Harwood
Engineering Company, Inc.
A Contemporary Version of the Bridgman-Birch 30-kb Apparatus and Certain
Ancillary Devices
by D.H. Newhall and L.H. Abbot
The paper briefly describes the basic Harwood apparatus, particularly its
deviations from the original Bridgman model and the elaboration thereof
by Birch for uses with gases at high temperature. Nearly all the physical
measurements made by Bridgman at 12 kb can be made with this equipment
at 30 kb, in some cases extending the temperature range.
The press portion of this apparatus lends itself to other high pressure
procedures, extending the pressure range and increasing the scope of experiments
without great additional cost. New designs and techniques, some of which
can be employed in a 14-kb isostatic environment, sometimes even at elevated
temperature, open many new paths of experiment, including some studies
of extrusion, and the acoustic measurement of elastic constants.
Two new concepts, the Harwood spherical anvils and the spherical cavity
press are described. They are designed for pressures extending to 100 kb
or more. The Harwood version of the Boyd-England apparatus, which can also
be accommodated in the press, is described. Other more conventional devices
have been accommodated in the press. Some are briefly mentioned, such as
large chambers for more modest pressure.
Some of them are equipped with internal furnaces for temperatures ranging
from 600 to 1500° C.
Evolution of the 30,000 bar Press
After a long period of research within the range 0-12,000 bars Bridgman
constructed two presses for higher pressures. The first (1), capable
of 50,000 bars in a small working cavity, was restricted to only a few
operations. The second (2), designed for 30,000 bars, accepted larger
samples and could be used for nearly all experiments previously performed
at 12,000 bars.
The essential components (shown in Figure 1)
of this latter press were the high pressure vessel, a, whose outside
surface was in the form of a truncated cone fitting into a heavy reinforcing
ring, b, a substantial jack, c, forcing the pressure vessel
into the reinforcing ring, wedge-fashion, to provide lateral support as
internal pressure was built up by a packed piston, d, driven down
into the pressure vessel; and a double press frame whose lower portion
consisted of a base plate, e, middle plate, f, and three
connecting tie rods, g, for containing the thrust of the lower jack
against the reinforcing ring, while the upper portion, an upper plate,
h, the middle plate and three connecting tie rods, buttressed the
upper jack, i, as it advanced the piston.
Subsequently Birch (3) built a 30,000 bar press similar in most
respects to that of Bridgman, but having a high pressure cylinder whose
bore was ¾ inch instead of ½ inch, in order to accommodate
the furnaces reaching 1400°C in gaseous environments. At the request
of a few experimenters, Harwood began the production of the Birch apparatus,
slightly modified. In the first models the bore of the high pressure cylinder
was ¾ inch: bores of 1 or 1½ inch are now available that
increase potentialities in many ways. To extend its usefulness the press
frame has been enlarged. A current model can be operated to 30,000 bars
in the conventional manner or, using the lower portion of the press and
the associated jack, a wide variety of apparatus may be substituted for
the standard high pressure cylinder and its reinforcing ring.
Bridgman remarked that there was no reason why his entire programme
of investigation to 12,000 bars should not be repeated with this 30-kb
apparatus and he intended to do so. Because of the intrusion of other projects
this intent was never completely accomplished. In fact, his 30,000 bar
work included only measurements of electrical resistance (4), of
linear compressibility (5), of viscosity (6), of volume compressibility
to 100,000 bars by means of a secondary internal piston and cylinder (7),
and a great deal of work on plastic flow (8) under tension, compression,
or shear, principally in steels but also in many other metals and some
non-metallics. The principal improvement of the Harwood models is the substitution
of a one-piece massive reinforcing ring instead of the three-piece construction
of Bridgman and Birch: this single piece can now be heat-treated and autofrettaged
fully as effectively as the three pieces originally preferred. Use of the
single ring and the availability of higher-grade steels, including the
maraging variety, has resulted in a very considerable increase in the life
of the high pressure vessels. So far as the authors are aware, none has
ruptured; their usefulness usually terminates because of scoring of the
bore by the piston. A further refinement is the provision for the positive
return of the two rams, which adds to the ease of the operation.
Furnaces
Harwood has supplied with its 30,000 bar units an assortment of furnaces
following the Birch pattern as illustrated in Figure
2a. These have a cavity, a, 3 inches long, and for the 900°C
range an internal diameter of ¼ inch, for the 1000°C range 3/16
inch, and for the 1500°C range, 1/8 inch. These
furnaces are self-contained units that can be loaded outside the press
and lowered into place from the top of the vessel by means of a rod threading
into the removable end piece, b. An essential point is that they
are constructed on a ceramic base, c, consisting of a commercial
electron tube socket. Hollow electrical clips, d, connected to the
heater winding, e, or the thermocouple, f, mate with pins
extending upward from the bottom closure. This feature is common to other
instruments used within the Harwood 30,000 bar apparatus. A special detail
of the furnace is the arrangement of the thermocouples. In order to obtain
maximum working space, and at the same time to secure reliable temperature
readings, the thermocouple wires, encased in capillary alumina tubing,
g, are introduced in slots milled lengthwise in the cores, h,
usually alumina, upon which the heater wire, usually platinum rhodium,
is wound. This assures full availability of the space within the core.
At the high temperatures for which these furnaces were designed an inert
gas must be used as a pressure medium. Iso-pentane, almost the only liquid
which does not freeze at ambient temperature and 30,000 bar, can be heated
to 250°C with hardly perceptible discoloration (9), and because
of its lesser compressibility and greater ease of handling would have some
advantage over gas within that range.
Tension and Compression Devices
There are also in operation the Harwood versions of the Bridgman tension
and compression devices. The former is shown in Figure
2b. It consists of two parts. The first is a specimen holder, a,
of hard steel, so constructed that the applied test force pushes the shoulders
of the dumb-bell-shaped testpiece, b, apart. The second is a combined
load cell and extensometer, c, whose body is, again, of hard steel.
The extensometer shows the elongation of the testpiece as the increment
of resistance of a segment of a small slide wire, d, as it is pushed
past a stationary terminal, e, by the stretching testpiece. The
load on the testpiece is provided by the high pressure piston that advances
to develop a predetermined pressure an then makes contact with a solid
space, f, resting on the testpiece. Further advance of the piston
stretches the testpiece slightly increases the environmental pressure.
The slide wire is a short length of Karma™, an alloy whose high resistivity
is little affected by temperature or pressure. The load cell incorporates
four strain gages, g, applied longitudinally and two tangentially,
bonded to the steel base by the "Rokide" (10) process. For this
technique we are indebted to Mr. T.E. Davidson, of the Watervliet Arsenal.
Because the strain gauges are matched and arranged in a bridge, tension
members opposite each other, and compression members likewise oppose each
other, the effects of temperature on the load cell output are canceled
out and the environmental pressure is effective only after the piston makes
contact with the spacer and load is applied. The pressure at which deformation
begins is controlled by use of a spacer of selected length. The compression
device is not shown. It differs from the tensile only in the manner of
loading the testpiece and relating its deformation to the slide wire.
Cell for 100,000 bars
Since Bridgman has adequately described his apparatus for measuring volume
as a function of pressure to 100,000 bars the briefest treatment will suffice
here. The sample, as shown in Figure 2c is contained
in a cemented carbide cylinder, b, of known cross-section, reinforced
by a shrunk on steel sleeve, c, and is compressed between two miniature
carbide pistons, d. The lower piston rests on a load cell, e,
similar to that used in the tension apparatus, and a slide wire, f,
indicates the compaction of the sample. The whole system is placed within
the tapered cylinder and pressurized in the conventional way. At a selected
pressure the packing head encounters a space, g, whose length is determined
by previous experience, and further advance drives the two miniature pistons
together and motivates the second stage of compression, extending to 100,000
bars, of the sample. The outer surface of the carbide cylinder has a slight
taper and the steel sleeve is driven over it for a few thousandths of an
inch of interference. The greater compressibility of the steel provides
extra lateral support. The carbide cylinder should be ductile, while the
pistons are selected for rigidity. With a tapered cylinder of ¾
inch bore and keeping the same proportions, it is possible to experiment
on a sample more than 3 times as large as that of Bridgman.
Lever Piezometer
The Harwood version (Figure 2d) of the lever piezometer
for measuring linear compressibility differs from that of Bridgman only
in the manner in which it is mounted on the bottom closure. Here again
the plug, h, and socket, k, arrangement adds considerably
to the convenience of the device. In the drawing, a shows the sample,
b the end caps, c the pivot for the lever, d, which
is constrained by the helical spring, e, to move the slide wire,
f, over the fixed contact, g, as the sample contracts under pressure.
Bellows Piezometer
The implement preferred by Bridgman to measure liquid compressibility (11)
(Figure 2e), consisted in essence of a closed bellows,
a, fixed at one end, and having attached at the other a slide wire,
b, mounted axially, Compression of the liquid contracts the bellows,
carrying the slide wire past a fixed contact, c, in the manner described
before.
Free Piston Piezometer
The piston piezometer, Figure 2f, is superior in
two respects to the bellows type; it is not injured by freezing of the
liquid contained, and in combination with a liquid of known compressibility
yields measurement of the compressibility of solid samples. Whereas Bridgman
(12) relied on an extremely good fit between the piston, a,
and the wall of the container, b, Harwood uses an O-ring, c,
seal with good results.
Viscometer
Bridgman's viscosity measurements at 30,000 bar were made by an adaptation
of the falling weight method, in which his press was disposed horizontally
and rotated two-thirds of a turn about its axis, flopping a vane between
two fixed electrical contacts, viscosity being measured as a function of
the time of transit. Even in Bridgman's smaller press it was an awkward
procedure. A new technique (13) has been developed (Figure
2g). It consists in connecting two closed metal bellows, a,
with a capillary, b, dumb-ball fashion. One of the bellows is extended
by means of a solenoid, c, and environmental pressure forces the
second bellows to contract, shifting some of the liquid through the capillary
into the extended bellows. The solenoid is de-energized and return flow
is timed. A slide wire, d, attached axially to the second bellows,
serves as an indicator of the displacement. According to Poiseuille's equation
the dynamic viscosity is given by
m = (pDpR4t)
/ 8Vl
where m is the viscosity, Dp
is the pressure differential between the ends of the capillary caused
by the elastic deformations of the bellows, R is the radius of the
capillary, l the length of the capillary, and V the volume
of liquid flowing through the capillary in the measured time t after
the de-activation of the solenoid.
Extrusion Apparatus
An important variant of the Bridgman-Birch apparatus is the press developed
for the extrusion of metals. One of us recalls seeing a few experiments
wherein Bridgman extruded slugs of various materials into a bucket of sawdust.
It appears that his original interest was in strengthening of the die by
pressure (14). Six years later he devoted a chapter in a book (15)
on wire drawing and extrusion, describing some experimental apparatus and
a few results. He was principally impressed by he large reduction of area
possible in a single pass, but saw no great advantage in the method, though
remarking that there was further work to be done. In recent years, however,
several investigators have realized the practical possibilities of extruding
from an environment of high pressure as an industrial technique. Most of
the work in this field appears to have been done by Pugh* (16) and
his staff at the National Engineering Laboratory and by Vereschagin et
al. in the Physics of SuperPressure Laboratory, Moscow. In America
probably the greatest capabilities now for extrusion under pressure are
that of Davidson and Noland at the Watervliet Arsenal, using a press built
by Harwood, and that of the staff at Battelle Memorial Institute. Watervliet
has been extruding from pressures up to over 20,000 bars. Similar equipment
is just coming into operation at two other laboratories. The principal
features are shown in Figure 3. The billet, a,
is placed within a tapered cylinder, b, that is supported in the
conventional way. The conical lower end of the billet fits into the extrusion
die, c, and forms a seal. A liquid of low viscosity surrounds it
and is compressed by a piston driven down by a jack, d, at the top
of the press. At a selected pressure, which may be as much as 30,000 bars
(determined by the volume of liquid to be compressed), the billet begins
to extrude into a receiving cylinder, e, directly below the die.
This cylinder may be pressurized to 20,000 bars by a piston driven by a
jack, f, carried on the same ram, g, which forces the tapered
cylinder into its massive supporting ring. With the more brittle materials
a homogenous product can only be achieved by extruding into a medium at
high pressure.
Larger Vessels for 20,000 bars
Along with the conventional tapered cylinder for 30-kb service alternative
pressure vessels of 1¼ inch bore, designed for 20 kb, have been
made available. In this instance the pressure vessels resemble the one
for 30 kb, but have only a slight taper and the thick supporting ring is
shrunk on permanently. The lower jack merely provides enough thrust to
hold the closure firmly in place, and to oppose the descending high-pressure
piston. These are useful, for example, where a rock sample, to be reasonably
representative, must have a diameter of 3/8 inch
or more. Furnaces for 1300°C similar to those previously described
have been provided with these cylinders, and in further elaboration there
has been constructed a system involving quartz oscillator plates for measuring
wave velocity in rock specimens (17) at temperatures approaching
600°C, the limit set by quartz. This is depicted in Figure
2h, where a indicates the specimen, b the oscillator
plate(s), c the electrical input, d the ground connection,
and e the furnace core.
Techniques with Lower Portion of Press
Many high pressure techniques have been developed, such as the familiar
opposed anvil pairs (18), the General Electric belt (19),
the Battelle girdle (20), and the National Bureau of Standards tetrahedral
press (21), which require only a single activating force. The lower
portion of the 30-kb press and its jack have been enlarged to accommodate
these as well as other apparatus developed by Harwood.
Spherical Anvils
Prominent among these is a class of devices wherein the sample is compressed
between two anvils. Since Bridgman (22) showed the possibilities
of securing significant results by quasi-hydrostatic pressure developed
between anvils, a host of succeeding experimenters has developed the technique
in the directions of increased pressure range, better understanding of
the pressure distribution across the anvil face, and variety of effects
observed. Except in multianvil devices, the anvils, so far as the authors
are aware, have invariably taken the form of a cylindrical body terminating
in a truncated cone. In certain respects a spherical configuration is superior
to this, for example, for similar diameters of face it yields somewhat
greater strength. Hertz (23) has provided a full treatment of the
situation. The capability of spherical anvils is suggested by Figure
4a which shows the results of crushing loads on anvils made from commercial
bearing balls of SAE 52100 steel. Similarly, it permits accessory devices
to be placed in closer proximity to the sample, as in the apparatus constructed
for Riecker's studies of rock deformation under high temperature and pressure
(24). Figure 4b shows the arrangement of
a high frequency induction coil, a, and the sample, b, with
respect to the anvils c. To some, the most important advantage would
be in economy. Riecker (25) relates a comparison wherein a cemented
tungsten carbide anvil costs 10 times as much as an equivalent spherical
anvil. A more extensive treatment of spherical anvils is in progress, to
be reported hereafter. An application of spherical anvils at present in
use is shown in an apparatus (Figure 4c) designed
to use ultrasonic techniques in measuring the effect of pressure on the
elastic properties of solids. The apparatus is positioned between the lower
ram of the 30-kb press and a spacer buttressed by the middle plate. The
sample, d, may be subjected to as much as 14 kb environmental pressure
within a thick-walled cylinder, g, and is also end-loaded by the
spherical anvils, a, whose optically flat bases against the lower
and upper closures, b and e, provide a seal for the environmental
pressure. The lower closure rests on the lower ram, not shown. The upper
closure abuts the spacer, and is long enough to accommodate any compression
of the sample. The quartz transducer plates, c, are held in place
by springs. Strain gages, f, are bonded to the lower anvil to constitute
a loaded cell. Electrical connections to the transducer and to the strain
gages are not shown; for the latter they are brought through the closure
by means of conventional conical anode couplings. There is room for additional
connections leading to a heating element surrounding the sample; such an
element has been designed but not built.
Hemispherical Press
The prototype of the hemispherical press (26), shown in Figure
4d, was used in several experimental runs to between 60 and 70 kb.
It was made entirely of steels selected for their qualities and properties
to cope with the combined stresses imposed by the design, which in turn
as conceived to limit the stresses imposed on the steel. The only indication
of overload was found in the ball piston that was made of SAE 52100 steel;
it broke subsequent to the test, while resting in storage. Obviously, pressures
still greater could be reached if the ball material were changed to either
a more suitable steel or to tungsten carbide, and further if the chambered
piece were also made of tungsten carbide.
The apparatus is composed of a massive support ring, a, an intermediate
ring, b, the chambered piece, c, with its hemispherical cavity
symmetrically disposed at the top of the centre piece, and a ball piston,
d. Both the intermediate and the chamber pieces are tapered so that
when the sample is under pressure from the ball piston, the tapered pieces
are wedged into each other and into the massive support ring. The greater
the piston force the greater the support forces.
Thick-walled spheres subjected to internal and external pressures develop
stress patterns not unlike those of thick-walled cylinders loaded in the
same way except that the equations involve third-order functions which
are due to Lamé (27). The amount of external pressure necessary
to completely cancel the tangential stress developed by the internal pressure
is given by:
Po = Pi (2 + W3)
/ 3W3
where Po is the external and Pi the
internal pressure acting on a thick hollow sphere of wall ratio (W = outer
diameter/inner diameter). It is seen that for heavy wall ratios Po
has only to be one-third of Pi.
There are other advantages to use of the hemispherical configuration.
It lends itself to scale-up more than other similar devices because less
motion is required of the piston for a given change of volume. For instance,
comparing the strokes required equally to compress equal volumes in cylindrical
and hemispherical cavities, it will be found that the stroke is three halves
greater in the first case. The angles that are useful for developing support
pressure are not necessarily self-locking. When the tapered pieces are
properly lubricated with flake graphite, glycerin and lead foil, they are
easily removed after a run and the charge is readily replaced.
Boyd-England Device
The Harwood version of the Boyd-England (28) device differs in the
design of the massive support cylinder. The original device used concentric
steel rings of identical height pressed together to form a massive support
ring for a tapered inner cylinder of tungsten carbide. The Harwood design,
Figure 4e, also uses concentric rings, a,
but of varying height, increasing toward the outer diameter of the massive
support. The purpose is to make a more rapid diminution of the intensity
of radial stress through the massive support ring and thereby increase
its stiffness. The middle plate of the press is shown at e. Simplicity
and equal effectiveness is achieved by mounting the smaller piston-loading
ram, b, on the larger one, c, used in end-loading the inner
sleeve of tungsten carbide. Otherwise the Harwood device provides a better
opportunity for cooling by means of a water jacket, d.
Conclusion
"There is thus plenty to be done in the field of presently attainable pressures,"
and "...the domain of less than astronomical pressures. The experimental
physicist will hardly be able to exhaust this lower domain in the foreseeable
future." These are the words of Professor Bridgman in the survey which
was his last published paper on the subject of high pressure (29).
The apparatus and devices discussed above are in most instances intended
for use in what we now generally regard as the moderately high pressure
range, the broad domain to which Bridgman referred. It is hoped that in
describing the many possibilities related to the 30-kb press we have suggested
favorable routes for further exploration of this domain.
References
(1) Bridgman, P.W. Phys. Rev. 1935, 48, 893-906; Proc.
Am. Acad. Arts Sci. 1940 74, 21-51.
(2) Bridgman, P.W. ibid. 1938 72, 157-205.
(3) Birch, F., Robinson, E.C., and Clark, S.P. jun. Ind. engng Chem.
1957 49, 1965.
(4) Bridgman, P.W. Proc. Am. Acad. Arts Sci. 1938 72,
157-205.
(5) Bridgman, P.W. ibid., 1940 74, 11-20; 1948 76,
89-99.
(6) Bridgman, P.W. ibid., 1949 77, 117-128.
(7) Bridgman, P.W. Phys. Rev. 1941 60, 351-354; Proc.
Am. Acad. Arts Sci. 1942 74, 425-440.
(8) Bridgman, P.W. Rev. mod. Phys. 1945 17, 3-14; J.
appl. Phys. 1946 17, 201-212.
(9) Kafalas, J. Personal communication.
(10) "Rokide ceramic spray coatings," data sheet, 1961 (Norton Co.,
Worcester, Mass.).
(11) Bridgman, P.W. Proc. Am. Acad. Arts Sci. 1931 66,
185-233.
(12) Bridgman, P.W. Z. Kristallogr. 1928 67, 363-376.
(13) Abbot, L.H. paper delivered to Chem. Inst. of Canada, Chem. Eng.
Div., Montreal, Oct. 1963; U.S. Patent No. 3,263,494.
(14) Bridgman, P.W. Rev. mod. Phys.1946 18, 58.
(15) Bridgman, P.W. Studies in large plastic flow and fracture
1952, 174 (McGraw-Hill, New York and London).
(16) Pugh, H. Ll. D. NEL Rep. No. 142, 1964.
(17) Schreiber, E. and Anderson, O.L. J Am. cer. Soc. 1965 49,
No. 4, 184-190.
(18) Bridgman, P.W. Proc. Am. Acad. Arts Sci. 1952 81,
167-251.
(19) Hall, H.T. Rev. Scient. Instrum. 1960 31, 125; also
Progress in very high pressure research (ed. Bundy, F.P., Hibberd,
W.R.jun., and Strong, H.M.) (Proc. Conf at Bolton Landing, Lake George,
New York, June 1960) 1961, 1-9 (John Wiley and Sons, New York).
(20) Young, A.P., Robbins, P.B., and Schwartz, C.M. High pressure
measurement (ed. Giardini, A.A., and Lloyd, E.C.) 1963, 262-273 (Butterworths,
Washington, and London).
(21) Lloyd, E.C., Hutton, V.O., and Johnson, D.P. J Res. Nat. Bur.
Stand. 1959 63C (No. 1), 59.
(22) Bridgman, P.W. Phys. Rev. 1935 48, 825-847; J.
appl. Phys. 1941 12, 461-469.
(23) Hertz, H. J. Math. (Crelle's J.) 1881 92, 156-171;
Gesammelte Werke 1895 1, 155 (Leipzig).
(24) Riecker, R.E. Rev. scient. Instrum. 1964 35, 1234-1236;
Instrumentation Paper No. 36, Air Force Cambridge Res. Lab., Office of
Aerospace Research, U.S. Air Force, L.G. Hanscom Field, Mass.
(25) Riecker, R.E. personal communication.
(26) U.S. Patent No. 3,268,951.
(27) Lamé, G. and Clapeyron, B.P.E. "Mémoire sur l'équilibre
intérieur des corpes solides homogènes," Memoires par
divers savans 1833 4 (Acad. des Sciences, Paris).
(28) Boyd, F.R. and England, J.L. J geophys. Res. 1960 65,
741.
(29) Bridgman, P.W. Solids under pressure (ed. Paul, W. and
Warschauer, D.M.) 1963, 5,11 (McGraw-Hill, New York and London). Reprinted
as last of P.W. Bridgman's Collected Experimental Papers 1964 (Harvard
University Press, Cambridge, Mass.).
Karma™ is a trade mark of the Driver Harris Company
* This reference includes a bibliography of 51 items, including 6 by
Vereschagin et al.
Harwood Engineering Company, Incorporated – © June, 1999
455 South Street, Walpole, Massachusetts 02081
phone: 508-668-3600
fax: 508-660-2276
http://www.ultranet.com/~harwood/
email: harwood@ma.ultranet.com
