SP

MJa

b

a

ARRA

KPISACN

1

cBpcmpootltcoous

rc

0d

J. of Supercritical Fluids 49 (2009) 369–376

Contents lists available at ScienceDirect

The Journal of Supercritical Fluids

journa l homepage: www.e lsev ier .com/ locate /supf lu

upercritical CO2 as a reaction and impregnation medium in the synthesis ofd–SiO2 aerogel inverse opals

aría José Tenorioa, María José Torralvob, Eduardo Encisoa, Concepción Pandoa,uan Antonio R. Renuncioa, Albertina Cabanasa,∗

Departamento de Química-Física I, Universidad Complutense de Madrid, Ciudad Universitaria s/n, 28040 Madrid, SpainDepartamento de Química Inorgánica, Universidad Complutense de Madrid, Ciudad Universitaria s/n, 28040 Madrid, Spain

r t i c l e i n f o

rticle history:eceived 9 October 2008eceived in revised form 17 March 2009ccepted 19 March 2009

a b s t r a c t

Pd–SiO2 aerogel inverse opals were prepared for the first time using supercritical CO2 (scCO2). Colloidalcrystals formed by 3D-arrays of monodisperse spherical polymer particles (opals) were used as tem-plates. In one approach, a pre-made large surface area SiO2 aerogel inverse opal prepared in scCO2 wasimpregnated with palladium hexafluoroacetylacetonate [Pd(hfac)2] in scCO2 and thermally treated to get

eywords:alladiumnverse opalsupercritical fluidserogelsatalyst

Pd–SiO2. In another approach, tetraethylorthosilicate and Pd(hfac)2 dissolved in scCO2 reacted directlyon the polymeric template in one step. Both, impregnation and reaction experiments were carried outat 40 ◦C and 85 bar. After the removal of template, large surface area porous materials replicating thestructure of the original template were obtained (inverse opals). These materials presented hierarchicalporosity with ordered macropores of mesoporous walls. Palladium (between 1 and 3 mol% by EDX) wasincorporated uniformly throughout the SiO2 matrix forming small clusters (by TEM). The effect of Pd

gel p

anomaterials incorporation on the aero

. Introduction

Opals are formed by monodisperse spherical colloidal parti-les assembled in three-dimensional ordered arrays (3D-arrays).ecause of the high contrast between the refractive index of thearticles and the voids of the crystal, these materials exhibit opales-ence and resemble the natural opals. Synthetic opals formed byonodisperse SiO2 or polymer particles (latex) can be used as tem-

lates to infiltrate precursors, which after reaction and eliminationf the template yield the inverse replica (inverse opal). Inversepals exhibit a number of interesting properties that derive fromheir special structural features. If the skeletal structure is on theength scale of optical wavelengths the inverse opal exhibit pho-onic properties, which are important for the design of photonicrystals, pigments and sensors, among others. Furthermore, inversepals are highly structured porous systems which allow transportf species through the pores as well as in the solid phase and aresed in fuel cells, microreactors, electrochemical cells, catalysis and

eparation [1,2].

Recently, supercritical fluids have received much attention aseaction and processing medium in material synthesis [3]. Super-ritical CO2 (scCO2) is by far the most frequently used fluid because

∗ Corresponding author. Tel.: +34 91 3945225; fax: +34 91 3944135.E-mail address: a.cabanas@quim.ucm.es (A. Cabanas).

896-8446/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.supflu.2009.03.011

orosity was studied.© 2009 Elsevier B.V. All rights reserved.

it is cheap, non-toxic and non-flammable and has relatively lowcritical temperature and pressure (Tc = 31 ◦C, Pc = 73.8 bar) [4]. Fur-thermore many ceramic precursors such as metal alkoxides dissolveat moderate pressure and temperature in scCO2 [5]. The lowviscosity, high diffusivity relative to liquids and very low sur-face tension of scCO2 promote infiltration in complex geometriesand mitigate mass transfer limitations common to liquid-phaseprocesses. CO2 is a gas at ambient pressure and is eliminatedcompletely upon depressurization. All these advantages have beenexploited in a new method to produce SiO2 aerogel inverse opalsin scCO2 developed at our laboratory [6–8]. The method involvesthe reaction of silicon alkoxides dissolved in scCO2 on 3D-latexarray templates. In this paper we show that the method canbe extended to produce PdO–SiO2 and Pd–SiO2 aerogel inverseopals.

Supported Pd catalysts are used in reduction and oxidation reac-tions as well as in hydrogenation, dehydrogenation, debenzylation,hydrocracking, carbonylation and other carbon–carbon couplingreactions [9]. In order to improve the catalytic activity and selec-tivity of the catalyst, the metal must be evenly dispersed on a largesurface area support such as a mesoporous SiO2 aerogel. The pres-

ence of macropores in the catalyst facilitates the transport of thereactants to the catalyst surface. On the other hand, structured cat-alysts of defined geometry and pore size distribution are necessaryto model catalytic processes. Preparing Pd–SiO2 inverse opals oflarge surface area may well serve to these purposes.

3 rcritica

olaisaaptam0Mouaw8aPcrtlPmtT

otn(dct3sho

obcwmiuo

70 M.J. Tenorio et al. / J. of Supe

Supercritical fluids have been previously used to deposit metalsn different supports. The large solubility of many organometal-ic compounds in scCO2 [10,11] has allowed their impregnationnd/or reaction onto different supports of complex geometry andnto polymeric materials. Blackburn et al. and Fernandes et al. havehown that Pd films can be deposited on Si wafers, polyimidend porous alumina by H2 reduction of �-allylpalladium hex-fluoroacetylacetonate [(�-CH3)Pd(hfac)], allyl(cyclopentadienyl)-alladium(II) [CpPd(�-C3H5)] and palladium hexafluoroacetylace-onate [Pd(hfac)2] dissolved in scCO2 at temperatures between 40nd 80 ◦C and pressures between 100 and 140 bar [12,13]. Confor-al filling of Pd onto silicon test wafers with features as small as

.1 �m × 1 �m was achieved. This process is of great interest inicroelectronics [14,15] and has been also used in the preparation

f membranes for H2 separation [13]. The same technique has beensed by Ye et al. to deposit Pd nanocrystals on carbon nanotubesnd SiO2 nanowires for catalytic applications [16–18]. Palladiumas deposited by H2 assisted reduction of Pd(hfac)2 in scCO2 at0–150 ◦C and P > 300 bar. Morley et al. [19] have prepared Pd–SiO2erogel composites by the impregnation of a silica aerogel withd(hfac)2 at 40 ◦C and 275 bar. The reduction of the organometallicompound in this case was carried out after depressurization of theeactor with H2 at 40 ◦C and 69 bar. scCO2 extraction was necessaryo remove the ligands in the precursor from the substrate. Simi-arly, Jiang et al. [20] have impregnated nafion membranes withd(hfac)2 in scCO2 at 60 ◦C and 136 bar. After depressurization theembranes were treated in H2 at 80 ◦C and 110 bar. Pd content in

he sample decreased with increasing distance from the surface.he Pd-nafion membranes can be used in direct methanol fuel cells.

Other authors have used liquid solvents to impregnaterganometallic compounds into mesoporous materials, which werehen treated in scCO2. For example, Lee et al. [21,22] have impreg-ated mesoporous SBA-15 with Pd(hfac)2 using tetrahydrofuraneTHF) or ethanol. The solvent was removed under vacuum and theried materials were treated in scCO2 at 60 ◦C and 100 bar. Theomposite material was calcined in air at 400 ◦C to decomposehe organometallic compound and reduced in a H2/Ar mixture at00 ◦C. The best dispersion was obtained using THF followed bycCO2 treatment. The favourable transport properties of scCO2 canighly improve the dispersion of Pd nanoparticles in the mesoporesf SBA-15 without any surface functionalization.

In this paper, we present different approaches to the synthesisf Pd–SiO2 inverse opals using scCO2. Unlike the examples shownefore, these materials are highly structured and present hierar-hical porosity with ordered macropores of mesoporous walls, as

ell as a regular composition. Initially, we have impregnated pre-ade large surface area SiO2 aerogel inverse opals with Pd(hfac)2

n scCO2 and thermally decomposed the precursor (Scheme 1). These of supercritical fluids and in particular scCO2 in the synthesisf metal-composite materials favours the metal dispersion on the

Scheme 1. Synthesis of SiO2 and Pd–SiO2 inv

l Fluids 49 (2009) 369–376

support, especially within tortuous geometries such as those foundin aerogel inverse opals. In order to reduce the number of stepsin the synthesis, the one-step synthesis of Pd–SiO2 aerogel inverseopals in scCO2 was also attempted (Scheme 2). In this case scCO2acted at the same time as reaction and impregnation medium forthe Si and Pd precursors, respectively. This is the first time that thesematerials have been produced in scCO2.

2. Materials and methods

2.1. Materials

For the synthesis of the polymeric templates, reagent gradematerials from Sigma–Aldrich and distilled water were used.Tetraethylorthosilicate (TEOS, 99+% pure), tetramethylorthosilicate(TMOS, 99+% pure), benzosulfonic acid (BSA, 99+% pure) and palla-dium(II) hexafluoroacetylacetonate (Pd(hfac)2) were obtained fromSigma–Aldrich. All chemicals were used as received. CO2 purity>99.99% and forming gas (5% H2 + 95% N2) purity >99.99% weresupplied by Air Liquide.

2.2. Template preparation

Polystyrene (PS) particles modified with methacrylic acid (MA)and itaconic acid (IA) groups on their surface were prepared bysurfactant-free emulsion copolymerisation in water [23]. The molarratio of styrene (S) to MA and IA in PS–MA–IA samples, PS:MA:IA,was 50:1:2 yielding monodisperse spherical particles of diam-eter 350 nm. The polymeric particles were impregnated with asmall amount of an acid catalyst (BSA) in the aqueous phase(0.01–0.001 mM acid/g latex) and the colloidal crystal was formedby centrifugation or evaporation and further dried under air at roomtemperature and at 45 ◦C in an oven. The monodisperse spheri-cal polymer particles crystallize in a face centered cubic (fcc) typeof packing. These colloidal crystals were used as templates in thesynthesis of SiO2 and Pd–SiO2 aerogel inverse opals.

2.3. Synthesis of SiO2 aerogel inverse opals

SiO2 aerogel inverse opals were synthesized in scCO2 from thereaction of TEOS or TMOS dissolved in scCO2 at 40 ◦C and 85 bar on3D-latex array templates following the procedure outlined in thefirst two steps of Scheme 1 and described elsewhere [6–8]. Afterdepressurization, the template was removed by heat treatment inair yielding highly porous materials (aerogels) replicating the struc-

ture of the original template (inverse opals). These materials wereused as supports in the synthesis of Pd–SiO2 aerogel inverse opals.

The major problem encountered to the synthesis of inverse opalsin scCO2 using 3D-latex array templates is the high solubility of CO2in polymers [24] which decreases the glass transition temperature

erse opals in scCO2 using approach (1).

M.J. Tenorio et al. / J. of Supercritical Fluids 49 (2009) 369–376 371

SiO2 i

opiw

2

ar

gawaopoao

srvgcwdlcTatm(vtdCrpvw(

irawf

Scheme 2. Synthesis of PdO–SiO2 and Pd–

f the polymer and can affect the mechanical stability of the tem-late due to the aggregation of the polymer particles. By introducing

n the polystyrene template different moieties, particle aggregationas suppressed at the reaction conditions [25].

.4. Synthesis of Pd–SiO2 aerogel inverse opals

In the synthesis of Pd–SiO2 aerogel inverse opals, two differentpproaches are compared. The experimental procedure is summa-ized in Schemes 1 and 2.

In approach (1) shown in Scheme 1, pre-made SiO2 aero-el inverse opals were impregnated in scCO2 with Pd(hfac)2. Inpproach (2) shown in Scheme 2, a mixture of TEOS and Pd(hfac)2as reacted on the polymeric template in scCO2. In both cases

fter reaction the material was thermally treated to decompose therganometalic compound and, in approach (2), to remove the tem-late. All the experiments (impregnation and reaction) were carriedut in scCO2 at 40 ◦C and 85 bar. These conditions of temperaturend pressure were chosen to assure solubility of reactants, reactivityf the precursors and stability of the template.

The experiments were conducted in a ca. 70 mL custom-madetainless-steel high-pressure reactor in the batch mode [8]. Theeactor was loaded with the template and reactants in differentials. In approach (1) two vials containing the pre-made SiO2 aero-el inverse opal and Pd(hfac)2 were used. In approach (2) four vialsontaining the 3D-latex array template, TEOS, Pd(hfac)2 and H2Oere used. Pd(hfac)2 can be previously dissolved in TEOS and intro-uced as a solution in the same vial. The amount of TEOS was in

arge excess for the reaction conditions. In approach (2) the con-entration of Pd(hfac)2 relative to TEOS was between 2 and 11 mol%.he reactor was then placed into a thermostatic bath (PolyScience)nd was filled with CO2 using a high-pressure syringe pump athe reaction temperature (Isco, Inc. Model 260D). The pressure was

easured using a pressure transducer (Druck Ltd.). A safety valveSwagelok) was fitted to the reactor. Solubility of Pd(hfac)2 was pre-iously studied using a high-pressure variable volume view cell andhe procedure described elsewhere [26]. At the experimental con-itions, 40 ◦C and 85 bar, solubility of Pd(hfac)2 is above 1 wt.% inO2. Impregnation of SiO2 was carried out for 18 h. Simultaneouseactions and impregnations of TEOS and Pd(hfac)2 on the tem-lates were performed for 3 h. Depressurization through a needlealve was carried out in 0.5–2 h. The weight change after reactionas determined by weight difference using an analytical balance

AND GR-200).In approach (2), the template was removed by thermal treatment

n air or N2 at 500 ◦C yielding PdO–SiO2 or Pd/C–SiO2 inverse opals,espectively. To avoid the presence of residual carbon in the samplesnd, for comparison purposes, in most cases the samples obtainedere heated in air at 500 ◦C for 3 h. Temperature was increased

rom room temperature to 300 ◦C at 1 ◦C/min and from 300 ◦C to

nverse opals in scCO2 using approach (2).

500 at 0.5 ◦C/min. The weight percentage remaining after templateremoval was also determined by weight difference.

After heat treatment in air, PdO–SiO2 brown composite mate-rials were obtained. PdO is not the active form in most catalyticapplications, so a further reduction step was required. Sampleswere reduced in forming gas (5% H2 + 95% N2) at 400 ◦C for 18 h. Inapproach (1), we also studied the direct reduction of the Pd(hfac)2impregnated SiO2 aerogel with a H2/N2 mixture at 300 and 400 ◦Cfor 18 h. Heating rate from room temperature was 1 ◦C/min.

2.5. Materials characterization

Materials were characterized using scanning electronmicroscopy (SEM), transmission electron microscopy (TEM),N2 adsorption, X-ray diffraction (XRD) and Thermo GravimetricAnalysis (TGA). SEM images were taken on a JEOL-6400 electronmicroscope working at 20 kV. Prior to analysis, samples weregold coated or evaporated with graphite when elemental analysiswas conducted. TEM were carried out on a JEOL 2000FX electronmicroscope operating at 200 kV equipped with a double tilting(±45◦) and a JEOL-JEM 3000F electron microscope operatingat 300 kV equipped with a double tilting (±25◦). Samples weredispersed in water or 1-butanol over copper grids and driedin air. Energy-dispersive detection X-ray analysis (EDX) wasconduced on selected samples using both SEM and TEM. N2adsorption–desorption isotherms at 77 K were obtained using anASAP-2020 equipment from Micrometrics. Prior to adsorption mea-surements, samples were out-gassed at 110 ◦C for 3 h. Isothermswere analysed using standard procedures. The BET equation wasused for specific surface calculations and the external surfacearea and the micropore volume were determined using the t-plotmethod [27]. The pore size distributions were calculated using theBarrett, Joyner and Halenda (BJH) method for a cylindrical poremodel [28]. XRD patterns were collected using a Siemens D5000diffractometer with Cu K-� radiation. TGA were obtained on aPerkin-Elmer Pyris 1 at a heating rate of 10 ◦C/min in O2 or N2 flow(20 mL/min).

3. Results and discussion

Table 1 summarizes the experiments performed following theapproaches (1) and (2) and the previous experiments carried outto obtain the supports in approach (1) together with some of theproperties of these materials.

3.1. Synthesis of SiO2 aerogel inverse opals

SiO2 aerogel inverse opals (samples 1–3) were synthesized fromthe reaction of TMOS or TEOS on colloidal crystals formed byPS–MA–IA (350 nm) latex particles impregnated with different con-

372 M.J. Tenorio et al. / J. of Supercritical Fluids 49 (2009) 369–376

Table 1Summary of SiO2 and Pd–SiO2 aerogel inverse opal experiments showing experimental conditions, phase composition, % Pd mol determined by EDX in SEM images and BETsurface area (SBET).

Samplenumber

Template/support Precursor Approach Thermal treatment Composition % Pd mol by EDX (SEM) SBET m2/g

1 PS–MA–IA TMOS Air 500 ◦C SiO2 – 5180.004 mM BSA/g latex

2 PS–MA–IA TEOS Air 500 ◦C SiO2 – 4170.008 mM BSA/g latex

3 PS–MA–IA TEOS Air 500 ◦C SiO2 – 2440.014 mM BSA/g latex

4 SiO2 sample 1 0.13 wt.% Pd(hfac)2 1 Air 500 ◦C PdO–SiO2 <1 mol% PdH2/N2 300 ◦C Pd–SiO2 0.6–1.2 mol% Pd

5 SiO2 sample 2 0.15 wt.% Pd(hfac)2 1 Air 500 ◦C PdO–SiO2 0.5–1.3 mol% 340Air 500 ◦C + H2/N2 400 ◦C Pd–SiO2

6 PS–MA–IA 10 mol% Pd(hfac)2 + TEOS 2 Air 500 ◦C PdO–SiO2 1.8–3.2 mol% 4470.014 mM BSA/g latex separated vials Air 500 ◦C + H2/N2 400 ◦C Pd–SiO2

7 Air ◦

Air8 Air

Air

cishr

smAhimdmttm

gtp[d

Fuptd

PS–MA–IA 2.3 mol% Pd(hfac)2 + TEOS 20.014 mM BSA/g latex solutionPS–MA–IA 11 mol% Pd(hfac)2 + TEOS 20.014 mM BSA/g latex solution

entrations of BSA in the aqueous phase [7,8]. In contrast to thenfusion in the liquid phase, when the process was carried out incCO2 the particles were only coated with the ceramic and the tetra-edral and octahedral voids in the original fcc packing of spheresemained empty in the inverse opal.

N2 adsorption–desorption isotherms of these materials pre-ented type IV hysteresis loops indicative of mesoporousaterials with BET surfaces area values from 518 to 244 m2/g.

dsorption–desorption isotherms of sample 2 in Fig. 1 show aysteresis loop due to the presence of mesopores. The adsorption

sotherms were studied using the BJH method for a cylindrical poreodel [28]. Analysis of the adsorption branch gave a monotone

ecreasing pore size distribution with a broad and not well definedaximum in the mesoporous range (not shown here). Analysis of

he desorption branch was more informative and pore size distribu-ions with defined maxima in the mesopore range were obtained,

aking these materials ideal to use as catalytic supports.There are many factors influencing the porosity of the SiO2 aero-

els produced in scCO2. Among them, we have identified as crucialhe precursor (TEOS or TMOS), its concentration in the supercriticalhase, the reaction time and the acid concentration of the template8]. A comparison of the adsorption isotherms and the pore sizeistributions of two SiO2 aerogel inverse opals synthesized under

ig. 1. Adsorption–desorption isotherms of aerogel inverse opals produced in scCO2

sing a PS–MA–IA 3D-latex array template following approach (1): (©) SiO2 (sam-le 2); (�) Pd–SiO2 obtained by the impregnation of sample 2 with Pd(hfac)2 andhermal treatment (sample 5). Inset shows pore size distributions obtained from theesorption branch of the isotherm using the BJH method.

500 C PdO–SiO2 Below detection limit 206500 ◦C + H2/N2 400 ◦C Pd–SiO2

500 ◦C PdO–SiO2 0.5–1.3 mol% 376500 ◦C + H2/N2 400 ◦C Pd–SiO2

the same experimental conditions using the same PS–MA–IA tem-plate but two different acid concentrations, 0.008 and 0.014 mMBSA/g latex (samples 2 and 3), shows that the aerogel obtainedusing the template with the lower acid concentration presents amuch higher surface area, 417 m2/g versus 244 m2/g. However, thepore size distributions are very similar and the maxima in the dis-tribution obtained from the desorption branch of the isotherms are13 and 15 nm for the 0.008 and 0.014 mM BSA/g latex, respectively(Figs. 1 and 8); the distribution is slightly narrower for the materialsobtained with the lower acid concentration on the template.

3.2. Impregnation of Pd(hfac)2 on SiO2 aerogel inverse opals

In approach (1), pre-made SiO2 aerogel inverse opals wereimpregnated with Pd(hfac)2 in scCO2 following Scheme 1. Sample4 was obtained by the impregnation of sample 1 with a 0.13 wt.%Pd(hfac)2 solution in CO2 at 40 ◦C and 85 bar. After slow depres-surization of the reactor, the white SiO2 aerogel support turnedbrownish black. Weight gain after impregnation was close to 21%.After heating in air at 500 ◦C a PdO–SiO2 composite material was

obtained. Fig. 2 displays the XRD of sample 4 heated in air showinga broad band at 2� ca. 20◦ due to amorphous SiO2 and a very weakreflection at 2� ca. 34◦ which indicates the presence of PdO [JPDS00-043-1024].

Fig. 2. XRD of PdO–SiO2 and Pd–SiO2 inverse opals obtained in scCO2 at 40 ◦C and85 bar and different heat treatments: (a and b) sample 4 obtained by the impregna-tion of a pre-made SiO2 aerogel inverse opal with Pd(hfac)2 following approach (1);(c and d) sample 6 obtained by simultaneous reaction and impregnation of TEOSand Pd(hfac)2 into a PS–MA–IA 3D-latex array template following approach (2).

M.J. Tenorio et al. / J. of Supercritica

Fma

FosF[(wtma

ilnyspd

FS

the SiO2 aerogels was further studied. Fig. 1 compares the N2-adsorption isotherms of a SiO2 aerogel inverse opal obtained fromthe reaction of TEOS on a PS–MA–IA (0.008 mM BSA/g latex) tem-plate (sample 2) and the material obtained by impregnation of

ig. 3. TEM of a PdO–SiO2 inverse opal obtained by the impregnation of a pre-ade SiO2 aerogel inverse opal with Pd(hfac)2 in scCO2 at 40 ◦C and 85 bar following

pproach (1) and heating in air at 500 ◦C (sample 4).

TEM images of the PdO–SiO2 composite material are shown inig. 3. The material is composed of SiO2 hollow spheres (macrop-res) arranged in a fcc packing. TEM images show that the orderedtructure of the inverse opal is not altered upon impregnation.ig. 3a and b shows the projections of the inverse opal along the1 1 1] direction in the fcc packing. Octahedral (Oh) and tetrahedralTh) holes in the structure are clearly observed. The darker spotshich appear in these images are PdO clusters formed following

he impregnation and heat treatment. Fig. 3c and d shows higheragnification images of this sample, where PdO clusters of ca. 2

nd 10 nm can be clearly observed.EDX analysis of this sample was conduced in both SEM and TEM

mages and revealed a loading below 1 mol% Pd on SiO2, which isower than expected considering the amount of Pd(hfac)2 impreg-ated on the support (21% weight gain upon impregnation would

ield ca. 2 mol% Pd on SiO2). The lower incorporation of Pd into theample can be explained considering the thermal behaviour of therecursor. TGA of Pd(hfac)2 and the impregnated samples were con-ucted in N2 and O2 flow and are shown in Fig. 4. Similar results are

ig. 4. TGA of Pd(hfac)2 and sample 4 (obtained by the impregnation of a pre-madeiO2 aerogel inverse opal with Pd(hfac)2 in scCO2 at 40 ◦C and 85 bar) in N2 and O2.

l Fluids 49 (2009) 369–376 373

obtained in N2 and O2 atmosphere. Pure Pd(hfac)2 sublimes at ca.110 ◦C. TGA of the impregnated samples presents a first drop at ca.70 ◦C ascribed to adsorbed H2O and at least two other drops, one atca. 110 ◦C and the other between 235 and 250 ◦C due to the sublima-tion and decomposition of Pd(hfac)2, respectively. The weight lossesindicate that half of the Pd(hfac)2 impregnated on the SiO2 aerogelsublimes between 110 and 165 ◦C and the other half decomposesbetween 235 and 250 ◦C. The partial sublimation of Pd(hfac)2 in N2and O2 explains the lower Pd loading obtained after calcination.

A portion of sample 4 was subjected to a different heat treatmentin a reducing H2/N2 atmosphere at 300 ◦C, rendering a dark/blackpowder. XRD of the reduced sample (Fig. 2) showed peaks at 2� val-ues ca. 40◦ and 46.7◦, ascribed to synthetic Pd [JPDS 46-1073]. Minorpeaks between 32◦ and 38◦ were also observed but not identifiedsuggesting the presence of impurities when the sample is reducedat these conditions. The same was also observed when the reduc-tion was performed at 400 ◦C. Weight loss of the sample during theheat treatment seems to indicate that sublimation of Pd(hfac)2 inthe H2/N2 mixture during heating is not as important as in N2 andO2. The problem of performing the reduction directly in H2/N2 isthe possible carbon contamination of the sample. EDX analysis inSEM images gave a Pd loading between 0.6 and 1.2 mol% in SiO2.TEM images of this sample are shown in Fig. 5.

Fig. 5a and b shows the projection along the [1 1 1] directionof the fcc packing, whilst Fig. 5c and d shows the projection alongthe [1 1 0] direction. All the images show darker spots uniformlydistributed through the sample, ascribed to Pd clusters of varyingsizes from 5 to 20 nm. The largest particles are a little larger thanthose observed in Fig. 3, which may be related to the slightly higherPd loading of the sample reduced in H2/N2.

To avoid film formation and promote metal dispersion, thesimultaneous impregnation and reduction of Pd(hfac)2 on SiO2 ina supercritical CO2/H2 mixture was not attempted.

The effect that the incorporation of Pd has on the porosity of

Fig. 5. TEM of a Pd–SiO2 inverse opal obtained by the impregnation of a pre-madeSiO2 aerogel inverse opal with Pd(hfac)2 in scCO2 at 40 ◦C and 85 bar followingapproach (1) and heating in H2/N2 at 300 ◦C (sample 4).

3 rcritical Fluids 49 (2009) 369–376

taAwhas

atawPrmo

uarameTsc

3P

trtttrfwiiwaass

mtror(olsthpvAa

P5t

74 M.J. Tenorio et al. / J. of Supe

his aerogel with a solution 0.15 wt.% Pd(hfac)2 in CO2 at 40 ◦Cnd 85 bar during 18 h, followed by the heat treatment (sample 5).lthough the weight gain of the SiO2 aerogel upon impregnationith Pd(hfac)2 was close to 22%, the Pd content in SiO2 after theeat treatment determined by EDX (SEM) was only between 0.5nd 1.3 mol% due to the partial sublimation of the precursor. Poreize distributions of both samples are shown in the inset of Fig. 1.

BET surface areas of the SiO2 and Pd–SiO2 aerogels (samples 2nd 5) are 417 and 340 m2/g, respectively and the pore size distribu-ions of the adsorption and desorption branches of the isothermsre almost coincident. There is a slight decrease in the BET areahen the pre-made SiO2 aerogel inverse opal is impregnated with

d(hfac)2 in scCO2 at 40 ◦C and 85 bar, heated in air at 500 ◦C andeduced in H2/N2 at 400 ◦C. This reduction in the surface area of theaterial is not surprising as non-porous Pd clusters are depositing

n the surface of the porous SiO2 aerogel.After impregnation and heat treatment, the small micropore vol-

me of the SiO2 aerogel disappears completely. The surface areascribed to the micropore volume is, however, smaller than the areaeduction upon impregnation indicating that some Pd clusters arelso deposited in the mesopores, particularly in those below theaximum in the pore size distribution as shown in Fig. 1. This is

xpected as Pd clusters between 2 and 10 nm were observed in theEM images of a similar sample (Fig. 5). The differences in the poreize distribution of both samples are very small due to the smalloncentration of Pd in the samples.

.3. Simultaneous reaction of TEOS and impregnation ofd(hfac)2 on 3D-latex array templates

In an effort to reduce the number of steps involved in the syn-hesis of these materials we have also attempted the simultaneouseaction and impregnation of TEOS and Pd(hfac)2 on the polymericemplates as shown in Scheme 2. PS–MA–IA (350 nm) latex par-icles were impregnated with different concentrations of BSA inhe aqueous phase and arranged in 3D-arrays. TEOS and Pd(hfac)2eacted on these templates in humidified scCO2 at 40 ◦C and 85 baror ca. 3 h (samples 6–8). Upon reaction, the materials turned fromhite to light blue or brownish blue. After the removal of template

n air at 500 ◦C, the materials became brown, which indicates thencorporation of PdO into the inverse opal (SiO2 inverse opals are

hite). XRD analysis of sample 6 (Fig. 2) showed a very broad bandt 2� ca. 20◦ due to amorphous SiO2 and peaks at 2� values ca. 34◦

nd 42◦ ascribed to PdO [JPDS 00-043-1024]. SEM images of theseamples showed 3D-arrays of hollow SiO2 spheres reproducing thetructure of the original 3D-array of latex spheres (not shown here).

As previously mentioned from a catalytic point of view, theaterial of interest is Pd–SiO2, so the material obtained by heat

reatment in air was further reduced in H2/N2 at 400 ◦C. XRD of theeduced sample is also shown in Fig. 2. Fig. 6 shows TEM imagesf the PdO–SiO2 (Fig. 6a) and Pd–SiO2 (Fig. 6b–d) composite mate-ials obtained using a PS–MA–IA template (0.014 mM BSA/g latex)sample 6). Images show that the material replicates the structuref the template. It is interesting to note that the diameter of the hol-ow spheres gets ca. 18% smaller than the original polymer particleize according to TEM due to the network shrinkage upon the reac-ion and the removal of template. For pure SiO2, the shrinkage isowever much smaller than when the infusion and reaction of therecursor are carried out in the liquid phase [7]. The darker spotsisible in the TEM images correspond either to PdO or Pd clusters.s in sample 4 obtained using approach (1) (Figs. 3 and 5), small

nd large clusters are observed in the TEM images.

Fig. 6a shows the projection along the [1 0 0] direction of thedO–SiO2 material obtained after heating in air. In this image,–10 nm PdO clusters are clearly observed. Fig. 6b and c correspondso the reduced Pd–SiO2 material and shows projections along the

Fig. 6. TEM of an inverse opal obtained by simultaneous reaction and impregnationof TEOS and Pd(hfac)2 in scCO2 into a PS–MA–IA 3D-latex array template at 40 ◦C and85 bar following approach (2) and heating in air at 500 ◦C (sample 6): (a) PdO–SiO2;(b–d) Pd–SiO2 obtained by the reduction of PdO–SiO2 in H2/N2 at 400 ◦C.

[1 1 0] and [1 1 1] directions of the fcc packing, respectively. SmallPd clusters are homogeneously distributed throughout the sample.Fig. 6d shows a higher magnification image of this sample, with Pdclusters of ca. 2 and 10 nm. There are no major differences betweenthe images before and after reduction. Because of the low temper-atures employed in the reduction and the small concentration ofPd in the samples, there is no sintering of the particles and nicelydispersed Pd–SiO2 composites are obtained.

In sample 6, the initial Pd(hfac)2 content relative to TEOS wasclose to 10 mol%, whilst the average Pd content estimated by EDXin the SEM images of the calcined material was between 1.8 and3.2 mol%. EDX analysis was also performed in the TEM images ofboth Pd–SiO2 and PdO–SiO2 and showed Pd loadings between 1 and1.5 mol% in SiO2. The differences found between the EDX analysisperformed in the SEM and TEM images suggest larger loadings onthe surface of the materials (SEM) in comparison to the bulk (TEM).At the same conditions, Pd(hfac)2 does not react on these templatesin pure scCO2 or in a supercritical ethanol/CO2 mixture. The smallPd content in the inverse opal relative to the initial concentrationof Pd in the reaction mixture seems to indicate that Pd(hfac)2 getsonly incorporated into the SiO2 network as it is being formed. Thepartial sublimation of Pd(hfac)2 during heating, already shown forthe materials produced following approach (1), would explain thelower incorporation of Pd in the sample.

We have also investigated different ways of eliminating the poly-meric template and conducted TGA of the as reacted sample indifferent media. Elimination of most of the template in N2 and O2atmospheres took place at temperatures of 405 and 372 ◦C, respec-tively. In N2, there was a slight weight loss up to 700 ◦C indicatingthe formation of some carbonaceous residue during the templatedecomposition at 500 ◦C. The template can be also removed by dis-solution in THF. In this solvent, Pd(hfac)2 is reduced to metallicpalladium yielding a grey powder, but there is some leaching of

the metal in the solution, and, as a result, the Pd loading in thissample is lower than that obtained using any thermal treatment.

Very similar results are obtained if the experimental procedureis slightly modified by introducing in the reactor directly a solutionof Pd(hfac)2 dissolved in TEOS instead of the reactants in separated

rcritical Fluids 49 (2009) 369–376 375

vtPTtttbEtPmrtaiifcpiP

wsa3eiptf3tfstP(

cm

Fu3(

Fig. 8. Pore size distributions of aerogel inverse opals produced in scCO2 using a

M.J. Tenorio et al. / J. of Supe

ials. Further experiments were conducted using the PS–MA–IAemplate (0.014 mM BSA/g latex) and varying the concentration ofd(hfac)2 in TEOS (2.3 and 11 mol%, samples 7 and 8, respectively).he colour of the calcined material was used as a quick indication ofhe Pd incorporation into the SiO2. As expected, the Pd loading intohe composite materials was much lower for the 2.3 mol% than forhe 11 mol% Pd(hfac)2/TEOS solution yielding very light and darkrown materials, respectively. These findings were confirmed byDX analysis in the SEM images, with Pd loadings below the detec-ion limit for the 2.3 mol% Pd(hfac)2 and between 0.5 and 1.3 mol%d in SiO2 for the 11 mol% Pd(hfac)2 sample. Comparison of theaterials obtained when TEOS and Pd(hfac)2 were loaded in the

eactor in separated vials (10% mol Pd(hfac)2) or premixed in solu-ion (11% mol Pd(hfac)2) showed that the former approach affordslarger incorporation of Pd (between 1.8 and 3.2 mol% Pd by EDX

n SEM images). Even though the solubility of Pd(hfac)2 in scCO2s high, it is not as high as the solubility of Pd(hfac)2 in TEOS and,or similar Pd(hfac)2 and TEOS concentrations, at the experimentalonditions, the amount of Pd(hfac)2 dissolved in the supercriticalhase without stirring seems to be lower when Pd(hfac)2 dissolved

n excess TEOS. The phase diagram of the system formed by TEOS,d(hfac)2 and CO2 deserves a further study.

N2-adsorption isotherms of the different PdO–SiO2 inverse opalsere compared to that of a pure SiO2 aerogel inverse opal synthe-

ized using the same template (PS–MA–IA 0.014 mM BSA/g latex)nd TEOS under the same conditions (40 ◦C, 85 bar and 3 h), sample. Fig. 7 shows that the isotherms for bare SiO2 and the differ-nt PdO–SiO2 composite materials present type IV hysteresis loopsndicative of mesoporous materials. Furthermore, all the materialsresent high surface areas and very little microporosity. Except forhe material with the lowest Pd content (sample 7) whose BET sur-ace area is slightly lower than that of the bare SiO2 material (sample), the BET surface area of the composite materials increases withhe Pd content from 244 m2/g for bare SiO2 (sample 3) to 376 m2/gor the materials produced using the 11 mol% Pd(hfac)2 in TEOSolution (sample 8) and reaches a higher value of 447 m2/g forhe composite material with the highest loading obtained using

d(hfac)2 and TEOS in separated vials (10% mol Pd(hfac)2 in TEOS)sample 6).

Size distributions calculated from the desorption branch areompared in Fig. 8. Distribution curves are relatively broad withaxima between 13 and 15 nm. The macropore walls in the SiO2

ig. 7. Adsorption–desorption isotherms of aerogel inverse opals produced in scCO2

sing a PS–MA–IA 3D-latex array template following approach (2): (©) SiO2 (sample); (�) PdO–SiO2 prepared from Pd(hfac)2 previously dissolved in TEOS (sample 8),�) PdO–SiO2 prepared from Pd(hfac)2 and TEOS in separated vials (sample 6).

PS–MA–IA 3D-latex array template following approach (2) obtained from the des-orption branch of the isotherm using the BJH method: (©) SiO2 (sample 3); (�)PdO–SiO2 prepared from Pd(hfac)2 in TEOS solution (sample 8), (�) PdO–SiO2 pre-pared from Pd(hfac)2 and TEOS in separated vials (sample 6).

and PdO–SiO2 composite materials are highly mesoporous and thepore size distributions are very similar. The second peak at ca.45–50 nm in each distribution corresponds to the equivalent porediameter formed by three adjacent hollow spheres (triangular win-dow) in the inverse opal [7]. The higher the palladium content, thehigher the shrinkage of the network and the smaller the size of thetriangular window.

Although the pores size distributions are very similar, theincrease in the BET surface area of the samples with the Pd con-tent indicates that the presence of Pd(hfac)2 during the reaction isaffecting the reaction of TEOS. It is likely that the adsorption of TEOSon the acidic sites of the template competes with that of Pd(hfac)2.In fact, we have observed that the impregnation efficiency of thesame polymeric templates with Pd(hfac)2 without TEOS at the sameconditions increases with the acid concentration of the template.Keeping all the variables constant, Pd(hfac)2 seems to effectivelydecrease the acid concentration of the support. Reduction of thePdO–SiO2 composite material with the highest Pd loading in N2/H2at 400 ◦C did not change the surface and pore size distribution ofthe sample. We conclude that the optimum acid concentration onthe template to achieve the highest surface area is different for pureSiO2 and the Pd–SiO2 composite materials.

Because of the higher reactivity of TMOS in comparison to TEOS,if TMOS is used as the silicon precursor, the incorporation of Pdinto the composite materials is much lower than that for TEOS atthe same concentrations.

4. Conclusions

PdO–SiO2 and Pd–SiO2 aerogel inverse opals can be producedin scCO2 either by direct impregnation of Pd(hfac)2 into pre-madeSiO2 aerogel inverse opals, or by the simultaneous reaction andimpregnation of TEOS and Pd(hfac)2 into 3D-latex array templatesat 40 ◦C and 85 bar. In both cases, after decomposition of theprecursors and/or elimination of the template, Pd–SiO2 aerogelinverse opals with Pd loadings between 1 and 3 mol% Pd in SiO2were obtained. Pd or PdO clusters (depending on the heat treat-ment) were uniformly distributed throughout the SiO2 matrix. Thematerials have high surface areas (up to 450 m2/g) and pore size

distributions with maxima between 13 and 15 nm in the meso-pore range. The presence of Pd(hfac)2 during the reaction seems toaffect the TEOS reaction and the BET surface area of the materialsproduced by the one-step synthesis changes with the Pd incorpora-tion. Pd(hfac)2 seems to effectively decrease the acid concentration

3 rcritica

oltnttmaptarm

A

tUPPcd

R

[

[

[

[

[

[

[

[pressure phase behaviour for the binary system carbon dioxide + dibenzofuran,

76 M.J. Tenorio et al. / J. of Supe

f the support, so the optimum acid concentration on the templateeading to the largest surface area for pure SiO2 is different fromhat of the Pd–SiO2 composite materials. In contrast, the impreg-ation of pre-made SiO2 aerogel inverse opals with Pd(hfac)2 leadso a slight reduction in the surface area, but there are no changes inhe pore size distribution at the given loadings. In both cases, the

aterials present hierarchical porosity with ordered macroporesnd uniform mesopores. The combination of macropores and meso-ores in the structure and the uniform distribution of Pd clustershroughout the SiO2 matrix, make these materials good candidatess catalyst for a large number of catalytic reactions. We are cur-ently exploring the potential of this technique to produce otheretal–SiO2 structured composite materials.

cknowledgements

We gratefully acknowledge the financial support of MEC (Spain)hrough Research Projects CTQ2006-07172 and MAT2007-65711,niversidad Complutense de Madrid (UCM) through Researchroject Santander-UCM PR34/07-15789 and CAM (Spain) throughroject CCG08-UCM/MAT-4247. M.J. Tenorio thanks MEC for finan-ial support through a FPI predoctoral grant. We also thank “Centroe Microscopía Electrónica” at UCM for the technical assistance.

eferences

[1] A. Stein, F. Li, N.R. Denny, Morphological control in colloidal crystal templatingof inverse opals, hierarchical structures, and shape particles, Chem. Mater. 20(2008) 649–666.

[2] A. Stein, R. Schroden, Colloidal crystal templating of three-dimensionallyordered macroporous solids: materials for photonics and beyond, Curr. Opin.Solid State Mater. Sci. 5 (2001) 553–564.

[3] C. Aymonier, A. Loppinet-Serani, H. Reverón, Y. Garrabos, F.J. Cansell, Review ofsupercritical fluids in inorganic materials science, J. Supercrit. Fluids 3 (2006)242–251.

[4] M.A. McHugh, V.J. Krukonis, Supercritical Fluid Extraction: Principles and Prac-tice, Butterworths, Boston, 1986.

[5] O. Sphul, S. Herzog, J. Gross, I. Smirnova, W. Arlt, Reactive phase equilibria in sil-ica aerogel synthesis: experimental study and prediction of the complex phasebehavior using the PC-SAFT equation of state, Ind. Eng. Chem. Res. 43 (2004)4457–4464.

[6] A. Cabanas, E. Enciso, M.C. Carbajo, M.J. Torralvo, C. Pando, J.A.R. Renuncio, Syn-thesis of ordered macroporous SiO2 in supercritical CO2 using 3D-latex arraytemplates, Chem. Commun. (2005) 2618–2620.

[7] A. Cabanas, E. Enciso, M.C. Carbajo, M.J. Torralvo, C. Pando, J.A.R. Renuncio,Synthesis of SiO2–aerogel inverse opals in supercritical carbon dioxide, Chem.Mater. 17 (2005) 6137–6145.

[8] A. Cabanas, E. Enciso, M.C. Carbajo, M.J. Torralvo, C. Pando, J.A.R. Renuncio, Stud-ies on the porosity of SiO2–aerogel inverse opals in supercritical CO2, Micropor.Mesopor. Mater. 99 (2007) 23–29.

[

[

l Fluids 49 (2009) 369–376

[9] H.U. Blaser, A. Indolese, A. Schnyder, H. Steiner, M.J. Studer, Supported palla-dium catalysts for fine chemical synthesis, Mol. Catal. A: Chem. 173 (2001)3–18.

[10] A.F. Lagalante, B.N. Hansen, T.J. Bruno, R.E. Sievers, Solubilities of copper(II) andchromium (III) �-diketonates in supercritical carbon dioxide, Inorg. Chem. 34(1995) 5781–5785.

[11] W.C. Andersen, R.E. Sievers, A.F. Lagalante, T.J. Bruno, Solubilities of cerium(IV),terbium(III) and iron(III) �-diketonates in supercritical carbon dioxide, J. Chem.Eng. Data 46 (2001) 1045–1049.

12] J.M. Blackburn, D.P. Long, J.J. Watkins, Reactive deposition of conformal pal-ladium films from supercritical carbon dioxide, Chem. Mater. 12 (2000)2625–2631.

[13] N.E. Fernandes, S.M. Fisher, J.C. Poshusta, D.G. Vlachos, M. Tsapatsis, J.J. Watkins,Reactive deposition of metal thin films within porous supports from supercrit-ical fluids, Chem. Mater. 13 (2001) 2023–2031.

[14] J.M. Blackburn, D.P. Long, A. Cabanas, J.J. Watkins, Deposition of conformalcopper and nickel films from supercritical carbon dioxide, Science 293 (2001)141–145.

[15] A. Cabanas, D.P. Long, J.J. Watkins, Deposition of gold films and nanostructuresfrom supercritical carbon dioxide, Chem. Mater. 16 (2004) 2028–2033.

[16] X.R. Ye, Y.H. Lin, C.M. Wai, Decorating catalytic palladium nanoparticles oncarbon nanotubes in supercritical carbon dioxide, Chem. Commun. 5 (2003)642–643.

[17] X.R. Ye, H.F. Zhang, Y.H. Lin, L.S. Wang, C.M. Wai, Modification of SiO2 nanowireswith metallic nanocrystals from supercritical CO2, J. Nanosci. Nanotechnol. 4(2004) 82–85.

[18] X.R. Ye, Y.H. Lin, C.M. Wai, J.B. Talbot, S.H. Jin, Supercritical fluid attachment ofpalladium nanoparticles on aligned carbon nanotubes, J. Nanosci. Nanotechnol.5 (2005) 964–969.

[19] K.S. Morley, P. Licence, P.C. Marr, J.R. Hyde, P.D. Brown, R. Mokaya, Y.D. Xia, S.M.Howdle, Supercritical fluids: a route to palladium–aerogel nanocomposites, J.Mater. Chem. 14 (2004) 1212–1217.

20] R.C. Jiang, Y. Zhang, S. Swier, X.Z. Wei, C. Erkey, H.R. Kunz, J.M. Fenton, Prepara-tion via supercritical fluid route of Pd-impregnated nafion membranes whichexhibit reduced methanol crossover for DMFX, Electrochem. Solid-State Lett. 8(2005) A611–A615.

21] S.S. Lee, B.K. Park, S.H. Byeon, F. Chang, H. Kim, Mesoporous silica-supported Pdnanoparticles; highly selective catalyst for hydrogenation of olefins in super-critical carbon dioxide, Chem. Mater. 18 (2006) 5631–5633.

22] S.S. Lee, H.I. Park, B.K. Park, S.H. Byeon, Influence of solvents on the forma-tion of Pd and PdO nanoparticles in SBA-15, Mater. Sci. Eng. B 135 (2006)20–24.

23] M.C. Carbajo Moreno, Ph.D. Thesis, Universidad Complutense de Madrid (Spain),2003.

24] D.L. Tomasko, H. Li, D. Liu, X. Han, M.J. Wingert, J.L. Lee, K.W. Koelling, A review ofCO2 applications in the processing of polymers, Ind. Eng. Chem. Res. 42 (2003)6431–6456.

25] A. Cabanas, E. Enciso, M.C. Carbajo, M.J. Torralvo, C. Pando, J.A.R. Renuncio, Effectof supercritical carbon dioxide on the ordering and aggregation of polystyrenelatex particles, Langmuir 22 (2006) 8966–8974.

26] E. Pérez, A. Cabanas, Y. Sánchez-Vicente, J.A.R. Renuncio, C. Pando, High-

J. Supercrit. Fluids 46 (2008) 238–244.27] M. Jaroniec, M. Kruk, J.P. Olivier, Standard nitrogen adsorption data for charac-

terization of nanoporous silicas, Langmuir 15 (1999) 5410–5413.28] E. Barret, L.G. Joyner, P.P. Halenda, The determination of pore volume and area

distributions in pure substances, J. Am. Chem. Soc. 73 (1951) 373–380.