Multiphase imaging of freezing particle suspensions by confocal microscopy
MMultiphase imaging of freezing particle suspensions by confocalmicroscopy
Dmytro Dedovets and Sylvain Deville ∗ Ceramic Synthesis & Functionalization Lab,UMR3080 CNRS-Saint-Gobain, 84306 Cavaillon, France (Dated: December 22, 2017)
Abstract
Ice-templating is a well-established processing route for porous ceramics. Because of the struc-ture/properties relationships, it is essential to better understand and control the solidification mi-crostructures. Ice-templating is based on the segregation and concentration of particles by growingice crystals. What we understand so far of the process is based on either observations by optical orX-ray imaging techniques, or on the characterization of ice-templated materials. However, in situobservations at particle-scale are still missing. Here we show that confocal microscopy can providemultiphase imaging of ice growth and the segregation and organization of particles. We illustratethe benefits of our approach with the observation of particles and pore ice in the frozen structure,the dynamic evolution of the freeze front morphology, and the impact of PVA addition on thesolidification microstructures. These results prove in particular the importance of controlling boththe temperature gradient and the growth rate during ice-templating. ∗ [email protected] a r X i v : . [ c ond - m a t . s o f t ] D ec . INTRODUCTION Ice-templating is a well-established processing route for porous materials in materialsscience in general [1–6] and in ceramics in particular [7–9]. Hundreds of papers are nowpublished every year on the topic, and the structural or functional properties of ice-templatedmaterials are systematically explored [10]. Although a considerable number of applicationshave been proposed, moving ice-templating from the lab towards industrial applications willdepend, to a large extent, of our ability to finely understand and control the process toensure reproducible, reliable architectures.Ice-templating is based on the segregation of matter by growing crystals, which can thenbe concentrated between the later. Removal of the ice–whereby “ice” is a generic termfor crystals grown from the solvent–provides a macroporous scaffold where the pores are areplica of the ice crystals, and the organization of particles in the scaffold is obtained duringfreezing. The structure of ice-templated materials, and thus their properties, are thereforelargely controlled by the phenomena that takes place during freezing.A lot of attention has thus been paid to in situ observations of the freezing of particlesuspensions. Several techniques have been proposed, each having its advantages and limi-tations [11]. Most of what we understand from the interactions of particles with growingcrystals have been obtained by optical microscopy [12, 13]. However, it only provides 2Dobservations, and the spatial resolution is not sufficient when small particles are used. X-ray imaging can provide 3D reconstruction of the grown [14, 15] or growing [16] crystals.However, its spatial resolution is not sufficient either to image particles. Artifacts inducedby the beam are also still problematic [16]. Transmission electron microscopy has also beenused [17] but does not provide 3D observations, and is not appropriate for systematic ex-periments. As the sample is fixed and of small dimensions, only a few interactions eventscan be imaged.In absence of appropriate experimental observations, efforts have been put in modellingthe redistribution of particles by growing crystals. Discrete elements modelling, in particular,provided numerous insights into the physics of ice-templating. The role of the growth rateof the crystals on the ordering of monodispersed spherical particles [18] or the alignment ofanisotropic (platelets) particles [19], was assessed. These results can be used experimentallyto produce a variety of materials with controlled microstructures and functional or structural2roperties [20]. Further progress in our understanding now depends on in situ observationsof these phenomena.Ideally, we need thus a technique able to image in situ and without artifacts the growth ofice crystals and the redistribution of particles, as well as the later stages of freezing when iceinvades the pores between concentrated particles. We recently demonstrated how confocalmicroscopy could be used to image in situ in 3D the growth of ice crystals [21]. Here webuilt on this preliminary work and show that particles can also be imaged during freezing.We developed a cooling stage that provides an independent control of the temperaturegradient and the growth rate of the ice crystals, making systematic investigations possible.In this paper, we demonstrate the benefits of this approach to investigate the physics ofice-templating.
II. METHODS
We used 2 µm diameter PMMA/TEFMA particles, marked in fluorescence with Pyrromethene546 (emission wavelength: 519 nm). The fluorescent dye (Sulforhodamine B) is dissolved inwater at − M . The suspensions were then prepared by incorporating 1 vol.% of particlesin this aqueous solution. The particle concentration is lower than that typically used inice-templating (5–40 vol.%) because we cannot properly image a volume if the particle con-centration is too high, because of light scattering. Using index-matched particles would bemore ideal for imaging, but this means not using pure water as index matching is usually ob-tained in water/DMSO systems. We prefer to use suspensions which formulation (aqueous)is close to that of ice-templating. The suspensions were sonicated in an ultrasound bath fora few minutes to ensure a good dispersion of the particles. A few suspensions were preparedby dissolving 1 wt.% of PVA (POLYVIOL SOLUTION LL 2830, 25%) in the suspension.We developed a cooling stage to perform in situ freezing experiments under the confocalmicroscope. The setup (Fig. 1) is composed of two Peltier modules that provide a constanttemperature gradient in the gap d between them. The particle suspension is introduced ina Hele-Shaw cell made of two glass slides, separated by two stripes of double-side stickytape which act as spacers to ensure a constant sample thickness of 100 µm . The sample issealed on both sides. This assembly is translated along the x -axis through the temperaturegradient at a constant velocity V by a stepper motor (Micos Pollux Drive stepper motor3 igure 1. Experimental setup to perform in situ freezing experiments in the confocal microscope.A constant temperature gradient ∆ T = T h − T c is established in the gap d between the Peltiermodules. The sample is translated through the temperature gradient at a constant velocity V ,which induces a growth of the ice crystals at a velocity V . The interface is thus kept at a con-stant position in the observation frame. Samples are 100 µm thick. T f is the freezing pointof the suspension. c (cid:13) (2017) Sylvain Deville (10.6084/m9.figshare.5722675) CC BY 4.0 licensehttps://creativecommons.org/licenses/by/4.0/ with VT-80 translation stage (PI, USA)). Because the sample is thin (100 µm ), thermalequilibrium is achieved in the range of growth rate investigated here (1–50 µm/s ); thesolidification front is thus at a constant position in the observation frame. We can thereforevary independently the solidification front velocity (adjusted by the stepper motor) and thetemperature gradient (established by the Peltier modules).Confocal imaging is achieved by two different fluorophores: the dye incorporated in theparticles, and a second dye (Sulforhodamine B) dissolved in water at − M , which fluorescesat 586 nm. Experiments performed with and without Sulforhodamine B revealed that it hasno noticeable impact on the freezing behavior of the system [22]. Images are acquiredthrough two photodetectors, operating at the respective emission wavelengths of the dyes.For image acquisition, we used long working distance non-immersive objectives (Leica HCPL APO 20x/0.70 CS and 10x/0.40 CS2) to minimize the effect of the microscope thermalmass on the freezing process. These objectives have free working distances of 0.59 mm and2.2 mm respectively.In a typical experiment, the sample is put in place on top of the Peltier elements, thermal4nsulation is achieved by covering the sample with a piece of polyurethane foam. A hole inthe foam let the objective come in close contact to the sample. The desired temperaturegradient ∆ T is established by setting the temperatures of Peltier elements and the sample isthen put in motion by the stepper motor. The interface velocity stabilizes within a minuteafter the beginning of the sample translation. The temperature gradient was varied from ◦ C/mm to ◦ C/mm in the experiments. The solidification front velocities were variedfrom 1 µm/s to 40 µm/s .Depending on the experiments and features investigated, 2D or 3D images were acquired.Image reconstruction was done with Fiji (ImageJ 1.51h) [23].
III. RESULTS AND DISCUSSION
The benefits of confocal microscopy to investigate the ice growth and the segregation ofparticles in a suspension are four-fold: • we can image individual particles. It is thus possible track the dynamics and organi-zation of particles during and after freezing. • because the dye is expelled from the growing ice, we can easily discriminate betweenthe water and the ice phases on the images. We can thus image simultaneously theparticles, the water, and the ice. • because of the rapid imaging mode of the microscope, we can image the process in2D at a rapid time resolution: up to 40 Hz at × pixels. We can thus takesnapshots of the solidification microstructures even at fast growth velocities (up to40 µm/s here.) • we can use the confocal mode to reconstruct the 3D solidification microstructures.Such combinations provide unprecedented insights into the phenomenon investigatedhere, as we demonstrate below. Particle-scale observations
The observation of a partially-frozen structure (Fig. 2) already provides a fresh look at theorganization of particles and the late stages of freezing. Elongated regions of concentrated5 igure 2. 3D reconstruction of the solidification microstructure. Although most of the structure isfrozen, some residual liquid pockets can be observed between particles. Several features of interestare pointed out here: A: isolated engulfed particles. B: Pore ice in larger pores. The ice alreadyinvaded the larger inter-particle pores. C: Locally dense packing of particles: pore ice is movingthrough the packing, into the smaller pores. D: residual liquid pockets in a small agglomerateof particles. Growth rate: 20 µm/s , temperature gradient: ◦ C/mm . c (cid:13) (2017) Sylvain Deville(10.6084/m9.figshare.5722675) CC BY 4.0 license https://creativecommons.org/licenses/by/4.0/ particles, typically encountered in ice-templated materials, can be observed between thecrystals. The directionality of growth induced a directional segregation pattern.If most of the particle reorganization took place here between adjacent ice crystals, someof the particles were also segregated between the crystal and the lower glass slide, resulting inthe particle monolayer regions seen for instance in the insets B and C of Fig. 2. Although thisis just a boundary limit effect, it provides an ideal configuration to investigate the penetrationof ice in the dense packing of particles. Several configurations, highlighted by the differentinsets can be observed. In inset A, isolated particles were engulfed by the growing crystals.We could not find evidence of a liquid film around these particles. Although a premeltedfilm is probably still present [24, 25], its expected thickness (a few nm) makes it too smallto be observable by confocal microscopy. Inset B shows a partially frozen region. Localdefects in particle packing can be observed, resulting in pores of different sizes between the6articles. The pores between closely packed particles still contain liquid water (as seen bythe fluorescence of the dye), while the water in the larger pores is frozen (no fluorescencevisible anymore). This observation can be explained by the Gibbs-Thomson depression ofthe freezing point. The ice entry temperature is lower as the pore size diminishes [26, 27]. Ininset C, which corresponds to a region at a lower temperature, pore ice can be observed in adensely-packed region. The upper part of the inset is mostly liquid, while the lower part ismostly frozen. The local temperature is therefore probably close to the ice entry temperaturefor the pore size that corresponds to the close packing of particles. The same behavior canbe observed in isolated 3D agglomerates of particles (inset D). A group of particles wereconcentrated and engulfed by the ice, forming a small agglomerate. The fluorescence signalclearly indicates that liquid water is still present in the agglomerate.Fig. 2 also shows that, for these solidification conditions, two fates are possible for theparticles. Most of the particles are segregated by the growing crystals in the inter-crystalspace. This mechanism is the basis for the development of ice-templated microstructures.However, some particles were also engulfed by the growing crystals without being concen-trated. The particles are therefore isolated in the ice in the frozen structure. Such particleswould not be observed if subsequent freeze-drying and sintering was performed, as theywould just randomly fall into the structure upon removal of the ice. In situ observationsof the frozen structures are therefore required to identify them. This means that the seg-regation of particles by the growing crystals is not 100% efficient. Such behavior was notreported before, and knowing about it is important as this could affect the ice-templatedmicrostructures.An increase of the growth rate to 40 µm/s (for the same temperature gradient) enhancesthis behavior (Fig. 3). We can observe that the fraction of isolated, engulfed particles isincreased. This can be explained by two phenomena: • the increase of the growth rate favors the engulfment of particles. The rejection orengulfment of the particles by the freeze front is governed by the force balance onthe particles [28]. In its most simple description, the force balance takes into accountthe viscous drag on the particle (exerted by the flux of water towards the interface)and the intermolecular interactions between the solidification front and the particle.When the drag is greater than the intermolecular interactions, the particle is engulfedby the front. There is therefore a critical velocity for engulfment at a given particle7 igure 3. 3D reconstruction of the solidification microstructure. The engulfment of isolatedparticles is enhanced by the increase of growth rate from 20 µm/s to 40 µm/s , compared tofig. 2. Growth rate: 40 µm/s , temperature gradient: ◦ C/mm . c (cid:13) (2017) Sylvain Deville(10.6084/m9.figshare.5722675) CC BY 4.0 license https://creativecommons.org/licenses/by/4.0/ size or, conversely, a critical particle size at a given freeze front velocity. However, theparticle-particle interactions modify the force balance, and favors the engulfment. Ina concentrated system like here, the critical velocity for engulfment is therefore lowerthan for isolated particles. The most important aspect here is that the concentrationof particles at the tip of the crystals will favor their engulfment. • the increase of the growth rate destabilizes the interface, as illustrated later in thepaper (see Fig. 7). Local variations of the ice growth velocity favors the engulfmentof isolated particles.Further work is now required to investigate the dependency of this phenomenon upon thefreezing conditions (particle concentration, growth rate, temperature gradient). Particle organization induced by ice growth
The understanding and control of the organization of particles or more generally objectsinduced by the growth of crystal is a central question in solidification studies [29]. In partic-8 igure 4. 2D cross-section that shows the organization of the particles induced by lateral growthof the crystals. The spatial organization can be compared with the output of discrete elementsmodelling. The first few layers of particles close to the crystal surface are organized. The par-ticles further show less organization. 1 wt.% of PVA, growth rate: 2 µm/s , temperature gra-dient: ◦ C/mm . c (cid:13) (2017) Sylvain Deville (10.6084/m9.figshare.5722675) CC BY 4.0 licensehttps://creativecommons.org/licenses/by/4.0/ ular, it is necessary to understand how the ordering, packing, and orientation of particles isrelated to the freezing conditions and morphology of the growing crystals. However, becauseof the typical space and time scale associated to the ice-templating conditions, these infor-mation’s were not accessible experimentally so far. Here, confocal microscopy is a suitabletool to investigate these phenomena. An horizontal cross-section taken during freezing isshown in Fig. 4. Parallel ice crystals (in black) repel and concentrate the particles in theinter-crystals space. The regions still liquid appear in magenta. The particles and theirorganization can clearly be observed in the regions adjacent to the crystals surface. In par-ticular, we can see how the first 2–3 layers of particles, in direct contact with the growingcrystals, organize into dense hexagonal packings. When the thickness of the layer of collectedparticles increases, the ordering is progressively lost.These observations are in good agreement with the output of discrete elements mod-elling [18, 19], which predicted such segregation patterns. The particle organization was9redicted to be strongly dependent on the displacement velocity of the interface in contactwith the particles [18]. If a growth velocity of 2 µm/s was set in this particular case, thelateral growth of the crystals, which is responsible for collecting and concentrating the par-ticles, is much lower than this (less than 2 µm/s in this case). Previous X-ray tomographyperformed in situ during freezing also reported [16] that the lateral growth velocity of icecrystals in a colloidal suspension is 2 to 3 times lower than the growth velocity along thetemperature gradient. The lateral growth velocity is therefore probably less than 1 µm/s here.Confocal microscopy appears thus as a powerful tool to investigate particle reorganizationduring freezing. Following this initial assessment, future work will focus on systematic vari-ations of the freezing conditions to investigate its impact on the reorganization of particles.
Effect of crystal tilt on particle segregation
Ice crystals, similar to other solidification systems such as succinotrile [30], do not al-ways grow perfectly aligned with the temperature gradient. Depending on the growth rate,temperature gradient, and the presence of additives, ice crystals can grow titled. The tip ofthe crystals can, in such case, adopt an asymmetric morphology which may, in turn, impactthe particle redistribution. An example is shown in Fig. 5, with a snapshot taken duringthe growth of ice. The asymmetric morphology can clearly be observed. The tip of thecrystals is partly faceted, with a facet facing the incoming particles. The projected area (orlength, in 2D) that collects the particles is shown in inset A. The projected length d r of theright-hand side of the crystal is greater than the projected length d l of the left-hand side ofthe crystal. Because of this tilt and so the different surfaces available to collect the particles,the relative fraction of particles segregated to the right-hand side of the crystals is greaterthan that segregated to the left-hand side. This can be observed with more details in theinset B, that shows both sides of the same crystal. The thickness of the layer of particlesrejected by the lateral growth of the crystal is 2 to 3 times larger on the right-hand side thanon the left-hand side. In a given pore channel, such as shown in inset C, the segregation ofparticles is thus asymmetric.This behavior was not reported before, and may impact the particle segregation pattern.The particle organization (packing density, orientation in case of anisotropic particles) be-10 igure 5. Snapshot of the system during ice growth with titled crystals. A differential lateralsegregation of particles is induced by the tilt. The tilted tip of the crystals pushes particlesto the right-hand side of the crystals. A: the tip of the crystal is asymmetric, and its relativesurfaces (or length, in 2D) that collect the particles on both sides are therefore different. B:Different concentration of particles are found on the right-hand side and left-hand side of thecrystals, because of the tilt-induced uneven segregation. C: progressive concentration of particlesinduced by the lateral growth of the ice crystals. 1 wt.% PVA, growth rate: 1 µm/s , temperaturegradient: ◦ C/mm . c (cid:13) (2017) Sylvain Deville (10.6084/m9.figshare.5722675) CC BY 4.0 licensehttps://creativecommons.org/licenses/by/4.0/ tween the crystals depends, among other parameters, on the thickness of the accumulatedlayer of particles. Several recent studies have reported that a controlled segregation of par-ticles by ice crystals can be used to induce 2D or 3D structures with structural or functionalproperties [20, 31–33]. In such cases, the functional response may be optimized by a bettercontrol of the particle organization. The effect identified here could thus potentially be usedto tune these properties. 11 igure 6. Effect of the temperature gradient on solidification pattern and particle segregationat the tip of the crystals. Increasing the magnitude of the temperature gradient for a givengrowth velocity stabilizes the front. Even at such rapid velocity (40 µm/s ), a thin layer ofparticles repelled by the crystal tip is visible. The corresponding temperature gradient is indi-cated on each picture. c (cid:13) (2017) Sylvain Deville (10.6084/m9.figshare.5722675) CC BY 4.0 licensehttps://creativecommons.org/licenses/by/4.0/
Impact of temperature gradient
The temperature gradient, along with the growth velocity, is the parameter with themost dramatic impact on the solidification microstructure. However, in the majority of theice-templating setups, the temperature gradient is not controlled and evolves as freezingproceeds. The usual approach in ice-templating is to put the suspension in a mold, the baseof which is cooled at a constant cooling rate [34]. In such a configuration, the temperaturegradient varies during the experiment. Most studies have thus focused on the impact of thegrowth rate [35, 36], which is a much easier parameter to control.Our temperature-controlled setup provides an independent control of the cooling rate andthe temperature gradient. We can thus assess separately the impact of the later.We investigated the impact of the temperature gradient on the solidification microstruc-ture for a constant growth velocity (Fig. 6). The temperature gradient was varied from ◦ C/mm to ◦ C/mm , for a growth rate of 40 µm/s . Two features should be noticed here.We can observe in Fig. 6 that even at such rapid velocity (40 µm/s ), a thin layer ofaccumulated particles is found at the tip of the crystals. This was not reported before,12 igure 7. Dynamic evolution of the crystal morphology, in an unstable regime. The solidificationfront in these conditions constantly evolve, with new crystals formed and older ones disappear-ing. This impacts the homogeneity of the solidification microstructure, but also results in a lotof isolated particles being engulfed. The arrows on the first frame indicate the tip of crystalsthat stopped growing. The arrows on the other frames indicate the same region in the succes-sive frames; it illustrates how a grain boundary between adjacent ice crystals disappear fromthe freeze front and is replaced by another crystal tip. Growth rate: 40 µm/s , temperaturegradient: ◦ C/mm . c (cid:13) (2017) Sylvain Deville (10.6084/m9.figshare.5722675) CC BY 4.0 licensehttps://creativecommons.org/licenses/by/4.0/ and could explain some of the previously published results. When the tip of the crystalsreach the top of the sample, the layer of concentrated particles will be retained in the frozensample, and thus in the final ice-templated material. This is usually not a problem as thelayer is only a few microns thick, and the top and bottom of the sample are often removed(cut) before being used. However, ice-templated thin films have recently gained traction fora number of applications such as dye-sensitized solar cells or solid oxide fuel cells [9, 37–39].The top layer of an ice-templated film that is only a few microns thick cannot be removed,and may impact the final functional properties. The presence of accumulated particles atthe top of the film can be seen for instance in Fig. 2c and 2d of ref. [40].The second result is the stabilization of the crystals as the temperature gradient increases.When the crystals grow at 40 µm/s in a ◦ C/mm temperature gradient, the crystals arenot stable. A time-lapse sequence, shown in Fig. 7, reveals how the interface constantlyevolves. New crystals appear at the front, while other ones disappear. This behavior has twoconsequences. The first one is the heterogeneous redistribution of particles. Ice-templatingunder such conditions would result in discontinuous pore channels, as a crystal that stopsgrowing results in the closing of an ice-templated pore. The second consequence is the13ngulfment of a large fraction of isolated particles by the front, as shown in Fig. 3 in thefrozen structure. Again, from an ice-templating point of view, such behavior is not desirable.Increasing the temperature gradient from ◦ C/mm to ◦ C/mm and ◦ C/mm , for thesame growth velocity, stabilizes the front and should thus result in more homogeneous ice-templated architectures.These results prove the importance of a proper control of both the growth rate andtemperature gradient, because their impact on particle redistribution, but also on the ho-mogeneity of crystals and thus of the ice templated structures. This should be kept in mindwhen designing ice-templating setups.
Effect of PVA
Several additives are commonly used in suspensions for ice-templating. Additives includesurfactants, but also binders, crystal growth modifiers, anti-foaming agents, or plasticizers.The impact of these additives on the development of ice-templated architectures is difficultto understand, as they typically affect several characteristics of the suspension (viscosity,surface tension, etc. . . ). With our setup, we can investigate the impact of additives on themorphology of ice crystals and the segregation of particles while keeping all other parametersconstant. We illustrate here the approach with the impact of polyvinyl alcohol (PVA), whichis one of the most used binder in ice-templating studies [41–44].The impact of the addition of 1 wt.% of PVA in the particle suspension is shown inFig. 8. At a growth rate of 20 µm/s , the crystals are titled with respect to the direction ofthe temperature gradient and solidification front. The tilt here is therefore a consequence ofthe PVA addition and is not an artifact due to a poorly controlled temperature gradient. Apossible explanation is that the increased undercooling of the freeze front, resulting from therejection of PVA by the growing ice, enhances the growth anisotropy of the ice crystals. Thisshould be taken into account if PVA is added to prepare ice-templated materials, as tiltedcrystals will result in tilted pores in the final architecture. A tilt might not be desirable inapplications that involve fluid or gas transport, for instance [45].Crystals also become thinner with the addition of PVA. The periodicity of the lamellarsolidification pattern in a ◦ C/mm is of 38 µm (at 40 µm/s ) and just 15 µm (at 20 µm/s )with the addition of PVA (and should thus be even lower if we further increase the growth14 igure 8. Impact of PVA on the solidification microstructure. In presence of PVA, the crystals aretilted. The temperature gradient is vertical, the tilt is thus not an artifact that could be induced bya titled temperature gradient. Such a tilt is typically obtained in solidification system that exhibitsa growth anisotropy. The crystals are also thinner, compared to the experiment without PVA.1 wt.% PVA, growth rate: 20 µm/s , temperature gradient: ◦ C/mm . c (cid:13) (2017) Sylvain Deville(10.6084/m9.figshare.5722675) CC BY 4.0 license https://creativecommons.org/licenses/by/4.0/ rate). The thinning of the ice crystals in presence of PVA can be used to align particles in2D, as shown in Fig. 9. Here, the spacing between adjacent crystals is so thin that it canaccommodate only 2 layers of particles. The particles used here sediment in a few minutesand thus rapidly rest at the bottom glass surface, before getting organized by the passing icefront. After freezing, we are thus able to obtain a 2D grid of particles aligned in thin threads.This has been used previously with
T iO and V O suspensions [46, 47], with applicationsin conducting platforms for electrical stimulation and thermochromic films. Our approachcould thus be used to better understand and control such architectures. IV. CONCLUSIONS AND PERSPECTIVES
The 3D, multiphase imaging of confocal microscopy, combined with its rapid imagingcapacities and submicronic spatial resolution, makes it the almost perfect tool to investigate15 igure 9. 2D grid of particles aligned in thin threads by the ice crystals. The ini-tial suspension contained 1 wt.% of PVA. Growth rate: 20 µm/s , temperature gradi-ent: ◦ C/mm . c (cid:13) (2017) Sylvain Deville (10.6084/m9.figshare.5722675) CC BY 4.0 licensehttps://creativecommons.org/licenses/by/4.0/ the physics of ice-templating. Our specially designed temperature-controlled stage, withits independent control of growth rate of the crystals and temperature gradient, will nowlet us investigate in details the experimental parameters relevant to ice-templating, help usbetter understand the development of solidification microstructure, and guide us to adjustthe experimental conditions based on the desired architecture and microstructure. Theresults shown here also highlight the importance of a good control over the temperaturegradient, to improve the stability of the freeze front and the homogeneity of the ice-templatedarchitectures and microstructures.
ACKNOWLEDGEMENTS
The research leading to these results has received funding from the European ResearchCouncil under the European Union’s Seventh Framework Programme (FP7/2007-2013) /ERC grant agreement 278004 (project FreeCo). We acknowledge Tom Kodger for providingthe particles used in this study. 16
EFERENCES [1] J. Zou, J. Liu, A. S. Karakoti, A. Kumar, D. Joung, Q. Li, S. I. Khondaker, S. Seal, L. Zhai,Ultralight multiwalled carbon nanotube aerogel., ACS Nano. 4 (2010) 7293–302.[2] A. Ahmed, R. Clowes, P. Myers, H. Zhang, Hierarchically porous silica monoliths with tuneablemorphology, porosity, and mechanical stability, J. Mater. Chem. 21 (2011) 5753.[3] L. Qiu, J. Z. Liu, S. L. Y. Chang, Y. Wu, D. Li, Biomimetic superelastic graphene-basedcellular monoliths., Nat. Commun. 3 (2012) 1241.[4] C. Walter, S. Barg, N. Ni, R. C. Maher, E. Garcia-Tuñón, M. M. Zaiviji Ismail, F. Babot,E. Saiz, A novel approach for the fabrication of carbon nanofibre/ceramic porous structures,J. Eur. Ceram. Soc. 33 (2013) 2365–2374.[5] K. M. Pawelec, A. Husmann, S. Best, R. Cameron, Ice-templated structures for biomedicaltissue repair: From physics to final scaffolds, Appl. Phys. Rev. 1 (2014) 021301.[6] M. Naviroj, S. Miller, P. Colombo, K. T. Faber, Directionally aligned macroporous SiOC viafreeze casting of preceramic polymers, J. Eur. Ceram. Soc. 35 (2015) 2225–2232.[7] Y. Tang, S. Qiu, Q. Miao, C. Wu, Fabrication of lamellar porous alumina with axisymmetricstructure by directional solidification with applied electric and magnetic fields, J. Eur. Ceram.Soc. 36 (2016) 1233–1240.[8] C. Ferraro, E. Garcia-Tunon, S. Barg, M. Miranda, N. Ni, R. Bell, E. Saiz, SiC porousstructures obtained with innovative shaping technologies, J. Eur. Ceram. Soc. 38 (2018) 823–835.[9] P. L. Rachadel, D. F. Souza, E. H. Nunes, J. C. D. da Costa, W. L. Vasconcelos, D. Hotza, Anovel route for manufacturing asymmetric BSCF-based perovskite structures by a combinedtape and freeze casting method, J. Eur. Ceram. Soc. 37 (2017) 5249–5257.[10] S. Deville, Properties and Applications of Ice-Templated Materials, in: Freez. Colloids Obs.Princ. Control. Use, 2017, pp. 439–548.[11] S. Deville, Investigating the Freezing of Colloids: Experimental Techniques to Probe Solid-ification Patterns, Crystal Growth, and Particle Movement, in: Freez. Colloids Obs. Princ.Control. Use, 2017, pp. 47–90.
12] C. Körber, Phenomena at the advancing ice-liquid interface: solutes, particles and biologicalcells., Q. Rev. Biophys. 21 (1988) 229–298.[13] B. Saint-Michel, M. Georgelin, S. Deville, A. Pocheau, Interaction of Multiple Particles witha Solidification Front: From Compacted Particle Layer to Particle Trapping, Langmuir 33(2017) 5617–5627.[14] S. Deville, E. Maire, A. Lasalle, A. Bogner, C. Gauthier, J. Leloup, C. Guizard, In Situ X-RayRadiography and Tomography Observations of the Solidification of Aqueous Alumina ParticleSuspensions-Part I: Initial Instants, J. Am. Ceram. Soc. 92 (2009) 2489–2496.[15] B. Delattre, H. Bai, R. O. Ritchie, J. De Coninck, A. P. Tomsia, Unidirectional Freezingof Ceramic Suspensions: In Situ X-ray Investigation of the Effects of Additives, ACS Appl.Mater. Interfaces 6 (2014) 159–166.[16] S. Deville, J. Adrien, E. Maire, M. Scheel, M. Di Michiel, Time-lapse, three-dimensional in situimaging of ice crystal growth in a colloidal silica suspension, Acta Mater. 61 (2013) 2077–2086.[17] K. Tai, Y. Liu, S. J. Dillon, In situ cryogenic transmission electron microscopy for character-izing the evolution of solidifying water ice in colloidal systems., Microsc. Microanal. 20 (2014)330–7.[18] S. A. Barr, E. Luijten, Structural properties of materials created through freeze casting, ActaMater. 58 (2010) 709–715.[19] F. Bouville, E. Maire, S. Deville, Self-assembly of faceted particles triggered by a moving icefront., Langmuir 30 (2014) 8656–63.[20] J.-W. Kim, K. Tazumi, R. Okaji, M. Ohshima, Honeycomb Monolith-Structured Silica withHighly Ordered, Three-Dimensionally Interconnected Macroporous Walls, Chem. Mater. 21(2009) 3476–3478.[21] M. Marcellini, C. Noirjean, D. Dedovets, J. Maria, S. Deville, Time-Lapse, in Situ Imaging ofIce Crystal Growth Using Confocal Microscopy, ACS Omega 1 (2016) 1019–1026.[22] D. Dedovets, C. Monteux, S. Deville, Freezing Emulsions, arXiv Prepr. (2017).[23] J. Schindelin, I. Arganda-Carreras, E. Frise, V. Kaynig, M. Longair, T. Pietzsch, S. Preibisch,C. Rueden, S. Saalfeld, B. Schmid, J.-Y. Tinevez, D. J. White, V. Hartenstein, K. Eliceiri,P. Tomancak, A. Cardona, Fiji: an open-source platform for biological-image analysis., Naturemethods 9 (2012) 676–82.
24] J. G. Dash, H. Fu, J. S. Wettlaufer, The premelting of ice and its environmental consequences,Reports Prog. Phys. 58 (1999) 115–167.[25] J. S. Wettlaufer, M. G. Worster, Premelting Dynamics, Annu. Rev. Fluid Mech. 38 (2006)427–452.[26] J. S. Wettlaufer, Ice surfaces: macroscopic effects of microscopic structure, Philos. Trans. Ser.A (1999) 3403–3425.[27] Z. Liu, K. Muldrew, R. Wan, J. A. W. Elliott, Measurement of freezing point depression ofwater in glass capillaries and the associated ice front shape, Phys. Rev. E 67 (2003) 1–9.[28] A. W. Rempel, M. G. Worster, Particle trapping at an advancing solidification front withinterfacial-curvature effects, J. Cryst. Growth 223 (2001) 420–432.[29] S. Deville, Freezing Colloids: Natural and Technological Occurrences, in: Freez. Colloids Obs.Princ. Control. Use, 2017, pp. 1–46.[30] J. Ghmadh, J.-M. Debierre, J. Deschamps, M. Georgelin, R. Guérin, A. Pocheau, Directionalsolidification of inclined structures in thin samples, Acta Mater. 74 (2014) 255–267.[31] X. Shen, L. Chen, D. Li, L. Zhu, H. Wang, C. Liu, Y. Wang, Q. Xiong, H. Chen, Assemblyof colloidal nanoparticles directed by the microstructures of polycrystalline ice., ACS Nano. 5(2011) 8426–33.[32] F. Bouville, E. Maire, S. Meille, B. Van de Moortèle, A. J. Stevenson, S. Deville, Strong, toughand stiff bioinspired ceramics from brittle constituents., Nat. Mater. 13 (2014) 508–14.[33] C. Ferraro, E. Garcia-Tuñón, V. G. Rocha, S. Barg, M. D. Fariñas, T. E. G. Alvarez-Arenas,G. Sernicola, F. Giuliani, E. Saiz, Light and Strong SiC Networks, Adv. Funct. Mater. 26(2016) 1636–1645.[34] S. Deville, Ice-Templating and Freeze-Casting: Control of the Processes, Microstructures, andArchitectures, in: Freez. Colloids Obs. Princ. Control. Use, 2017, pp. 351–438.[35] T. Waschkies, R. Oberacker, M. J. Hoffmann, Control of Lamellae Spacing During FreezeCasting of Ceramics Using Double-Side Cooling as a Novel Processing Route, J. Am. Ceram.Soc. 92 (2009) S79–S84.[36] S. Deville, E. Saiz, R. K. Nalla, A. P. Tomsia, Freezing as a path to build complex composites.,Science. 311 (2006) 515–8.[37] L. Ren, Y.-P. Zeng, D. Jiang, Fabrication of Gradient Pore
T iO Sheets by a Novel Freeze-Tape-Casting Process, J. Am. Ceram. Soc. 90 (2007) 3001–3004.
38] S. W. Sofie, Fabrication of Functionally Graded and Aligned Porosity in Thin Ceramic Sub-strates With the Novel Freeze-Tape-Casting Process, J. Am. Ceram. Soc. 90 (2007) 2024–2031.[39] Y. Chen, J. Bunch, T. Li, Z. Mao, F. Chen, Novel functionally graded acicular electrode forsolid oxide cells fabricated by the freeze-tape-casting process, J. Power Sources 213 (2012)93–99.[40] X. Wang, B. Yang, K. Zhou, D. Zhang, Z. Li, C. Zhou, Ordered porous TiO2 films obtained byfreezing and the application in dye sensitized solar cells, Curr. Appl. Phys. 15 (2015) 662–668.[41] K.-h. Zuo, Y.-P. Zeng, D. Jiang, Properties of Microstructure-Controllable Porous Yttria-Stabilized Ziroconia Ceramics Fabricated by Freeze Casting, Int. J. Appl. Ceram. Technol. 5(2008) 198–203.[42] C. M. Pekor, B. Groth, I. Nettleship, The Effect of Polyvinyl Alcohol on the Microstructureand Permeability of Freeze-Cast Alumina, J. Am. Ceram. Soc. 93 (2010) 115–120.[43] K.-h. Zuo, Y.-P. Zeng, D. Jiang, Effect of polyvinyl alcohol additive on the pore structure andmorphology of the freeze-cast hydroxyapatite ceramics, Mater. Sci. Eng. C 30 (2010) 283–287.[44] H.-J. Yoon, U.-C. Kim, J.-H. Kim, Y.-H. Koh, W.-Y. Choi, H.-E. Kim, Macroporous Alu-mina Ceramics with Aligned Microporous Walls by Unidirectionally Freezing Foamed AqueousCeramic Suspensions, J. Am. Ceram. Soc. 3 (2010) 3–5.[45] J. Seuba, S. Deville, C. Guizard, A. J. Stevenson, Gas permeability of ice-templated, unidi-rectional porous ceramics, Sci. Technol. Adv. Mater. 17 (2016) 313–323.[46] H. E. Romeo, F. Trabadelo, M. Jobbágy, R. Parra, 2D-ice templated titanium oxide films asadvanced conducting platforms for electrical stimulation, J. Mater. Chem. C 2 (2014) 2806.[47] X. Cao, N. Wang, J. Y. Law, S. C. J. Loo, S. Magdassi, Y. Long, Nanoporous thermochromicVO (M) thin films: controlled porosity, largely enhanced luminous transmittance and solarmodulating ability., Langmuir 30 (2014) 1710–5.(M) thin films: controlled porosity, largely enhanced luminous transmittance and solarmodulating ability., Langmuir 30 (2014) 1710–5.