Phase selection in hypercooled alloys

Highlights

• Combination of ESL and X-ray diffraction allows to study phase formation of hypercooled intermetallic alloys in-situ.

• These alloys crystallize into a cubic B2-type (CsCl-type) phase primarly, independent from undercooling nucleation temperature.

• Newly discovered effect of the hypercooling limit on the crystal growth velocity of intermetallic alloys remains unexplained.

Abstract

The effect of the hypercooling limit on the crystal growth velocity v as a function of the undercooling temperature $\text{Δ}T$ for intermetallic binary and ternary alloys remains unexplained yet. In combination with the electrostatic levitation technique, we performed in-situ high energy synchrotron X-ray diffraction measurements at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France, to investigate the phase formation of the binary alloy $\text{NiTi}$, the $\text{Zr}$-based ternary alloys (Ni_xCu_1−x)Zr ($x=0.3,0.4$) and the $\text{Ni}$-based ternary alloy $\text{Ni}\left({\text{Zr}}_{0.5}{\text{Ti}}_{0.5}\right)$ for solidification events at undercooling temperatures $\text{Δ}T<\text{Δ}{T}_{\text{hyp}}$ and $\text{Δ}T>\text{Δ}{T}_{\text{hyp}}$, whereas $\text{Δ}{T}_{\text{hyp}}$ are the corresponding hypercooling limits of the systems. Our results indicate that the cubic B2-type ($\text{CsCl}$-type, space group Pm3 m $\#221$) phase is always the primary solidifying phase for all investigated alloys. This holds for all undercoolings obtained. Thus, the deceleration effect observed in the crystal growth velocity for all investigated alloys at their hypercooling limits seems to be caused by something different than the formation of different phases.

Introduction

Recently, we reported about the effect of the hypercooling limit on the crystal growth velocity of undercooled melts [1]. The hypercooling limit is thermodynamically defined as $\text{Δ}{T}_{\text{hyp}}=\text{Δ}{H}_{\text{f}}/c_{\text{p}}$ and describes the undercooling temperature, from where complete isenthalpic solidification is achieved. For every alloy investigated, we observe a change in the $v\left(\text{Δ}T\right)$ relation at the corresponding hypercooling temperature. An explanation for this behaviour of the crystal growth velocity has yet to be found. The solidification front geometry does not exhibit any change close to the hypercooling transition and thus cannot be taken as an explanation. A natural explanation would be given by the solidification of the congruently melting intermetallic systems in two different crystal structures below and above the hypercooling limit. Different crystal structures can exhibit even for the same system largely different growth velocities. This has already been shown for ${\text{Fe}}_{60}{\text{Co}}_{40}$ by Herrmann et al. [2] and $\text{Fe}-\text{Cr}-\text{Ni}$ by Moir et al. [3]. Hartmann et al. identified the sharp increase of the crystal growth velocity of NiAl at $\text{Δ}{T}^{\text{∗}}=250\text{K}$ as a transition from ordered to disordered growth by in-situ X-ray diffraction measurements [4] and described it quantitatively by an extended sharp interface model. However their so-called critical undercooling temperature $\text{Δ}{T}^{\text{∗}}$ is far away from the system’s hypercooling limit ($\text{Δ}{T}_{\text{hyp}}=400\text{K}$). Several works have shown that the electrostatic levitation (ESL) technique offers the opportunity to investigate the solidification behaviour of metallic melts down to deep undercoolings [[5], [6], [7], [8]], some of them even close to the hypercooling limits [9,10]. Notably [11], claims to have exceeded the hypercooling limit for ${\text{Cu}}_{0.46}{\text{Zr}}_{0.54}$. In Ref. [1] we showed recent levitation experiment results: For the binary alloys Cu–Zr and $\text{NiTi}$ as well as the ternary alloys $\left({\text{Ni}}_{0.3}{\text{Cu}}_{0.7}\right)\text{Zr}$, $\left({\text{Ni}}_{0.4}{\text{Cu}}_{0.6}\right)\text{Zr}$ and $\text{Ni}\left({\text{Zr}}_{0.5}{\text{Ti}}_{0.5}\right)$ their crystal growth velocities versus undercooling temperature were investigated exceeding the corresponding hypercooling limits of the systems. The achieved undercooling temperatures are of a few hundred degrees below the liquidus temperature.

For the experimental results presented in this publication we combine the containerless electrostatic levitation technique with in-situ synchrotron X-ray diffraction at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France. We show that phase selection during rapid solidification can be monitored in deeply undercooled melts in real-time. As a well-suited binary intermetallic glass-forming system Cu–Zr has been very well studied in recent years [7,[12], [13], [14]]. It has already been shown in the past that it undercools below its hypercooling limit while freely suspended in an ESL [15]. Our results showed that the $v\left(\text{Δ}T\right)$ relation clearly changes at the hypercooling limit, however the solidification front geometry does not change. Gegner et al. carried out in-situ observations of the phase selection from the undercooled melt in Cu–Zr [15]. They found that for $\text{Cu}50\text{Zr}50$ the primary solidifying phase is always of B2-type over the entire investigated undercooling range, even beyond the hypercooling limit, up to $\text{Δ}T=310\text{K}$. ${\text{Cu}}_{0.46}{\text{Zr}}_{0.54}$ should always solidify as a combination of $\text{CuZr}$ (B2-type) and $\text{CuZr}2$ (tetragonal C11_b-type, $\text{MoSi}2$-type, space group I4/mmm $\#$ 139). Carrying on, in this work we have investigated the formation of the primary solidifying phases at the liquid-solid transitions for the binary intermetallic alloy $\text{NiTi}$ and for the ternary compositions $\left({\text{Ni}}_{0.3}{\text{Cu}}_{0.7}\right)\text{Zr}$, $\left({\text{Ni}}_{0.4}{\text{Cu}}_{0.6}\right)\text{Zr}$ and $\text{Ni}\left({\text{Zr}}_{0.5}{\text{Ti}}_{0.5}\right)$. The measurements have been performed for undercooling temperatures below and above the hypercooling limits of the alloys, i.e. $\text{Δ}T<\text{Δ}{T}_{\text{hyp}}$ (hypocooled) and $\text{Δ}T>\text{Δ}{T}_{\text{hyp}}$ (hypercooled), respectively.