CRYOCHEMICAL METHODS FOR MANUFACTURING NANOSIZED … · when it comes to sensors, turbine engines, and fuel cells fabrication. Being studied over the whole com-position range, zirconia

112 O.Yu. Kurapova, V.G. Konakov, S.N. Golubev, V.M. Ushakov and I.Yu. Archakov

© 2012 Advanced Study Center Co. Ltd.

Rev. Adv. Mater. Sci. 32 (2012) 112-132

Corresponding author: O.Yu. Kurapova, e-mail: [email protected]

CRYOCHEMICAL METHODS FOR MANUFACTURINGNANOSIZED CERAMICS AND CERAMIC

PRECURSOR POWDERS WITH LOW AGGLOMERATIONDEGREE: A REVIEW

O.Yu. Kurapova1,2,3,V.G. Konakov1,2,3, S.N. Golubev3,

V.M. Ushakov3 and I.Yu. Archakov1,2,3

1Chemical Department, St. Petersburg State University, St. Petersburg, Russia2Institute for Problems of Mechanical Engineering, St. Petersburg, Russia

3Science and Technical Center “Glass and Cera]ics”, St. Petersburg, Russia

Received: September 02, 2012

Abstract. Available methods of cryochemical treatment for nanopowders and nanoceramicsproduction were briefly described. The basics and theory of freeze-drying were discussed. Re-cent achievements and efforts in this field have been summarized. The proper choice ofcryochemical treatment methods appeared to be the enough easy route for producing nanosizedceramic powders with high dispersity and desired structure of agglomerated particles. Morphol-ogy and structure of stabilized zirconia precursors obtained via different cryochemical treatmenttechniques has been investigated in details. It was shown that physical and chemical propertiesof final powders (such as mechanisms of phase formation, agglomeration degree, etc.) as wellas the kinetics of it crystallization strongly depend on the way of it initial treatment.

1. INTRODUCTION

It is well known that such parameters as precursordispersity and their degree of agglomeration are criti-cal for nanoceramics manufacturing. Required prop-erties of final ceramics, such as its fractural strength,reduction of the critical defects size and refining ofthe matrix grain size during sintering, are often as-sociated with the elimination of the agglomerationin the precursor powders. It’s the particle size ofprecursors (i.e. dispersity of powder) that, in manyrespects, predetermines following compaction andsintering of samples. One of the possible ap-proaches giving the opportunity to manufacture theceramics with maximal density at lower tempera-ture is the use of nanosized precursors as a rawmaterial [1].

So far, post-precipitation treatment of gels, ob-tained by sol-gel co-precipitation technique, is con-sidered as the most efficient tool for producing ofceramic nanopowders with low degree of agglom-eration. Effect of treatment conditions on the par-ticle agglomeration degree, morphology, and sin-terability of powders has been reported in modernliterature, see Refs. [2-5]. However, most of con-ventional methods require evaporation of dispersemedium at elevated temperatures as a commondewatering method. This way of water removing pro-motes particles in close contact to each other. Itmay lead to a particle bridging via the reaction be-tween the hydroxyl groups on the surface of nearbyparticles according to reaction (1):

2R-OH + HO-R = R-O-R + H O. (1)

113Cryochemical methods for manufacturing nanosized ceramics and ceramic precursor powders...

Further condensation i.e. formation of the solidbridges between two particles results in formationof hard three-dimensional agglomerates and, hence,severe irreversible agglomeration [6]. The presenceof such agglomerates in precursors means non-uni-form compaction of the green bodies, i.e. lower ex-tent of particles packing and its lower coordinationin sample. As a result, shrinkage and cracking ofsamples during calcination is observed [3]. Hence,low quality and shorter shelf-life of final ceramicsmay take place. In order to reduce the formationand uncontrolled growth of hard three-dimensionalagglomerates pan-drying under pressure is conven-tionally used; it yields soft two-dimensional agglom-erates. Such agglomerates are weak and could beeasily broken mechanically. Another possible wayof agglomeration elimination is the exchange of ini-tial solvent to an organic fluid or an addition of SAS(surface active substances), for example, sodiumlauryl sulfate (SLS).

Surface active substances form a thin film onthe surface of each particle and prevent its adhe-sion. However, this film can not be removed by acommon cycle of washing and filtration. Thus, finalprecursor may contain some admixture of organ-ics; further precursor purifying is required, henceand the cost of the final ceramics could increaseconsiderably.

A number of cryochemical techniques (i.e. di-rect freezing, instant freezing, synthesis insupercritical carbon dioxide, spray granulation) couldserve a good alternative to conventional evaporationtechniques: pan-drying, spray-drying, oven-drying,etc. These mild methods do not imply elevated tem-perature during drying process (see Fig. 1). It can

Fig. 1. Two possible conceptions of water removal.

provide some other benefits like the uniformity of allthe methods, rather simple and inexpensive equip-ment.

New concepts of treatment help to eliminate hardthree-dimensional agglomeration due to limited con-tact between nearby particles. The environmental-friendly and non-toxic substances (water, liquid ni-trogen or carbon dioxide) are used as a dispersemedium. It corresponds to the modern tendency of“green” che]istry. It is i]portant to note that re-]oving of disperse ]ediu] by subli]ation doesn’tbring any impurities to the sample. Latter over-spreads the area of freeze-drying potential applica-tion to biological and medical engineering. Freezedrying is of particular interest as a completely auto-matic and simple enough route for producingnanopowders with narrow size distribution. In gen-eral, the suggested technique is based on freezingof sample with following removing of solidified sol-vent (water, alcohol or camphene) by the direct sub-limation of the disperse medium under reduced pres-sure. Thereby, both the tendency toward 3D agglom-eration and compression into hard agglomerates iseliminated [7]. Porous network is obtained wherepores replicate the shape of the solvent crystalsremoved. Different pore sizes (from 1 to 100 m),porosity (from 10% to 90%), and pore morpholo-gies can be obtained [8]. As all other cryochemicaltechniques, freeze-drying is based primary on physi-cal interactions (phase transitions) and slightly de-pends on the chemical nature of the sample. In or-der to mold green bodies of different shape, freezedrying can be combined with such casting tech-niques as slip-casting or gel-casting. Some appli-cations of freeze-drying are described in [9-11],


sometimes it is also called as freeze-casting. Incontrast to convectional drying of solvent-saturatedsamples (as gel, obtained via reverse co-precipita-tion technique), removal of disperse medium bysublimation eliminates a tendency to cracking andwarping [12].

Many systems had been already tested as aperspective ceramic materials for the industrial,medical and biological applications including alu-mina [8,13-15], titanium dioxide [16], yttria [6], sta-bilized zirconia [7, 17-23], hydroxyapatite [24], andsilica [16]. It was reported that stabilized zirconia(SZ) is a material, possessing unique combinationof fractural toughness, thermal, and chemical sta-bility [25-27]. The properties of zirconia-based ma-terials depend on the microstructure and the crys-talline phases, which are a function of particle sizedistribution, the degree of agglomeration, and theremoval of defects during processing [7].

As it was shown in recent reports [6,7],cryochemical methods have been successfully ap-plied to SZ ceramics and SZ ceramic powders fab-rication. However, recent reports mainly concernmanufacturing of highly porous ceramic structures[28], nanotubes, and nanowires [29]. Nanosizedceramic precursors powders can be also producedby these techniques [3,4,6,7]. Latter is significantwhen it comes to sensors, turbine engines, and fuelcells fabrication. Being studied over the whole com-position range, zirconia systems, nevertheless, maystill present several problems like slow degradationin the presence of moisture or other phase transfor-mations. Major control of such structures shouldbe carried via correct choice of the way of synthe-sis and further treatment. To date, the possibility ofstabilization of cubic zirconia is of great interest interms of oxygen conductivity of this phase. To thisend, the effect of treatment conditions on the mor-phology and agglomeration strength of final ceramicpowders had been widely investigated. In this re-view, which can by no means be exhaustive, atten-tion is largely confined to the basics of cryochemicalmethods. Recent achievements in this field are dis-cussed and summarized. The focus is made on thedata on production of nanosized stabilized zirconiapowders with low degree of agglomeration and dif-ferent morphology.

2. THEORETICAL ASPECTS

2.1. Freeze-drying technique

The freeze-drying method (i.e. ice-templating) wasfirst successfully applied by E. W. Flosdorf in 1935

in the instant coffee industry [13]. Already in 1954,freeze-casting has been described as a shapingtechnique for refractories [10]. Then, M. Bechtoldand W. Mahler have reported a freeze-forming pro-cess for ceramic fibers. In their study, phase sepa-ration in a gelled aqueous polysilicic acid was used.Gel was surrounded by ice and isolated in it duringfreezing; the direction of ice formation was controlledunidirectionally in the solution. Following meltingyielded ceramic fibers formation [30]. Method wasconsidered as unpro]ising and didn’t find wide co]-mercial application. However, in 2001, it was re-dis-covered by T. Fukasawa in terms of porous ceramicmaterials [30]. Alumina has been used as a modelmaterial in this study. The opportunity for the appli-cation of freeze drying to other types of ceramicswas pointed out. Since that time, ice-templating drewconsiderably more attention; number of studies andreports on freeze-casting and drying rose signifi-cantly. Nowadays it is a promising method for fabri-cation of ultra-fine ceramic powders and ceramicswith a complex shape. Furthermore, freeze-dryingis demonstrated as a preferred powder recoverytechnique since particles are kept in a dispersedstate during drying, and compression in hard three-dimensional agglomerates is eliminated [7].

In general, the discussed technique consists offreezing of a liquid suspension (aqueous or not),followed by sublimation of the solidified dispersemedium from the solid to the gas state under re-duced pressure and subsequent calcination or sin-tering [28,29]. The efficiency (i.e. the rate) of dryingdepends on degree of initial supercooling. It is oftenconsidered that degree of supercooling is the differ-ence between the “real” freezing point of the liquid(Tm) and the temperature at which solidification takesplace. In order to prevent melting, the temperatureof the frozen sample should be kept at least 20 de-grees lower than that for the pure disperse medium.

All the procedure can be divided into severalsteps: preparation and freezing of the sample; pri-mary drying; secondary drying, and sintering. As itis widely known, final properties of the powders arein strong dependence with the initial treatment con-ditions. In order to understand the nature of this ef-fect, every step of powder production should be dis-cussed in details.

2.2. Preparation of the samples

Preparation of the sample is in fact the most impor-tant step in a freeze-drying process, since it prede-termines resulting pore structure of the materials. Itincludes selection of disperse medium, preparation


of the suspension or obtaining gel by sol-gel co-precipitation technique, and freezing itself.

2.2.1. Selection of the dispersemedium

This is the expected consequence of literature re-views analysis that both the morphology of pow-ders and the conditions of drying strictly depend onchoice of the disperse medium. Varying this param-eter makes possible to perform freezing and dryingat close to room temperatures. Frozen dispersemedium temporarily acts as a binder and allows tohold the parts of green body together for demolding[12]. This, therefore, minimizes the concentrationof necessary organic binder. Latter enhances purityand makes faster binder burnout cycles. It is impor-tant to note that frozen disperse medium acts asnot only a binder between the particles of dispersedphase but also a template of the pore channels.Different porosity and unique shape of pore chan-nels can be achieved. To underline this feature, theter] ‘te]plate’ will be used instead of ‘disperse]ediu]’ further. The relevant characteristics of thecurrently tested templates are summarized andlisted in Table 1.As can be seen fro] Table 1, the ‘working’ te]-

perature of the suspension should correspond tothe range where the template is liquid. Room tem-perature is usually used in the case of water. Differ-ent temperatures are necessary for camphene-basedand tert-butyl alcohol suspensions (60 °C and 8 °C,respectively).

Due to its unique properties, water is concernedas the most common template in freeze-drying (FD).Starting with a well-dispersed suspension, the sur-rounding water is frozen, effectively immobilizingeach particle of dispersed phase. The expansion of

Solvent type Liquid Boiling Saturated Freezing Surface Temperature,density temperature vapor temperature tension at which vapor(g/ml) °C pressure at (°C (dyn/cm) pressure is 130 Pa

20 °C (kPa

Water 1 100 2.3 0 73.1 253ethanol 0.79 78 5.83 -114 24.1 242tert-butyl 0.78 82.5 4.1 25.3 20.7 198alcoholacetone 0.79 56,1 24.46-24.6 -95 26.2 214camphene 0.84 155 0.33 44.9 sublimes -glycerol 1.26 290 < 0.01 18 58 253

Table 1. Physical properties of water and some other solvents as templates for cryotreatment.

water during the crystallization (see Table 1) pre-serves particles from coming into close contact [5,7].On the other hand, phenomenon mentioned createsmechanical pressure and promotes compaction ofparticles of disperse phase, it is usually called as“freeze pressing”. The effect of physical and che]i-cal parameters under freezing on agglomerationdegree and texture of final powders has been dis-cussed in details in case of aqueous alumina sus-pensions in [5].

While water is, obviously, treated as being veryfavorable from an economic and environmental view-point (since the debinding step can be canceled),the disadvantages of this disperse medium are vastfor further manufacturing of ceramics [31]. Thesedisadvantages include: high surface tension (poorwetting), low vapor pressure (slow drying), sensitiv-ity to pH (less stable, higher viscosity suspensions),and organics packages that is dramatically affectedby residual humidity (shorter shelf-life and greenbody stability).

However, another strong advantage of water inrelevance to FD is a possibility to affect its solidifi-cation behavior. In order to modify the structure ofice various co-solvents (i.e. cryoprotectants) maybe added to the aqueous system. These additivesare used extensively for freezing of biological tis-sues and can be used to reduce the effects of sol-ute rejection and crystallite size during the solidifi-cation of water in suspensions of ceramic powders.The desired properties of cryoprotectants for freeze-casting ceramics are low toxicity, solubility in wa-ter, low freezing point depression, and cost effec-tiveness. It should be noted that manycryoprotectants give significant shift of freezingpoints in very low temperature area which does notpresent practical interest for freeze-drying tech-niques [12].


Ethylene glycol, propylene glycol, glycerol, ac-etone, and various alcohols are often used to con-trol the freezing behavior of aqueous ceramic sus-pension [10,19]. Such additives prevent the growthof large ice crystals, and hence, diminish the de-fects in freeze-casted green body. For example,mixing of glycerol with water (in proper proportions)yields shift of freezing te]perature just fro] -1.6 °Cto -9.6 °C. Further]ore, glycerol reduces the volu-metric expansion of water from 10% to 5% by vol-ume at 20 wt.% glycerol in water [11]. Kinetics ofsolidification of the zirconia suspensions in waterwith addition of tert-Butyl alcohol (TBA) was investi-gated by S. Sofie in [31]. It appeared that TBA lim-its the size of the solidified crystals for all solidsloadings and freezing temperatures evaluated. It wasalso reported that the addition of TBA to water re-sults in faster drying [32].

Several organic solvents with high melting pointsmay be used instead of water. The requirements forthe alternative disperse medium are following [9]:- appropriate solidification temperature, which

should be higher than the room temperature butshould not be too high (see Table 1);

- liquid viscosity similar to water so that concen-trated slurries can be made;

- small volume change during solidification to avoidproblems associated with solidification shrinkage(or expansion);

- higher vapor pressure in solid state than that forfreeze-drying. Moreover, the vehicle should be safeand inexpensive.

Recently, such compounds as TBA and cam-phene have been tried as a disperse medium infreeze-drying. It gave new challenges to freeze-cast-ing processing. Such disperse mediums not onlysuit mentioned requirements but also make it pos-sible to perform drying at room temperatures. Thesetemplates also offer the ability to cast ceramics thatare reactive with water, with melting temperaturesthat exceed room temperature [31]. Ultra-high po-rosity and completely interconnected pore channelsin final samples could be formed thus.Ca]phene – C10H16 - is a bicyclic monoterpene.

It is safe and inexpensive natural material which iscommonly used in fragrance compounds [9]. Cam-phene has a ]elting point around 44°– 48°C, at roo]temperature it is a crystalline solid material. Whensolid, it has a vapor pressure high enough for con-venient freeze-drying (see Table 1). At room tem-perature and ambient pressure there exist a largegradient of camphene vapor pressure between thesurface of camphene and the surrounding atmo-sphere. Hence, camphene can be removed via sub-

limation without the assistance of suction. It helpsto avoid the collapse of the green sample both dur-ing and after drying process. Latter makes it easierto use suspensions with lower solid loading, whichis required for high porous ceramics manufacturing[14]. During freeze drying, dendrites of camphenegrow in certain crystallographic directions. Suchfreezing behavior results in formation of abicontinuous structure, in which each separatedphase (camphene or network made of concentrateddispersed phase) is interconnected in 3D space [21-23].

Tert-butyl alcohol can allow a flexible freezingprocess and results in relatively long pore channelsafter sublimation, as a consequence of its high freez-ing temperature (see Table 1) [18,33]. Unlike thedendrite structure of frozen water or camphene, TBAnormally presents a form of long straight ice prismswithout any branches. As it was observed by R.Chen in [11], pore channels obtained after sublima-tion had straight polygonal cross sections. In caseof water or camphene-based suspensions, poreshave circular cross sections. To date, it is impor-tant for specific applications such as fabrication ofceramics with unidirectional pore structure such as,for example, water filters. It has been shown thatTBA can be removed also by rapid volatilization at80 °C which is faster than freeze-casting. It is inter-esting to note that TBA-based systems with 10vol.% of solid loading show the same freezing be-havior as aqueous based YSZ (yttrium stabilizedzirconia) suspensions at 30 vol.% solids and gen-erate the same level of porosity. At 30 vol.% of solidmore tortuous (substantial interconnected) poros-ity is generated [31].

Among other candidates that could replace wa-ter and organics and serve as a disperse medium inFD, are supercritical fluids and, in particular,supercritical carbon dioxide (scCO2). Supercriticalcarbon dioxide is utilized as a porogen or foamingagent for producing of various types of porous ma-terials [29,34,35]. It is nontoxic, nonflammable,cheap, chemically inert, and has a low viscosity.This last property leads to a high diffusivity for reac-tants in scCO2, which increases the reaction rates.When reactions are carried out in scCO2, reactionproducts can be easily recovered by simply ventingthis disperse medium [19]. The structures obtainedusing carbon dioxide are, somewhat, similar to thoseobtained with camphene, with a complex dendriticstructure [28]. Furthermore, a scCO2-based routemay allow the in-situ preparation of porous materi-als in containment vessels where solvent separa-


tion is problematic, e.g., direct formation of porouspacking within narrow-bore silica capillaries [36].

2.2.2. Preparation of the suspension

This step is very similar to the preparation of slur-ries for such common processing rout as slip cast-ing [28]. All the parameters should be carefully op-timized. Ceramic precursor and disperse mediumare mixed in proper proportions. To avoid microstruc-tural inhomogeneities, caused by agglomeration ofthe dispersed phase, prolonged stirring and de-air-ing is commonly carried out [30]. Researches haveobserved that ultrasonic impact is an efficient wayto obtain well-dispersed mixtures. Ball-milling hadbeen recently applied to ceramic slurries and is of-ten used in camphene freeze-casting techniques.Using the war] ball ]illing process at 55-60 °C, itis possible to disperse the ceramic powdershomogenously in the molten camphene with the aidof a dispersant [9,16,23]. The alternative route is toutilize nanoparticles that are initially in a well-dis-persed state. For example, gels obtained by reverseco-precipitation technique may be used here [6,7].

The effect of solid content of starting suspen-sions/gels on resulting pore structure seems to befully investigated for different disperse mediums[9,12,14,20,23]. It has been shown that the idealsuspension had a solid content of 15 vol.% [20]. Anincrease of solids content in aqueous suspensionscan potentially eliminate the growth of crystals and,consequently, modify freezing behavior of water [12].Higher concentration of particles in suspensioncauses increasing of nucleation centers which, inturn, causes earlier nucleation. Latter leads to aheterogeneous solidification of smaller ice crystals.The increased solid content also raises the freez-ing rate of the ice front. Three reasons are given inthe literature to explain this trend. First, in suspen-sions with a higher solid content less water needsto be frozen, this leads to faster freezing. Second,with lower water content, less crystallization heatneeds to be conveyed and, third, the heat conduc-tivity of water is lower than that of ceramic particles[20]. However, suspensions with solid content morethan 60% become difficult to mix and are no longersuitable for pressureless mold infiltration.

It was notified that the pore volume fraction andthe size of pores may be controlled by adjustingthe initial solid content in starting suspensions. Itwas shown that the size of the pores increases withthe decrease in the slurry concentration [14,21-23].To achieve moderate porosity (30% - 70%), both inaqueous and camphene-based freeze-casting meth-

ods, ceramic suspensions with a solid loading of20 vol.% have normally been used. In order toachieve ultra-high porosity, solid loading in the ce-ramic suspension should be as low as possible.Recently, new route for achieving ultrahigh porosityceramics was developed by Young-Hag Koh et al.As they report in [9], the decrease of the initial solidloading from 20 to 5 vol.% induces the increase inporosity linearly from 66% to 90% with an increasein the pore size.

2.3. Freezing step

According to Frenkel [37], nucleation of new phaseis due to heterophasic fluctuations and pre-transi-tional states. According to Eq. (2)

I S

s

LT T T T

T

( )

I 0

0

( ) ( ) , (2)

where l and s – che]ical potentials of particles inphases liquid and solid phases , L(l S) – heat oftransition from liquid to a solid state. Hence, nophase transition is observed at the equilibrium tem-perature. Shift of equilibrium state and, consequently,phase transition, is observed because of theoversaturation of the system under cooling. Fluc-tuations in the range of transition l S cause for-mation and crush of nuclei of different size. Microinhomogeneity of the system increases with the in-crease of the heterophasic fluctuations probability.To reach their critical size, nuclei have to overcomecertain potential barrier.

Under cooling, ceramic particles in suspensionare excluded from the frozen disperse medium andconcentrated in-between pure crystals of the sol-vent. The control of solvent crystal growth as wellas the interaction between particles and moving ofthe solidification front during freezing, gives one aunique opportunity to design numerous morphologi-cal features in the structure. Normally, for freezing,suspension is poured into a mold, which undergoesisotropic or anisotropic cooling to induce homoge-neous or directional solidification. Shape of the moldis also of importance because it should accommo-date the solidification volume change. In case ofceramic powders producing, simple ampoule or thePetri dish is suitable as a mold for freezing.

Recent studies investigate rheological propertiesof the aqueous suspensions, as well as porosityand mechanical strength of freeze-casted ceram-ics in relation to the freezing temperature and thecooling rate, and the particle size of starting mate-rials [8,24,25,29].


Fig. 2. Morphology of final precursors in dependence on the way of freezing and drying.

In case of aqueous solutions, slow cooling (withthe rate of 5-15 K/min) leads to slow ice nucleation.Large ice crystals are formed here and are oftenconnected with each other in dendritic shapes [32].Herein, each particle of dispersed phase acts as afuse and diminishes activation potential barrier forformation of critical nuclei, in other words, it worksas a nucleation site. Narrower particle size distribu-tion of starting powder causes simultaneous appear-ing of many small ice nuclei that grow with the samerate. Hence, smaller particles of dispersed phasein suspension can enforce ice nucleation processand potentially limit the time for growing of largedendritic ice crystals. It means less volume changeof disperse medium during freezing and higher de-gree of homogeneity. Fewer defects (i.e. cracks)should be expected in final ceramics. Thus, to in-duce ice nuclei and reduce the time available forgrowing of large ice crystals freezing rate should beas high as it is possible [29, 38]. To summarize,the following conditions are necessary to producesmaller crystals:- freezing should take place under extremely low

temperature/high pressure conditions;- starting materials should be nanosized;- particle size distribution of starting materials

should be mono-dispersed.

3. FREEZING METHODS

As it was mentioned above, the way of freezing (i.e.its temperature and cooling rate) predetermines thefinal structure of materials. According to freezing

conditions, all freezing methods can be divided intodirectional (i.e. controlled) and unidirectional (i.e.homogeneous) freezing. It can be followed eitherdrying under reduced pressure or melting as it isrepresented in Fig. 2.

3.1. Directional freezing

The process of freezing where the growth of crystalof disperse medium (generally, water) is controlledin one direction is called directional freezing[29,36,39]. This can be achieved by applying a hightemperature gradient across the sample. For that,a bottom of a vessel containing the liquid sample isplaced into a cryochemical bath (e.g., liquid nitro-gen). As vessel is frozen only from the bottom, theice growth occurs in one direction. Ice crystals growfrom the low to the high temperature end. Particlesof dispersed phase are concentrated with the mov-ing of the freezing front. Under certain conditions,growth of ice crystals excludes the particles of dis-persed phase from the freezing front, in turn, it leadsto aligned structures.

3.2. Unidirectional freezing

Unidirectional freezing is an alternative techniquewhere the sample is frozen with a rate of 10-103

K/min. When the freezing rate is too fast, the par-ticles of dispersed phase may be covered by a thinlayer of ice i.e. encapsulated in ice [12]. It can beachieved, for example, by spraying of suspensioninto the cryogen (i.e. liquid nitrogen) and is calledspray-freezing into liquid (SFL).


Spray freeze drying (SFD) is an indirect freezingtechnique which has been used in ceramic process-ing to insure the uniformity of starting materials onsubmolecular level or to improve the chemical reac-tivity of the powder. In order to freeze instantaneously,suspension is sprayed by drops beneath the sur-face of a cryogen at high constant pressure main-tained by pump. Intense atomization of the liquidtakes place when the drop reaches cryogenic liq-uid. The layer of vaporized cryogen is formed be-tween each drop and the surface of cryogen. Thisdoes not let drops sink and freezing takes place onthe surface of the cryogen [40]. Frozen granulesare then separated and subjected to a FD. It wasreported that SFD is more efficient step for produc-ing of precursor of high dispersity, suitable for sub-sequent compaction than spray drying (SD). As itwas observed by Bala P. C. Raghupathy, uniaxialdie pressing of FD nanopowders at 380 MPa dem-onstrated much more homogeneous microstructuresthan one of spray-dried nanopowders [17]. Freezedrying eliminates agglomeration and yields spheri-cal soft nanosized granules of precursors [5,17,41].These granules could be easily broken mechani-cally. Precursors of high dispersity are obtainedthen.

Unidirectional freeze casting of ceramics in asingle cooled system has been widely investigatedby many research groups [5,8,12,17,18,40]. How-ever, the structure of pores in ceramics manufac-tured via this technique is easily influenced by smalltemperature changes. A relatively new attempt wasused by S. Deville et al. A closed system had beenbuilt in their works. Both, bottom and top, rod tem-peratures were controlled [8,42-44]. It appeared thatthis achievement is crucial for the reproducibility ofthe porous network. The idea of S. Deville was takenup and developed by A. Preiss et al. They used acombination of electrophoretic deposition (EPD) anddouble-side cooling freeze casting to fabricate ce-ramics with graded pore structures [20]. The intro-duction of EPD enhanced the density and thick-ness of the walls. Porous structure of final ceram-ics was affected by it. As shown in these works,pulsed EPD had led to a finer lamellar porous struc-ture and a decrease in wavelength gradient over thewhole height.

4. DRYING STEP

During drying, sample, which was frozen under cer-tain conditions, can undergo either freeze-drying ordrying at room conditions.

4.1. Freeze-drying

Freeze-drying of the sample is carried out into afreeze-drier or a modified vacuum chamber. FD it-self consists, in fact, of primary drying and second-ary drying. When disperse medium is water, pri-mary drying of the sample is usually conductedbelow -20 °C to prevent the ice fro] ]elting and atlow atmospheric pressure, less than 0.3 kPa. Topromote sublimation, pressure in the drying cham-ber should be lower than vapor pressure of water.

Secondary drying is carried out to remove struc-turally bound disperse medium (mostly water) fromthe unfrozen sample. Here, a lower vacuum level isrequired than for primary drying [29].

4.2. Drying at room conditions

Room conditions are sufficient for successful subli-mation of TBA and camphene. These templatescan be removed by simple vaporization (see Table1). In this case, there is no need in the special equip-ment. In case of water, drying at atmospheric con-ditions include thawing, i.e. formation of undesiredliquid phase. For that, frozen sample is retained onthe air for convectional water evaporation. Herein,growth of agglomerates in 3D space occurs simul-taneously with polymerization processes. It shouldyield hard tree dimensional agglomerates and,hence, poor dispersity of final precursors. It is rightin case of powders, obtained by drying of gel, ob-tained by sol-gel co-precipitation technique withoutany pre-treatment. However, it is interesting to notethat pre-freezing modify the size and structure ofagglomerates. When freezing of gel/suspension isinstant (it is possible at temperatures of liquid nitro-gen), there is no time available for particles rear-rangement. So particles of the dispersed phase areencapsulated in ice. Further thawing and drying inair lead to the formation of soft and weak agglomer-ates [45].

5. FORMATION OF THE STRUCTURE:THEORY

Many features of the freeze-dried powders or freeze-casted materials can be understood by applyinggeneral principles of solidification processes. Thepoint of view of researches both on the basics offreezing and the kinetics of the solidification under-went certain evolution during last decade.

First explanation of the freezing behavior of theceramic suspensions was based on capillary theoryof frost damage [4-6]. According to this theory, cap-


illary forces were considered as the main drivingforce for the agglomeration phenomena. Freeze dry-ing, therefore, was regarded as the technique, wherecapillary forces were eliminated due to direct subli-mation of the disperse medium from the solid to thegas phase. Almost full absence of these forces pre-vented close contacts between particles of dis-persed phase and formation of chemical bonds.Removal of the disperse medium was associatedwith the phenomena of dissolution-reprecipitation.Mass transfer was due to reprecipitation of dissolvedsalts that occurred continuously following the freez-ing front. Any dissolved material (with any solubilityin the disperse medium) was therefore thought asbeing reprecipitated homogeneously on the surfaceof the substrate. Thus, the concentration of agglom-erated particles at contact points was limited. There-fore, the main feature of FD appeared in the inhibi-tion of solid bridges formation at contact points be-tween agglomerates. To date, these bridges areformed during traditional convectional dewateringunder elevated temperature [22].

Crash of this theory was, in fact, evident. Re-cently, S. Deville et al. have proposed anothermechanism of solidification kinetics. A combinationof in-situ methods of analysis (X-ray radiography andtomography) and analytical calculations was used.Application of this novel complex approach demon-strated that the morphology of the investigated sys-tem is in dependence with the initial choosing ofthe temperature profile at the bottom of the mold ofthe suspension [28,43,44,46]. Therefore, freezingduring ice-templating occurs in two stages: an ini-tial transient regime followed by a steady state (i.e.undercooling). The undercooling at the base of thesuspension is considered as a driving force of so-lidification. At slow solidification rates, the particleseasily diffuse away from the interface and the tem-perature of the suspension ahead of the interface isalways higher than the freezing temperature.

Obviously, the diffusion can not take place atfaster solidification rates. When the interface veloc-ity is too high, the particle redistribution is obtainedby the direct interaction with the solidifying inter-face. As the concentration gradient at the interfaceis steep enough, the gradient in the freezing tem-perature is larger than the temperature gradient. Forthus reason, the suspension ahead of the interfaceis below its freezing temperature (i.e. constitution-ally undercooled).

Sometimes, the freezing process of ceramicsuspension is regarded to be similar to the freezingof binary alloys, where the disperse medium playsa role of figurative second phase. The constitutional

undercooling here is closely related to morphologi-cal instability; it is also known as a phenomenon ofMullins-Sekerka instability [28,29]. It is one of themechanisms that trigger formation of the irregularstructures, i.e., cellular, lamellar, or even more com-plex dendritic morphologies. However, one shouldbe careful with such analogies: the fact is that, incontrast to frozen alloys, frozen ceramics is no moreplastic.

Another mechanism is related to the presenceof the particles in suspension. In that case, the in-stability is due to the reversal of the thermal gradi-ent in the liquid ahead of the freezing front and be-hind the particle. This is associated with the factthat the freezing rate decreases during the ice frontgrowth [20,46].

Freezing generally starts when the nuclei havereached their critical size [20]. Until freezing oc-curs, the system is in the undercooled state andthe temperature of the system is much lower thanthe freezing temperature of the liquid phase. Thus,the system is out of equilibrium. A system alwaystends to reach equilibrium which can be achievedby ice crystal growth. Therefore, the initial freezingrate should be faster as it can be in order to obtainthe equilibrium. It means that the higher degree ofundercooling will be observed.

Large ice crystals with a dendritic morphologygrow very quickly into the suspension [44]. Theirgrowth rate is five to fifteen times higher than that ofthe freezing front appearing later. Further away fromthe cold surface, the ice growth rate became muchlower. It’s due to the evolved ice which operates asan insulator and reduces the temperature gradient[20]. The ice crystals have more time to grow hori-zontally at freezing slower rate; so, the freezing frontadvances quickly in the direction Z, imposed by thethermal gradient [46]. The obtained microstructureis characterized by two populations of ice crystals:R-crystals (R for randomly oriented) and Z-crystals(for lamellar ice crystals, oriented along Z-direction).Due to their different orientation and morphology,these two populations of crystals do not concen-trate ceramic particles with the same efficiency. Thetransition zone was identified between these twodifferent types of crystals; due to the existence ofthis transitional zone, it becomes difficult to controlthe porosity and structure obtained. The facts men-tioned are undesirable for the further applications ofthe manufactured materials. Several parametersshould be kept to control the graded structure ofthe transitional zone: the degree of undercooling,the freezing temperature of the suspension andnucleation conditions. In order to manufacture po-


rous ceramics three general requirements shouldbe taken into account [28,46]:1. Thermodynamic condition.In order to achieve porous structure, the ceramicparticles in suspension must be excluded from theadvancing solidification front and entrapped betweenthe growing ice crystals. This aspect can be under-stood in terms of surface energies and interactionsbetween solvent-solidification front and the particlesin suspension. If the particle is encapsulated by thesolid, overall increase of the surface energy shouldbe observed i.e.:

sp lp sl0, (3)

where sp, lp, and sl are the interfacial free ener-gies associated with the solid-particle, liquid-par-ticle, and solid-liquid interface, respectively. Whenthis criterion is satisfied, a liquid film should existbetween the solidification front and the particle inorder to maintain the transport of molecules towardsthe growing crystal. When the velocity of the frontincreases, the thickness of the film decreases.Under certain critical velocity, the flow of the mol-ecules becomes insufficient to feed the crystals ofpure template growing behind the particle. So far,particles are encapsulated by the frozen template.Besides, it was shown that the minimal extensionof the transition zone is obtained with parabolic pro-files. As the position of the transition zone is un-known, it is convenient to apply a parabolic profilesince the beginning of the process.2. Morphology of the freezing front.Morphology of the front dictates the architecture ofthe final materials. The ice front must be non-pla-

nar. Indeed, if solidification is achieved using a frontwith planar morphology, all particles are collectedon the one side of the sample. This effect is beingused in the purification of pollutants. However, par-ticles redistribution must occur to form porous struc-tures; the particles should be excluded from thesolidification front and collected between the armsof the solidification front.3. Rate of freezing.According to the experimental tests and analyticalmodeling [46], the velocity of freezing front shouldbe as high as possible. A successful approach con-sists in applying a cooling rate starting from 2-4°C%]in; graded porous structures are obtained atthese rates.

6. EXPERIMENTAL

The experimental part of this study focuses on thecombination of sol-gel precipitation technique andfurther gentle cryothemical treatment of the obtainedgels. To compare the degree of agglomeration andmorphology of final powders, pan-drying under pres-sure was chosen as a common evaporation tech-nique. In order to stabilize cubic modification of zir-conia, the composition with 9 mol.% of calcia waschosen according phase diagram of that system.Fig. 3 presents a flow chart of the precipitation andtreatment procedures for the calcia stabilized zir-conia synthesis.

Sol-gel co-precipitation technique was reportedmany times as a safe and inexpensive way to pro-duce nanopowders with well controlled compositionand morphology. Commercially available saltsZrO(NO3)2

.2H2O and Ca(NO3)2 were used to prepare

Fig. 3. Flow chart of the precipitation and drying procedures.


0.1M aqueous solution. 1M ammonium aqueoussolution was used as a precipitant. A dilute saltsolution was added to ammonium one by drops witha rate of ~ 1-2 ml/min. The precipitation has beenperformed at a temperature of 1-2 °C in an ice bath;pH of the solution was kept at ~9-10 during the syn-thesis. To remove reaction byproducts, the obtainedgels were rinsed using a water jet pump. Washedcakes of hydroxides were divided into three equalparts and underwent different further treatment: uni-directional freezing followed by freeze drying ormelting and pan-drying.

6.1. Cryotreatment

One part of sample was freeze-dried. For that, col-lector of freeze-dryer was cooled with the rate of 1-2 °C/min. When the temperature had reached itsminimum (i.e. -50 °C), a Petri dish with a certainamount of the precipitate was placed right underthe collector for unidirectional freezing. After thesample was completely frozen (in ~40 min), it wasrapidly replaced into a storage chamber (20 °C) anddisperse medium was removed by sublimation un-der reduced pressure of 0.018 Torr (freeze-drierLabconco, 1l chamber, USA). Dried precursor, bysmall portions, was calcined at 400, 600, 800, and1000 °C for 3 hours, respectively.

A number of cosolvents was added to thesamples to reduce the effect of water expansionduring freezing, which appeared in significant crack-ing of the solidified gel. According to the referencedata, glycerol and acetone were chosen ascryoprotectants in this study. 10, 20, and 30 wt.%of glycerol and the same amounts of acetone wereadded to the samples just before freezing.

Another part of recently sintered precipitate wasrapidly frozen into the liquid nitrogen. Small amountof gel was placed into the Dewar flask upon vigor-ous stirring with a glass multiblade agitator. Afterthe nitrogen had evaporated, a frozen granulate wasleft for melting under the room conditions.

After treatment, the obtained powder was pan-dried at 110 °C under the pressure of 2 kg/cm2. Forthat, a small amount of gel was placed betweentwo smooth glass surfaces (plates) and pressed bya cylinder. A sample was placed into an ovenequipped with a fan to enhance mass transport.Drying was considered to be completed when nomore weight loss could be observed.

Dried precursor, by small portions, was calcinedat 400, 600, 800, and 1000 °C for 3 hours, respec-tively.

6.2. Pan-drying

Third part of precipitate was directly pan-dried at110 °C under pressure without any pre-treatmentas was described above.

6.3. Characterization

A set of experimental approaches was used for thedetailed investigation of the above mentionedsamples; as a result, the exhaustive data on thestructure, morphology, and agglomerate size in thesamples was obtained.

Thermal decomposition of the precursors wasstudied by simultaneous thermal analysis (STA 449F1 Jupiter Netzsch) in flowing nitrogen at a heatingrate of 10 °C/min.

Phase compositions of the samples were iden-tified by X-ray diffraction analysis (XRD, SHIMADZUXRD-6000 diffractometer, Cu K = 1.54 Å, Ni filter,2 = 2-80°) in air. The crystallite size dxrd was esti-mated from the XRD peaks broadening usingScherrer’s equation:

xrdd k

57.3,

cos (4)

where is the Bragg angle of diffraction lines, K is ashape factor taken as 1, and is the wavelength ofincident X-rays.

The average agglomerate size was determinedusing Horiba LA-950 laser particle size analyzer.Their morphologies were observed via scanning elec-tron microscopy (SEM, Supra 40 electron micro-scope) and X-ray nanotomography.

X-ray tomography is a nondestructive imagingtechnique in which the three dimensional structureof the sample is reconstructed from two dimensionalX-ray projection images. The measured absorptionof X-rays is based on the linear absorption coeffi-cients and the thicknesses of the material compo-nents along the X-ray paths, which allow us to re-solve the distributions of components of different lin-ear absorption coefficients inside the sample [49].By adjusting the X-ray energy with respect tosample composition and size, the sensitivity of themeasurement can be optimized [50]. Also, the in-crease in the number of projection images enhancesthe quality of the reconstructed three dimensionaldistribution of linear absorption coefficients. XradiaMicroCT-400 device was used in the tomographicanalysis. The maximal energy of the X-rays was 40keV, and 1505 projection images of each samplewere recorded for the reconstructions. The size ofreconstructed images was 1.17 m, which is close


Sample number Way of post-treatment

1 Freeze-dried precursor1.1 Freeze-drying of gel with the addition of 10 wt.% of glycerol1.2 Freeze-drying of gel with the addition of 10 wt.% of acetone1.3 Freeze-dried precursor with further calcination at 380 °C, 3 hours1.4 Freeze-dried precursor with further calcination at 500 °C, 3 hours1.5 Freeze-dried precursor followed by further consequent calcination at 380 °C,

19 hours1.6 Freeze-dried precursor with further calcination at 400 °C, 3 hours1.7 Sample 1.5 additionally calcined at 500 °C, 2 hours1.8 Freeze-dried precursor with further calcination at 1100 °C, 3 hours2 Precursor, obtained by instant freezing in liquid nitrogen followed by drying

at room conditions2.1 Sample 2 + following pan-drying and calcination at 400 °C, 3 hours2.2 Sample 2 + following pan-drying and calcination at 600 °C, 3 hours2.3 Sample 2 + following pan-drying and calcination at 800 °C, 3 hours2.4 Sample 2 + following pan-drying and calcination at 1000 °C, 3 hours2.5 Sample 2 + following pan-drying and calcination at 1100 °C, 3 hours

Table 2. Numeration and brief description of manufactured samples.

Fig. 4. STA analysis of sa]ples 1, 1.1, 1.2, 2, and 2.1 (4a - DSC curves, 4b – TG curves .

(a) (b)

to the maximal resolution that can be reached byconventional X-ray CT scanners.

7. RESULTS AND ITS DISCUSSION

The results of experimental tests consist of struc-tural and morphological observations, the data aboutthermal decomposition and agglomeration degreeof final powders. Since a number of samples wasmanufactured and investigated in this study, for thesake of clarity, Table 2 presents their numerationand brief description. Since the composition of allthe sa]ples is the sa]e, i.e. 9 ]ol.% CaO – 91mol.% ZrO2, it was dropped out hereafter.

7.1. Thermal decomposition

Results of STA analysis for samples 1, 1.1, 1.2, 2,and 2.1 give the opportunity to understand the com-plex nature of decomposition process (see Fig. 4).The information about mass losses and tempera-tures, and the types of effects for examined samplesis presented in Tables 3 and 4, respectively.

The endothermic peak on DSC curves of allsamples corresponds to water removal from thesamples extends up to 400 °C. It is accompaniedby significant mass loss according to TG curves(see Fig. 4b). Such profile of water evaporation dur-ing heating is common for all zirconia composites


Sample number 1 1.1 1.2 2 2.1

Mass losses during dehydration. Mass. % 25.1 25.1 20.35 34.7 3.34

Table 3. Mass losses of the samples during dehydration.

Sample 1 1.1 1.2 2 2.1 Type of the effect

T, °C 135 142 134 130 103 Endo-357 - 365 350 - Exo-496 510 502 524 495 Exo-

Table 4. Characterization of the endo- and exothermal effects observed on STA curves.

[47,51]. For samples 1, 2, and 2.1 this removal con-sists in simultaneous dehydration and bonded wa-ter removal according to the reaction:

4 2 2Zr(OH) = ZrO(OH) + H O. (5)

According to a referred literature, reaction 5 canbe considered as the main cause of the mass lossat T 400 °C as decomposition of Ca(OH)2 to CaOtakes place at ~ 550 °C. Formation of amorphousZrO2 is due to the loss of one more molecule ofwater, see reaction (6). The process of the bondedwater loss is assumed to be close to temperatureof zirconia oxide crystallization [44].

2 2 2ZrO(OH) = ZrO + H O. (6)

To prove considerations above let us make anassumption that the composition of precipitate isZr(OH)4 and it doesn’t contain any dispersed water.In other words, all the mass loss would be due tothe structurally bonded water. Further calculationsshow that in this case precursor would contain 22.1mass.% of water. According to TG curves, masslosses of the samples lie in range of 3.34 - 34.7mass.% (see Table 3). Hence, not all the testedapproaches provide efficient dehydration. Accord-ing to Table 3, mass loss of samples 1 and 2 is25.1 and 34.7 mass.%, respectively. It means thatboth freeze-drying and rapid freezing in liquid nitro-gen followed by drying at roo] conditions don’t al-low to remove structurally bonded water from thesamples. Further pan-drying and calcination at 400°C performed for sample 2 demonstrates the mostdewatered precursor. It’s interesting to note that fur-ther addition of glycerol and acetone during freeze-drying significantly modifies the process of waterremoval (DSC curves of samples 1.1 and 1.2 onFig. 4a). Mass losses of samples 1 and 1.1 are

equal to each other and consist 25.1 mass.%. Thedifference seems to be in overall tendency of massloss. As it can be seen from Fig. 4a, the slight exo-thermal effect is observed on the DSC curve ofsample 1 at temperature ~ 360 °C (see Table 4).Additives of glycerol and acetone reduce this effectand there is no peak observed on DSC curves ofsamples 1.1 and 1.2. To date, the endothermic peakof sample 1.2 (with the addition of acetone) is shiftedto 142 °C (see Table 4) and can be attributed todehydration via bonded water removal only; this factis proved by TG data. It also in agreement with theexisting data about the beginning of hydroxide de-composition at ~ 140 °C (see reaction 4). Note thatmass loss of all samples continues in all tempera-ture range.

In spite of the fact that sample 2.1 is the mostdewatered precursor, the exothermal effect at 495oC corresponding to crystallization of stabilized zir-conia isn’t profound at all (see Fig. 4a .

Exothermic peaks observed in DSC curves forall samples has a characteristic temperature ~ 500°C, it can be attributed as being an indicator of theformation of zirconia based fluorite-like solid solu-tions. In case of sample 2, this effect is slightlyshifted to high temperature region; its temperatureis 524 °C. This shift of a crystallization peak is re-lated to agglomeration processes during drying inair, which is proved by PSD-analysis. For freeze-dried samples (curves 1, 1.1, and 1.2 in Fig. 4a)crystallization effect is more profound than the samefor nitrogen treated samples (curves 2 and 2.1 inFig. 4). Obviously, it means more exhaustive pro-cess of crystallization, i.e. narrower distribution ofagglomerates of the particles in samples; see PSDdata below.

One more exothermal peak at ~ 360 °C (seeTable 4) appears in DSC curves of all the samples


Fig.5. XRD patterns of samples 1 (after freeze-dry-ing), 1.3 (1 following calcination at 380 °C, 3 hrs)and 1.4 (1 following calcination at 500 °C, 3 hours).

Fig. 6. SEM images showing the microstructure of sample 1. Scale bars correspond to 200 m (a), 20 m(b) 1 m (c), and 200 nm (d), respectively.

except 1.1 and 2.1; this effect is the most profoundin case of freeze-dried sample. It may be attributedto phenomenon of slow crystallization at low tem-perature. In order to prove this assumption, a num-ber of small portions of freeze-dried powder wascalcined at 330-380 °C for 3 hours. Referring to XRDpattern of sample 1, see Fig. 5, the powder of thissample is amorphous just after freeze-drying.

Let us discuss Fig. 5 in more details. The begin-ning of crystallization process was observed after 3hours of calcination of sample 1 at 380 °C (pattern1.3). The increase in calcination temperature up to500 °C results in full crystallization and accompa-nied the formation of fluorite cubic structure of zir-conia based solid solution (pattern 1.4). Therefore,the data obtained prove the existence of the phe-

nomena of slow crystallization at temperatures near360 °C that lead to formation of cubic solid solution.Exothermic peak on DSC curves in this region canbe considered in terms of slow kinetics of crystalli-zation. To date, a phenomenon of crystallization withslow kinetics was discussed by Gianfranco Dell’Agliet al. [4]. In this work slow thermal crystallization at360 °C was induced by hydrothermal treatment ofzirconia gels. It resulted in formation of monocliniczirconia with partial addition of tetragonal phase.

However, slow thermal crystallization of zirconiafrom amorphous gels to stable cubic phase at lowtemperature without any additions is reported therefor the first time. Slow crystallization at tempera-tures near 360 °C could be attributed to applying ofsuch gentle method of dehydration as freeze-dry-ing. Low temperature formation of fluorite-like zirco-nia based solid solution proves the thesis about theshift of phase transitions temperatures for nanosizedceramic powders in comparing with the precursorsmanufactured via classic quenching method.

Examination of the microstructure of freeze-driedpowders could, likely, give a key to mechanism ofthis process.

7.2. Morphology and microstructureof the precursors

SEM observations presented in Fig. 6a revealed thatfreeze-dried powder is made up of fibers. Suchneedle-like shape is likely a result of high degree ofundercooling during gel freezing since needles pro-vide a maximum of heat transfer [40].


Fig. 7. Visualized orthoslices of sample 1 from f X-ray tomography; the size of the presented part ofthe sample is 1. 17 m in all directions.

These fibers consist of agglomerates ofnanosized particles with a typical size of 26-38 nm(see Fig. 6d), this fact is also proved by X-ray to-mography data (see below).

The use of such nondestructive imaging tech-nique as X-ray tomography in materials science isrelatively new [49,50]. This new approach gives aunique opportunity to examine the structure in thebulk of the sample without any damage of initialmorphology and investigate in-situ the processesand reactions across the sample (for example, ki-netics of freezing).

For tomographic measurement, sample 1 wasgently pre-compacted into the pills. It should benoted, that the morphology of the powder was pre-served during compaction. Reconstructed three-di-mensional image of the microstructure of sample 1is given in Fig. 7. As it can be seen from the figure,the particles of sample 1 are nanosized. Freeze-drying resulted in slight agglomeration of final pow-der, the average size of the agglomerates is about250-300 nm. Note that these data agree with theresults of PSD-analysis.

Detailed SEM examination of sample 1 shows,that fibers of the precursor possess characteristicsponge-type structure with the voids on their sur-face. In general, it is a result of particle rearrange-ment during solidifying and consequent dispersemedium removal via sublimation. During freezing,ice crystals appear first at the surface of the gel.Particles of disperse phase at the cooling rate of 1-

2 °C are excluded by growing freezing front andcollected between the pure crystals of water. Lattergrow in a dendritic way in the bulk of the sample.With the temperature decrease, ice crystals pen-etrate progressively into the smaller pores and leadto solidification of liquid here [3]. However, a frac-tion of the smallest pores may still contain liquideven after appreciable undercooling. The degree ofundercooling is significant at these freezing rates,i.e. the conditions are quiet far from equilibrium.System tends to reach equilibrium and, since thefree energy of the solid below the freezing point isless than that of the liquid, there is a thermody-namic driving force for the liquid to flow out of thesmaller pores to form a solid. Latter, likely, modifiesthe kinetics of material transport, i.e. particle rear-rangement during the solidification. Spheroidizationcan proceed on the surface of porous network dur-ing further sublimation due to the weak coordina-tion of the system, see [7]. It may result in globesobserved in Fig. 6c.

These globules possibly act as the nuclei cen-ters to induce further slow crystallization that wasobserved by X-ray analysis at 380 °C after calcina-tion of sample 1 for 3 hours (see Fig. 5). The factthat the nuclei are quiet far from each other makesit easier for many independent crystal regions toform and then grow simultaneously upon heating.Because the amount of outer heat is not enough,just some certain amount of these globes can over-come the potential barrier and then act as nucleicenters. However, the amount heat energy given toa system is not enough for the beginning of rapidcrystallization as it takes place at 500 °C (Curve1.4 in Fig. 5). Thus, more time is required for a pro-cess of crystal formation.

Information about the crystallization kinetics wasobtained from X-ray patterns of freeze-dried powderduring ther]al evolution of the sa]ple at 380 °C viaquenching method. Dependences of the intensivelyof 30° peaks as well as the crystallinity of final pow-ders on ti]e of treat]ent at 380 °C are presentedin Fig. 8.

Degree of crystallinity (the ratio of crystal andamorphous phases) was calculated using the soft-ware of SHIMADZU XRD-6000 diffracrometer. As itcan be seen from figure, both curves have a platoafter 12 hours of thermal treatment. Another words,constant heating from 3 to 12 hours promotes aslow formation of cubic calcia stabilized zirconiasolid solution. Then the process is inhibited becausea lack of external heat. Indeed, the additional calci-nation of freeze-dried powder after heating at 380°C for 19 hours at 500 °C for 2 hours proves this


Fig. 8. Dependence of the intensively of 30° peaks (a as well as the crystallinity of final powders (b on ti]eof treat]ent at 380 °C.

Fig. 9. XRD patterns of samples 1 and 2 after calcination at 1100 °C - a) sample 1.8 and b) sample 2.5,respectively.

hypothesis and induces further increase in the in-tensity of 30° peak up to 7200 c.u. (the point in Fig.8a). Hence, exothermal effect at temperatures about360 °C corresponds to a slow formation of fluorite-like cubic zirconia solid solution. This assumptionidea does not contradict the hypothesis of the glob-ules that acts as nuclei for crystallization. It seemsthat rapid crystallization in all volume of the samplerequires overcoming higher energy barrier; in otherwords calcination temperature should be increased.

7.3. Structure of final powders andsize of crystallites

Powders both after freeze-drying and nitrogen treat-ment are amorphous. Calcination of samples 1 and2 resulted in the increase in the characteristic 30°peak intensity and formation of zirconia based cu-bic solid solution. Further calcination of thesesamples showed sticking difference in phase for-

mation of stabilized zirconia. Nitrogen treatedsa]ple shows “nor]al” (usual picture of phase evo-lution with a temperature (see Fig. 9). The coexist-ence of cubic and monoclinic phases can be ob-served from XRD-pattern of sample 2 after calcina-tion at 1100 °C. Extent of monoclinic phase is 15%.Calcination of freeze-dried sample reveals unex-pected stabilization of fluorite structure of zirconiabased solid solution. Such behavior of sample 1 withtemperature proves the idea about strong depen-dence of final properties of powders on the choiceof dewatering method.

The size of crystallites in samples 1-1.8, 2-2.5after calcination was estimated using Sherrer equa-tion, see Table 5.

It can be seen from Table 5 that all the precur-sors after calcination are nanosized. In case offreeze-drying, formation of small crystalline domainsis observed at 600 °C. Powders after treatment byliquid nitrogen show the same size of crystallites


Calcination te]perature, °C Size of crystallites, nmFreeze-drying Treatment by liquid nitrogen

400 <5 <5600 5 <5800 15 61000 30 221100 95 72

Table 5. Size of crystallites esti]ated by Sherrer’s equation fro] XRD patterns of calcinated sa]ples.

Sample number Way of treatment Calcination Averagete]perature, °C agglomerate

size, m

1 Freeze-drying of rapidly frozen gel (-50 °C) - 0.861.6 400 0.761.9 600 0.651.10 800 0.711.13 1000 1.091.1 Freeze-drying with the addition of - 0.63

10 mass.% of glycerol1.11 Freeze-drying with the addition of - 0.78



10 mass.% of acetone1.21 Freeze-drying with the addition of - 0.85

20 mass.% of acetone1.22 Freeze-drying with the addition of - 0.86

30 mass.% of acetone2 Freezing of gel in liquid nitrogen following - 0.95

thawing and pan-drying of powder obtained2.1 400 0.342.2 600 0.282.3 800 0.182.4 1000 0.24

Table 6. Average agglomerate size in the final powders depending on the way of chosen treatment.

only at 800 °C. As it was illustrated by SEM im-ages (see Fig. 6), globes-nuclei are quiet far fromeach other. It helps to avoid more frequent and oftencontacts between domains during it growing. In com-mon, freeze-drying results in faster formation of crys-talline domains and finer crystal structure.

7.4. Size of agglomerates

Size of agglomerates of all investigated sampleswas measured by laser scattering method (PSD-analysis); the average agglomerate size is presentedin Table 6.

Analyzing the data listed in Table 6, one cansee that both FD and liquid nitrogen treatment yieldspowders of high dispersity. Size of agglomerates isquiet equal (0.86 and 0.85 m, respectively). A bithigher agglomeration is observed in case of liquidnitrogen treatment. It is likely due to the fact thatthis technique involves thawing and further pan-dry-ing steps. When water is removed by evaporation,particle rearrangement occurs during drying due tothe condensation reaction on the surface of the dis-persed phase particles. When it comes to FD, dis-perse medium is removed directly by sublimation


from solid to gas state without passing through liq-uid phase. Thus, oxygen bonding is eliminated.However, ice expansion during fast freezing with arate of 1-2 °C/min provides considerable mechani-cal pressure across the sample. It appears in sig-nificant cracking of gel. The addition of 10 mass.%of glycerol and acetone modifies freezing behaviorof water; this fact was confirmed by STA data dem-onstrated above (see Fig. 4). Cryoprotectants re-duce the growth of large ice crystals. It results inoverall diminishing of freezing defects and, finally,in the formation of de-agglomerated powder with thesize of agglomerates about 0.63 and 0.64 mm. Fur-ther increase in cryoprotectant loading have no ef-fect on agglomeration degree of final precursors. Itis, likely, due to the fact that cryoprotectants con-siderably modify mass transport during sublimation[3,12]. Examination of STA results for samples withdifferent loadings of cryoprotectants (not shown here)proves this assumption.

Resulted agglomerates are soft and can be eas-ily broken by ultrasonic impact. For FD, their sizeis 710 nm, see Fig. 10.

The curves of the dependence of final precursoragglomerates on the calcination temperature wereplotted using the PSD-data on the agglomeratessize distributions in studied powders. The resultswere compared with those obtained earlier for pre-cursor powders of 8Y2O3-25TiO2-ZrO2 composition[45]. The curves are depicted in Fig. 11. As it isclearly seen from the figure, the difference betweentreatment methods manifests itself considerablyafter calcination. Freeze-dried powder gives moreclassic tendency of thermal evolution, whereas pow-

Fig.10. Results of PSD-analysis of sample 1 with and without ultrasonic impact.

Fig.11. Thermal evolution of powders after differentdewatering treatment.

ders after freezing in liquid nitrogen show de-agglom-eration up to 1000 °C. This fact can be understoodfrom the general positions of evolution of powderduring thermal treatment. In common, de-agglom-eration is observed up to 700 °C. Stabilization ofparticles is due to dewatering across the sampleand formation of zirconia based fluorite-like solidsolutions. Further temperature increase promotesquick growth of crystallite domains. At the tempera-ture 700 °C their evolution is ahead of de-agglom-eration of powder due to dewatering. As a result,the caking of agglomerates particles is observedduring calcination.

Instant freezing in liquid nitrogen results in iso-lation of particles of dispersed phase in ice. Thus,growth of dendritic ice crystals is limited. It should


lead to formation of many randomly oriented smallice crystals. Thawing induce a collapse of ice andslight agglomeration of powder after such treatment.Following calcination leads to de-agglomeration ofmelted granules. Then dehydratation processes takeplace on the surface and then in bulk of agglomer-ated particles. The size of final agglomerates ofsample 2 is 180 nm after calcination at 800 °C, seeTable 6.

In case of freeze-drying, significant agglomera-tion of particles is observed already at 800 °C.Herein, the formation of the final structure is, likely,observed at lower temperature than in the case ofinstant freezing. It could be understood using com-parison of XRD patterns of samples 1 and 2 treatedat 1100 °C, Fig. 9, which was discussed above.Sherrer’s esti]ation of crystallines gives 15 n] forfreeze-drying and only 6 nm for liquid nitrogen treat-ment (Table 5). Evidently, freeze-drying leads tofiner crystal structure at lower temperature.

8. CONCLUDING REMARKS

Present work briefly describes available methods ofcryochemical treatment for nanopowders andnanoceramics production. The basics and theory offreeze-drying are discussed. To summarize, milddewatering by direct sublimation eliminates hardthree-dimensional agglomeration and yields softgranules of agglomerates that can be easily brokenmechanically, for example, by ultrasonic impact.Instant pre-freezing of aqueous suspensions reducescondensation processes and allows further convec-tional drying of water-containing samples at roomconditions. It has been also shown that freezing stepis the most important among other steps (i.e. prepa-ration of suspension/gel, primary or secondary dry-ing) as it predetermines the size and structure ofagglomerates in final powders. The required poros-ity can be also achieved by correct choice of cool-ing rate and profile of freezing front. Fine particlesize distribution of starting materials as well asnanosized particles of dispersed phase results inmore uniform structure of freeze-casted green body.

Morphology and structure of stabilized zirconiaprecursors obtained via different cryochemical treat-ment techniques has been investigated in details. Itwas shown that physical and chemical propertiesof final powders (such as mechanisms of phase for-mation, agglomeration degree, etc.) as well as thekinetics of it crystallization strongly depend on theway of it initial treatment. Freeze-drying induces slowcrystallization of stabilized zirconia at 380 °C ac-companied by formation of cubic zirconia solid so-

lution. Powders obtained using liquid nitrogen treat-ment demonstrated de-agglomeration up to 1000°C followed by formation of cubic zirconia solid so-lution with certain amount of monoclinic phase at1100 °C. Freeze-drying of gel followed by calcina-tion of nanoparticles leads to the formation of finecrystal structure and stabilization of cubic phase atthis temperature. The data obtained proves strikingdifference in behavior between nano- andmicroparticles. The shift of phase transitions tem-peratures is observed clearly in case of nanosizedpowders.

These studies show that the proper choice ofcryochemical treatment methods is the enougheasy route for producing nanosized ceramic pow-ders with high dispersity and desired structure ofagglomerated particles. It allows a flexible treatmentprocess and, thus, provides unique opportunity todesign complex pores and pore channels, andachieve required morphology of final precursors. Theadvantages mentioned and versatility of cryothemicalmethods open a wide prospective for further investi-gations of nanosized ceramic powders and improve-ment of the nanoceramic technology. Authors be-lieve that cryomethods could find new developmentin following fields:- engineering of new functional materials and

graded pore structures, in particular, i.e. materi-als for microturbines, new engines, and solid ox-ide fuel cells;

- fundamental studies of the effects of suspensionsolid contain on the final agglomerate size of pre-cursors and green bodies;

- experimental study and modeling of freezing ki-netics i.e. new attempts to eliminate the effectsassociated with the existence of transition zone.

ACKNOWLEDGEMENTS

The authors are grateful to Prof. Jussi Timonen andDr. Markko Myllys ( Department of Physics, Univer-sity of Jyväskylä, Jyväskylä, Finland for fruitful co-operation and experimental help with the X-ray to-mography study.

This work was, in part, supported by the Rus-sian Ministry of Education and Science (Contracts14.740.11.0353 and 8025).

REFERENCES

[1] M. Mazaheri and Z. Razavi Hesabi // J. Amer.Cer. Soc. 92 (2009) 990.

[2] V.G. Konakov, S. Seal, E.N. Solovieva, S.N.Golubev, M.M. Pivovarov, A.V. Shorochov //Rev. Adv. Mater. Sci. 13 (2006) 71.


[3] Jalajakumari Nair, Padmakumar Nair, JanG. Van Ommen, Julian R. H. Ross and

Anthonie J. Burggraaf // J. Am. Ceram. Soc. 81(1998) 1487.

[4] Gianfranco Dell Agli, Giuseppe Mascolo, MariaCristina Mascolo and Concetta Pagliuca //J. Am. Ceram. Soc. 91 (2008) 3375.

[5] J. Zheng, L. Winnubst, Velianti, S. Fang andD. Salamon // Adv. Eng. Mater. 13 (2011) 77.

[6] Johanne Mouzon, Magnus Oden, OlivierTillement and Yves Lorand // J. Am. Ceram.Soc. 89 (2006) 3094.

[7] Ch. J. Szepesi, J. Cantonnet, R. Kimel andJames H. Adair // J. Am. Ceram. Soc. 94(2011) 4200.

[8] Audrey Lasalle, Christian Guizard, JeromeLeloup, Sylvain Deville, Eric Maire, Agne‘sBogner, Catherine Gauthier, Jerome Adrien andLoýc Courtois // J. Am. Ceram. Soc. 95 (2012)799.

[9] Kiyoshi Araki and John W. Halloran // J. Am.Ceram. Soc. 87 (2004) 1859.

[10] T. Moritzw and H.-J.Richter // J. Am. Ceram.Soc. 89 (2006) 2394.

[11] R. Chen, C. Wang, Y. Huang, L. Ma andW. Linz // J. Am. Ceram. Soc. 90 (2007)3478.

[12] Stephen W. Sofie and Fatih Dogan // J. Am.Ceram. Soc. 84 (2001) 1459.

[13] Regina Villanueva, Andres Go]ez, PedroBurguete, Eduardo Martýnez, Aurelio Beltranand Fernando Sapina // J. Am. Ceram. Soc.94 (2011) 236.

[14] Young-Hag Koh, Ju-Ha Song, Eun-Jung Leeand Hyoun-Ee Kim // J. Am. Ceram. Soc. 89(2006) 3089.

[15] Carolina Tallon, Rodrigo Moreno and IsabelNieto // Materials Research Bulletin 41 (2006)1520.

[16] Yoram de Hazan // J. Am. Ceram. Soc. 89(2006) 2394.

[17] Bala P. C. Raghupathyw and Jon G. P. Binner// J. Am. Ceram. Soc. 94 (2011) 42.

[18] Hong J. Choi, Ta Y. Yang, Seog Y. Yooa,Byung K. Kim and Hong Ch. Park //Materials Chemistry and Physics 133 (2012)16.

[19] Mei-Hwa Lee, Hung-Yin Lin and JamesL. Thomas // J. Am. Ceram. Soc. 89 (2006)3624.

[20] Annemarie Preiss, Bo Su, Simon Collins andDavid Simpson // J. Europ. Ceram. Soc. 32(2012) 1575.

[21] Changqing Hong, Xinghong Zhang, JiecaiHan, Jiancong Du and Wenbo Han // ScriptaMaterialia 60 (2009) 563.

[22] Jiecai Han, Changqing Hong, XinghongZhang, Jiancong Du and Wei Zhang //J. Europ. Cer. Soc. 30 (2010) 53.

[23] Changqing Hong, Xinghong Zhang, JiecaiHan, Jiancong Du and Wei Zhang // MaterialsChemistry and Physics 119 (2010) 359.

[24] Kechao Zhou, Yan Zhang, Dou Zhang,Xiaoyong Zhang, Zhiyou Li, Gang Liu and TimW. Button // Scripta Materialia 64 (2011) 426.

[25] Carolina Tallon, Rodrigo Moreno and M.Isabel Nieto // Int. J. Appl. Ceram. Technol.6 (2009) 324.

[26] Richard H. J. Hannink, P. Kelly and B.Muddle // J. Am. Ceram. Soc. 83 (2000) 461.

[27] J. Chevalier and L. Gremillard // J. Amer. Cer.Soc. 92 (2009). 1901.

[28] Sylvain Deville // Adv. Eng. Math. 10 (2008)155.

[29] Lei Qian and Haifei Zhang // J. Chem TechnolBiotechnol 86 (2011) 172.

[30] Takayuki Fukasawa, Motohide Ando, TatsukiOhji and Shuzo Kanzaki // J. Am. Ceram.Soc. 84 (2001) 230.

[31] Stephen W. Sofie // J. Am. Ceram. Soc. 90(2007) 2024.

[32] Sakchai Wittaya-Areekul and Steven L. Nail// J. Pharm. Sci. 87 (1998) 491.

[33] J.H. Kim, J.H. Lee, T.Y. Yang, S.Y. Yoon,B.K. Kim and H.C. Park // CeramicsInternational 37 (2011) 2317.

[34] H. Zhang, J.Long and J. Cooper // J. Am.Chem. Soc. 127 (2005) 13482.

[35] Andrew L. Cooper // Adv. Mater. 15 (2003)1049.

[36] Andrew L. Cooper // Adv. Mater. 13 (2001)1111.

[37] I. M. Vasserman, Chemical deposition fromaqueous solutions (Mir, Moscow, 1980), inRussian.

[38] Haifei Zhang and Andrew I. Cooper // Adv.Mater. 19 (2007) 1529.

[39] Haifei Zhang, David Edgar, Patricia Murray,Aleksandra Rak-Raszewska, LaurenceGlennon-Alty and Andrew I. Cooper // Adv.Funct. Mater. 18 (2008) 222.

[40] M. Generalov, Cryochemichal nanotechnology(Akademkniga, Moscow, 2006), In Russian.

[41] N. Uchida, T. Hiranami, K. Tanaka andS. Uematsu // Am. Ceram. Soc. Bull. 81(2002) 57.


[42] S. Deville, E. Maire, A. Lasalle, A. Bogner,C. Gauthier, J. Leloup and C. Guizardz //J. Am. Ceram. Soc. 92 (2009) 2489.

[43] Sylvain Deville, Eric Maire, Audrey Lasalle,Agnes Bogner, Catherine Gauthier, JeromeLeloup and Christian Guizardz // J. Am.Ceram. Soc. 92 (2009) 2471.

[44] Sylvain Deville, Eric Maire, Audrey Lasalle,Agnes Bogner, Catherine Gauthier, JeromeLeloup and Christian Guizard // J. Am.Ceram. Soc. 93 (2010) 2507.

[45] V.G. Konakov, O.Y. Kurapova, N.V. Borisova,S.N. Golubev, E.N. Solovieva and V.M.Ushakov // Bulletin of SPBU 4 (2011) 49.

[46] Andrea Bareggi, Eric Maire, Audrey Lasalleand Sylvain Deville // J. Am. Ceram. Soc. 94(2011) 3570.

[47] A.S. Sinitsky, V.A. Ketsko, I.V. Pentin,G.P. Murav’eva, A.L. Il’inskii and N.N.

Oleynilov // Russ. J. Inorg. Chem. 48 (2003)484.

[48] Xiaodi Wang, Ying Ma, Shanghua Li, Bin Zhuand Mamoun Muhammed // InternationalJournal of Hydrogen Energy (2011),doi:10.1016/j.ijhydene.2011.10.061.

[49] MicroComputed Tomography: Methodologyand Applications (CRC Press Boca Raton,2008).

[50] Jussi Timonen, Markko Myllys, VladimirG. Konakov, Anatoly V. Soudarev and IvanYu. Archakov // Rev. Adv. Mater. Sci. 29(2011) 13.

[51] L. Barin and O. Knacke, Thermochemicalproperties of inorganic substatces (Springer,1973).

Documents

CRYOCHEMICAL METHODS FOR MANUFACTURING NANOSIZED … · when it comes to sensors, turbine engines, and fuel cells fabrication. Being studied over the whole com-position range, zirconia