Rapidly Quenched Metals

LXVII

INTRODUCTION

INTRODUCTORY REMARKS

Hans WARLIMONT,     Vacuumschmelze GmbH, 6450 Hanau, and University of Stuttgart, 7000 Stuttgart, Federal Republic of Germany

This conference series on rapidly quenched metals has been attended by an increasing number of participants and has attracted a greater number of contributions every time it was held. It is notable that this increase has continued here, even though less drastically than last time in Sendai, if it is realised that the subject of rapidly quenched metals with all their structural and property aspects such as the structure of amorphous metals, rapid solidification processing, amorphous magnetism, micro-crystalline materials etc have become topics of regular meetings of metallurgical societies, physical societies, magnetism conferences and special national and local conferences of all kinds. This shows the liveliness of the subject as a topic of scientific study, as a topic which still provides many open and controversial aspects and as a topic of great technological potential and interest.

What is the present state of the field? Based on taking a broad view and considering the research and progress presented at RQ 5, it may be appropriate to focus on some points of particular relevance for present discussions and future work.

STRUCTURE OF AMORPHOUS METALS

The model of dense random packing of hard spheres has been recognised to fail as a general basis to describe the structure of amorphous metals. It is rather a special case among several structural concepts and experimentally determined structural features. Structure determination by conventional and even by more advanced scattering methods proves to be extremely difficult and limited to simple, mostly binary systems. The present state of structure description is utterly dissatisfactory. Short-range order parameters which are defined in analogy to those adopted for scattering analyses on crystalline and liquid metals are insufficient for anything close to a complete structure description, and other forms of structure description have not yet been found. The question is: will amorphous structures ever be described by means similar to those applied to crystalline states in terms of well-defined ideal structures and well-defined defects?

The strong potential for more comprehensive and succinct structure analysis appears to lie in a combination of scattering experiments and nuclear measurements by techniques such as NMR, Mössbauer spectroscopy etc. They permit the recognition of the elements of the short-range ordered structure and a determination these structure elements supplemented by a description of their interconnection, perhaps by topological parameters, may be an answer.

PROPERTIES AND APPLICATIONS OF AMORPHOUS AND MICRO-CRYSTALLINE MATERIALS

Many properties of rapidly quenched amorphous and micro-crystalline materials have been measured in attempts to understand the physics as well as to just find the difference from properties of conventionally processed crystalline materials. Much of the work which is the major part of the studies has been interpreted qualitatively only. What is the reason?

The classical synthesis of results in solid state physics and in physical metallurgy, namely to relate structure to properties, is not yet possible because structures and microstructures are not sufficiently well-known in qualitative and quantative detail and thus the relation cannot be established.

The application of amorphous and of other rapidly quenched materials has begun and it is like at the beginning of all other metallic materials: they are being produced, processed and used successfully before the structure-property relations as such have been fully understood; but in the present days of high technology we need to understand and to master all materials as well as we can to manufacture products which satisfy high specifications and are fully reliable. Therefore, the complete understanding of structure-propertiy relations is indespensible.

CONCLUSIONS

It can be concluded that for both scientific and technical reasons much remains to be studied and understood in the field of amorphous and crystalline rapidly quenched metals. RQ 5 has shown that considerable steps have been achieved but that at the same time, new questions are opening up.

The time of easy experiments in rapid quenching is over, that of serious comprehensive studies has only just begun.

HISTORICAL REMARKS AND HONOURS

U. Gonser

This is a special session for reminiscence about highlights of the past, and in particular we want to talk about amorphous metals. In this connection it seems necessary to make you acquainted with the German word Stammbaum. In the dictionary you will find the translation family tree which applies to humans, dogs and other animals. However the word Stamm has a wider meaning than family and in general, a Stammbaum represents the multiplicity of a development with all its branches and facets.

Here, of course, we are interested in the Stammbaum of the amorphous metals. In fig. 1 we have a picturesque exhibit of this wonderful Stammbaum and, indeed, it is a tall tree which is now in full bloom. This meeting exemplifies on the one side the vigorous roots producing these alloys and on the other the various strong branches representing magnetic, electrichal and mechanical properties and their applications; principles of glass forming, thermodynamics, phases and structures, short range order, surfaces, corrosion, catalysis, hydrogen and other additions – in short, the manifold branches which have developed.

fig. 1

If we were to chop the tree down – of course such a splendid Stammbaum should never really be hurt by cutting – but just imagine such a cut and what we would see are the annual rings, roughly a few over thirty. Thus it is remarkable that this young tree has grown to be so tall and magnificent in such a short time.

If we go back in time to the middle of this century, we discover a situation as shown in fig. 2: a very hard-working gardener taking good care of our very young Stammbaum.

fig. 2

The environment for studying amorphous or disordered systems was not at all favourable. To point this out one might quote the probably most published book in physics, Introduction to Solid State Physics, by Charles Kittel. His words and meanings are demonstrated in fig. 3. Thus, at that time the interest was directed solely toward the perfect crystal. Courage was needed to go in the other direction and swim against the current, producing disordered and amorphous material which some people regarded as garbage.

fig. 3 MEASUREMENTS ON SINGLE CRYSTALS ARE ALWAYS MUCH MORE SIGNIFICANT AND INFORMATIVE THAN ARE MEASUREMENTS ON POLYCRYSTALLINE SPECIMENS.

Taking a closer look at our gardener, we recognize the familiar face of Werner Buckel. Actually, there were originally two gardeners, but unfortunately the other one, R. Hilsch, passed away twelve years ago. Effectively this team produced amorphous metal films by vapour deposition on cold substrates and made the first systematic investigations which were published in a series of articles in Zeitschrift für Physik.

These early investigations had a great influence on the development of the field of amorphous metals. The spirit of those days is reflected in a discussion which took place after a talk given by Werner Buckel in the late fifties:

Shockley got up and said: I’d like to ask a question about the temperature rise in the reconstitution of the films. It seems to me that there must be a situation something like this: If you could lay down a film by putting atoms down one by one very gently with tweezers at random places, there should be a great deal of disorder in it. Now if the film is very thick, say a foot thick, if you should shoot a cosmic ray into it, and thereby nucleate a rearrangement, then the heat would not be able to get out. It seems to me that the temperature rise might be fairly substantial and the whole mass warms up because I should think you would have something like one atom in 10 to 20 with a bad bond, and this would mean some percentage of the melting temperature would be available.

Buckel replied: Yes, maybe there is a lot of energy developed at one point. However, I wouldn’t call it a temperature rise in the sense of thermodynamics. The heat is dissipated quite quickly.

After recalling a few highlights of the past it is now my great privilege and honour to announce the contribution of our gardener and pioneer, Werner Buckel. He is going to tell us about the early days of amorphous metals and demonstrate his ability to be both witty and philosophical at the same time.

THE EARLY DAYS OF QUENCH CONDENSED AMORPHOUS FILMS

Werner BUCKEL,     Physikalisches Institut, Universität Karlsruhe, D-7500 Karlsruhe, West Germany

It was a great honour for me to be invited to give this lecture. I very much enjoyed having the chance to give a short report on the beginning of our work on disordered metals which led us to the amorphous metals. It is also a great honour for me that our work found this international recognition. Many colleagues and co-workers made essential contributions to our work. They all have part of this honour. Many thanks to all who have been involved.

In 1939, Professor R Hilsch, my highly respected teacher, came from Göttingen to the University of Erlangen. Here he started his work on very low temperatures. I was a young physics student and had the great privilege of working with him. He built a small helium liquifier based on the principle proposed by E F Simon. A single expansion of helium gas from 150 bars pre-cooled to about 12 K by solid hydrogen produced approximately 50 ccm liquid He. This liquifier enabled us to study materials from 1.3 to 400 K. Hilsch’s original idea – as far as I can remember – was to cool single crystals of alkali halides to very low temperatures in order to study the effects which may occur if the mean free path of opitically excited electrons becomes larger than the crystal. From his previous work with Professor Pohl at Göttingen, he could extrapolate that this should happen at He temperatures.

But when he had this nice He liquifier, Hilsch decided to study disordered metals, in particular superconducting metals. Here, the idea was to prepare metal films with various degrees of disorder by condensing the metal vapour onto a substrate at various temperatures. At that time we already knew that this preparation – we call it quench condensation – strongly influences the superconductivity of tin. Shalnikov published in 1938 the interesting observation that a thin film condensed onto a substrate at He temperatures becomes superconducting at 4.6 K, almost 1 K higher than normal bulk tin. This surprisingly large increase of Tc stimulated us to start a systematic study of quench condensed metal films¹.

In this paper I will discuss some of our earlier experiments. Towards the end I will try to give a little outlook on the future of research on amorphous and glassy metals.

First of all, we repeated Shalinkov’s experiment. The first figure shows the resistance of a quench condensed tin film. The substrate was at 4 K during condensation². The resistivity is rather high due to the high degree of disorder produced by quench condensation. The transition temperature is 4.6 K. The disorder decreased with increasing annealing temperature. Simultaneously, the transition temperature Tc shifted to the value of bulk tin; the residual resistivity diminishes monotonously.

At that time we were mainly interested in superconductivity and we asked ourselves whether it would be possible to increase the disorder even more and, simultaneously, to increase Tc as well. Almost at the same time, we started in situ electron diffraction studies on such quench condensed films. From now on I will present the results irrespective of the historical development.

FIGURE 1 Resistance behaviour of a quench condensed Sn film. The substrate was at about 6 K during condensation.

From the electron diffraction one learns that the quench condensed Sn films grow in the normal white tin structure. Figure 2 shows the electron diffraction pattern taken from such a film. These are transmission pictures. Pattern (a) shows the intensity we had from the substrate; a little hole (20 μm in diameter) covered by a thin collodion film at 4 K. Pattern (b) was taken immediately after the quench condensation of pure tin. All lines of the white tin structure appear³.

FIGURE 2 In situ electron diffraction diagrams of a quench condensed Sn film. The first pattern is taken from the substrate before condensation. The different patterns are taken after annealing to the temperatures given in the figure.

After this result, the question was: how can one increase the disorder in the crystalline state or even prevent crystallisation? The answer was rather obvious. One had to condense simultaneously with the tin atoms some foreign atoms which do not fit into the host lattice but which should have some affinity to tin so that they are not precipitated completely during the condensation process. We looked in Hansen’s book on binary systems and found that Cu could be a good candidate. Tin has almost no solubility for Cu; however, some intermetallic compound exists in the system SnCu.

Therefore, we quenched condensed tin with 10 at% Cu onto a substrate at 4 K. The result of this experiment – which was done in the late forties – is shown in figure 3. pattern (b) shows the electron diffraction diagram of the as-quenched condensed film. This film grew amorphous. The amorphous state is rather unstable. The crystallisation temperature is near to 20 K. Crystallisation forms the white tin structure. Let us now look at the resistance of such an alloy film. In figure 4 the resistance is plotted versus temperature as for the pure tin film. The resistance of the as-quenched film is a factor of two larger than that of a corresponding pure tin film. As expected, a sharp drop occurs with crystallisation. Also, the superconductivity has drastically changed. The amorphous film becomes superconducting at about 7 K³.

FIGURE 3 In situ electron diffraction diagrams of a quench condensed Sn film with 10 at% Cu. a) Substrate before condensation; b) SnCu film immediately after condensation; c) after annealing to 20 K; d-f) after annealing to 50 K, 200 K and 300 K respectively.

FIGURE 4 Resistance behaviour of a quench condensed Sn90Cu¹⁰ film. The substrate was at about 6 K during condensation.

Many interesting questions arose from this result. As that time we did not have a microscopic theory of superconductivity. Problems concerning the superconductivity could not be solved until the sixties. I cannot go into details here.

Summarising these results, one can say that we succeeded ċ.in preparing amorphous SnCu samples. We had to use a second component to stabilise the amorphous state. We also found that other materials, such as Au, Al Zn and even Ge, can act as stabilisers. We studied the crystallisation temperature as a function of the composition. These experiments have many common aspects with the problems discussed at this conference.

Let me close my paper with some personal remarks. Hilsch and his co-workers did not search for amorphous materials. However, we did systematic experiments on the influence of disorder in metals. These experiments led us to the amorphous state of metals and some surprising results concerning superconductivity, the superconductivity of amorphous bismuth, for instance. From this experience one should keep in mind that basic research has to be done and has to be supported outside any project with well-defined goals.

Of course, we are presently in a situation where we can start special projects to look for materials which may have special advantages for application. I am sure that we will make great progress in preparing tailored materials. We will learn to produce glassy metals in three-dimensional pieces. The preparation by solid reaction points in this direction. The formation of an amorphous state in this way discovered by Johnson and Samwer is an exciting new facet of the large field of glassy and amorphous metals⁴.

Besides the search for new materials, we will gain more and more understanding of basic principles and methods to describe amorphous materials. Here, I think, we are just at the beginning. As the translation symmetry forms the basis for describing the crystalline state, a new basis is necessary to describe the disordered state adequately. Perhaps topological arguments will help here. We need a better – I should say a more basic – understanding of the stability of the amorphous state. The role of short-range order, chemical or topological short-range order has to be cleared. The influence of the free electron system is by no means satisfactorily understood.

I am convinced that the amorphous and glassy state of metals will remain a challenge for solid state physicists for at least the next ten years.

REFERENCES

1. Shalnikov, A. Nature, London. 1938; 142:74.

2. Buckel, W., Hilsch, R. Z Physik. 1952; 132:420.

3. Buckel, W. Z Physik. 1954; 138:136.

4. Yeh, XL, Samwer, K., Johnson, WL. Appl Phys Letters. 1983; 42:242. Schwarz, RB, Johnson, WL. Phys Rev Letters. 1983; 51:415.

OPENING LECTURE

THE SCOPE AND TRENDS OF DEVELOPMENTS IN RAPID SOLIDIFICATION TECHNOLOGY

N.J. GRANT,     Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA

Publisher Summary

This chapter discusses the scope and trends of developments in rapid solidification technology (RST), which is a young field. Not much is expected because RST is synonymous with structure control. The research coverage is quite unbalanced with heavy overconcentration in some areas and near total neglect in others. With some exception, industry seems reluctant to make positive moves both into RS particulate-related equipment and facilities and into alloy programs. Despite the youthfulness of RST, a significant number of alloys and products have already emerged as commercial entities. There are growth prospects in glassy alloys, microcrystalline alloys, and the combination of the two. The bigger field would be based on the microcrystalline state in part because it is less composition restricted. RST offers almost unlimited potential for new structures, new alloys, and new processing and processes. Joining of RST-based products has been almost totally neglected. Lack of an advanced joining technology proves to be a major barrier to growth. High conductivity, high strength, high temperature copper-base alloys based on the systems Cu–Zr, Cu–Cr–Zr, C–Ni–Ti have been produced from RS–PM alloys that have excellent strength and ductility properties from 20°–450°C. Considerably stronger stainless steels are possible, with promise of improved intermediate temperature creep strength.

1 INTRODUCTION

Having been deeply involved in almost all aspects of rapid quenching (RQ) and rapid solidification (RS) very nearly since the meaningful awakening and growth of this highly exciting and highly promising field, it is perhaps quite proper and timely to look at the several different phases of RST (RSP) and attempt an appraisal, an evaluation of where we are going, how rapidly or slowly we have made progress, how much we have accomplished, and most importantly, what have we missed, how good or bad is our balance of research, technology, industry. From such reviews it is my hope that we might initiate new efforts, intensify some existing ones and generally rebalance our talents, our resources, our industrial efforts to move this new science and technology more aggressively into the production of the better materials which this world clearly needs.

Before undertaking this task, it is only proper to set a reference line as to where we are and how effectively and efficiently we got there. Of the many hundreds of visitors I’ve had from all over the world during these past ten years, from scientists, engineers, govenment people and industry, the most unfair and unrealistic question which is asked is, Why, in view of the unusual, excellent combinations of alloy structures and properties, is the field of RST growing so slowly?

RQ-I, Yugoslavia, 1970¹ was the first effort to make an appraisal of rapid solidification. Two hundred papers could be identified as of that date. Very few of these papers dealt with engineering properties and then only in a cursory manner. There was tremendous acceleration of research, publications, engineering data, etc.; (over 10,000 papers have now been published) but in terms of total engineering coverage, in terms of fabrication and processing, the net effort is relatively small.

In the metals arena, a useful guideline states that it takes about ten years to move from the laboratory to production; that guideline is well documented and supported. On that basis, RST is an infant. The industrial successes which are recorded represent excellent progress for a new field: high speed tool steel production, jet engine superalloy disc production, glassy brazing alloy ribbon, various soft magnetic glassy ribbon and strip, stainless steel seamless tube production, concrete reinforcement pins, continously hot rolled particle sheet aluminum, etc.

By far the more fair and proper question to ask ourselves is, What’s missing; what aren’t we doing correctly or adequately to achieve commercialization more rapidly? I hope that we can find some answers, or at least point out areas of new or intensified efforts, and directions to help find the answers. For now, my opinion is that RST is ahead of schedule, ahead of traditional metal based commercialization rates, but, clearly, we should have been doing better than we are.

2 PARTICULATE PRODUCTION

2.1 Atomization

On the positive side, there have been definite increases in melt and atomization sizes so that now the more common size is 1 to 2.5 tons. Nevertheless even these increased melt sizes are not ideal for competitive pricing. After several attempts to succeed with air and water atomization, rapidly solidified powders are now predominantly inert gas atomized, which should be of no great surprise since RS powders tend to be extensively and complexly alloyed.

Since subsonic gas atomization appears to be restricted to about 10² K sec −1 at best, which for many alloys and structures is below the desired level of 10⁴ to 10⁵ K sec−1, it is surprising that so few new atomization developments have been reported. There are two exceptions, namely the Osprey process² and the Kohlsva Ultrasonic Gas Atomization process (USGA) which has been under major modification and development in our laboratories at MIT³,⁴. The Osprey process appears to be able to achieve solidification rates up to about 10⁴ and the USGA up to 10⁵ K sec−1. Other atomization processes such as the Rotating Electrode (centrifugal) Process and the Soluble Gas Process (H2 evolution) at best achieve rates lower than 10² K sec−1. The cup and dish RSR centrifugal atomization, augmented by dynamic helium quenching, can achieve rates up to 10⁵ K sec−1 but the powders are very coarse and frequently are multi-structured (fine and coarse dendritic zones)⁴.

Of note, the USGA process has been used to produce glassy alloy powders⁵ in a number of compositions. For the alloys prepared to date, the finer fraction, typically finer than -45μ or -63μm is fully glassy, whereas the coarser powders show progressively more crystalline phases.

There are potentially great rewards in structure control in many glassy and microcrystalline alloys if the solidification or quench rates could be increased to 10⁶ K sec−1. One advantage would be the potential for glassy alloy production in powder form versus chopped ribbon (melt spinning); other rewards would be easier raw material handling, better packing density, less surface area, and ultimately higher production rates. In view of the interesting and practical structures and properties being reported for crystallized prior glassy alloys⁶,⁷,⁸ quench rates higher than 10−5 K sec−1 are of great interest and commercial potential.

One obvious way to achieve the desired high RS rates is to produce much finer powders. Early Kohlsva studies⁹,¹⁰ reported high yields of powders in the range of 1 to 10 microns, with average particle sizes of 3 to 4μm. These were results primarily for low melting metals and alloys but the results suggest that such much finer powders are possible where very high solidification or quench rates are desired. Production of these ultra fine powders is bound to be more expensive; however, where unusual structures and properties are desired, a working process to accomplish this should have high priority in RST. We are just beginning a program to try to produce RS powders finer than about 10μm.

2.2 Substrate Quenching

Considering that substrate quenching is the newcomer in rapid solidification, the developments here have been quite spectacular, going from the Duwez gun technique¹¹, which produced fractional gram lots of splats, to various piston and anvil devices, to rotating devices starting with the Pond technique¹², to melt spinning, to melt extraction, to the twin roller technique. All of these substrate quenching devices have the potential for achieving solidification and quench rates of 10⁵ to 10⁶ depending on the splatted particulate thickness. The gun technique can achieve rates up to 10⁹ K sec−1 because of its ability to produce near micron thick splats¹³, however, in very small quantities. Several of these processes will probably never move beyond the laboratory experimental stage, but there are variations which are bound to achieve higher RS and RQ rates, which I predict will emerge within the next five years.

For now, where ribbon and strip are desired for the glassy state, the major limitations of the versatile melt spinning process are thickness (maximum basically of about 50μm) and surface finish where a final as quenched product is desired. For the moment there do not appear to be any break-thrus in achieving higher quench rates in thicker glassy products. Note that the upper thickness limit of about 50μm is also the same for the USGA atomized glassy alloy powders. Melt extraction is more versatile in producing a broader range of shapes than is possible by melt spinning; these include semi-powders, dog-bone shapes, and a variety of particulates in general.

The twin roller technique has never achieved the importance many of us foresaw. It turned out to be a better device for producing non-continuous particulates (foils, flakes, short ribbons) and is not competitive with melt spinning. The twin roller process does achieve high solidification rates, perhaps up to an order of magnitude greater than by melt spinning, but the dwell time in contact with the rolls is far too short to achieve the desired solid state quench rates. Further, when the molten stream contacts either roll surface before passing through the roll gap, partial to complete solidification can occur, resulting in an undesired hot rolling step which results in severe peening of the soft copper roll surfaces.

Tried early in RS history was the technique of atomization against a flat disk to produce splatted particulates. Excellent solidification rates were attained on average, but the early atomization processes were not outstanding. Stripping of the particulates was a problem¹⁴. Various centrifugal techniques which splatted particulates against a confining wall also encountered problems of atomized droplet size and stripping problems¹⁵.

The combination of highly improved atomization techniques with the twin roller device now looks attractive for producing thin splats with potential for attaining solidification rates in excess of 10⁶ K sec−1. Fig. 1 shows a schematic view of the simple process of splat atomizing onto the internal roll surfaces. Fine liquid metal droplets are desired, moving at high velocity and splatting against the moving roll surface. Dwell time should be long enough not only to achieve high SR’s or QR’s, but also very high solid state quenching rates. Using the USGA process, Rai¹⁶ has produced splat flakes as thin as 50μm by 4 to 5mm in the other two dimensions, see Fig. 1. It was observed that high quenching rates were enhanced as the flakes passed through the roll gap ċin intimate contact with the two roll surfaces. Current developments seek both thinner flakes (higher droplet velocity and finer droplet size) and greater uniformity of thickness. To date, stripping of the flakes has not been a problem, but too few compositions have been attempted to be confident that stripping will not become a problem. It is anticipated that twin roller splats should be able to produce glassy alloys in many compositions of interest. Further, production of twin roller splats will by-pass the chopping or hammer milling of long ribbons, which works particularly well for the more brittle glassy alloys but is difficult with ductile or tough microcrystalline alloys. Contamination from the hammer milling operation may be a problem with certain alloys.

FIGURE 1 Splat formation by atomization into twin roller gap. Thicknesses are less than 50μm.

The melt extraction process has some of the same potential as the twin roller for the equivalent of splat formation but will require wheel dressing, will have greater difficulty in controlling thickness, will not achieve as high production rates, and will probably experience lower quench rates. It is not too much of an extrapolation to anticipate further improvements in particulate size and shape control, and the attainment of higher SR’s and QR’s than 10⁵ K sec−1 through equipment modifications or newly designed liquid droplet delivery systems.

In the meantime, melt spinning is being applied to ribbon production at quench rates of at least 10⁵ K sec−1, with subsequent comminution, crystallization (if formed as a glass) and consolidation, as will be discussed below.

The final problem of concern is the degree of oxidation which particulates undergo both in being formed and in subsequent handling and processing. It appears that oxide formation is poorly monitored or controlled. In alloys which form protective, refractory oxides, the amount and types of films can present major problems: alloys are those based on aluminum, all of the stainless steels, most of the superalloys, the reactive metals Ti, Zr and their alloys, and any alloy containing significant amounts of Al, Cr, Si, Ti, Zr, Li, Be, Ce and other rare-earth metals, etc. In the presence of reactive (to oxygen) metals or alloying elements, the powder size and shape are important, as is the oxygen level of the atmosphere where the particulates are formed. Further contamination will take place during final exposure of the particulates during handling and processing. For example, in Al-Li alloys, the presence of 2 to 3 weight percent Li increases the oxide content for a given powder size and may change the oxide composition to AlLi02 from Al203.

We do not know how to remove the oxides once formed in these classes of alloys since reduction of the oxide is out of the question, and chemical cleaning of very fine particulates with large surface areas is unattractive for many reasons, including cost. In view of subsequent problems with ductility and toughness, these are not trivial problems and significant studies must be undertaken to minimize the role of non-reducible oxides in all particulate based alloys. Further discussion of particulate based oxides will follow below.

3 CONSOLIDATION

It appears safe to state that the most neglected sector in RST is consolidation. Nevertheless a close look at RST shows that several far reaching developments are closely associated with the consolidation step; these are hot isostatic pressing (HIP), hot extrusion (from the viewpoint of a preferred hot working process), and superplasticity (by virtue of the easily attainable, highly refined grain sizes in multi-phase alloy structures.

3.1 Hot Isostatic Pressing

HIP developments owe their progress to a potential for achieving near-net shape configurations, or merely to an ability to produce useful intermediate forms at full density for efficient subsequent processing: for example, fully dense right cylinders for superplastic shaping of superalloys to near net shape turbine disks.

What are some of the new HIP developments?

1. An increase of gas pressure to 45,000 psi (310 MPa), which will permit the use of lower HIP temperatures, therefore permitting greater retention of the refined structure of the particulates.

2. Significantly faster heat-up schedules as well as faster cooling cycles after HIPing. There will be significant savings in costs associated with these more rapid total cycle times. External heating and hot charging, and earlier (warm) discharge schedules appear quite practical.

3. Larger vessels. The increase in pressure vessel size, particularly the diameter, is quite remarkable for such a new development. From working diameters of 4 to 8 inches (about 10 to 20cm) in the 1960 period, there are numerous 40 inch diameter (1m) HIP vessels, one 47 inch (1.2m unit), and a commitment to a 60 inch vessel (1.52m) for the immediate future. In terms of cost per unit of weight or volume processed in these much larger HIP units, savings will be quite remarkable and of extreme interest to industrial groups.

4. High temperatures have never been a critical issue except perhaps in terms of very large particulate aggregates. Resistance windings for HIP heating furnaces have not been a limitation in terms of temperature. To use higher temperatures than, for example, 1500°C (1773 K), one normally uses a smaller furnace (volume) to permit more effective insulation within the HIP unit. Thus high melting temperature ceramics, intermetallics and refractory metals and their alloys can be accomodated. In fact the amount of HIP processing of ceramics has been increasing very rapidly.

Perhaps the most serious problem in trying to produce near-net shapes by HIP is the limited plastic deformation that the particulates undergo. This is especially true when the particulate compositions are based on alloys which form protective refractory oxide films or skins. Figures 2 and 3 show the structure of an RS-PM HIPed IN-100 nickel-base superalloy (which contains about 5% each of Al and Ti, plus 10% Cr) before and after stress rupture testing at 982°C (1255 K)¹⁷. The fully dense alloy, which is completely recrystallized, was HIPed at 1204°C (1477 K) for 2 h. A very wide range of recrystallized grain sizes is observed, with the recrystallization being characteristic of each powder particle and constrained from growing across prior powder boundaries due to the presence of fine fragments of the oxide films originally coating the powders. The resultant grain size variations can be as large as 1 to 100μm, commensurate with the range of powder sizes used for HIPing. In the structures there are grain boundaries which are clean (internal boundaries), and inter-powder-particle grain boundaries, which are pinned by the fine oxides residing there. Note that the intercrystalline fractures take place along prior powder boundaries, and not on the clean grain boundaries. Both the large variations in grain size, which result in large differences in strength and ductility across common grain boundaries, and the oxide enrichment at the prior powder boundaries are detrimental to achieving maximum strength, ductility and toughness. The much stronger coarse grains, which interface with the very fine, weaker grains, lead to early fracture at such sites, as seen in the micrographs. See Table I for a comparison of as-cast properties, HIPed properties, and hot extruded tensile properties for two well-known cast superalloys: IN-100 and Mar-M-509.

Table I

Room Temperature Tensile Properties of IN-100 and Mar-M-509 for Several Conditions

FIGURE 2 Fully dense, as HIP’ed (1204°C, 1477 K), RS-PM IN-100 alloy showing wide range of grain sizes.

FIGURE 3 Interparticle (also intercrystalline) fractures of structure shown in Fig. 2; stress rupture tested at 982°C (1255 K). 10% elongation.

The use of a narrower range of powder sizes would help by producing a more uniform grain size, but the yield of useful powders would undoubtedly suffer, as would the cost.

A combination of HIP plus hot work to achieve significant area reductions enhance the structures (which become uniform) and the properties. As discussed below, hot extrusion is equally effective. See Table I.

3.2 Hot Extrusion

For the consolidation of particulate masses, the hot extrusion process is nearly ideal. The use of extrusion ratios of about 4 to 1 (diameter) or 16 to 1 (area) literally guarantees full consolidation. Greater area reductions are always desired to obtain optimum hot working of the dendritic powders, and to achieve the finest possible uniform grain sizes. In particular, for alloys which carry along non-reducible, refractory oxide films, extrusion plus other forms of hot and cold working to achieve reductions of area of 200:1 and greater are highly desirable to obtain optimum distribution not only of oxides but other phases.

The single biggest problem with hot extrusion is the lack of sufficiently large presses to permit working with large particulate billets. A 1400 ton press can process a 7 inch (about 18 cm) diameter powder billet with an area extrusion ratio of about 16 to 30:1, hardly large by industrial standards. But, in terms of extrusion consolidation to provide an adequately strong, workable billet for subsequent processing by more conventional rolling and forging techniques, an area extrusion reduction of only 4 or 8:1 may be adequate in most cases and would require much less energy and would permit larger initial working diameters at relatively low temperatures.

Extrusion presses are used routinely to densify canned particulate billets or compacts by stalling the press against a closed block, machining off the can and re-extruding the bare, almost fully densified billet. Using the extrusion press to hot upset a low density powder billet is not as effective for producing useful structures and properties as is the low area extrusion reduction described above, which has the advantage of achieving partial breakdown of the surface oxide films which are present and recrystallizing the dendritic cast structure of the powders.

Clearly, for RST to grow, much larger hot extrusion presses will be needed. Initial canned powder compacts of 20 inches (about 50 cm) diameter would permit consolidated extrusion of 4 to 9:1 area reductions (2 to 3:1 diameter reductions) as a minimum. HIP consolidation can easily exceed 50cm diameter but would have greater potential if done in series with a large extrusion press for a great many complex alloys of interest.

Issues of temperature, strain rates and strain for extrusion are largely unanswered because the experience base is small. From the point of view of structure/property benefits, temperature, area reduction and strain rates are extremely important variables for which answers do not exist in the great majority of cases. The lowest temperature, high strain rates and the largest attainable strains are desirable for achieving the finest structures; unfortunately available extrusion equipment offers little control of strain rate. One selects the lowest temperature to obtain the desired area reductions for consolidation and dispersion of phases, accepting the resultant deformation rates. Temperatures typically are 5 to 10% lower than required for ingot hot working; this is possible because the RS particulate-based alloys are much less segregated, have a much finer grain size, and finer and fewer intermetallic phases. The selected hot working temperature is lower and the time at temperature to attain maximum solution and homogenization is much shorter than for ingot materials.

3.3 High Energy Rate Solid State Consolidation

Explosive compaction (cold)¹⁷, and high velocity projectiles and shock waves¹⁸ (usually cold) have been studied as methods of solid state compaction of RS particulates. In small sizes densification is successful but apparently the damping effect of a collapsing particulate mass is sufficiently effective to prevent full compaction of even relatively smallish compacts, for example, a 5cm cube. This is unfortunate since the structures of RS particulates cannot be preserved in large bodies, and some recrystallization, grain growth and phase changes will take place in many alloy systems on any subsequent heating and hot working.

The consolidation of particulates by cold, high energy rate methods also suffers the same problems as does HIP, namely, if oxide films are present on the particulates, the films will undergo some fracturing to provide metal to metal bonding, but the negative aspects of having oxide dispersions along prior particulate boundaries will be detrimental to ductility, toughness, corrosion and pitting problems, etc. This is unfortunately true for both glassy and microcrystalline metastable alloys.

3.4 Oxide Dispersions in Consolidated Particulate Products

Alloys based on reactive metals which form stable. refractory oxides, or alloys containing significant amounts of such alloying elements will form continuous oxide films or precipitates in quantities large enough to present problems in the fully consolidated state. These problems include decreased ductility, impaired toughness, pitting corrosion and selected general corrosion in proportion to the amount and distribution of the oxides. Aluminum alloys in particular have encountered these problems¹⁹,²⁰,²¹, but the problems also exist for the stainless steels, and for superalloys, to name a few. It means that HIP’ed aluminum cannot be a final product if strength, ductility and toughness are requirements. It also means that superalloys, such as the IN-100 alloy, may be subject to premature failure in the HIP’ed condition (see properties in Table I).

Unfortunately, if these problems exist for hot mechanically pressed and HIP’ed products, they would also be expected to exist to some degree in hot worked products which have undergone a limited amount of hot and cold work.

In spite of the fact that RS-PM alloys, aluminum alloys, for example, may have significantly higher yield and tensile values than a comparable ingot alloy, the much more highly refined RS-PM alloy will have equal or even slightly poorer ductility values which are related to high oxide contents, as seen in Table II²²,²³.

Table II

Tensile Properties of Ingot versus RS-PM Hot Extruded Aluminum Alloys²²,²³

*Coarse powder produced (>75μm)

**Fine powder product(<53μm)

In toughness tests, in spite of greater yield and tensile values, combined with a very fine grained structure, RS-PM aluminum alloys in fatigue crack growth rate tests are not superior to their less homogeneous, coarse grained ingot counterparts, as seen in Fig. 4.

FIGURE 4 Fatigue crack growth rates as a function of stress intensity range for 2024 plus lithium aluminum alloys (A-T4 with 1% Li and B-T6 with 3% Li) versus ingot material of T4 temper.

The reason seems quite clear, namely, the presence of fine oxide stringers in the extruded RS-PM alloy. In longitudinal tests these oxide stringers are of no concern, but in transverse tests, with the fracture short circuiting along oxide stringers, premature failures are common. For lithium containing aluminum alloys, which appear to have both a larger quantity of oxides and somewhate coarser oxides, ductility and toughness behavior are relatively worse than for high strength lithium free alloys. A severe case, but not a rare one, is demonstrated for an RS-PM 7075 alloy containing 1% Ni plus 1% Zr: The fracture path of the notched compact specimen changed from its normal direction by 90°, along a severe oxide stringer (Fig. 5)²¹.

FIGURE 5 Precracked fatigue specimen with fracture path shifted 90° due to oxide stringers in extruded RS-PM 7075 alloy containing 1% Ni + 0.8% Zrn (21).

As a particularly sensitive test of the role that oxide dispersions can play on fracture initiation and growth in RS-PM alloys, Kang and Grant²⁰ processed an X2020 alloy containing 1.6% Li by using the common practice of cold working (stretching) of the bar about 1.5 to 2% after solution heat treatment but prior to aging. Cold stretching increases the yield strength and enhances the elongation slightly for ingot grade alloys. In the X2020 RS-PM alloy, extruded at 400°C (673K), with an extrusion ration of 30:1 (area), the change in yield strength was small but generally positive; however, the ductility almost always decreased at least 1%. Examination of several cold stretched test bars, in the longitudinal direction, always showed isolated volumes of intercrystalline cracking: see Fig. 6. These fracture sites were associated with fine powder clusters in the extrusion; such fine powder volumes are richer in total oxide content and apparently are somewhat shielded during the hot extrusion step. In a tensile test the results are minor; in a notched toughness test the results are indicated to be severe.

FIGURE 6 Optical micrograph of longitudinal section of Al-Cu-Li (X 2020) alloy in T6 condition, with 2% cold stretch after solution heat treatment, showing intercrystalline cracks associated with oxide concentrations (20).

The alloy and alloy production steps which will determine the degree of damage due to oxide presence are the following:

1. Alloy composition, as discussed above. Specific alloying elements can have a large effect.

2. Powder size and shape (powder surface area, actually).

3. Atmosphere quality during particulate production, and in subsequent handling: time in a given atmosphere.

4. Consolidation variables.

The last item is more important than generally admitted. For a given average powder size, including powder size range, each particle undergoes a somewhat different reduction during hot extrusion. Assuming a right cylinder of equal length and diameter as an approximation of a powder sphere, Table III shows the changes in length and diameter of two representative powder particles.

Table III

Change in Diameter and Length of Two Powder Particles in Selected Extrusion Conditions

For purposes of this discussion, we are’interested in the final diameters of the extruded powders. In actual fact the coarser powders will extrude more ideally than the finer particles, and will respond better to higher extrusion ratios. Fine particles, particularly clusters of fine particles, will tend to do more sliding and less plastic extruding. Further, the finer particles tend to be more heavily oxidized than are the coarser particles. Nevertheless, Table III demonstrates that the finer oxide coated powder particles, for a given extrusion reduction, will tend to have the dispersed oxide fragments closer together, that is, more homogeneously dispersed. Further, the higher the reduction, the closer and more homogeneous will be the oxide dispersion frequency.

It has in fact been observed that larger powder compacts, say 7 to 10 inches diameter (18 to 25cm), when extruded to a given final bar diameter with a reduction ratio (area) of 200:1 or larger, will have better ductility and importantly better toughness properties than a 3 inch (7.5cm) diameter powder compact extruded at a nominal 30:1 (area) reduction. Ingots, which are of much greater cross section than any current powder compacts, combined with a coarse grain size and coarse constituent phases, but which are low in oxide volume, unexpectedly have good fracture toughness properties. The evidence suggests that the very large reductions necessary to produce final bar sizes are able to overcome the negative aspects of the coarse constituent phases, and therefore infer similar benefits to be derived through large reductions of RS-PM alloys.

3.5 Liquid Dynamic Compaction (LDC)

Liquid Dynamic Compaction is the alternate to Solid Dynamic Compaction (explosive compacttion, high energy impact processes) and offers a number of advantages that are worthy of close study. Liquid Dynamic Compaction has parallel roots with Plasma Gun Techniques, the Osprey Process² and the CSD-Aurora Steel Process²⁴, but is a more definitive term for spray atomization and collection.

One way to avoid the issue of oxide entrainment associated with processing of solid particulates would be to collect the atomized metal, preferable in the partially liquid (or solid) state in a protective atmosphere. There remain very many variables to be assessed and controlled, but recent results confirm and extend the promising results reported for the Osprey and CSD processes.

Using a 7075 + 1% Zr alloy for which a fair amount of data have been collected from small RS-PM extrusions (1.5 to 3 in. diam.: 3.8 to 5.6cm), with an extrusion ratio of 30:1 (area), and one large RS-PM extrusion (7 in. diam.: 18cm) with a combined extrusion ratio of about 200:1 (area), plates approximately 1 inch (26 mm) thick by 6 × 6 inches (15 × 15cm) in cross section were deposited against a water cooled copper substrate. The atomization was done in an argon atmosphere, using argon for atomization. The typical density was 96% with well scattered, siolated near micron pores. A hardness traverse across the plate thickness showed no variation of consequence, confirming that the porosity was fine and well scattered, and suggesting that the cooling rate from top to bottom of the plate was quite constant²⁵.

The high velocity atomized droplets were in fact splat quenched first against the copper substrate, and then, for the bulk of the thickness, against its own composition. Of particular interest was the uniform, equi-axed, fine recrystallized grain size as shown in Fig. 7. Since one of the aims of LDC processes is to avoid solid powder processing steps, with the unavoidable exposure of the powders to an oxidizing atmosphere, it is interesting to compare the fractures of this 7075 + Ni + Zr alloy extruded from solid powders versus that made by the LDC process. Figure 8 shows the typical delaminated tensile fracture of the RS-PM product and the delamination-free fracture of the LDC product. The absence of delaminations in the LDC produced material confirms the very much lower oxide content that is possible by controlled spray atomization and collection. Property comparisons for this alloy produced by several techniques are shown in Table IV (for equal hot extrusions and heat treatments)²⁵.

Table IV

Comparison of Tensile Data for RS-PM and LDC Produced 7075+Ni+Zr Aluminum Alloys and Alloy 7075

FIGURE 7 Microstructure of liquid dynamically compacted (LDC) 7075 aluminum alloy containing 1% Ni + 0.8% Zr. Density is 96%. Structure is fully recrystallized in as-deposited condition (25).

FIGURE 8 Tension fracture surfaces in an RS-PM 7075 + 1% Ni + 0.8% Zr aluminum alloy. a) 400°C (673 K) extruded powder product; b) LDC product with same reduction ratio (25).

LDC processes have much to offer, especially when oxidation of RS-particulates is a problem. The process is now beginning to gain attention in view of the excellent results reported from the several referenced sources. We believe that issues of toughness and poor transverse properties can be significantly improved via the LDC processes.

4 ALLOYS AND ALLOY SYSTEMS

If a criticism is to be voiced regarding developments in alloying, it’s that too much of the effort is concentrated, first, on reproducing existing alloys but with improved properties, or, second, making minor modification in existing alloys to improve some one specific property, for example, corrosion. The first of these approaches has little opportunity for success because the new process (RST) will start as a low volume effort, requiring changes in production facilities, and will be more expensive. Furthermore one usually finds that most customers are not unhappy with the conventionally produced product and have installed all the necessary facilities to optimally process and use the product.

The second approach is more realistic but still not optimal. There is always the danger that an alloy producer will find itself with two products of lesser volume, two production methods and two groups of customers.

The message would appear to be: as a first choice, seek to develop new alloys, not a better one or a slightly different one, but a new alloy, with different chemistry, very different structure, and a better set of properties. This portion of the paper will examine several interesting new alloy developments, primarily to point the way.

Alloy research in the RST programs has been concentrated on aluminum alloys and superalloys among microcrystalline materials, not too surprisingly. High speed tool steels found early industrial support and have emerged as successful commercial products. Some recent activity has also been apparent in the stainless steels, but the effort is much narrower and quite new.

Glassy alloy developments are of note because there were no glassy alloys to improve or mofify, and the effort was quite original in terms of new alloys, processes and products. Major emphasis was on the soft magnetic alloys, and it’s worthy of comment that hundreds of new compositions were studied, far beyond that possible with ingot based alloys. Brazing alloys have also been successfully developed and are worthy of note. Other developments will be discussed below.

4.1 Aluminum Alloys

The effort to develop RS-PM aluminum alloys has been very large and the literature is heavy with references. The most notable developments were those based on AL-Li²⁸, a reasonably new effort in view of the lack of prior success with ingot materials before about 1970. The preferred alloys were originally found in the Al-Cu-Li and Al-Mg-Li systems, with the former preferred for high strength. The Al alloy 2024 was modified²³ with 1 to 3 w/o Li (3 to 11 a/o), not a small change in view of the high atomic percent of lithium and the large volume content of the AL3Li phase introduced into the structure. From an aircraft or aerospace point of view these lithium alloyed aluminum alloys offer not only high strength but high specific strength and specific modulus as a result of the lithium additions. Typical best properties of these lithium alloys are shown in Table V, with corrections for density.

Table V

*Mechanical Properties of Selected Al-Li RS-PM Alloys

*Reference state for density corrections for Al-Li alloys

1. Soviet Al-Mg-Li ingot alloy.

2. Contains 6.7% Mg and 1.6% Li. Ref. 29.

3. Contains 4.4% Cu and 1.55% Li. Ref. 20.

A series of more complex Al-Li alloys containing additions of Zr, Cu + Zr, Cu + Mg + Zr, etc., are now undergoing detailed study in an effort to achieve not only high specific strength and modulus but also to improve corrosion resistance and toughness. As was noted above, under the Consolidation Section, the high lithium alloys oxidize more readily than their lithium free counterparts, and form a less protective oxide of AlLi02 as contrasted to γ-Al203.

Al-Li alloys continue to merit major additional attention, and have much to offer for light weight, high strength, tough alloys to the entire transportation and aero-industries.

The other area of prime interest among All alloys is that of high temperature performance. Again the choice of alloy compositions shows marked departures from conventional ingot materials with major additions of combinations of Ce, Fe, Ni, Mn, Co in quantities of 3-10% or more, in an effort, through rapid solidification, to attain a high volume content of finely dispersed stable intermetallics phases³⁰. To date, although useful high temperature strength and ductility values have been achieved, ductility and toughness values at room temperature need further improvements.

The SAP (0DS Al alloys) technique of oxide dispersions in simple Al alloys has been bypassed for now, but the use of simple solid solutions plus oxides (or carbides and other intermetallics) using pre-alloyed RS powders, and blended with stable, refractory particulates, remains a viable³¹, inadequately explored area of development.

4.2 Superalloys

The history of alloy development among the superalloys has been a competition between the Ni-base γ-γ¹ alloys and the Co-carbide alloys, first among wrought alloys and more recently among the cast alloys, with a current appraisal which shows that the strongest high temperature alloys have been the very coarse grained precision cast materials, which are classifiable as being non-hot or cold workable, but also lacking in toughness and forgiveness of defects. Competitive, and emerging at a much later date, were the directionally solidified superalloys, and still more recently the single crystal cast blades. Alloys such as IN-100 and Mar-M509 are typical of these materials.* Due to problems of severe segregation, coarse structures and the presence of borides and soluble chromium carbides, the temperature of first liquid formation, generally at grain boundaries, is too close to the maximum operating or exposure temperatures of these alloys. In this respect the single crystal alloys offered a safety cushion.

In what was an interesting and exciting development, RS-PM superalloys such as IN-100, Mar-M509, etc. were produced as hot extrusions yielding, in contrast to the cast condition, ultra-fine-grained alloys (4-10μm) which were tough, had high strength at low and intermediate temperatures (to about 700°C) (973K), high ductility at room temperature, and were superplastic above about 1050°C (1323K) with remarkably low strength, at strain rates lower than about 10−3 sec−1. On the basis of these properties, totally new applications evolved for alloys such as IN-100, namely turbine disc alloys with outstanding strength to about 700°C (973K), combined with excellent toughness, and capable of near net shape forming by superplastic deformation.

Similarly, the non-forgeable cast Co-base Mar-M509 alloy (and allied compositions) was found to be readily hot extrudable when prepared from RS powders, with excellent room temperature strength and ductility (see Table I), and superplastic tendencies at temperatures above about 1050°C (1323K); the grain size was 4-8μm. The remarkable role of grain size on the high temperature strength, ductility and deformability is shown in Table VI. What emerges from Table VI is the realization that ultra-fine grained alloys require lower temperatures and much lower forces to achieve desired deformations, for example, attainment of near net shapes by superplastic or nearly superplastic deformation. If a coarser grain size is later desired, the alloy can be recrystallized. And, the awareness that increased, complex alloying does not mean a lack of hot workability or low temperature ductility will encourage development and production of stronger, tougher alloys. The rationale here applies to all classes of alloys, not just to superalloys.

Table VI

The Role of Grain Size and Structure at 1255K on the Strength and Ductility of IN-100 Superalloy At a Strain Rate of:

Note that at the very slow strain rates, deformation of the cast alloy requires 10X the stress required for the fine grained RS-PM alloy, and the hot plasticity of the fine grained structure is 15X greater. But even at 10−4 sec−1, the stress requirement for the cast coarse grained alloy is 3.5X greater than for fine grained RS-PM alloy.

In view of the vastly improved cold and hot plasticity of the fine grained RS-PM superalloys, and the low stresses necessary to hot deform the alloys, the cast Mar-M509 alloy, with nominal carbon content of 0.6% and containing 3.5% Ta for formation of the stable TaC carbide, was modified by increasing the carbon content to 1.2%, eliminating the Ta, and adding enough Hf to combine exclusively with the carbon.

It is very important to note that the high HfC cobalt-based RS-PM alloy shows tensile properties at room temperature equal to those of the coherently aged Ni-base IN-100, a comparison which few materials experts would have anticipated. See Table VII.

Table VII

Tensile Properties of RS-PM High HfC Dispersion Strengthened Co-Base Superalloy Vs Cast and Wrought Mar-M509 and IN-100

At 982°C (1255K) and higher, whereas the cast, coarse grained IN-100 alloy is significantly stronger in stress rupture than the cast, coarse grained Mar-M509 alloy, the reverse is true for the alloys with grain size finer than 10μm. Here the carbide strengthened Co-base alloy is very much stronger than the equivalent RS-PM IN-100 alloy.

4.3 0D and 0DS Ni-base Superalloys

The alternate approach to γ-γ¹ Ni-base alloys and carbide strengthened Co-base alloys (γ – carbide) is the 0DS class of alloys. Benjamin et al³² have successfully produced a series