M A T E R I A L S C H A R A C T E R I Z A T I O N 6 2 ( 2 0 1 1 ) 2 6 3 – 2 6 7

ava i l ab l e a t www.sc i enced i r ec t . com

www.e l sev i e r . com/ loca te /matcha r

Short communication

Macroporous glass monoliths prepared from powderedniobium phosphate glass by fast sintering

Vitor Lacerda Mauricio, Oswaldo Luiz Alves, Italo Odone Mazali⁎

Instituto de Química, Universidade Estadual de Campinas - UNICAMP, Campinas, SP, Brazil

A R T I C L E D A T A

⁎ Corresponding author. Tel.: +55 19 35213164E-mail address: mazali@iqm.unicamp.br

1044-5803/$ – see front matter © 2011 Elsevidoi:10.1016/j.matchar.2011.01.003

A B S T R A C T

Article history:Received 4 February 2010Received in revised form10 November 2010Accepted 6 January 2011

Macroporous monoliths were prepared by very fast sintering (between 3 and 15 min) ofniobophosphate glass powders at low temperature (1018 K) using cellulose as a foamingagent. The porous materials were analyzed by thermal analysis, Raman spectroscopy,scanning electron microscopy and powder X-ray diffraction, and further investigated usingX-ray microtomography, a non-destructive technique capable of reconstructing three-dimensional models of samples and providing structuralmeasurements. The progression ofthe porosity of the monoliths depends on the sintering time (3 to 15 min) and the amount(up to 50% in mass) of cellulose used. The macroporous glass monoliths may findapplication in integrated chemical systems and in filtering processes.

© 2011 Elsevier Inc. All rights reserved.

Keywords:Porous glassX-ray microtomographyScanning electron microscopyRaman spectroscopy

1. Introduction

Porous ceramicmaterials have extensive applications inmanyimportant industrial processes and research fields, such as:construction of membranes, filters, catalyst supports, in themanufacture of integrated chemical systems and in hetero-geneous multiphase systems involving different componentsarranged for specific functions, processes or reactions [1–11].In particular, porous glasses are often employed in thefabrication of porous supports, particularly for application inoptical sensors and enzyme immobilization [12,13].

Porous glasses are usually prepared by applying pressureand/or heat to a glass source (typically in powdered form)mixed with a foaming agent and subsequent submission toappropriate processes to separate the foaming agent from theglass matrix. Ducheyne and coworkers prepared porousmonoliths employing Na2CO3, NaHCO3, CaCO3 and NH4H2PO4

; fax: +55 19 35213023.(I.O. Mazali).

er Inc. All rights reserved

as foaming agents, and found that the best results wereobtained using CaCO3 with hot-pressing at 100 MPa and 853 K,followed by removal of the foaming agent by acid leaching [14].A simple procedure has been reported for the preparation ofporous ceramics from α-quartz powder (from industrial waste)together with CaCO3 and cellulose by applying pressure andheating at 1573–1773 K [15].

Although porous glasses have generally been producedfrom silica-based sources, there is considerable interest indeveloping processes based on materials other than silica,since, as well as generating final composites with bettercharacteristics (high thermal expansion coefficient, transpar-ency, mechanical and thermal stability), it is possible that theuse of alternative raw materials (lower melting point) mayprovide easier routes to glass production. Phosphate-basedglasses, with additional network forming oxides and one ormore network modifying oxides, have attracted special

.

264 M A T E R I A L S C H A R A C T E R I Z A T I O N 6 2 ( 2 0 1 1 ) 2 6 3 – 2 6 7

attention in recent years since they exhibit increased chemicaldurability and thermodynamic stability, and have applicationin sealing processes, immobilization of nuclear waste and inregenerative medicine. Wang and coworkers described theproduction of a macroporous calcium phosphate glass-ce-ramic employing sucrose as foaming agent with two-steppressing and sintering at 1123 K [16]. A procedure for theformation of porous niobophosphate glasses has been de-scribed that involves impregnation on a polymeric spongefollowed by heat-treatment to remove the organic template[17]. Mazali and coworkers studied a large number of glassesbased on the Li2O–Nb2O5–CaO–P2O5 system and demonstratedthat for all samples tested the devitrification event is precededby a pronounced endothermic effect associated with thesoftening and sintering of the powdered glass [18]. Thesoftening phenomenon of these powdered glasses occursrapidly at relatively low temperatures (between 923 and1048 K) and gives rise to dense vitreous monoliths.

Thermal analysis, X-ray diffraction (XRD) and scanningelectron microscopy (SEM) are the techniques most commonlyapplied to characterize porous glasses. However, X-ray com-putedmicrotomography (micro-CT) has proved to be successfulin resolving the internal microstructure of materials in a non-destructive manner and has already been used to investigateporous glasses [19,20]. The procedure requires no prior prepa-ration of the sample and enables three-dimensional (3-D)modeling and morphological analysis to be carried out withina few hours. Moreover, micro-CT can be used to elucidate theremaining 3-D pore structure of the glass following removal ofthe foaming agent, thusprovidinga fastqualitativeanalysis, theresults of which may be related directly to SEM images [21].

In the present study, porous glassmonoliths were preparedby very fast sintering of powdered niobophosphate glass in thepresence of cellulose as foaming agent. A niobophosphateglass having the composition 6Li2O–18Nb2O5–43CaO–33P2O5

(LNCP glass) was chosen for study by virtue of its betterchemical durability in comparison with other compositions[18] and because it exhibits chemical durability similar tomany silica-based glasses. The resulting porous material wasanalyzed and physically characterized with particular empha-sis on the use of micro-CT to reveal the internal structure ofthe glass monoliths.

2. Experimental Procedure

A niobophosphate glass with composition 6Li2O–18Nb2O5–43CaO–33P2O5 (LNCP glass) was prepared by the methoddescribed elsewhere [18]. LNCP glass monoliths were groundto a powder and sieved in order to yield particles <300 μm.Powdered glasses were mixed with cellulose powder (averageparticle size of 20 μm,Aldrich) in an amount corresponding to 0,30 or 50% (in mass) of the glass, and uniaxially pressed at19.6 MPa into disks (10 mm diameter; 3 mm thick). The cellu-lose/LNCP disks were heated for 1 h at 823 K, at whichtemperature the cellulose thermally decomposed. At the endof this step, the cellulose-free/LNCP monolith kept its originalstructure, although with much larger free volume and higherfragility to handling. The cellulose-free/LNCP monoliths were

removed from the furnace at 823 K and immediately submittedto sintering at 1018 K for 2 to 30min to yield sintered LNCP(sLNCP) monoliths.

The thermal decomposition of cellulose was evaluated bythermogravimetric analysis, using a TA Instruments modelSDT Q600 thermal analyzer, under an atmosphere of syntheticair (flow rate of 20 mL min−1) and at a heating rate of 10 Kmin−1. The non-crystallinity of the monoliths was confirmedby XRD analysis with CuKα radiation on a Shimadzu modelXD3Adiffractometer. The Raman spectrawere recordedwith aRenishaw RamanMicroprobe System 3000 using a He–Ne laser(632.8 nm; 8 mW). SEM was performed on a Jeol JSM 6360–LVmicroscope. The mass density (ρ) of each glass sample wasmeasured at 293 K by Archimedes' method using deionizedwater as the buoyant liquid. The dissolution rate of LNCP glassis 5×10−8 g m−2 min−1 in 1.0 mol L−1 aqueous HCl solution at298 K [18].

A Skyscan Portable Micro-CT 1074 instrument wasemployed to analyze the sLNCP samples. The image datasetwas obtained by rotating each sample through 360° in steps of0.9° to yield 400 images, each obtained at an exposure time of720 ms, a voltage of 40 kV and a current of 1000 μA. Cross-sections were reconstructed with the aid of a cone-beamalgorithm and back projection implemented on Skyscan ConeReconstruction software [22]. Morphological measurementswere calculated using Skycan CT-Analyzer software withappropriate threshold leveling on interpolated regions ofinterest. The 3-D models were obtained from the recon-structed cross-sectional images with the aid of Skyscan Antsoftware. Color and 30% transparency were chosen so that thestructural features could be better observed.

3. Results and Discussion

Fig. 1a clearly shows that complete decomposition of celluloseoccurs in cellulose/LNCP disks heated for 1 h at 823 K, atemperature well below the softening temperature of theLNCP glass (1018 K; Fig. 1b), which was previously determined[18]. Even though the test was not conducted under isothermalconditions, at 800 K there is 100% of mass loss, indicating that1 h is enough time to completely decompose the cellulose.Moreover, following 3 min of treatment at the softeningtemperature, sLNCP monolith presented significant increasein the mechanical resistance to handling, in comparison tocellulose-free/LNCP, indicating that the sintering and densifi-cation procedure was very effective. In the absence ofcellulose, a dense (non-porous) monolith is obtained within5 min whereas, in its presence, the evident coalescence ofparticles eventually demonstrates the sintering process.

SEM analysis revealed that formonoliths without cellulose,the microstructure was quite dense after heating at 1018 K for5 min, whilst after 15 min of treatment there was no appre-ciable residual porosity (Fig. 2a). The effects of increasing thecellulose content to 30% (sLNCP70) and to 50% (sLNCP50), inmass, on the microstructure of the monolith after sintering at1018 K for 15 min are shown in Fig. 2b and c, respectively. Ahigher magnification SEM micrograph of the fracture face of asLNCP50 monolith that had been sintered for 15 min (Fig. 2d)

Fig. 1 – (a) Thermogravimetric analysis curve for celluloseutilized as foaming agent in the preparation of sLNCP samples,and (b) differential thermal analysis curve for LNCP glass.

Fig. 2 – SEM micrographs of fracture faces of (a) sLNCP(without cellulose), (b) sLNCP70 (with 30% in mass ofcellulose) and (c) and (d) sLNCP50 (with 50% cellulose inmassof cellulose) after sintering at 1018 K for 15 min.

265M A T E R I A L S C H A R A C T E R I Z A T I O N 6 2 ( 2 0 1 1 ) 2 6 3 – 2 6 7

clearly shows the formation of a neck between glass particles,which is characteristic of the sintering process. The distinctiveporous structure of the sLNCP50 monolith (produced withglass powder of average particle size<300 μm)wasmaintainedeven after 30 min of sintering. It is worth mentioning that theaverage particle diameter is a key factor in determining thefinal porosity of the sintered monolith.

Since the sintering process was performed at a tempera-ture between the glass transition (Tg=914 K) and crystalliza-tion (Tc=1103 K) temperatures, it is possible that the glassmayhave sustained surface nucleation followed by crystallization[18]. The Raman spectra of LNCP glass and a sLNCP50monoliththat had been sintered for 15 min at 1018 K are presented inFig. 3a and b, respectively. In the spectrum of LNCP, an intenseband centered at 914 cm−1 is distinctive for the Nb–O bond ofNbO6 octahedra in the glass structure. The start of thecrystallization process in LNCP is accompanied by theappearance of low intensity bands at 1047 cm−1, assigned to[νs(PO3

2−)], and at 739 cm−1, assigned to [νs(POP)], both absor-bencies being associated with the separation of the β–Ca2P2O7

phase [23]. Complete crystallization of LNCP is characterizedby the presence of a band at 805 cm−1, unequivocally assignedto niobyl groups (Nb=O) [23]. As can be seen in Fig. 3, theRaman spectra of LNCP and sLNCP50 have the same profileand show an absence of any bands characterizing theoccurrence of crystallization. Therefore, the monolithsobtained using the described sintering processes are indeedvitreous materials. This finding is corroborated by the XRDpatterns (not shownhere) of sLNCP50monoliths that had beensintered for 5, 10 and 15 min, all of which were characterizedby the absence of peaks attributable to crystalline phases.

The mass densities (ρ) and porosities (P) of sLNCP70 andsLNCP50 monoliths, plotted as a function of sintering time at1018 K, are presented in Fig. 4. The estimated density (ρestimated)of LNCP was 3.14 g m−3, as determined by the additive factormethod based on the known composition of the glass, and thisvalue represents the maximum density that can be achievedby sintered monoliths. P was calculated from the equation:

P = 100⋅ 1− ρ= ρestimatedð Þ½ � ð1Þ

Fig. 3 – Raman spectra of (a) LNCP glass and (b) sLNCP50sintered at 1018 K for 15 min.

Fig. 4 – Mass density (filled symbols) and porosity (opensymbols) of sLNCP70 (circle) and sLNCP50 (triangle) asfunctions of sintering time at 1018 K.

Fig. 5 – Micro-CT 3-D models of (a) sLNCP70 and (b) sLNCP50sintered at 1018 K for 5 min. The top image in each set(marked *) represents the most detailed 3-D model available.

266 M A T E R I A L S C H A R A C T E R I Z A T I O N 6 2 ( 2 0 1 1 ) 2 6 3 – 2 6 7

Independent of the amount of cellulose employed, the ρvalues of sLNCP monoliths increased, and the P valuesdecreased, with increasing sintering time, thus confirmingthe densification phenomenon. The role of cellulose asfoaming agent, or porosity controller, can be verified bycomparing ρ and P values of sLNCP70 and sLNCP50 at aparticular sintering time. Thus, after 5 min of heating at1018 K, sLNCP70 (formed with 30% in mass of cellulose) had arelative density of 0.82 and a porosity of 18.8%, while sLNCP50(formed with 50% in mass of cellulose) had equivalent valuesof 0.73 and 27.3%, respectively.

Samples of sLNCP monoliths produced with differentamounts of cellulose and sintered at 1018 K for 5 min wereanalyzed by micro-CT. 3-D measurements of the sinteredmonoliths were obtained using CT-Analyzer software withoptimal threshold values for the binarised images selected bytrial and error. The porosity (PCT) can be estimated from themicro-CT results using the equation:

PCT = 100− Obj:V =TVð Þ ð2Þ

in whichObj.V represents the volume of the object determinedfor regions considered to be solid by the CT-Analyzeralgorithm, and TV is the total volume of the object includingboth solid and non-solid regions. The latter includes anyregion that was not identified as solid by the software duringthe threshold adjustment. The PCT values so obtained were inagreement with the porosities calculated from Eq. (1). Thus,after 5 min of heating at 1018 K, the PCT for sLNCP70 wasdetermined to be 17.9%while for sLNCP50 the PCTwas 27.7%, inagreement with the porosity values shown in Fig. 4, calculatedby Eq. (1). Although such computational measurementscannot be taken for granted, they do provide a reasonableinsight into the properties of the sample. Moreover, valuesthat are closer to those given by established characterizationtechniques can be obtained for individual glass systems byrefining the threshold parameters.

Three-dimensional models derived from micro-CT recon-structed cross-sections of sLNCP monoliths that had been

heated for 5 min at 1018 K are shown in Fig. 5. Carefulobservation of the internal macrostructure of these modelsreveals an increasing porosity as the content of celluloseincreased. Although the monoliths presented pores through-out their entire structure (as shown in Fig. 2b), only macro-pores can be resolved using micro-CT, mainly because ofspatial resolution and algorithm restrictions. Because of theselimitations, the 3-D models showed no porosity at all at theedge regions of the samples. However, this constraint

267M A T E R I A L S C H A R A C T E R I Z A T I O N 6 2 ( 2 0 1 1 ) 2 6 3 – 2 6 7

diminishedwhenmorphological calculationswere consideredas can be verified by the porosity.

The upper images presented in Fig. 5a and b show detailed3-D models of sLNCP70 and sLNCP50, respectively. Thesemodels provided the highest pixel detail available and revealsomesurface topographyand innermacroporositywith smallerpores. The simpler 3-Dmodels that form the remaining imagesof Fig. 5 provideoverviewsof the innermacroporous structureofthe monoliths. The macropore architecture of the sinteredmonoliths is truly remarkable, and this suggests that micro-CTimaging can be used to characterize materials with potential inspecific applications such as, for example, catalysis. Sincemicro-CT imaging is a non-destructivemethod, sLCNP samplescan be analyzed before and after a specific treatment ormodification, thus revealing interesting features that mayhave been incorporated into the glass. Furthermore, becausethe technique is based on X-rays, any modification that altersthe composition of the sample to an appreciable extent can bedetected byX-ray attenuation and can eventually be revealed bycross-section imaging.

4. Conclusion

Macroporous monoliths were prepared by very fast sinteringof powdered niobophosphate glass with the composition6Li2O–18Nb2O5–43CaO–33P2O5, using cellulose as foamingagent. The subsequent sintering and densification processeswere very rapid, such that the monoliths already presentedhigh mechanical resistance after 5 min of heating at 1018 K, atemperature that is low in comparison with that required bysilica-based glass systems. The progression of the porosity ofthe sLNCP monoliths was dependent on sintering time andamount of cellulose used. The non-crystallinity of macropor-ous sLNCP monoliths was confirmed by Raman spectroscopyand XRD. X-ray microtomography was successfully applied inmodeling the macrostructure of the monoliths and incalculating their porosity. This analytical technique repre-sents a useful tool for elucidating glass macrostructures in anon-destructive manner. Further studies are being carried outon the application of the described porous materials assupports for integrated chemical systems and in filteringprocesses.

Acknowledgements

The authors are grateful to FAPESP, CNPq, and CAPES forfinancial assistance and to Prof. C.H. Collins (IQ-UNICAMP,Brazil) for English revision. This is a contribution of NationalInstitute of Science and Technology in Complex FunctionalMaterials (CNPq-MCT/FAPESP).

R E F E R E N C E S

[1] Innocentini MDM, Antunes WL, Baumgartner JB, Seville JPK,Coury JR. Permeability of ceramic membranes to gas flow.Mater Sci Forum 1999;299/300:19–28.

[2] Palmeri J, Blanc P, Larbot A, David P. Theory of pressure-driventransport of neutral solutes and ions in porous ceramicnanofiltration membranes. J Membr Sci 1999;160:141–70.

[3] Sondhi R, Bhave R. Role of backpulsing in foulingminimizationin crossflow filtration with ceramic membranes. J Membr Sci2001;186:41–52.

[4] Anpo M, Wada T, Kubokawa Y. Photochemistry in theadsorbed layer. V. Effects of surface pretreatments upon thephotolysis of adsorbed 2-pentanone. Bull Chem Soc Jpn1975;48:2663–6.

[5] Anpo M, Aikawa N, Kubokawa Y, Che M, Louis C, Giamello E.Photoluminescence and photocatalytic activity of highlydispersed titanium oxide anchored onto porous Vycor glass. JPhys Chem 1985;89:5017–21.

[6] Yamashita H, Ichihashi Y, HaradaM, Stewart G, Fox MA, AnpoM. Photocatalytic degradation of 1-octanol on anchoredtitanium oxide and on TiO2 powder catalysts. J Catal 1996;158:97–101.

[7] Nakagaki S, Ramos AR, Benedito FL, Peralta-Zamora PG,Zarbin AJG. Immobilization of iron porphyrins into porousVycor glass: characterization and study of catalytic activity. JMol Catal A 2002;185:203–10.

[8] Hultman HM, Lang M, Arends IWCE, Hanefeld U, Sheldon RA,Maschmeyer T. Chiral catalysts confined in porous hosts: 2.Catalysis. J Catal 2003;217:275–83.

[9] Kawachi EY, Bertran CA, Reis RR, Alves OL. Bioceramics:tendencies and perspectives of an interdisciplinary area.Quim Nova 2000;23:518–22.

[10] Melde BJ, Stein A. Periodic macroporoushydroxyapatite-containing calcium phosphates. Chem Mater2002;14:3326–31.

[11] Bard AJ. Integrated chemical systems – a chemical approachto nanotechnology. New York: Wiley; 1994.

[12] Sotomayor PT, Raimundo Jr IR, Zarbin AJG, Rohwedder JJR, NetoGO, Alves OL. Construction and evaluation of an optical pHsensor based on polyaniline–porous Vycor glass nanocomposite.Sens Actuators B 2001;74:157–62.

[13] Hosono H, Abe Y. Porous glass-ceramics with skeleton oftwo-dimensional layered crystal Ti(HPO4)2.2H2O. J NonCrystSolids 1992;139:86–9.

[14] Ducheyne P, El-Ghannam A, Shapiro I, US5676720-A 1997.[15] Furuta S, Nakao H, Katsuki H. Preparation of porous ceramics

from industrial waste silica mineral. J Mater Sci Lett 1993;12:286–7.

[16] Wang C, Kasuga T, Nogami M. Macroporous calciumphosphate glass-ceramic prepared by two-step pressingtechnique and using sucrose as a pore former. J Mater SciMater Med 2005;16:739–44.

[17] Rambo CR, Ghussn L, Sene FF, Martinelli JR. Manufacturing ofporous niobium phosphate glasses. J NonCryst Solids2006;352:3739–43.

[18] Mazali IO, BarbosaLC,AlvesOL. Preparationandcharacterizationof new niobophosphate glasses in the Li2O–Nb2O5–CaO–P2O5system. J Mater Sci 2004;39:1987–95.

[19] Maire E, Buffière YJ, Salvo L, Blandin JJ, Ludwig W, Létang JM.On the application of X-ray microtomography in the field ofmaterials science. Adv Eng Mater 2001;3:539–46.

[20] Luyten J, Mullens S, Cooymans J, De Wilde AM, Thijs I, KempsR. Different methods to synthesize ceramic foams. J EurCeram Soc 2009;29:829–32.

[21] Kerckhofs G, Schrooten J, van Cleynenbreugel T, Lomov V,WeversM.ValidationofX-raymicrofocuscomputed tomographyas an imaging tool for porous structures. Rev Sci Instrum2008;79:013711–9.

[22] Feldkamp LA, Davis LC, Kress JW. Practical cone-beamalgorithm. J Opt Soc Am 1984;A1:612–9.

[23] Mazali IO, Alves OL. Porous glass-ceramic of α-NbPO5 skeletonwith three-dimensional network structure. J Mater Sci Lett2001;20:2113–7.