RUDZIAK MARK K
WONG TERENCE
| WO1992020076A1 | 1992-11-12 | |||
| WO1993002222A1 | 1993-02-04 |
| US5160794A | 1992-11-03 | |||
| US5158620A | 1992-10-27 |
In the manufacture of superconducting wire for very high field magnets, such as those used in the proposed Superconducting Super Collider (SSC), or the proposed Large Hadron Collider (LHC), the superconductor must remain superconducting in very high magnetic fields in the range of 5 to 9 tesla. It is also recognized in the art that such material must be produced from large billets, up to 300mm in diameter, in order to achieve cost effective economies of scale. Further, as has been described in our earlier patents the final reacted Nb + Ti layer thickness must be in the range of 100 nm in order to achieve optimal conductor J c 's. Thus for a given final wire design, the ratio of the final wire diameter to the optimal layer thickness (100 nm) must be the same for the ratio of the starting multifilamentary restack billet diameter (300mm) to the initial layer thickness because the reduction in diameter is the same as the reduction in layer thickness. For example, the conductor design for the SSC called for a final wire size of 0.648mm in diameter. The reduction in diameter from 300mm multifilamentary billet to 0.648mm wire is a factor of 463 times. Thus, the initial Nb + Ti layer thickness in the multifilament billet must be 46.3 microns because the final layer thickness must be around 100 nm for optimal J c ' s (463 x
100 nm) . Such a large initial layer thickness poses special problems for fabricability and current density. These problems can be overcome through modifications described in the present invention.
When the layers within a Nb/Ti composite filament are very thick, it is neither practical nor effective to generate much of the requisite degree of diffusion within the filament prior to extrusion in the manner of a preferred embodiment of the invention described in our earlier patents. Under the laws of solid state diffusion, the distance of diffusion at a given temperature increases with the square root of time. To diffuse twice the distance within a composite at a given temperature, one would require four times the length of time. So when the niobium and titanium layers of a filament are very thick, it can require prohibitively long periods of time to achieve the required amount of diffusion. Of course, one can increase the temperature to speed diffusion. (By the solid state diffusion laws, increasing the temperature increases the rate of diffusion in an exponential manner.) While in principle this is an effective approach to thick layers, in practice the temperature must be so high that grain growth and grain boundary diffusion occurs, especially within the titanium layers. This reduces the homogeneity and the degree of order within the structure. The consequence is a degradation of J c performance for the final conductor due to less efficient flux pinning.
As it is not desirable to optimally diffuse thick niobium and titanium layers prior to extrusion, one is left with a processing route wherein much of the diffusion heat treatment is performed after extrusion. Such a scheme entails extrusion of filaments containing essentially undiffused niobium and titanium layers (Figure 2A) . This presents problems because the Nb/Ti structure does not flow during extrusion in the same way as an optimally diffused structure. In fact, it has been found
that filaments containing essentially undiffused niobium and titanium are unusually susceptible to shear fracture during extrusion. If not actually fractured, the filaments often display highly distorted layers. In either case, J c performance is greatly depressed.
For those familiar with the art, the design of the proposed SSC conductor poses significant difficulties in producing the fine filaments (6 microns) due to problems of sausaging and intermetallic formation of the filaments in the copper matrix. The same concerns exist within the present invention, thus the same care must be taken in avoiding these problems.
The niobium and titanium system shall be used as the exemplar throughout this discussion. It will be understood, however, that the principles discussed could be applied to many alternate systems as discussed in our earlier patents.
When the present invention is applied to the NbTi superconductor system, it involves the partial diffusion of Nb and Ti layers by means of a diffusion heat treatment. This diffusion of Nb into Ti lowers the alpha to beta transformation temperature of the Ti rich layer and allows extrusion in the beta phase at a lower extrusion temperature. At some point in the reduction, when the layers are between 2.5 - 15 microns, a further diffusion heat treatment is given to form optimal superconducting NbTi for the best J c at the magnetic field of interest.
One method by which to avoid the mechanical extrusion problems of the pure niobium and pure titanium composite filaments is to extrude under conditions where all beta- phase NbTi exists. Beta-phase NbTi is a body centered cubic (bcc) phase that has greater ductility than the hexagonal close packed (hep) alpha-phase of room temperature unalloyed titanium. The Ti-Nb phase diagram of figure 3 shows that increasing Nb concentration (via diffusion) lowers the alpha to beta transformation
temperature, thus decreasing the required extrusion temperature. As described above, higher temperatures lead to deleterious grain growth and grain boundary diffusion. Partial diffusion within the filaments prior to extrusion reduces the required extrusion temperature to transform the thick Ti layers to the beta-phase, mechanically more like a conventional NbTi alloy. Additional heat treatment after extrusion is necessary to reach optimal superconducting alloy composition and J c values.
The diffusion heat treatment step prior to extrusion may consist of heating the monofilament prior to restack, a separate billet heating step after restack (e.g., hot isostatic pressing), billet preheating prior to extrusion, or any combination of these operations.
The time at temperature of the heat treatment prior to extrusion will depend upon the layer thickness within the filaments, being shorter for smaller layer thicknesses. It must be emphasized that the diffusion caused by this heat treatment is intended to be partial; the diffusion heat treatment applied after extrusion is where sufficient superconducting NbTi is formed to produce optimal conductor J c 's. While leaving sufficient non superconducting material to serve as the APC layer.
Fig. 1 is a schematic diagrammatic sectional view of one preferred type of monofilament billet.
Figs. 2a, 2b, and 2c are composition scans made with a scanning electron microscope showing atomic percent of Nb and Ti as a function of scanning distance (in microns) across a section of a filament layer during various stages in the process.
Figs. 3a and 3b are NbTi phase diagrams.
Figs. 4a and 4b are schematic views of additional arrangements of Nb and Ti layers, and comparing a conventional monofilament (Fig. 4a) with a monofilament made in accordance with Example IV (Fig. 4b).
The major drawback to partial diffusion heat treatment prior to extrusion is that some grain growth and grain boundary diffusion inevitably occurs. This tends to increase grain size and decrease the homogeneity within the layers and ultimately reduce current density. Depending upon the application, the reduction in J c might be an acceptable trade-off for the improved ductility. Experience has shown, however, that it may not be necessary to heat treat the filaments or the secondary billet for substantial lengths of time, so long as the extrusion temperature is high (816°C-871°C) . At high temperatures, a gross mechanical failure of the filaments is not generally observed and layer uniformity can be excellent. It is believed that this processing route is made possible by the fact that the high temperature alone can promote sufficient Beta-phase formation to allow extrusion of uniform layers. Overall ductility is further enhanced by the high temperature serving to anneal and soften the niobium and titanium.
Whichever pre-extrusion diffusion processing is selected, the post-extrusion processing remains essentially the same. A layer thickness (niobium or titanium) of 2.5 - 15 microns is achieved within the composite and a further heat treatment is applied. The heat treatment will typically be at a temperature of 649°C-760°C for 10-20 hours, but will vary depending upon the demands of the application. The heat treated composite is then drawn to a final diameter where the superconducting and non-superconducting layers are around 100 nm in thickness. The non-limiting examples below describe various processing routes that fall within the scope of the invention.
EXAMPLE I Monofilament
Starting with a 153.7mm diameter copper can 10 having a 123.9mm diameter I.D., a monofilament billet is assembled using 0.381mm niobium and 0.508mm titanium
sheets 12 over a 22.2mm APC core 14. The sheets 12 are arranged in alternating sets of three so that the effective niobium thickness is 1.143mm and the effective titanium thickness is 1.524mm (see Figure 1). These thicknesses are required to achieve a 100 nm Nb + Ti thickness at final wire size (0.648mm). This requires a reduction in thickness of each layer by a factor of about 26,670. A niobium barrier 16, typically comprising six turns of 0.381mm niobium sheets, totalling about 4% of the non-copper volume, is interposed between the outermost niobium layer and the copper can. The barrier prevents copper contamination within the filaments during subsequent processing. A core of NbTi, Nb or stacked Nb and Ti sheets are inserted into the center of the assembly. When the sheet assembly is complete, a copper nose and tail are tungsten inert gas (TIG) welded into place, the billet is evacuated at 482°C to a pressure below 10- torr, and is then sealed. The billet is preheated for 5 hours at 871°C prior to extrusion to a diameter of 21.8mm, a thickness reduction of 7 times. The extrusion conditions for the monofilament billet does not create significant diffusion between Nb and Ti layers due to the large relative layer thicknesses and because the layers are not in intimate contact due to the lack of a hot isostatic pressed (HIP) or cold isostatic pressed (CIP) operation.
The extruded rod is rough cropped and then drawn to 9.27mm diameter, a thickness reduction of 2.4. At this point, it is pickled in nitric acid to a diameter appropriate for a filament spacing-to-diameter ratio (s/d) of about 0.17. After two further drawing passes aimed at smoothing the surface, the wire is given a final cropping to ensure uniform material properties. At this stage, there is limited diffusion and the thickness of each layer has been reduced by a factor of 16.5.
Multifilamentary Restack
A portion of the cropped material is drawn to 1.448mm hex diameter wire, (a further reduction of 6.4 times) where the Nb + Ti layer thickness is about 25 microns (Nb + Ti thickness x 1.448mm/153.7) . The hex wire is then straightened and cut to 127mm lengths. At this point, the compositional scan (Fig. 2A) shows little diffusion of Nb into the Ti layers. 811 of these filaments are placed into a furnace for heat treatment in an argon atmosphere. The heat treatment temperature is 871°C and the time is 4 hours. Diffusion occurs as a result of this heat treatment. As determined by EDS compositional scan, the compositional profile changes from that shown in Figure 2A to that shown in 2B. (The material forming the subject of these scans was processed in accordance with the present example.) This treatment is meant to anneal the monofilament and create partial diffusion so that the pure Ti is diluted by Nb. The more ductile Beta-phase NbTi will form at a lower temperature with increasing Nb content (see Figures 3a and 3b). This heat treatment is not meant to form the optimal superconducting alloy composition.
After heat treatment, the filaments are cleaned in nitric acid and 811 filaments are then stacked into a copper can having a 44.45mm internal diameter and 50.8mm external diameter. Copper filler rods are inserted around the stack to help minimize void space. A copper nose and tail are TIG welded into place and the billet is then evacuated and sealed. After a 2-2 1/2 hour preheat at 649°C, the billet is then extruded to 12.7mm diameter (a thickness reduction of 4 times). A post extrusion diffusion heat treatment of 649°C for 4 hours is then applied to the extruded rod in order to form sufficient superconducting NbTi for optimal conductor J c 's. At this point the compositional profile approximates that shown in Figure 2C. After heat treatment, the rod is drawn to a final size where the superconductor and non-superconductor
thickness is below 100 nm. At a final size of 0.208mm diameter, it displayed a non-copper current density (at 4.2K) of 2859 A/mm 2 at 5T and 1251 A/mm 2 at 7T. The initial Nb and Ti layer thickness had each been reduced by a factor of 25,888 at this final stage.
While preferred examples of the invention have been described above, they are only illustrative and are not to be interpreted as limiting the scope of the invention.
EXAMPLE II
A monofilament billet is assembled and processed as described in Example I, above. However, the 1.448mm hex diameter filaments are not given a heat treatment, but are directly stacked into the 50.8mm diameter secondary billet. After evacuation and sealing, the secondary billet is heated for 2-2 1/2 hours at 816°C and is then extruded to 12.7mm diameter. The pre-extrusion heating is meant to raise the temperature of the billet to a uniform level, anneal the monofilaments, and partially diffuse the Nb and Ti. A post extrusion heat treatment is then applied in order to form sufficient superconducting NbTi for optimal conductor J c . A composite processed in this manner had a post-extrusion heat treatment of 649°C for 8 hours. At a final size of 0.208mm diameter, it displayed a non-copper current density at (at 4.2K) of 3061 A/mm 2 at 5T and 1564 A/mm 2 at 7T.
It is not necessary that the pre-extrusion heat treatment be combined with the billet heating time. For example, the billet may be heated in argon atmosphere for 2 hours at 816°C and then separately heated for 2-2 1/2 hours at 649°C prior to extrusion. A diffusion heat treatment after extrusion is followed by drawing as described in the above Example.
EXAMPLE III
It is not always necessary to apply the diffusion heat treatment immediately after extrusion. It can be applied at an intermediate stage of the drawing process or even at several stages, provided that the layer thicknesses within
the composite are large enough (2.5-15 microns) to allow controlled diffusion. The following example describes the case where an APC composite is heat treated at an intermediate wire diameter:
A monofilament billet is assembled and processed as described in Example I, except that it is hexed to a diameter of 3.302mm. At this size, the Nb + Ti layer thickness is about 60 microns, having been reduced in thickness by a factor of 46.5. This wire is then straightened and cut to 127mm lengths optimal conductor J c 's. The filaments are etched clean in nitric acid and 127 of them are stacked in a hexagonal array inside a copper can having a 44.45mm internal diameter and a 50.8mm external diameter. Copper filler rods are inserted around the stack to help minimize void space. A copper nose and tail are welded into place and the billet is then evacuated and sealed. After a 2-2 1/2 hour preheat at 816°C, the billet is extruded to 12.7mm diameter, thickness reduction of 4 times. The extruded rod is cold drawn to 7.67mm diameter, where the Nb+Ti layer thickness is about 9 microns. At this point, the material is diffusion heat treated and drawn to final size as described previously. A composite processed in this manner had a diffusion heat treatment of 649°F for 12 hours at the 7.67mm diameter. At a final size of 0.095mm diameter, it displayed a non-copper current density (at 4.2K) of 2917 A/mm 2 at 5T and 1156 A/mm 2 at 7T. The initial monofilament Nb and Ti layers had each been reduced by a factor of 28,891 in this final state.
EXAMPLE IV In yet another embodiment of the invention, it is possible to overcome the problem of processing thick Nb/Ti layers by increasing the ease with which the titanium layers are diluted by Nb in order to lower the alpha to beta transformation temperature. This can be accomplished by introducing niobium into the titanium layer within the
Nb/Ti monofilament - i.e. by subdividing the titanium in such a way as to create a fine scale Nb/Ti structure within the larger structure such that the Nb has a shorter distance to diffuse. (Compare Figures 4a and 4b).
During thermal processing, the fine scale niobium quickly diffuses into the titanium, improving the overall ductility of the filament and thus the viability of extrusion. The thinner layer structure of this nature does not face the same extrusion difficulties as the thicker layers do as previously discussed. Thus, enough diffusion can occur in a pre-extrusion heat treatment in order to improve ductility and the viability of extrusion without deleterious grain growth or grain boundary diffusion. A post-extrusion heat treatment is provided to further create sufficient superconducting NbTi for optimal conductor J c . Another alternative is to provide all the diffusion prior to extrusion, removing the need for the additional post extrusion heat treatment step. The overall heat treatment is significantly less then that required for thicker layer structures due to the shorter diffusion paths. For this reason, it is possible to maintain niobium layers essentially undiffused, ultimately serving to enhance flux pinning and, consequently, improve current density.
A major advantage to this fine-scale Nb/Ti approach is that high temperatures need not be employed in the processing, since the diffusion of the niobium within the titanium will be relatively rapid even at moderate temperatures. Grain growth and grain boundary diffusion is thus eliminated, ensuring layer uniformity. Another advantage is that the normal/superconductor spacing within such a material can be adjusted by adjusting the relative width of the fine-scale Nb/Ti zone. This allows improved matching to the flux line lattice spacing at a given field and, hence, improved J c at that field. A composite processed in accordance with the fine-scale Nb/Ti approach is described in the following non-limiting example.
A monofilament billet is assembled using Nb and Ti sheets. The sheets are cut and stacked to produce an approximately cylindrical assembly. The sheets are ordered so as to form a repetitive structure with the following sheet sequence: 0.301mm thick Nb + 0.203mm Ti + 0.025mm Nb + 0.203mm Ti + 0.025mm Nb + 0.203mm Ti. Approximately 34 of these compound layers comprise the stack. The stack of sheets is surrounded by a 0.762mm thick Nb barrier and this assembly is then inserted into a copper can having a 44.45mm internal diameter and a 50.8mm external diameter. Strips of copper sheet are placed around the assembly to help minimize void space. A copper nose and tail are TIG welded into place, the billet is evacuated, and then sealed. It is then heated for 2-2 1/2 hours at 871°C, followed by extrusion to 12.7mm diameter. No significant diffusion occurs during the extrusion preheat due to the large layer thickness and the fact the layers are not in intimate contact due to the lack of a HIP or CIP operation.
The extruded rod is cropped and then cold drawn to 1.143mm hex diameter, a factor of 44.4 thickness reduction, where the titanium layers are about 5 microns thick. This wire is straightened and cut to 127mm lengths. 1,165 of these filaments are etched clean in nitric acid and then stacked into a copper can having a 44.45mm internal diameter and a 50.8mm external diameter. Copper filler wire is used to help fill the void space around the stack. A copper nose and tail are TIG welded into place and the billet is evacuated and sealed. The billet is preheated for 2-2 1/2 hours at 760 C C prior to extrusion to a diameter of 12.7mm. The billet preheat and extrusion are sufficient to drive the thin niobium layers into solution with the titanium layers without overly involving the thicker niobium layers. The greater ductility developed in the titanium region due to the formation of a ductile NbTi alloy helps ensure adequate filament quality and layer uniformity.
The extruded rod is given a further heat treatment of 10 hours at 649°C to form sufficient superconducting NbTi for optimal conductor J c . The rod is then drawn to a final size of 0.232mm, where optimum flux pinning and current density is observed. The Nb + Ti layer thickness had each been reduced by a factor of 9732 in this final stage.
Although the above examples describe a NbTi system the same concepts may be applied to alternative superconducting systems. For example, the ideas embodied in examples I, II and III may be applied to the NbTiTa ternary system by replacing the Nb within the monofilament with a NbTa alloy with the rest of the processing remaining essentially unchanged. Alternatively, the NbTa alloy could be replaced in the monofilament with thin layers of Ta adjacent to the Nb layer (as described in our earlier U.S. Patent 5,160,550) such that the diffusion time and temperature completely diffuse the Ta into the Nb and Ti regions. The concepts of example 4 could be utilized in such a manner by placing a 0.025mm Ta sheet adjacent to the 0.025mm Nb sheets within the monofilament.
While preferred examples of the invention have been described above they are only illustrative and are not to be interpreted as limiting the scope of the invention.