Transcript
Preservation of the allophanic soils structure by supercritical drying Thierry Woignier a,b,* , Juan Primera b,c , Laurent Duffours d , Philippe Dieudonne´ b ,Abdelazize Raada e a SeqBio IRD-Pole de Recherche Agronomique de la Martinique, Quartier Petit Morne, BP 214, 97285 Le Lamentin Cedex 2, France b CNRS, University of Montpellier 2, Montpellier, France c Departamento de fisica, FEC, LUZ, Maracaibo, Venezuela d PrimeVerre, im. A. Einstein, Parc du Mille´ naire, Montpellier, France e UMI, FSTE, De´ partement de Chimie, BP 509 Boutalamine, Errachidia, Morocco Received 26 September 2006; received in revised form 9 May 2007; accepted 11 May 2007Available online 26 May 2007 Abstract Allophanic soils are interesting in terms of environmental properties especially because of their potentialities as sinks for ‘‘greenhousegases’’: Allophane clays are natural mesoporous materials exhibiting organic carbon and nitrogen contents 3–4 times higher than thosemeasured in other clay soils. We present results on the preservation of the porous features of allophanic soils, at the nanometer scale, bysupercritical drying technique (SD).We show that textural properties such as specific surface area are systematically higher for the supercritical dried samples compared tothe classical dried samples, indicating the preserving effect of supercritical drying. Pore size distribution and small angle X-rays scatteringdata confirm specific surface area results. The structure at the nanometer scale is affected by classical dying, which reveals the interest of the SD method to correctly characterize natural and complex mesoporous material such as allophane. Thanks to this approach we showcorrelation between the true specific surface area and the carbon content, in allophanic soils. 2007 Elsevier Inc. All rights reserved. Keywords: C-sequestration; Supercritical drying; Allophane; SAXS; Specific surface area 1. Introduction The net emissions of greenhouse gases may be reducedeither by decreasing the rate at which they are emitted intothe atmosphere or by removing of greenhouse gases,through sinks (carbon sequestration [1,2]). In this way,agricultural soils which are large planet’s reservoirs of car-bon provide a prospective way of reducing the increasingatmospheric concentration of CO 2 .Volcanic soils containing allophanes (a mesoporousamorphous mineral) are interesting for their environmentalproperties. These soils called allophanic soils or andosols[3] exhibit higher organic carbon concentration (up to afactor of 4) [4] than other clay soils (kaolinic or smectitic)and can be considered as C sinks. They comprise weather-ing products such as allophanes srcinating from the leach-ing of volcanic ash and glasses and in a previous study [5]we have confirmed the clear influence of the allophane con-centration on C concentration. A paper recently published[6] shows also a strong correlation between the C contentand the poorly crystalline phase (allophane, protoimogolite).There is then a need to establish correlation between thesequestration mechanism in allophanic soils and physicalparameters in these soils. It is reasonable to admit thatthe ability of a soil for C sequestration could be relatedto the soil structure and pore features but the experimentaltechniques able to provide such information generallyrequire dried solids samples. Unfortunately, during a 1387-1811/$ - see front matter 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.micromeso.2007.05.019 * Corresponding author. Address: SeqBio IRD-Pole de RechercheAgronomique de la Martinique, Quartier Petit Morne, BP 214, 97285Le Lamentin Cedex 2, France. E-mail address:
[email protected] (T. Woignier). www.elsevier.com/locate/micromeso Available online at www.sciencedirect.com Microporous and Mesoporous Materials 109 (2008) 370–375 classical drying, allophanic soils exhibit an important irre-versible shrinkage which modifies the soil structure andporous features.This irreversible shrinkage comes from capillary stressesin the pores and from the large compliance of the poroussolids.In the literature we can read that the same kind of prob-lem (large shrinkage) has been solved in the case of syn-thetic sol–gel materials by supercritical drying (SD) [7].The gel is a two-phase medium containing the solid net-work and the liquid (alcohol, water). The structure of thesolid network can be described as an assembly of fractalclusters ( 50 nm) [8], built by the aggregation of small par-ticles ( 1–2 nm). During classical drying, capillary forcescollapse the gel structure resulting in a significant shrink-age. The magnitude of those stresses is dependent on theinterfacial energy c of the liquid. By supercritical dryingtechniques (SD) it is possible to suppress c and capillarystresses if the pressure and temperature pass over the crit-ical point of the liquid and preserve the porosity of thesolid network [7].So, it exist a similar behaviour during drying betweensynthetic gels and allophane aggregates and we proposeto use the supercritical drying to preserve the pores featuresof allophanic soils. This new way to dry the soil beforeanalysis would allow getting a more accurate and morerealistic description of soil organization at the nanometerscale.Thanks to this approach we present results on the struc-tural and textural properties of the allophanic clays. Wewill also discuss the C content of these soils and find corre-lation between the allophane content, the ‘‘real’’ porousfeatures (preserved by SD) and the carbon content in allo-phanic soils. 2. Experimental We selected several (19) allophanic soils, allophane con-tent is measured by the method of Mizota and van Reewijk[10] (Al and Si content are extracted by oxalate and pyro-phosphate). In this study the term ‘‘allophane content’’could correspond to different poorly crystalline alumino sil-icate compounds: allophane (hollows spheres of around 3– 4 nm); proto imogolite allophane (allophane with a Al/Siclose to 2) and proto imogolite (same short range structurethan imogolite (Al/Si 2) but with a weak fibrosity).Imogolite forms hollow tube (2 nm diameter). However,in a previous study [11] the IR spectra of the studied ando-sols do not present doublets of the band at 577 and967 cm 1 characteristic of the imogolite structure and theTEM micrographs presented in this study (Fig. 2) will showthat fibrous structure are not observed. We can assume thatthe major part of the allophane content measured is neitherproto imogolite nor imogolite.Soils have been sampled in A (humus surface) horizonor B (buried) horizon which correspond to two differentkinds of soils samples. Samples were conserved in closedcontainers to avoid evaporation. Carbon contents weremeasured with a CHN (thermofinnigan) chromatographanalyzer with a relative error lower 4%.To check the importance of the supercritical drying pro-cess, two series of dried soils samples have been prepared.The first set is obtained by a classical drying in an oven at45 C during 2 days and the second set of samples is driedby CO 2 supercritical drying (45 C and 80 bars). The super-critical drying process requires over passing the criticalpoint of the liquid. In this study, the apparatus used is aCritical Point Dried Balzers, CO 2 supercritical dryingneeds a previous solvent exchange and the whole procedureis similar to that previously published [11,12].In the following samples will be labeled as cdX or sdX,cd and sd referring to classical and supercritical drying,respectively, and X to the allophanic weight%.To compare the influence of the supercritical drying onthe soils properties different kinds of features have beenstudied or measured. The specific surface area ( S ) was mea-sured by N 2 adsorption–desorption techniques (BET anal-ysis) with a micromeritics ASAP 2010. The estimatedrelative error is 5%. The pore size distribution is calculatedusing the BJH method. The out gassing conditions are 24 hat 50 C, with a vacuum 2–4 l m Hg. Transmission electronmicroscopy was performed on samples SD18 and CD 18with a TEM JEOL Type 1200 EX (100 kV).Structure at the nanometer scale will be studied by smallangle X-ray scattering (SAXS). SAXS experiments werecarried out on solid powders in 1 mm diameter glass capil-laries. We worked in a transmission configuration. A cop-per rotating anode X-ray source (functioning at 4 kW) witha multilayer focusing ‘‘Osmic’’ monochromator givinghigh flux (10 8 photons/s) and punctual collimation wasemployed. An ‘‘Image plate’’ 2D detector was used.X-ray diagrams relating scattered intensity to the wave vec-tor q were obtained. Scattered intensity was corrected bytransmission and intensity background coming from scat-tering by an empty capillary. 3. Results Fig. 1a and b show the evolution of the carbon contentas a function of the allophane weight% in the samples.These figures confirm that the carbon concentrationincrease with the allophane weight% and the A horizons(more rich in humus) contain more carbon. These resultsare in agreement with the results announced in the litera-ture [4–6].Fig. 2a and b shows 2 TEM micrographs of sample sd18at different focus. The allophane structure is very openmade of aggregated small particles (3 nm) (Fig. 2a), build-ing clusters with size close to 8–10 nm (Fig. 2a). This clus-ters can stick and form larger aggregates ( 100 nm,Fig. 2b). The micrographs are similar to TEM micrographsof synthetic gels forming fractal structures. This descriptionis in agreement with literature results [3,4] and with a qual-itative fractal description of allophane aggregates [13–15]. T. Woignier et al. / Microporous and Mesoporous Materials 109 (2008) 370–375 371 As explained in the introduction section during a classi-cal drying, allophanic soils exhibit a much larger shrinkagethan other clay soils. Fig. 3 compares the drying curvesbetween an allophanic soil and a typical clay soil (kaolinic)and shows clearly a specific volume and water content loss4–5 times larger than for kaolinic soils. Because of this highamount of liquid and the associated important shrinkageupon classical drying, it is clear that the porous featureswill be strongly affected.Because SD preserves the porous structure, the specificsurface areas measured after supercritical drying (Ssd) aresystematically higher than those measured after classicaldrying (Scd). Fig. 4 displays plot of the difference Ssd– Scd. This difference increases with the allophane content,suggesting that the allophane aggregates probably are spe-cially affected by the classical drying.Fig. 5 compares the pore size distribution curves corre-sponding to the samples sd18 and cd18. The sd18 curveshows a large pore size distribution covering the range10–60 nm which is in a qualitative agreement with a fractaldescription of the allophane aggregate. This figure showsclearly that the pores structure is strongly affected by theclassical drying in the nanometer pore range. The majorpart of the porosity between 10 and 60 nm has been col-lapsed by capillary forces.The comparison of the sd18 (Fig. 2b) and cd18 (Fig. 2c) confirms the results of Figs. 4 and 5. TEM micrographsshow that the proposed fractal structure made of aggre-gated allophane particles forming aggregates (8–10 nm)which built larger aggregates (50–100 nm) (Fig. 2b) isdestroyed upon CD (Fig. 2c). Fig. 2c shows a more homo- geneous structure made of ‘‘particles’’ lower than 10 nmresulting from the collapse, upon classical drying.To precise the mesostructure of the allophane aggregateswe have performed SAXS experiments on the samples sd18and cd18. Fig. 6 shows the evolution of the scattered inten-sity ( I ) versus wave vector ( q ) and according to the litera-ture [8,9], the main features of the structure can beanalysed in terms of fractal geometry. SAXS experimentsprovides three different information of the structure: themean size of fractal clusters ( n ) the mean size of primaryparticles ( a ) which stick together to build the cluster, andthe fractal dimension D which express the clusters com-pactness. For each curve, the position of the two cross over(black and white arrows) are related to the inverse of thecluster size (2 p / n ) and the inverse of the particle size(2 p / a ), respectively. The power law part appearing as linearin a log–log scale has a slope related to D .In Fig. 6, the SAXS curve shows a power law behaviorlike q D (the slope is in the range 2.2–3). Sample sd18 showsa hump at q close to 2–3 nm 1 attributed to the allophaneparticle size. In the low q range, the cross over correspond-ing to the cluster size (the black arrow means that the fractalrange could cover more than one order of magnitude,between at least 30 and 3 nm.These results are in goodagreement with the fractal description of allophane aggre-gates in the literature [13–15] and the above results of trans-mission electronic microscopy. The confirmation of theimportant effect of classical drying on the structure of theallophane aggregates is also shown in Fig. 6 which com-pares the SAXS curves measured on CD18 and SD18.SAXS results demonstrate the intuitive explanation thatthe loss of specific surface area is the result of a collapseat the allophane aggregate scale. Sample cd18 displays thesame kind of hump at q 2–3 nm 1 attributed to the allo-phane particles size (3 nm) but in the low q range the crossover appears for q close to 1 nm 1 (black arrow) which cor-respond to a cluster size n 6 nm. SAXS data shows clearlythat the classical drying affects the structure of the allo-phane aggregates and decreases the aggregates size. 4. Discussion One of the important points of this study is that the allo-phane structure is a natural mesoporous material quiteclose to synthetic gel and behaves in a similar way duringdrying. This point must be discussed in this part of thepaper.It is clear that the formation of allophane and the gela-tion process are different. The allophanes are obtained bythe transformation of volcanic ashes and glass particlesand the weathering leads to over-saturated solutions whichprecipitate. The fine particle size and the predominance of glass favor preferential formation of poorly crystallinephase such as allophane and/or imogolite [16,17]. In thecase of the silica sol–gel process, the silicic acid solutionpolymerizes into discrete particles that aggregate intochains and networks [18]. However, in both cases the 0510150510 allophane (%) C ( % ) 02468100102030 allophane (%) C ( % ) ab Fig. 1. C weight% versus the allophane content in the soils: A horizons (a)and B horizons (b).372 T. Woignier et al. / Microporous and Mesoporous Materials 109 (2008) 370–375 network is obtained by the aggregation of colloidal parti-cles whatever, the way of formation of these particles.The fact that a corrosion process of glass structure couldleads to a ‘‘gel’’ is not surprising. The corrosion of glasseshas been largely studied in the case of nuclear glasses [19](the glasses for the containment of the radioactive wastes)and archeological glasses [20] and the alteration layerbetween the glass and water formed by precipitation of amorphous silicates is called and described as a gel.As explained in the introduction, it has been demon-strated that the supercritical drying is able to preserve thefragile network of synthetic gels [7,9]. Moreover, previousworks have also shown the interest of the SD to preservethe fine layer of gel developed by lixiviation at the interfaceof buried ancient glasses [20] and nuclear glasses [19]. These previous studies show the potentiality of the SD to preservethe texture of allophane.With the results of Figs. 2–6 we have shown that classi-cal drying modifies the textural properties of andosols,shrinking and destroying the allophane aggregates.Because of the transformation at the nanometer scale thespecific surface area is strongly affected and this effect Fig. 2. (a, b) TEM micrographs of sample sd18. (c) TEM micrographs of sample cd18. T. Woignier et al. / Microporous and Mesoporous Materials 109 (2008) 370–375 373 depends on allophane content. The comparison of Figs. 7and 8 proves without ambiguity the interest of SD to havea precise and more realistic description of the allophanicsoils structure.Without SD no positive correlation between specific sur-face area and C content is observed (Fig. 7). In contrast,the plot the carbon content versus specific surface area(Fig. 8) shows that the carbon content is dependent onSsd, for the two different kinds of soils studied (horizonsA and B).These results allow proposing an explanationfor the C content in allophanic soils (Fig. 1). High specificsurface areas will increase the possible adsorption and cer-tainly participate to the sequestration mechanism. How-ever, classical clays may develop high specific surfacearea but they do not contain so large C and N. For usthe specific structure (fractal) of the allophane aggre-gate [3,4,13–15] could play a role in the sequestrationprocess.The large specific surface area is the signature of smallpore size. Associated to the allophane tortuous structure,the permeability at the scale of the allophane aggregateswill be very low. For C species located in or near the allo-phane aggregates, possible exchanges or chemical reactionswith others chemical species will be difficult because of thislow permeability. The fluids will have more difficulty tomigrate inside the porous and tortuous porosity of theallophane aggregates compared to classical clays. The fluidwill be confined and trapped in the porosity and thechemical exchanges inside the allophane aggregates willbe poor. This low permeability can be a part of theexplanation concerning the high carbon content seques-trated in allophanic soils. Besides the well known chemicalaspect of the C accumulation in allophanic soils (stablehumus – Al, Fe complexes) [3] we propose that a physicalparameter like tortuosity and permeability, related to thepeculiar structure of allophane aggregates [13–15] couldplay a role. 01234503 water content (g/g) V s p e c ( c m 3 / g ) allophanic soilskaolinic clay 12 Fig. 3. Drying curves of a classical allophanic soil and a classical clay soil:specific volume ( V spec ) versus water content. 0501001502000102030 allophane content % S s d - S c d ( m 2 / g ) Fig. 4. Specific surface areas difference Ssd–Scd versus allophanic content. 00,0050,010,0150,020,0250,030,0350,040 20 40 60 80 100 Diameter (nm) d V / d R ( c m 3 / g / n m ) sd18cd18 Fig. 5. Pore size distributions of samples sd18 and cd18. 11010010001000010000010000000,1 1 10 q (nm-1) I ( q ) ( a . u . ) sd18cd18 Fig. 6. SAXS curves of samples sd18 and cd18. 02468101214050100150 specific surface area( m 2 /g) C c o n t e n t ( % ) B horizon Fig. 7. C weight% versus the specific surface area (Scd) measured on thecd set. B horizon (diamond) and A horizon (squared). 02468101214050100150200 specific surface area( m 2 /g) C c o n t e n t ( % ) A horizonB horizon Fig. 8. C weight% versus the specific surface area (Ssd) measured on thesd set. B horizon (diamond) and A horizon (squared).374 T. Woignier et al. / Microporous and Mesoporous Materials 109 (2008) 370–375
