influence of the sol-gel synthesis on the formation of spinel mgal2o4

8
INFLUENCE OF THE SOL-GEL SYNTHESIS ON THE FORMATION OF SPINEL MgAl 2 O 4 Julien Parmentier, Mireille Richard-Plouet, and Serge Vilminot* Groupe des Mate ´riaux Inorganiques, IPCMS, UMR 46 CNRS, 23, rue du Loess, 67037 Strasbourg Cedex, France (Communicated by J. Rouxel) (Received February 17, 1998; Accepted February 18, 1998) ABSTRACT Spinel precursors have been obtained by sol-gel processing using either magnesium nitrate and chemically modified aluminum alkoxide or a double alkoxide as raw materials. The xerogels have been characterized by differen- tial thermal analysis (DTA), X-ray diffraction (XRD), infrared (IR) spectros- copy, and transmission electron microscopy (TEM). It is proposed that the size of the particles is related to the functionality of the aluminum alkoxide. The use of the double alkoxide allowed pure spinel nanosized materials to be obtained. © 1998 Elsevier Science Ltd KEYWORDS: A. oxides, B. sol-gel chemistry, C. electron microscopy, C. X-ray diffraction, D. microstructure INTRODUCTION Magnesium aluminate spinel (MgAl 2 O 4 ) is a well-known mineral usually used as a refrac- tory. Indeed, its high melting temperature (2135°C), good chemical inertness, and mechanical properties give this material very attractive properties and applications. Nevertheless, the range of applications for this compound can be limited by the presence of impurities (in the mineral sample) or by fabrication difficulties (e.g., for coating). To resolve these problems, different synthesis methods have been developed. The use of mixed powders is the classical and cheapest route, but high temperature heat treatments must be carried out to achieve complete reaction (at least 100 h at 1300°C even with submicron particles [1]). Other *To whom correspondence should be addressed. Materials Research Bulletin, Vol. 33, No. 11, pp. 1717–1724, 1998 Copyright © 1998 Elsevier Science Ltd Printed in the USA. All rights reserved 0025-5408/98 $19.00 1 .00 PII S0025-5408(98)00169-X 1717

Upload: julien-parmentier

Post on 15-Sep-2016

219 views

Category:

Documents


6 download

TRANSCRIPT

INFLUENCE OF THE SOL-GEL SYNTHESIS ON THE FORMATION OFSPINEL MgAl 2O4

Julien Parmentier, Mireille Richard-Plouet, and Serge Vilminot*Groupe des Mate´riaux Inorganiques, IPCMS, UMR 46 CNRS, 23, rue du Loess,

67037 Strasbourg Cedex, France

(Communicated by J. Rouxel)(Received February 17, 1998; Accepted February 18, 1998)

ABSTRACTSpinel precursors have been obtained by sol-gel processing using eithermagnesium nitrate and chemically modified aluminum alkoxide or a doublealkoxide as raw materials. The xerogels have been characterized by differen-tial thermal analysis (DTA), X-ray diffraction (XRD), infrared (IR) spectros-copy, and transmission electron microscopy (TEM). It is proposed that thesize of the particles is related to the functionality of the aluminum alkoxide.The use of the double alkoxide allowed pure spinel nanosized materials to beobtained. © 1998 Elsevier Science Ltd

KEYWORDS: A. oxides, B. sol-gel chemistry, C. electron microscopy, C.X-ray diffraction, D. microstructure

INTRODUCTION

Magnesium aluminate spinel (MgAl2O4) is a well-known mineral usually used as a refrac-tory. Indeed, its high melting temperature (2135°C), good chemical inertness, and mechanicalproperties give this material very attractive properties and applications. Nevertheless, therange of applications for this compound can be limited by the presence of impurities (in themineral sample) or by fabrication difficulties (e.g., for coating). To resolve these problems,different synthesis methods have been developed. The use of mixed powders is the classicaland cheapest route, but high temperature heat treatments must be carried out to achievecomplete reaction (at least 100 h at 1300°C even with submicron particles [1]). Other

*To whom correspondence should be addressed.

Materials Research Bulletin, Vol. 33, No. 11, pp. 1717–1724, 1998Copyright © 1998 Elsevier Science LtdPrinted in the USA. All rights reserved

0025-5408/98 $19.001 .00

PII S0025-5408(98)00169-X

1717

methods have been investigated, such as the coprecipitation of Mg and Al hydroxides [2,3]and the sol-gel process [4–6]. These are based on the use of soluble chemical precursors,leading to a molecular level mixing of Al31 and Mg21 and allowing a decrease of thereaction temperature (800–900°C for spinel). More recently, potential applications of thiscompound in relation to its infrared transparency [7,8] chemical inertness (for protectivecoatings, ultrafiltration membranes) [9,10], and superplasticity (nanoparticles) [11] promotedfurther studies. The properties of the final material are highly dependent on the synthesisparameters used, especially during the liquid phase step. A better knowledge of the influenceof these parameters would allow improved control of the final form of the material. The aimof this study is to determine the effects of sol-gel synthesis conditions on some physico-chemical properties of spinel such as particle size and crystallographic purity.

EXPERIMENTAL

Synthesis.Two routes have been considered, the first one using a double alkoxide asprecursor (DA route) and the second using magnesium nitrate, Mg(NO3)2z6H2O, and alu-minum sec-butoxide. In the first route, the double alkoxide, MgAl2(OR)8 [12], has theadvantage of having the same cation ratio as the spinel phase. Moreover, these doublealkoxides appear to be soluble in their parent alcohol, which is not always the case for Mgand Al single alkoxides. The use of a double alkoxide could also prevent the differentialhydrolysis currently observed when starting from a mixture of simple alkoxides. In thesecond route, called the semi-alkoxide route (SA route), the gel network is built up fromhydrolysis–condensation reactions of one alkoxide, the second element being dissolved inthe solvent. The gel formation can be controlled using a chemical modification of thealkoxide,b-diketones (AcacH:C5H8O2; EtacH:C6H10O3) being currently used for this mod-ification. Magnesium acetate and nitrate are both very soluble in alcohols. The latter wasselected due to its good oxidizing capability; unlike acetate, it does not yield organic residuesafter drying. For both routes, hydrolysis was performed by simultaneously adding analcohol/water mixture and stirring vigorously. The hydrolysis ratio ([H2O]/[OR]) was equalto 3 and the alcohol volume was adjusted to achieve a final Al concentration of 0.5 M. Thegels were dried at 40°C. Subsequent heat treatments were carried out at a rate of 1°C/min.

Semi-alkoxide route. Aluminum sec-butoxide and magnesium nitrate (Mg(NO3)2z6H2O)were dissolved separately in isopropanol (IPA). Water was then added to the magnesiumnitrate solution in order for it to be used as a hydrolyzing solution with the correct hydrolysisratio. Where necessary, the required amount ofb-diketone modifier (ModH: EtacH orAcacH) was diluted in IPA and added while stirring vigorously, to give the modified alkoxide[13] according to the following reaction:

Al(OsBu)3 1 xModH3 Al(OsBu)3 2 xModx 1 xsBuOH

According to the degree of modification (x) and the nature of the modifier, the samples werelabeled, for example, Mod0, Etac1, Etac2, and Acac1.

Double alkoxide route. Aluminum and magnesium in stoichiometric quantities were re-acted under reflux with excess methoxyethanol under flowing argon. A little I2 was used asa catalyst for the alcohol/metal reaction [14]. After 72 h, a clear solution was obtained withsmall amounts of residual metals, which were separated by centrifuging. Hydrolysis was

1718 J. PARMENTIERet al. Vol. 33, No. 11

performed by the addition of a water/methoxyethanol solution (with a hydrolysis ratio [H2O]/[OR] equal to 1 for IR spectroscopy and 1/3 for TEM studies). A gel formed immediately.

Characterization methods. Different methods of characterization were used on the dry geland the calcined xerogel. Differential thermal analysis (DTA) coupled with thermogravimet-ric analysis (TGA) was performed on a Setaram TGDTA92 using a heating rate of 6°C/minin air. With some of the dried gel samples, the crystallization peak of the spinel phase couldnot be detected. This was most probably due to the small amount of sample remaining at thistemperature. Therefore, in such cases, the samples used were heat-treated at 500°C before-hand. Infrared spectra were recorded using the KBr pellet technique on a Nicolet FTIRapparatus in the range 4000–400 cm21. X-ray diffraction powder patterns (u–2u) wereobtained on a Siemens D500 apparatus using Co Ka1 monochromatized radiation. Thecell-parameter calculation for the spinel phase was performed with the program ERACEL[15]. The apparent crystallite size was determined using the Laue-Scherrer formula (eq. 1):

b(2u ) 5 l/[Lcos(u0)] (1)

wherel is the wavelength andu0 is the Bragg angle. The width of the diffraction lineb(2u)(in radians) was taken as the experimental half-width (bexp) and was corrected for experi-mental broadening (binstr) according to eq. 2 [16]:

b(2u ) 5 (b2exp 2 b2

instr)1/ 2 (2)

binstr was determined experimentally with an MgAl2O4 sample composed of large crys-tallites (.500 nm) and without defects to minimize the diffraction line broadening due tograin size or lattice strain.

Five diffraction lines—(220), (311), (400), (511), and (440)—were chosen for the mea-surement of the apparent crystallite size. These have the advantage of being well separatedand have high intensities (40, 100, 65, 45, and 55, respectively). The diffraction patterns wererecorded using a 2u step of 0.02° and a step time of 15 s. The half-widths were obtained bya deconvolution of the diffraction line using the PEAKFIT software [17]. Crystallite sizeswere also determined by transmission electronic microscopy on a Topcon apparatus operat-ing at 200 kV.

RESULTS AND DISCUSSION

DTA-TGA. The exotherm related to spinel crystallization appeared for all samples in thesame temperature range (770–820°C for the exothermic maximum) and was always asso-ciated with a weight loss (Fig. 1). This behavior has been attributed to the presence ofresidual OH groups distorting the tetrahedral and octahedral sites within the future spinelnetwork. The elimination of these groups (as H2O) could then initiate the crystallization.However, no clear relationship between the processing of the powders and the correspondingcrystallization temperature was apparent.

Infrared spectroscopy. IR spectra were recorded for the xerogels dried at 40°C (Fig. 2). For SAsamples, common bands appear in all cases, such as the broad OH band centered around 3400cm21, the 1630 cm21 H2O vibration band, and the bands related to NO3 groups at 1384 cm21

(strong) and 829 cm21. Below 800 cm21, broad bands are present that are related to the formation

1719SPINEL SYNTHESISVol. 33, No. 11

of the inorganic network. For modified samples, the sharp lines between 1600 and 400 cm21,excluding those related to the NO3 group, are related to the presence of theb-diketonate group.Such lines are clearly seen in the case of the Acac1 sample. Moreover, their positions correspondto those obtained from pure aluminum acetylacetonate, Al(Acac)3.

The spectrum of the Mod0 gel dried at 40°C does not reveal the presence of residual OsBugroups. This indicates the complete hydrolysis of the butoxide groups by water, yielding aninorganic Al–O network by condensation, as confirmed by the large peak at 640 cm21. Acomparison of the spectra obtained with modified Al alkoxides clearly indicates an increase in theb-diketone content following the sequence Etac1 , Etac2 , Acac1. This sequence could beexplained by the labile character of the Etac group in the presence of water, yielding the formationof an inorganic network as shown by the two broad bands in the range 800–400cm21. UnlikeEtac, Acac is a strong chelating agent of Al31 and cannot be removed by water. Thus, Al(Acac)3

was formed in the Acac sample, as detected by X-ray diffraction, but not Al(Etac)3 [18] in theEtac1 and Etac2 samples. The presence of Acac decreases the functionality of the alkoxide,limiting the growth of the inorganic polymer and resulting in a stable solution as observed. Thiscan be related to the IR spectrum, which is very similar to that for Al(Acac)3, where no broadpeaks (in the range 400–800 cm21) characteristic of the inorganic network were detected.

For the DA dried gel, the IR spectrum indicates clearly that hydrolysis was not completeand that some methoxyethoxy groups were still present. Nevertheless, condensation reactionsled to an inorganic network identifiable by the large peaks at 604 and 468 cm21.

X-ray Diffraction

Spinel phase characterization.The dried gels were calcined at 800, 1000, and 1400°C for15, 15, and 3 min, respectively (heating rate 1°C/min) and then analyzed by X-ray diffraction.The unit-cell parameter values obtained for the spinel phase were all between those expectedfor MgAl2O4 (a 5 8.083 Å, JCPDS 33-853) andg-Al2O3 (a 5 7.924 Å, JCPDS 29-63),indicating the formation of a phase within the solid solution MgAl2O4–g-Al2O3, which canbe written as Mg12yAl2O42y. They values were determined from a linear evolution of theunit-cell parameter vs. composition, according to Vegard’s law, for different annealingtemperatures. They values (for different temperatures) and the presence of the residual MgOphase are reported in the Table 1.

As expected the biggest departures from spinel composition are observed for the lowest

FIG. 1DTA and TGA curves for the DA sample.

1720 J. PARMENTIERet al. Vol. 33, No. 11

annealing temperature. However, considering the short annealing time, 15 min, quite lowyvalues (y , 0.08) are observed. The formation of a spinel phase close to the stoichiometricone indicates a good local homogeneity of the samples. The decrease ofy with temperatureis mainly due to the reaction between Mg12yAl2O42y and MgO.

For the DA route, homogeneity between Al and Mg is normally obtained at the molecular

FIG. 2Infrared spectra of the xerogels dried at 40°C: DA (a), Mod0 (b), Acac1 (c), Etac1 (d), andEtac2 (e).

1721SPINEL SYNTHESISVol. 33, No. 11

scale. For these DA samples, whatever the heat treatment, no MgO was detected. Neverthe-less, the highest value ofy was obtained for the DA sample calcined at 800°C, and a decreasein y values with increasing temperature was observed. Therefore, one can propose thatheterogeneities can be present in the sample, but the deviation from ideal composition ishomogeneously distributed in the whole sample, avoiding MgO crystallization. The hetero-geneities could result from preferential hydrolysis of one element (Al or Mg), yielding to thedisappearance of some Mg–O–Al bridges.

In the other cases, MgO was always detected for some of the thermal treatment steps. Thiscould result from heterogeneities that are much more concentrated in some parts of the startinggel. Annealing for longer time periods would be necessary to complete the reactions. Theinfluence of the initial chemical modification was not clearly determined, even though, at thehighest annealing temperature, better results were obtained with increasing Etac amounts,(y(Etac2) , y(Etac1) , y(Mod0)). Modification by Acac seems to be less effective; indeed in thiscase, precipitates of Al(Acac)3 were present, concentrating Al31 at the bottom of the beaker.

Particle-size determination. The relatively low values ofy, as shown in Table 1, particu-larly for 1000°C and 1400°C annealings, seems to reflect a good homogeneity with acomposition close to MgAl2O4. The line broadening is mainly due to the small crystallite size

TABLE 1Spinel Mg(12y)Al2O42y Cell Parameters andy Values After Heat Treatment

at 800°C for 15 min, 1000°C for 15 min, and 1400°C for 3 min

Sample

800°C 15 min 1000°C 15 min 1400°C 3 min

a0 (Å) y (%) a0 (Å) y (%) a0 (Å) y (%)

DA 8.071(10) 8(7) 8.079(1) 2.7(9) 8.082(1) 0.4(7)Mod0 8.078(2) 4(1) 8.079(1)a 2.7(7) 8.080(2)a 2.1(13)Etac1 8.081(3) 1(2) 8.082(3) 0.5(16) 8.080(2)a 1.8(14)Etac2 8.072(4) 7(3) 8.080(1)a 2.1(8) 8.081(1) 1.1(8)Acac1 8.074(2) 6(1) 8.079(2)a 2.5(14) 8.078(2)a 3.1(14)

aPresence of MgO.

TABLE 2Apparent Crystallite Sizes Determined

by XRD

Sample

Apparent crystallite size (nm)a

800°C 15 min 1000°C 15 min

DA 6(2) 19(2)Mod0 14(1) 25(2)Etac1 11(2) 19(2)Etac2 10(2) 34(2)Acac1 8(4) 28(4)

aStandard deviation given in parentheses.

1722 J. PARMENTIERet al. Vol. 33, No. 11

rather than to a variation of composition between the crystallites, as confirmed by TEM. Theresults of the apparent particle size determination are reported in Table 2.

For samples prepared by the semi-alkoxide route and heat-treated at 800°C near thecrystallization temperature, the apparent crystallite size increased following the sequenceAcac1, Etac2, Etac1, Mod0. For the Acac1 sample, Al(Acac)3 was observed; therefore, thediscussion is limited to the other samples. This sequence is similar to that observed fromthe IR spectra and is related to a decrease in the amount of modifier present in the driedgel. We could hypothesize that a decrease of the initial amount of modifier will increasethe functionality of the aluminum alkoxide, promoting the growth of the polymer andthereby the particle size, supposing that the particles are single crystals. This consider-ation seems to be verified by the TEM observations, particularly after annealing at1000°C .

TEM Study. All of the samples annealed at 800 and 1000°C for 15 min were examined by TEM,to confirm the apparent sizes obtained from X-ray diffraction. As illustrated by Figure 3, thecrystallites exhibit a plate-like shape with poorly defined edges. As the platelets overlap, it wasnot possible to observe one of them individually. A clear growth after annealing at 1000°C isobserved in all cases, but particularly in the case of the Acac1 sample, which reveals the presenceof well-formed edges. A mean particle size for each sample was calculated by averagingapproximately ten particle measurements. The results, given in Table 3, are only rough estima-tions; however, they confirm the size domain determined by X-ray diffraction, even if they do notconfirm an evolution within the different samples annealed at the same temperature.

FIG. 3Transmission electron micrograph of the Acac1 sample calcined at 800°C for 15 min (a) and1000°C for 15 min (b); 1 cm5 67 nm.

1723SPINEL SYNTHESISVol. 33, No. 11

CONCLUSION

Various techniques were used to characterize spinel sol-gel precursors and to assess the effectof the processing route on the final materials. A different degree of homogeneity was seen inthe double alkoxide and semi-alkoxide samples. The absence of MgO in the XRD results forthe DA sample indicates a more even distribution of Mg21 is obtained by the DA route thanby the SA route. Increasing the annealing temperature promoted the formation of nearly purespinel phase in the former case, but only reduced the amount of MgO in the latter. However,when the annealing time was increased, complete conversion to spinel was observed in allcases. The reactivity has been related to the presence of nanocrystalline particles in the firststages of the calcination experiments. It is proposed that the size of these particles isdetermined by the ability of the inorganic polymer to grow according to its functionality.

REFERENCES

1. O. Yamagushi, H. Taguchi, and K. Shimizu,Polyhedron6, 1791 (1987).2. R.J. Bratton,Am. Ceram. Soc. Bull.48, 759 (1969).3. R.J. Brattton,Am. Ceram. Soc. Bull.48, 1069 (1969).4. H. Dislich,Angew. Chem., Int. Edn. Engl.10, 363 (1971).5. J.C. Debsikar,J. Mater. Sci.20, 4454 (1985).6. D.Lepkova and A. Batarjav,J. Mater. Sci.26, 4861 (1991).7. R. Raynal, Thesis (Engl. Transl.), University of Limoges, 1987.8. Y. Trambouze,New Treaty of Inorganic Chemistry(Engl. Transl.), No. 6, pp. 596–608, Pascal (1972).9. O. Varnier, N. Hovnanian, A. Larbot, P. Bergez, L. Cot, and J. Charpin,Mater. Res. Bull.29, 479 (1994).

10. K.B. Planz, R. Riedel, and H. Chmiel,Adv. Mater.10, 4 (1992).11. C.R. Bickmore, K.F. Waldner, D.R. Treadwell, and R.M. Laine,J. Am. Ceram. Soc.79, 1419 (1996).12. J. Sassmanshausen, R. Riedel, K.B. Planz, and H. Chmiel,Z. Naturforsch.48b, 7 (1993).13. J.H. Wengrovius, M.F. Garbauskas, E.A. Williams, R.C. Going, P.E. Donahue, and J.F. Smith,

J. Am. Chem. Soc.108, 982 (1986).14. D.C. Bradley, R.C. Mehrotra, and D.P. Gaur,Metal Alkoxides, p. 300, Academic Press, London (1978).15. J. Laugier and A. Filhol,ERACEL(computer program), 1978.16. J.P. Eberhart,Methodes Physiques d’E´ tude des Mine´raux et des Mate´riaux Solides, Doin, Paris

(1976). (in French)17. PEAKFIT, version 4.0, AISN Software Inc., 1995.18. A.G. Charles, N.C. Peterson, and G.H. Franke,Inorg. Synth.9, 25 (1967).

TABLE 3Mean Apparent Crystallite Sizes Determined by TEM

Sample

Apparent crystallite size (nm)a

800°C 15 min 1000°C 15 min

DA 7 26Mod0 9 19Etac1 7 18Etac2 12 23Acac1 13 31

aThe values represent average of ten measurements for eachsample.

1724 J. PARMENTIERet al. Vol. 33, No. 11