www.nature.

com/npjmgrav

REVIEW ARTICLE OPEN

Microgravity studies of solidification patterns in model

transparent alloys onboard the International Space Station
S. Akamatsu 1 ✉, S. Bottin-Rousseau1, V. T. Witusiewicz 2, U. Hecht2, M. Plapp3, A. Ludwig4, J. Mogeritsch4, M. Şerefoğlu 5
,
N. Bergeon6, F. L. Mota 6, L. Sturz 2, G. Zimmermann2, S. McFadden7 and W. Sillekens 8

We review recent in situ solidification experiments using nonfaceted model transparent alloys in science-in-microgravity facilities
onboard the International Space Station (ISS), namely the Transparent Alloys (TA) apparatus and the Directional Solidification Insert
of the DEvice for the study of Critical Liquids and Crystallization (DECLIC-DSI). These directional-solidification devices use innovative
optical videomicroscopy imaging techniques to observe the spatiotemporal dynamics of solidification patterns in real time in large
samples. In contrast to laboratory conditions on ground, microgravity guarantees the absence or a reduction of convective motion
in the liquid, thus ensuring a purely diffusion-controlled growth of the crystalline solid(s). This makes it possible to perform a direct
theoretical analysis of the formation process of solidification microstructures with comparisons to quantitative numerical
simulations. Important questions that concern multiphase growth patterns in eutectic and peritectic alloys on the one hand and
single-phased, cellular and dendritic structures on the other hand have been addressed, and unprecedented results have been
obtained. Complex self-organizing phenomena during steady-state and transient coupled growth in eutectics and peritectics,
1234567890():,;

interfacial-anisotropy effects in cellular arrays, and promising insights into the columnar-to-equiaxed transition are highlighted.
npj Microgravity (2023)9:83 ; https://doi.org/10.1038/s41526-023-00326-8

INTRODUCTION different phases—via the so-called interfacial anisotropy. Such

Solidification microstructures in alloys largely determine the challenging issues call for fundamental research based on in situ
properties of materials, and their characterization is of utmost experimental diagnostics, using systematic protocols guided by
interest in industrial research. It is common practice to identify the general concepts of the nonlinear physics applied to
those microstructures ex situ under the microscope and to solidification phenomena4.
measure their morphological features with full knowledge of the In this review, we will report on recent observations made
chemical nature of individual compounds and the physical during microgravity solidification experiments using model
properties of the mixture. Most often, however, information is transparent alloys. Focus is placed on unprecedented results
lacking on the actual solidification path during the cooling obtained with two devices installed onboard the International
process, and a clear interpretation of how the distribution of the Space Station (ISS), namely, the Transparent Alloys (TA) apparatus
microstructures in the bulk solid occurred cannot be achieved. The of the European Space Agency (ESA)5 and the Directional
central question—how, upon cooling, a heterogeneous crystalline Solidification Insert of the DEvice for the study of Critical Liquids
solid forms from a homogeneous liquid mixture—is actually and Crystallization (DECLIC-DSI) developed by the French National
strikingly complex when considered on a fundamental level1,2. Center for Space Studies (CNES)6 in close collaboration with NASA.
Frozen-in or as-cast microstructures form out of equilibrium. They Two distinctive features and strengths of this scientific research
arise from self-organizing processes during growth at the are worth underlining. First, considering the long characteristic
advancing interface between the solid and the liquid. Some times of the targeted phenomena, the ISS is the only facility that
characteristic lengths and their approximate scaling with micro- provides a stable reduced-gravity environment for experimental
scopic properties and control parameters can be derived campaigns that typically require several weeks. Access is thus
theoretically. A scaling analysis is however insufficient: spatiotem- given to dynamic phenomena that are difficult to reproduce in a
poral phenomena during solidification depend on boundary laboratory due, in particular, to the interaction of the solidification
conditions, initial conditions, and the whole history of the with convective motion in the liquid. High-quality experiments in
process3. Theoretical challenges are primarily associated with essentially diffusion-controlled crystal growth conditions are key
steady-state shapes and patterns, their morphological stability to a direct comparison with numerical simulations and an
against symmetry breaking, and the formation of long-lived unequivocal identification of the relevant physical and geome-
stacking defects within basically periodic arrangements. In trical parameters. Second, transparent alloys that freeze like metals7
addition, this dynamics can be influenced by instrumental have been used for many decades as model systems for in situ
characteristics and by the crystalline structure of the growing experimental studies of solidification patterns8–12. The shape of
solid—being a single- or a polycrystal or made of crystals of the solid growing from the liquid mixture can then be followed in

1
Sorbonne Université, CNRS-UMR 7588, Institut des NanoSciences de Paris, case courrier 840, 4 place Jussieu, 75252 Paris, Cedex 05, France. 2Access e.V., Intzestr. 5, 52072
Aachen, Germany. 3Laboratoire de Physique de la Matière Condensée, CNRS, Ecole Polytechnique, Institut Polytechnique de Paris, 91120 Palaiseau, France. 4Department
Metallurgy, University of Leoben, Franz-Josef-Str. 18, 8700 Leoben, Austria. 5Department of Metallurgical and Materials Engineering, Marmara University, 34854 Maltepe, Istanbul,
Turkey. 6Aix Marseille Univ, Université de Toulon, CNRS, IM2NP Marseille, France. 7School of Computing, Engineering, and Intelligent Systems, Ulster University, Northland Road,
Derry/Londonderry, Northern Ireland BT48 7JL, UK. 8European Space Agency, ESTEC – Research and Utilisation Group, Directorate of Human and Robotic Exploration
Programmes, Keplerlaan 1, 2201 AZ Noordwijk, Netherlands. ✉email: [email protected]

Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA
 S. Akamatsu et al.
2
real time over a large range of times and distances with flexible speaking, during the growth of nonfaceted crystals from the melt
and technically light means of optical microscopy. Transparent- —is primarily governed by the interplay of the moving solid-liquid
alloy solidification methods have evolved considerably over the interface with the diffusion fields of chemical species and heat in
last few decades4,13–17, thus strengthening their unique demon- the liquid1,2. The diffusion gradients—the very ones that are
stration power. In parallel, the continual development of time- responsible for the morphological instabilities of interest—create
resolved numerical-simulation techniques has enabled increas- thermosolutal density gradients in the liquid. In the gravity field,
ingly direct, quantitative comparison of experimental observations the density gradients can induce convective flow, which perturbs
with theory18,19. This synergy has already led to many new the solidification dynamics on a large scale (see, e.g., refs. 30,31 and
insights, with strong scientific impact in both materials science1,2 references therein; also see below). Microgravity in free-fall and
and the physics of nonequilibrium pattern formation3,20, and is orbiting facilities offers a unique way to suppress or drastically
also central to the scientific advances that will be reviewed here. damp out thermosolutal convection, whatever the alloy composi-
We will present first the scientific context of in situ solidification tion and the geometry of the experiment (see, e.g., ref. 31).
using transparent alloys, including the effect of convection in the Solidification patterns can then be observed under purely
liquid that may occur on ground. Second, the main innovative diffusion-controlled growth conditions. The in situ experimental
features of the TA and DECLIC-DSI apparatuses and the phase-field conditions and the corresponding observations can be more
models for numerical simulations will be described. Third, we effectively analyzed with the help of theoretical and numerical
report on representative results on multiphased growth patterns models.
in eutectic (lamellar-to-rod morphological transition, effect of a Diffusion-controlled crystal growth in nonfaceted alloys leads to
varying solidification velocity) and peritectic alloys (transient various pattern formation phenomena, depending first on alloy
coupled growth) and on single-phased cellular and dendritic characteristics. In a sufficiently dilute binary alloy a single-phased
patterns (interfacial anisotropy, curvature of the isotherms, solid forms. For thermodynamic reasons, the solid and the liquid
columnar/equiaxed transition). Finally, general conclusions and in contact do not have the same composition. This entails a
prospects are proposed in the “Outlook and Summary” section. redistribution of chemical species by diffusion in the liquid. A
nonlinear coupling of the diffusion gradients with the shape of the
advancing solid-liquid interface is responsible for the morpholo-
1234567890():,;

REVIEW gical instabilities of the growing solid. In contrast, the capillary

Preliminary remarks response of the solid-liquid interface prevents strongly curved
Most metals and some other compounds of various chemical deformations of the solid-liquid interface, which is a stabilizing
nature, such as salts, rare gases, and some organic molecules, factor. The so-called cellular or Mullins-Sekerka instability of the
crystallize from the melt without faceting, in contrast to the vast planar front above a critical value Vc of the pulling velocity has
majority of compounds21. On the atomic scale, nonfaceted crystals been predicted on the theoretical basis just described2. By
present a “rough” structure of the solid-liquid interface. The increasing V above Vc, cellular patterns and dendritic arrays form,
surface free energy (surface tension) of the solid-liquid interface is the space-time dynamics of which is sensitive to boundary and
weakly anisotropic, and the shape of the crystals, both at initial conditions. This complexity is common to a large class of
equilibrium and during slow growth, remains smooth, without nonequilibrium pattern-forming systems3,20. It motivates labora-
facets or sharp edges. The kinetics of attachment of atoms or tory research involving well-calibrated in situ experiments and
molecules at the interface (interfacial kinetics) is relatively fast, and time-resolved numerical simulations that can be compared
during growth, the interface can be considered locally at, or very quantitatively with each other. This general comment also holds
near, equilibrium. Based on general statistical-physics arguments, true in the case of eutectic and peritectic solidification, which
a criterion was proposed by K. Jackson in the 1960s, which states presents the particularity to deliver a multiphased solid.
that simply speaking, nonfaceted materials are such that their Large-scale solute redistribution gradients, along with tempera-
entropy of melting ΔSm (per atom or molecule) is very close to, or ture gradients, can trigger convective motion in the liquid phase
not much greater than the Boltzmann constant kB22. Using that during solidification. In ordinary conditions, natural convection in
criterion, Jackson and Hunt found a series of nonfaceted materials the melt perturbs both the temperature and solute concentration
made of small-sized, globular-shaped organic molecules that fields, which are no longer homogeneous on length scales on the
solidify from the liquid and form plastic-crystal phases7. The order of the size of the container. This entails a substantial
transparent compounds, such as succinonitrile (SCN), neopentyl- modification of the microstructure. This has been evidenced, in
glycol (NPG), and d-camphor (DC), which are the focus of the particular, by comparing experiments performed with the DECLIC-
present review, belong to this class of materials. The term “plastic” DSI apparatus on-orbit on the one hand and with a twin apparatus
refers here to the large ductility of the molecular crystals in on-ground on the other hand32–34: convection can induce the
question. Plastic crystals present a high-symmetry (cubic or appearance of localized patterns and superstructures or modify
hexagonal) lattice with well-defined positional order but have a the primary cell spacing distribution. Interestingly, forced or
rotational disorder of the molecules23,24. Their crystal structure has enhanced convection by centrifugal motion35 and electromag-
been studied in various compounds, including SCN25 and carbon netic forces36 (also see ref. 37) have been shown to trigger phase
tetrabromide (CBr4)26, which are among the most common separation or influence microstructural transitions. These effects
transparent analogs in solidification studies. Except for their lower are naturally absent in the microgravity conditions existing in
thermal conductivity, nonfaceted organic compounds present solidification facilities onboard the ISS.
physical properties relevant to solidification that are very similar to
those of metals to within scale factors. This property justifies the Methods and apparatuses
use of low-melting nonfaceted organic alloys as excellent model The multi-user TA program encompasses the study of (1) steady-
systems for fundamental solidification studies. state and transient multiphased growth patterns in eutectic and
Model alloys with low melting temperatures are particularly peritectic alloys and (2) the columnar-to-equiaxed transition in
suitable for the implementation of automated solidification single-phased alloys. The DECLIC-DSI setup was designed to study
instruments intended for being installed onboard microgravity the spatiotemporal dynamics of cellular and dendritic arrays. Both
facilities27–29. There is a fundamental motivation for carrying out apparatus (as well as the numerical simulations) use the
solidification studies in reduced gravity. The formation of directional-solidification configuration (also referred to as the
remarkable shapes and patterns during solidification—generally Bridgman method), during which the sample is displaced at an

npj Microgravity (2023) 83 Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA
 S. Akamatsu et al.
3

Fig. 1 The DIRSOL method in the TA apparatus. a Schematic representation. V: pulling velocity. θ: viewing angle. b In situ TA images during
directional solidification of a eutectic SCN-DC alloy (V = 0.007 µms−1). Top: detail of an oblique-view image (rescaled to 1:1 aspect ratio) of a
rod-like eutectic pattern, with DC rods appearing white and the SCN matrix dark59. Bottom: side-view image (liquid on top, eutectic solid on
bottom).

imposed velocity V in a fixed temperature field with a controlled is obtained (the imaging plane coincides with the image of the
axial temperature gradient G (axis z). In the experiments, the growth front through the liquid and the glass). The small aperture
displacement (also called “pulling”) velocity V typically ranges and a finely tuned inclination of the transmitted-light illumination
from 0.010 to 100 µms−1, and G between 10 and 80 Kcm−1, make it possible to select rays coming from one of the two solid
depending on the system under study. The pulling velocity V is phases only15. This also limits the astigmatism due to the plane
the principal experimentally accessible control parameter. In diopters—the TA optics includes an additional astigmatism
steady-state, the solidification rate is equal to V, and the solid- correction device for optimizing the sharpness of the images.
liquid interface stays, on average, at a constant position in the The effective resolution of about 3 µm is sufficient to study the
temperature field. Ideally, the isotherms remain flat, perpendicular overall dynamics of the structure. A numerical camera is coupled
to the pulling axis z, and stationary in the laboratory frame. In to the optics for real-time imaging. The TA setup also possesses a
many cases, this “frozen-temperature” assumption is essentially second camera for side-view observations (Fig. 1b). During
valid. In other cases, the motion or a distortion of the gradient eutectic growth experiments, this observation mode was used
field must be considered, as will be mentioned below. In the to follow the average position of the solidification front and to
experiments, the solidified alloys are contained within transparent, monitor its shape (essentially planar in Fig. 1b) along the x-axis.
glass-wall cartridges. That the position of the solidification front is The side-view imaging was the main mode of observation for
essentially constant in the laboratory frame allows continuous peritectic and columnar/equiaxed structures.
observation with a fixed optics. More detailed characteristics The TA apparatus was launched to the ISS in December 2017 in
specific to the TA and DECLIC-DSI instruments, respectively, as the first CRS (Commercial Resupply Services) flight of the SpaceX
well as the phase-field numerical-simulation method, are Dragon using a Falcon 9 launcher whose first stage was reused. It
presented below. was installed in the Microgravity Science Glovebox (MSG) facility
The design of the (multi-user) TA apparatus, developed by onboard the ISS in January 2018. The astronauts were in charge of
QinetiQ Space (now Redwire Space), was based on the DIRSOL the introduction of the cartridges into TA, in addition to the
method previously developed at INSP (Fig. 1a)15. Its principle is installation of the apparatus in MSG. The major parts of the
derived from the conventional thin-sample directional solidifica- operations are essentially automatic, with telescience (telemetry
tion technique. Flat-wall (fused silica) samples with a rectangular control of parameters, image frequency, etc.) used from the
cross-section (6 × 1 mm2 in the case of TA eutectic-alloy operation center (E-USOC, Madrid, Spain) permitting near real-
cartridges) and a length of several cm are filled under a protective time feedback. Three TA cartridges of SCN-DC-based eutectic
Ar atmosphere with the molten alloy prepared with purified alloys were used successfully onboard the ISS. Two of them were
compounds and sealed. For certain applications, a crystal selector processed during the SEBA (Solidification along a Eutectic Path in
at the cold end is used to grow a single eutectic grain, except for a Binary Alloys) campaign and a third one during the SETA
few sub-boundaries. In the apparatus, the temperature gradient (Solidification along a Eutectic Path in Ternary Alloys) campaign.
(G = 30–60 Kcm−1) is established between two metallic blocks Peritectic-alloy cartridges have been used for the METCOMP
separated by a 7-mm gap (also known as the adiabatic zone in the (Metastable Solidification of Composites: Novel Peritectic Struc-
scientific literature). Each of those blocks is made of two pieces tures and In-Situ Composites) program. Some preliminary results
(also called clamps) with independent thermal regulation and obtained during the CETSOL (Columnar to Equiaxed Transition
assumed to have good thermal contact with the cartridge walls. during SOLidification processes) campaign will also be presented.
The observation is made from the outside, in oblique incidence It can be noted that SETA/SEBA, CETSOL and METCOMP were
with respect to the solidification axis. Large working distance former MAP (Microgravity Application Promotion) programs of
optics with a small, tunable numerical aperture is used under dark- ESA. The TA results reported here have been obtained during TA
field observation mode (Fig. 1b). In the TA apparatus, the campaigns spanning between the years 2018 and 2022.
inclination angle θ of the optics is fixed to a value (43.6°) close The DECLIC facility is a compact, multi-user facility for
to an optimum value such that a sharp image of the entire pattern conducting experiments in the fields of fluid physics and materials

Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA npj Microgravity (2023) 83
 S. Akamatsu et al.
4

Fig. 2 The DECLIC-DSI. a Optical diagnostics (schematics). Examples of images using the different diagnostics (SCN-0.46 wt%DC alloy; G = 12
Kcm−1; microgravity experiments). b Top-view image of a dendritic pattern (V = 8 µms−1); c Interferometric observation of a dendritic pattern
(V = 1 µms−1); d Side-view observation (V = 1.5 µms−1). Color online.

science, and, more generally, on transparent media within the ISS A major step forward in the applications of phase-field models
environment. The main part of DECLIC is common to all for solidification was the development of the thin-interface limit
experiments and contains electronics (for regulation, data for pure substances45, binary alloys46,47 and eutectic alloys48. With
acquisition and management, communication, etc.) and some this method, it is possible to perform quantitative (accurate)
optical resources (laser, optics, cameras). Different inserts that simulations using a thickness of the diffuse interfaces that can be
contain elements specific to each experiment complete the orders of magnitude larger than the physically realistic value. The
description of the facility. The DSI is the insert dedicated to associated gains in computational performance have made it
solidification studies to image the dynamics of cellular and possible to perform realistic simulations, both in two and three
dendritic arrays—for a detailed description, see ref. 38. In the dimensions, on the scale of the microstructure.
DECLIC-DSI, a cylindrical, quartz-wall cartridge (inner diameter of To perform the simulations, in-house codes are commonly
1 cm; typically with a solidification length of 10 cm) is inserted in a written in C or FORTRAN, and modern methods are used for
Bridgman furnace (V = 0.1 to 30 µms−1). A system of volume computational speedup (preconditioning, GPU parallelization,
compensation accommodates the density variations associated multi-grid methods). These codes typically run on computational
with phase changes in the alloy. This configuration permits the
clusters on the laboratory scale (a few GPUs or a few hundred CPU
study of extended patterns from initial stages up to steady-state
cores) within processing timeframes of hours to days. This allows
growth conditions being established. Temperature gradients G
for a detailed and direct confrontation between experiments and
ranging from 10 to 30 Kcm−1 are imposed between the hot and
cold zones6,39. simulations.
A schematic of the optical diagnostics is given in Fig. 2a. The
crucible is equipped with a circular, optically flat glass window at Eutectic growth
the bottom and a lens immersed in the melt at the top. The axial Eutectic alloys present the specific case of solidifying into a
observation mode takes advantage of the complete transparency multiphase solid with self-organized composite microstructures2.
of the experimental cartridge. The light coming from the axial In a binary eutectic alloy, the liquid and two distinct solid phases,
LEDs is transmitted through the solid and the liquid, crossing the generically noted as α and β, can coexist at equilibrium at the
interface, the image of which is formed on a CCD camera. On the (unique) eutectic temperature TE. Their respective concentrations
same axis, a Mach-Zehnder interferometer (He-Ne laser) produces CE, Cα and Cβ are fixed (Cα < CE < Cβ). During directional
interferometric images used for high-resolution, three- solidification, the two solids can grow simultaneously and form
dimensional shape reconstruction14,40. Additionally, transverse quasi-planar, two-phased solidification patterns. The coupled-
observation optics provides side-view images of the interface on growth dynamics is mainly governed by solute diffusion in the
a millimeter scale39. liquid on the scale of the (interphase) spacing λ, and capillary
The main instrument monitoring was operated from the
effects. In steady-state, coupled-growth patterns basically exhibit
CADMOS center, the French User Support and Operation Centre
banded or hexagonal arrangements, thus delivering lamellar or
(USOC) in Toulouse, France. Taking advantage of the provided
rod-like microstructures in the solid. As shown by Jackson and
telescience capabilities, scientists were given the possibility to
follow the solidification pattern in near real-time conditions and to Hunt in the 1960s, the average spacing remains close to a
remotely control the experiments. characteristic length λJH, which varies as V−1/2 with the pulling
For about 20 years, the phase-field method has become the velocity V9,10. Rod-like and lamellar patterns are morphologically
method of choice for modeling and simulation of solidification, stable for λ values ranging within finite intervals, the limits of
crystal growth, and microstructure evolution41–45. Its advantage is which approximately follow the same V−1/2 scaling as λJH49.
that it avoids front tracking by representing interfaces as steep but Outside their stability domains50, lamellar and rod-like patterns
smooth (i.e., diffuse) variations of one or several scalar fields, undergo secondary, symmetry-breaking instabilities and complex
which are the so-called phase fields. The equations of motion for spatiotemporal phenomena51,52. Anisotropic behavior due to
the phase fields are coupled to the transport equations for all crystal orientation effects in eutectic alloys has also been carefully
relevant quantities (heat, solute, momentum) following the studied17,53,54. In the current review, we assume an essentially
principles of out-of-equilibrium thermodynamics. isotropic system.

npj Microgravity (2023) 83 Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA
 S. Akamatsu et al.
5

Fig. 3 Lamellar-to-rod transition observed in situ. Directional solidification of a eutectic SCN-DC alloy with tilted isotherms
(V = 0.007 µm s−1) in microgravity (TA apparatus)59.

observed for markedly departing phase fractions (typically

η < 0.25). Due to mass conservation, η is determined by the lever
rule in the phase diagram10. A lamellar-rod transition is thus
expected to occur in a given system upon varying the
concentration (and hence varying η). This has been essentially
confirmed experimentally57 and numerically58, but complex,
unsteady dynamic processes involving hybrid shapes were mostly
revealed.
When turning attention to an alloy with fixed composition, it
has been shown numerically that rod-like and lamellar stability
intervals overlap with each other. Within the bistable region of
Fig. 4 Lamellar-to-rod transition in a numerical simulation. Three- parameters, a coexistence between well-ordered lamellar and rod-
dimensional phase-field simulation of a model alloy with symmetric
phase diagram and volume fraction η = 0.34. The tilt of the like patterns is shown to be possible. In addition, numerical
isotherms with respect to the pulling direction is 4°. Zero-gradient simulations suggest that the transformation of lamellae to rods
boundary conditions are applied to phase and concentration fields occurs in a propagative way. As reported shortly, the main
at the front and rear (corresponding to the presence of the glass observations made during the first SEBA campaign confirmed this
plates in the experiments) and periodic boundary conditions on the scheme and brought additional pieces of information59.
sides. Lamella form at the rear side and slowly elongate, while Rod-like patterns in the SCN-DC eutectic system have been
pinchoff of new rods occurs periodically at the lamella terminations. studied previously on ground using the DIRSOL technique60–62.
Color online. Systematic experiments consisting of observing the response of
steady-state patterns to stepwise, downward or upward changes
Space experiments permit the study of the morphological of V were used to identify the upper and lower stability limits of
stability of lamellar, rod-like, or more complicated patterns and hexagon patterns (rod splitting and rod elimination instability
their dependence on the alloy concentration C0 ranging over a thresholds, respectively)62. Observations made during the SEBA
finite interval about CE55,56. For an off-eutectic alloy (C0 ≠ CE), there campaign59 were in full agreement with previous conclusions—
exists a diffusion layer of extension ld = D/V (with D being the the high-level performance of the TA apparatus was also attested
solute diffusion coefficient in the liquid), which can reach several (see Fig. 1b). In practice, a stretching of the pattern (thus an
millimeters for solidification velocities below 1 µms−1. That layer, increase of λ) in the transverse direction was also observed, both
which superimposes the λ-scale modulation of the diffusion field, during on-ground and space experiments. This phenomenon was
has little influence on the coupled-growth dynamics but can due to a mild curvature of the isotherms, as imposed by different
trigger convection in the liquid. The composition and thermal thermal conductivities of the cartridge walls and the alloy. At long
fields in the liquid are then perturbed on a large scale, and vastly times, the increase of the spacing was counterbalanced by rod
different morphologies can be seen simultaneously in the same splittings, and the system converged toward a quasi-steady
sample57. spacing distribution (with λav ≈ λJH)61.
Two SEBA cartridges and one SETA cartridge were used In order to observe straight lamellae and their mode of
successfully with the TA apparatus onboard the ISS. The two transformation into rods on long timescales, a mild transverse
main questions that were addressed are described as follows. The component was imposed on the temperature gradient. This was
first question (SEBA) concerns the so-called lamellar-to-rod made possible by the four hot- and cold-clamp temperatures
transition: in simple terms, in a given system, is it possible to being regulated independently. The isotherms were inclined by an
induce a morphological transformation of rods into lamellae and/ angle of about 6.3 ± 0.3°. This entailed a global drift of the
or the reverse by changing the pulling velocity? The second coupled-growth pattern in the direction perpendicular to the
question (SETA) relates to the response of rod-like patterns to an sample walls63. A large lamellar domain was observed to grow
imposed variation of the pulling velocity and whether there exists (Fig. 3). The lamellae were stabilized in contact with the sample
a spacing and/or order selection mechanism induced by such wall. At the free ends, the lamellae underwent a breakup
temporal “velocity ramps”. instability and emitted new rods in a loosely regular way. The
process was propagative, but in the configuration with tilted
The first study presented here concerns the lamellar-to-rod transition isotherms, the breakup instability was slower than the geometrical
in binary eutectics. Eutectic solidification microstructures are drift of the pattern. This permitted the extension of a stable
often either of the rod-like or the lamellar type. Lamellar patterns lamellar domain. A long-lived coexistence of straight lamellae and
prevail in alloys with nearly equal phase fractions (η ≈ 0.5, where η rods separated by a sharp domain wall was thus evidenced. There
is the volume fraction of the β phase), while rod-like patterns are was strong evidence that, at low velocity, a lamellar pattern of

Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA npj Microgravity (2023) 83
 S. Akamatsu et al.
6
infinite extension would remain absolutely stable64.
This scenario was tested by three-dimensional phase-field
simulations with the model that was developed in ref. 48 and
used previously to study the stability of rods58 and lamellae65, as
well as the effect of a transverse temperature gradient63. A model
phase diagram that was completely symmetric with respect to the
exchange of the two solid phases was used rather than the SCN-
DC phase diagram in order to maximize computational efficiency.
An example of a simulation with a tilt angle of 4° that was started
from an initial condition with a random distribution of α and β
phases in the solid is shown in Fig. 4. As in the experiments,
lamellae start to form on one side of the system and slowly
propagate toward the other side, while they are shortened by the
pinchoff of new rods at the lamella terminations. A systematic
variation of the alloy concentration, pulling velocity, and tilt angle
yielded the following main results. (1) For a given velocity and tilt
angle, lamellae propagate across the system above a critical
volume fraction (or concentration), below which only rods are
present. This critical composition depends on V. (2) For a given
composition and tilt angle, lamellae are formed below a critical
pulling velocity which depends on the composition. (3) The critical
composition or velocity also depends on the tilt angle. These
results are fully consistent with the experimental observation and
demonstrate that this mechanism of the rod-to-lamella transition
is generic and not limited to the SCN-DC alloys system.
There are still many pending questions, in particular concerning
the precursory instability mode, which resembles a Rayleigh-
Plateau instability of a liquid jet, and the fact that the stability of
the lamellae does not follow the ordinary V−1/2 scaling. Finally,
additional observations were made in a second SCN-DC cartridge
in the TA apparatus, following a nearly similar experimental
protocol but without using the crystal selector. Several eutectic
grains were present during solidification, and smooth but clear
crystallographic effects were observed. Ongoing numerical
simulations are expected to bring more detailed information on
those questions.

The second study concerns the dynamics of eutectic growth under

imposed velocity transients. Rod-like patterns with local hexago-
nal order most commonly contain a large number of topological
defects, mostly heptagon/pentagon pairs in chain-like arrange-
ment delimiting small hexagon domains. The SETA experimental
program was inspired first by an observation reported in an Al-
Al3Ni eutectic alloy66, bringing experimental evidence that the
Fig. 5 Rod-like eutectic growth patterns with varied solidification
degree of order (or disorder) of rod-like eutectic patterns can velocity. Directional-solidification of a eutectic SCN-DC alloy with a
differ strongly depending on whether a given V value is reached third compound (NPG) in microgravity (TA apparatus). a Nominal
upon either increasing or decreasing the pulling velocity. A (dots) and actual (red line) growth velocity as a function of time.
similar conclusion was drawn in ref. 58, as well as in a recent b Images (details) of the rod-like eutectic structure (see labels in a).
phase-field simulation study67. However, by comparing those c Average rod spacing λav as a function of the actual growth velocity
studies, it appears that the way order or disorder is reached during the same experiment (black dots; the black line is a guide for
depends strongly on the system and the geometry of the the eye). Arrows: time variation. Large red disks: average spacing λqss
experiment. In brief, a practical but complex question was to measured during (separate) quasi-steady experiments. Red dotted
determine whether transient dynamics during a solidification line: λJH. Color online.
experiment with a controlled temporal variation of the pulling
rate can lead to a rod-like pattern with improved quality, that is, a Upon decelerated growth, the average spacing λav increases,
better organized hexagonal arrangement. as expected from the approximate λJH scaling. However, the
A series of transient solidification experiments with different spacing adjustment is slow and lags behind the scaling law
acceleration and deceleration profiles were performed on ground predictions, which is primarily due to the increasingly slow
and onboard the ISS68 using a binary eutectic SCN-DC alloy and a dynamics of the rod elimination process. The slow-V, large-λ
univariant (two-phase) ternary eutectic SCN-DC alloy with a small pattern obtained at the end of the decelerating stage experi-
amount (0.5 wt%) of the third compound (NPG)69. Positive or ences a re-acceleration phase without having established a stable
negative (temporal) V-ramps, that is, accelerated or decelerated steady-state distribution. Upon accelerated growth, λav
V(t) profiles with stepwise, linear and exponential variations, were decreases, and the spacing evolution nearly follows the scaling
imposed. This made a huge number of observations that are law predictions, which is due to the fast rod-splitting dynamics. It
currently under analysis, mainly in terms of the evolution of the appears thus that an imposed V(t)-profile program with alternat-
rod-spacing distribution (histogram) and the degree of hexago- ingly negative and positive V-ramps results in hysteresis-like
nal order. The following conclusions have been tentatively drawn dynamics that can be clearly visualized in a λav(V) diagram. The
—they are illustrated in Fig. 5. width of the hysteresis loop and also the improvement (or

npj Microgravity (2023) 83 Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA
 S. Akamatsu et al.
7
regular eutectic ones. Such PCG patterns were found to depend on
various factors (see, e.g., refs. 72,73), including convection ahead of
the front74,75. To suppress the latter, a first microgravity investiga-
tion of PCG was performed in hypo-peritectic Cu-Sn alloys onboard
the ISS (M. Rappaz, private communication). Indeed, PCG micro-
structures were found, but the presence of large gas pores—one of
the most prohibitive disturbances during solidification studies76—
made the analysis difficult, if possible at all.
Besides the requirement that the primary phase is morpholo-
gically stable, little is known about the conditions under which
PCG is possible and about the transient that leads to its
emergence. Experiments on a transparent peritectic system under
microgravity offer an unprecedented possibility to study these
questions further.
The transparent organic peritectic model system Tris(hydrox-
ymethyl)aminomethane-Neopentylglycol (TRIS-NPG) was used for
in situ microgravity studies (METCOMP program) in the TA
apparatus77,78. Three TRIS-NPG alloys of nominal concentrations
(50, 52, and 53 mol%NPG, respectively) close to the peritectic
concentration Cp have been processed and 18 individual
solidification experiments were realized. A typical experiment
was performed (after a 2–3 h of annealing time) at constant
velocity (V = 0.04–0.18 µms−1) in a temperature gradient G of
Fig. 6 Peritectic coupled growth. Formation of PCG during about 40 ± 10 Kcm−1. The solid-liquid interface was observed in
directional solidification of a near-peritectic TRIS-51 mol%NPG alloy side view. Thanks to a slight transverse temperature gradient, the
in microgravity (TA instrument, ISS). The side-view images were structure of the solidification front could be imaged by using focus
taken with middle focus except for the insets, which were taken with series. The challenge for studying PCG patterns in the TRIS-NPG
front focus. The photograph in (a) was taken 1.5 h after a pulling
velocity of 0.11 µms−1 was activated. The snapshots in panels (a–f) system is that the primary and peritectic phases have nearly equal
were evenly spaced in time, with a time interval of 1 h. refraction indices. As indicated below, the distinction between the
two phases could nevertheless be made by observing their
different dynamical behaviors.
degradation) of the hexagonal order depend on the shape of the During the initial annealing stages (V = 0), two processes
V(t)-profile. occurred simultaneously: (1) the grains of the initial (primary-
phase) polycrystal coarsened, and (2) inclined grain boundaries
Peritectic growth migrated toward the solid-liquid interface in the direction of the
gradient axis (temperature gradient zone melting or TGZM
Peritectics are found in many industrial alloys such as steels and
process1,2,79). Figure 6 shows an example of how PCG formed
other Fe-based alloys, Cu- and Ti-alloys, some magnetic materials,
during solidification. At the moment the pulling velocity was set to
and superconductor alloy70. A peritectic transition involves a liquid
V = 0.11 µms−1, the α polycrystal was made of elongated grains
and two solid phases, which coexist at the peritectic temperature
(as in Fig. 6a). After a certain time of pulling, an elongated crystal
Tp, the concentration of the liquid (CL), that of the primary phase α (blue band in Fig. 6b) was observed to grow close to the back wall.
(Cα) and that of the secondary (also called peritectic) phase β (Cβ, We could get evidence that this new crystal appeared at the
or Cp) being such that Cα < Cp < CL. The peritectic growth interface at a time when the temperature of the front was below
dynamics depends strongly on whether the average concentra- the peritectic temperature and was identified as being of the
tion of the alloy C0 is larger or smaller than Cp and on peritectic phase β. It propagated in the thickness of the sample
experimental conditions. while leaving some α crystals growing simultaneously (red patches
During directional solidification, the primary phase commonly in Fig. 6c). In other words, the two (primary and peritectic) phases
grows with dendritic morphology, and the peritectic phase β were observed to grow in a coupled way. This is, to the best of our
forms at the rear of the dendrite tips, in contact with the liquid at knowledge, the first real-time observation of a transient solidifica-
the peritectic temperature Tp (peritectic reaction). There is also a tion regime leading to PCG.
strong driving force for the secondary phase to continue to grow An interesting morphological evolution from large rods to
below Tp so that the primary phase can be eventually absent from lamellae (possibly drifting or with a zigzag shape) was observed
the final bulk microstructure. (Fig. 6c–f). This occurred while the front was recoiling toward
For solidification velocities that are slow enough to avoid cellular lower temperatures along the temperature gradient axis. The
instability, the primary solid initially grows with a planar α-liquid change of the microstructure in the solid can also be noted in the
interface at a temperature that continuously decreases with time last image (Fig. 6f). During the following stages (not shown), a
due to solute redistribution. For temperatures below Tp, the liquid decrease of the α phase fraction was observed until complete
becomes metastable against the nucleation of the peritectic phase disappearance. Thus, in brief, during the recoil of the solidification
β. For a hyper-peritectic alloy (Cp < C0 < CL), this may lead to a front, we observed a transition from α-phase to β-phase growth
steady-state growth of the β phase. For hypo-peritectic alloys via a transient α-β PCG regime. A more accurate analysis of the
(Cα < C0 < Cp), no stable conditions for planar growth of single- recoil of the front still remains to be made. It obviously
phased—either α or β—solids exist. Two complex solidification corresponds to an unsteady solute redistribution process, which
regimes have been observed and analyzed in previous studies. In might involve impurity pile-up, a plausible large-scale composition
the first regime, thin α and β crystals grow laterally and form thin gradient in the cartridge, and a possibly temperature-dependent
bands that alternate in the direction of the temperature gradient solute diffusion coefficient in the liquid80. Nevertheless, the actual
(banded growth; see ref. 71 and references therein). In the second solidification dynamics was favorable to the occurrence of a PCG
regime, the two solid phases grow simultaneously, thus forming sequence in the course of a long-lasting recoil of the solidification
peritectic coupled-growth (PCG) microstructures that resemble front.

Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA npj Microgravity (2023) 83
 S. Akamatsu et al.
8

Fig. 7 Dendritic patterns in DECLIC-DSI. In situ directional solidification of a SCN-0.46 wt% DC alloy (V = 1.5 μms−1; G = 12 Kcm−1) under
microgravity (DECLIC-DSI). Raw images (a) are analyzed to determine each dendrite center and to build Voronoi skeletons (b) that are used to
calculate local spacings (scale in µm) and draw maps of primary spacing (c)89. Color online.

The front recoil complicates the theoretical analysis of PCG. In individual tip shape. Microgravity experiments aim at clarifying
fact, similarly to what was observed in ground experiments on Cu- that dynamics, essentially free of convection, thus allowing
Zn81,82, PCG is initiated before the solute boundary layer has fully identification of the main phenomena and key parameters that
built up. Further evolution of this layer over time therefore creates enter into play during diffusion-controlled growth pattern
growth conditions for the coupled-growth structure that vary on a formation. Such experiments also contribute to the creation of a
slow time scale and ultimately leads to the termination of PCG in benchmark experimental database which is required for the
the present experiments. This non-steady nature is a general improvement and validation of numerical simulations.
feature of PCG in alloys with large solidification intervals. Two series of DECLIC-DSI microgravity experiments that are
Regarding the initiation of coupled growth, in comparison to reported here were performed using dilute SCN-DC alloys of
eutectic coupled growth, there are both similarities and differ- concentration well outside the coupled zone (0.24 and 0.46 wt%
ences. In both cases, coupled growth is initiated starting from DC, respectively). The analysis of the first series of experiments
single-phase growth when the second phase emerges at the front, shed light on complex thermal conditions within the experimental
either by nucleation or by growth from a seed deep within the setup39. It was evidenced that heat transport, including latent heat
solid. In eutectics, coupled growth emerges either by the rejection inside the sample, must be considered in analyses and
instability of a finger of the second phase that propagates along numerical simulations. The (most commonly nonplanar) shape of
the front83 or by the immediate emergence of a two-phase the isotherms depends on V and the position along the adiabatic
pattern84, and the initial spreading of the two-phase pattern is zone. Thus the problem cannot be described using the classical
followed by coarsening. For the transparent peritectic system frozen-temperature approximation (stating that the thermal field
studied here, the dynamics of initiation is rather slow, and the is fully fixed externally). On this basis, a new modeling approach
structure refines rather than coarsens after its emergence. This was developed, and a 3D phase-field model was coupled with a
mechanism for the emergence of peritectic growth is very time-dependent calculation of the thermal diffusion in the
different from the successive nucleation and overgrowth that adiabatic zone86. Taking such non-ideal thermal conditions into
has been postulated for Fe-Ni in previous works85. account was key to improved agreement between experimental
To make further progress, the dynamics of the transient leading and simulated pattern dynamics and characteristics. It will be
to PCG needs to be studied in a more systematic way. This will highlighted below how far several factors of “non-ideal” solidifica-
certainly also shed new light on the question of which alloys and tion, not only unfrozen thermal fields and macroscopic interface
under which conditions PCG can occur. Three-dimensional phase- curvature but also crystal orientation and polycrystallinity,
field simulations of peritectic growth would be very useful to intervene in the cellular and dendritic pattern dynamics. Such
supplement the experiments. deviations from the minimal, diffusion-controlled growth model
are intrinsically attached to the elaboration of materials by
solidification in bulk configuration. It is imperative to identify
Cellular and dendritic growth and analyze their impacts in terms of microstructure character-
In directional solidification of a (semi) dilute alloy, increasing the istics and dynamics.
pulling velocity above the threshold value Vc leads to a It is well recognized that the shape of deep-cell and dendritic
morphological instability of the planar, single-phased interface1,2. patterns is strongly influenced by the orientation of the growing
As mentioned above, a fingering, cellular microstructure first crystal with respect to the main solidification axis and the
develops for V not too far above Vc. For higher V values, deep cells isotherms87,88. In contrast, the consequences on large scales in
undergo a sidebranching instability and evolve toward a dendritic spatially extended arrays in bulk samples remain essentially
array pattern. The growth dynamics of both cellular and dendritic unexplored. In DECLIC-DSI experiments, great care has been taken
arrays is complex and involve various length scales that are to be to select single crystals with a [100] axis of the SCN crystal lattice
characterized. It is common to start with the most evident pattern (the growth axis of freely growing dendrites) parallel to z –to
characteristic size, namely the primary spacing λ, which corre- within a few degrees (< 3°). This led to unprecedented
sponds to the distance of repetition in a periodic structure, and to observations of large arrays of (essentially) axial cells and dendrites.
proceed with the analysis of sidebranching characteristics and The larger concentration of the alloy (SCN-0.46 wt%DC) resulted in

npj Microgravity (2023) 83 Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA
 S. Akamatsu et al.
9

Fig. 8 Drifting-cell grains. a Primary spacing and drifting velocity of the cells as a function of their distance from a divergent sub-boundary
between two subgrains (SG1 and SG3, respectively), as indicated in the top-view image. b Time evolution of the spacing profile over 8.1 h (the
experimental data points were fitted to smooth curves). Alloy: SCN-0.24 wt%DC. V = 2 μms−1. G = 19 Kcm−1. Color online.

well-developed dendritic patterns at rather low velocities (Fig. 7). dynamics near the container walls84. Ongoing analysis aims at a
We have developed several image analysis methods to achieve a deeper understanding of the influence of sub-boundaries on the
robust identification of each dendrite position and size during spatiotemporal organization of the array structure, as well as other
solidification. Figure 7 illustrates the analysis process performed on dynamic phenomena such as sidebranching and the dependence
a raw image to extract primary spacing values. In this example, a of dendrite tip shapes on growth conditions.
major effect of the macroscopic interface curvature on the primary Complex phenomena also occur along the sub-boundaries. In a
spacing was identified (Fig. 2d)89. Such a macroscopic curvature recent work, Song et al. 95 report that during polycrystalline
tended to induce a radial drift of the pattern, which the anisotropic growth, a sub-boundary often adopts a wavy shape on the scale of
growth of dendrites resists. In addition, continual branching and λ (see Fig. 8a). A few cells belonging to one grain can eventually
elimination events prevented the formation of a well-ordered invade an adjacent grain, thus ending up with an intertwining of
pattern in steady-state. the two crystals. Systematic numerical simulations over a wide
Cellular patterns with disordered hexagonal arrangement were range of grain misorientations bring evidence for the general
formed at lower velocity for the less concentrated alloy40,90,91. relevance of this experimental observation. Those results funda-
Interestingly, not only steady-state but also oscillatory patterns mentally alter the traditional view of well-separated cellular/
were observed. A breathing-mode oscillation was identified, dendritic grains in three dimensions.
sometimes with a local coherence that is not generalized,
however, over long distances due to the large-scale disorder of The columnar-to-equiaxed transition in dendritic growth
the pattern. An in-depth phase-field study92 led to a detailed In metal alloy castings, columnar and equiaxed grains of the
description of the stability limits of steady-state and oscillatory primary phase are common. In practice, columnar patterns
patterns in a complete morphology diagram. The oscillations were designate dendritic arrays that usually grow during directional
also found to be sensitive to the misorientation of the crystal91,93. solidification, as they have been studied with the DECLIC-DSI.
A criterion was proposed to estimate a critical misorientation Equiaxed grains nucleate in the liquid or grow from detached
angle above which oscillations are systematically damped out. In primary-dendrite arms ahead of the main solidification front due
addition, it was possible to extract the complete cell tip shape to constitutional supercooling. The so-called Columnar to
evolution during the oscillation cycle from high-resolution Equiaxed Transition (CET) describes a change in local grain
interferometric images40. The shape variation is associated with structure that occurs in multicomponent alloys on a macroscale
an evolution of the concentration field, inaccessible experimen- (the scale of the container) of a casting. It results in a textural
tally but mediating the diffusive interactions between the cells. change in the microstructure from elongated (columnar) to
Full support was gained from 3D phase-field simulations that symmetrical (equiaxed) grain morphologies1,2. A complete CET
showed that transverse diffusive solute fluxes between neighbor- occurs during transient conditions—varying external solidification
ing cells are essential in the oscillation dynamics. conditions, changes of thermal and solutal fields—that favor the
A higher degree of complexity was induced in large single growth of equiaxed grains in a sufficient number and size so that
crystals containing a few low-misorientation grain boundaries. the progression of the initial columnar dendritic front is blocked.
Those “sub-boundaries” delimit subgrains with slightly converging In addition to the extension of the constitutionally supercooled
or diverging [100] crystal lattice axes—the ordinary growth liquid (and its evolution), important factors determining the CET
direction of free dendrites (Fig. 8a). Due to the interfacial are the nucleation rate and distribution of new crystals and/or the
anisotropy, cellular patterns in non-axial grains drift laterally. ability for solid fragments to detach from the columnar mushy
Cell-spacing measurements revealed markedly different effects of zone and grow as equiaxed grains. Taking external conditions,
divergent and convergent sub-boundaries on the spatiotemporal nucleation and alloy parameters into account, Hunt96 proposed a
evolution of the patterns94. In brief, close to a divergent sub- simple criterion based on blocking fraction of equiaxed grains to
boundary (Fig. 8b), the drifting motion of the cells away from each derive a microstructure map depending on thermal gradient G
other induces a peaked, nonuniform spacing distribution that and solidification velocity V. Explained simply, columnar grains are
extends over typically twenty cell spacings. In contrast, a favored by intermediate to high-temperature gradients and
convergent sub-boundary induces a localized decrease of λ due slower growth rates. Equiaxed grains are favored by low-
to cell elimination. temperature gradients and higher solidification or growth rates.
On a larger scale, drifting cell patterns exhibited a complex However, it is also known that convection in the liquid, and
spacing adjustment mechanism due to a sink-and-source buoyancy effects as well, have a strong influence on the evolution

Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA npj Microgravity (2023) 83
 S. Akamatsu et al.
10
a uniform liquid, the system will be cooled down continuously to
induce the growth of equiaxed dendritic grains (in isothermal
mode). In situ observation of grain growth in bulk samples will
provide the opportunity to characterize the multiple grain
interactions that determine the final grain structure. Characteriza-
tion of individual grains will also be performed (global grain
shape, internal spacings, evolution and coarsening), and specific
attention will be given to the selection of the dendritic growth
direction as the chosen alloy compositions are similar to those of
CETSOL-1. Attention will be paid to stereological corrections that
Fig. 9 Columnar-to-equiaxed transition. Directional solidification are needed to account for the side-view projection imaging of the
of an NPG-30wt.%DC alloy in microgravity (TA apparatus). The three-dimensional structures99. The experimental results will be
pulling velocity was increased from 14 µms−1 (a) to 32 µms−1 (b). compared to a volume-averaged model of equiaxed growth,
Vertical dimension of the images: 5.1 mm. In this figure, the images including a suitable grain nucleation model100.
are presented with the solidification axis z (that is, the axis of the
temperature gradient) horizontal (the liquid is on the right, the solid
on the left). OUTLOOK AND SUMMARY
of equiaxed grains (distribution of nucleation sites, motion of It is well known and widely experienced within the scientific
dendritic fragments). Obtaining reference observations with community that there is a long lapse of time between the
markedly reduced gravity effects is therefore key to gaining a moment when a microgravity experiment campaign in the ISS is
better understanding of the CET dynamics. defined and when it is actually carried out. As hopefully
Studies on CET in microgravity have been performed previously demonstrated above, this delay is actually useful to carry out in-
using metallic alloys97, where it was shown that a CET can occur depth on-ground experiments. This research has already yielded
either as a sharp or a progressive phenomenon. Modeling numerous important scientific results and has helped to refine the
suggested that a progressive CET occurs by simultaneous design of the experiments for an optimal use of the microgravity
columnar and equiaxed growth, the nucleation rate of equiaxed resources.
dendrites being an important factor. However, for obvious The need for solidification microgravity experiments with in situ
reasons, the analysis of microgravity experiments with the metallic diagnostics, including real-time imaging and monitoring, as well
system was performed by ex situ observations of cross-section as telescience and remote control of experimental parameters, is
micrographs. Model transparent alloy experiments offer the key evidenced. Technological innovation and transfer for the design of
advantage of in situ observation of the microstructure develop- the TA and DECLIC-DSI instruments is also a key factor for
ment, which, in the context of CET, is required to directly attest to obtaining accuracy and novelty. In this context, the long-term
the absence of convective transport or sedimentation/floatation of opportunity given by the European (ESA), American (NASA),
equiaxed grains. French (CNES), German (DLR), and Austrian (FFG-ASAP) space
During the CETSOL campaigns carried out onboard the ISS with agencies to the scientific partners for developing the research
the TA apparatus, the binary NPG-DC system within a hypoeutec- projects presented in this review has been highly beneficial.
tic composition range, that is, mainly in the single-phased NPG Major scientific objectives that concern morphological transi-
region of the phase diagram, was used as a model alloy. The tions, interfacial-anisotropy effects, transient processes and
cartridges were of the same dimensions as for SEBA and SETA nucleation-and-growth phenomena and their interaction with a
experiments, except for a substantially larger inner thickness competition between diffusion and convection during solidifica-
(6 mm), making them suitable for the observation of nucleation tion have been reached or are about to be reached in the next few
and growth processes in a bulk container. Simply speaking, three years. Many new phenomena have been observed directly with
groups of questions were defined, along with the design of time and space resolutions that are allowed by the use of model
dedicated experiments. The first one is related to the occurrence transparent alloys. This research has had an impact on our
of columnar and equiaxed microstructure as a function of the fundamental understanding of nonequilibrium pattern formation
thermal gradient, the pulling velocity and the alloy concentration and has increased our knowledge of engineering-directed aspects
(performed as part of the CETSOL-1 experiment). The second of great relevance for alloy casting and crystal growth. A major
group of questions, to be addressed during future experiments progress is shown with the quantitative match between carefully
(CETSOL-2), will be focused on obtaining observations close to performed experimental observations and accurate numerical
casting conditions (that is, with a low-temperature gradient) but simulations and theoretical concepts. The increasing use of phase-
without convection. The third aspect concerns crystal orientation field models and the upscaling of optimized simulation models
effects and will be investigated in detail in both experiments. are certainly some of the most salient outputs of solidification-in-
During CETSOL-1, different values of the thermal gradient were microgravity research and opens interesting perspectives for the
applied, and stepwise increasing pulling velocities were imposed modeling of casting and advanced manufacturing processes.
while observing the evolution of microstructures in a fixed field of New phenomena or long-standing questions are to be
view (side-view mode). The compositions of the NPG-DC alloys addressed, which could benefit from the accumulated knowledge
were chosen such that within the interval of three samples, a in solidification—including rapid solidification, melting processes,
change of crystallographic orientation from <100> to <111> was the effect of impurities and trace elements, and the solidification
expected98. Figure 9 shows a preliminary result from CETSOL-1. of multicomponent alloys. This is of great importance, in full
Here, a CET was induced by increasing the pulling velocity of the awareness of a growing need for innovation in energy-saving and
cartridge. The analysis of the ensemble of observations is an recycling processes to be implemented in industry—on Earth or
ongoing work. in space.
During CETSOL-2 (to be performed tentatively in 2024), the
central area of the sample will be placed at an (ideally) uniform
temperature. The temperatures of the so-called hot and cold DATA AVAILABILITY
blocks will be identical, or a small temperature gradient will be Data sets relevant to the present study are available from the corresponding author
applied. Starting with temperatures above the liquidus, thus from upon reasonable request.

npj Microgravity (2023) 83 Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA
 S. Akamatsu et al.
11
CODE AVAILABILITY 30. Beckermann, C. Modelling of macrosegregation; applications and future needs.
The code for phase-field simulations is available from the corresponding author upon Int. Mater. Rev. 47, 243–261 (2002).
request. 31. Roósz, A. & Tomolya, K. (eds) Solidification and Gravity VI, Vols. 790–791. Mate-
rials Science Forum (Trans Tech, 2014).
32. Jamgotchian, H. et al. Localized microstructures induced by fluid flow in
Received: 15 December 2022; Accepted: 17 September 2023; directional solidification. Phys. Rev. Lett. 87, 166105 (2001).
33. Mota, F. L. et al. Convection effects during bulk transparent alloy solidification in
DECLIC-DSI and phase-field simulations in diffusive conditions. JOM 69,
1280–1288 (2017).
34. Bergeon, N., Reinhart, G., Mota, F. L., Mangelinck-Noël, N. & Nguyen-Thi, H.
REFERENCES Analysis of gravity effects during binary alloy directional solidification by com-
1. Kurz, W., Fisher, D. J. & Rappaz, M. Fundamentals of Solidification (Trans Tech, parison of microgravity and Earth experiments with in situ observation. Eur.
2022). Phys. J. E Soft Matter 44, 98 (2021).
2. Dantzig, J. A. & Rappaz, M. Solidification (EPFL, 2016). 35. Zimmermann, G., Hamacher, M. & Sturz, L. Effect of zero, normal and hyper-
3. Fleury, V., Gouyet, J.-F. & Leonetti, E. M. Branching in Nature (Springer, EDPS- gravity on columnar dendritic solidification and the columnar-to-equiaxed
ciences 2001). transition in Neopentylglycol-(D)Camphor alloy. J. Cryst. Growth 512, 47–60
4. Akamatsu, S. & Nguyen-Thi, H. In situ observation of solidification patterns in (2019).
diffusive conditions. Acta Mater. 108, 325–346 (2013). 36. Steinbach, S. & Ratke, L. The effect of rotating magnetic fields on the micro-
5. Ludwig, A. et al. Advanced solidification studies on transparent alloy systems: a structure of directionally solidified Al–Si–Mg alloys. Mater. Sci. Eng. A 413-414,
new European solidification insert for material science glovebox on board the 200–204 (2005).
International Space Station. JOM 64, 1097–1101 (2012). 37. Al-Omari, K., Roósz, A., Rónaföldi, A., Sved ́ a, M. & Veres, Z. Macrosegregation
6. Bergeon, N. et al. Dynamics of interface pattern formation in 3D alloy solidifi- evolution in eutectic Al-Si alloy under the influence of a rotational magnetic
cation: first results from experiments in the DECLIC directional solidification field. Metals 12, 1990 (2022).
insert on the International Space Station. J. Mater. Sci. 46, 6191–6202 (2011). 38. Mota, F. L., Medjkoune, M., Strutzenberg Littles, L., Karma, A. & Bergeon, N.
7. Jackson, K. A. & Hunt, J. D. Transparent alloys that freeze like metals. Acta Met. Solidification furnace for in situ observation of bulk transparent systems and
13, 1212–1215 (1965). image analysis methods. Rev. Sci. Instrum. 94, 065111 (2023).
8. Kurz, W., Fisher, D. J. & Trivedi, R. Progress in modelling solidification micro- 39. Mota, F. L. et al. Initial transient behavior in directional solidification of a bulk
structures in metals and alloys: dendrites and cells from 1700 to 2000. Int. Mater. transparent model alloy in a cylinder. Acta Mater. 85, 362–377 (2015).
Rev. 64, 311–354 (2019). 40. Pereda, J. et al. Experimental characterization and theoretical analysis of cell tip
9. Hunt, J. D. & Jackson, K. A. Binary eutectic solidification. Trans. Metall. Soc. AIME oscillations in directional solidification. Phys. Rev. E 102, 032804 (2020).
236, 843–852 (1966). 41. Boettinger, W. J., Warren, J. A., Beckermann, C. & Karma, A. Phase-field simula-
10. Jackson, K. A. & Hunt, J. D. Lamellar and rod eutectic growth. Trans. Metall. Soc. tion of solidification. Annu. Rev. Mater. Res. 32, 163–194 (2002).
AIME 236, 1129–1142 (1966). 42. Steinbach, I. Phase-field models in materials science. Model. Simul. Mater. Sci.
11. Jackson, K. A., Hunt, J. D., Uhlmann, D. R. & Seward, T. P. On origin of equiaxed Eng. 17, 073001 (2009).
zone in casting. Trans. Metall. Soc. AIME 236, 149–155 (1966). 43. Provatas, N. & Elder, K. R. Phase-field Methods in Materials Science and Engineering
12. Huang, S. C. & Glicksman, M. E. Fundamentals of dendritic solidification. 1. (Wiley-VCH, 2010).
Steady state tip growth. Acta Met. 29, 701–715 (1981). 44. Plapp, M. Phase-field models. Handbook of Crystal Growth 2nd edn, Vol. 1A (ed.
13. Huang, W. D. & Wang, L. L. Solidification researches using transparent model Nishinaga, T.) 631–668 (Elsevier, 2015).
materials. A review. Sci. China Tech. Sci. 55, 377–386 (2012). 45. Karma, A. & Rappel, W.-J. Quantitative phase-field modeling of dendritic growth
14. Bergeon, N., Weiss, C., Mangelinck-Noel, N. & Billia, B. Interferometric method for in two and three dimensions. Phys. Rev. E 57, 4323–4349 (1998).
the analysis of dendrite growth and shape in 3D extended patterns in trans- 46. Karma, A. Phase-field formulation for quantitative modeling of alloy solidifica-
parent alloys. Trans. IIM 62, 455–460 (2009). tion. Phys. Rev. Lett. 87, 115701 (2001).
15. Bottin-Rousseau, S., Perrut, M., Picard, C., Akamatsu, S. & Faivre, G. J. Cryst. 47. Echebarria, B., Folch, R., Karma, A. & Plapp, M. Quantitative phase field model of
Growth 306, 465–472 (2007). alloy solidification. Phys. Rev. E 70, 061604 (2004).
16. Witusiewicz, V. T., Hecht, U. & Rex, S. Top-view approach for in-situ observation 48. Folch, R. & Plapp, M. Quantitative phase-field modeling of two-phase solidifi-
of growth morphology in bulk transparent organic alloys. J. Cryst. Growth 353, cation. Phys. Rev. E 72, 011602 (2005).
17–24 (2012). 49. Akamatsu, S. & Plapp, M. Eutectic and peritectic solidification patterns. Curr.
17. Akamatsu, S., Bottin-Rousseau, S., Şerefoğlu, M. & Faivre, G. Lamellar eutectic Opin. Solid State Mater. Sci. 20, 46–54 (2016).
growth with anisotropic interphase boundaries: experimental study using the 50. Akamatsu, S., Plapp, M., Faivre, G. & Karma, A. Overstability of lamellar eutectic
rotating directional solidification method. Acta Mater. 60, 3206–3214 (2012). growth below the minimum-undercooling spacing. Metall. Mat. Trans. A 35,
18. Ohno, M. Quantitative phase-field modeling and simulations of solidification 1815–1828 (2004).
microstructures. ISIJ Int. 60, 2745–2754 (2020). 51. Karma, A. & Sarkissian, A. Morphological instabilities of lamellar eutectics. Met.
19. Karma, A. & Tourret, D. Atomistic to continuum modeling of solidification Trans. A 27, 635–656 (1996).
microstructures. Curr. Opin. Solid State Mater. Sci. 20, 25–36 (2016). 52. Ginibre, M., Akamatsu, S. & Faivre, G. Experimental determination of the stability
20. Cross, M. C. & Hohenberg, P. C. Pattern formation outside of equilibrium. Rev. diagram of a lamellar eutectic growth front. Phys. Rev. E 56, 780–796 (1997).
Mod. Phys. 65, 851–1112 (1993). 53. Hecht, U. et al. Multiphase solidification in multicomponent alloys. Mater. Sci.
21. Woodruff, D.P. The Solid-liquid Interface (Cambridge University Press, 1973). Eng. Rep. 46, 1–49 (2004).
22. Jackson, K. A. Kinetic Processes: Crystal Growth, Diffusion, and Phase Transitions in 54. Khanna, S., Aramanda, S. K. & Choudhury, A. Role of solid-solid interfacial energy
Materials (Wiley-VCH, 2004). anisotropy in the formation of broken lamellar structures in eutectic systems.
23. Sherwood, J. N. (ed.) The Plastically Crystalline State (Wiley, 1979). Metal. Mater. Trans. A 51, 6327–6345 (2020).
24. Shalaev, E., Wu, K., Shamblin, S., Krzyzaniak, J. F. & Descamps, M. Crystalline 55. Plapp, M., Bottin-Rousseau, S., Faivre, G. & Akamatsu, S. Eutectic solidification
mesophases: structure, mobility, and pharmaceutical properties. Adv. Drug Deliv. patterns: interest of microgravity environment. C. R. Mecanique 345, 56–65 (2017).
Rev. 100, 194–211 (2016). 56. Akamatsu, S. et al. In situ experiments in microgravity and phase-field simula-
25. Derollez, P., Lefebvre, J. & Descamps, M. Neutron diffuse scattering in the plastic tions of the lamellar-to-rod transition during eutectic growth. C. R. Mecanique
phase of succinonitrile. J. Phys. Condens. Matter 2, 9975–9987 (1990). https://doi.org/10.5802/crmeca.142 (2023).
26. More, M., Lefebvre, J. & Fouret, R. Solid-state phase transition in carbon tetra- 57. Liu, S., Lee, J. H. & Trivedi, R. Dynamic effects in the lamellar-rod eutectic tran-
bromide. II. The crystal structure of phase I. Acta Cryst. B33, 3862–3865 (1977). sition. Acta Mater. 59, 3102–3115 (2011).
27. Glicksman, M. E., Koss, M. B. & Winsa, E. A. Dendritic growth velocities in 58. Parisi, A. & Plapp, M. Defects and multistability in eutectic solidification patterns.
microgravity. Phys. Rev. Lett. 73, 573–576 (1994). EPL 90, 26010 (2010).
28. Tennenhouse, L. A., Koss, M. B., LaCombe, J. C. & Glicksman, M. E. Use of 59. Bottin-Rousseau, S. et al. Coexistence of rod-like and lamellar eutectic growth
microgravity to interpret dendritic growth kinetics at small supercoolings. J. patterns. Scr. Mater. 207, 114314 (2022).
Cryst. Growth 174, 82–89 (1997). 60. Akamatsu, S. et al. Real-time study of thin and bulk eutectic growth in Succi-
29. Kauerauf, B., Zimmermann, G. & Rex, S. Cellular growth in transparent model nonitrile-(D)Camphor alloys. J. Cryst. Growth 299, 418–428 (2007).
substances under microgravity and ground based conditions, Solidification and 61. Perrut, M., Akamatsu, S., Bottin-Rousseau, S. & Faivre, G. Long-time dynamics of
Gravity 2000. Mater. Sci. Forum 329-3, 271–276 (2000). the directional solidification of rodlike eutectics. Phys. Rev. E 79, 032602 (2009).

Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA npj Microgravity (2023) 83
 S. Akamatsu et al.
12
62. Perrut, M., Bottin-Rousseau, S., Faivre, G. & Akamatsu, S. Dynamic instabilities of 91. Pereda, J. et al. Experimental observation of oscillatory cellular patterns in three-
rod-like eutectic growth patterns: a real-time study. Acta Mater. 61, 6802–6808 dimensional directional solidification. Phys. Rev. E 95, 012803 (2017).
(2013). 92. Tourret, D. et al. Oscillatory cellular patterns in three-dimensional directional
63. Perrut, M. et al. Role of transverse temperature gradients in the generation of solidification. Phys. Rev. E 92, 042401 (2015).
lamellar eutectic solidification patterns. Acta Mater. 58, 1761–1769 (2010). 93. Mota, F. L., Bergeon, N., Karma, A., Trivedi, R. & Debierre, J. M. Oscillatory-
64. Şerefoğlu, M., Bottin-Rousseau, S. & Akamatsu, S. Lamella-rod pattern transition nonoscillatory transitions for inclined cellular patterns in three-dimensional
and confinement effects during eutectic growth. Acta Mater. 242, 118425 directional solidification. Phys. Rev. E 102, 032803 (2020).
(2023). 94. Mota, F. L. et al. Effect of sub-boundaries on primary spacing dynamics during
65. Parisi, A. & Plapp, M. Stability of lamellar eutectic growth. Acta Mater. 56, 3D directional solidification conducted on DECLIC-DSI. Acta Mater. 204, 116500
1348–1357 (2008). (2021).
66. Ratke, L. & Alkemper, J. Ordering of the fibrous eutectic microstructure of Al- 95. Song, Y. et al. Cell invasion during competitive growth of polycrystalline soli-
Al3Ni due to accelerated solidification conditions. Acta Mater. 48, 1939–1948 dification patterns. Nat. Commun. 14, 2244 (2023).
(2000). 96. Hunt, J. D. Steady state columnar and equiaxed growth of dendrites and
67. Kellner, M., Kunz, W., Steinmetz, P., Hötzer, J. & Nestler, B. Phase-field study of eutectic. Mater. Sci. Eng. 65, 75–83 (1984).
dynamic velocity variations during directional solidification of eutectic NiAl- 97. Mooney, R. P. et al. Concurrent model for sharp and progressive columnar to
34Cr. Comp. Mat. Sci. 145, 291–305 (2018). equiaxed transitions validated by directional solidification experiments pro-
68. Witusiewicz, V. T., Hecht, U., Fernandez, J., Rodriguez, J. & Ezquerro, J. M. In-situ cessed in microgravity conditions. Comput. Mater. Sci. 210, 111436 (2022).
observation of eutectic growth during directional solidification of succinonitrile– 98. Zimmermann, G., Hamacher, M. & Sturz, L. In-situ observation of equiaxed
(D)camphor–neopentyl glycol alloys under imposed velocity transients. Acta dendrites solidifying under reduced gravity conditions during the TEXUS-55
Mater. 203, 116469 (2021). mission–the MEDI-2 experiment. ESA-SP 742, 250–257 (2019).
69. Witusiewicz, V. T., Hecht, U., Sturz, L. & Rex, S. Phase equilibria and eutectic 99. Mooney, R. P. & McFadden, S. Theoretical analysis to interpret projected image
growth in ternary organic system (D)camphor–neopentylglycol–succinonitrile. J. data from in-situ 3-dimensional equiaxed nucleation and growth. J. Cryst.
Cryst. Growth 286, 431–439 (2006). Growth 480, 43–50 (2017).
70. Kerr, H. W. & Kurz, W. Solidification of peritectic alloys. Int. Mater. Rev. 41, 100. McFadden, S., Mooney, R. P., Sturz, L. & Zimmermann, G. A Nucleation Pro-
129–164 (1996). genitor Function approach to polycrystalline equiaxed solidification modelling
71. Dobler, S., Lo, T. S., Plapp, M., Karma, A. & Kurz, W. Peritectic coupled growth. with application to a microgravity transparent alloy experiment observed in-
Acta Mater. 52, 2795–2808 (2004). situ. Acta Mater. 148, 289–299 (2018).
72. Hunziker, O., Vandyoussefi, M. & Kurz, W. Phase and microstructure selection in
peritectic alloys close to the limit of constitutional undercooling. Acta Mater. 18,
6325–6336 (1998). ACKNOWLEDGEMENTS
73. Lograsso, T. A., Fuh, B. C. & Trivedi, R. Phase selection during directional soli- We are grateful to the European (ESA), American (NASA), French (CNES), German
dification of peritectic alloys. Met. Mater. Trans. A. 36, 1287–1300 (2005). (DLR), and Austrian (FFG-ASAP) space agencies for their long-term support. We thank
74. Park, J.-S. & Trivedi, R. Convection-induced novel oscillating microstructure the E-USOC of Madrid (Spain) and the CADMOS of Toulouse (France) for their support
formation in peritectic systems. J. Cryst. Growth 187, 511–515 (1998). to TA and DECLIC-DSI operations, respectively.
75. Liu, S. & Trivedi, R. Effect of thermosolutal convection on microstructure for-
mation in the Pb-Bi peritectic system. Met. Trans. A. 37, 3293–3304 (2006).
76. Nabavizadeh, S. A. et al. Spurious grain formation due to Marangoni convection
AUTHOR CONTRIBUTIONS
during directional solidification of alloys in µ-g environment of International
Space Station. J. Cryst. Growth 574, 126334 (2021). All authors contributed equally to this work. S.A. took charge of the main
77. Ludwig, A. & Mogeritsch, J. Observation of peritectic couple growth for a hyper- coordination, supervision, and writing of the paper. W.S. was the initiator of the
peritectic alloy under microgravity conditions. IOP Conf. Ser. Mater. Sci. Eng. review. All authors discussed the results and implications and commented on the
1274, 012032 (2023). manuscript at all stages.
78. Mogeritsch, J., Sillekens, W. & Ludwig, A. In-situ observation of coupled growth
morphologies in organic peritectics under pure diffusion conditions. In Proc.
TMS 2022 151st Annual Meeting & Exhibition, 1429–1441 (2022). COMPETING INTERESTS
79. Nguyen-Thi, H. et al. Preparation of the initial solid–liquid interface and melt in The authors declare no competing interests.
directional solidification. J. Cryst. Growth 253, 539–548 (2003).
80. Ludwig, A., Mogeritsch, J. & Witusiewicz, V. T. Hypo-peritectic TRIS–NPG in a
stationary temperature gradient: thermodynamics, grain boundary migration ADDITIONAL INFORMATION
and phase identification. J. Cryst. Growth 604, 127052 (2022). Correspondence and requests for materials should be addressed to S. Akamatsu.
81. Valloton, J., Wagnière, J. D. & Rappaz, M. Competition of the primary and
peritectic phases in hypoperitectic Cu-Sn alloys solidified at low speed in a Reprints and permission information is available at http://www.nature.com/
diffusive regime. Acta Mater. 60, 3840–3848 (2012). reprints
82. Valloton, J., Dantzig, J. A., Plapp, M. & Rappaz, M. Modeling of peritectic coupled
growth in Cu-Sn alloys. Acta Mater. 61, 5549–5560 (2013). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims
83. Akamatsu, S., Moulinet, S. & Faivre, G. The formation of lamellar-eutectic grains in published maps and institutional affiliations.
in thin samples. Met. Mater. Trans. A 32, 2039–2048 (2001).
84. Şerefoğlu, M. & Napolitano, R. E. On the role of initial conditions in the selection
of eutectic onset mechanisms in directional growth. Acta Mater. 59, 1048–1057
(2011). Open Access This article is licensed under a Creative Commons
85. Lo, T. S., Dobler, S., Plapp, M., Karma, A. & Kurz, W. Two-phase microstructure Attribution 4.0 International License, which permits use, sharing,
selection in peritectic solidification: from island banding to coupled growth. adaptation, distribution and reproduction in any medium or format, as long as you give
Acta Mater. 51, 599–611 (2003). appropriate credit to the original author(s) and the source, provide a link to the Creative
86. Song, Y. et al. Thermal-field effects on interface dynamics and microstructure Commons license, and indicate if changes were made. The images or other third party
selection during alloy directional solidification. Acta Mater. 150, 139–152 (2018). material in this article are included in the article’s Creative Commons license, unless
87. Akamatsu, S. & Ihle, T. Similarity law for the tilt angle of dendrites in directional indicated otherwise in a credit line to the material. If material is not included in the
solidification of non-axially-oriented crystals. Phys. Rev. E 56, 4479–4485 (1997). article’s Creative Commons license and your intended use is not permitted by statutory
88. Deschamps, J., Georgelin, M. & Pocheau, A. Growth directions of microstructures regulation or exceeds the permitted use, you will need to obtain permission directly
in directional solidification of crystalline materials. Phys. Rev. E 78, 011605 from the copyright holder. To view a copy of this license, visit http://
(2008). creativecommons.org/licenses/by/4.0/.
89. Mota, F. L. et al. Influence of macroscopic interface curvature on dendritic
patterns during directional solidification of bulk samples: experimental and
phase-field studies. Acta Mater. 250, 118849 (2023). © The Author(s) 2023
90. Bergeon, N. et al. Spatiotemporal dynamics of oscillatory cellular patterns in
three-dimensional directional solidification. Phys. Rev. Lett. 110, 226102 (2013).

npj Microgravity (2023) 83 Published in cooperation with the Biodesign Institute at Arizona State University, with the support of NASA