Perovskite membranes by aqueous combustion synthesis: synthesis

Separation and Purification Technology 25 (2001) 117–126

Perovskite membranes by aqueous combustion synthesis:synthesis and properties

Alexander S. Mukasyan, Colleen Costello, Katherine P. Sherlock,David Lafarga, Arvind Varma *

Department of Chemical Engineering, Uni�ersity of Notre Dame, Notre Dame, IN 46556 USA

Abstract

The objective of this work is to identify optimum synthesis, compacting and sintering conditions in order to achievea pure phase fully densified La0.8Sr0.2CrO3 (LSC) perovskite membrane. The aqueous combustion synthesis of LSCpowders was investigated over a wide range of synthesis conditions by using the metal nitrates (oxidizer)–glycine(fuel) system. The powders were pressed and sintered to create dense materials, which were characterized. It wasshown that depending on fuel/oxidizer ratio, �, the reaction can proceed in three different modes: SmolderingCombustion Synthesis (SCS), ��0.7, with maximum temperature, Tm�600°C; Volume Combustion Synthesis(VCS), 0.7��1.2, 1150°C�Tm�1350°C; Self-propagating High-temperature Synthesis (SHS), 1.2��1.6,800°C�Tm�1100°C. In turn, the characteristics of synthesized powders depend on the combustion mode. Thecrystalline structure of as-synthesized powders becomes more defined as � increases (amorphous for SCS; crystallinefor VCS and SHS). The specific surface area decreases slightly when mode changes from SCS (�25 m2 g−1) to VCS(�20 m2 g−1), however, it increases substantially under SHS conditions (up to 45 m2 g−1). It was also shown thatcalcination is beneficial only for SCS powders, while VCS and SHS powders may be sintered directly as synthesized,thus bypassing the time and energy consuming calcination step. The measured oxygen permeation values for themembranes are comparable with the best candidate materials reported in the literature. © 2001 Elsevier Science B.V.All rights reserved.

Keywords: Perovskite membranes; Combustion synthesis; Membrane microstructure; Sintering; Oxygen selective transport

www.elsevier.com/locate/seppur

1. Introduction

Perovskites are inorganic complex oxides withthe empirical formula ABO3, where A-site cationsare typically rare earth metals, while B-sites areoccupied by transition metals with mixed valence

states. Some perovskites exhibit high mixed elec-tronic and oxygen ionic conductivities, and forthis reason are being widely studied for applica-tions in solid oxide fuel cells (SOFCs), oxygensensors and pumps, batteries, and oxygen-perme-able membranes [1,2].

Among these materials, Lanthanum chromite(LC) perovskite shows high electronic conductiv-ity, and stability under both oxidizing and reduc-ing conditions. However, its applications are

* Corresponding author. Tel.: +1-219-6316491; fax: +1-219-6318366.

E-mail address: [email protected] (A. Varma).

1383-5866/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.

PII: S 1383 -5866 (01 )00096 -X

A.S. Mukasyan et al. / Separation/Purification Technology 25 (2001) 117–126118

limited by poor sinterability in air, which can beimproved by use of dopants such as Strontium andCalcium [3], but their effective use requires rela-tively high concentrations and very high sinteringtemperatures.

In the present work, we focus on the Sr-dopedlanthanum chromites (LSCs). These perovskitesare prepared by a variety of wet-chemical methods,including hydrothermal sol-gel synthesis, coprecip-itation, spray-pyrolysis, liquid and amorphous cit-rates, and freeze-drying [4]. However, it remains achallenge to synthesize such powders rapidly, withdesired phase composition, high purity, high sur-face area and narrow particle size distribution. Thecombustion synthesis (CS) technique offers a po-tential solution to this problem.

In general, CS is an energy-efficient method forthe production of advanced materials, includingpowders and net-shape articles [5,6]. Characterizedby fast heating rates, high temperatures, and shortreaction times, CS occurs typically by two modes.In the self-propagating high-temperature synthesis(SHS) mode, powder reactants are heated locallyand the reaction follows a wave-like propagationthrough the medium, while in volume combustionsynthesis (VCS), reactants are heated uniformlyand reaction occurs simultaneously throughout themixture.

A modification of the CS method is aqueouscombustion synthesis, which uses precursoraqueous solutions containing nitrates of the de-sired metals and a fuel, such as glycine, hydrazine,or urea. This process has been used to synthesizea variety (more than 100) of fine particle complexoxides, such as chromites, ferrites and zirconia[7–9]. Powders prepared by aqueous CS generallyhave high surface areas, as well as pure phase andchemical compositions. Chick et al. [10] have re-ported the only work available in the literature onthe aqueous CS of LSC powders. They investi-gated a relatively narrow range of reaction mixturecompositions (corresponding to the VCS mode),and did not report data on sinterability and otherproperties of materials.

In this paper, we investigate the aqueous CS ofLa0.8Sr0.2CrO3 powders over a wide range of syn-thesis conditions. The resulting powders are char-

acterized for phase composition, microstructureand surface area. The powders are pressed andsintered to create dense materials, which are alsocharacterized. The objective of this work is toidentify optimum synthesis, compacting and sin-tering conditions in order to achieve a pure phasefully densified membrane.

2. Experimental

2.1. Aqueous combustion synthesis of lanthanumstrontium chromite powders

Aqueous CS of La0.8Sr0.2CrO3 perovskite wasconducted by using the metal nitrates–glycinesystem. In this case, the overall combustion reac-tion can be represented as follows:

0.8La(NO3)3·6H2O(c)+0.2Sr(NO3)2(c)

+Cr(NO3)3·9H2O(c)+nH2N(CH2)CO2H(c)

+ (2.25n−7.2)O2(g) � La0.8Sr0.2CrO3(c)

+ [2nCO2+ (13.8+2.5n)·H2O

+ (n/2+2.9) N2](g)�.

Some characteristics of the chemical reagents(Aldrich) are given in Table 1. The ratio betweennitrates (oxidizers) was maintained constant corre-sponding to the desired phase composition, whileratio between glycine (fuel) and oxidizer (metalnitrates), �, was changed. Note that the stoichio-metric ratio (�=1) corresponds to the situationwhere La0.8Sr0.2CrO3 composition can be formeddirectly from the reaction between fuel and oxi-dizer, with no additional oxygen required (n=3.2).

The fuel and oxidizers were mixed in distilledwater, thoroughly stirred to reach complete disso-lution of all solid reagents, and placed on a hot

Table 1Some characteristics of used powders

Powder Melting point (°C)Purity (%)

240–24598H2N(CH2)CO2H40La(NO3)3·6H2O 99.9

Cr(NO3)3·9H2O 609957099Sr(NO3)2

A.S. Mukasyan et al. / Separation/Purification Technology 25 (2001) 117–126 119

plate (Cole-Parmer) to initiate the reaction. Atwo-wave length pyrometer (Ircon MR-OR10-99C) and thermal video system (TVS-2000MkII,Cincinnati Electronic Corp.) were used to recordthe process temperature– time history. As temper-ature reached 100°C, water started to boil andevaporate from the solution, which increased so-lution viscosity substantially, culminating in rapidreaction. Depending on the value of �, synthesisof perovskite powder occurred in three differentregimes (see Section 3.1 for details).

The as-synthesized powders were generallyheat-treated (calcined) in air atmosphere. Specifi-cally, powders were mounted in a hollow boronnitride cylinder, and calcined at 850°C for 12 h.There are two main reasons for this procedure.The first is to remove residual water and othereasily volatile components, while the second is toimprove crystalline structure of the powders. Thelatter is especially important in the case of pow-ders produced from initial mixtures with ��0.7,which have amorphous structure immediately fol-lowing CS. However, it appears that calcinationalso influences the microstructure of the powders(see Section 3.2).

2.2. Pero�skite membrane preparation

Following calcination, the powders werepressed in stainless steel cylindrical dies with di-ameters 13 and 30 mm, to produce green diskswith thickness 1–2 mm. Using automatic hy-draulic press (Carver, Inc., ‘Auto’ series), thepowder was subjected to uniaxial pressure in therange 2000–25 000 lbs, for a period 30–60 s. Theobtained disks had initial densities in the range20–60% of theoretical value.

To achieve higher final density, the green sam-ples were sintered in air atmosphere. For this, thedisks were mounted on a boron nitride platformand heat treated for 10 h at temperature 1600°C,using a high-temperature split-tube furnace(Thermcraft).

2.3. Materials characterization

Several techniques were used for comprehensiveanalysis and characterization of the synthesized

powders and materials. Phase composition analy-sis was conducted using a Rigaku X-Ray Diffrac-tometer (XRD). To investigate the productmicrostructure, a scanning electron microscope(SEM, JEOL) was utilized, primarily in the Sec-ond Electron Images (SEI) mode. Finally, theBET surface area of powders was measured byusing the nitrogen adsorption/desorption method(Autosorb-1, Quantachrome).

A gas-tight electrochemical cell was used tomeasure the oxygen permeability in steady-stateconditions. Details about the setup are given else-where [11]. The device allows achieving reducedoxygen environments in the cell, using flowing airas the surrounding reference atmosphere. The sin-tered membrane was polished to the appropriatethickness (�1.3 mm), and its sides were coveredwith Pyrex glass to prevent oxygen from entering.Under steady state conditions, the amount ofoxygen entering the cell (by permeating throughthe membrane) equals that pumped out. In thiscase, oxygen flow through the membrane can bedetermined as follows:

jO2=I

4FS,

where jO2 is oxygen flux (mol cm−2 s−1), I thepumping current (A), S the effective cross-sec-tional area of membrane (cm2) and F the Fara-day’s constant.

3. Results

3.1. Combustion synthesis

The plot of maximum reaction temperature as afunction of fuel/oxidizer ratio, �, as well as thecharacteristic combustion regimes is shown in Fig.1. As noted above, depending on �, the reactioncan proceed in three different modes:1. Smoldering Combustion Synthesis (SCS), ��

0.7, with maximum temperature, Tm�600°C;2. Volume Combustion Synthesis (VCS), 0.7�

��1.2, 1150°C�Tm�1350°C;3. Self-propagating High-temperature Synthesis

(SHS), 1.2��1.6, 800°C�Tm�1100°C.


Fig. 1. Maximum combustion temperature achieved for different combustion modes as a function of the fuel/oxidizer ratio.

The SCS mode is characterized by relativelyslow, essentially flameless reaction, with forma-tion of an amorphous ash as solid product. Ex-tremely fast (explosion-type) reactioncharacterized the VCS mode. In this case, thereaction occurs essentially simultaneously in thesolution volume. Finally, the characteristic featureof the SHS mode is that reaction initiates locallyand propagates as a combustion wave in a self-sustained manner through the reaction volume.

3.2. Characterization of as-synthesized powders

It appears that different powders synthesized inthe SCS mode are amorphous; no peaks can beidentified on the XRD patterns (Fig. 2a). How-ever, the VCS and SHS powders have identicalwell-defined crystalline structures (Fig. 2b). Asearch-match database analysis confirmed thatthese powders have phase composition ofLa0.8Sr0.2CrO3 perovskite.

The characteristic microstructure of SCS pow-der is presented in Fig. 3a. It may be seen thatagglomerates involved thin (100 nm) flakes withsmooth surface, perforated by many large pores.Fig. 3b shows typical morphology of VCS pow-

ders. It is interesting that despite different synthe-sis conditions, this microstructure is similar to theprevious one. On the other hand, the SHS powderhas somewhat different morphology (see Fig. 3c).In this case, the surface pores decrease substan-tially and conglomerates are not thin flakes butwaffles, which have complex high surface areanetwork structure. The data for specific surfacearea of different powders are presented in Table 2.It can be seen that SHS powder has larger specificsurface as compared to the SCS and VCSproducts.

3.3. Characterization of calcined powders

The mass of SCS powder decreased after calci-nation, up to 20%, while mass loss of VCS andSHS powders was not more than 1%. The X-raydiffraction analysis shows that upon calcination,the amorphous as-synthesized SCS powders be-came well crystalline with phase composition sim-ilar to those for VCS and SHS powders. Thelatter remained essentially unchanged in phasecomposition during calcination. However, thepeaks of all calcined powders exhibited a reduc-tion in broadness as compared to the correspond-ing as-synthesized patterns.


It is remarkable that microstructures of allpowders changed dramatically with calcination(compare Fig. 3 a–c and d– f). It can be seen thatthe calcined powders do not exhibit flake orwaffle type microstructure and are extremely uni-form with well defined morphology of each parti-cle. Also, the characteristic size of particlesincreased after calcination. This leads to largedecreases in specific surface area of powders asshown in Table 2. Note that the powder synthe-sized in SHS mode is the most homogeneous andwith the highest surface area among all calcinedsamples.

3.4. Compacting and sintering

To produce fully dense membrane by sintering,it is favorable to have high initial density of greendisks. The latter can be achieved by increasingapplied load during pressing. However, above acertain threshold load, disks could not maintainintegrity and would fracture or cleave. For exam-ple, for the 13 mm diameter press-die, samplescompacted from as-synthesized SCS powders wereobserved to have threshold load �20 000 lb,while for uncalcined VCS and SHS powders itwas only 12 000 lb. Disks pressed from calcined

Fig. 2. XRD patterns for as-synthesized powders by (a) SCS, and (b) VCS/SHS modes.


Fig. 3. Characteristic microstructures of (a–c) as-synthesized and (d– f) calcined LSC powders produced by (a, d) SCS, (b, e) VCS,and (c, f) SHS modes. Magnification: 5000× .

powders had much lower threshold (�2400 lb).Note that for maintaining final material purity, nobinder was used. In Table 3, relative densities ofgreen disks at the threshold load are presented. Itcan be seen that different as-synthesized powderscan be compacted to essentially the same initialdensity. Also, calcination in general leads to in-creasing initial sample density, which is related tothe microstructural transformations occurringduring calcination.

The pressed disks were sintered following theprocedure described in Section 2.2. A shrinkage inheight and diameter was observed in all sintereddisks. The data for final sample densities reachedafter sintering are presented in Table 3.

3.5. Microstructure of pero�skite membranes

Cross sections of sintered membranes were pre-pared and examined in the SEI mode. The charac-teristic microstructure of membrane sintered fromuncalcined SCS powder is shown in Fig. 4a. Thismembrane has a final relative density of 69%, andaverage grain size 2 �m. It may be seen that anumber of relatively large pores (�10 �m) char-

acterize the structure. Essentially no large poresare observed when disk of the SCS calcined pow-der was sintered (Fig. 4d) which yields higher finaldensity (77%); however, the grain size is larger (4�m). Although the microstructure is uniform,there are noticeable gaps between particles andoverall particle boundaries are distinct. A highermagnification of this sample is also shown in Fig.4d, and reveals only initial stages of neck forma-tion between the grains.

Fig. 4b represents the typical cross-section of amembrane (relative density 72%) produced fromuncalcined VCS powder. This uniform mi-crostructure is characterized by grains of 1.5 �mdiameter, narrow particle size distribution and

Table 2Characteristic surface areas of powders synthesized in differentcombustion modes

Specific surface areaPowder type Specific surface areaof calcined powderof as-synthesized(m2 g−1)powder (m2 g−1)

2SCS 27VCS 620

2441SHS


Table 3Relative initial and final density of perovskite membranes

Load (lbs) Initial relativeCalcination Final relativePowder type Increase of density after sinteringdensity (%) (%)density (%)

20 000 43No 59SCS 1612 000 46VCS 72No 2612 000 44No 77SHS 33

2400 52 77 25SCS Yes2400 52Yes 90VCS 382400 47 85 38SHS Yes

Fig. 4. Characteristic microstructures of (a, c, e) uncalcined and (b, d, f) calcined sintered LSC powders produced by (a, b) SCS,(c, d) VCS, and (e, f) SHS modes. Magnification: ×1000.

absence of large pores. Fig. 4e shows the mi-crostructure of membrane sintered from calcinedVCS powder, which reached relative density of90%. A higher magnification of this sample showsthat in this case formation of necks leads to lossof particle distinction.

Fig. 4c shows the sintered cross-section of amembrane produced from SHS powders that werenot previously subjected to calcination. Thismembrane reached a sintered relative density of77%. There are virtually no distinct particles visi-ble in this sample, however several relatively large(2–5 �m) irregular-shaped pores can be observed.

The sample (relative density 85%) sintered formSHS calcined powders is shown in Fig. 4f, whichexhibits similar characteristics as membrane fromthe calcined VCS powder (Fig. 4e).

3.6. Oxygen permeability of pero�skitemembranes

The oxygen transport rate through the per-ovskite membranes was measured in the range650–850°C using the approach discussed in Sec-tion 2.3. The oxygen flux through the sample (1.3mm-thick disk; SCS-calcined) as a function of


oxygen pressure inside the cell at five differenttemperatures is shown in Fig. 5. It may be seenthat oxygen permeability increases with both in-creasing temperature and oxygen partial pressuregradient. For example, at 850°C, the oxygen per-meability is �2×10−7 mol cm−2 s−1 at�log(pO2)=1.5.

4. Discussion

4.1. Combustion synthesis of powders

In Fig. 6, it is seen how characteristics ofsynthesized powders depend on the combustionmode (or �). Progressive increase of fuel amount

Fig. 5. Oxygen permeability of sintered LSC disk (�=0.5) as a function of oxygen partial pressure inside the cell and differenttemperatures.

Fig. 6. Characteristics of synthesized powders, combustion modes and BET surface areas as functions of the fuel/oxidizer ratio; opensymbols — as-synthesized, filled symbols — calcined.


(�) leads to the following sequence of reactionmodes: SCS�VCS�SHS. Simultaneously, thecrystalline structure of as-synthesized powders be-comes more defined (amorphous for SCS�crys-talline for VCS and SHS). The specific surfacearea decreases slightly, when mode changes fromSCS to VCS, however under SHS conditions itincreases substantially. All these changes can beexplained based on the temperature-time charac-teristics of the reaction modes. Thus, relativelylow temperature (600°C) during SCS does notallow for crystalline structure to form. However,long time of reaction (minutes) leads to the rela-tively low (�25 m2 g−1) specific surface area. Athigh temperatures (�1300°C) of VCS mode, thecrystalline structure of formed perovskite is welldefined, which occurs even during very short (�1–5 s) reaction time. On the other hand, thesehigh temperatures lead to some decrease of pow-der specific surface area (up to 20 m2 g−1). De-crease of reaction temperature as well as shortreaction time (�5 s), which characterizes theSHS mode, results in large increase of powdersurface area (up to 45 m2 g−1). It is interestingthat in this case the temperature is still highenough to produce desired crystalline structure.Thus powders with both fine microstructure (�100 nm; 45 m2 g−1) and developed crystallinitycan be directly formed only during reaction in theSHS mode. It is remarkable that SHS powdersproduced at higher temperatures have muchhigher specific surface area as compared to SCSpowders synthesized at much lower temperatures.

4.2. Calcination

The first effect of calcination process is in gen-eral favorable, predictable and related to the im-provement of crystalline structure ofas-synthesized powders. In fact, XRD analysisrevealed that this additional heat treatment trans-formed the SCS powder from amorphous to crys-talline form. This process also removed residualcarbon and other impurities. For VCS and SHSpowders, it was shown that after calcination allcharacteristic peaks became more intense and nar-rower. The second effect is not obvious and isrelated to the substantial changes in all powder

microstructures. On one hand, calcined powdershave more uniform morphologies (see Fig. 3),which is a positive effect. On the other hand,specific surface areas of powders decrease (seeTable 2), which is negative because it diminishessinterability. All these observations lead to theconclusion that calcination may be beneficial onlyfor SCS powders, while VCS and SHS powdersmay be sintered directly as synthesized, thus by-passing the time and energy consuming calcina-tion step.

4.3. Membrane preparation

It was observed that calcined powders, as com-pared to as-synthesized ones, have much lowerthreshold loads above which defects formed dur-ing compacting. This is related to changes ofmorphology and specific surface area of powdersthat occur during calcination. However, even un-der these lower loads, higher initial densities wereobtained for calcined powders (see Table 3). Notethat this effect is more pronounced for powderssynthesized from solutions with lower � composi-tions. For example, the differences in initial sam-ple densities, between calcined and as-synthesizedpowders, are 9, 6 and 3% for SCS, VCS and SHSmodes, respectively. Thus, compactability of SHSpowders is not much enhanced by the calcinationprocedure, again suggesting that this step may beomitted.

It is apparent that two main factors influencethe final membrane density reached after sinteringat high temperature (1600°C): initial sample den-sity and powder sinterabilty. The former is higherfor powders with lower surface area, while thelatter is favored for those with higher surface area(Tables 2 and 3). Because these two factors haveopposite dependence on surface area, some opti-mization is required to reach the highest densityof sintered membrane. From Table 3, it appearsthat the calcined VCS powder leads to the highesttheoretical density (90%). However, microstruc-ture of sintered uncalcined SHS powder is ingeneral more dense than all others (see Fig. 4),except for isolated regions containing large pores.The reason for these pores may be small amountof low volatile impurities that remain in as-syn-


Table 4Permeation data of representative oxygen conductors

Temperature (°C) Thickness (mm) �PO2 (atm) ReferenceMaterials Flux (mol cm−2 s)

850 1.3La0.8Sr0.2CrO3 0.202×10−7 This workSrFeCo0.5Ox 10−7 850 2.9 0.20 [12]La0.3Sr0.7CoO3−� 1.6×10−6 1100 0.6 0.21 [13]

1150 14×10−7 0.21La0.4Sr0.6CoO3−� [14]900 2 0.21 [15]Y0.05Ba0.95CoOx 5×10−8

1000 0.01 0.20 [16]3×10−8CeO2–YSZ

thesized SHS powders. It is expected that theseimpurities can be removed by low temperatureheat treatment (�400°C), which simultaneouslyshould not significantly decrease powder surfacearea. The optimization of conditions for synthesisof fully dense perovskite membranes preparedfrom SHS powders is currently in progress.

4.4. Membrane oxygen permeability

As observed in Fig. 5, oxygen permeabilityincreased with increasing temperature and oxygenpartial pressure gradient. The increase is sharp atlow gradients and asymptotic when oxygen pres-sure inside the cell approaches zero. The tempera-ture dependence and magnitude of oxygen fluxindicate that the membrane is fully dense forpermeation, suggesting that remaining porosity isclosed.

Table 4 shows different oxygen permeation val-ues from the literature for other representativecandidate materials and LSC synthesized in thiswork. Taking into account the different condi-tions (thickness and temperature), LSC exhibitspermeation performance comparable with thebest.

Acknowledgements

We gratefully acknowledge financial supportfrom the National Science Foundation (grants

CTS-9900357 and CTS-9907321). We also thankDrs U. Balachandran and B. Ma from ArgonneNational Laboratory for conducting the oxygenpermeation measurements.

References

[1] B.C.H. Steele, Mater. Sci. Eng. B13 (1992) 79.[2] N.Q. Minh, J. Am. Ceram. Soc. 76 (1993) 563.[3] M. Mori, Y. Hiei, N.M. Sammes, Solid State Ionics 103

(1999) 123.[4] P. Cousin, R.A. Ross, Mater. Sci. Eng. A130 (1990)

119.[5] J.J. Moore, H.J. Feng, Prog. Mater. Sci. 39 (1995) 243.[6] A. Varma, A.S. Rogachev, A.S. Mukasyan, S. Hwang,

Adv. Chem. Eng. 24 (1998) 79.[7] K.C. Patil, S.T. Aruna, S. Ekambaram, Curr. Opin. Solid

State & Mat. Sci. 2 (1997) 158.[8] M.M.A. Sekar, A. Halliyal, J. Am. Ceram. Soc. 81 (1998)

380.[9] S. Bhaduri, S.B. Bhaduri, E. Zhou, J. Mater. Res. 13

(1998) 156.[10] L.A. Chick, L.R. Pederson, G.D. Maupin, J.L. Bates,

L.E. Thomas, G.J. Exarhos, Mat. Lett. 10 (1990) 6.[11] B. Ma, U. Balachandran, J.H. Park, C.U. Segre, J. Elec-

trochem. Soc. 143 (1996) 1724.[12] B. Ma, U. Balachandran, C.-C. Chao, J.H. Park, C.U.

Segre, Ceram. Trans. 73 (1997) 169.[13] C.H. Chen, H.J.M. Bouwmeester, R.H.E. van Doorn, H.

Kruidhof, A.J. Burggraaf, Solid State Ionics 98 (1997) 7.[14] N. Itoh, T. Kato, K. Uchida, K. Haraya, J. Memb. Sci.

92 (1994) 239.[15] H.W. Brinkman, H. Kruidhof, A.J. Burggraaf, Solid

State Ionics 68 (1994) 173.[16] J. Han, Y. Zeng, G. Xomeritakis, Y.S. Lin, Solid State

Ionics 98 (1997) 63.

Documents

Perovskite membranes by aqueous combustion synthesis: synthesis