‘h . ,7?- JM If)!» H ”‘53... Mo“ 57 4" «0'; «0" -_ c n ' W ‘1‘) u! gig/(v, I ‘ f’ u ,, . u w". _a 'v. lav} ""“=‘.,'31-I f 1 r ”1.75:, ._ . ~..‘~~*‘ | 1w.“ 4%“.- c o 9 "A. CA I v.7 Lg,,. .4r'unfi . 3L4;;h§l;l'v'un bury-inwa- Lan'ig?1.a}’ This is to certify that the dissertation entitled STRUCTURE AND PROPERTY CHANGES IN 1VIETGLAS® 26OSSC INDUCED BY PULSE LASER INTERACTION. presented by Chandrashekhar G. Wakade has been accepted towards fulfillment of the requirements for Doctor of EhilQSQphLdegree in MaLBLialLScience. Major pro essor Date _June_Zl._l_9_8_5__ MSU is an Affirmative Action /Equal Opportunity Institution 0-12771 RETURNING MATERIALS: 1V1£31_J Place in book drop to LJBRARJES remove this checkout from “ your record: FINES .wiH be charged 1f book TS returned after the date stamped beiow. STRUCTURE AND PROPERTY CHANGES IN METGLAS® 26058C INDUCED BY PULSE LASER INTERACTION BY Chandrashekhar G. Wakade A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Metallurgy, Mechanics, and Materials Science 1985 ABSTRACT STRUCTURE AND PROPERTY CHANGES IN METGLAS® 26OSSC INDUCED BY PULSE LASER INTERACTION. By Chandrashekhar G. Wakade Metglas 26OSSC foils were subjected to a high energy ruby laser irradiation to investigate the spot-welding characteristics and associated microstructural changes. Laser fluences ranging from = 105 W/cm2 to = 107 W/cm2 were used for this study. The inherent rapid cooling, associated with pulse laser interaction, preserves the primary amorphous structure of the resolidified metallic glass. However, even at power levels insufficient to produce substantial melting, spherulitic surface crystals were observed to grow. These spherulites upon coales- cence form a network-type microstructure. Crystallization products of laser irradiated Metglas 260580 aretx-Fe, Fe3B and FezB. Pulse duration, repetition and energy input determine the nature and formation sequence of these stable and metastable crystalline phases. Laser welding of two foils of the metallic glass is possible, however, it is accompanied by some embrittlement and crack formation in the heat affected zone. The nature of these deformations depends on the sample geometry and the boundary conditions associated with the hold-down device. For the first time, a direct experimental mapping of a crack—tip nonlinearity, in the form of a shear band zone, has been recorded. A theoretical model, which predicts such a shear band zone at the crack—tip, is used to discuss the elastic-plastic response of the metallic glass. Metglas® is a trademark of Allied Chemical Corporation. To Bharati and Covind, my mother and my father, who have been ideal parents to me throughout my life and to Pranjali who has been an ideal wife. ii ACKNOWLEGEMENTS I would like to eXpress my appreciation and gratitude to Dr. Kali Mukherjee, my major advisor, for his constant support, encouragement, patience, guidance and friendship, and especially invaluable assistance in the preparation of this thesis. Dr. Gary Cloud, Dr. C.-M. Hwang, and Dr. Jack Bass, my thesis committee members, for their support, suggestions and understanding. Dr. Rohan Abeyaratne, for introducing me to the theoretical aspects of nonlinear crack problems and for his useful comments and discussions. Dr. H. H. Liebermann of Allied Chemical Corporation, for providing Metglas 26OSSC and the useful technical data on the metallic glass. The Division of Engineering Research (DER) of the Michigan State University, for partial support of this research. Mr. Leo Szafranski, Ms. Michelle Ward, and Ms. Arlene Klingbiel, for their technical assistance. Mr. Narendra Dahotre, Mr. Subhasish Sircar, and Mr. Krishnan Narashimhan, my colleagues, for their technical assistance and help whenever needed. Mr. Saeid Niroumand and Mrs. Grace Niroumand, my friends, for their assistance and help in preparing this manuscript. Mrs. Pat Mukherjee for proof-reading this manuscript. Metallurgy, Mechanics, and Materials Science Faculty and Staff members for their help and cooperation. My daughter, Zuie, whose early arrival in this world, gave me a new burst of energy and desire to work harder. Finally, to my friends and family, for their love, encouragement, and support, and especially to my mother-in-law, Mrs. Pramila M. Thote, for coming all the way from India to help my wife with our new born daughter, so that I could devote my time to complete this dissertation. TABLE OF CONTENTS LIST OF TABLES 0.0.000.......00.0.00.........OOOOOOOOOOOOOOOO LIST OF FIGURES ......OOOOOOOOOOOOOOOOO....OOOOOOOOOOOOOOOOOO I II INTRODUCTION ........................................... BACKGROUND ............................................. 2.1 Nature of Glassy State ............................ 2.1.1 Liquid-Glass Transition .................... 2.1.2 Structural Relaxation ...................... 2.1.3 Kinetics of Structural Relaxation .......... 2.1.4 The Glassy-to—Crystalline Transformation ... 2.2 Class Formation ................. ....... ........... 2.2.1 Kinetic Criteria ........................... 2.2.2 Methods of Preparation ..................... 2.2.3 Solidification Processes and Quenching Rates 2.2.4 Glass-Forming Alloys ....................... 2.2.5 Factors Affecting Glass Formation .......... 2.3 Structure ......................................... 2.4 Corrosion Behaviour ............................... 2.5 Magnetic Properties ............................... 2.5.1 Effect of Deformation and Magnetic Annealing 2.6 Mechanical Properties ............................. 2.6.1 Density and Thermal Expansion .............. 2.6.2 Elastic Constants .......................... 2.6.3 Anelasticity ............................... iv Page VI VII 1 10 12 15 16 18 26 3O 33 35 37 39 49 50 50 51 Page 2.6.4 Strength and Hardness ...................... 55 2.6.5 Fatigue Behaviour and Toughness ............ 57 2.6.6 Creep ...................................... 58 2.6.7 Effect of Deformation ...................... 60 2.6.8 Annealing Embrittlement without cryStallization .......O...‘................ 62 2.6.9 Stress—Strain Behaviour .... ....... ......... 63 2.6.10 Deformation Characteristics ................ 68 2.6.11 Deformation Mechanism ...................... 70 III EXPERIMENTAL PROCEDURE ................................. 75 IV RESULTS AND DISCUSSION ................................. 78 4.1 Laser Melting and Microstructural Characteristics Of the Metallic Glass .............OCCOOOOOOOCOOOOC 78 4.2 Spot—Welding Characteristics ...................... 82 4.3 Crystallization Behaviour of Metglas 26OSSC ....... 85 4.4 Crack Formation in the Metallic Glass ............. 91 4.5 Wavy Deformation and Shear Band Formation ......... 95 4.6 High Temperature Mechanical Testing ............... 99 4.7 Crack-Tip Nonlinearity ............................ 102 V CONCLUSIONS ..... ......... .............................. 109 APPENDIXES .................................................. 110 LIST OF REFERENCES ................ ..... ..................... 112 LIST OF TABLES Page Magnetic properties of FRMDBZO glass and comparable commeer-al materials ......OOOOOOOOOOOOOO......OOOOOOOOO 41 Maximum saturation magnetization obtained in various amorphous alloy systems ................................ 44 Mechanical properties of metallic glasses: Vicker's hardness Hv (kg mm‘z), fracture strength of (kg mm-2 ), Young's modulus E (103 kg mm-2 ) ...................... 56 Nominal composition and properties of Metglas 26OSSC ... 75 Comparison of measured and calculated Go values ........ 107 vi 10. 11. 12. 13. 14. 15. 16. LIST OF FIGURES Temperature dependence of viscosities n and time constant T for structural relaxation in a Pd-Cu-Si alloy ............. The specific volume V of a Pd-Cu—Si alloy ................. Time-temperature—transformation (TTT) curves of metals .... Single foil quenching technique ........................... Piston and anvil technique ................................ Continuous casting processes: a) twin-roller quenching; b) melt-spinning (outside); c) rotating crucible spinning; d) melt—spinning (inner); and e) melt—extraction method ... Predicted cooling rate q as a function of section thickness t, and heat transfer coefficient h (w m"2 3'1) ............ Repersentative phase diagrams of glass-forming alloy systems Bernal's idealized holes describing the topology of dense random packing: a) tetrahedron; b) octahedron; c) trigonal prism capped with three half-octahedra; d) Archimedean anti-prism capped with two half—octahedra; and e) tetragonal dodecahedron .............................................. Moment per Fe atom in amorphous Fe8OBZO alloys as B is replaced by C, Si, For Ge 000............OOOOOOIOOOOOOOOOO Temperature dependences of apparent yield and tensile strengths for amorphous Fe—P-C alloy ...................... Relationship between the strain rate and the strength properties of amorphous Pd-Si alloy ....................... Effect of strain rate on the fracture stress at various temperatures ......OOOOCOO0.000000......................... Effect of stress on the start of crystallization in an amorphous Pd‘Si alloy ooooooooooooo 0000000000....0...000.00 Effect of cold rolling on diffraction patterns of an amorphous Pd—Si alloy ........ ............................. Effect of cold rolling on hardness and tensile properties 0f the Pd‘Si alloy oooooooooooo 0 ooooo 00.000.000.0.0....00.0 vii Page 18 22 22 25 29 31 36 42 54 54 55 59 61 61 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. Representative stress—strain curves of a Pd—Si alloy ...... Stress-strain curves of a Pd-Si alloy as a function of temperature OO0......0.0.0.0..........OCOOOOOOOOO00......O. Compression stress-strain curves for a Pd-Cu-Si alloy ..... Schematic diagram of the experimental set-up .............. SEM picture of a hole drilled in Metglas 26OSSC. Laser fluence :107w/Cm2 O0.00............OCOOOOOOOOOOOOOOOO.... Laser damage in Al foil. Laser fluence : 107' W/cm2 ...... Dendritic resolidification structure at the lip of a laser induced hole. Laser fluence = 107 W/cm2 ................ a) SEM micrograph showing distribution of spherulites in the HAZ, b) An enlarged view of the selected area in (a) .. Dendritic resolidification structure at the center of the laser irradiated spot. Laser fluence 2 106 ‘W/cmz ........ a) Circumferential cracks in spot—welded region; b) trans- verse cracks across spot-welded region; c) Laser spot welded region showing surface rippling and crack formation; d) magnified view of region in (c) showing recrystallized polygonal domains; and e) intergranular crack along crystallized zone ......................................... XRD results from as-received and isothermally annealed Met818826OSSC OO...CO....0..........OOOOOOOOOOOOOOOOOO.... XRD results of various isothermal annealing treatments on the metallic glass showingci-Fe and Fe3B peaks ............ A comparison of x-ray diffraction intensity profile from an as—received sample and a laser annealed foil ........... A comparison of x-ray diffraction intensity profile from an as-received sample and a laser annealed foil ........... a) SEM picture showing crack formation; b) magnified view of selected region from (a) showing concentric bands ..... SEM picture showing tearing steps .............. . ........ .. a) SEM micrograph showing tearing steps and a crack—tip shear band zone induced by laser irradiation in Metglas 26058C when covered with a transparent overlay; b) magnified view of the crack-tip shear band zone seen in (a) ......... viii Page 65 65 67 76 79 79 81 81 83 84 86 86 88 88 93 94 94 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. SEM micrograph showing wavy deformation bands ............. Coarse (“BIJm) shear bands seen on the fracture surface near the h01e 00.0.0.0......000......0.000000000000000..... Finer («0.0511m) shear band spacing observed on the crack surface away from the laser irradiated zone ............... Shear bands in Metglas 26OSSC produced by mechanical tearing at ambient temperature O...0.0.0.0.........COCOOOOOOO...... Crack-tip shear band zone induced by laser damage in the metallic glass. Angle of tilt 30o ........................ Stress-strain response curves of the metallic glass as a function Of temperature OO0.0.0.0000.........OOOOOOOCOOO... Tensile fracture surface showing characteristic vein pattern Response cueve in simple shear for a power-law material ... Cauchy shear stress-strain curve assumed by Knowles and Sternburg for a special class of materials ................ Crack-tip shear band zone predicted corresponding to the response curve shown in Figure 42 ......................... SEM picture showing evidence of normal sueface displacement in the laser irradiated metallic glass specimen ........... ix Page 96 97 97 98 98 100 100 104 104 106 106 I INTRODUCTION In practical applications involving metallic glasses it may become necessary to join two foils by using a welding technique. Unfortunately at elevated temperatures (T :Tg, the glass transition temperature) the physical and mechanical properties of metallic glasses are degraded (1-9). The mechanical properties of metallic glasses are extremely sensitive to the degree of crystallinity. The onset of crystallization causes a number of structural changes in metallic glasses due to the formation of various stable and metastable phases. It is well known (4,10,11,12,13,14) that the crystallization mechanism in metallic glasses is of nucleation and growth type. In certain metallic glasses, the crystallization process is very sensitive to the precise initial composition of the material, temperature, and the heating rate (15,16). In conventional welding techniques, such as resistance spot-welding for example, the heat affected zone (HAZ) is relatively large, and the heating cycle lasts for several seconds. Morris (14) has shown that a heating cycle as short as half-second, can cause crystallization in Metglas 2826. Thus, a successful spot-welding of such a material may only be possible when the heating and cooling cycles are much shorter. A pulse laser might be a suitable candidate for such a purpose for the following reasons: i) the heat affected zone would be quite narrow, ii) energy input to the welding region could be precisely controlled, and iii) cooling rate of the molten pool could be comparable to or greater 1 2 than the cooling rate in melt spinning and thus the amorphous structure could be preserved. Interaction of a high energy laser pulse with a solid produces two nearly simultaneous effects: (i) a temperature spike within the irradiated volume and (ii) an elastic/plastic deformation front which propagates through the solid. The nature and magnitude of these two events depend on the thermophysical properties of the solid, input energy, spot size and energy-time profile of the laser pulse (17). Highly localized thermal expansion, in the wake of a rapid heating pulse at the surface of an elastic solid, promotes a stress wave which propagates into the interior of the solid. The magnitude of the peak stress is a strong function of the pulse duration, and for a pulse lasting a few nanoseconds, fracture and spallation are possible even in a very ductile metal (18). Thus, a ruby laser with normal pulse operation was preferred over a Q—switched operation. Results of such a pulse laser interaction with Metglas 26OSSC (Fe80B14.5 Si3.5 C2) are reported in this investigation. II BACKGROUND 2.1 Nature of Glassy State The word "glass", in its original meaning refers, to an amorphous or a non-crystalline solid formed by continuous cooling of a liquid. A solid is defined as any body having a viscosity greater than 1014 P. A glass, like the liquid from which it is formed, lacks three dimensional atomic periodicity, however, has compositional homogeneity. Lack of three dimensional long range order can be verified easily by the presence of a limited number of diffuse halos in x-ray, electron, and neutron diffraction patterns. A x-ray diffractometer generally exhibits broad peaks centered in the range in which strong peaks are also seen in the diffraction patterns of the corresponding crystalline phase(s). High resolution electron microscopy reveals no diffraction contrast and, in particular, reveals none of the contrast effects normally associated with grain and lattice defect structures. Such effects are characteristic of amorphous solids indicating that any coherently scattered diffraction domains present are smaller than about 20 A in diameter (19). The expression "glass" is sometimes also used to describe amorphous solids produced by electrodeposition, vapor condensation, sputtering, and ion implantation based on the similarity in diffraction patterns and crystallization behavior between amorphous solids formed by atomic 3 4 condensations and ones formed by liquid solidification (19). Noncrystal- line solids produced by different methods are often indistinguishable in x-ray measurements, but their physical properties may differ drastically in as—prepared state, Ni—P glasses produced by liquid quenching method are ductile whereas electrodeposited amorphous Ni-P alloys are very brittle (20). In this dissertation, the term "metallic glass" or "glassy metal" is used to describe non-crystalline solids formed by continuous cooling of a melt to distinguish it from amorphous solids formed by other methods which are referred to as amorphous metals for the sake of clarity. The term "metallic glass", however, is somewhat misleading because the majority of the glassy metals synthesized are in reality alloys containing metallic as well as nonmetallic elements. Their physical properties though are typically metallic. During the solidification of a melt to form a glass, no essential change in the spatial atomic configuration occurs. Hence, a liquid and a glass can be considered to belong to the same phase structurally and thermodynamically. In other words, a glass may be considered as a solid with frozen-in liquid structure. The structure of a glass can be clearly distinguished from that of a liquid, because glass structure is independent of temperature but depends on the means of preparation and heat treatment. Whereas for a liquid, both structure and properties are independent of thermal history. A glass, when given sufficient time and atomic mobility, relaxes structurally towards a more stable equilibrium state, hence it is considered to be metastable with respect to crystal- line phase and transforms to stable configuration through the process of nucleation and growth. In the following sections important attributes of the glassy state, the glass—to—liquid transition, structural V .4 5 15 /(52 /, 9:1K/S ’ 6 I’q31 T ” 0Kl§ O a) 10 r— X" T92 ’I’ a .9 X l ,r m a x : 33 v ’1 "’1’ p :1 ,’ f -1-5 0) g 5 +- (' “I .9 __ , , 1 ITgr 1 I l I I I :L’I’ ~40 O "/:I I 1 1 l 1 2 3 Tm/ T Figure 1: Temperature dependence of viscosities and time constant for structural relaxation in a Pd—Cu—Si alloy (21). relaxation, and crystallization behaviour are described. 2.1.1 Liquid—Glass Transition The temperature dependence of viscosities of a Pd-Cu—Si glass forming alloy (21) is shown schematically in Figure 1 to illustrate the liquid—solid transformation. In this figure, the solid lines represent the experimental data. The logarithm of viscosityrwand structural relaxation time T are plotted against the reciprocal of reduced temperature Tm /T, where Tm is the thermodynamic equilibrium melting temperature. T is the average time required for atomic rearrangement and is related approximately to viscosity n and shear modulus U by: 6 r = n/u 2.1 withlJ= 1011 dynes cm‘2 for amorphous alloys. The viscosities of molten metals above Tm are low (10’2 P) and increase slowly on cooling with a relatively low activation energy 0 2 3 kTm. 0n super cooling a liquid below Tm, it either crystallizes or forms a glass. 0n crystallization, n increases discontinuously by many orders of magnitudes and then follows a path X with activation energy Q = 3eV. If crystallization is bypassed, n increases continuously with decreasing temperature following the path L. The rate of increase in n and the time constant T of the liquid increase with increasing under- cooling. Below the melting temperature, the viscosity of supercooled liquids can be described empirically by Vogel-Fulcher expression (21-23) n = no exp [ B/ (T-T.)] 2.2 where n and B are constants depending upon materials and To is an ideal glass transition temperature. This form of temperature dependence of n has also been observed to be true in many oxide glasses and other nonmetallic glassy materials near the glass transition. For liquid metals and organic liquids, B is small and represents a small fraction of the chemical bonding energy, while TLrepresents a large fraction (1/3 to 2/3) of T . In network forming liquids, values of B are large and are of the order of chemical bonding energy, and To is low. In pure silica or germania, the viscosities are well described by an Arrhenius equation (24). n of the oxides is high ( 2107 P) even above Tm and increases rapidly with decreasing temperature. At sufficient undercooling, the time constant of the liquid,T , becomes comparable to or greater than the duration of experimental measurement. Below this temperature, the atomic configuration remains virtually unchanged, thus giving rise to the discontinuity in the n against T relationship and is observed to follow an Arrhenius expression (25) r] a exp (Q/kT) 2.3 The temperatures, T 81 and T 82’ which correspond respectively to fast and slow cooling and mark transition from equilibrium liquid state to frozen-in non-equilibrium structure of iso-configurational structure, are known as glass transition temperatures T8. At T8, the structural relaxation time Tg is related to the rate of cooling q E -dT/dt by (26) Ig = (kTg/Qaq) 2.4 where 08 is the apparent activation energy for structural relaxation of the supercooled liquid, T8 a exp (Q/kTg) and k is Boltzmann's constant. For most metallic glasses, 0a 2 4 eV, T8: 700 K and hence (kTg,/Qa) 2 2-10 K. At different cooling rates, the glass freezes in a different state e.g. Cl and CZ. The glass has a lower viscosity and greater specific volume and internal energy when produced at a faster cooling rate. Figure 2 shows the corresponding specific volume of a Pd-Cu-Si glass (27). At the fast cooling rate of 106 K/s , usually experienced during melt spinning of metallic glasses, the glass freezes at Té, at which T8 = 10“5 s , into a glass denoted by GI: At slower cooling rates 2 1 K/s , the glass freezes at low temperatures T82 and the resultant glass 02 has higher viscosities below T8 by at least six orders of magnitude than does the glass Cl produced at higher cooling rates. If the crystallization of a glass is sluggish enough in a temperature range just above T8, then a reversible transition from a glass to the undercooled state can be observed experimentally (28,29). ,’ : 9.0 - i a : B P ,1 E P "’ In, E E: _ > 85 - 1 1 J l 1 1 1 l l l 1 500 1000 T(K) Figure 2: The specific volume V of a Pd—Cu—Si alloy (27). In many other glassy metals, however, the glass-liquid transition is unobservable at slow rates of measurements due to the intervention of rapid crystallization process. 2.1.2 Structural Relaxation Classes, in as-prepared state, are thermodynamically unstable with respect to the fully relaxed glassy state and tend to relax structurally to the latter stage at a rate which depends on the previous thermal history and temperature. The glasses produced at higher cooling rates which possess a greater frozen-in structural disorder and thus high diffusivity, would relax structurally at relatively lower temperature compared to glasses produced at slower cooling rates. Evidence for structural relaxation upon annealing (30,31) at temperatures below the crystallization temperature and at times insufficient to cause 9 detectable crystallization was observed by such phenomena as internal friction (32,33), stress relief(34), stress relaxation and creep (25, 35), recovery from cold work (36), specific heat (37), and magnetic annealing (38). A direct indication of structural changes was obtained by x-ray diffraction studies by Waseda and Masumoto (39). Upon heating a glass undergoes structural relaxation in the temperature range for which the time constant for structural relaxation of the glass is given by T8 3 tmea = Aq’1 2.5 where A = kTZ/Qg : 10K. Q is the activation energy for the structural 8 relaxation of the glass and is usually low (~l-2 eV). At slow heating rates of :1 K/s , the glass G1 would have value of T8 = 10 s and would undergo structural relaxation at temperatures far below Tgl. The glass Gz , which is produced at slower cooling rates or preannealed near TS' would have T8 > tmea z 10 s and would not show structural relaxation below T82. Both G1 and G2 , however, undergo glass—liquid transition at temperature T8, upon slow heating. If the glass G1 is heated at rates higher than the quenching rate of 106 K/s , e.g. during laser heating, the glass would exhibit negligible structural relaxation and transform to liquid at or above T81° As shown in Figure 2, the glass Gl, produced at a faster rate of cooling, has a specific volume greater by “’0.5% than does the slowly cooled glass 02 . At extremely slow cooling rates, volume contraction follows path d. In practice, length of path d is limited by the increasingly slow rates of cooling that are required. Further extra— polation of the glassy volume, V, would intersect the crystalline value «200 K below T8, corresponding approximately to the ideal glass 10 transition temperature To. Upon heating the quenched glass G1, at rates slower than the rates of quenching, e.g. 1 K/s , the glass at just above room temperature exhibits an irreversible contraction that is caused by structural relaxation. The initial contraction is small and becomes more pronounced at higher temperatures. At the glass transition, the glass transforms to a liquid, and coefficient of thermal expansion increases by a factor of two. Upon further heating, the glass crystallizes and its volume contracts by 'V l—ZZ. The volume contraction associated with structural relaxation is about one third the volume change that accompanies crystallization. This fraction ( N1/3) is also the ratio of heat evolution and elastic constants associated with structural relaxation to those of crystallization (27,37). Many physical and mechanical properties such as mechanical ductility (31,40), microhardness (41), electrical resistivity (41), atomic diffusivity (42) and magnetic anisotropy (40) alter significantly upon structural relaxation. This was attributed to local structural and compositional fluctuations at temperatures well below crystallizatiOn temperature. The change in density associated with metallic glasses is small, 'V0.5%; however, change in other physical properties is appreci- able, e.g. Young's modulus by N7Z (27,32,33,43,), internal energy by ”200 cal/mol (37), Curie temperature by as much as 35 K (44), and atomic diffusivity by many orders of magnitude (42). 2.1.3 Kinetics of Structural Relaxation The rate of isothermal isobaric volume change can be written as d(V-Ve)/dt=(V-Ve)/Tv 2.6 11 where t is the experimental time, V and V8 are actual and equilibrium specific volumes of the glass respectively and TV is the volume retard- ation time. The form of equation (2.6) is applicable to the rate of change in various physical properties such as viscosity, magnetic anisotropy and Curie temperature. The temperature dependence of the relaxation time for various properties is similar. The rate of stabil— ization depends on the experimental procedure because the relaxation time is determined not only by temperature of measurement but also by overall free volume. The rate of structural relaxation is greater for glasses produced at higher rates of quenching and for glasses with lower Tg (45,46). Isothermal structural relaxation of metallic glasses near Tg has been studied from the change in x-ray intensity function (30), from enthalpy recovery, and Curie temperature aging (47). The rate of equilibrium obeys approximately logarithmic time law. This relationship is different from the exponential time dependence frequently observed in magnetic aging (48,49) in which the relaxation kinetics are described by the first-order rate reaction with a single relaxation time. In stress relief and magnetic aging experiments at temperatures below T3, the kinetics of annealing is given by equation (2.6) with a single thermally activated process having Q N 1 eV. The stress relief of as-quenched glasses exhibits two processes, the initial short-time relaxation and the final long—time relaxation (50). The rates of stress relief are reduced drastically, but the apparent activation energies (Qre W 1 eV) are not affected much in glasses produced at slower cooling rates. 12 2.1.4 The Classy—to—Crystalline Transformation If an amorphous solid is heated to high enough temperature or annealed for a longer time, it will transform to crystalline state. The process of crystallization involves nucleation and growth of the crystalline phase. Liquid-to-crystalline transformation is an extremely fast phase change and takes place at well defined temperatures, whereas glass-to-crystalline trasformation is a rate process. The rates of glass-crystal transformation are usually dominated by frequency of nucleation and hence depend not only on atomic diffusivities but also strongly on glass-crystal interfacial energy and entropy of fusion. In other words, the transformation kinetics depends not only upon the transport properties, but also on the relative stability of the crystal- line phase. The ease of glass formation and the stability of alloy glasses thus, in general, occur in parallel. Empirically, glasses with eutectic compositions are most stable, and ternary alloy glasses that form readily at slow cooling rate are more stable than the corresponding binary alloy glasses (29). The kinetics of transformation are analyzed using the generalized theory of phase transformation (51,52), x = 1 — exp(-kntn) where x is the volume fraction transformed in time t, kn is a kinetic constant, and the exponent n is determined by transformation mode. n ranges from 2.5 to 4 for Pd-Si glasses (53—55), whereas values from 3 to 5 are obtained in Fe-based glasses (56-58). In a Pd-Ag—Si alloy value of n reported is equal to unity (59). The activation energy, Qx , for crystallization varies widely with composition and temperature of transformation. In several stable glasses such as Pd-Cu—Si (53), Pd—Ni-P (60), Fe-Ni—P-B (61), Zr-Cu (31) which crystallize above T8, Qx is high (> 4 eV) and l3 comparable to the On for the viscous flow in an equilibrium glass. For most metallic glasses which crystallize near or below Tg, Qx ranges from 4 eV to 2 eV (1,9,61,62). Most monoatomic amorphous metals which are thermally least stable exhibit a very low thél eV and crystallize at about 20 to 50 K, which is just above the vapor deposition temperature (63). Pd-Si (1,64) and Cu—Zr (4,65) alloys undergo crystallization during annealing below Tg, whereas Fe—Ni-based alloys crystallize near Tg . The occurrence of crystallization in many metallic glasses far below TS' apparently reflects the high nucleation frequency and high diffusivities in the unrelaxed glassy state so that crystallization proceeds concurrently with the structural relaxation. The crystallization process is marked by an exothermic peak in the differential scanning calorimeter (DSC) trace, by abrupt changes in mechanical and magnetic properties, and by presence of sharp peaks in a diffraction pattern. Crystallization behaviour of amorphous ferro- magnetic glassy alloys is studied by Mossbauer spectroscopy. Measuring electrical resistivity as a function of temperature is a simple technique used extensively to determine the approximate temperature at which crystallization proceeds rapidly. The onset of crystallization is marked by one or several rapid drops in resistivity—versus—temperature plot. Appearance of multiple steps in the plot indicates formation of metastable crystalline phases which later are replaced by more stable phases at higher temperatures. Various crystallization morphologies are observed to develop depending on the annealing temperature and composition. In vapor deposited amorphous Ni—P films, the mode of crystallization was found to be dependent on phosphorus content (66). The facet and dendritic l4 morphologies are expected modes of crystallization. For a glass with a continuous random type of structure, whereas a microcrystalline body might be expected to crystallize by grain growth process. However, an extremely fine-grained crystallite structure might arise from copious homogeneous nucleation of the crystal followed by crystallite coarsening. In FeaoNiAOPlaBé crystallization was observed to take place by nucleation and growth of crystal having a barrel shape (11,14,40,67), whereas for FeSONi3OBZO and F880B2O alloys, crystallites exhibited a characteristic cylindrical shape with rounded ends and flattened sides (68). Yeh and Maddin (56) have reported often appearance of different crystallization morphologies in the same sample. Shimomura et al. (10) have reported marked changes in the morphology of crystalline phases as a result of a small change in the composition of the amorphous Fe—based alloys. Crystallization processes of metallic glasses are quite complex. Crystallization behaviour of various metallic glasses has been reported in the literature (4,9-15,40,65,67—9l). Thus, upon heating, a glass exhibits three characteristics: structural relaxation, a glass-liquid transition, and crystallization. Structural relaxation occurs when the glass attains appreciable mobilities such that Tg(T, q, etc.) 3 t , while the glass transition mea takes place when the time constant of the relaxed glass T1(T) approaches tmea“ q‘l. The observed glass transition temperature T8 is thus determined by the heating rate during measurement and is less susceptible to previous heat treatments. In contrast, the rates of structural relaxation depend critically on previous history. Glasses produced at high quenching rates or subjected to cold—rolling or irradiation processes exhibit highly disordered structure and high 15 atomic mobilities, and thus may undergo structural relaxation far below Tg (50,92). Annealing drastically reduces atomic mobilities and the rate of structural relaxation. Since the kinetics of crystallization processes are governed not only by atomic diffusivities but also by thermodynamic parameters, there is no unique correlation between the stability of a glass as indicated by the crystallization temperature Tx and the glass transition temperature Tg. In general, however, more easily formed glasses are more stable. 2.2 Glass Formation Structural and kinetic criteria have been proposed to describe the process of glass formation. Rawson (93) and Cahn (94) have reviewed many of the structural theories. These theories take into consideration the atomic structure, atomic size effects, and bonding to predict glass formation tendency. The kinetic criterion, developed mainly by Turnbull and Cohen (95) and by Turnbull (96), considers rate of cooling relative to the kinetics of crystallization. This theory assumes that any material may be rendered glassy if quenching rates are high enough to avoid crystallization. The two criteria are complementary. The structural criteria based on chemical bonding, coordination or chemical structure merely imply indirectly that the free energy of particularly stable glassy structure will be low enough to make the driving force for crystallization negligible. The kinetic criteria which have been successfully employed to predict ease of glass formation are described next. 16 2.2.1 Kinetic Criteria Adapting simple nucleation theory and assuming the avoidance of a single nucleus as the criteria for glass formation, Turnbull (96) estimated quantitatively the condition for glass formation. The homogeneous nucleation frequency I for normal metals is expressed by I(cm-3S“1) = -—%239 exp(-ba3B/TpATr2) 2.7 where n is viscosity, b is the shape factor (e.g. for a spherical nucleus b=l6n/3), the reduced temperature Tr=T/Th and ATr=l-Tr with Tm as the melting temperature. Here a and B are dimensionless parameters related to the liquid-crystal interfacial tension 0 and the entropy of fusion ASf and are given by a = (NOV)1/3o/AHf and B = ASf/R, where No is Avogadro's number, V is the molar volume in the crystal, AHf is the molar heat of fusion, and R is the gas constant. For constant n=lO-'2 P, the nucleation frequency I increases sharply with ATr from zero to a broad peaked maximum at Tr = 1/3. The peak nucleation frequency Imax 2 1032 exp(-ll3a38) depends strongly on thermodynamic parameters and decreases sharply from 1031 to 10'4 l/cm38 as aBl/B increases from 0.25 to 0.9. Turnbull concluded that liquids with a81/3 = 0.9 should readily form glasses, while for liquids with 061/3 < 0.25, it should be impossible to suppress crystallization. Glass formation is thus enhanced greatly by a large interfacial energy and entropy of fusion. 8 = l for metals, Si02, 0e02, organic plastic materials, and many strong acids and bases, while for typical organic and inorganic compounds, ranges from 4 to 10. Droplet nucleation experiments indicate that for typical metals a = 0.4 to 0.5, and for most non metals 0 = 0.33. The progressive increase in B improves the glass-forming tendency. The network liquids such as Si02, B203, P205, Ge02 etc. form glasses easily 17 because of their extremely high viscosity at Tm. Uhlmann (97) and Davies (98) have carried out a qualitative estimation of the critical quenching rate for glass formation. For this they have considered the theory of nucleation and growth and the Johnson-Mehl treatment of transformation kinetics. In this approach time—temperature-transformation (TTT) curves corresponding to a small volume fraction of crystalline phase are constructed, and then these curved are used to estimate minimum quenching rates, QCr , for the formation of various glasses. The reduced glass transition temperature, Tro = To/Tm, is the most important factor determining the glass forming abilities for metals, where To is an ideal glass transition temperature. Increase in Tro drastically lengthens the time at the nose of the TTT curves, sharpens and shifts the nose to smaller undercooling as shown in Figure 3. With an assuption that viscosity is dependent on temperature, qcr is observed to vary from 2109 K/s to 102 K/s as Tr0 increases from 0.3 to 0.6. These values agree reasonably with the values estimated experimentally e.g. 1010 K/s for pure metals Ni and Pd; 106 K/s for glass forming alloys Au-Si, Ni-P, Pd-P; and 102 K/s for easy glass formers Pd-Cu-Si, Pd-Ni-P. For the above systems, T = 1/4, 1/2 and 2/3 respectively. re The glass transition temperature Tg depends weakly on composition. The increase in Tro and hence the drastic enhancement of glass-forming tendency in these alloy systems is due to lowering of the liquidus temperature. Near eutectic compositions have lowest liquidus tempera— ture, hence form glasses quite easily as evidenced for most metallic glasses and ionic glasses in which bonding character is not altered drastically. The estimation of qcr by the construction of TTT curves 18 04“ _ X2105 02 9cr=AIN/'N 0 1 1 1 1 .1 __1 -3 -5 -4 -2 2 4 5 log 1 (sec) Figure 3: Time-temperature—transformation (TTT) curves of metals. implicitly assumes that the crystallization kinetics over the full range of temperature is as rapid as at the temperature of the nose and thus overestimates qcr by about a factor of three, estimated from the continuous cooling curves (99). 2.2.2 Methods of Preparation Glassy metals have been known for many years. As early as 1845, Wurtz has reported a method of obtaining a nickel deposit on iron by chemical deposition of a Ni solution with hypophosphide. From the details given in that paper, it is most likely that the metallic deposit was amorphous, but it could not have been confirmed until the discovery 19 of x-rays 65 years later. A number of amorphous alloys were prepared by chemical deposition method in early 19503, however, emphasis was given to the mechanical and chemical properties. In 1960 Klement et al. (100) reported the production of Au—Si glassy metal by quenching from a melt. Since then the liquid method has been extensively developed and has become the preferred method because it is faster, applicable to a wider range of composition, probably gives material of more uniformity and few unknown contaminants, and is adaptable to large scale production. The field of glassy metals is expanding so rapidly that it is difficult to describe all the techniques in detail that are being used. 2.2.2.1 Electrochemical Methods These methods can be considered as standard electrolytic or electroless techniques for the following reasons: (i) the metallic deposit is glassy and resides in the chemical composition of the bath, (ii) its pH, and (iii) its temperature. When a potential is applied to the cell, the current density is also important. The number of glassy metals obtainable by electrolysis is very limited. The truly amorphous structure of these alloys has been established for transition metals Ni (101,102), Co (103), and Fe (104) with P. 2.2.2.2 Rapid Quenching Techniques With the exception of easy glass forming alloys such as Pd-Cu-Si, Pd-Ni-P, and Pt-Ni-P, most metallic glasses are produced at a relatively high cooling rate of 105 K/s or higher. The high quenching rates are accomplished by spreading a thin layer of liquid in good contact with a 1'3‘ Hm- w. .11 20 highly conductive substrate, e.g. metals or sapphire. The quenching rate is determined by the rate of heat transfer at the liquid—substrate interface and the thickness and thermal conductivity of the liquid layer. Usually the thickness of the quenched metallic glass in which crystallization is prevented, is limited to «50 um. Amorphous solids can also be formed by methods in which the liquid state is bypassed completely. These processes, known as atomic deposition or atomic condensation, involve growth from the vapor phase by thermal evapor- ation, sputtering or from decomposition of gaseous compounds by radio frequency discharges. These techniques provide a very high effective quenching rate and yield amorphous solids which cannot be obtained as a glass by liquid quenching. 2.2.2.2A Vacuum Evaporation and Sputtering Methods Vacuum deposition offers a method of very fast quenching rate from vapor to solid phase. The atoms in the vapor phase strike the substrate more or less independently of their neighbors, but once on the substrate the diffusion paths are very short (of the order of atomic distances), and as soon as stable nuclei are formed, the growth of a crystal can be very rapid. Hence amorphous metals or alloys have been obtained (including superconducting amorphous films of Bi, Ga, As, Sb, and Be), only on substrates maintained at liquid-helium temperature (4 K). Many amorphous films are very unstable and crystallize with a temperature range of 20-50 K. Thus it is necessary to study the structure and properties of such films within the evaporation chamber. In spite of this difficulty, vacuum evaporation is the only method by which pure metals, particularly high melting point transition metals can be e 21 obtained in the amorphous state. It has been suggested that the unavoidable presence of traces of impurities may stabilize an amorphous structure. The crystallization temperature of amorphous Co and Fe films increased with an increase in residual gas pressure, as reported by Ichikawa (105), strongly support the view. Many alloys of metals or metals and metalloids that have not been prepared in glassy phase by liquid quenching can also be prepared in amorphous form by vapor deposition or by sputtering. Binary alloys of metals such as Cu-Ag and Co-Au (106), Au-Ni (107), Pb-Au (108), Fe-Au (109), and Fe-Si (110) were produced in the form of amorphous films by deposition on targets at 80 K and are stable at room temperature. Solid-state immiscibility and a large (> 10%) difference in atomic radii of the constituents has been proposed as the criteria for amorphous phase formation by Madder (111). In liquid quenching, the glass forming compositions generally lie near a low melting eutectic. As different processes are involved in vacuum evaporation and liquid quenching, the different criteria for glass formation in each case are justified (112). In sputtering method, atoms of the source material are bombarded by high energy inert gas atoms. Atoms are removed from the source by collision with the inert gas ions instead of thermal evaporation. Amorphous alloys can be obtained by sputtering on a substrate at room temperature and these alloys are more stable than vapor deposited ones. 2.2.2.2B Liquid Quenching The principle behind liquid quenching technique is to retain in the solid state the atomic arrangement present in liquid state at the MYLAR rat DIAPHRAM—c— Q VALVE :0 RF COM. Q/‘A‘ CU BLOCK Figure 4: Single foil quenching gun technique (100). E51 PIS TONS VACUUM ENVELOPE ———o RF(XML A fit ”PHOTOCELL Figure 5: Piston and anvil technique (113). 23 quenching temperature. If the liquid alloy could be cooled fast enough, so as to avoid the normal process of crystallization, then the liquid structure can be said to be "frozen—in". This requires very rapid rates of heat transfer to the cooling medium and are achieved by conduction from the liquid to a highly conductive metallic substrate. In addition, the condition that the layer of liquid alloy in contact with the substrate must not exceed a certain thickness must be satisfied. This is necessary since the rate of cooling at a point away from the substrate decreases with the distance from the substrate. Using these conditions as criteria, Klement, Willens and Duwez (100) synthesized the first metallic glass Au.,SSizs from a liquid by the now classical gun technique (also splat cooling) (Figure 4). In this process, a small liquid globule was propelled via a gaseous shock wave onto a copper substrate. The substrate was curved so that centrifugal force promoted good thermal contact between the liquid layer and the substrate. Quenched samples produced were irregular in shape with thickness varying from about 1 um to 10 um. In the gun technique, heat is extracted from the melt from one side only. Pietrokowsky (113) designed a piston and anvil technique in which the liquid globule was squeezed between a fast moving piston and fixed anvil (Figure 5). Both piston and anvil were lined with copper to insure fast heat removal. The rate of cooling obtained by the gun technique is often mentioned as being about 106 K/sec. The rate of cooling mentioned here is only a rough average, since the actual rate varies not only from point to point but also with time during quenching. The average rates of cooling in gun technique differ by one order of magnitude. The piston and anvil method gave geometrically better samples, e.g. 2 cm in diameter and 20-501Jm in thickness. 24 Methods of making long ribbons since first reported, have been extensively developed. The ribbon-making methods are basically similar; a continuous jet of liquid metal is directed on a moving metal surface, or between two metal rolls, or sometimes into a moving liquid stream. The techniques are called by various names: melt-spinning, liquid quenching, chill-block spinning, roller quenching etc. The first continuous casting process was reported by Chen and Miller (114) and investigated extensively by Babic et al. (115). In this two-roller or roll-quenching process, a molten metal stream was directed into a gap between a pair of rapidly rotating rollers lined with copper. By proper adjustment of mass flow of the liquid alloy, temperature, the peripheral velocity of the rolls, and the size of the gap, the resultant product obtained was a metallic glass ribbon about 2 II in width and 50 pm in thickness. Other commonly used techniques are melt-spinning processes of Strange and Pim (116) (Figure 6b), and Pond and Maddin (117) (Figure 6c) in which the molten stream is cast respect— ively onto the outside or inside of a rotating drum. These techniques were developed originally for fabricating crystalline filaments and later adopted for continuous production of metallic glass ribbons (1, 118). In comparison with Strange and Pim process, the Pond and Maddin process yields a higher quenching rate because centrifugal force allows better thermal contact between the melt and drum surface. The disadvan— tage in this process is that quenched samples are difficult to remove from the drum. A process developed by Chen and Miller (119) is shown in Figure 6d. By impinging a stream of melt onto the inner sloped surface of a rotating drum, good heat transfer between the melt and substrate was ensured and continuous fabrication of glassy filaments was achieved. 25 Figure 6: Continuous casting processes: a) twin—roller quenching (114, 115); b) melt-spinning (outside) (116); c) rotating crucible spinning (117); d) melt—spinning (inner) (119); and e) melt- extraction mrthod (120). 26 In a melt-extraction method (120) the periphery of a spinning disc is brought into contact with the surface of a melt or a molten drop at the end of continously fed rod of alloy (Figure 6e). The liquid alloy solidifies upon contact with the disc and separates from it because of the radial acceleration. The roller quenching method has an advantage over melt casting by extracting heat from both sides and in principle yields metallic glasses twice as thick for a given cooling rate and with smooth surfaces of uniform thickness. Other methods of quenching liquid alloys, such as plasma spraying (121), explosive forming (122) and melt atomization (123,124) which yield glass surface layer have not been much explored for fabrication of metallic glasses. Laser surface melting (125) was first successfully employed to demonstrate formation of a glassy Pd-Cu-Si surface layer on a crystalline alloy of glass-forming composition. The quenching rates reported were higher than 1010 K/s. Poate et al. (126) reported production of amorphous layer of Pt—Si on a silicon single crystal substrate with a very short pulse laser interaction. Since then vitrification of various metallic glasses by laser glazing has been reported (127-132). 2.2.3 Solidification Processes and Quenching Rates Basic apparatus designs and processing techniques for manufacturing amorphous alloy ribbons have been developed with consideration for opti— mization of heat transfer and fluid dynamics. In other words, in various continuous casting processes used in the manufacture of rapidly solid— ified alloys, the important objectives are (i) to maximize the average quench rate and (ii) to control the geometrical diamensions and finish 27 of the cast product. The melt jet upon impingment on the circumferential surface of a rapidly rotating wheel, forms a puddle and then departs under the action of radial acceleration. During its period of contact with the substrate, the melt cools rapidly and its viscosity increases. The kinetic energy of the melt jet causes spreading of the liquid on the substrate to an extent determined by melt surface tension and other fluid properties as well as momentum balance on impact. The width and length of the puddle, which are governed by the nozzle size, velocity of the melt jet, and the rapid substrate velocity, determine the width w, and thickness t, of the ribbon. Therefore the ribbon width may be increased by increasing the melt jet diameter and velocity, and by using steep angle of melt jet incidence on the moving substrate surface. Ribbon thickness may be reduced by increasing the substrate surface velocity, reducing the melt jet diameter and velocity, and by using a steep angle of incidence of melt jet on the moving substrate surface. Excessively high melt jet velocity may result in thickened ribbon edges, while extremely high substrate velocities may result in continuously- cast product having serrated edges and surface perforations. The high surface tensions, low viscosities of the metallic melts, and interaction between the gas boundary layer on the moving substrate surface and the melt puddle, cause significant puddle turbulance and subsequent ribbon edge serrations and surface perforations. The turbulance in the puddle may greatly be smoothened by impinging the melt jet with a significant component of motion along the direction of substrate motion or by quenching the melt in a He atmosphere or in a crude vacuum (133). The ribbon width can be increased to strip or sheet dimensions by using a long slit nozzle or by employing a series of 28 uniformly spaced nozzles. In each case it is necessary to maintain a stable, elongated melt puddle in contact with rapidly moving substrate surface, such that merging of the individual melt puddles yields the desired elongated composite puddle. The use of multiple nozzles has yielded a ribbon as wide as 10 mm (133, 134). A common characteristic of glassy alloy ribbons manufactured by this method is the appearance of a line texture along the length of the ribbon. The initial quench rate of a twin-roller quenching method is greater than in a single roller method; however, contact time or distance over which the ribbon touches the substrate is extremely short, thereby rendering no significant improvement in overall quench rate. Although the twin—roller method produces ribbons which are flat and have parallel surfaces, undesirable cold work of the ribbon during casting as well as in—plane curvature is found to occur. The rates attained in any quenching process are determined by the thickness of the product and the mode of heat transfer between the melt and the quenching medium. The quenching rate achieved varies with t—n with n=1 for Newtonian and n=2 for ideal quenching. Ideal cooling is achieved when the Nusselt number, Nu (=hd/k), is greater than 30, while Nu < 1, Newtonian cooling is obtained. Here h is the heat transfer coefficient, and k is the thermal conductivity of the melt. Figure 7 shows predicted cooling rate q as a function of t and h (135, 136). Comparison of experimental data with calculated values of Figure 7 reveals that (i) in case of splat quenched foils ( < l m thick) the mode of cooling is Newtonian and rates of cooling are very sensitive to the thermal contact with substrate; (ii) for melt-spun ribbons (‘QOlJm thick) q is less sensitive to thermal contact when h :10 WmZ/s. 29 lal SPLAT QUENCHmG h:‘07 ‘. lblMELTSPHNMNG mm 106 I P ICIWATEROUENCHING / o . COOLING RAIE q Ks“ THICKNESS 1AM Figure 7: Predicted cooling rate as a function of section thickness t, and heat transfer coefficient h (W/mzsl) (135,136). Non-wetting of the substrate by the melt contamination or oxide formation on the substrate, however, may greatly reduce the rates of quenching and hamper glass formation. The quenching rates attained are m 109 K/s for splat quenching, N106 K/s for melt quenching, and 103 K/s for the water quenching processes. The maximum thickness tmax obtainable for glassy ribbons is determined by the thermal diffusivities of the melt Dtv and the critical quenching rate required for glass formation. For ideal qcr ’ quenching: tmax rt( DtTm/qcr )1/2 where Tm is melting temperature. For most alloy liquids, Dt = 0.2 cmZ/s, Tm = 1000 K, and tmax 2:0.1 mm for qcr = 106 K/s. 30 2.2.4 Glass—Forming Alloys The ability to form glass depends critically on quenching conditions. Attempts to form metallic glass from pure metallic elements have never been materialized. Computer simulations have indicated glass formation by atomic deposition (137) or by liquid quenching at rates of lOlZK/s (138). In practice, however, pure monatomic metal liquids have not been quenched into glasses due to failure to attain such high rates of quenching to avoid crystallization. Formation of a glassy phase is reported in literature (139), in the thinnest portion of a splat quenched Ni foil in oxidizing atmospheres. It is suggested that impurity contamination played a significant role in stabilizing the glassy phase. Glass—forming binary alloys may be divided into three groups; transition metal or noble metal alloys containing about lO—30Z semimetal (P, B, Si, C), alloys of early transition metals (Zr, Nb, Ta, Ti) and late transition metals (Fe, Co, Ni, Cu, Pd), and alloys containing IIA metals (Mg, Ca, Be). Phase diagrams of the selected systems from each group are shown in Figure 8 (140). The first group, commonly known as metal—metalloid systems, are represented by Au758i25, PdBOBZ)’ FeBOBZJ' N180P20 and Pt75PQS, whose glass-forming compositions fall into a narrow deep eutectic region (Figure 8a). The second group referred to as inner-transition systems (Figure 8b,c) forms glasses generally near an early transition metal compositon of 60%, e.g. ZrKfNi, Fe, Co, Pd)33 , Ti65 Ni35 , NbEK)Rh40. In alloys of Zr-Cu, Ti-Cu, Nb—Ni (141), however, glasses can be formed for a broad range of composition. The third group consisting of IIA metals is represented by Mg-Zn, Ca—Mg and Ca—Al or Ca-Zn systems (142). TEMPERATURE (°c) Figure 8: Representative phase diagrams of glass—forming alloy systems (140). 1600 1000 i 1 l Pd Si 2000 + 1000 l Zr Ni 2000 K r%‘ h————-—1 \ ._ h [c 1 \ 0 \ \ F- | 1 | I '-' 1 E “ O"‘P“ 4 1 1 ‘1 Zr (:0 1600 __ + Id] 1000\ " \l 1 1 1 1 U 1000 500 l l l 1 Co Al zooo — ‘\ 1-———-1 \‘ I _ x (1) IL \ I ' s 0 ' \ o ‘. _ \“ ’0' E — a ' o — ’1'! Hf Be 32 These alloy systems are characterized by a relatively high melting intermetallic phase of ABZ type; e.g. Man2 , CaMg2 and CaAl2 , participate in a relatively low melting eutectic near 40% of B element (Figure 8c). In Be—containing alloys, Ti-Be alloys form glasses in a narrow range (37—41% Be), while Zr-Be and Hf—Be alloys exhibit a wide range (30-50% Be and 30-60% Be respectively) (5). In addition to these three types of glass-forming systems, various alloy systems have been identified which form glasses, e.g. U alloys containing 20-40% V and Cr (143), U-30Z Mn, Co, Fe and Ni, and Pt-lO-ZOZ transition metal (144) (Figure 8d), A1-17Z Cu (145), Al-30Z Ge (146), Al-7Z Cr (147), Al—7Z Ni (148), La—24Z Au (149), Gd-M with M Co, Ni, Cu and Pd (150), Y60FeM)and ThSOFeSO(151). The common features present in all glass—forming alloys are (i) a strong interaction between the constituent atoms as indicated by a negative heat of mixing, and (ii) a low lying eutectic (Figure 8). The heats of mixing,,Afin, range generally from -2 to —10 K Cal (g-atom)“1 for the glass-forming alloys mentioned above. The strong atomic interactions are also evident from the formation of stable intermetallic phases which play a role in a relatively low eutectic reaction. The eutectic temperatures T8 are generally about 0.6 of the melting temperature Tm as compared with Te/Tm = 0.8 for eutectic systems. The additional depression of melting points for the glass-forming alloys is because of the strong atomic interaction (152). Most binary metallic glasses, with the exception of Pd-Si, Zr—Cu, and Zr-Be systems, are not stable and crystallize at temperatures below the effective glass transition temperature. These glasses produced only as foils ’b40 um thick, require a critical quenching rate of 105 K/s or 33 higher. The admixture of glass-forming binary alloys or the addition of third elements, having different atomic radii and different inter— metallic compound symmetry, enhances sharply the thermal stability and the glass-forming tendency, mainly by lowering the eutectic temperature. Ternary alloy systems, Pd—Cu—Si, Pd-Ni—P and Pt—Ni-P, have been prepared with consideration to above mentioned criteria. These glasses requiring low quenching rates ( 102 K/s ) exhibit excellent thermal stability and remain glassy at the glass transition temperature for several hours. The eutectic point of ternary alloys is lowered by 50-300 K from that of the constituent binary alloys (29, 153). The increased ease of formation and thermal stability upon alloying is a common phenomenon and has been observed for many metallic glasses. 2.2.5 Factors Affecting Glass Formation The kinetic approach indicates occurrence of a deep eutectic as a likely signal for glass formation. The efforts to correlate eutectic compositions with the liquid structure complement the kinetic approach. The first such attempt by Hume—Rothery and Anderson (154) suggests that at special compositions such as A128, A53 and A3B, eutectic structures might be most stable structurally due to low free energy. Bennett et al. (155) suggested that in transition metal—metalloid systems small and softer metalloid atoms occupy the large voids in the random dense network of large transition metal ions and stabilize the random configuration. Geometrical arrangement based on this proposal yields an alloy composition consisting 80% large metal atoms, a composition frequently observed suitable for most metal-metalloid alloys that have been produced. It was indicated (156), however, that none of the holes 34 were large enough to accomodate the metalloid atoms. In some cases it was observed that the glass—forming composition of the metal-metalloid system departs significantly from the predicted range, and in many cases the softer metalloid atoms are larger than the metal atoms, as in case of Pt65be3(157) and (Au, Ag)25Pb75 (158). Considering the glass as a nearly free electron solid, Nagel and Tauc (159) proposed that at a composition where the Fermi level is at a minimum density of states, or where the Fermi vector 2Kf equals qp , the first peak of structure vector, the glassy phase should be stabilized. Thus an alloy with valence electron concentration (VEC) Z=l.7 should be most stable against crystallization. At such composition in liquid alloys, electrical resistivity has a negative temperature coefficient (NTC). The Nagel and Tauc model was further modified by Guntherodt and Kunzi who suggested that glass formation would be favored for compositions which show NTC in liquid state. This VEC value and NTC criterion compare favorably with some compositions of metal-metalloid alloys, and the wide glass-forming composition range in Zr—Cu and Nb-Si systems, however, fail to justify the large difference in glass-forming tendency among alloys that satisfy the criteria. In many glassy alloys, no minimum density of states is found. For (Mn, Fe, Co, Ni)- Ge alloys the concentration ranges of NTC are centered near the equiatomic composition; however, the observed glass-forming range is near 80% metal—20% metalloid composition. These models are based on the assumption of stabilization of liquid phase and ignore the composition dependence of the stability of crystalline phase. Very little eXperimental evidence, as suggested by Chadwick (160), is available to argue in favor of stabilization of liquid phase. Chen (152) proposed consideration of the destabilization of the crystalline mixture, rather than the stabilization of the glassy phase, as a criteria for predicting useful glass-forming range. Masumoto and Maddin (161) have suggested a correlation between the glass—forming elements and its position in the periodic table. Elements of different valence often participate in strong A—B bonding, which is important in promoting glass formation. Valence difference often coincides with the differences in atomic size. Thus, so far there exists no theory of glass formation in metallic systems that is truly predictive. Appreciable atomic size differences and strong interaction of component atoms are hallmarks of deep eutectic systems and thus of potential glass—forming alloys. 2.3 Structure To characterize the structure of the non—crystalline materials, x—ray, electron and neutron diffraction methods are commonly used. The x-ray diffraction patterns of glassy alloys consist of a series of broad diffraction maxima of decreasing intensity and increasing width with increasing Bragg angle, indicating clearly that the metallic glasses we are discussing are truly amorphous, as opposed to microsrystalline. The structural models for amorphous solids used most frequently are quasi-crystalline model and the Bernal model. Quasi—crystalline models are heterogeneous in which the misoriented crystallites are very small and are separated by regions of less-ordered non—crystalline arrangements. Microcrystalline models for amorphous solids have been proposed for monatomic (162) as well as polyatomic (163) systems. The second model used for describing the structure of metallic glasses is Figure 9: 36 Bernal's idealized holes describing the topology of dense random packing: a) tetrahedron; b) octahedron; c) trigonal prism capped with three half—octahedra; d) Archimedean anti— prism capped with two half-octahedra; and e) tetragonal dodecahedron (164). 37 known as the dense random—packing of hard spheres model (DRPHS) and was first proposed by Bernal (164) who applied it to monatomic liquids. Later this model was modified and applied to amorphous solids (165). The hard sphere model predicts splitting of the second peak in the interference and in the atomic distribution functions that has been observed in large majority of amorphous alloys. This feature is normally prominent for amorphous solids but not for liquids, indicating at least some increase in structural order for the amorphous solid compared with the liquid state. In crystalline state, the metalloid atoms; e.g. P, B, Si and C, have smaller atomic size than the transition metal atoms. Hence the hard sphere model with two sizes of spheres was analyzed (166). Polk (167) suggested that smaller metalloid atoms fit into the "holes" of the five basic Voronoi polyhedra present in this model (Figure 9). Assuming all these holes were occupied by the metalloid atoms, the atomic fraction of the metalloid in one alloy was obtained to be 19.1%, which is remarkably close to the metalloid fraction (20 at %) observed in most metal-metalloid systems. Experimental data and model studies of metallic glasses have been reviewed extensively in the literature (156, 168). Also critical reviews of the structural models are published (169—171). 2.4 Corrosion Behavior Results of corrosion studies on metallic glasses were first reported by Naka et al. (172, 173). It was observed that metallic glasses containing certain amount of Cr and P exhibited extremely high corrosion resistance in acid and neutral solutions such as l M H2804, 6% FeCl3 and l N HCl. Glassy Fe80P13C7 alloy, which is fairly unstable in 38 corrosive environments, exhibits remarkable corrosion resistance on addition of chromium. Fe-Cr—P-C glasses containing more than 8% Cr showed extremely high resistance exceeding 18Cr—8Ni stainless steel, particularly in FeCl solution in which the latter suffers pitting and crevice corrosion. This superior corrosion resistance of the Fe-Cr-P—C glasses has been attributed to the rapid formation of thick, uniform, and highly corrosion resistant passive films. Hashimoto et al. (174) found that the corrosion potential of the glassy alloys was high and showed higher reactivity as compared with the 18Cr-8Ni stainless steel, apparently due to the presence of a large amount of semimetallic elements. This high reactivity of a glassy alloy was considered responsbile for promoting the rapid formation of thick and highly corrosion—resistant passive films enriched in chromium via rapid dissolution of other elements in the alloys. The high reactivity of glassy alloys also was credited for a rapid recovery of rupture sites of the passive films. Thus it was concluded that the chromium enrichment process in forming passive films enables the glassy alloy to exhibit a superior corrosion behaviour at lower Cr content as compared with 18Cr- 8Ni stainless steel. The rate of formation of chromium-enriched passive film in the active state depends on the rate of active dissolution. Consequently the effectiveness of Fe—Cr-based glasses simply reflects whether the individual element accelerates the active dissolution. The alloys containing P and C have the highest corrosion potential and thus are the best corrosion—resistive while the alloys with Si and B are less reactive and exhibit poor corrosion resistance (175). Masumoto and Hashimoto (176) observed that the corrosion rate of 39 the Fe-Cr glassy alloy progressively decreased by addition of silicon, boron, carbon and phosphorus in 0.1 N H2804; the progression was boron, carbon, silicon and phosphorus in 3% NaCl. The addition of chromium without phosphorus to Fe-based glassy alloys (e.g. Fe—lOCr—13P—7C and Fe-lOCr-13P—7Si) was observed to be ineffective in improving their corrosion resistance properties (175, 177). On the other hand, the glassy Fe—Cr alloys with phosphorus revealed a negligible corrosion rate in all cases except the alloy containing silicon. The addition of various metallic elements such as Ni, Mo, W, etc., to amorphous Fe—P—C alloys improved the corrosion resistance of the alloys. The effects of additive metallic elements in improving corrosion resistance were observed to be remarkable when a small amount of chromium was present, m 3% Cr, in the Fe-P-C alloys (178). A small amount of Mo addition, < 3%, to Fe—P—C alloys was more effective than a chromium additive in reducing the corrosion rates (179, 180). Corrosion behaviors of glassy nickel-base alloys were found to be similar to the glassy iron-base alloys (176). 2.5 Magnetic Properties The existence of long—range ferromagnetic ordering in glassy alloys may appear striking; however, ferromagnetism is supposed to arise from nearest neighbor interactions. The very short—range order structure in a glassy state does not differ significantly from that in a correspond- ing crystalline material. Glassy alloys are not structurally and magnet— ically isotropic as evidenced by the presence of macroscopic magnetic anisotropy. Residual strain (181), field annealing (182) or directional ordering during solidfication (183) induces anisotropies which determine 40 the domain structure of the glasses and thus their magnetic behavior. The domain structures in glassy alloys are essentially similar to those observed in crystalline counterparts. Most models developed for crystal- line magnetic materials to describe the static and dynamic properties are shown to be applicable to glassy alloys (184-186). Magnetic properties of various metallic glasses have been studied extensively (34,44,50,181—217). Amorphous alloys based on iron or cobalt are ferromagnetic at room temperature (186) and generally have coercive fields less than 100 mOe in the as-cast state. These low coercivities are the result of the high homogeneity of the amorphous structure which has no grain boundaries or precipitate particles to act as pinning centers for domain walls. The lowest coercive fields and highest permeabilities are attained in alloys with zero magneto— striction. The magnetostriction is small only in alloys of high Co and/or Ni content (182). The Ni—rich members of this series have TC below room temperature (187) and are not of much engineering interest, and the Co-rich alloys are too expensive for use in large equipment. Iron-based metallic glasses have been the focus on intensive research because of their high magnetic inductions, soft magnetic properties and potentially low raw material costs. For practical applications as soft magnetic materials the ribbon form is most useful. Hence, studies of magnetic behaviour of glassy alloys were initially centered on glassy ribbons prepared by continuous melt—spinning (181,182,188). 0f the iron-metalloid binaries, only Fe—B and Fe—P have been successfully prepared in ribbon form. Since Fe—B binary exhibited higher TC (=377OC) and magnetic moment per iron atom, further studies were concentrated on Fe8OBZO glassy alloy. F980320 alloy showed good 41 soft magnetic properties, i.e., high permeability, low coercivity and high room temperature saturation magnetization. In addition, high resistivity and low gage of the glassy alloy made it suitable for use as a core material in power transformers. Table I lists the saturation induction, coercivity, remanence, and permeability of toroidal Feaszo samples, as well as handbook values for commercial alloys (189): 50% Ni- Fe and grain oriented Fe-3.2% Si alloy (189). The properties of the field-annealed toroid of Fegngo glassy ribbons compare favorably with those of the commercial alloys. Although this alloy has BS value lower than grain—oriented Fe—Si alloys for which BS reaches «.20 kG, its a.c. core loss at 60 Hz is one—third that of the crystalline counterpart. Thus further efforts were directed to attempt various compositional modifications of Fenggo glass in order to find other suitable high induction alloys. Table I: Some properties of FeBOBZO glass and comparable commercial materials. FeaOBKDglass toroid Commercial 2—mil alloys As-cast field annealed 50% Ni-Fe Oriented Si—Fe 4nMs (kG) 16.0 16.0 16.0 19.7 HC (Ga) 0.08 0.04 0.1 0.5 Br (kG) 8.15 12.3 14.6 14.4 u (20) 1700.00 4000.00 500.00 1000.00 Umax(=-E—) 102000.00 320000.00 146000.00 29000.00 Loss(W/Eg) .... 0.53 0.77 1.5 60 Hz,13kG Loss(W/kg) 0.33 0.10 0.22 0.26 1 kHz,1kG 42 2.2m .... ,——’C) ' 2.1 17" 03A:8__—A———A—-A‘-—S| lug /Fe \ 971‘0~ 2.0 »——_ V \\ —4 \\ P ~\ V- 1111111 8 1012 141618 20 o 1 °/o 1.9 O N»- AL— 0).... Figure 10: Moment per Fe atom in amorphous FeaoBzo alloys as B is replaced by C, Si, P or Ge (192). The addition of metalloid atoms is expected to alter magnetic moment,uB, and the Curie temperature, TC, of the glassy alloys result— ing from the modification of local atomic environment. In Fe—based glassy alloys, replacing Fe with Co or Ni in general reduces the moment. Mizoguchi et al. (190) found that replacing either Fe or Go with Mn, Cr and Mo drastically reduces both HB and TC. The usual assumption is that the moments of these atoms align antiferromagnetically with the dominant iron moment. The influence of composition on TC has been investigated by Chen et al. (191) for a number of glassy alloys systems. It was reported that the addition of a metalloid which expands Fe—Fe inter— atomic distance raises TC while the increase in Co—Co and Ni—Ni pair distances lowers TC. The effect of replacing B by other metalloids has been systemati— cally investigated by Kazama et al. (192) with the results shown in Figure 10. It is seen that Ge and Si increase the moment per atom, but 43 C and P decrease it. When these results are recalculated in terms of moment per gram, all the substitutions lower the moment, but Si and C have only a small effect. The effect of replacing part of B by C on magnetic properties has been investigated by Hatta et al. (193), Luborsky et al. (194) and Mitera et al. (195). It was found that replacing B by C reduced the thermal stability in an alloy having more than 80 at % Fe, increased the TC slightly, and decreased the slope of the magnetization-temperature curve. The coercivity was always higher for the ternary Fe-B—C alloys than for the binary Fe—B alloys. The coercivity, both as—cast and after stress—relief annealing, increased on replacement of B by C. The optimum composition was determined to be FeBlBl3C6' Thus, although the low—temperature saturation magnetization decreased on replacement of B by C, as expected, the room temperature magnetization exhibited a broad ridge of constant saturation magnetization per gram extending from approximately F880B20 to approximately F983Bllc6' Replacement of B by Si has been investigated by O'Handley et al. (196) and by Luborsky et al. (197). Luborsky et al. (197) observed that TC increased slightly on replacement of boron by silicon. The crystallization temperature increased with increasing silicon and with decreasing iron and boron content. The alloys with silicon were generally easier to prepare in the amorphous state than binary Fe—B alloys. Also, Fe—B—Si alloys exhibited higher saturation magnetization, with minimum H and hence lowest losses between compositions Fe81817512 and Fe82BIZSi6. Hatta et al. (198) and Mitera et al. (195) have also studied the effects of replacing boron by a combination of silicon and carbon. The l, 4 results showed that the stability of the Fe—P-Si-C alloys is greatly improved. Starting from F081B13C6 Mitera et al. (195) replaced carbon by silicon upto Fefiplflej and found a very slight linear drop in Ort (about 1%) and a modest increase (28 K) in TC occuring mainly with the few % silicon added. It was also observed that silicon addition is very effective in reducing coercive force and core loss as well as improving magnetic stability during aging at low temperatures. The composition, optimally balanced between coercive force and saturation magnetization, which gives desirable soft magnetic properties, was identified to be Fe81Bl3Si4 C2 . The results of this study agree nearly with those reported by Luborsky and Walter (199). Luborsky and Walter (199) obtained maximum saturation magnetization for an alloy composition of Fe81813Sizo5C2.5. The crystallization temperature, Curie temperature, saturation magnetization, and density all appeared to be average values of the ternary Fe-B—C and Fe—B-Si properties. Maximum saturation magnetization obtained in various amorphous alloy systems is summarized in Table II. The best properties reported for Fe818138i4C2 composition are oS(R.T.) = 176 emu/g, H = 8 mOe, Br/BIOO = 0.88 and the core loss, C Table II: Maximum Saturation Magnetization at Room Temperature Alloy system As-cast Annealed References 08 4an Us 4an emu/g kG emu/g k6 Fe-B 180 16.7 18 17.0 (200) Fe—B-C 180 16.9 184 17.3 (194,201) Fe—B-Si 170 16.5 183 16.9 (197) Fe-B-Si-C 182 17.0 186 17.3 (199) 45 w/f = 8x10-4 W-sec/kg (at 12.5 kG and 50 Hz). This alloy, however, has maximum value of magnetostriction («J50x10-6) reported for any amorphous transition metal alloy. Thus, small magnetostriction is clearly not an absolute requirement for low HC. Due to its lowest core loss, this alloy has found application in low and high frequency transformers as a core material, as well as in magnetic amplifiers. Investigations on Be additions up to 2-4 at % to Fe—B system have also been reported (202). Such additions can occupy the transition metal sites when Be is introduced in glassy Fe-B alloys. This leads to a drastic reduction of the saturation magnetostriction without reducing the magnetization. Substitution effect studies of Fe by Cr in amorphous Fe—Si-B alloys (203) showed that Cr decreases Curie temperature quite greatly, but it also slightly increases Tx. Cr improves soft magnetic properties, namely it decreases coercive force and core loss, and increases initial magnetic permeability. 2.5.1 Effect of Deformation and Magnetic Annealing A quantity of considerable engineering importance is the power loss per cycle under a.c. excitation. Generally it is found that the losses in as-prepared glassy alloys are somewhat higher than in conventional crystalline permalloys, but after a suitable heat treatment the glassy alloys are superior to the crystalline alloys. If the high-induction (high Fe—content) glassy alloys are compared with conventional crystal- line Si-Fe, the glassy alloys are substantially lower in losses except at values of maximum induction approaching the saturation limit. Thus, the development of the excellent soft magnetic properties of glassy 46 alloy ribbons requires annealing at temperatures too low, or for times too short, to cause crystallization and cooling in a magnetic field. The annealing treatment relieves the stresses resulting from the casting operation, and cooling in the magnetic field induces a uniaxial anisotropy in the ribbon plane. Being amorphous, the ribbons lack the magnetocrystalline anisotropy of crystalline materials. Magnetic annealing of glassy alloys is effective in reducing the stress anisotropy to lower values. It also substantially reduces coercive force, increases hysteresis loop squareness and thereby increases permeability and decreases losses. Magnetic annealing also increases TC of glassy alloys. The property changes during annealing are primarily associated with changes in internal stresses originally present as a result of the quenching and preparation of the ribbon. Annealing in demagnetized state deteriorates magnetic properties. Although the losses in glassy alloys are low, the classical eddy current loss component is only 1-15% of the measured eddy current loss (204,205). (Eddy current losses vary inversely with resistivity and linearly with the square of the thickness). The large "anomalous" eddy current loss is attributed to the relatively coarse domain structure of glassy alloys (184). The domain structure of a glassy alloy ribbon in the stress-relief condition is composed of 180 domains parrallel to the ribbon axis when the applied magnetic field is parallel to sheet thickness. The ratio of domain wall spacing to sheet thickness is observed to be 20 to 40 (206). This larger ratio than that found in crystalline metals contributes to high anomalous losses in the ferro- magnetic amorphous alloys. Presence of domain walls, inferred from the large Barkhausen discontinuities observed in d.c. hysteresis loops, can 47 be observed by the Bitter technique using colloidal magnetic particles (207), by magneto-optical Kerr effect (208), and by scanning electron microscopy (183). O'Handly et al. (209) observed a reduction in core loss by annealing in a crossed magnetic field (one field parallel to the ribbon axis and the other perpendicular to the plane of the sample). Similarly, Luborsky et al. (34) annealed in either a parallel or perpendicular field to control the magnitude of the anisotroy (in both cases the field was in the plane of the sample). The resultant losses were found to be sensitive to the magnitude and the direction of the induced anisotropy. Fujimori et al. (210) observed that magnetic annealing in the plane of the strip but oblique to the major strip axis caused a decrease in losses within the frequency range of 50 Hz—20 kHz. Domain observations confirmed that the oblique magnetic anneal resulted in as easy direction of magnetization at an angle to the strip axis and a general decrease of the domain wall spacing. Thus, domain refinement reduced high anomalous loss. Washko et al.(211) investigated the origin of losses during a.c. excitation in Metglas 26OSSC. The calculated losses: using equilibrium domain spacing and measured d.c. hysteresis losses, account for 70-90% of the total core loss. Results of this study thus conclude that losses are dominated by hysteresis losses in the material. The results of the study by Krause and Werner (206) on Metglas 26OSSC show that domain refinement by scratching, annealing in a transverse field, precipitation of crystallites, and applying a bias transverse field improve the high frequency magnetic properties. Thus, there appears to be disagreement as to what factor contributes most to the core losses of glassy alloys. 48 However, power losses of a glassy alloy, except at values of maximum induction approaching the saturation limit, are considerably lower than the conventional grain—oriented silicon steels. Liebermann et al. (212) have studied the effects of annealing in air on various properties of several glassy alloys. All the alloys studied showed a steady increase in TC on annealing at low temperatures, but some compositions showed a smaller increase in TC on annealing near the crystallization temperature than on annealing at lower temperatures. All the alloys showed a clearly measurable decrease in length on annealing. Annealing experiements on the shape of 60 Hz hysteresis loops showed a decrease in anisotropy associated with non—uniform internal stresses, but in some cases also showed the slow development of a fairly strong uniaxial anisotropy with its easy axis perpendicular to the ribbon axis. This uniaxial anisotropy was attributed to the development of an oxide layer during annealing, which in turn produced uniform compressive stress due to differential thermal contraction and therefore a stress-magnetostriction anisotropy. Effects of deformation and annealing on magnetic properties of F880B20 (50) and Fe4ONi4OP14Bé (50,213) are reported in the literature. Williams and Egami (50) observed that deformation increases anisotropy and stress-relief rate, while annealing reduces it. Luborsky et al. (213) showed that cold—rolling of glassy alloy ribbon results in a large increase in coercive force and a large decrease in magnetization in low fields. This was attributed to both slip-like structures developed by rolling and strain magnetostriction—induced anisotropy. 0n annealing the rolled ribbon to temperatures below its TX, the magnetic properties recovered to values obtained on annealing as—cast ribbon, but higher 49 temperatures were required. The rolled ribbon showed no evidence for structural change, or change in the glass or crystallization temperatures. Studies of effects of mechanical deformation on magnetic properties of FeAONi40P14B6, Fe4ONi4OBZO and COSONiZOFe6Si12B12 by Gibbs et al. (214) also showed that as the amount of thickness reduction increased, coercive force increased. However, the large increase in the coercivity after rolling was thought to be due to an increase in the domain wall pinning interaction. Chen et al. (44) studied the effect of structural relaxation on Curie tempratures of Fe-based metallic glasses. It was observed that the TC of these glasses increased monotonically with the annealing temperature Ta' 2.6 Mechanical Properties Metallic glasses exhibit unique properties arising from unique characteristics of the glassy state. Though metallic glasses in tensile deformation at room temperature seem macroscopically brittle, they are inherently plastic to a considerable degree as has been demonstrated with Pdgosizo by bending of thin ribbons (1,25,218) and compression of short rods (219-221). Fracture in metallic glasses proceeds by highly localized shear deformations, which is unlike the brittle fracture commonly observed in non—metallic glasses (e.g. silicate glasses). The elastic moduli of glassy metals are about two—thirds those of typical crystalline transition metals. Because of the lack of translational periodicity, the fracture strength of glassy metals approaches the theoretical strength ( Of = E/50) as compared with observed for crystal- line metals. Metallic glasses exhibit cycle dependent, i.e. not time dependent, fatigue. 50 2.6.1 Density and Thermal expansion In Ni-P (102) and Pd-Cu-Si (222) metallic glasses density was observed to vary linearly with alloy compostion, whereas for (Pd,Ni)—P (223) and Fe-B (224) glasses negative deviation from linearity was observed. The density in glassy alloys is usually about 1-2% less than corresponding crystalline alloys. The volume change associated with structural relaxation is 0.5% (223). The volumetric thermal expansivity B = (l/V)dV/dT, of most glassy metals is approximately equal to the corresponding crystalline counterpart. The volume change associated with crystallization of metallic glasses is W 1%. A number of ferromagnetic metallic glasses exhibit spontaneous volume magnetostriction near the Curie temperature. The magnetostriction is large in Fe-based glassy alloys and increases with decreasing metalloid content. The thermal expansion coefficient is negative in Fengg and becomes positive for the higher boron-containing alloy F883B17. 2.6.2 Elastic Constants The Young's modulus E and shear stiffness u are generally lower by 20-40% but the bulk modulus K by only “7% in as-quenched glassy alloys than in the crystalline state (225). The relatively large decrease in shear stiffness in the glassy state has been attributed to the inter- atomic shear displacements inherent in a disordered structure (226,227). The shear stiffness softening Aug reduces considerably upon structural relaxation. The difference in shear stiffness of a completely relaxed glassy alloy and its crystalline counterpart is small ((10%). The large 51 observed in as—quenched glassy alloys is due to frozen—in excess volume (27). The increase in E associated with the decrease in volume V during structural relaxation is large ( «’15%) and is comparable to that accompanying crystallization ( m 17%). The Young's modulus of metal— metalloid glasses, (Pd1_xNiX)80P20 (228), [ Fe1_X(Ni,Co)x ]75P16B6Al3 (229), and [Fe1_x(Ni,Co)x]goB20 (230) is found to show a positive deviation with x from a linearity. The increased E in glassy alloys is correlated to the magnetic ordering by Chou (229). The influence of magnetic ordering, however, is claimed to be insignificant in a ferro- magnetic glass (231). Chen and Krause (232) have observed that the Young's modulus E of metal-metalloid glasses increases with boron content, however, is independent of phosphorus content. E decreases in order (Fe,Co), (Pd, Ni) and Pt glasses while the bulk modulus of the corresponding crystalline metals, increases in that order. 2.6.3 Anelasticity Metallic glasses exhibit viscoelastic behaviour. The glass deforms instantaneously upon applying tensile stress, followed by time dependent reversible deformation and steady—state viscous flow. The time required for attaining a steady flow is of the order of the structural relaxation time. The recoverable anelastic compliance, Ja at a stress level of m 1 MPa in Pd—Si (21) is small at low temperature (of the order of the elastic compliance) but increases rapidly in a narrow temperature range when viscosities reach 1012 P. Ja becomes as high as several hundreds times the elastic compliance. In this temperature range both isothermal J3 and n of the glass remain constant at small stress 0 g 1 MPa but decrease at a higher stress i.e. the glass becomes non—Newtonian. The total anelastic strain, Ea = lac, approaches a constant value «,10'2. Masumoto and Maddin (1) reported for a PdEMPiZO glass a relatively small €a_= 10"3 at 100‘? (far below Tg = 370°C) but a large value of Ea = 10-2 at about 200°C. The high anelastic strain implies the occurance of a considerable directional structural ordering. Anelastic behaviour of Pd-Au-Si (32) was examined thrOUgn internal friction methods. The study has indicated that the internal friction of the amorphous Pd—Au-Si alloy increases exponentially above 150°C suggesting that structural relax— ation occurs at temperatures far below the glass transition temperature (about 370°C). 2.6.3.1 Thermoelastic Behaviour Elastic relaxation may originate from the change in temperature caused during flexural vibration. Berry and Pritchet (43) obtained from thermoelastic relaxation measurements in glassy alloys the thermal diffusivities 0th = 0.05 cmZ/s for Pdgfifiilg and 0.012 for Fe75P15Cu3 which are only one-half and one-third of those for the crystalline phase(s). This is consistent with the higher electrical and thus thermal resistivity in the glassy state. 2.6.3.2 Hagnetoelastic behaviour When a substance is exposed to magnetic field, its diamensions change. This effect is called magnetostriction. The fractional change in length, A = Al/l, measured at saturation is called saturation magnetostriction, AS. Magnetostriction occurs in all pure substances. However, even in strongly magnetic substances, the effect is small: As 'lf— ..u.. C! '1 g. ‘4, n..\ ‘5 m, 6. is typically of the order of 10-5. The iron-rich glasses show large positive magnetostrictions, AS = (20—40) x 10'6 (182,233), while As is negative in Co glasses. In the Fe-Co system AS = 0 occurs at approxi— mately the Fe—Co ratio (“0.05) relatively independent of glass formers (234,235). The addition of Ni to an Fe alloy decreases AS. One of the consequences of magnetostriction is dependence of Young's modules E of a material on its s ate of magnetization. When an originally demagnetized specimen is saturated, its modulus increases by an amount AR. he value of AB effect, hS—Ed/hs, is about 0.06 for Ni and less than 0.01 for Fe. The striking AB effect in certain ferro— magnetic metallic glasses was first reported by Berry and Pritchet (43, 236). The AH effects in metallic glasses can be large and are measured at room temperature with moderate magnetic field N 5 0e. The large AE effect can not only reduce the elastic stiffness of materials but also can significantly alter or even reverse the sign of the temperature coefficient (1/E)dE/dT (237). The magnitude and field dependence of the AB effect can be varied considerably by prior magnetic annealing that reduces internal stress and induces a uniaxial magnetic anisotropy in the specimen. The maximum AH effect of metallic glasses is exception— ally large, «0.4, for the FGAOHiAOPIABfi, m0.8 for F875P15C10 and Fe80P13 C7 (239) and ml.9 for Fe788i10B12 (238). In the as—quenched state the value of AH effect is small (< 0.1) for Fe—based metallic glasses. As Ni replaces Fe, the Ab effect decreases in magnitude and the stress level required for saturation also decreases, as a result of decreasing magnetostriction and magnetic moment. MAXIMUM OR FRACTURE STRESS.Kg/mm2 Figure ‘00 54 100 r 200> O APPAR(NY "(to SYRESS F P C 'oo‘ss L 1 AH ‘ FRA-CYURE SHI£SS l ULHMAT( sme~cm o—USl 4 200 ‘00 —200 11: -‘OO O 1 4 l a O ‘00 200 TESTING TEMPERATURE. 'c Temperature dependences of apparent APPARENT YIELD STRESS. Kgxmm’ yield and strengths for amorphous Fe—P-C alloy (1). Figure 12: STRAIN RATE, sec" Relationship between the strain rate of amorphous Pd-Si alloys (25). n E o. \E r. ... _. -fi. __ 1103 E 0') x E _ 14() F- (n m ~+ FR4CrU 5!. LL] "" RES - (I m y— m 120 *- 3 Q I) [1.11 p— O ;: C) 100 r- -4 2 O i o g ------------ 4- 2 1— YI_E_L2 STRESS x . in < o o— —‘ — “'— ""— ru- ~ 6 o ”J (I: CC) - 5 g; 2 0 U ’— >— < -1 4 E 60 1 l J -4 - - 10 ‘0 3 10 2 tensile and strength properties kJ'l p1 lSO~~-—-——-—-—~~ «E ° uauuuu S'HISS W _ 0°C 6 ° N \01 IV\\ X ‘30 H ‘OO'C o“\° m. ”’4' r‘u \"\-\ m l/. \O \5 w *— ' l I I\Q; u 95 ISO‘C ‘ "' " "\. 9\\\.-l. V‘ : \f\y ”O *— #; __._________,__ (_____,.-.. T:__: 2 ' . / - 2 '\ '/' ' x // <1 2 90 ~ ' \ . 200‘C . J o 0: 300°C ° 0 -.. -—-.-- O >— \ w ‘ o 0 or I) 70 r— OUCYlLE of BRHYLE .— 9 ‘(J a. 8 \ or '— 0. ‘ u 50 s ‘ was 1111411174 1 11 IO IO IO IO STRAIN RATE, sec" Figure 13: Effect of strain rate on the fracture stress at various temperatures (25). 2.6.4 Strength and Hardness Metallic glasses exhibit a very high fracture strength and a high fracture toughness at temperatures well below T8° For Ni— and Fe-based alloys tensile strengths of 140 kg/mm2 (239) and 310 Kg/mm2 (161) have been reported. Fracture strength Of 2 370 Kg/mm2 exhibited by FBBOBZO metallic glass (240) is much higher than that for the convensional highest strength steels (maraging steel and a hard drawn piano wire). Such high strength in glassy alloys is in sharp contrast with that for the commom non—metallic glasses. It has been reported that temperature dependence of the tensile strength for the amorphous state is, in general, higher than for the fully crystallized state (after annealing), and the strength decreases sharply at temperatures well below the glass transition temperature, T8 56 Table 111: Mechanical properties of metallic glasses;(a) Vicker's hard— ness IIv (kgmm‘z), fracture strength of (kgmm‘z), Young's modulus E (x1000 kgmm’z). H(°) o- o H/o H/o E(C) E/o E/o v T y f y f y Pd77Cu68i17 455 157 — 2.91 - 9.3 59 - PTW.5CU6SiHi5 498 - 157 — 3.17 8.97 - - Pd 64Nil6p20 452 160 - 2.83 — 9.3 58.95 - Pd16Ni64P20 541 180 - 3.01 - 10.6 58.9 - Pt7SP25 344 - - - - 9.3 - - Fe75B25 1314 - - - - 17.9 - - FeBOBZO 1100 - 370 - 2.97 16.9 - 45 Fe80P16C3B1 835 - 249 - 3.35 13.8 - 55.4 ZrSOCuSO 580 - - - - 8.51 - - TiSOCUSO 610 — — - — 10.8 - - Fe8OP20 755 - 236 - 3.2 13.3 - 56.4 Pd4ONi4OP20 452 — 158 - 2.97 10.0 - 63.3 (°)Data obtained from literature. See refs. 5,245 and 246. (b)Accurate to 15% (C)Accurate to 12% 57 (1,241) (Figure 11). Masumoto and Maddin (25) have shown that as the strain rate decreases the fracture stress increases (Figure 12). This behaviour is in contrast with the general strength properties of crystalline metals. The effect of temperature and strain rate on the fracture stress and mode is shown in Figure 13 (25). The effect of temperature on flow and fracture of various metallic glasses is summerized by Pampillo (221). The hardness Hv , fracture strength Of, Young's modulus E, HV /of and E/cfi of some typical metallic glasses are listed in Table III. An indication of plastic strength of a material is given by measurement of its microhardness (242). For crystalline metals, the ratio of hardness (Vickers / 136° diamond pyramid) to yield stress is = 3. As metallic glasses do not work harden, their tensile strengths must be equal to or less than their yield strengths. Metallic glass Fay)B20 which shows remarkable strength has the H/of value close to the theoretical value. The ratio E/Of «.50 of metallic glasses may be compared with that of Fe whiskers, E/Of fl’15.5 (243). Masumoto (244) has reported that the influence of composition on Hv is similar to that on E. With increasing metalloid content (C, P, B or Si) in Fe- or Co-based glasses, HV increases. Hv decreases in the order of Fe, Co, Ni, Pd and Pt glasses as does E. 2.6.5 Fatigue Behaviour and Toughness The fatigue lifetime of metallic glasses Pd—Si (247), Pd-Cu—Si and Fe-Ni-P-B-Al (248) has been reported in the literature. Stress (S) Versus cycle-to-failure (N) behaviour was qualitatively similar to that Of other materials. The fatigue behaviour of metallic glasses was 58 observed to be cycle dependent, not time dependent as for hard, non- metallic glasses. The S-N curve had distinct fatigue limit 0*" with O*’/Of 2 0.3 at a critical number of cycles «105. The rate of fatigue crack propagation (dl/dN) in glassy alloys was described by a power law of the stress intensity factor with exponent m ranging from 2-4 (247, 249,250). A well-defined plastic zone consisting of slip bands around a fatigue crack propagated along slip bands formed ahead of the crack. Such a crack growth mechanism in a glass is very similar to that in crystalline metals. The fatigue slip bands which had a single slip step, however, were distinct from the so-called extrusions or intrusions as generally observed in crystalline metals. The mode III fracture toughness K111 values for Fe, Pd-Si and Zr—Cu glasses were 320, 140 and 190 1(gmm‘3/2 respectively (251). The tearing energy F was found to be proportional to the yield stress Oy' : F/Oy : 0.03 mm. The tearing energy of metallic glasses F 2 10 J/cm2 is many orders of magnitude larger than in crystalline metals ( “10.01 J/cm2 ) (245). 2.6.6 Creep Creep measurements in a metallic glass were first reported by Chen and Turnbull for Au-Ge-Si (252). The experiments demonstrated for the first time the existance of a rheological solid glass/metastable liquid transition. Creep measurements were conducted within a temperature range just below the glass transition temperature. Within this temperature range, the amorphous alloy behaves as a viscous liquid with a stress independent viscosity n ranging from 109 to 1012 P. The temperature dependence of viscosity follows a Fulcher-Vogel relation: n = no eXp[A/(T-To)]. The n -T relation was found to be independent of how the 59 V -- .- . - - .. “.... --.—..---._.,_.. .1 \ /--uL(.m~T~C. LINE OF CRY51ALLOZATION WITHOUT SIRESS \ \ \ rcc urusraact PHASF Q \ I. w (,mmm‘, (INT; 0! —/ \\ O S; (,nvsTm LIIAYION \ 3 WI!“ STRESS \ ‘— A 1 lm \ 2 200 F T 800 aexq m l O 3 ‘3; \ 1». \ 1.. \ >— O V) \ O O O \\ \ o o\\ o AMORPHOUS PHASE \ TOO >— O L1111111 1 1 1111111 1 Lilllll TIME (mm) Figure 14: Effect of stress on the start of crystallization in an amorphous Pd-Si alloy (25). temperature was reached. If the temperature is suddenly changed, the Viscosity is found to lag behind the temperature change. Chen and 301dstein (21) have reported similar results on Pd-Au-Si, Pd-Ag-Si, and PC1—Cu-Si metallic glasses near their glass transition temperatures. At low stress and below the temperature T3, at which n = 1012 P, the bGhaviours of the alloys were similar to that of Au-Ge—Si. However, above Ta creep became non—Newtonian, and the viscosity was, therefore, Stll‘ess dependent. Annealing the samples at T > T8 had no effects on the Creep behaviour near Ta, and hence phase separation was disregarded as the possible cause of the anomalous behaviour. Maddin and Masumoto (25) carried out extensive creep measurements on PngSiZO in the temperature range 373 to 473 K ('1‘8 = 655 K). It was obServed that an applied tensile stress promotes crystallization (Figure 60 14) (25). Creep experiments were conducted at temperatures and during times within the region where the solid remained amorphous. From the analysis of the steady—state creep it was concluded that creep was non-Newtonian. 2.6.7 Effect of Deformation Glassy alloys exhibit localized plastic deformation in tension, compression or in bending, suggesting that the strong metallic bonding remains though there is no long range order. Effects of deformation in cold rolled glassy alloys e.g. Pd-Si (253) and Fe—P—C (254) have been reported. It was observed that in x-ray diffraction pattern, the first peak position shifts slightly to the low angle side and the width of the peak at half height becomes larger by cold rolling (Figure 15) (253). Mossbauer spectroscopy results showed that cold rolling decreased the component of the magnetization axis lying in the plane of the specimen (254). On cold rolling, the Young's modulus, the fracture stress and the hardness decreased as the reduction in the thickness of the glassy alloy increased (Figure 16) (253). The electrical resistance, however, increased on heating after cold rolling. It was concluded that deform- ation by rolling at room temperature produces much more disordered structure than is present in the as-quenched state by introducing additional irregularities through defects. Results of differential specific heat measurement experiments showed increase in the crystal- lization temperature indicating stabilization of the internal structure of the amorphous phase. Chen (255) has reported effects of cold rolling on Young's modulus 61 I Wavy-watt .Cu fl-(Lz'l N‘-$-~ Anovm “To, / ". 600 4 - T. _——4 —L_. .. -.H WI“ ~ot Naduzz. a. ... . .-. . . . . I L 0 \ 40" 60$ . 3 N \ unn‘ 1m 0 to \ mn‘ SH 4 ‘vv _ ... ..'.._.. .--l___ r .3 j ) 9 L u / E 200 - '3“ z p/ \ S \ / N. -fl \TF. muss” o n )L )6 Jo LO 1: u 15 to Degree 10 Figure 15: Effect of cold rolling on diffraction patterns of an amorphous Pd—Si alloy (253). RW '3 60 run-sin Alloy 9 E 2 N ” SS * \ 1 O c g so 1 you...) ,_0_ _. .-.. , _.- -..—.1 "-4 Q“ ‘60 b T a. awn? E no q Tag 8‘ :5 m. : ‘20P ‘ 70 E g 4 T G C 100 . 4/ 2 M. g p 1 w E f an» E : § 0 “V ( § xn"b‘——**—-——~e\ii\“\\\fi E 1 L00 0 20 £0 Reduchon by Cold-RoH-nq,'l. FiSure 16: Effect of cold rolling on hardness and tensile properties of the Pd—Si alloy (253). 62 and structure of metallic glasses. The study concluded that cold rolling of metallic glasses induced two distinct atomic arrangements. Local atomic regrouping appeared to strengthen the structure and increase Young's modulus, whereas the unstable plastic flow lead to structural disorder and softening of glasses. Fe—based glass showed least ductility and smallest poisson's ratio among glasses studied, due to the highest degree of disorder created. 2.6.8 Annealing Embrittlement without Crystallization Many Fe-based glasses which are ductile in the as-quenched state become brittle upon annealing apperantly as a result of structural relaxation (41,45,214,229,250,256). In binary alloys such as Fe-B (45), Fe-P (45), and Zr—Cu (257,258), the alloys embrittle as a result of crystallization, whereas glass forming Pd, Ni, and Pt alloys exhibit a high ductility even in a partially crystalline state (259). The presence of crystalline dispersion in Ni-B alloys was shown to have less embrittling effect than it did in Zr—Cu (257). However, kinetics of embrittlement in many glassy alloys often correlate neither to the state of stress relief nor to the crystallization process (45,229,260,261). Chi et al. (262) showed that FeaoNigoBgo ribbons produced at slower quenching rate were stress relieved at slower rates but embrittled at much lower temperature with low apparent activation energy «J eV. Based on the observation that Fe80P13C7 (244) and Fe40Ni40P14B6 (214) lose ductility at lower temperatures of annealing than do Fe78B128110, IRESOBZOv FEAONiAOBZO which contain no phosphorus, it was proposed that tfiuosphorus is the embrittling agent in the two alloys containing it. lhilter et al. (263) found P enrichment on the fracture surface of 63 Fe40N140P1486 ribbons. Besides phosphorus segregation or clustering (256,263), phase separation of transition metals (45) and an increase in viscosity causing easier crack propagation (50) were prOposed as the possible reasons for observed embrittlement. Chen (45) showed that the enhanced loss of ductility of various glasses studied occured in the alloys containing two or more metalloid elements. Some P-containing glasses Fe80P17(B,Si,C)3 were found to exhibit a mechanical stability comparable to or surpassing those of B-containing glasses, Fe8OB20 and Fe4ONi4OB20° Thus, he concluded that in the Fe-based alloys, presence of phosphorus cannot be the only cause for such annealing embrittlement as suggested by Luborsky et al. (214). Chen (45, 229) proposed that local structural and compositional fluctuations analOgous to phase separation in regions of 4:20 °A accompanied by structural relaxation may be responsible for the enhanced embrittlement in glassy alloys. The resulting stress concentration around the clusters leads to embrittlement. This suggestion was strengthened by the fact that ternary alloys of Fe—P—B, Fe—P—Si and Fe-Co-P exhibited a higher tendency to phase separation and hence were more susceptible to embrittlement than were the corresponding binary alloys of Fe-P, Fe-B, and Co-P. 2.6.9 Stress-Strain Behaviour Tensile tests have been performed on several metallic glasses (1, 218,239,264,265). Generally, the specimens were in the form of thin strips 20-40 um thick and 1 mm wide. Typically (1,218,239,265) such ribbon specimens fail by shearing off (antiplane shear, slant fracture) 64 on a plane 450 to both the tensile axis and the thickness vector. Occasionally, at low temperatures, portions of square fracture are observed (264,266), i.e. fracture surface is perpendicular to the tensile axis -macroscopically (i.e. glass follows Tresca criteria (267)). Tensile failure of glassy alloy wires and strips (plane stress) is accompanied by intense plastic shear deformation. This is true whether the macroscopic mode of failure is brittle (tearing; antiplane shear, mode III failure) or ductile (yielding). Figure 17a shows a typical room temperature tensile load versus elongation curve for Pd-Si (1). Unloading experiments as in Figure 17b show that irrecoverable plastic strains occur before failure,and except for a small anelastic component (268), the glass behaves in an elastoplastic fashion. At room temperature tensile plastic elongations (measured after subtracting the linear elastic part) have been reported to be between 0.1 and 0.5% (1, 218,239) showing a macroscopically brittle behaviour. Masumoto and Maddin (l) have reported macroscopically ductile tensile behaviour in Pd-Si when pulled at temperatures above 473 K as shown in Figure 18. Below this temperature, the tensile elongation is 0.1% and increases rapidly above that temperature reaching 4% at 573 K. As metallic glasses do not work harden their tensile strengths must be equal to or less than their yield strengths. The former case is observed when yielding and failure occurs simultaneously. According to plasticity theory, a thin sheet will yield (following von Hises' criteria) in a zone whose normal makes an angle 0 = 35.3° with the tensile axis and 900 with the thickness vector. According to Argon (269), 0 will increase slightly if, as for metallic glasses, oy/E (oy.= Yield stress, E = Young's modulus) is large; for oy/E 2:0.02, typical of Figure 17: Figure SHess.Kg/mnfi 18: 65qumm L0AD,K9 ExtiufiON J 03mm .——_‘ Representative stress-strain curves of a Pd—Si alloy (1). 200 1" Pd 20 0/0 Si 0°C (In-r HosuMoIo ‘ Maddin) ' 100°C I 1: 200°C [I III '” 250°C .I 100 *‘ u I :1 I . ...! I3OC)C cg :I 11': 050°C II It 2 'o' F_————7T“;‘——~. ::,: 1 450 C 2 0 11111 1 1 1 11 {F 1 O l 2 3 4 21 Strain X Stress-strain curves temperature (1). of a Pd-Si alloy as a function of 66 glassy alloys, predicted angle is 37°. Accordingly, the included angle (90-0) measured on a wide face at the fracture tip of the specimen which fails coincident with yielding should be «53°. This failure geometry has been reported for Ni-Fe-P—B—Si ribbons (270). In Ni—Pd—P alloys it is reported that the fracture surface of the ribbon specimens made an angle of 55 :50 and 90° with the tensile axis on the narrow and wide surfaces respectively (271). Davis (270) has suggested that in order to observe the 53° mode of failure at T << T8, one must test specimens with a reduced area gage section with width to thickness (w/t) ratio of the order of 8:1. If smooth, uniform cross section ribbons are tested, failure is typically initiated at grips and occurs by tearing (mode III, antiplane strain (249)). If a reduced section specimen with w/t >> 8 is pulled, failure will occur by tearing across the gage section but at somewhat higher stress due to elimination of the grip constraint. Uniform cross section ribbons fail (at the grips) in 45° slant mode. Tomizawa and Masumoto (272) tested Cu-Zr alloy ribbons (0.5 mm wide and 201Jm thick) in tension. The fracture stress was observed to decrease with increasing strain rate. The fracture surface was at 45-50° to the tensile axis and contained about 30% of smooth, featureless region and 70% of a region of the characteristic 'vein' pattern. Upon increasing strain rate in the tension test, the smooth region increased and the vein pattern became more pronounced. Uniaxial compression deformation in Pd-Cu-Si (220,273), Pd-Ni-P and PhD—Ni-P (219) glassy alloys is reported in the literature. Figure 19 Shcnus the uniaxial compressive nominal stress—strain curves for Pd-Cu-Si (220) at different temperatures, which approximate the expected 67 3 o 1 6 o .0...“ (ow-(nut Suttqul'N" $0» Figure 19: Compression stress—strain curves for a Pd—Cu-Si alloy (220). behaviour for an elastic—perfectly plastic material. Davis (246) has attributed the roundness of these curves at low strain to extrinsic features such as slight misalignment of the specimen ends. In compression depending on temperature, the glass deforms initially in a smooth or stable way and after a certain strain continues by unstable or jerky flow and serrations (219,220). Serrated flow has been observed in Pd—Cu-Si (220), Pd—Ni-P and Pt-Ni—P (219) glassy alloys. Serrated flow in Pd—Cu-Si is supressed below 200 K (220). Below this temperature deformation proceeds in a stable manner. Pampillo and Chen (220 have suggested that the unstable flow found in these experiments could be due to partial relaxation of the atomic rearrangements produced by plastic deformation. If these arrangements lead to a 'softer glass' (e.g. 68 destruction of compositional short range order or increase in the mean atomic volume), then above a certain temperature, To, where there is enough atomic mobility, partial relaxation of these rearrangements (e.g. regeneration of the destroyed short range order) may occur as quickly as the rearrangements are produced: this leads to unstable flow. Below To (To:: 200 K), the atomic mobility is too small to allow this and plastic deformation becomes stable. The explanation suggested is similar to the model postulated to explain serrated flow in crystalline substitutional solid solutions by Cottrell (274). So far, the anology holds qualitatively, but more experimental data are required before reaching a conclusion. 2.6.10 Deformation Characteristics Plastic deformation of metallic glasses occurs by either of two mechanisms: a diffuse rearrangement of atoms or the nucleation and propagation of narrow sheared regions. The first type of deformation mode is homogeneous and occurs when metallic glasses are deformed above their glass transition temperatures T in absence of crystallization g' and at slow strain rates. Flow takes place uniformly with each volume element contributing an equal amount of strain. At low stresses, the strain-rate increases linearly with stress; at slightly higher stresses the strain-rate follows the sinh (stress) law (275). Fracture occurs when some section has necked down to a narrow thickness. Such a homogeneous (Newtonian) viscous flow in Au—Ge-Si metallic glass was first demonstrated by Chen and Turnbull (22). The second type of deformation mode which is extremely inhomogenous occurs in the form of highly localized shear deformation 69 bands at temperatures well below T8 and at high strain rates. Plastic flow in a metallic glass at temperatures T<) 4.8 where u, b and n are material constants. The corresponding response curve in simple shear is expressed by T(k) = u [ l + "E7 k2 ]“"1 k (0 g k (on) 4.9 where U is the shear modulus at infinitesimal deformations, and n is the hardening parameter. Figure 41 shows the response in shear for various values of n. A power—law material hardens or softens in simple shear depending on whether n>l or n<1 respectively. For ng 1/2, T(k) strictly increases monotonically with k such that T(k) —> as k —>a3. For n=l/2, T(k) approaches u//2b—as k -> m, whereas for n < 1/2, (k) has a maximum at k = ko=[n/b(1-2n)]1/2 and then gradually decreases to zero. Thus, all solutions of the displacement equation of equilibrium (4.2) are elliptic everywhere when n gl/Z. For n O as k —> w. Equation (4.2) suffers a loss of ellipticity at a solution U and a point (x1,x2) if |VU(x1,x2)l >1. The corresponding crack—tip nolinearity is shown by the schematic 106 x2 Figure 43: Crack- -tip shear band zone predicted corresponding to the response curve shown in Figure 42 (3O 9) Figure 44: SEM picture showing evidence of normal surface displacement in the laser irradiated metallic glass specimen. 107 diagram in Figure 43. Across the elastostatic shocks, displacement and traction are continuous but displacement gradient and stress suffer a jump discontinuity. Within the cross-hatched region, strain localization and shear band formation are predicted. The generalized form of the shear stress-strain response curve given by (4.10) and (4.11) is 1(k) — 1(k) k 4.12 l O IIA "A H k < m 4.13 II F. Q H IIA where a is a material constant, restricted by 0 g a < 1. The angle 00, defined in Figure 43, is related to the material constant,a, by (309) sine. = -- ----- 4. 14 For the special class of materials considered by Knowles and Sternburg (309) a =1/2 and the angle 00 determined from equation (4.14) is l9.47°. Table V: Comparison of measured and calculated 90 values. 00 measured a calculated 0 obtained by a value from measured Knowles and considered by values Sternburg (309) Knowles and Sternburg (309) 18.5:0.5° 0.5183 19.47° 0.5 l9.0:0.5° 0.50878 l9.5:0.5° 0.49394 20.5:0.5° 0.48125 Returning to the experimental results, Figures 33(b) and 38 show crack-tip nonlinearity observed in the form of a shear band zone in the laser irradiated metallic glass. Note the observed shear band zone is remarkably similar in nature to the theoretically predicted configur- ration. 00 values measured from the experimental results are summarized 108 in Table V. The average of measured values of 00 when substituted in equation (4.14) yields a = 0.50056, which is in close agreement with the value of a (=l/2) considered by Knowles and Sternburg (309). Thus the shear stress-strain response curve for the metallic glass studied can be represented by equations (4.10) and (4.11). Mode III nature of deformation in the laser irradiated metallic glass was inferred from the tearing steps observed as shown in Figure 32. Additional evidence in support of this argument comes from the normal surface displacement observed in laser irradiated metallic glass as shown Figure 44. Shear band formation being the primary mode of deformation in metallic glasses, deformations can be considered locally volume preserving, thus satisfying the condition of incompressibility. Above 523 K, results of the tensile stress-strain curves exhibit clearly that strain softening does take place (as no necking was observed. See Appendix 1). Finally, though the crack propagation in the laser irradiated metallic glass foil, occurs through a steep temperature gradient, a crack-tip shear band zone, away from the HAZ is possible when the crack arrest occurs in the region where the temperature is suitable for such an instability. Thus, the model predicted by Knowles and Sternburg (309) is applicable in this study. 109 V CONCLUSIONS In conclusion, it has been shown that a 12 msec duration laser pulse interacting with an Fe—based metallic glass causes crystalization in the HAZ, even at power levels which do not produce substantial melt or hole drilling. It is also shown that the crystallization occurs through the growth of spherulitic domains which coalesce to produce a network—type micro— structure. It is suggested that the morphology and kinematics of this surface crystallization may be influenced by preexisting nuclei. Joining or welding of two foils of the metallic glass is possible with the use of lower power density and multiple pulsing. However, in the HAZ, microcrystallinity is developed, causing some embrittlement and crack formation. Results of XRD experiments on isothermally annealed ribbons of metallic glass foils indicated that a—Fe was the first crystallization product followed by formation of Fe3B phase, which is a metastable one. X-ray diffraction studies on isothermally as well as isochronally annealed ribbons did not show any difference in the degree of crystal— linity from opposite surfaces. For the first time, a direct experimental mapping of a crack-tip shear band zone was recorded. A theoretical model, predicted by Knowles and Sternburg, was shown to be applicable in this study. 110 APPENDIX I Original cross sectional area of the specimen, A0, is A0 = 5 mm x 0.025 mm = 0.125 mm . As the load bearing capacity of the specimen does not change _____ = -____-__— (1) Assuming locally volume preserving deformations, i.e. constancy of volume, A01o = A 1 (11) max max where Amax and lmax represent the cross sectional area and length of specimen at the maximum load respectively. Thus, Aolo 0.125 x 20 2 A = -------- = = 0.1247 mm . (iii) max 1 20.045 max Now, substituting value of Amax in equation (1), 0f x Amax 91.94 x 0.1247 Af = = = 0.10156 mmz. (iv) 0 112.90 max Thus, if necking took place, then the cross sectional area at fracture should have been 0.10156 mm2 , which is about 18.75% smaller than the original cross sectional area. Such a reduction in area was not seen. This supports that strain softening occured in the metallic glass at the temperature. 111 APPENDIX II In order to obtain a pulse duration profile of the ruby laser used in this study following arrangements were made. A defocussed ruby laser beam was made incident on a silicon photodiode (Newport Research Corpo— ration, Model # 882) after passing the laser beam through an optical diffuser and a neutral density 3.0 filter. This was done to avoid the detector saturation. The output of the pulse detector was fed to a storage oscilloscope (Tektronix, model # 464) through a laser power meter (Newport Research Corporation, model # 820). The pulse duration profile obtained is shown in the figure below. Figure I: Voltage versus time duration profile of a ruby laser pulse. LIST OF REFERENCES 10. 11. 12. 13. 14. 15. 16. 17. 18. LIST OF REFERENCES T. Masumoto and R. Maddin, Acta Met., 19 (1971) 725. R. L. Freed and J. B. Vander Sande, ibid 28 (1980) 103. C. P. Peter Chou and F. Spaepen, ibid 23 (1975) 609. J. M. Vitek, J. B. Vander Sande, and N. J. Grant, ibid 23 (1975) 165. L. E. Tanner and R. Ray, Scripta Met., 11 (1977) 783. M. A. Marcus, Acta Met., 27 (1979) 879. F. E. Luborsky and J. L. Walter, J. Appl. Phys., 47(8) (1976) 3648 H. Chang and S. Sastri, Metall. Trans., 8A (1977) 1063. F. E. Luborsky, Mater. Sci. Eng., 28 (1977) 139. K. Shimomura, P. H. Shingu, and R. Ozaki, J. Mater. Sci., 15 (1980) 1175. T. Watanabe and M. Scott, ibid 15 (1980) 1131. U. Herold and U. Koster, in 'Rapidly Quenched Metals 111', (ed. B. Cantor), Vol.1, p. 281, Metals Society, London (1978). A. L. Greer, Acta Met., 30 (1982) 171. D. G. Morris, ibid 29 (1981) 1213. P. Duhaj, D. Barancok, and A. 0ndryka, J. Non-Cryst. Solids, 21 (1976) 411. B. C. Giessen, Ref. 12. p. 438. J. F. Ready, in 'Effects of High-Power Laser Radiation", Academic Press, New York (1971). K. Mukherjee, T. H. Kim and W. T. Walters, in 'Lasers in Metal— lurgy', (ed. K. Hukherjee and J. Mazumder), p. 137, The Metal— lurgical Society of AIME, Warrendale, Pa. (1981). 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40 41. B. G. Bagley, H. S. Chen, and D. Turnbull, Mater. Res. Bull., 3 (1968) 159. A. Staudinger and S. Nakaharas, Thin Solid Films, 45 (1977) 125. H. S. Chen and M. Goldstein, J. Appl. Phys., 43 (1972) 1642. H. S. Chen and D. Turnbull, J. Chem. Phys., 48 (1968) 2560. H. S. Chen, J. Non-Cryst. Solids, 29 (1978) 223. C. A. Angell, J. Am. Ceram Soc., 51 (1968) 117. R. Maddin and T. Masumoto, Mater. Sci. Eng., 9 (1972) 153. H. S. Chen, J. Non-Cryst. Solids, 27 (1978) 257. H. S. Chen, J. Appl. Phys., 49 (1978) 3289. H. S. Chen and D. Turnbull, Appl. Phys. Letts., 10 (1967) 284. H. S. Chen and K. A. Jackson, in 'Metallic Glasses', p. 75, Metals Park, Ohio:ASM (1978). T. Egami and T. Ichikawa, Mater. Sci. Eng., 32 (1978) 293. N. A. Pratten and M. G. Scott, Scripta Met., 12 (1978)137. H. S. Chen, H. J. Leamy, and M. Bartmatz, J. Non-Cryst. Solids, 5 (1970) 444. T. Soshiroda, M. Koiwa, and T. Masumoto, J. Non-Cryst. Solids, 22 (1976) 173. F. E. Luborsky, J. J. Becker, and R. 0. McCary, IEEE Trans Magn., MAG-11, (1976) 1644. T. Egami, J. Am. Ceram Soc., 60 (1977) 128. K. D. Krishnanand and R. W. Cahn, Scripta Met., 9 (1975) 1259. H. S. Chen and E. Coleman, Appl. Phys. Lett., 28 (1976) 245. F. E. Luborsky, R. O. McCary, and J. J. Becker, in 'Rapidly Quenched Metals', (eds. N. J. Grant and B. C. Giessen) Vol. 1, p. 467, MIT Press, Cambridge, MA (1976). Y. Waseda and T. Masumoto, Z. Phys., B 22 (1975) 121. Y. Waseda, H. Okazaki, and T. Masumoto, J. Mater. Sci., 12 (1977) 1927. H. S. Chen, in 'The Structure of Non—Crystalline Solids', (ed. P. H. Gaskell) p. 79, Taylor and Francis, London (1977). 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. H. S. Chen, L. C. Kimberling, J. M. Poate, and W. L. Brown, Appl. Phys. Lett., 32 (1978) 461. B. S. Berry and W. C. Pritchet, J. Appl. Phys., 44 (1973) 3122. H. S. Chen, R. C. Sherwood, H. J. Leamy, and E. M. Gyorgy, IEEE Trans. Magn., MAG-12, (1976) 933. H. S. Chen, Scripta Met., 11 (1977) 367. F. E. Luborsky and J. L. Walter, Mater. Sci. Eng., 35 (1978) 255. H. S. Chen, J. Appl. Phys., 49 (1978) 4595. F. E. Luborsky, AIP Conf. Proc., 29 (1976) 209. B. S. Berry amd W. C. Pritchet, J. Appl. Phys., 47 (1976) 3295. R. S. Williams and T. Egami, IEEE Trans. Magn., MAG-12, (1976) 927. J. W. Christian, in 'The Theory of Transformations in Metals and Alloys', p 525, Pergamon Press, Oxford (1965). D. Turnbull, Solid State Phys., 3 (1956) 225. B. G. Bagley and E. M. Vogel, J. Non-Cryst. Solids, 28 (1975) 29. C.-P. Chou and D. Turnbull, ibid 17 (1975) 168. J. J. Burton and R. P. Ray, ibid 6 (1971) 393. H. L. Yeh and R. Maddin, Ref. 38. p. 281. M. G. Scott and P. Ramachandrarao, Mater. Sci. Eng., 29 (1977) 137. P. H. Shingu, K. Shimomura, R. Ozaki, K. Osamura, and Y. Mukarami, as in Ref. 12. p. 315. N. Funakoshi, P. Kanamori, and T. Manabe, Japan J. Appl. Phys., 16 (1977) 515. P. G. Boswell, Scripta Met., 11 (1977) 701. E. Coleman, Mater. Sci. Eng., 23 (1976) 161. T. Masumoto, A. Inoue, and H. Kimura, J. Japan Inst. Met., 41 (1977) 730. K. L. Chopra, in 'Thin Film Phenomena', p. 195, McGraw Hill, New York, (1969). J. Megusar, J. B. Vander Sande, Ref. 38. p. 401. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. R. L. Freed and J. B. Vander Sande, J. Non-Cryst. Solids, 27 (1978) 9. B. G. Bagley and D. Turnbull, Acta Met., 18 (1970) 857. J. L. Walter, P. Rao, E. F. Koch, and S. F. Bartram, Metall. Trans. 8A (1977) 1141. J. L. Walter, S. F. Bartram, and R. R. Rusell, ibid 9A (1978) 803 L. E. Collins, N. J. Grant, and J. B. Vander Sande, J. Mater. Sci., 18 (1983) 804. A. S. Schaafsma, H. Snijder, and F. Vander Sande, Phy. Rev., B 20 (11) (1979) 4423. O. T. Inal, L. E. Murr, and F. G. Yost, Mater. Sci. Eng., 51 (1981) 101. H. N. 0k and A. H. Morrish, Phy. Rev., B 22(7) (1980) 3471. C. F. Chang and J. Marti, J. Marer. Sci., 18 (1983) 2297. H. N. 0k, K. S. Beak, and C. S. Kim, Phy. Rev., B 24(11) (1981) H. N. 0k and A. H. Morrish, J. Appl. Phys., 52(3) (1981) 1835. K. Osamura, K. Shibue, R. Suzuki, Y. Mukarami, and S. Takayama, J. Mater. Sci., 16 (1981) 957. F. L. Cumbrera, C. F. Conde, M. Millan, A. Conde, R. Marquez, J. Mater. Sci. Lett., 2 (1983) 499. T. Kemeny, I. Vincze, and B. Fogarassy, Phy. Rev., B 20(2) (1979) 476. 0. T. Inal, L. Keller, and F. G. Yost, J. Mater. Sci., 15 (1980) 1947. U. Koster and U. Herold, Scripta Met., 17 (1983) 495. Idem, ibid 12 (1978) 75. J. L. Walter, A. E. Berkowitz and E. F. Koch, Mater Sci. Eng., 60 (1983) 31. K. Muller and M. Von Heimendahl, J. Mater. Sci., 17 (1982) 2525. F. L. Cumbrera, H. Miranda, A. Conde, R. Marquez, and P. Vigier, J. Mater. Sci. 17 (1982) 2677. D. G. Morris, Scripta Met., 16 (1982) 585. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. J. L. Walter, S. F. Bartram, and I. Mella, Mater. Sci. Eng., 36 (1978) 193. J. C. Swartz, R. Kossowsky, J. J. Haugh, and R. F. Krause, J. Appl. Phys., 52(5) (1981) 3324. D. C. Tian, F. Fouquet, F. Guyot, C. Mai, and J. Perez, Mat. Sci. Eng., 53 (1982) 179. N. Saegusa and A. H. Morrish, Phy. Rev., B 26 (1982) 305. Idem, Phy. Rev., B 27 (1983) 4027. B. Bhanu Prasad, A. K. Bhatnager, and R. Jagannathan, J. Appl. Phys., 54(4) (1983) 2013. H. S. Chen, Appl. Phys. Lett., 29 (1976) 79. H. Rawson, in 'Inorganic Glass-Forming Systems', Academy Press, New York, (1967) R. W. Cahn, Fizika 2 Suppl. 2, papers 25 (1970) D. Turnbull and M. H. Cohen, in 'Modern Aspects of the Vitreous State', (ed. J. D. Mackenzie) Vol. 1, p 36, Buttersworth, London (1960). D. Turnbull, Contem. Phys., 10 (1969) 473. D. R. Uhlmann, J. Non-Cryst. Solids, 7 (1971) 337. H. A. Davies, Phys. Chem. Classes, 17 (1976) 159. P. I. K. Onorato and D. R. Uhlmann, J. Non-Cryst. Solids, 22 (1976) 367. W. Klement, R. Willens, and P. Duwez, Nature, 187 (1960) 869. J. Dixmier, K. D01, and A. Guinier, in 'Phys. of Non-Cryst. Solids', Proc. Int. Conf. p. 67, Interscience, New York (1964). G. S. Cargill, III, J. Appl. Phys., 41 (1970) 12. G. S. Cargill, III and R. W. Cochrane, J. Phys., Paris 35 (1974) CA 269. J. Logan, Phys. Stat. Solidi, A30 (1976) 73. T. Ichikawa, ibid A19 (1973) 707. S. Mader, A. S. Nowick, and H. Widmer, Acta Met., 15 (1967) 203. S. 1. Herd and P. Chaudhari, Phys. Stat. Solidi, A 26 (1974) 627. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. M. G. Scott, and R. Maddin, Ref. 38, p. 249. G. Marchal, P. Mangin, and C. Janot, Phil. Mag., 32 (1975) 1007. P. Mangin, G. Marchal, B. Rodmacq, and C. Janot, ibid 36 (1977) 643. S. Mader, in 'Recrystallization, Grain Growth and Texture', p. 523, Cleveland: ASM (1966). R. Roy, J. Non-Cryst. Solids, 3 (1970) 33. P. Pietrokowsky, Rev. Sci. Instrum., 34 (1963) 455. H. S. Chen and C. E. Miller, ibid 41 (1970) 1237. E. Babic, E. Girt, R. Krsnik, and B. Leontic, J. Phys., E. Sci. Instrum. 3 (1970) 1014. E. H. Strange and C. A. Pim, 1908 US Patent No. 905758. R. Pond and R. Maddin, Trans. Metall. Soc. AIME 245 (1967) 407. H. H. Liebermann and C. D. Graham, IEEE Trans. Magn., MAG-13, (1976) 921. H. S. Chen and C. E. Miller, Mater. Res. Bull., 11 (1975) 49. R. E. Maringer and C. E. Mobley, J. Vac. Sci. Tech., 11 (1974) 1067. P. H. Shingu, K. Shimomura, and R. Ozaki, Trans. Jpn. Inst. Met., 20 (1979) 33. C. F. Cline and R. W. Hopper, Scripta Met., 11 (1977) 1137. T. Yamaguchi and K. Narita, Appl. Phys. Lett., 35 (1978) 468. S. A. Miller and R. J. Murphy, Scripta Met., 13 (1979) 673. E. M. Breiman, B. H. Kear, and C. M. Banas, Phys. Today, Nov. p. 44 (1976). J. M. Poate, H. J. Leamy, T. T. Sheng, and G. K. Keller, Appl. Phys. Lett., 33 (1978) 918. R. Fastow, J. Gyulai, M. Nastasi, and P. G. Zielnski, J. Mater. Sci., 3 (1984) 109. H. Yoshioka, K. Asami, and K. Hashimoto, Scripta Met., 18 (1984) 1215. K. Asami, T. Sato, and K. Hashimoto, J. Non-Cryst. Solids, 68 (1984) 261. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. S. Yatsuya and T. B. Massalski, Mat. Sci. Eng., 54 (1982) 101. R. Becker, G. Sepold, and P. L. Ryder, Scripta Met., 14 (1980) 1283. C. -J. Lin, F. Spaepen, and D. Turnbull, J. Non-Cryst. Solids, 61 -62 (1984) 767. H. H. Liebermann, IEEE Trans. Magn., MAG-15, (1979) 1393. J. L. Walter, Ref. 12. p.30. H. Jones, Rep. Prog. Phys., 36 (1973) 1425. R. C. Ruhl, Mater. Sci. Eng., 1 (1967) 313. C. H. Bennett, J. Appl. Phys., 43 (1972) 2727. A. Rahaman, M. J. Mandel, and J. P. McTague, J. Chem. Phys., 64 (1976) 1564. H. A. Davies, J. Aucote, and J. B. Hull, Nature Phys. Sci., 246 (1973) 13. M. Hansen, in 'Constitution of Binary Alloys', McGraw Hill, New York (1958). R. Ray, B. C. Giessen, and N. J. Grant, Scripta Met., 2 (1968) 357. B. C. Giessen, J. Hong, L. Kabacoff, D. E. Polk, R. Raman, and S. Amand, Ref. 12. p. 249. R. Ray and E. Musso, US Patent No. 3981722 (1976). B. C. Giessen and R. O. Eliott, Ref. 12. p. 406. H. A. Davies and J. B. Hull, J. Mater. Sci., 9 (1974) 707. P. Ramachandrarao, M. Laridjani, R. W. Cahn, Z. Metall., 63 (1972) 43. P. Furrer and H. Warlimont, Mat. Sci. Eng., 28 ( 1977) 127. K. ChattOpadhyay, P. Ramachandrarao, S. Lele, and T. R. Anantha- raman, Ref. 38. p. 157. W. L. Johnson, S. J. Poon, and P. Duwez, Phy. Rev. B 11 (1975) 150. K. H. J. Buschow, N. M. Beekmans, Ref. 12. p. 133. K. H. J. Buschow, A. M. Van Diepen, N. M. Beekmans, and J. W. M. Biesterbos, Solid State Commun. 28 (1978) 181. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174. H. S. Chen, Acta Met., 22 (1974) 1505. M. Naka, Y. Nishi, and T. Masumoto, Ref. 12. p. 231. W. Hume —Rothery and E. Anderson, Phil. Mag., 5 (1960) 383. C. H. Bennett, D. E. Polk, and D. Turnbull, Acta Met., 19 (1971) 1295. G. S. Cargill, III, Solid State Phys., 30 (1975) 227. P. K. Srivastava, B. C. Giessen, and N. J. Grant, Met. Trans., 3 (1972) 977. P. Predecki, B. C. Giessen, and N. J. Grant, Trans. Metall., Soc. AIME 233 (1965) 1438. S. R. Nagel, and J. Tauc, Phys. Rev. Lett., 35 (1975) 380. G. A. Chadwick, in 'Liquids: Structure Prpperties Solid Interact- ions', p. 326, Elsevier, New York (1965). T. Masumoto and R. Maddin, Mater. Sci. Eng., 19 (1975) 1. N. F. Mott, and R. W. Gurney, Rep. Prog. Phys., 5 (1938) 46. N. Valenkov and E. Porai -Koshits, Z. Krist., 95 (1937) 195. J. D. Bernal, Nature, 185 (1960) 68. M. H. Cohan and D. Turnbull, ibid 203, (1964) 964. J. F. Sadoc, J. Dixmier, and A. Guinier, J. Non—Cryst. Solids, 12 (1973) 46. D. E. Polk, Acta Met., 26 (1972) 485. B. C. Giessen and C. N. J. Wagner, in 'Liquid Metals: Chemistry and Physics', (ed. S. V. Beer) p. 633, Dekker, New York (1972). J. L. Finney, Nature, 266 (1977) 309. Y. Waseda, J. Solid State Phys., 12 (1977) 181, P. Chaudhari and D. Turnbull, Science, 199 (1978) 11. M. Naka, K. Hashimoto, and T. Masumoto, J. Japan Inst. Met., 38 (1974) 835. Idem, Corrosion, 32 (1976) 146. K. Hashimoto, M. Naka, and T. Masumoto, Sci. Rep. Res. Inst. Tohoku Univ., A26 (1976) 48. 175. 176. 177. 178. 179. 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. M. Naka, K. Hashimoto, and T. Masumoto, J. Non-Cryst. Solids, 28 (1978) 403. T. Masumoto and K. Hashimoto, Ann. Rev. Mat. Sci., 8 (1978) 215. M. Naka, K. Hashimoto, and T. Masumoto, Sci. Rep. Res. Inst. Tohoku Univ., A26 (1977) 283. K. Hashimoto, K. Osada, T. Masumoto, and S. Shimomura, Science, 16 (1976) 71. M. Naka, K. Hashimoto, and T. Masumoto, J. Non-Cryst. Solids, 29 (1978) 61. K. Hashimoto, M. Naka, K. Asami, and T. Masumoto, Corrosion, 19 (1979) 165. T. Egami, P. J. Flanders, and C. D. Graham, AIP Conf. Proc., 24 (1974) 697. R. C. Sherwood, E. M. Gyorgy, H. S. Chen, S. D. Ferris, G.Norman, and H. J. Leamy, ibid 24 (1974) 745. H. J. Leamy, S. D. Ferris, G. Norman, D. C) Joy, R. C. Sherwood, E. M. Gyorgy, and H. S. Chen, Appl. Phys. Lett., 26 (1975) 259. E. M. Gyorgy, H. J. Leamy, R. C. Sherwood, and H. S. Chen, AIP Conf. Proc., 29 (1975) 198. E. M. Gyorgy, in 'Metallic Glasses', p. 275, Metals Park, OH: ASM (1978). C. D. Graham and T. Egami, Ann. Rev. Mat. Sci. 8 (1978) 423. S. Ohnuma, K. Watanabe, and T. Masumoto, Phys. Stat. Solidi, 44 (1977) K-151. H. Fujimori, T. Masumoto, Y. Chi, and M. Kikuchi, Japan J. Appl. Phys., 13 (1974) 1889. R. C. O'Handley, L. I. Mendelsohn, R. Hasegawa, R. Ray, and S. Kavesh, J. Appl. Phys., 47(10) (1976) 4660. T. Mizoguchi, K. Yamaguchi, and H. Miyajima, in 'Amorphous Magne- tism', p. 325. Plenum, New York (1973). H. S. Chen, R. C. Sherwood, and E. M. Gyorgy, IEEE Trans. Magn., MAG—13 (1977) 1538. N. S. Kazama, M. Mitera, and T. Masumoto, Ref. 12. 164. S. Hatta, T. Egami, and C. D. Graham, Jr., Appl. Phys. Lett., 34 (1979) 113. 194. 195. 196. 197. 198. 199. 200. 201. 202. 203. 204. 205. 206. 207. 208. 209 O 210. Ix) p—a Ix) F. E. Luborsky, J. J. Becker, J. L. Walter, and D. L. Martin, IEEE Trans. Magn., MAG—16 (1980) 521. M. Mitera, T. Masumoto, and N. S. Kazama, J. Appl. Phys., 50(11) (1979) 7609. R. C. O'Handley, C. -P. Chou, and Christofaro, ibid 50(5) (1979) 3603. F. E. Luborsky, J. J. Becker, J. L. Walter, H. H. Liebermann, IEEE Trans. Magn. MAG-15, 3(1979) 1146. S. Hatta, T. Egami, and C. D. Graham, Jr., ibid MAG-14, (1978) 1013. F. E. Luborsky and J.L. Walter, ibid MAG-16, (1980) 572. F. E. Luborsky, H. H. Liebermann, J. J. Becker, and J. L. Walter, Ref. 12. Vol. 2, p. 188. F. E. Luborsky, J. J. Becker, and H. H. Liebermann, Ref. 12. Vol. 2, p. 249. R. Hasegawa, J. Appl. Phys., 52(3) (1981) 1847. K. Inomata, M. Hasegawa, and K. Shimanuki, IEEE Trans. Magn., MAG -17, 6 (1981) 3076. M. Blundell, C. D. Graham, Jr., and K. J. Overshott, J. Magn. Magn. Matls., 19 (1980) 174. M. Blundell and K.J. Overshott, IEEE Trans. Magn., MAG-16, (1980) 1224. R. F. Krause and F. E. Warner, ibid 6 (1981) 2686. H. Fujimori, Y. Obi, T. Masumoto, and H. Saito, Mater. Sci. Eng., 23 (1976) 281. J. W. Shilling, AIP Conf. Proc., 29 (1976) 222. R. C. O'Handley and M. C. Narasimhan, IEEE Trans. Magn., MAG-15, (1979) 970. H. Fujimori, H. Yoshimoto, and H. Morita, ibid 5 (1980) 1227; H. Fujimori, H. Yoshimoto, T. Masumoto, and T. Mitera, J. Appl. Phys., 52(3) (1981) 1893. S. D. Washko, M. L. Osborn, and W. G. Veeraraghaven, J. Appl. Phys., 52(3) (1981) 1899. H. H. Liebermann, C. D. Graham, and P. J. Flanders, IEEE Trans. Magn. MAG-13, 5 (1977) 1541. 213. 214. 215. 216. 217. 218. 219. 220. 221. 222. 223. 224. 225. 226. 227. 229. 230. 231. 232. 233. F. E. Luborsky, J. L. Walter, and D. G. LeGrand, ibid 6 (1976) 930. M. R. J. Gibbs, J. E. Evetts, and N. J. Shah, J. Appl. Phys.,50 (11) (1979) 7642. K. Narita, J. Yamasaki, and H. Fukunaga, ibid 50(11) (1979) 7591. N. DeCristofaro, A. Freilich, and G. Fish, J. Mater. Sci., 17 (1982) 2365. C. D. Graham, and T. Egami, IEEE Trans. Magn. MAG-15, 6 (1979) 1398. H. J. Leamy, H. S. Chen, and T. T. Wang, Met. Trans., 3 (1972) 699. H. S. Chen, Scripta Met., 7 (1973) 931. C. A. Pampillo and H. S. Chen, Mater. Sci. Eng., 13 (1974) 181. C. A. Pampillo, J. Mater. Sci., 10 (1975) 1194. H. S. Chen and B. K. Park, Acta Met., 21 (1973) 395. H. S. Chen, J. T. Krause, and E. A. Sagety, J. Non—Cryst. Solids, 13 (1973/74) 321. R. Ray, B. C. Giessen, C.-P. Chou, and L. A. Davis, Scripta Met., 11 (1977) 973. B. Golding, B. G. Bagley, and F. S. L. Hsu, Phy. Rev. Lett., 29 (1972) 68. E. Orowan, Proc. lst National Congress of Appl. Mech., p. 453, New York: ASME (1952). D. Weaire, M. F. Ashby, J. Logan, and M. J. Weins, Acta Met., 19 (1971) 779. H. S. Chen, J. T. Krause, and E. Coleman, J. Non-Cryst. Solids., 18 (1975) 157. H. S. Chen, Mater. Sci. Eng., 26 (1976) 79. C.-P. Chou, Phy. Rev. Lett., 37 (1976) 1004. B. S. Berry and W. C. Pritchet, AIP Conf. Proc. 34 (1976) 292. H. S. Chen and J. T. Krause, Scripta Met., 11 (1977) 761. K. I. Arai, N. Tsuya, N. Yamada, H. Shirae, H. Fujimori, M. Saito and T. Masumoto, as in Ref. 38. p. 489. 234. 235. 236. 237. 238. 239. 240. 241. 242. 243. 244. 245. 246. 247. 248. 249. 250. 251. 252. 253. 254. H. A. Brooks, J. Appl. Phys., 47 (1976) 344. R. C. O'Handley, R. Hasegawa, R. Ray, and C.-P. Chou, Appl. Phys. Lett., 29 (1976) 330. B. S. Berry and W. C. Pritchet, ibid 34 (1975) 1022. B. S. Berry, in 'Metallic Glasses', p. 161, Metals Park, OH: ASM (1978). K. I. Arai and N. Tsuya, J. Appl. Phys., 49 (1978) 1918. C. A. Pampillo and D. E. Polk, Acta Met., 22 (1974) 493. Materials Research Report, Nov. 1975, Allied Chemical Corp. T. Masumoto and H. Kimura, J. Japan Inst. Metals. J. J. Gilman, in 'The Science of Hardness Testing & its Research Applications, p. 51, Metals Park, OH: ASM (1973). S. S. Brenner, J. Appl. Phys., 47 (1956) 344. T. Masumoto, Sci. Rep. Res. Inst. Tohoku Univ., A26 (1977) 246. H. S. Chen, Rep. Prog. Phys., 43 (1980) 353. L. A. Davis, Ref. 38. p. 369. T. 0gura, T. Masumoto, and K. Fukashima, Scripta Met., 9 (1975) 109. L. A. Davis, J. Mater. Sci., 11 (1976) 711. Idem, ibid 10 (1975) 1557. D. G. Ast and D. Krenitsky, Mater. Sci. Eng., 23 (1976) 241. H. Kimura and T. Masumoto, Scripta Met., 9 (1975) 211. H. S. Chen and D. Turnbull, Acta Met., 11 (1969) 104. T. Masumoto and M. Koiwa, J. Japan Inst. Metals. M. Ura, M. Kitaguchi, E. Fujita, and T. Masumoto, J. Japan Inst. Metals. H. S. Chen, Scripta Met., 9 (1975) 411. L. A. Davis, R. Ray, C.-P. Chou, and R. C. O'Handley, Scripta Met., 10 (1976) 541. P. E. Donovan and W. M. Stobbs, J. Non-Cryst. Solids, 55 (1983) 61. 258. 259. 260. 261. 262. 263. 264. 265. 266. 267. 268. 269. 270. 271. 272. 273. 274. 275. 276. 277. 278. 279. 280. N. A. Pratten and M. G. Scott, as in Ref. 12, p. 389. H. S. Chen and D. E. Polk, J. Non—Cryst. Solids, 15 (1974) 174. A. Inoue, T. Masumoto, S. Arakawa, and T. Iwadachi, as in Ref. 12 p. 265. A. Inoue, T. Masumoto, and H. Kimura, J. Japan Inst. Metals, 42 (1978) 303. G. C. Chi, H. S. Chen, and C. E. Miller, J. Appl. Phys., 49(1978) 1915. J. L. Walter, F. Bacon, and F. E. Luborsky, Mater. Sci. Eng., 24 (1976) 239. L. A. Davis and S. Kavesh, J. Mater. Sci., 10 (1975) 453. D. E. Polk and C. A. Pampillo, Scripta Met., 7 (1973) 1161. J. M. Vitek and N. J. Grant, Metall. Trans., 6 (1975) 1472. R. Hill, in 'Mathematical Theory of Plasticity', Oxford (1950). H. S. Chen and T. T. Wang, J. Appl. Phys., 41 (1970) 5338. . S. Argon, in 'The Inhomogeneity of Plastic Deformation', p. 61, ASM, Ohio (1973). L. A. Davis, Scripta Met., 9 (1975) 339. S. Takayama and R. Maddin, Metall. Trans., 7A (1976) 1065. S. Tomizawa and T. Masumoto, Sci. Rep. Inst. Tohoku Univ., See A, 26 (1977) 263. C. A. Pampillo, Scripta Met., 6 (1972) 915. A. H. Cottrell, Phil. Mag., 44 (1953) 829. A. I. Taub, Acta Met., 28 (1980) 633. C. A. Pampillo and A. C. Reimschussel, J. Mater. Sci., 9 (1974) 718. S. Takayama and R. Maddin, Acta Met., 23 (1975) 943. H. S. Chen, H. J. Leamy. and M. J. O'Brien, Scripta Met., 7(1973) 415. A. H. Cottrell, in 'The Mechanical Properties of Matter', p. 322, John Wiley, New York (1964). A. S. Argon, in 'Polymer Materials', p. 411, ASM, Ohio (1975). 281. 282. 283. 284. 285. 286. 287. 288. 291. 292. 293. 294. 295. 296. 300. 301. A. E. H. Love, in 'The Mathematical Theory of Elasticity', p. 221 Dover (1944). D. E. Polk and D. Turnbull, Acta Met., 20 (1972) 493. J. J. Gilman, J. Appl. Phys., 44 (1973) 675. J. C. M. Li, C. A. Pampillo, and L. A. Davis, in 'Deformation and Fracture of High Polymers', (eds. H. H. Kausch, J. A. Hassell and R. I. Jaffe, Plenum, New York (1974). A. L. Robinson, Research News; Glassy Metals: No Longer a Labora- tory Curiosity, Science, 182 (1973) 908. F. Spaepen and D. Turnbull, Scripta Met., 8 (1974) 563. D. G. Saffman and G. I. Taylor, Proc. Roy. Soc. London, A245 (1958) 312. L. A. Davis, in 'Metallic Glasses, p. 191, Metals Park: ASM, Ohio (1978). D. Lee and T. M. Devine, Ref. 38. p. 393. W. C. Leavengood and T. S. Vong, J. Appl. Phys., 31 (1960) 1416. J. C. M. Li, in 'Frontier in Materials Science', p. 527. Dekker, New York (1976). M. F. Ashby and J. Logan, Scripta Met., 7 (1973) 513. F. Spaepen, Acta Met., 25 (1977) 407. P. S. Steif, F. Spaepen, and J. W. Hutchinson, ibid 30 (1982) 447 A. S. Argon, ibid 27 (1979) 47. ) .. Becker, G. Sepold, and P. L. Ryder, 'Chemistry and Physics of Rapidly Solidified Materials', (ed. B. J. Berkwitz and R. O. Scattergood), TMS-AIME Warrendale (1983). G. A. Jones, S. F. M. Parker, P. J. Grundy, K. J. Overshott, and M. H. Price, IEEE Trans. Magn. MAG-19, 5 (1983) 1922. J. A. Leake and A. L. Greer, J. Non-Cryst. Solids, 38-39 (1980) 7, J. L. Walter, J. D. Livingstone, and A. M. Davis, Mater. Sci. Eng., 49 (1981) 47. H. N. 0k and A. H. Morrish, Phy. lev. B 23 (1981) 2257. A. L. Greer and J. A. Leake, Ref. 12. p. 299. 302. 303. 304. 305. 306. 307. 308. 309. 310. 311. L. A. Davis, V. DeCristofaro, and C. H. Smith, in Proc. of the Conf. on Metallic Glasses, p. 1, Science and Technology, Budapest H. N. 0k and A. H. Morrish, J. Phys. F, Metal Phys., 11 (1981) 1495. P. Duwez, Ann. Rev. Mater. Sci., 6 (1976) 83. S. Takayama, J. Mater. Sci., 11 (1976) 164. R. M. White, J. Appl. Phys., 34 (1964) 2123. P. E. Donovan and W. 1. Stobbs, Acta Met., 31 (1983) 1. J. L. Walter and F. E. Luborsky, Mater. Sci. Eng., 33 (1981) 91. J. K. Knowles and E. Sternburg, J. Elasticity, 10(1) (1980) 81. Idem, ibid, 11(2) (1981) 129. R. Abeyaratne, ibid 11(4) (1981) 373.