SMITHSONIAN MISCELLANEOUS COLLECTIONS
PART OF VOLUME XLIX
Ibobokins jfunb
RESEARCHES ON THE ATTAINMENT
OF VERY LOW TEMPERATURES
PART II.—FURTHER NOTES ON THE SELF INTEN-
SIVE PROCESS FOR LIQUEFYING GASES
BY
MORRIS W. TRAVERS, D.Sc, F.R.S.
Professor of Chemistry in University College, Bristol, England
AND IN PART BY
A. G. C. GWYER, B.SC, AND F. L. USHER
%S^
(No. 1652)
CITY OF WASHINGTON
PUBLISHED BY THE SMITHSONIAN INSTITUTION
1906

RESEARCHES ON THE ATTAINMENT OF VERY
LOW TEMPERATURES.
PART II.i—FURTHER NOTES ON THE SELF INTENSIVE
PROCESS FOR LIQUEFYING GASES.
BY MORRIS W. TRAVERS, D.SC, F.R.S.
PROFESSOR OF CHEMISTRY IN UNIVERSITY COLLEGE, BRISTOL,
AND IN PART
BY A. G. C. GWYER, B.SC, AND F. L. USHER.
I. Introduction.
In 1903 I submitted to the Smithsonian Institution an account of
some researches on the Hquefaction of air and hydrogen which I had
carried out with the aid of a grant from the Hodgkins Fund. Since
then, with the aid of a further grant from the same source, I have
extended these investigations, and have obtained the results which are
recorded in the following pages.
I may state at the outset that I have been obliged to confine my
attention mainly to points of practical interest. After completing the
first section of my work I began to make preparations for the complete
investigation of the process of liquefying gases by the self-intensive
process, and, in conjunction with Professor R. A. Lehfeldt, commenced
a research on the determination of the Joule-Thomson effect for gases
over a wide range of temperature and pressure. My removal to Bris-
tol prevented me from continuing to participate in this work, which
Professor Lehfeldt is now completing. My work was also consider-
ably retarded by the fact that a compressor and air liquefier, which
had been purchased by the Bristol University College shortly before
my appointment, was found to be practically worthless, and only after
the compressor had been completely reconstructed and a new air lique-
fier had been built, could any research be undertaken. With the means
al my disposal this work occupied more than a year.
II. Previous Result.
Before going further I will make one or two remarks as to state-
ments in my last paper. In calculating the quantity of the liquid pro-
^ Part I was published in 1904 in Volume XLVi, Smithsonian Miscellaneous
Collections, under the title : Researches on the Attainment of Very Low Tem-
perature, by Morris W. Travers, D.Sc.
I
2 TJIE ATTAINMENT OF VERY LOW TEMPERATURES
duced by a Hampson machine {loc. cit., p. 13), I have certainly over-
estimated the " loss due to heat absorption by liquefier." This is put
down as equivalent to 100 gms. of liquid per half-hour and from ex-
periments with a Hampson machine fitted with a vacuum vessel sur-
rounding the regenerator coil, so as to reduce the heat-absorption to
an almost negligible quantity, it appears that this amount is more than
twice as large as it should be. The theoretical yield of liquid air is
probably nearer 7.5 per cent, for the conditions investigated, or 0.045
per cent, per atmosphere pressure.
The formula given on page 14 of the same paper indicates that the
yield of liquid should be about 0.06 per cent, per atmosphere, a result
which is also too high, and this is due to the fact that the value k,
which represents the Joule-Thomson effect in the equation, is not, in
all probability, independent of the pressure.
I had intended to investigate the influence on the efficiency of the
apparatus of the volume of air flowing through it, and of the initial
pressure, temperature, etc. Since I commenced these experiments,
however, Messrs. Bradley and Rowe (Physical Review, XIX., pp. 330
and 387) have carried out an investigation on these lines with a lique-
fier of the Hampson type, and as their work has been carried out in
a most careful manner it is unnecessary for me to say more than that
my own results are in complete agreement with theirs. The greater
part of my work, however, refers to the relative behaviour of liquefiers
of different construction.
HI. The Efficiency of the Regenerator Coil in Self-Intensive
Liquefiers.
In the various forms of my apparatus which I have described in my
former communication to the Smithsonian Institution, the hydrogen has
been cooled to the temperature of liquid air boiling in z'actio before en-
tering the regenerator coil. The form and dimensions of the latter were
determined solely from considerations based on experiments with the
air liquefier, and it must be considered that my success with my first
hydrogen liquefier was largely due to good fortune. In this machine
the regenerator coil was made of copper pipe, 2 mm. inside and 3.5
mm. outside, wound as closely as possible in flat spirals to form a cylin-
der 180 mm. long and 50 mms. in diameter around a brass tube, which
supported the expansion valve at its lower end.
In a second liquefier, which was built for Professor Anschiitz of
Bonn, by Brins Oxygen Company, the regenerator coil was of the
same dimensions, but was constructed differently. It consisted of two
copper tubes, of the same diameter, wound together, so that the cylin-
der, so formed, consisted ultimately of two coaxial coils which were
connected in parallel, with the coil in the liquid air chamber above.
THE ATTAINMENT OF VERY LOW TEMPERATURES 3
and with the valve below. The doubling of the coil was intended to
diminish any tendency on the part of the pipe to become blocked with
impurity separated from the gas, and to reduce friction in the pipe,
and so increase the pressure at the expansion valve. These precau-
tions are, as a matter of fact, totally unnecessary.
The results obtained with this liquefier were far from satisfactory,
and a careful study of its behaviour showed that the fault lay entirely
in the regenerator coil. It was evident that the preliminary cooling
of the compressed gas was as complete as in the case of my first ma-
chine, probably even more so ; but that the heat-interchange between
the compressed and expanded hydrogen in the regenerator coil was
unsatisfactory. Careful consideration of the problem soon led to an
explanation of the inefficiency of the second regenerator coil, and it is
this point I shall next discuss.
The phenomena connected with the interchange of heat between
fluids in motion has been studied chiefly with a view to applying the
results to steam boiler problems. Osborne, Reynolds, Stanton, and
others have shown that when currents of water are made to flow in
opposite directions through concentric tubes, the change of tempera-
ture is independent of the velocity of the streams, or that the heat-
interchange is proportional to the relative velocities. The same re-
marks apply to the change of temperature and loss of heat by furnace
gases in passing through boiler tubes ; and to show how closely the
phenomena of heat regeneration in gas liquefiers resembles those I
have just referred to, it is sufficient to state that in the Hampson air
liquefier the difiference between the temperatures of the air which enters
and leaves the apparatus shows the same dift'erence of temperature
even when the velocity of the stream of air passing through it is greatly
increased.
It is also well known that the loss of heat that furnace gases un-
dergo is considerably diminished if the absolute velocity of the gas be
decreased. It can be shown, for instance, that by increasing the diam-
eter of the tubes of an ordinary boiler, or by spacing the tubes of a
water-tube boiler more widely the heat interchange between the gases
and the water may be considerably reduced.
Were it possible to consider the phenomenon of heat interchange
between two fluids moving in opposite directions on different sides of
a conducting surface one of conduction of heat through the fluid, the
study of it would be a comparatively simple matter. As a matter of
fact, however, this is the case only when the fluids were travelling
slowly, and on account of the low conductivity of gases the heat inter-
change is, under such circumstances, very small. When, however, the
velocity of the fluid bears a certain relationship to the dimensions of
the space through which it is flowing, the longitudinal components into
4 THE ATTAINMENT OF VERY LOW TEMPERATURES
which we may theoretically divide the stream no longer form straight
lines, but break up into eddies. As the result of this the heat passes
more rapidly from the gas to the surface in contact with it, or vice
versa, than it does when the " stream line " flow is maintained. It has
been stated that for the efficient working of a boiler both the furnace
gases and the water must " scrub " the surface separating them.
The formation of eddies in a fluid depends upon its velocity, viscos-
ity, density and on the form of the aperture through which it is flow-
ing. The greater the velocity when once the " critical velocity " is
passed, the greater the tendency to form eddies, which also increase
as the space through which the gas is flowing diminishes. Further,
it is well known that when a gas strikes against a sharp edge or angu-
lar obstacle eddies are formed much more readily than when the stream
of gas meets a curved surface.
The difference of the behaviour of the two regenerator coils re-
ferred to on page 2 may now be explained. The first coil consisted
of a single copper tube, so that the velocity of the compressed gas flow-
ing through it was twice as great as that of the gas in the second coil,
each component of which was half the length. Further, since it was
impossible to wind the double coil as closely as the single coil, the
velocity of the expanded gas passing over the outside of the pipes was
greater in the case of the first coil than in that of the second. As the
tendency to form eddies in the gas increases with the velocity, particu-
larly when the stream passes through narrow openings, such as exist
between the flat spirals of the coil, it is not surprising that the efficiency
of the second coil was lower than that of the first.
I was fortunately able to test this theory by means of trials carried
out on two air liquefiers of the Hampson type. In one of these two
copper tubes were wound together to form the regenerator coil, and
in the other four copper pipes of the same diameter were similarly
wound. The difference in the behaviour of the two liquefiers, which
were otherwise identical, under exactly similar conditions was suffi-
ciently marked to prove my point. The two-coil liquefier gave from
2 to 3 per cent, more liquid air than the four-coil liquefier, and whereas
in the first case the difference between the temperatures of the air as
it entered and left the apparatus was 0.4°, it rose in the second case
to 1.4°.
It appeared, therefore, that in constructing regenerator coils the
pipe should have as small an internal diameter as is possible, and that
the coil should be closely wound. It is practically impossible to use
copper tube of a diameter less than 2 mm. inside and 3.5 mm. outside
on account of the mechanical difficulties which arise in joining the
sections, and in winding the coil. The spacing of the spirals is theoret-
ically limited by the fact that the glass vacuum-vessel enclosing the
THE ATTAINMENT OF VERY LOW TEMPERATURES 5
coil can not be subjected to a back-pressure of over one atmosphere.
I have, however, found it to be mechanically impossible to wind
the spirals too closely.
It is practically impossible to further increase the rate at which
the compressed gas loses its heat, but by means of an arrangement
which I shall next describe I have under certain circumstances suc-
nOPOOOC
a
^.QQoooo
0"-^"^-
Fig. I.—Eddies formed by
passage of gas through
orifice.
D'
Fig. 2.—Regenerator coils.
oxg B
pooooo
ceeded in increasing the refrigerating efficiency of the expanded gas
as it passes over the outside of the coils.
I have already referred to the fact that when a stream of gas passes
across the edge of a plate at right angles to its path, or enters or leaves
an orifice with square-cut edges, eddies are formed as in the figure i.
It occurred to me that by placing sheets of perforated vulcanized fibre
between the horizontal spirals, which make up the coil, such eddies
would be set up in the gas, and that the jets of gas passing through
the holes would themselves form eddies both by their mutual action,
and by impinging on the copper coils. Such an arrangement would
be much more effective in overcoming the tendency of the gas to flow
in streams, parted rather than broken up by the curved surface of the
pipes. Further, the fibre being a highly non-conducting material
would insulate the adjacent sections of the regenerator coil, and thereby
increase its efficiency.
In the early part of the year 1904 I constructed two similar regen-
erator coils, each 20 cm. long and 7.2 cm. in diameter, wound about
a tube I cm. in diameter, surrounding the valve rod. One of the coils,
THE ATTAINMENT OF VERY LOW TEMPERATURES
oooooo
oooooo
Fig. 3.—Apparatus for liquefying air.
/^, had a disc of the vulcanized fibre between each horizontal helix,
the other, B, had not (fig. 2). During the experiments the coil in
use was placed inside a vacuum vessel, as in figure 3, arrangements
being made for measuring the air passing through the apparatus, and
THE ATTAINMENT OF VERY LOW TEMPERATURES 7
the air liquefied. To determine the temperature gradient in the coil
the ends of thermo-electric junctions, carefully insulated, were placed
at the top of the coil, near the valve, and at two points equi-distant
from the end and well within the coil. Copper-constantan junctions
were used; the potential difference between each of the junctions in
the coil and a junction immersed in ice was measured by the usual
potentiometer method ; the results were interpreted with the aid of
O THE ATTAINMENT OF VERY LOW TEMPERATURES
wires from the thermo junctions, which are not shown in the figure,
passed through a paraffined cork in the vertical opening of H, while
a rubber tube connected with the horizontal opening led to a gas-meter,
by means of which the volume of air was measured.
A brass reducing piece L and two rubber sleeves M and N con-
nected the cylindrical vessel B with the vessel P in which the liquid
air collected. The liquid ran directly from the nozzle of B into P, and
to allow of the escape of the gaseous air which resulted in the evapora-
tion of a little of the liquid, the pinchcock 6^ could, if necessary, be
opened from time to time.
The temperature gradients in the two coils are shown on the curve
in figure 4. The curve a represents the gradient in the coil A, in which
the perforated vulcanized discs had been inserted, and the curve h the
gradient in the simple coil B.
The result of this part of my experiments is interesting. It will
be observed that in the case of both coils the temperature gradient is
at first very steep, but curiously enough the effect of introducing the
fibre discs is, so far as the lower section of the coil is concerned, the
reverse of what was looked for, the temperature gradient is rendered
less steep, not more so. Over the upper section of the coil the fibre
discs produce the desired effect, with the result that as a whole the
coil A produces a slightly more effective heat interchange than the
coil B.
The explanation of these facts is, I believe, as follows : Over the
lower section of the coil, where the rate of heat interchange is normally
very high, though the fibre discs may increase the eddying effect, they
also shield the coils from close contact with gas, and hence their total
effect is negative. Over the upper two-thirds of the coil, where the
rate of heat interchange is low, the shielding factor is insignificant com-
pared with the action of the discs in producing eddy motion, and the
efficiency of the coil is increased.
Table I. Results of Experiments with Coils A and B.
Date.
THE ATTAINMENT OF VERY LOW TEMPERATURES
Quantity of Liquid AiR-CoNvtR.T£D,XX
10
IZ. 3456789
Quantity of Liquid Air Formed. O O O O
lO THE ATTAINMENT OF VERY LOW TEMPERATURES
expected, the efificiency decreases slightly in proportion as the mass-
flow of the air increases. The temperature measurements indicate,
however, that the gradient in the lower part of the coil is practically
independent of the quantity of air passing through it, though the tem-
perature difference at the exit became slightly greater. Bradley and
Rowe (he. cit.) arrive at a similar result.
IV. Application of These Results to the Construction of a
New Air Liquefier.
The result of these experiments seemed to show that the introduc-
tion of the vulcanized fibre discs increases the rate of heat interchange
in the upper part of the coil, where the temperature difference between
the two streams of gas is comparatively small, but is non-effective so
far as the lower third at least is concerned.
The lower part of the liquefier which I constructed is arranged as
in figure 3, but in place of the cap there is a metal cylinder filled with
a coil constructed of two pipes coaxially. This part of the coil is 25
centimetres long and 10 centimetres in diameter. The cylinder enclos-
ing it is secured to the vacuum vessel by means of a rubber sleeve, and
is enclosed in a wooden casing filled with animal wool.
The results obtained with it have been quite satisfactory. The
temperature difference at the inlet and outlet is smaller than can be
measured, and the yield of liquid air is, within the limits of experi-
mental error, as high as has been obtained with any liquefier.
Table II. Result of Experiment with New Air Liquefier.
Date.
THE ATTAINMENT OF VERY LOW TEMPERATURES II
published by Alessrs. Bradley and Rowe. The results of their experi-
ments are tabulated above.
They observe that when these results are plotted (figure 50) the
curve cuts the pressure axis at a point corresponding to a pressure
between 700 and 800 atmospheres, below which no liquid would be
formed. They remark that this may be in some measure due to the
increase in the interchange temperature with fall of pressure, but make
no attempt to connect the figures in the second and third columns.
Now if we take the heat of evaporation of liquid air to be 50 calories
per gram, and its specific heat at constant pressure to be 0.237 calories
per gram, we can assume that a fall of temperature of one degree in
the escaping air is equivalent to the evaporation and warming up to
the room temperature of ^ of its mass of liquid air, where
(50 + 0.237 X 200) A-= ( I — -v) 0.237,
x= 0.25 per cent, per dyne.
If then the interchange temperature were zero over all ranges of
pressure the quantity of liquid air produced in the machine would be
:
Pressure.
12 THE ATTAINMENT OF VERY LOW TEMPERATURES
low pressure by opening the expansion valve wide, in which case there
is a steady fall in pressure throughout the length of the coil, a very
much smaller yield of liquid is obtained than when the valve is kept
partially closed and supply of air is reduced. It appears that though
a slight fall of pressure in the gas between the top of the coil and
the expansion valve does not materially effect the working of the
machine, it is a factor which cannot be neglected in estimating its
efficiency.
VI. Application of the Perforated Discs to the Hydrogen
LiQUEFIER.
I now constructed a hydrogen liquefier, of which the regenerator
coil A, with the perforated discs, formed part. This machine did not,
however, yield satisfactory results, for though I obtained some liquid
hydrogen by means of it, the yield was poor. I may add, however,
that the liquefaction commenced very soon after the gas was first al-
lowed to expand. Further, experiments conducted with compressed
air, of which it is not necessary to give the details, pointed to the fact
that the conduction of heat down the coil was considerably decreased
by the introduction of the perforated discs.
The probable explanation of the behaviour of this coil is as follows.
The effect of the discs in producing eddy currents increases the rate
of heat interchange, so long as the temperature difference between the
compressed and expanded gas is not very great, thus increasing the
rate at which the coil cools down at the commencement of the experi-
ment. When, however, the temperature of the expanded gas reaches
the liquefaction point (20.5° abs.), the temperature of the compressed
gas is still above the critical temperature, and we arrive at a condition
similar to that which obtains in the lower part of the coil of an air
liquefier. It appears then that the effect of the perforated discs in
shielding the coils from contact with the expanded gas is greater than
in increasing the eddying and thereby the rate of heat interchange.
VII. Final Form of Hydrogen Liquefier.
In reviewing my experiments with the hydrogen liquefier I am
driven to the conclusion that the form of regenerator coil used in the
construction of my first machine gave better results than any which
I afterwards obtained. In this machine the coil was constructed of
a single tube wound into a coil 150 mm. long and 50 mm. in diameter,
the spaces between the components of each helix being as narrow as
possible.
It appears that the small diameter of the coil and the closeness of
the spacing are more effective in producing a rapid heat interchange,
THE ATTAINMENT OF VERY LOW TEMPERATURES 1
3
by merely increasing the velocity of the flow of the expanded gas, than
are any of the methods by which I have attempted to increase the sur-
face exposed to the gas, or its tendency to form eddies over the surface
of the coil. It is to this type that I have now reverted.
VIII. An Attempt to Employ a Metal Vacuum Vessel in the
Liquefaction of Hydrogen.
As there appeared to be many obvious advantages in employing a
metal vacuum vessel in place of the glass vessel which usually forms
part of the apparatus, I determined to make one, and to carry out some
experiments with it. The vessel was constructed of drawn brass tube
with spun copper connection, as in the figure. After the parts had been
soldered together the whole was silver-plated with the idea of render-
ing the metal absolutely non-porous. The vacuum vessel was secured
by means of a flange and screws to a horizontal plate, which forms part
of the annular space surrounding the liquid air chamber in the lique-
fier. A spiral copper tube connected the inside with the outside, and
terminated below in a valve similar to that one used in the Hampson
air liquefier.
The vacuum vessel was exhausted through a piece of fine copper
tube, hard-soldered into the outer tube. When the exhaustion was
complete this tube was melted in the oxyhydrogen flame and securely
sealed. When the vacuum vessel was half filled with liquid air the
outside did not become cold, but it was noticed that the liquid evap-
orated more quickly than from a glass vessel on account of the heat
supplied to it by conduction down the inner wall.
An attempt to liquefy hydrogen with a machine fitted with the
vacuum vessel proved a failure. This was probably due to the reason
mentioned above, and also to the evaporation of the liquid in passing
through the valve. Only a trace of liquid was collected, and the ex-
periment was not repeated.
IX. On the Preparation of Hydrogen for Use with the
Liquefier.
In my first memoir on the liquefaction of hydrogen I pointed out
that the difficulty introduced by the presence of impurity in the gas
was due rather to the separation of solid at the expansion valve than
to the actual blocking of the coil. I suggested that this was probably
due to the so-called " solvent " properties, which gases under high
pressure and in the neighborhood of their critical temperature are
known to possess.
I have come to the conclusion that while some impurities are highly
detrimental to the working of the apparatus, others are comparatively
14 THE ATTAINMENT OF VERY LOW TEMPERATURES
innocuous. I have observed that when using an air Hquefier, and using
certain kinds of oil to lubricate the compressor, the vacuum vessel in
which the liquid air is collected, when the latter is evaporated, contains
a trace of oil.
After an experiment with a hydrogen Hquefier, the moisture which
remains in the vacuum vessel has often an oily smell. I have also often
detected the odor of arsenuated hydrogen emitted from the solid con-
densed in the vacuum vessel as the apparatus becomes warm. I may
say that I have always obtained the best results by using pure zinc,
pure dilute sulphuric acid with a little copper sulphate, and taking care
to employ a good lubricating oil on the working part of the compressor.
In June, 1903, I was invited to give a demonstration of the lique-
faction of hydrogen before the Congress of Applied Chemistry in Ber-
lin. I was unable to obtain a compressor, but was supplied with a
number of cylinders of hydrogen, prepared by the electrolysis of cal-
cium chloride solution, and compressed to no atmospheres. Six of
the cylinders were at once connected with the Hquefier and the ex-
panded hydrogen was allowed to escape into the atmosphere.
Analysis of a sample of this gas showed that it contained 0.2 per
cent, of oxygen. There was, however, no tendency for solid to form
at the expansion valve and the result of my experiment was the most
satisfactory that I ever reached.
i