GB2346669A

GB2346669A - High-pressure piston-and-cylinder cell for pressurizing a sample

Info

Publication number: GB2346669A
Application number: GB9903438A
Authority: GB
Inventors: Ian Ross Walker
Original assignee: Individual
Current assignee: Individual
Priority date: 1999-02-15
Filing date: 1999-02-15
Publication date: 2000-08-16
Anticipated expiration: 2019-02-15
Also published as: GB9903438D0; GB2346669B

Abstract

A piston-cylinder pressure cell 1 for pressurising a sample, eg for phase transition investigations at very low temperatures, comprises a cylinder 2 having a bore in which the sample is located and a piston 15, eg of tungsten carbide, to increase pressure within the bore. Force may be transmitted to the piston 15 from a second piston 17 via a pad 16. The sample (not shown) may be mounted on the leading end 10 of electrical feedthrough 9. The cylinder is composed of a cobalt alloy or a cobalt-nickel alloy, eg 20% Cr, 35% Ni, 10% Mo, 35% Co and may be located within a sleeve 3 eg of beryllium-copper. The sample may be covered by a hollow non-metallic cap 12, eg of polytetrafluoroethylene, filled with a pressure-transmitting liquid. The cylinder 2 may be lapped or polished to reduce friction between the cap 12 and the cylinder. A sealing ring 13, composed of a cobalt alloy similar to that of the cylinder, may be located between the piston 15 and the cap 12.

Description

PRESSURE CELL. METHOD AND SEALING RING The present invention relates to a pressure cell, a method of preparing a pressure cell, and a sealing ring.

The frontiers of research in condensed matter physics increasingly involve the use of pressure as a clean and controllable means of altering the interatomic spacing of a given material. This makes it possible, for example, to bring a substance near a phase transition in order to study quantum critical behaviour. Such studies often take place at very low temperatures, and frequently involve the use of magnetic fields. Piston-cylinder cells are often used to create high pressures and have the advantage of providing a relatively large amount of space within which a sample of the material can be located.

A known"simple"long bore piston-cylinder pressure cell (i. e. one not requiring a tungsten carbide inner liner with complicated end supports, etc.) using a compound cell made of 350 and 300 maraging steels allowed a pressure of 50 kbar at room temperature to be achieved. Maraging steels are strongly ferromagnetic. Another known pistoncylinder device, capable of being used in a cryogenic environment, also used a compound arrangement which made use of several different types of maraging steel, and could reach a pressure of 40 kbar at a temperature of 1.4 K.

Various high pressure non-magnetic piston-cylinder cells are known. One of these cells was essentially a monoblock cell made of beryllium-copper and was capable of reaching 30 kbar at room-temperature, but, owing to pressure loss upon cooling, only 25 kbar at 4.2 K. In this known cell, only a fraction of the total length of the bore was used, so that the device is effectively being used in a"short-cell"configuration. Another known cell is a compound"short-cell"which makes use of a non-magnetic nickel"Russian alloy" (40HNU-VI) cylinder in the centre, surrounded by a beryllium-copper sleeve. It has been tested to a pressure of 31.5 kbar at room temperature and 25 kbar at 4.2 K.

Many of the known piston-cylinder cells for cryogenic applications employ beryllium-copper, and especially alloy 25 (C17200), in those components which are subjected to large stresses. This is largely because the material, like most metals with an FCC (face centred cubic) crystal structure, remains strong and tough when cooled to low temperatures. According to the Cryogenic Materials Data Handbook, the tensile strength of a particular 1/2 HT Alloy 25 rod rises from about 1.35 GPa at 293 K to 1.48 GPa at 20 K and the yield strength increases from 1.17 GPa to 1.35 GPa over the same temperature interval. This same source also indicates that the elongation in 2 inches (approx 5cm) rises slightly from 5% to 6%, while the impact energy (as measured using the Charpy K test) increases from about 7 J to 11 J, as the temperature is decreased from 293 K to 90 K. The present applicant is not aware of any information about the behaviour of the fracture toughness of this material as the temperature is reduced. Other properties, such as the modulus of elasticity, the fatigue life, and the fatigue strength, are also increased as the temperature is reduced from 300 K to 20 K. Hence, a reduction in temperature improves the mechanical properties of beryllium copper Alloy 25 in almost every way.

In order to reach the highest pressures at low temperatures, research workers have often made use of various types of maraging steel for the cores of their cells. Like many other steels, and in common with most materials with a BCC (body centred cubic) crystal structure, maraging steels lose toughness as the temperature is reduced. However, unlike other high strength steels with a BCC crystal structure, maraging steel does not seem to undergo a sudden transition from a ductile state to a brittle one when the temperature is lowered. Nevertheless, the ductility and toughness of these materials is very low near 4 K and below.

Data on the mechanical properties of maraging steel in the heat treated or"age hardened"condition at such temperatures is hard to obtain, probably because under these conditions they are nearly worthless for most engineering purposes. At room temperature, the 0.2% yield strength, tensile strength, elongation, reduction of area and Sharpy V-notch toughness of 300 maraging steel in the age hardened condition are respectively: 2.00 GPa, 2.03 GPa, 11%, 57% and 23 J. The fracture toughness (Kic) of 300-grade maraging steel in the age hardened condition is reduced from a room temperature value of 80 MPa. ml/2 to 66 MPa. ml/2 at 77 K. However, the value at 4 K is not known, except that it is known that the toughness of aged maraging steel alloys at low temperatures is marginal. It is also known that for a maraging steel with the designation EHP 921, in the age hardened condition, the 0.2% yield strength, ultimate tensile strength, total elongation, and reduction of area are respectively: 1243 MPa, 1356 MPa, 16.9%, and 55% at 300 K; 1572 MPa, 1869 MPa, 31.7% and 30% at 77 K; and 1763 MPa, 1848 MPa, 1.8% and 9% at 20 K. All these values are for transverse (as opposed to longitudinal) test samples cut from the original billet of material.

Thus, it can be seen that in the case of EHP 921, the ductility is greatly reduced at low temperatures. Since the room temperature elongation and reduction of area of 300 series maraging steel are close to those of EHP 921, one would expect that its behaviour at 20 K might also be similar. In spite of the low toughness of maraging steel in cryogenic environments, the present applicant has used 300 grade maraging steel for high pressure work at low temperatures on numerous occasions without any difficulties. However, the applicant is aware of one example of a maraging steel cylinder cracking while under pressure which probably took place while it was at low temperatures. The applicant's own present maraging steel cells use a compound arrangement, with a maraging steel core surrounded by a shell of beryllium copper Alloy 25.

A high-strength, non-magnetic alloy, with the designation 40HNU-VI and the composition 39-41% Cr, 3.33.8% Al, balance Ni, has been used in pressure cells by experimentalists in the former Soviet Union for many years.

It has an FCC crystal structure and, like beryllium-copper, remains ductile at low temperatures. It is capable of reaching a tensile strength at room temperature of 2.10 GPa, a yield strength of about 2.00 GPa, with an elongation of roughly 7%. Even higher tensile strengths can be attained (up to 2.37 GPa) if one is willing to accept reduced ductility and toughness. (It should be borne in mind that strength and toughness in metals are usually complementary qualities, and an increase of one parameter usually results in a decrease of the other.) The present applicant is not aware of any data concerning the low temperature mechanical properties of this material.

Like maraging steel, alloy 40HNU-VI is primarily strengthened by an age hardening process. However, if the material is cold worked (in order to produce"work hardening") before ageing, hardnesses of about 64-67 Rc can be attained. This compares with a hardness of up to 57 Rc for 40HNU-VI which has not been so treated. Such work hardened material presumably has effectively zero tensile strength, ductility and toughness. Nevertheless, it is understood that it works well in compression, and can be used for such things as indenters and pistons.

However, a major disadvantage of alloy 40HNU-VI is that it is practically unavailable outside the former Soviet Union and is difficult and expensive to obtain with the former Soviet Union.

According to a first aspect of the present invention, there is provided a piston-cylinder pressure cell for pressurising a sample, the cell comprising a cylinder having a bore in which a sample can be located, and a piston movable in the bore to increase the pressure within the bore, the cylinder being composed of a cobalt alloy or a cobalt-nickel alloy.

The cobalt alloy of the cylinder preferably has a composition of substantially 18-22% chromium, 33-37% nickel, 9-11% molybdenum, and balance cobalt, by weight.

The cobalt alloy of the cylinder most preferably has a composition of substantially 20% chromium, 35% nickel, 10% molybdenum, and 35% cobalt, by weight.

The preferred alloy for the cylinder is a high strength, non-magnetic alloy with the designation MP35N.

This has been available for about 30 years, but has not, to the best of the present applicant's knowledge, been used in high pressure piston-cylinder cells. The material has a composition 35% Co, 35% Ni, 20% Cr, 10% Mo. It obtains most of its strength from work hardening and the remainder from an ageing process. The AMS 5844E specification requirements (which cover MP35N bars) state that, following ageing at 538-648 C for 4 hours, the material should have an ultimate tensile strength of at least 1.79 GPa, a 0.2% offset yield strength of at least 1.59 GPa, an elongation in 4D of at least 8% and a reduction of area of at least 35%. However, the typical properties of an AMS 5844E product after ageing at 593 C for 4 hours are respectively: 2.03 GPa, 1.97 GPa, 9% and 45%, with a hardness of 51 HRc. All the above values, and throughout the present specification except where indicated, were measured at room temperature.

All the properties given here are determined by longitudinal tensile tests (i. e. tests in which the bar is subjected to forces along its axis of symmetry). Since the material is usually strengthened mostly by work hardening, it is to be expected that there will be some differences between the mechanical properties in the longitudinal direction and those in the transverse one. The present applicant has no precise information about what these differences might be. It is reasonable to suppose that they are small. Beryllium copper is also strengthened by work hardening, and in the case of Alloy 25 in the HT condition (and cold worked by up to about 38%), it is understood that the difference between the transverse and longitudinal yield strength is no more than 5%. Since MP35N is cold worked by a similar amount (about 50%), the directional dependence of its mechanical properties might be confined in roughly the same way.

The tensile properties of this material depend on its size. For example, a 8.38 mm diameter rod of MP35N has an ultimate tensile strength, 0.2% yield strength, elongation and reduction in area (in the age hardened condition and at room temperature) of respectively: 1.92 GPa, 1.81 GPa, 15% and 55.1%. A 22.87 mm diameter rod of the same material has tensile properties of respectively: 2.01 GPa, 1.92 GPa, 9.97%, and 53.9% in one part of the batch (or"heat"), and 1.99 GPa, 1.90 GPa, 10.6% and 54.3% in another part.

Finally, a 44.43 mm diameter rod of this material (which is the largest available) has tensile properties of respectively : 2.14 GPa, 2.09 GPa, 9.8% and 48.9% in one part of the batch, and 2.03 GPa, 1.97 GPa, 9.0% and 43.2% in another part. These variations are presumably due to the ease with which larger sizes can be work hardened.

There may be an advantage in using raw material of the largest possible size when making a pressure cell, regardless of the size of the cell itself. Furthermore, it is possible in principle to obtain even stronger material by carefully selecting from different parts of a batch, or from different batches.

The crystal structure of MP35N material is FCC, and so it is expected that (in contrast with maraging steel, but like beryllium copper) its cryogenic mechanical properties should be good. In fact, tests have been carried out on 51% work hardened material which has been aged at 566 OC for 4 hours. These revealed that the material had a room temperature (298 K) ultimate tensile strength of 2.08 GPa, 0.2% yield strength of 2.00 GPa, elongation of 9% and reduction of area of 52%; and cryogenic (77 K) values of: 2.46 GPa, 2.31 GPa, 9.5% and 45%, respectively. The present applicant does not have any cryogenic fracture toughness data, but Charpy V-notch impact toughness results for the same material are: 14.6 Joules at room temperature and 14.0 Joules at 77 K. On other material, which had been cold worked by 49% and aged at 650 C for four hours, the tests showed a room temperature (24 C) tensile strength of 1.93 GPa and Charpy V-notch test results of 25.6 J at 24 C, 21.8 J at 77 K and 18.3 J at 20 K. The present applicant does not have any data for lower temperatures, but the toughness of the material is unlikely to fall very much below its 20 K value, and the ultimate tensile and yield strengths are likely to be better than those measured at 77 K.

The material of the cylinder is preferably prepared by work hardening prior to age hardening.

If very high levels of work hardening are performed on this material, the tensile and yield strengths can be greatly improved. For example, work hardening the alloy by 65%, followed by ageing at 538 C for 4 hours, will result in a room temperature tensile strength of 2.48 GPa, and a 0.2% offset yield strength of 2.41 GPa. Unfortunately, the resulting elongation and reduction of area is only 4% and 8% respectively, which places it outside the limits set by the AMS 5844E specification. Such material is not available commercially, but it is possible in principle for the user to subject MP35N material to further work hardening in order to bring up the strength levels. In the case of pressure cells, such work hardening could be carried out by forcing an oversize mandrel through the bore of the cell before ageing. An alternative method would be to repeatedly pressurise and depressurise the cell so as to cause yielding and reverse-yielding, before subjecting it to the ageing process.

It is preferred to do the work hardening and ageing in the order given because, although work hardening is a necessary condition for age hardening, the reverse is not true. In other words, work hardening sets up the metallurgical conditions which are necessary for further hardening by ageing to take place, although the opposite does not occur.

The magnetic properties of a sample of AMS 5844E material in the age hardened condition were analysed with the aid of a commercial SQUID magnetometer. The results of these tests can be seen in Figure 1. The magnetisation curve of the material reveals no hysteresis, so it is not ferromagnetic at least down to 2 K. At a temperature of 4 K and a field of 1 kG, the magnetisation of AMS 5844E material is about six times greater than that of 40HNU-VI Russian alloy under the same conditions).

The electrical resistivity of MP35N decreases from 103 SQ-cm at 294 K to 99 pQ-cm at 89 K, and is unlikely to drop very much below this at 4 K. As a result, MP35N does not strongly screen changing magnetic fields, in contrast with beryllium copper alloy 25 in the age hardened condition, which has a resistivity of 7 pQ-cm at 293 K and 5.5 HQ-cm at 4 K. The 40HNU-VI Russian alloy has a resistance of 80 AE2-cm at 300 K and 64 pQ-cm at 4.2 K. It is evident that MP35N could be useful in experiments involving high pressures in rapidly changing magnetic fields.

It can be seen that the room temperature mechanical properties of AMS 5844E material are very similar to those of 300 series maraging steel. Consequently, it would appear that the alloy could be used as a direct replacement for maraging steel in pressures cells which are to be used at room temperature. Of course, its toughness is likely to be very much better than that of maraging steel in a cryogenic environment, and so from this viewpoint alone it is preferable to maraging steel for low temperature experiments. MP35N work hardens much more readily than maraging steel, and so repeated inelastic deformations of the cell following use at the highest pressures may lead to failure sooner than in maraging steel cells. Of course, beryllium copper also has this problem.

The room temperature mechanical properties of AMS 5844E are also similar to those of the Russian alloy 40HNU VI, although its levels of ductility (elongation) at the same strength values are somewhat higher. On the other hand, the magnetisation of AMS 5844E (at 1 kG and 4 K) is greater than that of 40HNU-VI. Although this would seem to weigh against the use of AMS 5844E, in most experiments at the highest pressure levels (where one is likely to use such materials in pressure cell construction), this advantage is mitigated by the need for slightly magnetic components (e. g. tungsten carbide, in such things as electrical feedthroughs and pistons) in other parts of the cell. Of course, very high strength non-magnetic materials, such as silicon-cemented diamond compacts, can be used for these items, but such materials are expensive and difficult to work with. Furthermore, in comparison with its Russian counterpart, AMS 5844E is very easy to obtain.

The cylinder is preferably located within a sleeve.

The sleeve is preferably composed of beryllium-copper.

The cell may comprise a cap located within the bore and arranged to cover in use a sample located within the bore, the piston being arranged to bear against the cap in use to increase the pressure within the cap.

The cap may contain a pressure-transmitting medium to transmit pressure to a sample located within the cap. The medium is preferably a liquid. The medium may be a soft solid.

The cap may be composed of polytetrafluoroethylene.

The cell may comprise a sealing ring between the piston and the cap to provide a seal between the piston and the cap in said bore, the sealing ring being composed of a cobalt alloy or a cobalt-nickel alloy. The cobalt alloy of the sealing ring preferably has a composition of substantially 18-22% chromium, 33-37% nickel, 9-11% molybdenum, and balance cobalt, by weight. Most preferably, the cobalt alloy of the sealing ring has a composition of substantially 20% chromium, 35% nickel, 10% molybdenum, and 35% cobalt, by weight.

The cell may comprise locking means for locking the piston in position to maintain the pressure within the bore.

According to a second aspect of the present invention, there is provided a method of preparing a pressure cell having a cylinder with a bore in which a non-metallic cap is located, the cap being arranged to cover in use a sample located within the bore and being pressurised in use to increase the pressure on a sample located within the bore, the method comprising the steps of honing or lapping or polishing the bore prior to inserting the cap, thereby to reduce friction between the cap and the cylinder during pressurisation of the cell.

Under normal conditions, friction between two hard metal surfaces is not influenced to any appreciable extent by surface roughness. However, in the case of plastics or elastomers which are forced to slide over rough surfaces, deformation energy ("hysteresis"and/or"grooving") losses caused by small-scale inelastic deformation of the material can lead to large levels of friction. So, for these materials, roughness has a very important effect on the friction level. In general, with sliding motion involving plastics or elastomers, friction due to adhesion of the two surfaces is relatively small compared with that due to deformation. Hence, unless the surfaces are very smooth, it is not normally possible to reduce friction significantly by using a lubricant. (It is assumed that the relative motion of the surfaces is slow enough that hydrodynamic lubrication does not come into play.) The cap may be composed of polytetrafluoroethylene.

The cylinder is preferably composed of a cobalt alloy or a cobalt-nickel alloy.

According to a third aspect of the present invention, there is provided a sealing ring for providing a seal in a piston-cylinder pressure cell having a cylinder having a bore in which a sample can be located, a piston movable in the bore to increase the pressure within the bore, and a cap located within the bore and arranged to cover in use a sample located within the bore, the piston being arranged to bear against the cap in use to increase the pressure within the cap, the sealing ring being arranged in use between a said piston and cap to provide a seal between a said piston and cap, the sealing ring being composed of a cobalt alloy or a cobalt-nickel alloy.

The cobalt alloy of the sealing ring preferably has a composition of substantially 18-22% chromium, 33-37% nickel, 9-11% molybdenum, and balance cobalt, by weight.

Most preferably, the cobalt alloy of the sealing ring has a composition of substantially 20% chromium, 35% nickel, 10% molybdenum, and 35% cobalt, by weight.

An embodiment of the present invention will now be described by way of example with reference to the accompanying drawings, in which: Figure 1 is a graph to scale showing plots of the magnetic moment of a 0.10 cm3 (0.85 g) sample of AMS 5844 material in the age hardened condition as a function of temperature from 2 K to 293 K and in various magnetic fields; Figure 2 is a cross-sectional view of a first example of a cell according to the present invention; Figure 3 is a schematic drawing of apparatus for vacuum impregnating an electrical feedthrough of the cell of Figure 2 with epoxy; Figure 4 is a detailed cross-sectional view of the electrical feedthrough; Figure 5 is a cross-sectional view of a second example of a cell according to the present invention; and, Figures 6A and 6B are respectively a plan view and a longitudinal cross-sectional view of an example of a sealing ring for a piston-cylinder pressure cell according to the present invention.

Referring to Figure 2 of the drawings, a first example of a pressure cell 1 according to the present invention has a hollow inner cylinder 2 of AMS 5844E alloy, which has the composition described above, and an outer sleeve 3 of beryllium copper in which the inner cylinder 2 is mounted.

This compound design is preferred for several reasons.

First, the boundary between the inner cylinder 2 and the outer sleeve 3 inhibits the propagation of any cracks from the core to the outer surface that might otherwise occur.

Secondly, beryllium copper has a smaller magnetic permeability than AMS 5844E alloy, and so the use of beryllium copper for the outer sleeve 3 will make it easier to use the cell 1 in a magnetic field. In addition, beryllium copper is less likely to have a large specific heat at low temperatures than AMS 5844E alloy, so that use of beryllium-copper in the bulk of the cell 1 makes it easier to cool the cell 1 to very low temperatures (below 100 mK, for example). Finally, beryllium-copper is easier to machine than AMS 5844E alloy, which is especially helpful when making the threads 4,5 which engage the beryllium-copper lock nuts 6,7 provided at each end of the cell 1. The outer surface of the beryllium copper sleeve 3 has a step 8 which permits the cell 1 to be suspended under a hydraulic ram (not shown) while pressure is applied by the ram, as will be described further below. This arrangement for holding the cell 1 (in contrast with, for example, supporting it on its base) reduces the pressure drop which may otherwise take place when the ram pressure is released. The bore of the sleeve 3 and the outer surface of the inner cylinder 2 are tapered with an angle of 1 towards the second lock nut 7.

A tungsten carbide electrical feedthrough 9 is secured in position in the cell 1 by one of the lock nuts 7 so that a leading end 10 of the feedthrough 9 projects into the hollow centre of the inner cylinder 2. A sample (not shown) of a material whose properties are to be investigated is mounted on the leading end 10 of the electrical feedthrough 9. Depending on the properties to be investigated, the sample is connected to wires 11 or for example a coil (not shown) which are connected to external instrumentation (not shown). For example, the electrical resistance, heat capacity or magnetic susceptibility of the material can be measured under high pressure conditions and at temperatures from ambient to cryogenic. Usage of cells of this general type is well known in itself and will not be described in detail. A hollow cap 12 is sealed at one end and open at the other. The open end is fitted over the leading end 10 of the feedthrough to cover the sample with the closed end of the cap 12 projecting into the hollow centre of the inner cylinder 2 with the cap being filled with a pressure transmitting liquid. A first sealing ring 13 is fitted around the closed end of the cap 12 and a second sealing ring 14 is fitted around the open and of the cap 12. The first sealing ring 13 is annular having a triangular cross-sectional shape as shown in Figure 6A to fit over the cap 12 and is composed of the AMS5844E alloy.

Thus, the first sealing ring 13 is strong in compression and non-magnetic.

A first tungsten carbide piston 15 is fitted into the opposite end of the inner cylinder 2 and abuts the closed end of the cap 12. A tungsten carbide pressure transmitting pad 16 abuts the piston 15 and is located within the other lock nut 6. A second tungsten carbide piston 17 abuts the other side of the pressure transmitting pad 16 and is operatively connected to a hydraulic ram (not shown). When the ram is operated, the force is transmitted via the second piston 17, the pad 16 and the first piston 15 to the sealing ring 13 and the cap 12. As a result, the cap 12 compresses to raise the pressure of the pressure transmitting liquid, thereby increasing the pressure on the sample. Typically, the lock nut 6 around the pad 16 is then tightened to lock the pad 16 and the first piston 15 in position. This allows the ram to be withdrawn and the second piston 17 to be removed whilst the pressure in the cap 12 is maintained, thereby allowing removal of the cell 1 to other apparatus where the operating conditions can be varied as required. For example, the cell 1 can be removed to a cryogenic apparatus where the temperature of the whole cell 1 can be controllably lowered.

Fabrication of the pressure cell 1 is straightforward.

The outside sleeve 3 of the cell 1 (without the inner taper) and the lock nuts 6, 7 are machined from a solid rod of beryllium-copper alloy 25 in the H condition (work hardened, but not aged). It is then age-hardened at 335 C for two hours, and allowed to cool in air. The inner cylinder 2 is then cut from a AMS 5844E alloy rod in the unaged condition. Machining this material is not difficult with carbide-tipped tools, and is roughly comparable to working with C-300 maraging steel. To begin, the bore of the inner cylinder 2 is drilled. The diameter of this hole is less than the final value, because the material changes dimensions slightly during the ageing process. The 1 tapers on the inner cylinder 2 and outer beryllium copper sleeve 3 are then machined. These operations are done in sequence, without altering the settings on the lathe, in order to ensure that the parts fit together precisely. The inner cylinder 2 is then aged at 593 C for 4 hours, and allowed to cool in air. The bore of the inner cylinder 2 is subsequently honed to the correct diameter. Finally, the inner cylinder 2 and the outer sleeve 3 are forced together, without lubricant, using a hydraulic ram operating at a load of several tonnes.

The feedthroughs used in piston cylinder cells 1 are typically the most vulnerable part of these devices, and a considerable amount of care must go into their construction if they are to be used reliably. The wires 11 are sealed into the bore of the feedthrough 9 using a commercial epoxy which is loaded with tiny particles of sapphire. These particles serve two functions. First, they lower the thermal expansion coefficient of the epoxy to a level which is not too far above that of structural materials, such as AMS 5844E alloy and tungsten carbide. This helps to prevent debonding of the epoxy due to differential thermal contraction as the cell 1 is cooled and warmed. Secondly, the sapphire particles greatly increase the compressive strength of the epoxy, and thereby help it to withstand the high pressure inside the cell 1.

Vacuum degassing of the epoxy, in order to remove trapped air, is very important in achieving high strength.

(The ratio of the compressive strength of epoxy which has been degassed to that of epoxy which has not been so processed may be at least 1.7). If there are any bubbles of air in the epoxy after it has been injected into the bo

The wires 11 which are used in the feedthrough 9 are made of copper magnet wire, with an insulation which is both strong and capable of adhering well to the epoxy.

Suitable insulations include polyvinyl formal (Formvar), and modified polyester. Insulations which should preferably be avoided, because of their lack of adhesion, include polytetrafluoroethylene (Teflon), polyimide (such as ML) and insulations which contain polyamide (of which nylon is an example). Formvar and modified polyester insulations are also more resistant to mechanical abrasion than Teflon and polyimide. Modified polyester is more resistant to solvents than Formvar, and is probably the best insulation for the feedthrough wires 11.

The wires 11 that are preferably used in the cell 1 are preferably twisted in order to reduce electromagnetic pickup. In order to ensure that the wires 11 are completely covered with epoxy, the twists are spread apart (transversely) in the region of a twisted pair which passes through the feedthrough 9.

Before inserting the wires 11 into the feedthrough 9 and adding the epoxy, both the wires 11 and feedthrough 9 are preferably thoroughly cleaned in order to promote good adhesion. The bore of the feedthrough 9 is first blasted with fine particles of sand or glass beads in order to remove loose films which may have been introduced during the fabrication process (e. g. during oil quenching of tool steel). Both the feedthrough 9 and the wires 11 are then ultrasonically cleaned in a sequence of solvents in order to remove organic contamination. In the case of the wires 11, such"contaminants"can take the form of light oil or fine wax which is added deliberately during their manufacture in order provide lubrication during coil winding. The following solvents, in the order in which they are used, may be used: 111-trichloroethane, acetone, and isopropanol. Since it is important not to recontaminate the wires 11 or the feedthrough 9 with oils from the hands after they have been cleaned and before the epoxy is introduced, gloves should be worn during this operation. Alternatively, the feedthrough 9 and its wires 11 can be given a final clean in isopropanol before the epoxy is applied; the assembly should be heated after this has been done in order to drive off excess isopropanol.

In order to ensure that the epoxy is injected into the feedthrough 9 without simultaneously introducing air bubbles, a vacuum impregnation technique is used.

Apparatus 20 for doing this is shown in Fig. 3. The feedthrough 9, with the wires 11 inserted, is mounted in an open (atmospheric pressure) end 21 of a glass vessel 22 of the apparatus 20. Degassed epoxy is carefully poured down the side of a funnelled opening of an aluminium vacuum impregnation fixture 23 which supports the feedthrough 9 at the atmospheric pressure end 21. When the epoxy has completely covered the hole 24 which leads down to the feedthrough 9, the vessel 22 beneath it is evacuated through an opening 25 with a vacuum pump (not shown). The vacuum is held until epoxy has been drawn through the bore of the feedthrough 9, and can be seen emerging from its lower end. In order to ensure that the wires 11 that are in contact with the bore of the feedthrough 9 are completely covered with epoxy, the vacuum is then released, and the wires 11 are gently pulled down through the epoxy and feedthrough 9 for a distance of about 5 mm. The vacuum is then reapplied until more epoxy can be seen flowing from the underside of the feedthrough 9. The application of moderate heating helps to lower the viscosity of the epoxy and assist its passage along the feedthrough 9. This can be done by heating the aluminium impregnation fixture 23 to about 50 C with a hot air gun.

When the epoxy is in place, the feedthrough 9 is detached from the vacuum impregnation fixture 23, and excess epoxy is removed from the wires 11 using mechanical means and, if necessary and with care, by squirting them with acetone. It is important that epoxy is completely removed from exposed wires 11, since these will otherwise become very brittle when the epoxy hardens. The feedthrough 9 is then placed in an oven, and the epoxy is cured at about 100 C for 3 hours.

When epoxy is drawn through the feedthrough 9 under vacuum applied to the opening 25 in the vessel 22 below the feedthrough 9, the particles of sapphire are filtered to a significant extent by the wires 11, and tend to be retained at the end at which the epoxy is introduced. This can be seen by subsequently examining the feedthrough 9 under a microscope. As a result, there is a sapphire particle density gradient along the bore of the feedthrough 9, with a high density of particles at the end at which the epoxy is introduced. Since the sapphire makes a very important contribution to the compressive strength of the epoxy, it is desirable that the sapphire density be highest at the high pressure end of the feedthrough 9.

For the pressure transmitting liquids typically used in the cap 12, such as 50: 50 n-pentane: isoamyl alcohol or 50: 50 n-pentane: isopentane, the seal which is produced by the sapphire-loaded epoxy is not always perfect. Small cracks may form between the epoxy and the wires 11, or between the epoxy and the walls of the feedthrough 9.

Since the epoxy is very rigid, these cracks may remain open when the pressure is applied, allowing the pressure transmitting fluid to escape. For this reason, as shown most clearly in Figure 4, a sealing layer 19 of some relatively soft material is applied to the high pressure end of the feedthrough 9 after the sapphire loaded epoxy 18 has cured. This material 19 flows into any cracks when the pressure is applied, and provides a reliable seal. The soft material also acts as a strain-relief for the wires 11, by preventing them from being bent too sharply. For this purpose, a semi-flexible epoxy may be used. A particularly convenient semi-flexible epoxy is one of the many quick-setting types, which hardens in 5 to 10 minutes.

In order to avoid the presence of bubbles, which can collapse upon pressurisation, such an epoxy should preferably be mixed very carefully with its hardener before applying it to the feedthrough 9. (Rapid-setting epoxies normally harden too quickly to permit vacuum degassing).

If it is not possible to use a rapid-setting epoxy in this way, consideration can be given to the use of a soft, slowsetting epoxy, which can be vacuum degassed. One such material is Stycast 2741. The hardness of this epoxy in the cured state can be varied by altering the ratio of the unmixed epoxy to catalyst.

Although the soft epoxies seem to be perfectly acceptable for use with a n-pentane: isopentane pressure transmitting medium, they tend to weaken in the presence of the n-pentane: isoamyl alcohol mixture. The use of silicone rubber is a possible solution to this problem, although it may swell in the presence of alcohols. However, it does not seem to completely degrade over time, as epoxies do, in the presence of isoamyl alcohol. It may be noted that the most common types of single component silicone rubber generate acetic acid while they cure, and these can etch copper. It is preferable to use a non-corrosive grade of single component rubber, which produces only alcohol during the curing process. An attractive alternative is a two component silicone rubber, such as RTV 655. This particular product has a low viscosity in the mixed state, and requires an elevated temperature cure. These characteristics make the vacuum degassing process fairly straightforward and, because the material is also completely transparent, help to ensure that the resulting solid is free of bubbles.

The straight tungsten carbide piston 15 shown in Figure 2 may cause problems at pressures near the compressive strength of the material. The reason for this is that under high loads, there will be very high local pressures near the edge of the upper end surface of the piston 15. This effect leads to chipping of the edge, which results in a reduction of the area of the upper surface. As a result, local pressures at the freshly exposed edge will become even greater, more chipping will take place, and the rod will proceed to disintegrate completely at a load which is below the true compressive strength of the tungsten carbide. In the case of the material (Sandvik H10F) which is often used in such cells 1, a pressure of roughly 40 kbar seems to be the maximum which can be used with such a piston 15, although there is substantial variation in the amount of pressure which must be applied before chipping takes place.

To reach higher pressures, a piston 15'having the shape illustrated in Figure 5 can be employed. The alternative piston 15'has a head 30 with rounded corners 31. The head is connected to a cylindrical body 32 of reduced diameter with a smoothly curved transition 33 between the head 30 and body 32. Regardless of whether the piston 15,15' is shaped as shown in Figure 5 or straight as in Figure 2, it will be much more durable and resistant to chipping if the corners 31 are rounded. It also helps to keep the piston 15,15'as short as possible relative to the diameter. Typically, the piston 15,15'will be approximately 30 mm in length.

Since the upper sealing ring 13 will expand to accommodate a large bore when it is pressurised, it is not necessary that this sealing ring 13 have an initial fit which is very tight; a light press fit is adequate. It is desirable that the lower (copper) sealing ring 14 have a heavy press fit onto the feedthrough 9 and into the bore of the cell 1. The lower sealing ring 14 will deform during the first pressurisation, and when it is subsequently reused, it will probably have to be forced into the bore of the cell 1 using a hydraulic ram. The cap 12 may be made undersized by about 100 ym so that it requires a force-fit onto the end of the feedthrough 9. When this is done, the outer diameter of the part of the cap 12 which overlaps the feedthrough 9 is slightly larger than the diameter of the cell bore. As the feedthrough 9 is inserted into the cell 1, this part of the cap 12 is squeezed between the cell bore and the feedthrough 9. All the above measures help to ensure that the initial seals are good enough that no pressure transmitting fluid or cap material escapes when the pressure is applied.

When samples, or other items, are being connected to the feedthrough wires 11, it is worth bearing in mind that the cap 12 may tear open when the pressure is applied.

Although it usually reseals itself as the ram is extended, the resulting volume inside the cap 12 may be considerably less than might otherwise be supposed based on the original length of the cap 12. It has been found that the likelihood of tearing is reduced if the surfaces of the cap 12 are made as smooth as possible. If a 50: 50 n-pentane: isoamyl alcohol mixture is used for the pressure transmitting fluid, and if no tearing takes place, the length of a cap 12 with a 38 mm inside length is reduced by about 15 mm at a pressure of 35 kbar. Since the walls of the cap 12 will also thicken as the pressure increases, it is also preferable to leave a generous amount of room between the inside wall of the cap 12 and delicate items such as samples. A space of about 350 Um between the sample and the walls of the cap 12 should be sufficient for most purposes.

One of the advantages of beryllium copper over other structural metals is that it is highly resistant to cold welding (galling or seizure) when sliding against itself.

Nevertheless, it is sometimes found that the lock nuts 6,7 will become stuck inside the cell 1 if they are not provided with a suitable lubricant and, on occasion, cannot be removed without being damaged in the process. To prevent this, the threads of the lock nuts 6,7 can be covered with a high pressure lubricant, such as one of the many anti-seizure formulations based on molybdenum disulphide.

It is beneficial to listen to a pressure cell 1 while is being pressurised. Normal pressurisation takes place very quietly, and the only sound which may be heard is that due to the pump which operates the hydraulic ram. If any other sounds, such as clicks, are detected, the pressurisation process should be stopped. These often indicate that the piston 15 has chipped or cracked.

Groaning or creaking sounds frequently signal the incipient catastrophic failure of a piston 15. In such a case, and if there is enough time, the ram pressure should be reduced immediately. Loud bangs normally indicate that either the piston 15 has disintegrated or the feedthrough 9 has blown out.

It is preferred to increase the applied pressure in 8 kbar increments, and wait for a few minutes before proceeding to the next level of pressurisation. At each interval, the upper lock nut 6 is tightened. It is very useful to fill a chart during the pressurisation process with information about the fluid pressure inside the ram, the displacement of the piston 15 (to ensure, for example, that the cap 12 is not collapsing onto the sample), the pressure which is applied to the piston 15 by the ram, and the actual pressure inside the cell 1.

After pressurisation to levels over about 20 kbar, the bore of the cell 1 may permanently deform. This will not necessarily lead to problems, as the sealing ring 13 between the piston 15 and the cap 12 will expand to a certain extent to accommodate distortions. What can cause problems is the roughening of the wall of the cylinder 2 which occurs when the cell 1 is pressurised and then taken apart. This results in friction between the cap 12 and the wall of the cylinder 2 during subsequent pressurisations, which can lead to intolerable pressure losses at the highest pressures (e. g. 40 kbar of applied pressure in order to achieve only 25 kbar inside the cell 1).

Under normal conditions, friction between two hard metal surfaces is not influenced to any appreciable extent by surface roughness. However, in the case of plastics or elastomers which are forced to slide over rough surfaces, deformation energy ("hysteresis"and/or"grooving") losses caused by small-scale inelastic deformation of the material can lead to large levels of friction. So, for these materials, roughness has a very important effect on the friction level. In general, with sliding motion involving plastics or elastomers, friction due to adhesion of the two surfaces is relatively small compared with that due to deformation. Hence, unless the surfaces are very smooth, it is not normally possible to reduce friction significantly by using a lubricant. (It is assumed that the relative motion of the surfaces is slow enough that hydrodynamic lubrication does not come into play.) Although the bore of the pressure cell 1 can be rehoned on a regular basis in order to reduce its surface roughness, the fact that it is normally slightly barrelled after being taken to very high pressures means that significant amounts of material must be removed before the surface is acceptably smooth. In order to avoid this, the inside surface may be polished with a felt polishing wheel.

Such a wheel is mounted in a high-speed hand-held drill (e. g. a"Dremel"machine) and coated with diamond paste.

The bore of the pressure cell 1 is covered with a suitable polishing lubricant, and the wheel is moved back and forth inside it while spinning at roughly 20,000 rpm. A sequence of three grades of diamond paste, such as 14 Um, 3 pm and 1 Hm can be used, with a separate wheel for each grade.

The process takes about an hour or so, and may have to be carried out every time the cell 1 is taken apart, if it has been used, and will again be used, at the highest pressures. The increase in bore diameter after polishing may amount to only 25 um (compared with perhaps 125 Hm after honing, depending on how badly the bore has barrelled). If the bore has been polished often enough, it will become very uneven, and honing will become necessary to restore an acceptably cylindrical shape.

After polishing, the pressure which must be applied by the piston in order to reach an actual internal pressure of 35 kbar can be for example 41.6 kbar. The pressure loss (6.6 kbar) could probably be reduced even more if the cylinder 2 were rehoned as well as polished, because it is difficult to remove the largest surface features by polishing alone. Pressure losses due to friction apparently decrease as the bore of the cell 1 increases in size. It has been found that pressure losses due to friction seem to be comparatively small in, for example, 6 mm diameter pressure cells, even without polishing.

The most unreliable elements in a pressure cell 1 are the electrical feedthrough 9 and the piston 15. Common feedthrough faults include short and/or open circuits in the wiring, and violent ejection of the wires 11 during pressurisation (often caused by inadequate outgassing of the epoxy 18 before placing it in the feedthrough 9).

If a piston 15 disintegrates inside the bore of a cell 1, it is likely that small shards of tungsten carbide may be imbedded in the wall. This can happen not only during the process of disintegration, but also while the remains of the piston 15 are removed from the bore. Such particles can be very difficult to detect and remove using normal cleaning procedures. When a new piston 15 is subsequently inserted into the bore, it can be damaged by these particles and fracture at a very low pressure. To prevent such accidents, the pressure cell 1 can be rehoned after a piston fracture in order to remove stray pieces of tungsten carbide.

As mentioned, the piston 15 may be made from tungsten carbide. It is important to note that not all types of tungsten carbide are the same. Tungsten carbide for which there is no data on compressive strength should be avoided.

Knowing the hardness is not enough, since defects in the material can lead to a reduction of strength, but not necessarily in the hardness, since hardness is a local property which may depend on the position of the indenter used to measure the hardness. Manufacturers are also more likely to pay attention to the quality of the tungsten carbide if their product is being sold on the basis of global parameters such as compressive strength and transverse rupture strength, which are sensitive to the presence of pores and other defects. Furthermore, although values of 50 to 60 kbar are often given as"the compressive strength of tungsten carbide"by standard references, the actual compressive strengths can vary from between, for example, 26 kbar (in the case of Sandvik's H27N grade) to 115 kbar (in the case of their 3UF grade), depending on the cobalt binder content and the size of the tungsten carbide grains. Normally, there is a trade off between compressive strength for fracture toughness and transverse rupture strength. The present applicant has not been able to find any information about the strength or toughness of tungsten carbide at low temperatures. However, because tungsten carbide parts are used in compression, problems of embrittlement will generally going be less of a problem than they would be in those parts of the cell 1 which are placed under tension or shear. Sandvik's H10F grade of material is suitable for most applications for the cell 1, because it is readily available, has a good combination of compressive strength (63 kbar) and fracture toughness (13 MN m~l5), and is very reliable. This material has a magnetic permeability of about 3 at room temperature. It is possible to reduce the permeability of the tungsten carbide by using grades with a smaller cobalt content. For example, Sandvik has a 3% Co material which is designated "H3F". This composition has a compressive strength of 78 kbar, a fracture toughness of 7 MN m~l5, and a permeability of about 1.5 at room temperature. However, it is difficult to obtain. Another type of tungsten carbide is available in which the grains are bound together with nickel, or an alloy in which nickel is the main constituent. Such material has a much lower permeability than the cobalt-cemented variety, although its mechanical properties are generally inferior. One grade is cemented with pure nickel and has a measured compressive strength of 47 kbar. Sandvik offer a type of non-magnetic material (C9M) which contains 91% tungsten carbide, 8.03% Ni, 0.69% Cr and 0.28%"other materials". It has an estimated compressive strength which is somewhat higher than the pure nickel-cemented variety (55 kbar), and a measured fracture toughness of 10.7 MN m-1-5 However, is difficult to obtain in the required shapes and sizes.

A material which has been used for the feedthrough 9 in the cell 1 during testing is high speed steel, which is strongly ferromagnetic (the reason for doing this is given later). The favoured material is a highly alloyed grade which is manufactured by Erasteel and designated"ASP 2060". This material can be machined in a soft condition, and subsequently heat treated to a hardness level of Rc 69, which corresponds to a compressive strength of about 39 kbar. Despite its hardness, the material is relatively tough. It is probably the strongest material (in compression) which can be shaped using ordinary machine tools.

In order to reduce the magnetic moment of the cell 1 to the lowest level, a tungsten carbide, such as one of the grades discussed above, is preferably used in the feedthrough 9. In general, this material must be either ground and/or spark eroded to the required shape.

The cell 1 of the invention has been taken to 35 kbar at room temperature. The upper limit was imposed not by bursting of the AMS 5844E alloy or Be-Cu parts of the cell 1, but by failure of the straight tungsten carbide piston 15 illustrated in Figure 2 at higher pressures. This could probably be avoided by using a profiled piston 15'such as the one illustrated in Figure 5.

The pressure was determined by measuring the change in the resistance of a strip of 99.9999% pure tin, and using the standard calibration data. After the lock nuts 6,7 are tightened and the ram pressure is removed, the pressure drops to 34 kbar. When the cell 1 is cooled to 2.5 K, the pressure is measured again (this time by determining the temperature of the superconducting transition in the tin), and it is found that it drops to 29 kbar. This loss of pressure upon cooling is a fairly frequent occurrence in clamp cells. It may be possible to prevent it by using springs to transmit the force, in order to take up some of the slack caused by the large thermal contraction of the pressure transmitting medium. This approach, using Belleville washers or"split springs", is often used with diamond anvil cells. The cell 1 described here seems to withstand thermal cycling without difficulty.

Three of the parts which have been used in testing the pressure cell 1 of the invention, namely the feedthrough, the sealing ring, and the pressure transmitting pad, have been made of steel and are therefore strongly ferromagnetic. These have been used because most of the work up until now has involved testing the capability of the MP35N alloy, and also since for some applications the presence of small magnetic parts of no importance.

However, replacing these parts with less strongly magnetic ones, such as tungsten carbide, presents no difficulty.

An embodiment of the present invention has been described with particular reference to the examples illustrated. However, it will be appreciated that variations and modifications may be made to the examples described within the scope of the present invention.

Claims

CLAIMS 1. A piston-cylinder pressure cell for pressurising a sample, the cell comprising a cylinder having a bore in which a sample can be located, and a piston movable in the bore to increase the pressure within the bore, the cylinder being composed of a cobalt alloy or a cobalt-nickel alloy.
2. A pressure cell according to claim 1, wherein the cobalt alloy of the cylinder has a composition of substantially 18-22% chromium, 33-37% nickel, 9-11% molybdenum, and balance cobalt, by weight.
3. A pressure cell according to claim 2, wherein the cobalt alloy of the cylinder has a composition of substantially 20% chromium, 35% nickel, 10% molybdenum, and 35% cobalt, by weight.
4. A pressure cell according to any of claims 1 to 3, wherein the material of the cylinder is prepared by work hardening prior to age hardening.
5. A pressure cell according to any of claims 1 to 4, wherein the cylinder is located within a sleeve.
6. A pressure cell according to claim 5, wherein the sleeve is composed of beryllium-copper.
7. A pressure cell according to any of claims 1 to 6, comprising a cap located within the bore and arranged to cover in use a sample located within the bore, the piston being arranged to bear against the cap in use to increase the pressure within the cap.
8. A pressure cell according to claim 7, wherein the cap contains a pressure-transmitting medium to transmit pressure to a sample located within the cap.
9. A pressure cell according to claim 7 or claim 8, wherein the cap is composed of polytetrafluoroethylene.
10. A pressure cell according to any of claims 7 to 9, comprising a sealing ring between the piston and the cap to provide a seal between the piston and the cap in said bore, the sealing ring being composed of a cobalt alloy or a cobalt-nickel alloy.
11. A pressure cell according to claim 10, wherein the cobalt alloy of the sealing ring has a composition of substantially 18-22% chromium, 33-37% nickel, 9-11% molybdenum, and balance cobalt, by weight.
12. A pressure cell according to claim 11, wherein the cobalt alloy of the sealing ring has a composition of substantially 20% chromium, 35% nickel, 10% molybdenum, and 35% cobalt, by weight.
13. A pressure cell according to any of claims 1 to 12, comprising locking means for locking the piston in position to maintain the pressure within the bore.
14. A method of preparing a pressure cell having a cylinder with a bore in which a non-metallic cap is located, the cap being arranged to cover in use a sample located within the bore and being pressurised in use to increase the pressure on a sample located within the bore, the method comprising the steps of honing or lapping or polishing the bore prior to inserting the cap, thereby to reduce friction between the cap and the cylinder during pressurisation of the cell.
15. A method according to claim 14, wherein the cap is composed of polytetrafluoroethylene.
16. A method according to claim 14 or claim 15, wherein the cylinder is composed of a cobalt alloy or a cobaltnickel alloy.
17. A method according to claim 16, wherein the cobalt alloy of the cylinder has a composition of substantially 18-22% chromium, 33-37% nickel, 9-11% molybdenum, and balance cobalt, by weight.
18. A method according to claim 17, wherein the cobalt alloy of the cylinder has a composition of substantially 20% chromium, 35% nickel, 10% molybdenum, and 35% cobalt, by weight.
19. A sealing ring for providing a seal in a pistoncylinder pressure cell having a cylinder having a bore in which a sample can be located, a piston movable in the bore to increase the pressure within the bore, and a cap located within the bore and arranged to cover in use a sample located within the bore, the piston being arranged to bear against the cap in use to increase the pressure within the cap, the sealing ring being arranged in use between a said piston and cap to provide a seal between a said piston and cap, the sealing ring being composed of a cobalt alloy or a cobalt-nickel alloy.
20. A sealing ring according to claim 19, wherein the cobalt alloy of the sealing ring has a composition of substantially 18-22% chromium, 33-37% nickel, 9-11% molybdenum, and balance cobalt, by weight.
21. A sealing ring according to claim 20, wherein the cobalt alloy of the sealing ring has a composition of substantially 20% chromium, 35% nickel, 10% molybdenum, and 35% cobalt, by weight.
22. A piston-cylinder pressure cell, substantially in accordance with any of the examples as hereinbefore described with reference to and as illustrated by the accompanying drawings.
23. A method of preparing a pressure cell, substantially in accordance with any of the examples as hereinbefore described with reference to and as illustrated by the accompanying drawings.
24. A sealing ring for a piston-cylinder pressure cell according to claim 19, substantially in accordance with the example as hereinbefore described with reference to and as illustrated by the accompanying drawings.