1. Physical Metallurgy 24 th Lecture MS&E 410 D.Ast dast@ccmr.cornell.edu 255 4140

2. Metallic Glasses

3. History Long produced - without being aware of it - in electrolytic plating Rapid solidification at ~ 10 6 C/sec cooling rate Splat cooling 1963 (P. Duwez and R. H. Willens) “Melt Spinning” 1969 (R. Pond, Jr. and R. Maddin) molten metal stream impinges on the inner surface of a rapidly rotating hollow cylinder. Ribbons of 10’s of micron in thickness ~ 1985..1990 “Bulk” Metallic Glass. First Au 80 Si 20 (Harvard, Turnbull), later Zr and Pd based (Tohoku, Cal Tec). Up to 5 mm. Rates as low as 1 o K/s

4. Typical Compositions Ribbons Fe 40 Ni 40 P 14 B 6 (Metglas 2826) Bulk Zr: 58.5 Cu: 15.6 Ni: 12.8 Al: 10.3 Nb: 2.8 (Vitreloy) Meltspinning

5. Properties of Metallic Glasses (Amorphous Alloys) High yield stress (586 KSI for 2826, similar to very best steel) Zero macroscopic ductility (more later) Extremely low Saturation Magnetization (< 0.3 Oersted) High corrosion resistance (but not better than good ss with 28%Cr)

6. Theory: Mechanical Properties (C.M. Li) Start with crystalline material Introduce 10 13 dislocation per cm 2 Average dislocation spacing 3 nm No long range order - amorphous in X-ray diffraction Deformation: Yield strength => extrapolate steel data to 10 13 Deformation => balance between dislocation multiplication and annihilation => zero work hardening

7. In the C.M.Li theory, the material has (already) by ultimately work hardened and can work hardened further. Any increase in dislocation density is balanced by mutual destruction of dislocations. The climb towards each other

8. Stress - Strain relationship Strain Stress Absence of work hardening or softening: Metallic Glasses behave like ideal elastic plastic solids.

9. Yield strength of a metallic glass based on modeling it with 10 13 dislocations per cm 2. The experimental value for 2826 is 586KSI (=>5.10 13 /cm2) Li model

10. The Li model not only predicts the yield strength but the very low ductility. DUCTILITY Material does not work harden Deformation confined to narrow zones (veins) Macroscopic ductility ~ zero Tensile deformation is unstable

11. Theories of glass formation Flat atomic potential potential (Turnbull) Deep Eutectics (close to T g ) (Gilman Allied Co) “Confusion” large unit cell in crystalline state Packing of spheres of unequal size (> 14% opposite of H.R. Solubility rules) Several mechanism can be at work simultaneously - the design of “bulk” amorphous metallic alloys is still largely hit and miss.

12. Turnbull theory (same Turnbull as from undercooling) Origin Au 80 Si 20 Observation : Au fast diffuser in Si - suggesting a “flat”potential In a flat potential there is no fixed interatomic distance => glass forming !

13. Deep Eutectic Theory (Duwez, Davies at Allied) Glass => Transition temperature weak function of composition Eutectic => strong function of composition Search for phase diagram with deep and narrow eutectics In this case, you only need a small  T to freeze in the liquid => which allows you to get by with relatively slow cooling rates. The philosophy that created Metglass 2826 An other way to look at this is that you are searching for a very stable liquid. Stable liquids are more likely to form glass

14. Complicated unit cell theory (various authors) Find a crystalline phase with a very large unit cell Liquid lacks time to assemble proper crystal If possible split elements further (e.i. Replace Fe with equal Fe and Ni as in 2826 which is break up of Fe 80 P 20 ) The trick to “split up” one element in several, chemically similar ones is well known in the glass industry

15. The “sphere of unequal size” mixing theory Bernal showed that a mixture of spheres of unequal size is much more stable against crystallization than a mixture of equal sized spheres. The polymer language equivalent is to say that mixing balls of different size reduces free volume (74% for equal size spheress) Computer Simulation of the structure of an amorphous bulk metallic glass. Note the mix of atom size (Ohio State)

16. Structure There have been never ending discussion on the structure of metallic glasses. To what degree does short range order exist ? Is five fold symmetry a key to metallic glass formation ? (this goes as far back to the Voronoi Amorphon) To large degree, the discovery of “order” in a disordered system is an expression of the need of human nature to classify nature into categories (read Plato!). Thus, there never will be agreement ! But it keeps physicists employed :-)

17. Diffraction A major problem is that diffraction does not work on disordered system The relation between diffraction and structure is not clear cut, because diffraction data lack phase information Note: Presence of sharp diffraction spots do NOT necessarily indicate an ordered material with a periodic array of atoms. The famous example is the superposition of two frequencies with an irrational ratio (e.g. SQRT of 2). They add up to a non-periodic signal. In diffraction language, a non periodic object can have a perfectly sharp diffraction pattern

18. A 2006 paper in Nature... “clearly there is a pressing need”

19. An attempt (in the Nature paper) to extract from experimental diffraction data (pair correlation's for Ni-Ni, P-P, Ni-P in Fe 80 P 20 ) the real space structure with M.C. methods. Unfortunately, the computation involves compromises. And likely is not unique

20. Magnetic Properties Fairly typical. Saturation B is ~ 16KG (pure Fe 20KG) because not pure Fe (15..~20 at% of glass formers such as P,B) Saturation H is extremely low (~ 10 Oe), Earth Magnetic Field is 0.3 Oe

21. 2004 work at the Institute for Materials Research, Tohoku University and the Component and System Development Center, Toyota Motor Corporation SPS is Spark-Plasma-Sintering of metal powder Sintered amorphous metal powder, B 3 13.5 KG, H c = 6A/m

22. Loss factors IR 2, eddy current loss in coil Eddy current loss in core Amorphous metals win “because they are lousier conductors” and because they are thinner (20  vs 0.2 mm) Magnetic Hysteresis

23. Eddy losses scale with thickness with a power 1…2, thus the fact that metallic glasses are 1/10 in thickness reduces eddy current in the core by a factor 10 to 100. The resistivity of metallic glass is 3 times higher (150 vs 50  cm), giving an other factor of 3

24. Due to increase in electronic switching power supplies, electronic dimmers etc, 60 cycle current has increasing amount of higher harmonic. Measured V and I, secondary winding

25. Measured and expected (---) loss in a 500 KW transformer under actual conditions. At 60% capacity, the calculated loss was 2 and 4 KW (4 and 8% efficiency) for amorphous and Fe-Si respectively.

26. Of course, the Fe-Si industry could make 20  thick transformer sheets - it is just more expensive. And upgrade the Fe-Si to superoriented Si Steel. All a question of price and market pressure. THE END