EFFECT OF DESIGN FACTORS ON THERMAL FATIGUE CRACKING OF DIE CASTING DIES. Final Technical Report. David Schwam John F. Wallace Sebastian Birceanu

Transcription

1 EFFECT OF DESIGN FACTORS ON THERMAL FATIGUE CRACKING OF DIE CASTING DIES Final Technical Report David Schwam John F. Wallace Sebastian Birceanu Department of Materials Science Case Western Reserve University Cleveland, Ohio Work Performed Under Contract DE-FC07-00ID US Department of Energy Assistant Secretary for Energy Efficiency and Renewable Energy Washington DC October 2004

2 TABLE OF CONTENTS Content Page TABLE OF CONTENTS 2 LIST OF TABLES 4 LIST OF FIGURES 5 ACKNOWLEDGEMENTS 8 ABSTRACT 9 1. Introduction Die Failure Modes Thermal Fatigue Failure Mechanisms Thermal Shock and Thermal Fatigue Resistance Evaluation Factors Thermal and Physical Properties that Affect Thermal Fatigue Resistance Thermal Conductivity Thermal Expansion Coefficient Mechanical Properties that Affect Thermal Fatigue Resistance Elastic Modulus and Strength The Effect of Thermal Cycling on Microstructural Stability The Temperature-Time Effect on the Structure of Martensitic Steel 20 2

3 Content Page Martensitic Transformation in Steel - Brief Overview Tempering of Martensite Materials and Experimental Procedures Materials The Thermal Fatigue Test Specimens and Equipment Thermal Fatigue Cracks Evaluation Procedure Temperature Measurement Microhardness Measurement Scanning Electron Microscopy Results and Discussion Softening During Thermal Cycling and Thermal Fatigue Resistance The Influence of Immersion Time on Softening and Thermal Fatigue dsadasdascracking The Influence of Cooling Line Diameter on Softening and Thermal sadadadadfatigue Cracking Stress Analysis at the Specimen Surface and Around the Cooling Line Microstructure Degradation that Promotes Softening During Thermal sasdsdcycling 4. Conclusions Bibliography 99 3

4 LIST OF TABLES Table Page 2.1. Chemical Composition of Experimental Material - Premium Grade H Typical Properties of Premium Grade H Characteristics of the Tested Specimens Measurement Data For Different Immersion Times Immersion Time Effect on Hardness Variation Across the Surface Measurement Data For Different Cooling Line Diameters Cooling Line Diameter Effect on Hardness Variation Across the Surface 50 4

5 LIST OF FIGURES Figures Page 1.1. Hysteresis Loop at the Surface of a Material Subjected to Cyclic Heating aaaaand Cooling CCT Diagram for H13 steel The Reference Specimen for Thermal Fatigue Test The Thermal Fatigue Test Equipment Temperature Measurement Relationship between Tensile Properties and Hardness The Effect of Thermal Cycling on Crack Area-Different Immersion Times The Effect of Thermal Cycling on Crack Length- Different Immersion Times The Effect of Thermal Cycling on Microhardness Distribution Across the aaaasurface- Different Immersion Times The Effect of Temperature on Crack Area-Different Immersion Times The Effect of Temperature on Crack Length-Different Immersion Times Effect of Elevated Temperature on Tensile Strength The Effect of Hardness Recovery on Thermal Fatigue Cracking Relationship Between Total Crack Area and Average Maximum Crack aaaalength The Relationship Between Maximum Crack Length and Microhardness at aaaaamaximum Crack Length The Effect of Temperature on Microhardness-Different Immersion Times The Effect of Microhardness at Average Maximum Crack Length on Crack aaaaaarea 74 5

6 Figures Page The Effect of Microhardness at Average Maximum Crack Length on Crack aaaaalength-different Immersion Times Microhardness Profile at the Corner of 12 Seconds Immersed Specimen Tempering Curve for H Maximum Temperature Cycle for 1.5" Cooling Line Specimen After 12 aaaaaseconds Immersion Time The Effect of Thermal Cycling on Crack Area-Different Cooling Line aaaaadiameters The Effect of Thermal Cycling on Crack Length-Different Cooling Line aaaaadiameters The Effect of Thermal Cycling on Microhardness Distribution Across the aaaaasurface-different Cooling Line Diameters The Effect of Temperature on Crack Area-Different Cooling Line aaaaadiameters The Effect of Temperature on Crack Length-Different Cooling Line aaaaadiameters The Effect of Temperature on Microhardness-Different Cooling Line aaaaadiameters The Effect of Microhardness at Average Maximum Crack Length on Crack aaaaaarea-different Cooling Line Diameters The Effect of Microhardness at Average Maximum Crack Length on Crack aaaaalength-different Cooling Line Diameters The Effect of Immersion Time on Temperature The Effect of Cooling Line Diameter on Temperature Cracks at the Corner of H13 Specimen Crack at the Cooling Line of H13 Specimen Stress Modeling at the Corner and Cooling Line 91 6

7 Figures Effect of Volume Percent Primary Carbides on the Transverse Charpy V- notch Impact Toughness of H13 Page Microstructure Sampling at the Corner of 12 Seconds Immersion Specimen a. The Effect of Temperature on Microstructure Unaffected b. The Effect of Temperature on Microstructure 0.2 from Corner c. The Effect of Temperature on Microstructure 0.1 from Corner d. The Effect of Temperature on Microstructure 0.06 from Corner e. The Effect of Temperature on Microstructure Corner Temperature Influence on Carbide Size and Distribution-Photomontage Effect of Austenitizing Temperature on the Weight Percentage of Isolated aaaaacarbide Residues in H13 Steel Small Carbide in Softened H Large Carbide in Softened H

8 ACKNOWLEDGEMENTS This research investigation was supported by the Department of Energy, Office of Industrial Technology through the Cast Metal Coalition program. The Die Materials Committee of the North American Die Casting Association provided guidance for this work. The efforts of Mr. Steve Udvardy, Director of Research and Education at NADCA and the members of the committee are gratefully acknowledged. This publication was prepared with the support of the U.S. Department of Energy (DOE), Award No. DE-FC07-00ID However, any opinions, findings, conclusions or recommendations expressed herein are those of the authors and do not necessarily reflect the views of the DOE. 8

9 ABSTRACT The thermal fatigue of steel die casting dies becomes more severe at higher operating service temperatures, reducing die life significantly. Consequently, to extend die life, die design has to address efficient cooling methods. A key issue in this respect is the size and location of cooling lines relative to the surface of the die. This subject was studied in detail, to elucidate the effect of die temperature on thermal fatigue cracking. The investigation correlates the thermal fatigue cracking in an immersion test specimen with the temperature attained near the surface and the corresponding softening of the steel. The effect of cooling line location vis-à-vis the surface temperature and the resulting cracking pattern are shown for various immersion times and different sizes of cooling lines. Higher temperatures induce faster and deeper softening of the steel leading to more thermal fatigue damage. Die design with cooling lines close to the surface can reduce this damage significantly. Since the thermal fatigue test has previously provided a remarkably accurate prediction of the relative thermal fatigue cracking, these results should have good applicability in die casting operations. 9

10 1. INTRODUCTION 1.1. Die Failure Modes The durability of materials in molten aluminum is an important consideration in engineering applications such as die casting, containment of liquid metal and semi-solid processing [1]. Die casting is the process of choice in many manufacturing industries - automotive, hardware, electrical and electronics, computers and many others. It provides high volume and cost effective aluminum, zinc and magnesium components with good properties. Some of the advantages of this technology over the traditional sand castings are [2, 3]: - Die casting is able to provide complex shapes within closer tolerances; - Higher rates of production with little or no machining required; - The die casting parts are durable, dimensionally stable, and have a good appearance; - Die castings are monolithic; they combine many functions in one, complex shaped part; The main failure modes of aluminum die casting dies are physical erosion (washing), chemical attack (corrosion), gross cracking (cleavage cracking) and thermal fatigue cracking (heat checking) [2,3,4,5]. Erosion occurs when a swift flow of melt exists relative to the surface of the solid, and becomes more severe when there are hard particles in the melt. Chemical corrosion refers to dissolution of materials by the melt as well as the formation of interphase layers, when the relative motion between the solid material and the melt is negligible [1]. 10

11 Gross cracking is usually catastrophic and may result in complete cracking through the die. Massive fracture of die casting dies occurs when the die material is stressed beyond its fracture strength. This can occur even when the applied stress is below the yield stress. It results from a combination of thermal and mechanical stresses [6]. This type of failure is related to the inherent resistance of the die material to fracture termed fracture toughness. 1.2 Thermal Fatigue Failure Mechanisms The life of dies used at elevated temperatures is often determined by their thermal fatigue properties [7]. The fatigue failure produced by fluctuating thermal stresses is known as thermal fatigue. Thermal stresses occur when the expansion or contraction of a part as a result of a temperature change is restrained [8]. The constraint may be internal or external [9]. External constraints produce forces that act on a component that is alternately heated and cooled. Internal constraints may result from temperature gradients across the section (simply because heat is not able to flow quickly enough in response to the external changes), structural anisotropy and different coefficients of expansion in adjacent phases or grains [10]. Thermal fatigue resulting from the action of internal constraints can also be defined as thermal cycling damage. Temperature gradients form as a result of heating and cooling of the surface during injecting the molten metal, ejection and the lubricant spraying stages of the die casting cycle [4, 11]. When molten aluminum is injected, the die surface heats up creating a steep temperature gradient between the surface and the cooler underlying mass 11

12 of the die. As a result, the surface wants to expand more than the interior. Because the interior is more massive, it prevents the surface layer from expanding. As a result of this internal restraint, the surface is placed under compression. As heat is conducted into the underlying layers, the surface temperature decreases rapidly. When the casting is ejected, the surface of the die cools down. The spray of die lubricant further decreases the surface temperature [12]. The surface then cools more rapidly than the interior, the compression stresses are relieved and tensile stresses may be created. The temperature gradient and the coefficient of the thermal expansion of the material determine the magnitude of the surface stress. For the simple case of a bar with fixed end supports, the thermal stress generated by a temperature change T is: σ = αe T (1-1) where α is the linear thermal coefficient of expansion and E is the elastic modulus [8]. For a biaxial condition, the stress is given by: σ=αe T /(1-ν) (1-2) where ν is Poisson s ratio. The thermal stresses never fully develop because α, E, ν and the thermal conductivity all change with temperature [13]. The fatigue damage of metallic materials can be subdivided into the stages of crack initiation, subcritical crack propagation, and final unstable fracture [14]. Crack initiation usually occurs at stress concentration sites originating from component geometry, machining irregularities or surface imperfections [15]. During the compressive part of the cycle, the increase in temperature lowers the yield strength of material, and the 12

13 compressive strain may become plastic when substrate prevents deformation. During the tension part of the cycle, the concentrated thermal stress is larger than the yield strength of the material, and reversed plastic deformation may occur. After sufficient number of cycles, the localized plastic deformation will cause a fatigue crack. Once a crack is initiated, propagation occurs along a plane perpendicular to the maximum tensile stress. Fatigue cracks in steels can deviate noticeably from the expected plane of propagation when large prior austenite grain boundaries are present or when another crack is in close proximity [15]. The influence of other cracks on crack propagation is explained by the interaction of the highly stressed regions at the tips of the cracks. When the crack tips are close enough, this interaction changes the general state of stress. This provides an attraction of cracks to each other until the cracks are joined by reversed crack branching or forking. When the side branches join, they produce the classical craze-cracking pattern [16]. Several opinions exist about the driving force for crack propagation. One opinion is that cracks widen and deepen by the wedging action of the molten metal or oxide that is forced into them in each shot [17]. Another is that crack propagation occurs only by the thermal fatigue mechanism. In this event, propagation has to occur during the cooling cycle, since that provides the tensile stress necessary for crack propagation. Once a fatigue crack is formed, it will continue to grow because of the stress concentration effect even when the tensile stress is low [18]. The damage caused by thermal cycling can be separated into stages [14]: - crack initiation at the surface; - crack linking at the surface; 13

14 - growing of small cracks in depth direction from the crack net; - growing of the largest crack to the complete failure. The generation and evolution of thermal stress have been explained based on the type of thermal transients that occur during the service of a part subjected to thermal shock-thermal fatigue [10]. Suppose a thick structure at low temperature is suddenly brought in contact with a hot fluid. As explained before, the surface tends to expand against the remaining material and goes into compression, yielding along OQ (Figure 1.1). Because of the heat transfer towards the core, the temperature gradient decreases and the system expands, taking the surface into tension at R. The residual tension is responsible for intergranular cracking. When the material is further subjected to rapid cooling this series of events is reversed. The surface now goes into tension, as it tends to contract, with a tension peak at point S that promotes transgranular cracking since the corresponding strain rate is high and the temperature decreases. Later, when the whole structure cools, the yielded section at A goes into compression. The compression stress at P (at the original strain level) is not particularly damaging, but when many cycles are repeated, the system goes into a hysteresis loop at the surface (PQRS), leading to thermal fatigue cracking due to the reversed deformation Thermal Shock and Thermal Fatigue Resistance Evaluation Factors A common measure of thermal shock resistance is the maximum sudden increase in the surface temperature that a material can sustain without cracking [19]. The thermal shock resistance and thermal fatigue resistance depend on a number of material 14

15 properties. These include the thermal expansion coefficient α, thermal conductivity k, thermal diffusivity K, elastic modulus E, fracture toughness K Ic, tensile (fracture) strength σ f and upon the additional parameters of heat transfer coefficient h, specimen size and duration of thermal shock [4,19,20,21]. Figure 1.1. Hysteresis Loop at the Surface of a Material Subjected to Cyclic Heating and Cooling [10]. 15

16 A commonly used thermal shock - thermal fatigue resistance parameter is the merit index of R=σ f /Eα or R'=kσ f /Eα [4,19,21]. For large values of Biot number (β = bh/k), i.e. large heat transfer coefficient h, radius or thickness r or b, respectively, and small thermal conductivity k, or when thermal strains are the result of the material being mechanically constrained, the thermal shock resistance is determined by R. For very small Biot numbers, i.e. small heat-transfer coefficients, small radius or thickness, and large conductivity, or when thermal strains are a consequence of thermal gradients resulting from rapid heating or cooling, the thermal shock resistance is determined by R'. If we consider the thermal fatigue as a series of repeated thermal shocks, these parameters can be used to describe the thermal fatigue resistance and for ranking of materials. In this respect, the effect of elements incorporated into the chemistry of an alloy should be considered based on their contribution to [22]: a) Thermal properties of the material: coefficient of thermal expansion, specific heat, and thermal conductivity b) Material strengthening through carbide formation, solid solution strengthening with consequent increase in the capacity for withstanding repeated strains and with improved creep performance. c) Microstructural stability and oxidation resistance 1.4. Thermal and Physical Properties that Affect Thermal Fatigue Resistance Thermal Conductivity 16

17 The thermal conductivity is the quantity of heat transmitted, due to unit temperature gradient, in unit time under steady conditions in the direction of the temperature gradient. This condition occurs when the heat transfer is dependent only on the temperature gradient [23]. Under the conditions described above, thermal conductivity will reduce thermal fatigue by establishing a low thermal gradient between the surface and the underlying layer. Equations (1-1) and (1-2) indicate that a lower temperature gradient will decrease the stress in the material. The successful use of a molybdenum insert in pressure die casting dies partly results from the high values of thermal conductivity. However, because of its variation with temperature, the influence of this parameter may be diminished by the operating conditions. For instance, ferritic steels have generally higher values of thermal conductivity than austenitic steels, but at high temperature, say at 1073 K, their thermal conductivities become similar [22] Thermal Expansion Coefficient The coefficient of linear thermal expansion is the ratio of the change in length per degree K to the length at 273 K. The coefficient of volume expansion is about three times the linear coefficient [23]. The combination of the temperature gradient and the coefficient of thermal expansion determine the magnitude of stress, as shown by the equation (1-1). Indeed, the amount of expansion in the axial direction of a slit from a blade-divided subjected to a temperature T x will be, according to Duhamel's analogy [20]: ε = αt x (1-3) 17

18 and the compressive stress induced by bringing the slit back at its initial dimension, will be: σ = -EαT x (1-4). Among metals, refractory metals have the lowest thermal expansion coefficients [22] Mechanical Properties that Affect Thermal Fatigue Resistance Elastic Modulus and Strength The elastic modulus is a measure of the stiffness of the material. It is defined as the ratio of the stress and strain in the elastic regime: E = σ/ε (1-5) A lower modulus results in lower stress at a given strain level. Some metallic, but especially structural ceramic materials are susceptible to failure when thermally shocked due to a high Young's modulus, combined with relatively high thermal expansion coefficient, low strength and low thermal conductivity [22]. In general, a material with a low Young's modulus and a high value of yield strength is desirable, as the elastic component of the strain is large and the plastic component is small during a typical thermal cycle. The best combination of properties is 18

19 a high strength-high ductility (high toughness) material, but unfortunately a high strength is often associated with a low value of ductility. A very important issue related to thermal fatigue resistance of materials is the hot hardness and the variation of strength with the temperature. At high temperatures, the surface loses strength and hardness, especially in steels. This fact will lower the thermal fatigue resistance. The parameters R and R' will therefore change their values with the temperature not much because of the variation of the Eα product, which is roughly constant with increasing temperature [10], but due to the sudden drop in strength at a certain temperature (which in the case of R'), may not be compensated by the rise in thermal conductivity The Effect of Thermal Cycling on Microstructural Stability The prolonged exposure to elevated, varying temperatures and repeated stresses as it takes place in thermal cycling naturally causes changes in the microstructure [22]. These changes in the metallurgical structure may contribute to failure by reducing strength and they are referred to as instabilities. Sources of instabilities include transgranular-intergranular fracture transition, recrystallization, aging or overaging, phase precipitation or decomposition of carbides. Borides or nitrides, intermetallic phase precipitation, delayed transformation to equilibrium phase, order-disorder transition, general oxidation, intergranular corrosion, stress-corrosion cracking, slag-enhanced corrosion, and contamination by some trace elements also cause instabilities [5]. 19

20 1.7. The Temperature-Time Effect on the Structure of Martensitic Steel Martensitic Transformation in Steel - Brief Overview The rapid cooling of a plain-carbon eutectoid steel, after it was heated in the austenite region, in such a manner that it misses the nose of the TTT diagram curve, will lead to the formation of the martensite [31]. The conversion of an austenitic microstructure to a martensitic microstructure in many steels takes place continuously with decreasing temperature during uninterrupted cooling. This is a unique characteristic of the transformation kinetics of martensite and is referred to as athermal transformation [24]. As a general definition, a martensitic transformation occurs by nucleation and growth and involves the coherent formation of a phase from another without a change in composition, by a diffusionless and homogeneous lattice shear [32]. Martensite in steels is a metastable body-centered tetragonal (or body-centered cubic, below 0.2% C) supersaturated solid solution of carbon and other alloying elements in Fe-α, in which the alloying elements remain locked into the position they occupied in the parent austenite [31,33]. types: From a morphological point of view, Fe-C martensites can be classified into two - lath martensite, typical of all low and medium carbon with up to 0.6 %C; - plate martensite, above about 1.0% C; its formation was found to be favored by austenite stabilizers, such as N, Ni, Pt or Mn, but prevented by ferrite stabilizers like Si, 20

21 Cr, W, V, and Mo. Between 0.6 and 1 % C, a mixture of lath and plate martensite occurs [31,34]. Another important issue for the material properties past transformation is the grain size of the parent phase, austenite. The austenitic grain size will not affect the number of martensite nuclei in a certain volume, but the plate size is a function of the grain size. In larger grain size material the strain associated with the transformation can cause large residual stresses to build between adjacent grains. This can eventually lead to grain boundary rupture. Fine grains will not be that susceptible to this phenomenon, due to self-accommodation, and together with a smaller martensitic plate size, will provide for a stronger and tougher material [33]. The general trends related to the austenitic grain size in heat treated products are: - Hardenability - deeper hardening for coarse-grain austenite, and shallower hardening for fine-grain; the addition of alloying elements, except for cobalt, will minimize the difference, because of the increase of hardenability and inhibition of the grain growth [40]. - Toughness - higher for small-grained material - Distortion, quench cracking, internal stress - less present or prevalent in fine- grained structure [39]. An important observation is that increasing the austenitizing temperature will produce an improvement in the thermal fatigue performance as a result of the higher tempering resistance [40], despite a larger grain size. Large grains were proved to be 21

22 detrimental to thermal fatigue resistance [42,43]. This effect is probably the result of the more effective dissolution of alloy carbides and the consequent increase of alloying elements in solid solution Tempering of Martensite The martensitic transformation is essential for the hardening of steel and induces a desirable hardness. It also increases brittleness, which results from factors such as lattice distortion caused by carbon atoms trapped in the octahedral sites, impurity atom segregation at austenite grain boundaries, carbide formation during quenching, and residual stresses. The hardness of martensite will increase with carbon content and/or alloying elements. In order to improve ductility and toughness (and sometimes even strength), most of the technological steels must be tempered. During the heating for the tempering process, a number of solid-state reactions may occur [24,30,33,34,35,37,38]: C ( F) Carbon segregation to dislocations and boundaries or preprecipitation clustering (in high-carbon steels), caused by the interaction energy created between carbon and strain field around dislocations. In low carbon-steel M s temperature is higher and can be sufficient time for carbon to segregate or even precipitate as ε carbide or cementite during quenching C ( F)- First stage of tempering - Precipitation of transition carbides - η(fe 2 C) or ε(fe 2.4 C) - in steels with carbon content above 0.2 %. The phenomenon is accompanied by a slight increase in hardness. Below 0.2 %C, the atoms 22

23 prefer to diffuse at the boundaries or dislocation sites during cooling. Consequently, not much carbon is left in solution to precipitate upon reheating C ( F)- Second stage of tempering - Decomposition of austenite retained after quenching especially in low-alloy steels with more than 0.4% C, into ferritic bainite and carbides. It is associated with tempered martensite embrittlement, since carbides replace the austenite in the spaces between the laths of martensite C ( F) - Beginning of the third stage of tempering - Lath-like orthorombic Fe 3 C precipitation C ( F) - Segregation of impurity and alloying elements, which is responsible for temper embrittlement. The temper embrittlement has been attributed to the segregation of impurity atoms such as P, Sb, As or Sn to prior austenite grain boundaries C ( F) - Recovery of dislocation structure; Lath-like Fe 3 C agglomerates to form spheroidal Fe 3 C, but the lath structure is maintained. During recovery, the cell boundaries and random dislocations contained between them are annihilated and a fine grain acicular structure is developed C ( F) - Formation of alloy carbides, also called the fourth stage of tempering. Occurs in steels containing sufficient carbide forming elements (Ti, Cr, Mo, V, Nb or W). Above about 500 C, substitutional diffusion becomes significant and alloy carbides replace the less stable cementite which dissolves as a finer alloy carbide dispersion forms. Two ways exist in which cementite-alloy carbide transformation can take place: 23

24 - in situ transformation - the alloy carbide nucleates at several points at the cementite/ferrite interfaces, and grow until cementite disappears and is replaced by a alloy carbide dispersion - by separate nucleation and growth - the alloy carbides nucleate heterogeneously within the ferrite on dislocations, lath boundaries, and prior austenite grain boundaries. The carbides then grow at the expense of cementite. The stable carbide forming elements like V and Mo are hence the promoters of the strengthening reaction that occurs in the temperature range from 500 to 600 C. This is known as secondary hardening, induced by the replacement of the coarse cementite by the finer alloy carbide, as described above C ( F) - Recrystallization and grain growth occur. The ferrite can recrystallize more readily in low rather than high-carbon steels, because the grain boundary pinning caused by carbide precipitates inhibits the process. After recrystallization is complete, growths of carbide particles and of ferrite grains are the only kinetic processes that continue. One of the major concerns in die steel selection is the softening that occurs due to the thermal cycle. Steels for aluminum die casting experience a high temperature that could reach 1200 F during the casting thermal cycling [4]. It has been shown in previous studies [4,11] that the thermal fatigue behavior is better for temper resistant steels. Alloying elements that help retard the rate of softening during tempering are desirable. 24

25 The most effective elements in this regard are strong carbide formers such as chromium, molybdenum and vanadium [24]. The decrease in hardness and strength of carbon steels during tempering is largely due to the coarsening of Fe 3 C with increasing temperature. Under these conditions an element with a greater affinity for carbon like those mentioned would form alloy carbide with high resistance to coarsening and therefore provide hardness retention, good creep and thermal fatigue resistance. The favorable influence of these alloying elements can turn into a deleterious one, when present in steels in too high of a quantity. Excess alloying elements produces large carbide particles on the grain boundaries in the quenched and tempered steel and increase the brittleness of the steel, resulting in gross cracking. A high austenitizing temperature can dissolve the carbides in the solid solution, but too high of a temperature will lead to a grain coarsening with same detrimental results. 25

26 2. MATERIALS AND EXPERIMENTAL PROCEDURES 2.1. Materials The material chosen for this work was the Premium Grade H13 steel, since this is the preferred die steel for the aluminum die casting industry. The composition of the steel is given in Table 2.1. H13 is a chromium hot work steel. It is basically a hypoeutectoid steel with high hardenability and a good combination of strength, hot hardness, toughness and ductility. It has good resistance to tempering. Some typical physical and mechanical properties of H13 are presented in Table 2.2. This steel has limited amount of alloy segregation, a fine grain size and a structure that has a low inclusion content and low concentration of sulfur and phosphorus. The following heat treatment procedure was chosen in order to obtain the strength and toughness combination required by the aluminum die-casting industry. The specimens were austenitized at 1875 F, oil-quenched according to the schematic CCT diagram in Figure 2.1, and then double tempered at 1100 F for 2 hours. Such a procedure with double tempering will tend to eliminate the residual austenite, and lead to a predominantly tempered martensitic structure with a hardness of Rc, high strength and good toughness. 26

27 2.2. THE THERMAL FATIGUE TEST Specimens and Equipment Specimens for the thermal fatigue test were processed to the dimensions shown in Figure 2.2. The reference specimen is 2 x2 x7, rectangular in shape with a 1.5 diameter and 6.5 long hole in the center for internal water-cooling. Three other specimens were designed with 1.6", 1.7", and 1.8" cooling line diameters. The four corners of the specimens were designed and fabricated with a radius of and the specimens' surface was hand polished with 240, 320, and 400 grit silicon carbide paper. The thermal fatigue test equipment is shown in Figure 2.3. The specimens were alternately cycled (dunked) in a molten aluminum alloy (380 grade) bath, which was maintained at 1350 F. A pneumatic system consisting of an cylinder automatically actuated was used to immerse and withdraw the specimens from the aluminum bath at different cycle durations consisting of 5, 7, 9 (reference) and 12 seconds immersion and 24 seconds withdrawn. Water flowed through the specimens at a rate of 1.5 gal/min through the internal cooling line shown in Figure 2.2. The outer surface of the specimen was sprayed with water just before it entered the molten aluminum bath. The specimens were turned 90 around their long axis every 1,500 cycles to insure the uniform spraying of the water. Table 2.3 summarizes the specimens used and their particular characteristics. 27

28 Thermal Fatigue Cracks Evaluation Procedure Specimens were removed from the test system after 5,000, 10,000 and 15,000 cycles and their cracks were measured. Since the temperature fluctuations and geometrical constraints are the greatest at the corners, cracks form mainly at the corners. For measuring the cracks, the surface of the specimens is polished with 240, 320 and 400-grit silicon carbide paper. A V-shaped fixture with 400-grit silicon carbide paper is used to polish the corner. Only cracks on the corners within a 3 central length were measured, to eliminate the end effect of the top and bottom areas. Two concepts are used to evaluate the thermal fatigue resistance of the steels, Average Maximum Crack Length and Total Crack Area [4]. The Average Maximum Crack Length L a is the average length of the longest cracks on the four corners, within the middle three inches of the corners. 4 1 L a = L mi 4 where i = indicates each of the four corners, and L mi is the maximum crack length of i corner. i= 1 The crack area of each crack is defined as the square of the crack length. The Total Crack Area is the sum of the products of the number of cracks in each 100 micron size range and the square of the midpoint of that range for all the four corners. 28

29 A t = 4 n i= 1 j= 1 N i, j L 2 j where L j = 100j-50 µm j = 1...n, corresponds to different crack length range and N i,j represents the number of the cracks of i corner in the crack length range of 100(j-1) to 100j µm. The number and length of all cracks were measured under an optical microscope attached to a Leitz microhardness tester Temperature Measurement In order to determine the temperature of the corner, a thermocouple hole was drilled in a specimen with an initial cooling line diameter of 1.5 (Figure 2.4). The drilling was performed at an angle from the vertical in order to reach as close as possible to the corner, at the middle of the specimen. The distance of the thermocouple junction from the corner was estimated at about After inserting the thermocouple, it had to be fixed in place in order to minimize the errors given by the eventual displacement of the tip from the center bottom of the hole. The temperature values for different immersion times were then recorded on a computer. After the first set of measurement on the 1.5 diameter, the cooling line diameter was increased by machining to 1.6, and subsequently to 1.7 and 1.8. This procedure ensures excellent relativity, since the thermocouple and its location were constant. 29

30 2.2.4 Microhardness Measurement The microhardness of the specimens was measured before testing and after testing at 5,000, 10,000, 15,000 cycles. A Buehler Micromet 2100 Microhardness Tester was used to obtain a profile distribution of hardness from the specimen corner to the center at both the surface and inside the specimen. A Vickers indenter was used, with a 500 g indentation load. The Vickers hardness was converted to Rockwell C scale directly by the tester's scale converter. The microhardness was taken at the middle of the specimen, starting from the corner towards the center. The first measurement was made at 0.01" from the edge, then at 0.02", 0.04" and so on until no further variation in hardness was obtained. A supplemental set of measurements were performed on the cross section of the 12 seconds immersion time specimen, as seen in the hardness distribution chart, Figure The 12 seconds specimen was chosen due to the severe conditions that it has been subjected compared to the other specimens SCANNING ELECTRON MICROSCOPY A Hitachi S-4500 Scanning Electron Microscope (SEM) was used to study the microstructure of the materials. The specimens were polished and then etched in 2 % Nital solution. The attached Energy Dispersive Spectrometer (EDS) was used to determine the composition of carbides. 30

31 Element C Si Mn Cr Mo V Ni P S Fe Weight % <0.001 bal Typical Composition of AISI/SAE H max max max bal TABLE 2.1. Chemical Composition of Experimental Material - Premium Grade H13 31

32 Density lb/in 3 (g/cm 3 ) Coefficient of Thermal Expansion, linear µin/in. F (µm/m. C) Thermal Conductivity BTU.in/ft 2.h.F (W/m.K) Elastic Modulus ksi (GPa) (7.8) 6.11 (11) C 169 (24.3) 215 C 6.39 (11.5) C (24.4) 350 C 6.89 (12.4) C (24.7) 605 C 30,500 (210) 22 Tempering Temperature F ( C) Tensile Strength ksi (MPa) Yield Strength ksi (MPa) Reduction in Area % Hardness Rockwell C Impact Energy ft-lbf (J) 980 (525) 284 (1960) 228 (1570) (16) 1120 (605) 217 (1495) 187 (1290) (30) TABLE 2.2. Typical Properties of Premium Grade H13 [44,45] 32

33 Specimen Cooling Line Diameter Immersion Time A (*) 1.5" 9 sec B C D E F G 1.5" 5 sec 1.5" 7 sec 1.5" 12 sec 1.6" 9 sec 1.7" 9 sec 1.8" 9 sec 23 (*) Reference Specimen TABLE 2.3. Characteristics of the Tested Specimens 33

34

35 A cooling - oil quench, martensite, no carbides B cooling - air cooling, martensite+carbides FIGURE 2.1. CCT Diagram for H13 steel [4] 35

36 (* ) FIGURE 2.2. The Reference Specimen for Thermal Fatigue Test (*) Three other specimens had the cooling line diameter 1.6", 1.7" and 1.8" respectively 36

37 FIGURE 2.3 The Thermal Fatigue Test Equipment 37

38 Φ0.08 FIGURE 2.4 Temperature Measurement 38

39 3. RESULTS AND DISCUSSION 3.1. Softening During Thermal Cycling and Thermal Fatigue Resistance During the aluminum die casting process, some parts of the die are subjected to very severe conditions of temperature and consequently, stress. Generally, these are thin sections, fingers and corners, where the heat transfer is two-dimensional and the amount of energy that the material must absorb is much higher than the average for the rest of the die. Often these are the sections that fail first. It becomes critical to create conditions for rapid heat extraction from the surface, dissipation inside the material, or transfer towards a heat conveyor such as a cooling line. The main mechanism and the most frequent manifestation of die failure is thermal fatigue cracking. It has been shown in previous investigations [15,40] that the strength is very important in controlling crack initiation. It is also known that the mechanical properties are directly related to the hardness of the material (Figure 3.1). The capability of a steel to preserve good mechanical properties during cycling at temperatures above the tempering temperature, is essential in establishing a satisfactory level of performance to be expected. This assertion leads directly to the interest in the phenomenon of softening during thermal cycling of die casting dies. This subject will be considered and analyzed in this work. Hot-work tool steels like H13 are used in quenched and tempered condition. In this work, the extent of softening and the means to minimize it were evaluated. The effect of immersion time and diameter of the cooling line on the temperature and 39

40 softening of the surface were studied. In both cases, there are differences regarding the heat supply and extraction to and from the surface. In the case of different immersion times, the amount of heat supplied to the surface of the specimen is limited by the time spent in contact with molten aluminum. The heat extraction capacity is determined by the size of the cooling line. If the immersion time is constant, a constant amount of heat is supplied to the surface. However, a larger cooling line diameter will enhance the capacity of heat extraction. In production, it is very difficult to control the time spent by the casting in the die. The cycle length is limited by solidification time, especially for large parts. Under these conditions, designing the die with cooling lines closer to the surface may be the only feasible solution. One must be cautious and consider the limits set by the hoop stresses, which are increasing with the temperature gradient. Nevertheless, varying immersion time is very useful for the proposed study due to the ability to simulate extreme conditions that may occur during the die life The Influence of Immersion Time on Softening and Thermal Fatigue Cracking The experiment involved testing of three specimens for which the immersion time in the molten aluminum was the only variable. The maximum and minimum temperatures reached at the corner and the temperature distribution inside the specimen (toward the cooling line) varied as a function of the time spent in the molten metal bath. Thermal fatigue behavior of the three specimens was compared with the reference 9 seconds immersion time specimen. The results are presented as the Total Crack Area and the Average Maximum Crack Length for each immersion time (Figure 3.2 and Figure 3.3). 40

41 The hardness measured at the corner of the specimens after 15,000 cycles is shown in Figure 3.4. There is a clear trend for the thermal fatigue cracking parameters (i.e. Total Crack Area and Average Maximum Crack Length) to increase with immersion time. This observation points at the main cause of thermal fatigue damage, which is the temperature variation during cycling. One of the direct effects of the temperature, in particular the maximum temperature reached at the corner, is the softening of the steel. The extent of softening during tempering is generally evaluated by a master parameter, known as Hollomon-Jaffe parameter. This value represents the combined effect of temperature and time. Since the temperature and time are interdependent variables in the thermally activated process of tempering, a trade-off of temperature for time or vice-versa is based upon a simple equation: P = T(C + logt) x 10-3 (3-1) where P is the Hollomon-Jaffe parameter T is the absolute temperature [K] t represents the time [hours] C is a material constant This equation yields a reasonably accurate prediction of hardness for carbon and alloy steels containing % carbon and less than 5% total alloying elements, irrespective of initial structure. It is not the scope of this work to investigate the hardness of steel as a function of temperature-time. The hypothesis is however that the temperature is the main 41

42 factor that causes hardness loss. The dependency of thermal fatigue cracking on the level of hardness is investigated in detail. The dependency of Total Crack Area and Maximum Crack Length on the maximum temperature at the corner of the specimen for different immersion times is presented in Figures 3.5 and 3.6. The results demonstrate that the higher the temperature, the more thermal fatigue damage will occur. As previously discussed, a higher temperature will produce a more severe and deeper softening of the surface and within the section of the specimen. Initially, the surface deformation (strain) is within the elastic capabilities of the die steel. The surface of the specimen has irregularities in the forms of corrosion pits or surface scratches. These sites serve as stress concentrations. Plastic deformation can therefore occur at stresses well below the yield strength of the parent material (it must be also noted that the strength of the material drops at high temperature, see Figure 3.7), and initiate fatigue cracks. In addition to the stress concentrations caused by surface imperfections, tempering weakens the surface material. A cumulative fatigue process occurs in the material, since plastic strain gradually increases during the test as a result of lower yield strength of the material. The compressive stress will eventually exceed the elastic limit of the steel and plastic deformation will take place after the initial elastic strain has occurred [50]. Under these conditions, it is therefore necessary for the material to drop below of certain strength level characterized by a lower hardness value in order for the crack to initiate. It has been experimentally demonstrated that if the strength properties of the material are reclaimed before the cracks initiated, the thermal fatigue behavior can be 42

43 markedly improved. The experiment consisted of cycling a H13 steel specimen for 2,500 cycles and then re-heat treating it to the original hardness value. The results compared to regular 51 HRC and 46 HRC H13 specimens are presented in the Figure 3.8. It is clearly shown that the re-heat treated specimen to 51 HRC after every 2,500 cycles exhibited better resistance against heat checking. The cyclic heat treatment reclaimed the strength of the material and its resistance against cracking, impeding crack initiation, as well as the propagation of the existent cracks. Based on this evidence, it is believed that for a certain combination of temperature/stress the crack initiation will occur at a correspondent value of hardness. Therefore, it is expected that in a specimen subjected to a higher maximum temperature the hardness will drop faster. The higher drop in strength during immersion will thus cause cracks to initiate earlier. The cracks have then more time to grow, and the Average Maximum Crack Length will presumably be higher. The relationship between Average Maximum Crack Length and Total Crack Area is presented in Figure 3.9. Longer cracks correspond to a higher value of Total Crack Area. In addition, more cracks may initiate at the weakened surface, grow faster, and contribute to a higher Total Crack Area. At the same time, the behavior of the propagating crack is influenced by the characteristics of the material at the crack tip, and hence by the ability to resist plastic deformation. A parameter was chosen, which could provide information about the properties ahead the crack tip/front, namely the microhardness at a distance equal to the Average Maximum Crack Length (Figure 3.10). It is asserted that this distance characterizes well the propagation of cracks inside the specimen. The dependency of this 43

44 new parameter on the temperature measured at the corner is presented in Figure 3.11 and appears to have a linear trend. The relationship between the cracking parameters and the Immersion Time 5 sec 7 sec 9 sec 12 sec Maximum Temperature [F] Minimum Temperature [F] Total Crack Area [x 10 6 µm 2 ] After 15,000 Cycles Average Maximum Crack Length After 15,000 Cycles [x 100 µm] Hardness at the Average Maximum Crack Length [HRC] After 15,000 Cycles TABLE 3.1. Measurement Data For Different Immersion Times 44

45 Distance From the Corner [in] 5 sec 7 sec 9 sec 12 sec TABLE 3.2: Immersion Time Effect on Hardness Variation Across the Surface 45

46 microhardness measured at the distance equal to the Average Maximum Crack Length is described in Figures 3.12 and Figure It can be concluded that a higher maximum temperature will accelerate the loss in hardness at the corner of the specimen. The crack will extend to a longer distance, as it will have more time to propagate. The hardness loss will be also more severe further inside the material. It appears that the longer the crack is, i.e. the higher the temperature at the surface, the lower the hardness ahead of it. A possible explanation of this phenomenon is that the thermal stresses decrease from the surface towards the interior as the temperature gradient drops, mainly due to the decrease in the maximum temperature. In order for the crack to advance, the strength of the material must decrease even more. However, in this particular configuration, the cooling line does not allow a very deep softening and the crack may eventually stop before it attains a critical length that will lead to instability. In addition to the surface microhardness evaluation, internal hardness measurements were taken on a center section of the 1.5" cooling line diameter and 12 seconds immersion time specimen. This specimen was under to the most severe test conditions among all the specimens used. The sample was sectioned at the center (about 3.5" from the both ends). The Rockwell hardness profiles obtained from these HV microhardness values are shown in Figure The hardness distribution plots indicate that the softening near the edge is significantly higher than inside the sample, which is predictable due to the higher temperature and heat transfer conditions. However, the degree of softening exceeds the typical tempering curve, shown in the Figure The temperature measured at about 46

47 0.06" from the corner of the specimen is almost 1150 F. The temperature cycle at this location has a minimum at 460 F and a maximum at 1147 F. If the peak portion of the thermal cycle is separated (from the cycle presented in Figure 3.16), it will show that the specimen resided a total of 5.5 sec x 15,000 cycles = 82,500 seconds ~ 23 hours, at a temperature between 1100 F and 1150 F. The value of the hardness at 0.06" from the corner was measured to be 30.6 HRC. According to the tempering curve of the steel (from Figure 3.4), such a drop in hardness from 45 to around 30 HRC would be produced after about 23 hours at 1150 F. It is concluded that another mechanism contributed to the softening, presumably cyclic stress softening The Influence of Cooling Line Diameter on Softening and Thermal Fatigue Cracking Frequently, critical sections (usually thin parts or complicated shape sections subjected to multidirectional heat transfer) occur within a die-casting die. These sections are under high temperature and severe stress conditions. The importance of the maximum temperature and its influence on softening, and hence on thermal fatigue cracking was discussed in the previous section. In this experiment the maximum and the range of temperature reached at the corner and the variation inside the specimen was investigated as a function only of the cooling line diameter. A larger cooling line will actually bring down the maximum temperature at the surface, and at the same time will keep the temperature range almost the same, since the minimum temperature drops as well, because of a higher heat extraction capability. 47

48 The thermal fatigue behavior of three specimens with different cooling line diameters, 1.6", 1.7" and 1.8", was compared with the reference 9 seconds immersion time - 1.5" cooling line diameter specimen. The results are presented as the Total Crack Area and the Average Maximum Crack Length for each cooling line diameter (Figures 3.17 and 3.18). The evaluation of softening or hardness loss at the corner of the specimens after 15,000 cycles is shown in Figure The same trend as in the previous experiment was observed for the thermal fatigue cracking parameters. The values of Total Crack Area and Average Maximum Crack Length decrease with the increase in the maximum temperature The effect of different cooling line diameter on Total Crack Area and Average Maximum Crack Length as a function of maximum temperature at the corner of the specimen is presented in Figures 3.20 and The curve seems to reach a plateau as the cooling line diameter becomes smaller. If the curves are compared with those obtained for varying immersion times, it will be noticed that the tendency of the curve to level around a certain maximum temperature is common for both situations. The variation of microhardness measured at the Average Maximum Crack Length is shown in Figure The dependency of the thermal fatigue cracking on the microhardness measured at the Average Maximum Crack Length is presented in Figures 3.23 and 3.24.The relationship between the cracking parameter and the microhardness measured at the distance equal to the Average Maximum Crack Length follows the temperature trend, confirming the observation made for different immersion times. The longer the crack is, because of the higher temperature at the surface, the lower the hardness ahead the crack. 48

49 Cooling Line Diameter [in] 1.8" 1.7" 1.6" 1.5" Maximum Temperature [F] Minimum Temperature [F] Total Crack Area [x 10 6 µm 2 ] After 15,000 Cycles Average Maximum Crack Length After 15,000 Cycles [x 100 µm] Hardness at the Average Maximum Crack Length [HRC] After 15,000 Cycles TABLE 3.3: Measurement Data For Different Cooling Line Diameters 49

50 Distance From the Corner [in] 1.8" 1.7" 1.6" 1.5" TABLE 3.4: COOLING LINE DIAMETER EFFECT ON HARDNESS VARIATION ACROSS THE SURFACE 50

51 3.2. Stress Analysis at the Specimen Surface and Around the Cooling Line The stresses developed at the surface are responsible for initiation and subsequent crack propagation. The required hardness loss for crack initiation and propagation varies function of the level of induced stress. These thermal stresses are generated by the difference between the maximum and minimum temperature (temperature gradient). Different testing or production conditions will result in different temperature and stress distributions. The effect of immersion time and cooling line diameter on maximum, minimum and range of temperature are presented in Figures 3.25 and More severe conditions (longer immersion time or smaller diameter of the cooling line) will not raise only the maximum temperature but also the minimum temperature, so that the temperature gradient (range, in the plots) will not increase too much. Consequently, a larger cooling line or a shorter immersion time will shift the overall cycle towards lower values, keeping the stress in about the same range. This observation is extremely important, because of the implications resulting from the capacity of a larger cooling line diameter to promote a lower softening-causing maximum temperature without a major increase in the stress level. The stresses in the thermal fatigue specimen are complex. The cracks initiate not only at the corner of the specimen, where softening favors the plastic strain accumulation (Figure 3.27), but also at the cooling line, due to high hoop tensile stresses created during immersion. The latter formation is promoted by the existence of severe stress concentrators caused by cooling water corrosive action. In extreme conditions, the cracks initiated at the cooling line can cause failure of the specimen, mainly because they initiate 51

52 and grow faster in the thinnest section of the specimen or die as a consequence of high tensile hoop stress induced by the extreme temperature gradient (Figure 3.28). The axial stress range at the corner can be estimated using equation 1-1. For 12 seconds immersion and 1.5" cooling line diameter specimen: σ = αe T = 6.9 µin/in F * 30,500 ksi * 687 F = ~145 ksi. where α is the coefficient of thermal expansion, E represents the elastic modulus and T is the temperature gradient. This estimation agrees relatively well with the value of stress range obtained by computer modeling (Figure 3.29). The computer modeling for the 1.5 cooling line diameter specimen and 12 seconds immersion time shows that during immersion in molten aluminum the compressive axial stress at the corner attains a high value. High compressive stress and low yield strength may generate plastic strain. The result is a residual tensile stress, which is well below the yield strength, but high enough to initiate fatigue cracks at stress concentrators. The axial stress at the cooling line is tensile. Because the temperature at the wall of the cooling line is low, the axial stress is tolerated. However, the hoop stress developed is markedly higher. In the presence of stress concentrators like machining marks and corrosion pits, cracks can initiate and propagate from the cooling line Microstructure Degradation that Promotes Softening During Thermal Cycling The alloying elements present in the steel affect the hardening, tempering characteristics and the carbides in steels. As a consequence, they have the ability to impart certain features to die steel [48]: 52

53 - Greater strength in large sections because of deeper hardening or increased hardenability. In steels, strength is virtually proportional to hardness. - Less distortion in the process of hardening by increased hardenability due to the ability to harden the steel with a less drastic quench. Less distortion, dimensional change, and quench cracking are direct results of this lowering of thermal stresses set up by large temperature gradients. - Greater resistance to abrasion at the same hardness by promoting the formation of hard, stable and wear-resistant carbides. - Alloying elements induce higher toughness in small sections by promoting fine grain size. They also lower the internal stresses through less drastic quenches, and permit a greater relief of internal stresses through the use of higher tempering temperature without much loss of hardness. As far as the individual effect of particular elements, it is known that molybdenum is effective in improving the hardenability and high temperature strength. It retards the softening of martensite at all tempering temperature and reduces susceptibility to tempering embrittlement. Above 1000 F, the presence of molybdenum keeps the size of carbides small. Like molybdenum, chromium also retards the softening of martensite. By substituting chromium for some of the iron in cementite, the coalescence of carbides is retarded. However, its effect on the hardenability is less than that of molybdenum. Vanadium is a stronger carbide forming element than the above two elements. The vanadium containing carbides are stable at elevated temperature. Thus the steel has to be 53

54 austenitized at a sufficiently high temperature and for a sufficient length of time to bring most of the carbides into solution. For instance, when the H13 tool steel is austenitized at 1010 C (1850 F) for an hour, the molybdenum and chromium carbides are dissolved in solid solution, but the vanadium carbide (VC or V 4 C 3 ) does not dissolve [29, 30, 35]. Also, the precipitation of vanadium carbides and carbonitrides in high strength low alloy (HSLA) steels raises the strength of these steels well above the normally processed mild steels [24, 25]. Silicon not only has its own potential in increasing hardenability, but also stabilizes the ε iron carbide upon tempering to such an extent that it is still present in the microstructure after tempering at 400 C in steels with 1-2 % Si. Silicon slows down the nucleation and growth of the carbide and also enters into the carbide structure, delaying the transformation of ε to Fe 3 C. However, if in large quantity, silicon precipitates on grain boundaries, martensite lath boundaries and/or martensite lath/carbide interfaces during tempering [26, 27], which enhances the embrittlement and lower the toughness of die steels. A lower content of silicon presumably minimizes the interfacial segregation and results in higher toughness and thermal fatigue resistance [6]. Thermal fatigue resistance is affected by the combination of primary carbides formed in as quenched condition and carbides precipitated during tempering. Smaller and fewer carbides in as quenched conditions make the crack initiation hard, and the well dispersed carbide precipitation pattern makes it harder for crack to propagate [11]. The influence of the amount of primary carbides on the impact toughness properties of H13 is shown in Figure

55 The effect of temperature and hardness on thermal fatigue behavior of quenched and tempered H13 steel is supported by the observations made on the microstructure of quenched and tempered H13 steel at different distances from the corner, in the crosssection of the 12 seconds immersion time specimen (Figures 3.31). As the temperature increases, the coarsening of the carbides becomes more severe (Figures 3.32, a d). The photomontage in Figure 3.33 illustrates the distribution of carbide in this crosssection. As the temperature decreases from the surface towards the cooling line, the carbides become finer. Figure 3.34 (a) shows the microstructure of H13 sample in the unaffected material. The base material shows agglomeration of carbides from the original temper and some larger carbides probably remained undissolved during the austenitizing. These carbides are concentrated primarily at the austenite grain boundaries and between the lathes of martensite [15]. Figure Effect of Volume Percent Primary Carbides on the Transverse Charpy V-notch Impact Toughness of H13 [41] Transverse CVN (ft - lbs.) o F - 25 min., Oil quenched 1135 o F - 2hrs. Air Cooled - Surface - Center Avg. Vol. Primary Carbides

56 Tracking the carbides in a complex alloy steel like 5CrMoV(H13) back to the annealed structure, it is found that the total weight percentage of carbides present is about 4.4 %. After austenitizing at 1850 F, only 2.3 % of weight represents undissolved carbides (Figure 3.34), of which most is vanadium and some molybdenum. Further increase in austenitizing temperature results in extensive dissolution of Mo and V. By F, most of the molybdenum is in solution and the weight percentage of undissolved carbides is about 1.5 % and most of this is vanadium carbide [49]. Previous studies on the effect of austenitizing temperature on the amount of carbides present in the quenched microstructure of other tool steels have shown the same dependency [3]. The important characteristic of H13 steel is the presence of a higher content of vanadium than in other tool steels. Vanadium carbide tends to be more stable at higher austenitizing temperatures. The hardness after quenching is a good indicator of the effect of dissolution of vanadium carbide. A jump from about 59 to 61 HRC has been observed by increasing the austenitizing temperature from F range to F. Alloying elements also affect the softening resistance during tempering. They restrain the coarsening of cementite in the range C (Si, Cr, Mo, W) either by entering into the cementite structure or by segregating at the carbide-ferrite interfaces. Secondly, in alloy steels such as H13, a number of alloying elements form fine carbides that are thermodynamically more stable than cementite. The alloying elements Cr, Mo, V, W and Ti form carbides with substantially higher enthalpies of formation [47]. When strong carbide forming elements are present in steel in sufficient concentration, their carbides will be formed in preference to cementite. However, during tempering of alloy steels, 56

57 alloy carbides do not form until F. Below this temperature range the metallic alloying elements cannot diffuse fast enough to allow alloy carbides to nucleate. The metallic elements diffuse substitutionally, in contrast to carbon that diffuses interstitially. Hence, the diffusivity of carbon is several orders of magnitude greater in iron than those of the metallic alloying elements. The coarsening of carbides in steel influences markedly the mechanical properties. The strengthening theories show that the yield strength of a dispersed alloy, controlled by the capacity of dislocations to move around spherical particles, varies inversely with the spacing between particles. If the carbide dispersion is coarsened by further heat treatment, the hardness and strength of the alloy falls [47]. The theory for coarsening of a dispersion shows that the coarsening rate is dependent on the diffusion coefficient of the solute: r t 3 r 0 3 = (k/rt) V m 2 D σ t (3-2) where r 0 = the initial mean particle radius r t = the mean particle radius at time t D = diffusion coefficient of solute in matrix σ = interfacial energy of particle/matrix interface per unit area V= molar volume of precipitate k = constant 57

58 Under any given temperature, cementite will coarsen at a higher rate than any of the alloy carbides. This is typical in alloy steels in which cementite and an alloy carbide coexist, where cementite dispersion is always much coarser. A basic Energy Dispersive Spectrometry (EDS) analysis of carbides in the over tempered structure of the H13 near the corner of the specimen after 15,000 cycles, has shown that the largest carbides in the microstructure are Cr-rich. The smaller carbides are Mo-rich carbide (Figures 3.35 and 3.36). Chromium diffuses more rapidly in ferrite than most metallic alloying elements, with the result that in chromium steels Cr 7 C 3 is detected during tempering at temperature as low as 500 C, and it coarsens rapidly compared to molybdenum or vanadium carbides [47]. Thus, in chromium steel, continuous softening will normally occur during tempering between C, although the addition of other elements, such as Mo, can reduce the rate of coarsening of Cr 7 C 3. Also, previous works have shown that during tempering at 1200 F of 5CrMoV steel, an iron-rich chromium carbide forms, (CrFe) 7 C 3 [16]. The small carbide appears to be Mo-rich, in the form of M 2 C or eventually M 6 C, explainable by the fact that molybdenum carbide is less sensitive to growth. Vanadium-rich carbides were not detected EDS, even though some vanadium was found to be present in Cr-rich carbides. This is due to the ability of vanadium to maintain a very fine carbide (VC or V 3 C 4 ) dispersion even at temperatures approaching 700 C. The detection of vanadium-rich carbides by EDS method is at best difficult. 58

59 4. CONCLUSIONS 1) For a configuration without severe stress concentrators, the softening of the steel is the most important factor for the crack initiation. Less thermal fatigue damage has been observed when the conditions promoted lower temperature at the surface, which preserved the hardness and hence the strength. A high value of yield strength means higher material resistance to plastic deformation. At the same time, elevated temperature at the surface will induce a deeper softening. It appears that a condition for the extension of the thermal fatigue cracking damage is the decrease in strength ahead the crack front. 2) In die-casting applications, the highest maximum temperature will occur in thin sections where the material capacity to absorb and transfer the heat from the surface is very different. From another point of view, high temperature - long resident time conditions are important, because of the similarity with the die casting of large components, when the die is subjected to elevated temperature for longer periods of time. The experimental results have shown an important decrease of the cracking when the cooling line is positioned closer to the surface. Moreover, the experimental data indicates the existence of a temperature threshold, below which the thermal fatigue damage is minimal. A cooling line closer to the surface will shift the maximum temperature towards lower values, and keep at the same time the stresses at a relative constant value. However, decreasing the maximum temperature at the surface by placing the cooling lines too close to the surface may be limited by the high level of hoop stresses created at the cooling line. 59

60 3) The presence of strong carbide-former elements like chromium, molybdenum and vanadium, will reduce the softening by preserving a fine distribution of carbides. These elements inhibit the coarsening of cementite in the range C. At the same time, these elements form fine carbides that are thermodynamically more stable than cementite. Among the three elements, chromium-rich carbide is the most susceptible to growth, but the presence of molybdenum and vanadium inhibits it to certain measure. 60

61 FIGURE 3.1. Relationship Between Tensile Properties and Hardness for H13 Steel [45, Reprinted with permission of American Society of Materials] 61

62

63 sec 7 sec The Effect of Immersion Time 12 sec Total Crack Area [x 10 6 µm 2 ] sec 12 sec 9 sec 20 0 all below sec 5 sec Number of Cycles FIGURE 3.2. The Effect of Thermal Cycling on Crack Area - Different Immersion Times - 63

64 Average Maximum Crack Length [x100 µm] sec 7 sec 9 sec 12 sec The Effect of Immersion Time 5 sec 7 sec 9 sec 12 sec Number of Cycles FIGURE 3.3. The Effect of Thermal Cycling on Crack Length - Different Immersion Times - 64

65 15,000 cycles Hardness HRC sec 7 sec 20 9 sec 12 sec Distance from the Corner [in] The Effect of Immersion Time FIGURE 3.4. The Effect of Thermal Cycling on Microhardness Distribution Across the Surface - Different Immersion Times - 65

66 , 000 cycles 12 sec Total Crack Area [x10 6 µm 2 ] sec 20 5 sec 7 sec The Effect of Immersion Time Temperature [F] FIGURE 3.5. The Effect of Temperature on Crack Area - Different Immersion Times - 66

67 Average Maximum Crack Length [x 100 µm] 18 15, 000 cycles sec 14 9 sec sec 5 sec 2 The Effect of Immersion Time Temperature [F] FIGURE 3.6. The Effect of Temperature on Crack Length - Different Immersion Times - 67

68 FIGURE 3.7. Effect of Elevated Temperature on Tensile Strength [45, Reprinted with permission of American Society of Materials] 68

69

70 H13 at 46 HRC H13 at 51 HRC Total Crack Area [x 10 6 µm 2 ] H13 Re-Heat Treated to 51 HRC After Every 2,500 Cycles 2BAR Quench+Double Temper to 51HRC 2BAR/46HRC OIL/51HRC 2BAR/51HRC Thermal Cycles FIGURE 3.8. The Effect of Hardness Recovery on Thermal Fatigue Cracking 70

71 Total Crack Area [x 10 6 µm 2 ] Average Maximum Crack Length [x 100 µm] FIGURE 3.9 Relationship Between Total Crack Area and Average Maximum Crack Length 71

72 T 1 < T 2 FIGURE The Relationship Between Maximum Crack Length and Microhardness at Maximum Crack Length 72

73 40 Microhardness at Maximum Average Crack Length [HRC] 15, 000 cycles 5 sec 35 7 sec 9 sec sec The Effect of Immersion Time Temperature [F] FIGURE The Effect of Temperature on Microhardness - Different Immersion Times- 73

74 , 000 cycles 12 sec Total Crack Area [x 10 6 µm 2 ] sec sec 5 sec The Effect of Immersion Time Microhardness at the Average Maximum Crack Length [HRC] 28 FIGURE The Effect of Microhardness at Average Maximum Crack Length on Crack Area - Different Immersion Times - 74

75 18 Average Maximum Crack Length [x 100 µm] , 000 cycles 5 sec 36 7 sec sec 9 sec The Effect of Immersion Time Microhardness at the Average Maximum Crack Length [HRC] FIGURE The Effect of Microhardness at Average Maximum Crack Length on Crack Length - Different Immersion Times - 75

76 FIGURE Microhardness Profile at the Corner of 12 Seconds Immersed Specimen 76

77 FIGURE Tempering Curve for 77

78 H13 Temperature [F] Time [sec] FIGURE 3.16: Maximum Temperature Cycle for 1.5" Cooling Line Specimen After 12 Seconds Immersion Time 78

79 " 1.7" The Effect of Cooling Line Diameter Total Crack Area [x 10 6 µm 2 ] " 1.5" 1.8" 1.7" 1.6" 1.5" 20 0 all below Number of Cycles FIGURE The Effect of Thermal Cycling on Crack Area - Different Cooling Line Diameters - 79

80 18 Average Maximum Crack Length [x100 µm] " 1.7" 1.6" 1.5" The Effect of Cooling Line Diameter 1.5" 1.6" 1.7" 1.8" Number of Cycles FIGURE The Effect of Thermal Cycling on Crack Length - Different Cooling Line Diameters - 80

81 15,000 cycles Hardness HRC " 1.7" 1.6" 1.5" Distance from the Corner [in] The Effect of Cooling Line Diameter FIGURE The Effect of Thermal Cycling on Microhardness Distribution Across the Surface - Different Cooling Line Diameters - 81

82 120 15, 000 cycles 1.5" 100 Total Crack Area [x10 6 µm 2 ] " 1.7" 1.6" 20 The Effect of Cooling Line Diameter Temperature [F] FIGURE The Effect of Temperature on Crack Area - Different Cooling Line Diameters - 82

83 Average Maximum Crack Length [x 100 µm] 14 15, 000 cycles 1.6" 1.5" " " The Effect of Cooling Line Diameter Temperature [F] FIGURE The Effect of Temperature on Crack Length - Different Cooling Line Diameters - 83

84 40 Microhardness at Average Maximum Crack Length [HRC] , 000 cycles 1.8" 1.7" 1.6" 1.5" The Effect of Cooling Line Diameter Temperature [F] FIGURE The Effect of Temperature on Microhardness - Different Cooling Line Diameters - 84

85 120 15, 000 cycles 1.5" 100 Total Crack Area [x10 6 µm 2 ] " 1.7" 1.6" The Effect of Cooling Line Diameter Microhardness at the Average Maximum Crack Length [HRC] 31 FIGURE The Effect of Microhardness at Average Maximum Crack Length on Crack Area - Different Cooling Line Diameters - 85

86 120 15, 000 cycles 1.5" 100 Total Crack Area [x10 6 µm 2 ] " 1.7" 1.6" The Effect of Cooling Line Diameter Microhardness at the Average Maximum Crack Length [HRC] 31 FIGURE The Effect of Microhardness at Average Maximum Crack Length on Crack Area - Different Cooling Line Diameters - 86

87 18 Average Maximum Crack Length [x 100 µm] , 000 cycles 1.8" " 1.6" " The Effect of Cooling Line Diameter Microhardness at the Average Maximum Crack Length [HRC] FIGURE The Effect of Microhardness at Average Maximum Crack Length on Crack Length - Different Cooling Line Diameters - 87

88 Maximum Temperature Minimum Temperature Range 1000 Temperature [F] sec 7 sec 9 sec 12 sec Immersion Time FIGURE The Effect of Immersion Time on Temperature 88

89 Maximum Temperature Minimum Temperature Range Temperature [F] Cooling Line Diameter FIGURE The Effect of Cooling Line Diameter on Temperature 89

90 Corner X 45 FIGURE CRACKS AT THE CORNER OF H13 SPECIMEN Cooling Line X 200 FIGURE CRACK AT THE COOLING LINE OF H13 SPECIMEN 90

91 Axial Stress Node A Axial Stress Node B Hoop Stress Node B Stress [psi] A Cooling Line, 12 Seconds Immersion Time [sec] B FIGURE Stress Modeling at the Corner and Cooling Line 91

92 Figure Effect of Volume Percent Primary Carbides on the Transverse Charpy V-notch Impact Toughness of H13 [41] Transverse CVN (ft - lbs.) o F - 25 min., Oil quenched 1135 o F - 2hrs. Air Cooled - Surface - Center Avg. Vol. Primary Carbides 92

93 Corner FIGURE Microstructure Sampling at the Corner of 12 Seconds Immersion Specimen 93

94 ~30 HRC 15,000 Cycles FIGURE 3.32(D) THE EFFECT OF TEMPERATURE ON MICROSTRUCTURE 0.06 FROM CORNER 15,000 Cycles ~27 HRC FIGURE 3.32(E). THE EFFECT OF TEMPERATURE ON MICROSTRUCTURE CORNER 94

95 15,000 cycles 12 Seconds Immersion Time 1.5 Cooling Line Diameter FIGURE Temperature Influence on Carbide Size and Distribution Photomontage 95