Phase and microstructure selection in peritectic alloys under high G-V ratio

A new model for the case of a superposition of a phase transformation and a microstructure transition from columnar plane front to equiaxed dendritic growth for peritectic alloys is presented. This model is applied to Fe-Ni alloys where a phase transition from gamma to delta takes place when the solute content decreases. The results of the model are compared with predictions of the nucleation constitutional undercooling model. They show that the peritectic two-(solid) phase region is restricted and may completely disappear depending on nucleation undercooling and dendrite growth kinetics.

Peritectic solidification of Cu-Sn alloys

Frédéric Kohler

Many commercially important alloys such as Fe-C, Fe-Ni, Cu-Zn, Cu-Sn and Ti-Al exhibit peritectic transitions on solidification. In the two-solid phase region of peritectic metallic systems, a wide spectrum of microstructures has been revealed during directional solidification experiments at low growth rates, where both α- and β-phases would grow independently as planar fronts [1, 2, 12–20]: discrete bands, islands, simultaneous growth of the α- and β-phases in the form of oriented lamellae or fibers and continuous oscillatory tree-like microstructures in highly convective peritectic systems [3, 21, 22]. Recently, Vandyoussefi et al. [27] and Dobler et al. [16] have done extensive solidification experiments on Fe-Ni alloys in the corresponding two-solid phase region. Depending on the growth conditions and local composition, a lamellar eutectic-like structure was observed. Despite recent works, the peritectic solidification at low growth rate is not fully understood. Additionally, coupled growth was observed only for peritectic systems for which the solidification interval of the primary phase ΔΤ 0α is fairly small and with negligible convection; but is it also possible for peritectic alloys with fairly large ΔΤ 0α such as Cu-Sn? If so, what would be the effect of convection? And finally, is the coupled growth front really isothermal? In the present contribution, the peritectic Cu-Sn system has been chosen because of its remarkable properties and technological importance. Indeed, two essential features distinguish Cu-Sn alloys from other peritectic systems: The equilibrium solidification interval of the a-phase decreases with the tin concentration in the hypoperitectic composition range. Additionally, the equilibrium solidification interval of the primary phase is about 25 times larger in the Cu-Sn system than in the Fe-Ni peritectic system. Since only few studies have been made on the detailed solidification of tin bronze alloys, the first goal of this study was to carry out thermal analyses in order to investigate precisely the solidification of Cu-Sn alloys and to measure temperatures of solid-state phase transformations. Therefore, an SPTA assembly has been built and successfully calibrated. Additionally, a heat flow model was developed in this work and coupled with a Cu-Sn thermodynamic database to treat solidification-dependent latent heat. Single pan thermal analyses on three Cu-Sn alloy compositions revealed that the corresponding phase diagram recently re-investigated in Liu et al. review [28] is the most reliable. Directional solidification experiments on Cu-Sn alloys of various compositions have been conducted at different velocities in a high gradient Bridgman furnace. In this context, the set-up already used by Dobler [14] has been modified in order to reduce the size of the samples and, accordingly, natural convection and the associated macrosegregation. During each solidification run, two different geometries were thus tested (3 mm diameter cylinder and 4/6 mm tube), allowing the observation of convection effects on solidifying microstructures. An essential outcome of this project is that α+β lamellar structures have been observed in both sample geometries, even when Vp is above the critical velocity for the α planar front growth, i.e., the α-phase grew with a cellular morphology. With the Cu-Sn phase diagram used in the present work and for positions close to the quenched interface, concentration profiles measured across the lamellar structures revealed that the corresponding operating temperature Τ*(x) seems to be non-isothermal, in contradiction with previous experimental observations in Fe-Ni peritectic alloys [14–16]. But, since the Cu-Sn phase diagram is not assessed accurately, Τ*(x) is still subject to large variations (±10°C). On the other hand, the peritectic transformation after solidification has been evidenced and modifies the compositions and fractions of phases. Additionally, local concentration showed that a partial banding mechanism and an unstable lateral propagation of both solid phases were responsible for the formation of lamellar structures. Various geometrical arrangements were revealed on transverse cuts of lamellar structures: straight α- and β-lamellae, labyrinth-like microstructures, disordered arrays of rods of one phase in the matrix of the other phase and uniform regions of β-phase. The dynamic evolution of lamellar structures was attributed to growth competition mechanisms and instabilities between phases originated from the same nuclei. Finally, it should be noted that the Kurdjumov-Sachs relationship was observed between the primary and the peritectic phases on both the straight lamellae and labyrinth-like microstructure. Finally, all lamellar structures ended by the termination of α-lamellae and the subsequent growth of a β planar front. In this context, the α-lamellae of lamellar structures observed in the 4/6 mm tube exhibited in all cases a "hammer-like" morphology before being overgrown by the peritectic phase. Careful investigations of this morphology revealed that it resulted effectively from growth competition mechanisms, which remain to be fully understood. In a simulated geometry corresponding to the 4/6 mm tube of Cu-Sn samples, 3D macrosegregation modeling revealed that a nearly helical convection pattern existed in the liquid phase ahead of a moving planar α-liquid interface. And between each plume, the fluid flow was mostly oriented parallel to the solid-liquid interface. As a consequence, the lateral propagation of the α- and β-phases during the 3D growth competition was surely affected by these convective motions. Finally, natural convection and the associated solute enrichment were assumed to be responsible for the destabilization of lamellar structures and the subsequent end of α-lamellae. The multi-phase field computations emphasized that the stability of the triple junction between α-, β- and liquid phases is function of the initial alloy concentration C0, of the lamellar spacing λ and of the thermal gradient Gl. In a near future, directional solidification experiments are planned to be conducted onboard of the International Space Station. Then, the obtained peritectic microstructures will have to be compared with the microstructures observed in the present work in order to assess more precisely the effect of natural convection on the formation of lamellar structures.

EPFL2008

Characterization and modeling of twinned dendrite growth

Mario Alberto Salgado Ordorica

The formation of feathery grains during semi-continuous casting of Al-alloys [1, 2] is an interesting problem from both practical and theoretical points of view. These structures are formed by a lamellar sequence of twinned and untwinned regions separated by straight and wavy-like boundaries. Each pair of lamellae contains twinned dendrites split in their trunk center by a coherent {111} twin plane, while lateral arms meet at an incoherent {111} boundary. In practice, feathery grains are considered as defects which reduce the mechanical properties of a solidified ingot. In theory, an understanding of twinned dendrite growth includes different solidification phenomena, e.g., interfacial energy anisotropy, crystallographic growth directions, twinning, growth competition mechanisms, etc. Although several studies have been performed in order to understand the physics leading to the nucleation and growth of twinned dendrites, various questions remain unanswered. In this work, a comprehensive study of twinned dendrite growth has been undertaken, with the main objectives being: i) to study the effect of different alloying elements and solidification conditions on twin formation in binary Al-alloys; ii) to establish a better understanding on the stability of twinned dendrites and the growth kinetic advantage that they exhibit over regular ones; and iii) to elucidate the stable shape of the twinned dendrite tip. In order to study alloying-element effects, binary Al-X alloys (where X = Zn, Mg, Cu and Ni), were produced under Directional Solidification (DS) conditions in the presence of a slight natural convection in the melt. Analysis of these castings has shown that feathery grains can form for all solute elements of interest, but not for all compositions. The probability of forming feathery grains is relatively high when the alloying elements are hcp (Zn or Mg), but decreases for fcc solute elements (Cu or Ni). A study on the effect of forced convection in the melt, performed using different experimental set-ups, confirms previous observations suggesting that it is the shearing components of the liquid slow which induce twin nucleation [3, 4]. However, the poor reproducibility of these experiments and the variable rate of feathery grains formation indicate that twin nucleation is governed by a highly stochastic behavior. The probability of such an event decreases as the melt slow is less complex and the associated Stacking Fault Energy (SFE) of the alloying element is increased. In terms of growth kinetics advantage, a characterization using various metallography techniques and X-ray synchroton tomography has shown that this is in part due to the complex morphology of twinned dendrites. Indeed, it has been confirmed that these dendrites grow along ‹110› directions with ‹110›, and also sometimes ‹100› secondary arms, the primary trunk spacing of these dendrites being much less anisotropic than previously thought [5, 6]. It has been shown that twinned dendrites grow in a stable manner at a lower undercooling than regular ones. In addition, the distribution and orientation of their side arms favors their growth at the expense of less developed regular dendrites. A mechanism to explain the lateral multiplication of twin planes is also proposed in this work. In terms of stability, observations after partial remelting of twinned DS specimens, then re-solidification in a Bridgman furnace, have shown that even if twin planes remain stable during partial remelting, independent regular non-twinned dendrites issued from the twinned and untwinned parts of the seed grow during solidification. This implies that the formation of twinned dendrites is not only related to the ability to nucleate a stacking fault, but also to the imposed solidification conditions. The favorable growth kinetics of twinned dendrites is also explained by their tip morphology. Three hypotheses have been evaluated in this work: i) the grooved tip [7], stabilized by the Young-Laplace equilibrium condition; ii) the doublon [5], i.e., a double tip dendrite that grows with a narrow liquid channel in its center that solidifies at a composition close to C0; and iii) the pointed (or edgy) tip [8], equilibrated by torque terms of γsl when it is too anisotropic. Observations performed at the twinned dendrite tip have revealed the presence of a small groove, which eliminates definitively the hypothesis of the edgy tip. Further examination of the stable growth morphology of twinned dendrites has been done by using a phase field model that reproduces the presence of a twin plane through an appropriate boundary condition. The results of these simulations show that twinned dendrites are doublons that grow with a narrow liquid channel (0.2 to 3 µm width) whose depth depends strongly on the alloy composition and the solidification conditions. In order to validate these simulations, Focused Ion Beam (FIB)-microtomography and X-ray chemical analysis (EDS) in a Scanning Transmission Electron Microscope (STEM) were performed on small specimens extracted from twinned dendrite trunks. These have shown the existence of a positive solute gradient in a region localized within 2 µm around the twin plane, i.e., as expected for a doublon morphology. Additionally, the presence of small particles aligned within the twin plane is in agreement with the formation of small liquid pockets below the doublon root as predicted by the numerical model, but further work is required to explain this remarkable feature. Finally, composition measurements performed after quenching a partially remelted specimen also seem to confirm the existence of a doublon. This work contributes to the understanding on twinned dendrite growth kinetics in several aspects, principally the effect of alloying elements on twin formation, their growth kinetic advantage, their stability and their stable tip morphology. However, further work must be carried out to overcome the limited knowledge on the mechanisms leading to twinned dendrite nucleation. In the same manner, the growth mechanisms of the doublon morphology should also be further investigated in future works.

EPFL2009

Texture Formation in Hot-Dip Galvanized Coatings

Aurèle Mariaux

Continuous hot-dip galvanization is a process during which steel sheets are coated with a layer of Zn-Al alloy to protect them against corrosion. As the name indicates, this is achieved by dipping the preheated sheet into a bath of molten alloy. The sheet is then withdrawn and cooled down in air. The solidification microstructure of such coatings is made of large zinc grains (spangles), which preferentially have their hexagonal ‹0001› axis perpendicular to the sheet plane (basal texture). The mechanisms leading to this preferred orientation were investigated in this study, with a particular focus on nucleation and dendritic growth. An experiment was designed to reproduce the solidification stage of hot-dip galvanization on a laboratory scale. Industrial coatings were remelted in an infrared image furnace and solidified under controlled cooling conditions. Thanks to the wide range of possible cooling rates, the influence of this parameter on the size and orientation of the zinc grains could be investigated. The thermal plateau induced by the solidification of the thin coating could be detected, even for fairly thick steel sheets. However, the signal was better for larger coating/steel thickness ratio. A minimum coating thickness and Al content are needed if Fe-Zn intermetallic outbursts are to be avoided during remelting. Additionally, skin-pass, as well as any deformation of the coated sheets, must be avoided prior to these laboratory experiments. A coupled cellular automaton — finite volume model was concurrently developed to simulate solidification. It was then used to obtain the nucleation parameters of the spangles by inverse modeling based on the remelting experiments. The model accounts for the effects of zinc crystal symmetry and of the confined coating geometry on nucleation and growth phenomena. In particular, the preferred orientation of the grains is included in the nucleation site distribution. Its basal part is still an estimate, but it could be refined from new experimental measurements with the infrared remelting device. On a more fundamental side, a mathematical formulation of the anisotropic solid-liquid interfacial energy of Zn-Al alloys was derived from the equilibrium shape of liquid droplets in a solid matrix. This property changes by more than 40% between the c axis and the basal plane. This very large variation has important effects on the solidification mechanisms. The interfacial energy, as well as the wetting conditions for an anisotropic grain, were then implemented in a phase field model, which was used to simulate the growth of zinc dendrites in coatings. It revealed that the growth directions and velocities are strongly influenced by the film thickness and by anisotropic wetting effects. The possibility that the preferred zinc orientation could be induced by the textured steel substrate was investigated by comparing the orientation of zinc grains with that of the ferrite grains underneath. It was observed that spangles with a random orientation distribution are independent of the substrate. Unfortunately, the observed specimen did not exhibit a basal coating texture and thus, the existence of a relation between basal grains and ferrite texture could not be investigated. The anisotropy formulation was also used to extend the classical theory of heterogeneous nucleation to anisotropic metals. These computations showed that a basal orientation of the nuclei is strongly favored from an energy point of view. Such developments can explain the occurrence of basal grains around holes that form during infrared remelting, so-called rose structures. Preferential basal nucleation and the wetting influence on dendrite growth are also responsible for the morphology transition observed in Galfan, a Zn-Al coating of eutectic composition: when appropriate cooling conditions are applied, a shiny layer of basal grains forms at the top surface before the onset of eutectic solidification. Finally, these mechanisms could also explain the preferential basal orientation of regular hot-dip galvanized coatings. However, the conditions on a galvanizing line do not meet the criteria that are requested to activate them in Galfan. Additionally, the texture of hot-dip coatings is reported to depend on the surface preparation of the sheet, although the anisotropic nucleation theory does not consider the steel substrate. Further investigations on these contradictions are needed before these mechanisms can be used to explain the preferred orientation in regular galvanized coatings.

EPFL2010