressed reversibly and adiabatically
(isentropic process) from the source temperature to the sink temperature.
106
The Carnot cycle thus consists of two reversible adiabatic processes and two
reversible isothermal processes as shown in Fig. 5.2.
A measure of the performance of a refrigerator is the coefficient of
performance (COP), defined as the energy removed from the source divided by the
work required to remove this amount of energy:
( ) ( )
a net a net
COP Q / W Q / m / W / m
(4.1 )
where Q
a
/m is the refrigeration effect or the energy absorbed by the
refrigeration medium per unit mass of refrigerant, and W
net
/m is the net work expended
per unit mass of refrigerant. Notice that Q
a
is a positive quantity (heat added to the
system is considered positive), while the net work expended is a negative quantity
(work done on the system is considered negative; work done by the system is
considered positive). For this reason, we have included the negative sign in eqn. (4.1)
so that the coefficient of performance will be a positive number.
To compare the performance of practical systems, we shall use the figure of
merit for the refrigerator, defined by
FOM = COP/COP
i
(4.2)
where COP is the coefficient of performance of the actual system and COP, is the
coefficient of performance of the thermodynamically ideal system. As in the case of
liquefaction systems and separation systems, the figure of merit is a number between
zero and unity. A FOM near unity implies a very "good" refrigerator, and a small FOM
implies a "poor" refrigerator compared with the ideal refrigerator.
107
Fig 4.2: carnot cycle
As a consequence of the First Law of Thermodynamics, the net work in a
cycle is equal to the net heat transferred because the initial and final states in a cycle
are identical. In the Carnot cycle, all processes are reversible; therefore, the net heat
transfer is given by
( ) ( )
net h 2 1 c 4 3
Q mT s s 0 mT s s 0 + + +
or
( ) ( )
net h c 1 2 net
Q / m T T s s W / m
(4.3)
where s is the entropy and the subscripts refer to the numbered points in Fig. 5.2. The
energy added to the system from the source at constant temperature is given by
( ) ( )
a c 4 3 c 1 2
Q / m T s s T s s
(4.4)
Using the expressions for the net work and the heat absorbed from the source,
we find the coefficient of performance for the Carnot refrigerator to be
c
i
h c
T
COP
T T

(4.5)
108
where the temperatures are absolute temperatures (K or OR).
From eqn. (4.5) we see that the coefficient of performance for the Carnot
system is independent of the refrigerant. That is, between the same temperature limits,
the COP would be the same if helium gas or liquid nitrogen or liquid argon were used
as the refrigerant. This was not the case for liquefaction systems.
One of the corollaries of the Second Law of Thermodynamics is that no
refrigeration system can have a COP larger than that of a Carnot refrigerator operating
between the same two temperature limits; otherwise, a perpetual-motion machine of
the second kind could be formed. All practical refrigerators require more work for a
given refrigeration effect than that required by a Carnot refrigerator. For an ambient or
sink temperature of 300 K (5400R or 80F), the coefficient of performance for various
source temperatures is shown in Table 4. 1. We note from Table 4.1 that the work
requirement for a given refrigeration effect increases as the source temperature is
lowered. Even for the thermodynamically ideal refrigerator, large expenditures of work
are required to maintain very low temperatures. The cryogenics engineer must pay
dearly to maintain a low temperature, and the "dearness" increases as the temperature
is lowered.
4.2. The thermodynamically ideal isobaric-source system
In cold-gas refrigerators or refrigeration systems in which the working
medium is not condensed, energy is absorbed at a varying temperature instead of a
constant temperature as in the Carnot refrigerator (Jacobs 1962). In this case, it would
be unfair to compare the actual system with the Camot system because the source
temperature for the real system is not constant. The thermodynamically ideal isobaric-
source refrigeration cycle is shown on the temperature-entropy plane in Fig. 4.3. This
is the ideal cycle that must be used in comparing real systems that absorb heat at
constant pressure.
109
Table 4.1. Coefficient of performance for a Carnot refrigerator operating between 300
K (5400R) and a low temperature T
c
Source Temperature
No K
o
R COP
i
= Q
a
/W
net
W
net
,Q
a
1 111.7 201.1 0.5932 1.686
2 77.4 139.3 0.3477 2.876
3 20.3 36.5 0.07258 13.778
4 4.2 7.56 0.01420 70.43
5 1.0 1.8 0.003344 299.0
6 0.1 0.18 0.0003334 2,999.0
7 0.01 0.018 0.0000 333 29,999.0
Suppose we let the sink temperature (usually ambient temperature) be T
0
, the
lowest source temperature be T
1
, and the highest source temperature be T
2
. Energy is
added reversibly to the refrigerant at constant pressure between the temperatures T
1
and T
2
. The energy-rejection process is a reversible isothermal process; therefore, the
energy rejected from the system is given by
( ) ( )
r 0 4 3 0 2 1
Q mT s s mT s s
(4.6)
110
Fig 4.3: Reversible isobaric-source refrigeration cycle
Because the heat-absorption process is a constant-pressure process, the total
energy absorbed from the source is given by
( ) ( )
2 2
a 2 1
1 1
Q mT ds m dh dp m h h

(4.7)
From the First Law of Thermodynamics applied to the entire cycle, we find
that the net work transfer is equal to the net heat transfer:
( ) ( )
net net r a 0 2 1 2 1
W Q Q Q m T s s h h + ]
]
(4.8)
Using the definition of the coefficient of performance, eqn. (4.1), we find the
following expression for the COP for an ideal isobaric-source refrigerator:
( ) ( )
a 2 1
i
net 0 2 1 2 1
Q h h
COP
W T s s h h



(4.9)
Equation (4.9) is valid for any working substance. However, for many cold-
gas refrigerators, the pressures are sufficiently low that the working fluid may be
111
assumed to behave approximately as an ideal gas. For an ideal gas, the enthalpy
change is given by
( )
2 1 p 2 1
h h c T T
For an ideal gas in a constant-pressure process, the entropy change is given by
( ) ( ) ( )
2 1 p 2 1 2 1 p 2 1
s s c ln T / T R ln p / p c ln T / T
If we make these substitutions into eqn. (4.9), we obtain the following
expression that is valid for an ideal gas with constant specific heats:
( ) ( )
2 1
i
0 2 1 2 1
T T
COP
T ln T / T T T

(4.10)
This expression may be written in terms of the temperature ratios T
2
/T
1
, and
T
0
/T
1
, as follows:
( )
( ) ( ) ( )
2 1
i
0 1 2 1 2 1
T / T 1
COP
T / T ln T / T T / T 1

+
(4. 11)
From eqn. (4.11) we see that the COP is independent of the ideal gas used as
the refrigerant. The COP
i
depends only upon the ratios of the highest Source
temperature to the lowest Source temperature and the sink temperature to the lowest
source temperature. Equation 4.11) is plotted in Fig. 4.4. It can be shown that if T
2
/T
1
approaches unity, eqn. (4.11) reduces to the expression for the COP
i
of a Carnot
refrigerator.
Example 4.1. Determine the ideal COP for an isobaric Source refrigerator
operating reversibly between a sink temperature of 300 K (80F) and a minimum
source temperature of 110 K (198R) and a maximum source temperature of 180 K
(324R). The working fluid is gaseous nitrogen, and the source pressure is 1.013 MPa
(10 atm).
112
Fig. 4.4. Coefficient of performance for an ideal isobatic source refrigerator. To is the
sink temperature: T
1
and T
2
are the minimum and maximum source temperatures,
respectively.
From the temperature-entropy diagram for nitrogen, we find the following
property values:
h
1
= 248 J/g at 110 K and 1.013 MPa
s
1
= 2.59 J/g-K f
h
2
= 332 Jig at 180 K and 1.013 MPa
s
2
= 3.18 J/g-K .
The ideal COP may be determined from eqn. (4.9):
( )
( ) ( ) ( )
i
332 248
COP 0.903
300 3.18 2.59 332 248

In this pressure and temperature range, nitrogen is not quite an ideal gas. Thus
we should expect to get a slightly different answer if the ideal-gas expression, eqn.
(4.10), is used:
113
( )
( ) ( ) ( )
i
180 110
COP 0.900
300 ln 180/110 180 110

This value differs by only about 0.3 percent from the answer obtained
previously.
We may compare the ideal COP obtained in this problem with the COP for a
Carnot refrigerator operating between 300 K (5400R) and 110 K (198R). Using eqn.
(4.5),
i
110
COP 0.579
300 110

In this case, there is considerable difference between the COP, of the Camot
refrigerator and that of the isobaric-source refrigerator.
REFRIGERATORS FOR TEMPERATURES ABOVE 2 K
4.3. Joule- Thomson refrigeration systems
Any of the liquefaction systems that do not use an expansion engine may be
classed as Joule-Thomson refrigerators because they depend upon the Joule-Thomson
effect to produce low temperatures. Instead of withdrawing the liquid formed in a
refrigerator, heat is absorbed from the low temperature Source to evaporate this liquid.
A, simple Linde-Hampson refrigerator is shown in Fig. 4.5, and its cycle is shown on
the temperature-entropy plane in Fig. 4.6 .
. As mentioned in Chap. 3, the compression from point I to point 2 in Fig, 4.6
would be isothermal in the Ideal case. In practice, the gas enters the compressor at
point I' and there is a small temperature difference because the effectiveness of the heat
exchanger is less than unity. The compressed gas is passed through the heat exchanger,
cooled to low temperatures by heat exchange with the cold outgoing gas stream, and
expanded through a Joule-Thomson valve into an evaporator. In the evaporator (which
corresponds to the liquid receiver in the liquefaction system), the liquid formed after
the expansion process is evaporated (at constant temperature) by absorbing heat from
the space to be refrigerated. The vapor then returns through I the heat exchanger to
the compressor.
114
Fig. 4.5. Linde-Hampson refrigerator.
Fig. 4.6. Thermodynamic cycle for the Linde-Hampson refrigerator.
If we apply the First Law of Thermodynamics to the system consisting of the
heat exchanger, expansion valve, and evaporator and assume no heat inleaks from
ambient as well as negligible kinetic-energy and potential-energy changes of the
working fluid, we obtain
115
( )
a 1 2
Q m h h
&
&
(4.12)
where
1
h
is the actual enthalpy of the fluid leaving the heat exchanger at the
warm end. The heat exchanger effectiveness is defined by
1 g
1 g
h h
h h

(4.13)
Using eqn. (4.13) to eliminate the enthalpy h~, the refrigeration effect may be
written in terms of the fluid properties and the heat exchanger effectiveness:
( ) ( ) ( )
a 1 2 1 g
Q / m h h 1 h h
&
&
(4.14)
where h
i
is the enthalpy of the fluid at the exit of the heat exchanger under
ideal conditions-that is, at the same temperature as at point 2.
We may make two observations from eqn. (4.14). First, we see that the Joule-
Thomson refrigerator cannot be used with neon, hydrogen, or helium as the working
medium, unless these gases are first precooled below their maximum inversion
temperatures. Because the heat absorbed by the refrigerator must be a positive quantity
(considering the refrigerator as the thermodynamic system), h
1
must be larger than h
2
for an ideal heat exchanger in order to have a positive refrigeration effect ( )
a
Q / m
&
&
.
This condition is not met if the working fluid enters the heat exchanger at the warm
end at a temperature above the maximum inversion temperature for the fluid.
Second, we see that there is a value of the heat exchanger effectiveness below
which the refrigerator will not work. This limiting effectiveness may be determined by
setting the refrigeration effect equal to zero in eqn. (4.14).
The work requirement for the system is given by
( ) ( )
0
2 1 2 1 2
a
T s s h h
W
m

&
&
(4.15)
where
0
a

is the overall efficiency of the compressor. From the definition of

the coefficient of performance, eqn. (4.1), we find the COP for the Linde-Hampson
refrigerator to be
116
( ) ( ) ( )
( ) ( )
c, 0 1 2 1 g
a
2 1 2 1 2
h h 1 h h
Q
COP
W T s s h h
]

]


&
&
(4.16)
The liquid in the evaporator boils at constant temperature; thus this
refrigerator is of the isothermal-source type. Liquid nitrogen is a suitable refrigerant
for maintaining temperatures in the region between about 65 K (117
o
R) and 115 K
(207
o
R). The temperature in the evaporator can be regulated by controlling the
pressure in the evaporator by means of the expansion-valve setting. At 65 K the
evaporator pressure would be 17.4 kPa (2.53 psia), and at 115 K the evaporator
pressure would be 1.939 MPa (281 psia), using nitrogen as the working fluid. The
temperature range for the refrigerator is limited on the lower end by the triple point of
the working fluid and also by the difficulty in maintaining low vacuum pressure with
large mass flow rates. If the pressure were lowered below triple-point pressure,
nitrogen snow would form in the evaporator and clogging of the expansion valve could
result. In addition, there would be poor heat transfer in the evaporator between the
evaporator wall and the porous solid cryogen. The temperature is limited on the high
end by the critical point. As the critical point is approached, the heat of vaporization of
the liquid approaches zero. If we are not concerned with having an isothermal source,
the temperature range of the Linde-Hampson refrigerator can be extended up to
ambient temperatures. However, when we get into the temperature region above about
200 K (3600R), other refrigerants such as the Freon compounds become more
attractive as refrigeration media.
Stephens (1970) and Buller (1971) have described miniature Joule-Thomson
refrigerators that utilize thermostatically operated expansion valves. The self-
regulating Joule-Thomson refrigerator has the advantage of rapid cool-down because
the initial gas flow rate is much larger than that for a fixed-orifice refrigerator. In
addition, the problem of solid contaminants in the gas stream is significantly reduced.
Reliable operation at pressures from 34.5 MPa (5000 psi) to 69 MPa (10,000 psi) are
made possible with the use of the self-regulating orifice.
Chan et al. (1981) described a miniature Joule-Thomson refrigerator using an
adsorption compressor. Large quantities of gas are adsorbed in an adsorbent, then the
117
adsorbent is heated in a closed system to produce a high pressure gas. The gas passes
through a Joule-Thomson system and is adsorbed in a second adsorbent chamber,
which is cooled. A minimum COP of 0.22 was reported for nitrogen gas as the
working fluid absorbing energy at 77 K (139R). The cold compressor temperature
was 150 K (2700R), and the hot compressor temperature was 470 K (846R or 386F).
Example 4.2. Determine the refrigeration effect, COP, and figure of merit for
a simple Linde-Hampson refrigerator operating from 300 K (80F) and 101.3 kPa (I
atm) to 10.13 MPa (100 atm). The overall efficiency of the compressor is 75 percent,
and the heat exchanger effectiveness is 0.960. The working fluid for the refrigerator is
nitrogen.
From the temperature-entropy diagram for nitrogen, we find the following
property values:
h
i
= 462 J/g at 300 K and 101.3 kPa (I atm) $, = 4.42 J/g-K
h
2
= 444 J/g at 300 K and 10.13 MPa (100 atm)
s
2
= 3.00 J/g-K i
h
g
= 229 J/g at 101.3 kPa (77.36 K) and saturated vapor conditions
h
r
= 29 J/g at 101.3 kPa (I atm) and saturated liquid conditions.
The work requirement per unit mass for the refrigerator is
( ) ( )
2 1 2 1 2 c, 0
W/ m T s s h h / n ]
]
&
&
= [(300) (4.42 3.00) (462 444)] / (0.750)
= (408)/(0.750) = 544 J/g (234 Btu/Ib
m
)
( ) ( )
a 1 2 1 g
Q / m h h 1 h h
]

]
&
&
= (462 444) (1 0.960) (462 229)
= 8.68 J/g (3.73 Btu/Ib
m
)
The coefficient of performance for the refrigerator is:
a
COP Q / W 8.68/ 544 0.01596
& &
118
The COP, for a Carnot refrigerator operating between T
h
= 300 K and
T
c
= 77.36 K is given by
( ) ( ) ( )
i c h c
COP T / T T 77.36 / 300 77.36
= 0.3475
The figure of merit for the system is
FOM = COP/COP
i
= 0.01596/0.3475 = 0.0459
4.4. Refrigerator optimization
In the design of a Joule-Thomson refrigerator, the designer may determine the
heat exchanger effectiveness at the design stage by selection of the heat exchanger
surface area. If an effectiveness near unity is selected, the heat exchanger surface area
is quite large and corresponding high heat exchanger costs results. On the other hand,
if an effectiveness near the lower limiting value is chosen, the mass flow rate for a
given heat absorption rate is quite large and a corresponding high compressor cost
results. It is apparent that there is an intermediate value of the heat exchanger
effectiveness (optimum value) that will minimize the total annual cost of the
refrigerator. This problem has been examined by Gifford (1960) for the case of a
balanced heat exchanger.
The primary components of cost of the refrigeration system are the
compressor costs, which includes both operating costs (energy costs) and capital costs,
and the heat exchanger costs. As an initial approximation, the compressor, costs are
proportional to the power requirement of the compressor,
( )
c 1 c 1 c
C C W C W / m m
& &
& &
(4.17)
where C
1
is the compressor cost per unit power requirement, including both operating
and capital costs. Similarly, the heat exchanger costs are proportional to the surface
area of the heat exchanger, as an initial approximation:
C
E
= C
2
A (4.18)
Ordinarily, such items as piping, valves, insulation and so on have costs that
are not dependent upon the heat exchanger effectiveness. Therefore, let us consider the
total system cost as the sum of the compressor and heat exchanger costs:
119
C
T
= C
c
+ C
E
The optimum condition may be found by setting the derivative of the total
cost equal to zero, or
c T
1 2
W dC dm dA
0 C C
di m di di
| `
+

. ,
&
&
&
(4.19)
where i = 1

t = heat exchanger ineffectiveness.

The required mass flow rate through the system may be determined from eqn.
(4.14):
( ) ( ) ( ) ( )
a a
1 2 1 g 1 g
Q Q
m
h h i h h h h H i


& &
&
(4.20)
where
( ) ( )
1 2 1 g max
H h h / h h i (4.21)
The quantity H is the upper limiting ineffectiveness above which the
refrigerator will not work at all. From eqn. (4.20), we find
( ) ( )
a
2
1 g
Q dm
di
h h H i

&
&
(4.22)
Because most heat exchangers used in cryogenic systems are counter flow
exchangers, the surface area may be found from eqn. (3.88):
( )
( )
c
R R
R
mc / U
1 C C i
A ln
1 C i
+ | `

. ,
&
(4.23)
where C
R
= C
c
/C
h
= capacity rate ratio
C
c
= mean specific heat of the cold stream
C
h
= mean specific heat of the warm stream
The term ( )
c
mc / U &
is generally a weak function of the mass flow rate
because the overall heat transfer coefficient is proportional to the mass flow rate raised
120
to a power near unity; therefore, this ratio may be considered as a constant as a first
approximation. The area derivative may be found from eqn. (4.23):
( )
( )
c
R R
mc / U
dA
di i 1 C C i

+
&
(4.24)
For convenience, let us define the parameter as follows:
( ) ( )
( )
2 1 g c
1 c a
C h h mc / U
C W / m Q

&
& &
&
(4.25)
Making the substitutions from eqns. (4.22), (4.24), and (4.25) into eqn. (4.19),
we obtain
( ) ( )
2 2
R R
C i 2 H 1 C i H 0 + + (4.26)
For the case in which is not equal to C
R
, we may define the following
dimensionless parameters:
( )
R
1
R
2 H 1 C
B
2 C
+

(4.27)
2
2
R
H
B
C

(4.28)
We may write eqn. (5.26) in the simple form,
2
1 2
i 2B i B 0 + (4.29)
Two values of the optimum ineffectiveness result from the solution of eqn.
(5.29); however, only one solution yields an ineffectiveness value less than i
max
= H.
The optimum ineffectiveness for
R
C
is given by
( )
1/ 2
2
opt opt 1 1 2
i 1 B B B (4.30)
If by chance,
R
C
, the optimum ineffectiveness may be determined
directly from eqn. (4.26):
2
R
opt
R R
C H
i
2C H 1 C

+
(4.31)
121
Note that if we are given the heat exchanger "free"; that is, if C2 = O. we find
that = a and B
2
= O. From eqn. (4.30), we find that the optimum ineffectiveness
would be i
opt
= 0, which is the result that we would anticipate for a free heat exchanger.
4.5 Cascade or precooled Joule-Thomson refrigerators
For temperatures lower than those obtainable with liquid nitrogen, the only
available working fluids are neon, hydrogen, and helium. For these fluids, precooling
must be used in any system that has no expansion engine. A typical precooled liquid-
neon or liquid-hydrogen refrigerator (Geist and Lashmet 1961) is shown schematically
in Fig. 4.7. The cycle for the system is shown in the temperature-entropy plane in Fig.
4.8.
Fig 4.7: Precooled Linde-Hampson refrigerator
122
Fig 4.8: Thermodynamic cycle for the Precooled Linde-Hampson refrigerator
If we apply the First Law of Thermodynamics to all components of the
system shown in Fig. 4.7 except the compressors, neglect heat inleaks from ambient,
and neglect kinetic-energy and potential-energy changes of the fluids, we obtain
( ) ( )
a 1 2 p a b
Q m h h m h h +
&
& &
(4.32)
where m& is the mass flow rate of the main refrigerant,
p
m&
is the mass flow rate of the
precoolant, and the subscripts on the enthalpy terms correspond to the points in Fig.
4.8. If we define the precoolant mass-flow-rate ratio as
z =
p
m / m & &
then eqn. (4.32) may be written
( ) ( )
a 1 2 a b
Q / m h h z h h +
&
&
(4.33)
We may introduce the heat exchanger effectiveness for the main heat
exchanger E and for the precoolant heat exchanger
p

, defined by
1 g
1 g
h h
h h

(4.34a)
123
a e
p
a e
h h
h h

(4.34b)
Making these substitutions into eqn. (4.33), we find
( ) ( ) ( )
a 1 2 1 g
Q / m h h 1 h h
&
&
( ) ( ) ( )
a b p a e
z h h 1 h h
]
+
]
(4.35)
where h
1
is the enthalpy of the main refrigerant at the temperature T
2
, and h. is
the enthalpy of the precoolant at the temperature T
b
.
Applying the First Law to the cold exchanger and the evaporator, we obtain
a 1 4
Q / m h h
&
&
(4.36)
Introducing the effectiveness of the cold exchanger,
1 g
c
7 g
h h
h h

(4.37)
the refrigeration effect may be written as follows:
( ) ( ) ( )
a 7 4 c 7 g
Q / m h h 1 h h
&
&
(4.38)
The required precoolant mass-flow-rate ratio may be determined by equating eqns.
(4.35) and (4.38), if we assume that the temperature at point 4 is practically equal to
the precoolant bath temperature so that h. and 117 are known quantities.
Lower temperatures may be attained by using a three-stage cascade
refrigerator with nitrogen (or argon), hydrogen (or neon), and helium as the working
fluids. An example of this type of refrigerator is shown in Fig. 4.9.
124
Fig 4.9 : Three stages Joule- Thomson liquid helium refrigerator
The Claude liquefaction system or the Collins liquefaction system could be
used as a refrigeration system. A schematic of a Claude refrigerator is shown in Fig.
4.10. If we apply the First Law to the three heat exchangers, the expansion valve, and
the evaporator as a unit, neglecting heat inleaks from ambient and kinetic-energy and
potential-energy changes, we obtain the following for the heat absorbed by the
refrigerant:
( ) ( )
a 1 2 3 e
Q / m h h x h h +
&
&
(4.39)
125
where
e
x m / m & &
= expander mass-flow-rate ratio,
e
m&
= mass flow rate
through the expander, m& = mass flow rate through the compressor, and the subscripts
refer to the points given in Fig. 4.11. If we let h
e
= enthalpy at the end of an isentropic
expansion from point 3 to the pressure at point e, then the expression for the
refrigeration effect may be written in terms of the expander adiabatic efficiency as
follows:
( ) ( )
a 1 2 ad 3 e
Q / m h h x h h +
&
&
(4.40)
The net work requirement, assuming that the expander work is utilized to help
in the compression of the gas, is given by
( ) ( ) ( )
net 2 1 2 1 2 c, 0 e, m ad 3 e
W / m T s s h h / x h h ]
]
&
&
(4.41)
Fig. 4.10. Claude refrigerator.
126
Fig. 4.11. Thermodynamic cycle for the Claude refrigerator.
where
c,0
is the overall efficiency of the compressor and
e,m
is the
mechanical efficiency of the expander.
Meier and Currie (1968) described the performance of a Claude refrigerator
used to maintain a low temperature of 4 K (7.2R) while providing I watt of
refrigeration. Whitter (1966) described the design of a refrigerator utilizing two
expansion engines, similar to the Collins liquefier.
Two significant modifications of the basic Claude system are the use of a
"wet" expander or expander operating in the two-phase region to replace the expansion
valve (Johnson et al. 1971), and the use of a low temperature compressor (Minta and
Smith 1982). A schematic of this system is shown in Fig. 4.12, and the cycle is shown
on the temperature entropy plane in Fig. 4.13. The two-phase expander is used
primarily for systems involving helium as the working fluid because the thermal
capacity of the compressed gas is generally larger than the latent heat of the liquid
phase. Unlike an air or nitrogen expansion engine in which the engine efficiency is
seriously affected by the presence of liquid in the engine, operation of the helium
127
expander in the two-phase region does not result in a serious deterioration of the
engine performance. The thermodynamic performance of the system is improved by
the use of the saturated-vapor compressor. In addition, the required heat exchanger
surface is less than that required for the conventional Claude system because of heat-
transfer coefficients are higher when the cold gas stream pressure is increased.
Fig. 4.12. Claude refrigerator with a wet expander and a saturated-vapor compressor.
Fig 4.13: Thermodynamic cycle for the system shown in fig 4.12
4.7. Philips refrigerator
The Philips refrigerator operates on the Stirling cycle, which was in vented in
1816 by a Scottish minister, Robert Stirling, for use in a hot-air engine. As early as
1834, John Herschel (Kohler 1960) suggested that this engine could be used as a
128
refrigerator. The first Stirling cycle refrigerator was constructed by Alexander Kirk
(Kirk 1874) around 1864. A schematic of -the sequence of operations of the Philips
refrigerator is shown in Fig. 4.14, and the cycle is shown on the temperature-entropy
plane in Fig. 4.15.
The Philips refrigerator consists of a cylinder enclosing a piston, a displacer,
and a regenerator. The piston compresses the gas, while the displacer simply moves
the gas from one chamber to another without changing the gas volume, in the ideal
case. The heat exchange during the constant-volume process is carried out in the
regenerator.
Fig 4.14 Philips refrigerator schematic
The sequence of operations for the system is as follows.
Process 1-2. The gas is compressed isothermally while rejecting heat to the
high-temperature sink (surroundings).
129
Process 2-3. The gas is forced through the regenerator by the motion of the
displacer. The gas is cooled at constant volume during this process. The energy
removed from the gas is not transferred to the surroundings but is stored in the
regenerator matrix.
Process 3-4. The gas is expanded isothermally while absorbing heat from the
low-temperature source.
Process 4-1. The cold gas is forced through the regenerator by the motion of
the displacer; the gas is heated during this process. The energy stored during process 2-
3 is transferred back to the gas. In the ideal case (no heat inleaks), heat is transferred to
the refrigerator only during process 3-4, and heat is rejected from the refrigerator only
during process 1-2.
If we assume that the heat transfers to and from the refrigerator are reversible,
the heat transferred may be determined by the Second Law of Thermodynamics.
Heat rejected = Q
r
= mT
1
(S
2
s
1
)
Heat absorbed = Q
a
= mT
3
(s
4
s
3
)
where m is the mass of gas it' the refrigerator cylinder. By the First Law,
W
net
= Q
r
+ Q
a
for a cycle, so the coefficient of performance of the ideal Philips
refrigerator is
( ) ( )
a 3
net 1 1 2 4 3 3
Q T
COP
W T s s / s s T

(4.42)
If the working fluid behaves as an ideal gas. we may write
( ) ( )
1 2 1 2 1 2
s s c ln T / T Rln /

+
( ) ( )
1 2 4 3
R ln / Rln /
= s
4
s
3
because
1 2
T T
and
3 4
T T
,
1 4

and
2 3

, where is the gas specific
volume. The coefficient of performance of an ideal Philips refrigerator with an ideal
gas as the refrigerant is
130
3
1 3
T
COP
T T

(4.43)
This is the same expression as that for the COP, of a Carnot refrigerator;
therefore. the ideal Philips refrigerator would have a figure of merit of unity. Frictional
energy dissipation, pressure drops through the regenerator, finite temperature
differences during heat rejection and heat absorption. and finite temperature
differences between the regenerator and the working fluid all tend to lower the figure
of merit in the actual refrigerator.
The first analysis of the details of operation of the Stirling system was
presented by Schmidt (1871). An excellent summary of Schmidt's analysis was given
by Walker (1983). The Schmidt analysis did not include the effects of non-isothermal
compression and expansion and regenerator Ineffectiveness. A more exact analysis
was presented by Finkelstein (1975); however, the application of this analysis requires
the use of a rather involved computer program.
Fig. 4.15. Thermodynamic cycle for the ideal Philips refrigerator.
The Philips refrigerator is constructed by the Philips Company of Eindhoven,
Netherlands. It has been used successfully in the liquefaction of air at a rate of 5.5
liter/hr (11.6 Ib
m
/hr), in gas separation systems, and in miniature cooling systems for
electronic components. The figure of merit for the actual system as contrasted to the
131
ideal Philips refrigerator is about 0.30 when the source temperature is that of liquid air
(79 K or 142R).
5.9. Vuilleumier refrigerator
The Vuilleumier refrigerator, first patented by Rudolph Vuilleumier in 1918,
is similar to the Stirling refrigerator, except the VM refrigerator uses a "thermal"
compressor instead of a mechanical compressor. A variation of the VM device was
also invented by Vannevar Bush (1938), and another version was patented by K. W.
Taconis (1951). A schematic of the VM refrigerator is shown in Fig. 4.16, and the
ideal VM cycle is shown on the temperature-entropy plane in Fig. 4.17.
In the ideal VM cycle, heat is added from a high-temperature source to the
gas in the "hot" cylinder, and the displacer moves downward to maintain the
temperature of the gas constant at T
h
(process 1-2). At the same time, near-ambient
temperature gas flows from the intermediate volume through the regenerator to the hot
volume (process 4-1). The displacer then moves upward and gas is displaced from the
hot volume to the intermediate volume (process 2-3). Heat is rejected from the
intermediate volume to maintain the temperature of the gas in the volume constant at
T. (process 3-4). As the cold displacer is moved to the left, heat is absorbed by the gas
in the cold volume from the low-temperature source to maintain the gas temperature
constant at T, (process 5-6). At the same time, gas from the intermediate volume flows
through the cold regenerator to the cold volume (process 4-5). The cold displacer then
moves back to the right, and gas is displaced from the cold volume through the cold
regenerator to the intermediate volume (process 6-3).
Assuming that all processes are thermodynamically ideal and that the working
fluid may be treated as an ideal gas, the heat-transfer terms may be determined as
follows. The heat added from the high temperature source is
132
( ) ( )
h h h 2 1 h h 2 1
Q m T s s m RT ln /
(4.47)
The heat added from the low-temperature source is
133
( ) ( )
c c c 6 5 c c 6 5
Q m T s s m RT ln /
(4.48)
where m
h
and m
c
are the mass finally within the hot and cold volumes,
respectively. and R is the specific gas constant. Finally, the heat rejected to the
intermediate-temperature sink is given by
( ) ( ) ( ) ( )
a h c a 4 3 h c a 3 4
Q m m T s s m m RT ln / + +
(4.49)
From the temperature-entropy diagram for the cycle, we see that
2
=
3
=

6
and
1
=
4
= U
5
.
Because the net heat transfer for the system is zero (there is no external work
done on or by the gas, for the ideal system), Q
h
+ Q
c
+ Q
a
= 0, and
( ) ( ) ( ) ( )
h h 2 1 c c 2 1 h c a 2 1
m RT ln / m RT ln / m m RT ln / 0 + +
or
( ) ( )
c h h a a c
m / m T T / T T
Because this system is driven by a heat transfer from a high-temperature
source, instead of by mechanical work, the coefficient of performance must be defined
a little differently. In this case, the COP is
c h c c h h
COP Q / Q m T / m T
Making the substitution for the mass ratio, we find
( )
( )
c h a
h a c
T T T
COP
T T T

(4.50)
This analysis of the performance of the Vuilleumier refrigerator is quite
simplified because the effects of harmonic motion of the displacers, regenerator
ineffectiveness, and other thermodynamic losses are not considered. A more complete
(and more complicated) analysis of the VM refrigerator has been presented by Rule
and Qvale (1969) and by White (1976). In general, the COP of the VM refrigerator is
less than that for a Slirling or Philips refrigerator because a fraction of the heat added
from the high temperature source in the VM refrigerator must be rejected, according to
134
the Second Law of Thermodynamics; whereas all of the mechanical work input can be
utilized in the Philips refrigerator, in the ideal case. Using eqn. (4.50) with
temperatures of T
h
= 800 K (14400R), T
a
= 360 K (648R), and T
c
= 78 K (140AOR),
the COP of the ideal VM refrigerator is 0.1521. For a Philips refrigerator operating
between T
h
= 360 K and T
c
= 78 K, the COP is 0.2766, which is about 82 percent
larger than the COP of the VM refrigerator.
One of the advantages of the Vuilleumier refrigerator is that the thermal input
may be provided by solar energy or isotope energy, which makes the VM refrigerator
attractive for cryogenic cooling in long-duration space exploration and in applications
where mechanical vibration of a drive engine must be avoided (Sherman 1982).
5.10. Solvay refrigerator
The Solvay refrigerator was invented in Germany about 1887 (Solvay 1887),
and was the first system planned for air liquefaction using an expansion engine
(Collins and Canaday 1958). Solvay's prototype apparatus was able to achieve a low
temperature of only 178 K (3200R or - 140F), - so the system was not considered for
cryogenic refrigeration until the late 1950s when Gifford and McMahon (1960) of A.
D. Little, Inc., described the use of the refrigerator in a miniature infrared cooler.
The Solvay refrigerator is shown schematically in Fig. 4.18. If we were to
consider a unit mass of gas as it flows through the system, if would trace out the path
on the temperature-entropy plane as shown in Fig. 4.19. The sequence. of operations
for the Solvay refrigerator is as follows.
Process 1-2. With the piston at the bottom of its stroke, the inlet valve is
opened. The high-pressure gas flows into the regenerator, in which the gas is cooled,
and the system pressure is increased from a low pressure PI to a higher pressure P2'
Process 2-3. With the inlet valve still open, the piston is raised to draw a
volume of gas into the cylinder. The gas has been cooled during its flow through the
regenerator.
Process 3-4. The inlet valve is closed, and the gas within the cylinder is
expanded (isentropically in the ideal case) to the initial pressure Pi- As the gas
135
expands, it does work on the piston, and energy is removed from the gas as work. The
temperature of the gas therefore decreases.
Process 4-5. The exhaust valve is opened, and the piston is lowered to force
the cold gas out of the cylinder. During this process, the cold gas passes through a heat
exchanger to remove heat from the cooled.
Fig. 4.18. Solvay refrigerator schematic.
Process 5-1. The gas finally passes out through the regenerator, in which the
cold gas is warmed back to room temperature.
Assuming that the work output during the expansion process is utilized in the
compression process, the net work requirement for this system is given by
( ) ( ) ( )
net 2 1 2 1 2 c,0 e, m ad 3 4
W / m T s s h h / h h ]
]
(4.51)
where the first term represents the compressor work and the second term represents the
work output during the expansion process. The enthalpy h
4
is the value that would be
achieved at the end of an isentropic expansion from point 3 to the pressure P, at point
4. The energy removed from the low-temperature Source is given by
136
( ) ( )
a 5 4 5 4 ad 3 4
Q / m h h h h 1 h h
(4.52)
Fig. 4.19. Path traced out by a unit mass of gas on the T-s plane for the Solvay
refrigerator.
The expansion piston in this system is similar to t'1e Heylandt piston shown
in Fig. 4.45. The piston is constructed of a poor heat conductor, such as micarta, so
that it may be sealed at the warm end, which helps to avoid the problem of a low-
temperature moving seal.
The miniature Solvay refrigerator described by Gifford and McMahon (1960)
was capable of attaining 55 K (99R) in a single stage. The cylinder had a diameter of
5.6 mm (0.22 in.) and a length of 51 mm (2.0 in.). The working fluid was helium gas,
which varied in pressure between 345 kPa (50 psig) and 1725 kPa (250 psig).
4.11. Gifford-McMahon refrigerator
A schematic of the Gifford-McMahon refrigerator (McMahon and Gifford
1960) is shown in Fig. 4.20. The path of a unit mass of the working fluid on the
temperature-entropy plane is shown in Fig. 4.21. This system consists of a compressor,
a cylinder closed at both ends, a displacer within the cylinder, and a regenerator. This
137
system differs from the Solvay refrigerator in that no work is transferred from the
system during the: expansion process. The displacer serves the purpose of moving the
gas from one expansion space to another and would do zero net work in the ideal case
of zero pressure drop in the regenerator.
The sequence of operations for the Gifford-McMahon refrigerator is as
follows.
Process 1-2. With the displacer at the bottom of the cylinder, the inlet valve is
opened and the pressure within the upper expansion space is increased from a low
pressure PI to a higher pressure p
1
. The volume OJ the lower expansion space is
practically zero during this process because the displacer is at its lowest position.
Position 2-3. With the inlet valve still open and the exhaust valve closed, the
displacer is moved to the top of the cylinder. This action moves the gas that was
originally in the upper expansion space down through the regenerator to the lower
expansion space. Because the gas is cooled as it passes through the regenerator, it will
decrease in volume so that gas will be drawn in through the inlet valve during this
process to maintain a constant pressure within the system.
Fig. 4.20. Gifford-McMahon refrigerator schematic.
138
Fig. 4.21. Path traced out by a unit mass of gas on the T-s plane for the Gifford-
McMahon refrigerator.
Process 3-4. With the displacer at the top of the cylinder, the inlet valve is
closed and the exhaust valve is opened, thus allowing the gas within the lower
expansion space to expand to the initial pressure Pi- The gas that is finally within the
lower expansion space does work to push out the gas that leaves during this process;
therefore, energy is removed as work from the gas finally left in the lower expansion
space. This causes the gas in the lower expansion space to drop to a low temperature.
This process is similar to the expansion process in the Simon liquefier (see Sec. 3.20).
Process 4-5. The low-temperature gas is forced out of the lower expansion
space by moving the displacer downward to the bottom of the cylinder. This cold gas
flows through a heat exchanger in which heat is transferred to the gas from the low-
temperature source.
Process 5-1. The gas flows from the heat exchanger through the regenerator,
in which the gas is warmed back to near ambient temperature.
The net work requirement for this system is given by
( ) ( )
1 1 2 1 2 c, 0
W/ m T s s h h / n ]
]
(4.53)
The energy removed from the low-temperature source is given by
( ) ( ) ( ) ( )
a e 5 4 ad c 5 4
Q / m m / m h h m / m h h
(5.54)
where m
e
, is the mass of gas within the lower expansion space at the end of
the expansion process 3-4, and m is the total mass of gas compressed. Because the
139
volume of the expansion space remains constant during the expansion process. the
mass ratio m
e
/m may be written in terms of the density ratio:
e 4 3
m / m p / p
(4.55)
There arc several factors that contribute to a loss in performance of the
Gifford-McMahon refrigerator (Ackermann and Gifford 1971), including the
regenerator ineffectiveness, thermal conduction down the displacer and Its housing,
"shuttle" heat transfer, and the finite volume within the regenerator. Measurements of
the performance of a small infrared cooler operatll1g on the Gifford-McMahon cycle
showed that the actual refrigeration effect was approximately 59 percent of the ideal
refngcratiui1 effect.
In both the Solvay and the Gifford-McMahon refrigerators, the regenerator is
a critical component, as in the case of the Philips refrigerator. For an efficient
refrigerator, the regenerator effectiveness should be 98 percent or better. Punched
copper or brass screens were used as the regenerator packing material, as shown in Fig.
4.22. To reduce the heat conduction along the length of the regenerator, the punched
screens were separated by a coil of stainless-steel wire. For very low temperature
regenerators, lead may be used instead of copper because lead has a 'higher specific
heat at low temperature due to its lower Debye temperature.
The Solvay and Gifford-McMahon refrigerators have several advantages in
common. The engine valves and displacer piston seals are at room temperature;
therefore, low-temperature sealing problems are eliminated. Through the use of a
regenerator instead of an ordinary heat exchanger, high effectiveness of the heat-
exchange component can be attained, and the system can operate with slightly impure
gas as the refrigeration medium. Because of the back-and-forth motion of the gas
through the regenerator, the impurities are deposited in the regenerator during the
intake process and are swept back out during the exhaust process. Regenerators are
generally less expensive, for a given surface area, than heat exchangers, also.
The Solvay system has two advantages over the Gifford-McMahon system:
(I) The coefficient of performance of the Solvay system is inherently higher than that
of the Gifford-McMahon system because more energy is removed from the working
140
fluid by the external-work-producing process. (2) In the Gifford-McMahon system, a
small motor is required to move the displacer back and forth while the expanding gas
moves the piston in the Solvay system. On the other hand, the Gifford-McMahon
system has some advantages over the Solvay system: (I) There is practically no
leakage past the displacer in the Gifford-McMahon system because of the small
pressure difference across the displacer seals. (2) The displacer and crank arm in the
Gifford-McMahon system need not be designed to support a large force; therefore, the
motion transmission system can be quite simple and subject to fewer problems with
vibration.
Fig. 4.22_ Regenerator schematic. The stainless-steel wire spacer is used to reduce the
longitudinal conduction heat transfer within the matrix
One of the major attractive features of the Gifford-McMahon system is the
ease with which it can be adapted to multi staging. A three-stage refrigerator is shown
in Fig. 5.23. By using helium gas as the working fluid, refrigeration may be achieved
at three different temperature levels with only a slight increase in the complexity of the
overall system. All the valves in the multi-stage system operate at room temperature,
and the 'three displacers are operated by a single actuator. By multi-staging,
temperatures near 15 K (27
o
R) can .be attained with less work than by using a single-
stage system.
141
Example 5.6. A Gifford-McMahon refrigerator operates between the pressure
limits of 101.3 kPa (1 atm) and 1.013 MPa (10 atm) using helium as the working fluid.
The maximum temperature of the space to be cooled is 70 K (126
o
R) and the
temperature of the gas leaving the compressor is 300 K (540
o
R). Assume that the
regenerator is 100 percent effective, and the compressor overall efficiency is 60
percent. The expansion efficiency is 90 percent. Determine the COP for the system.
From the temperature-entropy diagram for helium, we find the following
property values:
h
1
= 629 J/mol = 1572.7 J/g at 101.3 kPa (1 atm) and 300 K (540
o
R)
s
1
= 125.7 J/mol-K = 31.41 J/g
h
2
= 6308 J/mol = 1575.8 J/g at 1.013 MPa (10 atm) and 300 K
Fig. 4.23. Three-stage Gifford-McMahon refrigerator. All three displacers are moved
by the same actuator. Three different levels of refrigeration are possible with this
refrigerator.
s
2
= 106.6 J/mol-K = 26.63 J/g-K
h
3
= 1518 J/mol = 379.2 J/g at 1.013 MPa (10 atm) and 70 K (I 26R)
142
p
3
= 1.71 mol/L = 6.85 g/L
h
4
= 636 J/mol = 158.8 J/g at 101.3 kPa (I atm) and s
4
= s
3
= 19.05 J/g-K
h
5
= 1514 J/mol = 378.2 J/g at 101.3 kPa and 70K
The work requirement per unit mass for the compressor is
( ) ( )
2 1 2 1 2 c, 0
W/ m T s s h h / ]
]
W/ m [(300)(31.41 - 26.63) - (1572.7 - 1575.8)J/(0.60)
W/m = 1437.1/0.60 = 2395.2J/g(1030Btu/Ib
m
)
The actual enthalpy at the end of the expansion process may be determined
from
( ) ( )
4 4 ad 3 4
h h 1 h h +
4
h
= 158.8 + (I - 0.90)(379.2 158.8) = 180.8 J/g = 724 J/mol
At a pressure of 101.3 kPa (I atm) and
4
h
= 724 J/mol, we find the actual
density at the end of expansion to be
4
h
= 0.38 mol/L = 1.52 g/L
The mass ratio is
e 4 3
m / m p / p 1.52/ 6.85 0.2219
The refrigeration effect is
( ) ( )
a e 5 4
Q / m m / m h h
( ) ( )
a
Q / m 0.2219 378.2 180.8
= 43.80 J/g (18.83
Btu/Ib
m
)
The COP for the refrigerator is
COP = Q
a
/W = 43.80/2395.2 = 0.01829
The temperature limits for the system are: sink temperature, 300 K (540
o
R);
maximum Source temperature, 70 K (l26R); and minimum source temperature (at
143
point 4), 32.1 K (57.8
o
R). Assuming that the helium gas is nearly an ideal gas, we may
calculate the ideal COP, for an isobaric Source refrigerator using eqn. (4.9):
( ) ( ) ( )
70 32.1
COP 0.1934
300 ln 70/ 32.1 70 32.1

The figure of merit for this system is
FOM = COP/COP
i
= 0.10829/0.1934 = 0.0946
REFRIGERATORS FOR TEMPERATURES BELOW 2 K
4.13. Magnetic cooling
In the systems discussed previously, either a liquid or a gas was used as the
working substance. For these systems, we are limited to the regions of temperature
above about 0.6 K. To produce this low temperature, we can use only liquid He
4
or
liquid He
3
boiling under reduced pressure because all other materials are solid at 0.6 K.
The low temperature we can attain with boiling helium is determined by the pressure
above the liquid. At I K (1.8R) the vapor pressure of liquid He4 is 16 Pa (0.12 torr),
while the vapor pressure of liquid He4 at 0.6 K is only 37.5 mPa (2.81 X 10-4 torr).
Liquid He3 can be used to reach 0.6 K with a little less effort because the vapor
pressure of liquid He) at 0.6 K is 72.6 Pa (0.545 torr). About 0.4 K (O.7"R) is the limit
we can attain in a practical system because of the difficulties in maintaining such low
pressures with even moderate flow rates. It would be a sad situation for low-
temperature physicists, though, if this were the lowest temperature that could be
reached.
Giauque (1927) and Debye (1926) independently suggested a way to break
this "temperature barrier." They pointed out that a paramagnetic substance could be
used instead of a gas or liquid and that a magnetic field could be used instead of the
expansion of a fluid to attain the low temperatures. If we were to compress a gas at
constant temperature, we should increase the order (or decrease the entropy) of the
system because we move the molecules closer together without increasing their
random velocities. Then, if we were to expand the gas reversibly and adiabatically, we
should not change the degree of order (because the entropy remains constant) of the
system. We should move the gas molecules farther apart, however, so that the random
144
molecular velocities (and therefore the gas temperature) must decrease in order to
maintain the same degree of order (or entropy). When we get a gas to very low
temperatures, there is not much room left for any more ordering of the system because
it is almost as ordered as it can be.
A paramagnetic substance, however, has another way of ordering itself. In the
absence of an external magnetic field, the dipoles of the paramagnetic material are
more or less randomly arranged, even at low temperatures. If we apply a magnetic
field at constant temperature (analogous to compressing a gas isothermally), we shall
tend to align the magnetic moments of the atoms of the paramagnetic material, thereby
introducing order or decreasing the entropy of the material. If the magnetic field is
removed reversibly and adiabatically (corresponding to a reversible adiabatic
expansion of a gas), the entropy remains constant but the alignment of the dipole
moments is not as great as before. To preserve the degree of order (or maintain the
entropy constant), the temperature of the paramagnetic material must decrease. This
process is called adiabatic demagnetization and it is the process that allows us to enter
the temperature region below 0.6 K.
A schematic of an apparatus to carry out the adiabatic demagnetization
process is shown in Fig. 4.24. A paramagnetic salt pellet is suspended in a chamber by
silk or nylon threads. This chamber is initially filled with gaseous helium, and the
chamber is then immersed in a liquid-helium bath. The liquid helium is boiling under
reduced pressure, so its temperature and the temperature of the paramagnetic salt are
about I K. The helium bath is surrounded by a liquid-hydrogen or liquid-nitrogen
shield to reduce the heat transfer from ambient to the helium bath. This entire
assembly is placed between the poles of a powerful electromagnet, which is shaped so
that the field of the magnet is concentrated around the salt pellet. The magnetic field is
turned on and maintained for about an hour to allow the heat of magnetization (similar
to the heat of compression for a gas) to be conducted to the helium bath by the gaseous
helium in the small chamber, thereby maintaining the salt at its original temperature.
When thermal equilibrium is attained, the gaseous helium (which is called an exchange
gas) is pumped away to thermally isolate the paramagnetic salt. The magnetic field is
then removed, and the temperature of the salt drops to a very low value. Temperatures
145
as low as 0.0014 K or 1.4 mK (0.0025"R) have been attained by this method,
according to de Klerk, Steen land, and Gorter (1950). A detailed summary of the
adiabatic demagnetization process is given by White (1979).
Fig. 4.24. Apparatus for carrying out an adiabatic demagnetization process.
This process of adiabatic demagnetization will work only for very low
temperatures because of the magnitude of the lattice thermal effects at temperatures
much above 2 K or 3 K. The lattice entropy must be much smaller than the entropy
associated with the magnetic dipoles of the paramagnetic material if a significant
temperature change is to be achieved. At very low temperatures, the lattice entropy is
given by
s (lattice) = 77.9 R(T/
p
)
3
(4.79)
As we shall see later, the maximum dipole entropy for a simple spin system
(S = ) is given by
s(dipole, H = 0) = R In 2 (4.80)
To ensure the success of the adiabatic demagnetization process, we should
want the lattice, entropy to be 1 percent or so of the dipole entropy. The temperature
for this condition to be true can he found from eqn.
146
( ) ( )
3
p
77.9R T/ 0.01 R ln 2
or, solving for the upper limiting temperature T
0
,
( )
1/ 3
0 p p
T ln 2 / 7790 0.0446 ]
]
(4.81 )
This result indicates that a materia1 with a high Debye temperature
p

would
be advantageous for magnetic cooling.
4.14. Thermodynamics of magnetic cooling
The magnetic process may be analyzed thermodynamically if, in order to
simplify the situation, we consider pressure and volume changes sm.1I enough to be
neglected. In this case, we may write
T ds = du
0
H dI (4.82)
where
0
= 4 10
7
T m/A = permeability of free space in SI units, H
is the magnetic field intensity, A/m, and J is the magnetic moment per unit mass, A-
m
2
/kg The quantity
0
H dI represents the magnetic work per unit mass
corresponding to the volume-change work + p dv for a pure substance.
For pure substance, it can be shown from thermodynamic reasoning (van
Wylen and Sonntag 1976) that
p
p
Tds dh dp c dT T dp
T

| `

. ,
(4.83)
The analgoues expression for a paramagentic substance can be obtained by
replacing the specific volume by the magnetic moment per unit mass I and by
replacing the pressure p by the quantity
0
H:
H 0
H
I
Tds c dT T
T

| `
+

. ,
(4.84)
where c
H
is the specific heat at constant magnetic field intensity (analogous to
c
p
for a pure substance)
For the adiabatic demagnetization process, the entropy of the paramagnetic
material remains constant; therefore, ds = 0. Making this substitution into eqn. (4.84),
147
we may solve for the differential temperature change due to a differential change in the
magnetic field intensity while entropy remains constant.
0
M
H H
T T I
H c T

| ` | `



. , . ,
(4.85)
The magnetic moment may be determined from eqn. (4.89)
J = ng
B
B( )
J = ()( 1.6147)(10
24
)(1.992)(0.9273)( 10
23
)(5.9684)
J = 89.01 Am
2
/kg
We can compare this value with the one obtained by using the Curie law
(which does not hold, in this case, as we shall see). The Curie constant for gadolinium
sulfate may be found in Table 4.4:
C = 263.3 10
6
K-m
3
/kg
Using the Curie law to determine the magnetic moment, we obtain
J = CH/T = (263.3)(10
6
)(320)(10
3
)/(0.50)
J = 168.5 A-m
2
/kg
We see that the correct magnetic moment differs from the one obtained from
the Curie law by
(168.5 - 89.01)(100)/(89.01) = 89 percent which is not negligible.
4.16 Magnetic refrigeration systems
With this, background on the thermodynamic and magnetic properties of
paramagnetic materials, we can now look into the application of adiabatic
demagnetization in maintaining temperatures below 1.0 K (1.8
o
R). Such a
refrigeration system has been developed by Daunt et al. (1954) at Ohio State
University. A schematic of this refrigerator is shown in Fig. 4.26, and its ideal cycle is
shown on 'he tempecature-entropy plane in Fig 4.27. Because the working medium is a
paramagnetic material (iron ammonium a/urn), lines of constant magnetic field
intensity appear on the temperature-entropy diagram instead of lines of constant
pressure. In the ideal case, the refrigerator cycle is a Carnot cycle. However,
148
irreversibilitys due to beat transfer from ambient and the finite time rate of change of
the magnetic field introduce entropy increases during the adiabatic processes and
temperature increases during the ideal isothermal processes. Modifications on the basic
refrigerator have been made by Zimmerman et al. (1962), who used superconducting
magnets instead of ordinary magnets.
Fig. 4.26. Magnetic refrigerator schematic.
149
Fig. 4.27. Thermodynamic cycle for the magnetic refrigerator.
The sequence of operations for the magnetic refrigerator is as follows.
Process 1-2. The magnetic field is applied to the working salt while the upper
thermal valve is open and the lower thermal valve is closed. When the upper thermal
valve is open, heat may be transferred from the working salt to the liquid-helium bath,
thereby maintaining the salt temperature fairly constant. The thermal valve between
the working salt and the reservoir salt is closed so that heat will not flow back into the
low-temperature reservoir during this process.
Process 2-3. Both thermal valves are closed, and the magnetic field around
the working salt is reduced adiabatically to some intermediate value. During this
process, tee temperature of the working salt decreases.
Process 3-4. The thermal valve between the working salt and the reservoir salt
is opened, and the field around the working salt is reduced to zero while heat is
absorbed isothermally by the working salt from the reservoir salt.
Process 4-1. Both thermal valves are closed, and the magnetic field around
the working salt is adiabatically increased to its original value.
150
The energy absorbed as heat from the reservoir salt in the ideal case is given
by
Q
a
= mT
3
(s
4
s
3
) (4.100)
because the process 3-4 is reversible and isothermal ideally. The entropy values may
be determined from the Brillouin expression, eqn. (4.98). The quantity m is the mass of
the working salt, and the subscripts refer to the numbered points in Fig. 4.28. In the
ideal case, the energy rejected as heat from the working salt is given by
( ) ( )
r 1 2 1 2 4 3
Q mT s s mT s s
(4.101)
Applying the First Law to the entire cycle, we find the work requirement for
one cycle:
( ) ( )
net a r 1 3 4 3
W Q Q m T T s s +
(4.102)
From eqn. (4.100) and (4.102), we see that the coefficient of performance for
the ideal magnetic refrigerator is the same as that for a Carnot refrigerator.
Because of the irreversibilitys involved, the ideal performance of the
refrigerator is not attained in practice. The actual performance of the magnetic
refrigerator constructed by Daunt et al. is compared with the ideal performance in Fig.
4.28. The mass of the working salt is 15 g, and the helium bath is maintained at a
temperature of 1.11 K. The sequence of processes is carried out so that one cycle
requires about 2 minutes to complete.
151
Fig. 4.28. Actual and ideal performance of the magnetic refrigerator.
The working salt used in the magnetic refrigerator by Daunt et al. was an iron
ammonium alum salt, and the reservoir salt (which was used as a "thermal flywheel" to
smooth out temperature fluctuations in the space to be cooled) was chromium
potassium alum. These salts have low thermal conductivities; therefore, heat transfer to
and from the salts poses quite a problem. Copper fins and 3-mm (1/8-in.) lengths of
fine copper wire (0.05-mm to 0.08-mm diameter) were embedded in the salt pellets to
improve the heat-transfer situation. About I g of copper wire and I g of silicone
vacuum grease were mixed with the IS g of paramagnetic salt, and the pellet was
formed by pressing under a pressure of 20 MPa (3000 psi). The vacuum grease acted
as a binder and improved the mechanical stability of the salt pellet.
One of the advantages of the magnetic refrigerator is that it can operate
effectively in zero gravity. Because of this characteristic, magnetic refrigerators have
been used to cool infrared bolometers in space systems by NASA (Sherman 1982).
4.17. Thermal valves
One of the critical components of the magnetic refrigeration system is the
thermal valves. In the Daunt-Heer refrigerator, thin lead strips were used as the
thermal valves. It was observed that the thermal conductivity of lead was different in
the superconducting state compared with the normal state, as shown in Fig. 4.29.
152
When the material is below the transition temperature in zero field (and therefore in
the superconducting state), many of the electrons that would ordinarily take part in the
heat-transport process are restricted from doing so because of the quantum
considerations of the superconducting state. The thermal conductivity is not zero in the
superconducting state because there is still energy transport by the lattice (phonon
energy transfer).
When a magnetic field is applied to the lead strips, the lead is driven into the
normal state if the applied field is above the transition value, and the electrons are once
again free to take part in the heat-transport process. The difference between the normal
thermal conductivity and the superconducting thermal conductivity can be as much as
two orders of magnitude. The valve is thus in the "open" position (heat flow can take
place) when it is driven normal by the valve magnet, and the valve is in the "closed"
position (heat flow is restricted but there is some leakage) when the valve magnet field
is removed.
There are other thermal valves that might be used, but they are usually not as
effective as the superconducting valve. Collins and Zimmerman (1953) used a
mechanical contact switch in a magnetic refrigerator. This type of valve has the
advantage that there is zero thermal leakage when the contact is broken. The serious
disadvantage is that there is energy dissipated when the contact is made or broken, and
this energy is too large to be tolerated in the region below 1.0 K. It is difficult for the
refrigerator to remove energies on the order of 40 W (1.4 X 10
-4
Btu/hr) without
having to remove the energy dissipation in the valves, too.
4.18. Dilution refrigerators
The idea that cooling could be achieved by means of dilution of He
3
by super
fluid He
4
was first suggested by H. London (1951). Phase separation in He
3
He
4
mixtures had not been discovered at that time, so there was little interest expressed in
developing a practical dilution refrigerator. After Walters and Fairbank discovered the
phase separation phenomenon in 1956, London presented a practical technique for the
dilution refrigerator (London et al. 1962). Hall et al. (1966) and Neganov et al. (1966)
constructed and operated dilution refrigerators that reached 0.065 K (0.11
o
R) and 0.025
153
K (0.045R), respectively. Commercial dilution refrigerators are now available that
operate at 0.005 K (0.009R), such as shown in Fig. 4.30.
Fig. 4.29. Thermal conductivity of lead in the normal and superconducting states.
A schematic of a He
3
-He
4
dilution refrigerator is shown in Fig. 4.31. The gas
(which is practically pure He) is "compressed" in a vacuum pump from about 4 Pa
(0.03 torr) to a pressure on the order of 4 kPa (30 torr), then cooled in a heatexchanger
and a liquid helium bath at 4.2 K. The gas is next condensed in a bath of liquid helium
boiling at about 1.2 K. The liquid He
3
expands through a constriction (capillary tube)
and is cooled further in the still, which operates at about 0.6 K. The liquid is again
cooled in another heat exchanger before entering the mixing chamber, where the He
3
is
mixed with He
4
at temperatures between 0.005 K and 0.050 K.
In the mixing chamber, the liquid separates into two phases-a less dense
concentrated He
3
mixture and a more dense dilute He
3
mixture. The temperature-
composition diagram for He
3
-He
4
mixtures is shown in Fig. 4.32. Numerical values are
tabulated in Table 4.7. The He) molecules "expand" from the concentrated phase into
154
the dilute phase (actually, diffuse through the dilute phase), and the mixture
temperature would tend to decrease; however, heat is added to the mixing chamber
from the low temperature region to maintain a constant temperature. This process is
analogous to the isothermal expansion of a gas, except no external work is done by the
He) in the mixing chamber.
The dilute mixture returns through the heat exchanger to the still, where heat
is added to evaporate the He
3
from the mixture. The concentration of He
3
in the liquid
phase in the still is approximately 1.0 percent, and the composition of the vapor is
around 95 to 98 percent He)
3

The refrigeration effect of the dilution refrigerator may be determined by
application of the First Law to the mixing chamber:
( )
a 3 m i
Q n h h
&
&
(4.103)
where n
3
is the molar flow rate of He
3
, h
m
is the molar enthalpy of the He
3
in
the dilute phase leaving the mixing chamber, and h, is the molar enthalpy of the
practically pure He) entering the mixing chamber. Radebaugh (1967) noted that, for
temperatures below about 0.04 K, the enthalpies could be approximated by'
2
m 1 m
h C T , where C
1
= 94 J/mol-K
2
(4.104)

2
i 2 i
h C T , where C
2
= 12 J/mol-K
2
(4.105)

155

Fig. 4.30. He
3
He
4
dilution refrigerator unit. The system
allows samples with up to eight electrical contacts to be
top-loaded into the working refrigerator. Samples can be
cooled from room temperature to below 10 mK (0.018
o
R)
in 2 hours (courtesy of Oxford Instruments, Osney Mead,
Oxford, England).
Table 4.7: Phase separation temperature curve for liquid He
3
He
4
mixtures
(Radebaugh 1967)
Mole fraction He
3
Temperatur
e
Mole fraction He
3
Temperature
x (K) x (K)
0.0640 0.0 0.12 0.2902
0.0641 0.01486 0.13 0.3166
0.0642 0.0208 0.14 0.3412
0.0643 0.0253 0.15 0.3644
0.0645 0.0323 0.16 0.3863
0.0647 0.0377 0.17 004072
0.065 0.0445 0.18 004272
0.066 0.0606 0.20 0.4647
0.067 0.0721 0.25 0.5460
0.068 0.0817 0.30 0.6097
0.070 0.0970 0.35 0.6558
0.072 0.1102 0040 0.6843
0.074 0.1223 0.847 0.60
156
0.076 0.1335 0.876 0.55
0.078 0.1436 0.904 0.50
0.080 0.1528 0.928 0.45
0.085 0.1747 0.949 0.40
0.090 0.1946 0.968 0.35
0.095 0.2129 0.982 0.30
0.10 0.2301 0.9965 0.20
0.11 0.2616 1.000 0.10
where T
m
is the temperature of the dilute phase leaving the mixing chamber, and T, is
the temperature of the practically pure He) entering the mixing chamber.
A typical He) flow rate for dilution refrigerators is about 1.0 10
4
mol/so.
For an ideal heat exchanger (T
i
= T
m
) and a mixing chamber temperature of 0.040 K,
the maximum refrigeration effect should be
( )
( ) ( )
2
4 6
a
Q 10 94 12 0.040 13.1 10 W 13.1 W


&
One of the more critical aspects in the design of a dilution refrigerator is the
design of the heat exchanger between the mixing chamber and the still. From eqn.
(4.103), we note that the temperature ratio for zero heat addition is T
i
/T
m
= (94/12)

=
2.80. For T
m
= 0.04 K, the largest value of T, for finite heat addition is 0.112 K, or the
maximum temperature difference at the cold end of the exchanger is (T
i
- T
m
) = 0.072
K. Large surface areas per unit volume have been achieved for heat exchangers using
sintered metal elements within the flow passages.
157
MODULE V
CRYOGENIC-FLUID STORAGE AND TRANSFER SYSTEMS
After a cryogenic fluid has been liquefied and purified to the desired level, it
must then be stored and transported. Cryogenic-fluid storage-vessel and transfer-line
design has progressed rapidly as a result of the growing use of cryogenic liquids in
many areas of engineering and science. Cryogenic fluid storage vessels range in size
from small I-liter flasks used in laboratory work up to 106 ml (28,000 gal U.S.) and
larger vessels used to store liquid nitrogen, liquid oxygen, and liquid hydrogen for
industrial use and in space-vehicle ground support systems. Storage vessels range in
type from low-performance containers, insulated by rigid foam or fibrous insulation so
that the liquid in the container boils away in a few hours, up to high-performance
vessels, insulated by multilayer evacuated insulations so that less than 0.1 percent of
the vessel contents is lost per day.
Because the storage and transfer system is considered to be one of the critical
parts of any cryogenic system, many examples of cryogenic-fluid storage-vessel
design have appeared in the literature (Hallett et al. 1960; Wilson 1960; Eichstaedt
1960; Canty and Gabarro 1960; Zenner 1960). In this chapter, we shall consider the
basic design approach for conventional storage vessels and transfer lines, along with
the auxiliary components used in storage and transfer of cryogenic fluids.
CRYOGENIC-FLUID STORAGE VESSELS
5.1. Basic storage vessels
In 1892 Sir James Dewar developed the vacuum-insulated double-walled vessel that
bears his name today (Dewar 1898). The development of the dewar vessel (which is
the same type of container as the ordinary Thermos bottle used to store coffee, iced
tea, etc.) represented such an improvement in cryogenic-fluid storage vessels that it
could be classed as a "break-through" in container design. The high-performance
storage vessels in use today are based on the concept of the dewar design principle-a
double-walled container with the space between the two vessels filled with an
insulation and the gas evacuated from the space. Improvements have been made in the
158
insulation used between the two walls, but the dewar vessel is still the starting point for
high-performance cryogenic fluid vessel design.
Fig. 5.1. Elements of a dewar vessel.
The essential elements of a dewar vessel are shown in Fig. 5.1. The storage
vessel consists of an inner vessel called the product container, which encloses the
cryogenic fluid to be stored. The inner vessel is enclosed by an outer vessel or vacuum
jacket, which contains the high vacuum necessary for the effectiveness of the
insulation and serves as a vapor barrier to prevent migration of water vapor or air (in
the case of liquid hydrogen and liquid helium storage vessels) to the cold product
container.
The space between the two vessels is filled with an insulation, and the gas in
this space may be evacuated. In small laboratory dewars, the "insulation" consists of
the silvered walls and high vacuum alone; however, insulations such as powders,
fibrous materials, or multilayer insulations 'are used in larger vessels. Since the
performance of the vessel depends to a great extent upon the effectiveness of the
insulation, we shall devote a section to the discussion of insulations used in cryogenic-
fluid storage and transfer systems.
There is no need for fill and drain lines for small laboratory containers (the
fluid is simply poured in or out through the open end of the container); however, a fill
159
and drain line (which may be two separate lines or a single line) is necessary for larger
vessels. A vapor vent line must be provided to allow the vapor formed as a result of
heat inleak to escape. In addition, some method must be provided to remove the liquid
from the container. Liquid removal may be accomplished either by pressurization of
the inner container or by a liquid pump. If pressurization of the inner vessel is used, a
vapor diffuser must be incorporated in the vent line in order that that warm
presurrization gas be distributed within the ullage space (vapor space above the liquid)
and in order that the warm gas be directed away from the surface of the cold liquid to
reduce recondensation of the pressurization gas. Antislosh baffles are employed in
transportable vessels to damp motion of the liquid while the container is being moved.
A suspension system must be used to support the product container within the vacuum
jacket.
The design capacity and design pressure for a storage vessel is usually
established by the storage requirement of the user. When large storage vessels first
came into use, most were custom-tailored for the specific use. However, most
cryogenic-vessel manufacturers have reached the point that a set of standard-size
vessels is available. These standard units are generally more economical than specially
made vessels.
Cryogenic-fluid storage vessels are not designed to be completely filled for
several reasons. First, heat inleak to the product container is always present; therefore,
the vessel pressure would rise quite rapidly because of vaporization of the liquid if no
vapor space were allowed. Second, inadequate cool-down of the inner vessel during a
rapid filling operation would result in additional boil-off, and the liquid would be
percolated through the vent tube if no ullage spa~ were provided. A 10 percent ullage
volume is commonly used for large storage vessels. This means that a nominal 106 m)
(28,000 gal U.S.) dewar actually has an internal volume of 116.6 m) (30,800 gal U.S.).
Cryogenic-fluid storage vessels may be constructed in almost any shape one
desires-cylindrical, spherical, conical, or any combination of these shapes. Generally,
one of the most economical configurations is the cylindrical vessel with either dished,
elliptical, or hemispherical heads or end closures. Spherical vessels have the most
160
effective configuration as far as heat inleak is concerned, and they are often used for
large-volume storage in which the vessel is constructed on the site.
A cylindrical vessel with a length-to-diameter ratio of unity has only 21
percent greater surface area than a sphere of the same volume, so the heat-inleak
penalty is not excessive for a cylindrical vessel compared with a spherical vessel.
Cylindrical vessels are usually required for transportable trailers and railway cars
because the outside diameter of the vessel cannot exceed about 2.44 m (8 ft) for
normal highway transportation. For shop-fabricated stationary vessels that are shipped
to the site by rail, the maximum diameter depends upon the route taken, but generally
not more than 3.05 m (10 ft) to 3.66 m (12 ft) can be accommodated.
5.2. Inner-vessel design
The details of conventional cryogenic-fluid storage- vessel design are covered
in such standards as the ASME Boiler and Pressure Vessel Code, Section VIII (1983),
and the British Standards Institution Standard 1500 or 1515. Most users require that
the vessels be designed, fabricated, and tested according to the Code for sizes larger
than about 250 dm
3
(66 gal U.S.) because of the proved safety of Code design.
The product container must withstand the design internal pressure, the weight
of the fluid within the vessel, and the bending stresses as a result of beam-bending
action. The inner vessel must be constructed of a material compatible with the
cryogenic fluid; therefore, stainless steel, aluminum, Monel, and in some cases copper
are commonly used for the inner shell. These materials are much more expensive than
ordinary carbon steel, so the designer would like to make the inner-vessel wall as thin
as practical in order to hold the cost within reason. In addition, a thick walled vessel
requires a longer time to cool down, wastes more liquid in cool-down, and introduces
the possibility of thermal stresses in the vessel wall during cool-down. For these
reasons, the inner vessel is designed to withstand only the internal pressure and
bending forces, and stiffening rings are used to support the weight of the fluid within
the lower vessel.
5.3. Outer-vessel design
Because the outer shell of a dewar vessel has only atmospheric pressure
acting on it, one could erroneously think that the shell thickness could be made quite
161
small. Indeed, if were used to determine the shell thickness, ridiculously small
thickness values would be obtained. Actually, the outer shell would not fail because of
excessive stress, but it would fail from the standpoint of elastic instability (collapsing
or buckling). A thin cylindrical shell will collapse under an externally applied pressure
at stress values much lower than the yield strength for the material.
Table 5.1. Typical mechanical properties of metals
Density Youngs modulus
Metal kg/m
3
Ib
m
/in
3
GPa psi Poisson's
Ratio
Carbon steel 7720 0.279 200 29 10
6
0.27
Low-alloy
steel
7830 0.283 200 29 10
6
0.27
Stainless
steel
7920 0.286 207 30 10
6
0.28
Aluminum 2700 0.098 69 10 10
6
0.33
Copper 8940 0.323 117 17 10
6
0.33
Monel 8830 0.319 179 26 let 0.32
5.4. Suspension system design
One of the critical factors in the design of an effective cryogenic-fluid storage
vessel is the method used to suspend the inner vessel within the outer vessel. A poor
suspension system can nullify the effect of using a high performance insulation. Some
commonly used suspension systems are: (I) tension rods of high-strength stainless
steel, (2) saddle bands of metal or plastic, (3) plastic (Micarta, for example)
compression blocks, (4) multiple-contact supports (stacked discs), (5) compression
tubes, and (6) wire cables or chains.
The inner-vessel suspension system is subjected to the weight of the inner
vessel and its contents plus dynamic loads that arise in transporting the vessel,
earthquakes, and so on. Even if the storage vessel is a stationary vessel, it must
withstand dynamic loads when empty during shipment if the vessel is not constructed
on site. Seismic loads are possible for a vessel constructed on site.
162
Fig. 5.2. Typical methods of supporting the inner vessel within the outer
vessel in a dewar.
Some typical acceleration loadings that have been specified in design are
given in Table 5.2. These loadings are given in multiples of the local acceleration due
to gravity, so that a container subject to a 2-g acceleration load has a force equivalent
to twice the container weight acting on the suspension system.
Table 5.2. Acceleration loads specified in suspension system desi.gn for
cryogenic-fluid storage vessels
Vertical Vertical Transverse, longitudinal
Type of unit Up,g Down, g g g
Stationary storage vessels:
Empty 0.5 3 0.5 5
Full 0.5 1.5 0.5 0.5
Full with blast loading 3 5 4 4
Transport trailers:
Small (below 4 m' or 1060 2 5 4 8
gal U.S.)
Large (above 4 m') 4 2 4
Let us analyze the suspension system loads for the case of high-strength
tension rods. The suspension rods are arranged such that the rods support tensile loads
163
only. The forces for vertical-down loading, vertical-up loading, and transverse loading
(two cases) are shown in Fig
Fig. 5.3. Dynamic loading conditions for support system If we apply a force balance
for the case of the vertical-down loading, we obtain
2F W N
g
W = 0
Or, the force in one set of vertical rods is
F (1 + N
g
) W/2
where N
g
= acceleration load
W = weight of the inner vessel and its contents
By applying a similar force balance and moment balance (for the transverse
loading), the forces in the suspension system can be obtained for the other cases. The
results are summarized in Table
If the angle between the longitudinal rods and the vessel wall is denoted by
, then the acceleration load for the longitudinal rods is
1 g
F N W/ cos
164
Two sets of longitudinal rods must be used-one set for forward loading and
another set for rearward loading.
For safety reasons, generally two or more rods are used at each support point.
The support rods are attached at the main support rings on the outer vessel and at one
of the inner-vessel stiffening rings. Resistance to heat transfer down the rods may be
achieved by using springs or washers at either end or both ends of the support
members. Spherical washers may be used to reduce misalignment of the rods resulting
from thermal contraction of the inner vessel. A typical support bracket is shown in Fig.
5.5. Piping
Piping necessary to remove liquid from the container, vent vapor from the
vessel, and so on, introduces a source of heat in leak to the product container. With a
properly designed piping system, the heat transfer down the piping is due to
conduction along the pipe wall only. For this reason, the piping runs should be made as
long as possible, and thin walled pipe should be used. Schedule 5 pipe (the thinnest
piping available in 304 stainless steel) or schedule 10 pipe is typically used in many
larger cryogenic-fluid storage vessels. The thermal contraction of the piping runs must
be considered in the piping system design also.
Fig. 5.4. Cryogenic-fluid storage-vessel piping arrangements
Four piping arrangements are shown in Fig. 5.10. Arrangement 1 is
commonly used for multilayer-insulated vessels. A long length of pipe is obtained
between the warm outer vessel and the cold inner vessel by extending the vacuum
space around the pipe back into the inner vessel. Arrangement 2 is one of the "don'ts"
in cryogenic vessel design-a poor arrangement. Condensation of vapor will take place
165
along the top of the horizontal portion of the line exposed to fluid within the inner
vessel. This liquid will flow along the horizontal section of the line to the warmer
portion of the pipe, at which point the liquid will evaporate. Heat will be transferred to
the inner container by the quite efficient boiling-condensation convection process, and
a large heat-transfer rate will result. In addition, no provision is made for thermal
contraction of the line; therefore, high thermal stresses will be produced in the pipe. In
arrangement 3, a vertical rise in the vacuum space is used to prevent the convection
problem present in arrangement 2, and an expansion bellows is introduced to allow for
thermal contraction. Arrangement 3 presents a serious problem in reliability of the
container, however, and this arrangement should not be used if possible. If a leak into
the vacuum space should occur, it would most likely occur in the bellows because of
fatigue loading of the bellows during repeated cool-down. With the bellows located as
in arrangement 3, the leak would be quite difficult and costly to repair. The best piping
arrangement is shown in arrangement 4, in which the expansion bellows is located in
the stand off, where the bellows is easily accessible for repair. A vertical rise is also
incorporated in arrangement 4 to prevent percolation of the liquid within the line.
The minimum wall thickness for piping subjected to internal pressure is
determined according to the ASA Code for Pressure Piping by the of lowing
expression:
0
a
pD
t
2s 0.8p

+
where p = design pressure
D
0
= outside diameter of pipe
s
a
= allowable stress of pipe material
5.6. Comparison of insulations
A comparison of the advantages and disadvantages of the insulations used in
cryogenic systems is given in the following summary.
Advantages Disadvantages
1 Expanded foams Low cost. High thermal contraction.
166
No need for rigid vacuum jacket.
Good mechanical strength.
Conductivity may change with time.
2 Gas-filled powders and liberous
materials Low cost.
Easily applied 10 irregular shapes.
Not flammable.
Vapor barrier is required.
Powder can pack and conductivity is
increased.
3 Vacuum alone
Complicated shapes may be easily
insulated.
Small cool-down loss.
Low heat flux for small thickness
between inner and outer vessel.
A permanent high vacuum is
required. Low-emisivity boundary
surfaces needed.
4 Evacuated powders and fibrous
materials Vacuum level less stringent
than for multilayer insulations.
Complicated shapes may be easily
insulated.
Relatively easy to evacuate.
May pack under vibratory loads and
thermal cycling.
Vacuum filters are required. Must be
protected when exposed to moist air
(retains moisture).
5 Opacified powders
Better performance than straight
evacuated powders.
Complicated shapes may be easily
insulated.
Vacuum requirement is not as
stringent as for multilayer insulations
and vacuum alone.
Higher cost than evacuated powders.
Explosion hazards with aluminum in
an oxygen atmosphere.
Problems of settling of metallic
flakes.
6 Multilayer insulations
Best performance of all insulations.
Low weight.
Lower cool-down loss compared with
powders.
Better stability than powders.
High cost per unit volume. Difficult
to apply to complicated shapes.
Problems with lateral conduction.
More stringent vacuum requirements
than powders.
5.7. Vapor-shielded vessels
Another method of reducing the heat inleak to a cryogenic-fluid storage
vessel is to ~se the cold vent gas to refrigerate an intermediate shield, as shown in Fig.
7.17. T e escaping vent gas intercepts some of the heat that would otherwise d its way
to the product liquid. The effectiveness of this method' reducing the heat inleak
depends upon the ratio of sensible heat abs ed by the vent gas to the latent heat of the
fluid, as indicated in the following analysis.
The heat-transfer rate from ambient to the vapor shield through all paths may
be written
167
( ) ( ) ( )
2 s 2 2 s 2 2 1 s 1
Q U T T U T T T T

]
]
&
where the coefficient U
2
may be determined from
( ) ( ) ( )
2 t t t
ins sup piping
U k A/ x k A/ x k A/ x + +
where k
t
is the thermal
conductivity for the insulation, supports, or piping; A is the heat-transfer area for each
of these components; and x is the length of conduction path (thickness for the
insulation, support length for vessel supports, an~ piping length for piping). The heat-
transfer rate between the shield and the product container may be written in a similar
manner:
( )
s 1 1 s 1 g fg
Q U T T m h

&
&
where
g
m&
is the mass flow rate of boil-off vapor and h
fg
is the heat of vaporization of
the fluid. Assuming that the vent gas is warmed up to the shield temperature within the
shield flow passages, the sensible heat absorbed by the vent vapor is
( )
g g p s 1
Q m c T T
&
&
From an energy balance applied to the shield, we find
2 s s 1 g
Q Q Q

+
& & &
U
2
[(T
2
T
1
) (T
s
T
1
)] = U
1
(T
s
T
1
) + U
1
c
p
(T
s
T
1
)
2
/h
fg
Let us introduce the following dimensionless parameters:
( )
1 p 2 1 fg
c T T / h
2 1 2
U / U
( ) ( )
s 1 2 1
T T / T T
Making these substitutions, we obtain the following expression for the
dimensionless temperature of the shield:
( )
2
1 2 2
1 1 0 + +
or
( )

2 1 2
2
1 2
2
1 4
1 1
2
1

]
+
] +
' '

] +

]

168
Fig. 5.5. Vapor-shielded cryogenic-fluid storage vessel.
169