DOI: 10.1021/la901560h 11897Langmuir 2009, 25(19), 11897–11904 Published on Web 07/14/2009

pubs.acs.org/Langmuir

© 2009 American Chemical Society

High-Resolution Nanoprobe X-ray Fluorescence Characterization of

Heterogeneous Calcium and Heavy Metal Distributions in Alkali-Activated

Fly Ash

John L. Provis,*,† Volker Rose,‡ Susan A. Bernal,†,§ and Jannie S. J. van Deventer†

†Department of Chemical and Biomolecular Engineering, University of Melbourne, Victoria 3010, Australia,‡Advanced Photon Source, Argonne National Laboratory, Argonne, Illinois 60439, and §Materials Engineering

Department, Composite Materials Group (CENM), Universidad del Valle, Cali, Colombia

Received May 1, 2009. Revised Manuscript Received June 27, 2009

The nanoscale distribution of elements within fly ash and the aluminosilicate gel products of its alkaline activation(“fly ash geopolymers”) are analyzed by means of synchrotron X-ray fluorescence using a hard X-ray Nanoprobeinstrument. The distribution of calcium within a hydroxide-activated (fly ash/KOH solution) geopolymer gel is seen tobe highly heterogeneous, with these data providing for the first time direct evidence of the formation of discrete high-calcium particles within the binder structure of a geopolymer synthesized from a low-calcium (<2 wt % as oxides) flyash. The silicate-activated (fly ash/potassium silicate solution) sample, by contrast, shows a much more homogeneousgeopolymer gel binder structure surrounding the unreacted fly ash particles. This has important implications for theunderstanding of calcium chemistry within aluminosilicate geopolymer gel phases. Additionally, chromium and iron areseen to be very closely correlated within the structures of both fly ash and the geopolymer product and remain within theregions of the geopolymer which can be identified as unreacted fly ash particles. Given that the potential for chromiumrelease has been one of the queries surrounding the widespread utilization of constructionmaterials derived from fly ash,the observation that this element appears to be localized within the fly ash rather than dispersed throughout the gelbinder indicates that it is unlikely to be released problematically into the environment.

Introduction

Fly ash is a byproduct of coal combustion produced at a rate ofmore than 450 million tonnes per annum worldwide1,2 and sopresents significant challenges in disposal in addition to poten-tially highly profitable opportunities for valorization. Fly ashconsists of particles ranging in size from∼0.1 to>100 μm, whichare heterogeneousonboth interparticle and intraparticle levels,3-5

which makes it a particularly interesting and challenging materialto characterize. Here, the new hard X-ray Nanoprobe instrumentoperated by the Center for Nanoscale Materials in partnershipwith the Advanced Photon Source, Argonne National Labora-tory,6,7 is used to study the nanoscale elemental disposition withinfly ash particles and the hardened gel products, X-ray amorphous

aluminosilicate materials termed “inorganic polymers” or “fly ashgeopolymers”,8,9 formed by the reaction of fly ash with alkalinesolutions. Geopolymers are currently being developed as anenvironmentally beneficial replacement for Portland cement forconcrete production, offering comparable performance and costwhile reducing greenhouse gas emissions by a factor of approxi-mately 5.10 Given that cement production is responsible for up to8% of global anthropogenic CO2 emission,2 this provides theopportunity to reduce CO2 emission by at least tens of millions oftonnes per annum worldwide. However, a number of aspects ofgeopolymer structure are not yet well understood, in particular,the role of calcium within the geopolymer structure and thepossibility of the release of toxins from the fly ash into theenvironment while the material is in use.

There have been a number of previous studies of elementaldistributions within fly ash particles on a variety of length scales,using several different analytical techniques. Volatile elements arebelieved to become enriched at the surfaces of some large fly ashparticles, aswell as in the fine fractions of ashes.11-13Early studiesusing particle-inducedX-ray emission (PIXE) in a protonmicrop-robe showed strong correlations between specific elements, inparticular, a marked correlation between Fe and Ti (as well asvarious other elements, including Cr) and a weak correlationbetween these elements and Ca.14 Other studies have also beenaimed at classifying fly ash particles into different categories

*Towhom correspondence should be addressed. E-mail: jprovis@unimelb.edu.au. Phone: þ61 3 8344 8755. Fax: þ61 3 8344 4153.(1) Manz, O. E. Fuel 1997, 76, 691.(2) Humphreys, K.; Mahasenan, M. Toward a sustainable cement industry.

Substudy 8: Climate change, World Business Council for Sustainable Develop-ment, 2002.(3) Hemmings, R. T.; Berry, E. E. In Fly Ash and Coal Conversion By-Products:

Characterization, Utilization, and Disposal IV; McCarthy, G. J., Glasser, F. P.,Roy, D. M., Eds.; Materials Research Society: Warrendale, PA, 1988; pp 3-38.(4) Qian, J. C.; Lachowski, E. E.; Glasser, F. P. In Fly Ash and Coal Conversion

By-Products: Characterization, Utilization, and Disposal IV; McCarthy, G. J.,Glasser, F. P., Roy, D. M., Eds.; Materials Research Society: Warrendale, PA,1988; pp 45-54.(5) Nugteren, H. W. Part. Part. Syst. Charact. 2007, 24, 49.(6) Maser, J.; Winarski, R.; Holt, M.; Shu, D.; Benson, C.; Tieman, B.;

Preissner, C.; Smolyanitskiy, A.; Lai, B.; Vogt, S.; Wiemerslage, G.; Stephenson,G. B. In Proceedings of the 8th International Conference on X-ray Microscopy;Aoki, S., Kagoshima, Y., Suzuki, Y., Eds.; IPAP: Himeji, Japan, 2005; pp 26-29.(7) Shu, D.; Maser, J.; Holt, M.; Winarski, R.; Preissner, C.; Smolyanitskiy, A.;

Lai, B.; Vogt, S.; Stephenson, G. B. AIP Conf. Proc. 2007, 879, 1321.(8) Wastiels, J.; Wu, X.; Faignet, S.; Patfoort, G. In Proceedings of the

9th International Conference on Solid Waste Management; Widener University:Philadelphia, 1993; pp 8.(9) Duxson, P.; Fern�andez-Jim�enez, A.; Provis, J. L.; Lukey, G. C.; Palomo, A.;

van Deventer, J. S. J. J. Mater. Sci. 2007, 42, 2917.

(10) Duxson, P.; Provis, J. L.; Lukey, G. C.; van Deventer, J. S. J. Cem. Concr.Res. 2007, 37, 1590.

(11) Linton, R. W.; Loh, A.; Natusch, D. F. S.; Evans, C. A.; Williams, P.Science 1976, 191, 852.

(12) Linton, R. W.; Williams, P.; Evans, C. A. Anal. Chem. 1977, 49, 1514.(13) Cereda, E.; BragaMarcazzan, G.M.; Pedretti, M.; Grime, G.W.; Baldacci,

A. Atmos. Environ. 1995, 29, 2323.(14) Jak�si�c, M.; Watt, F.; Grime, G. W.; Cereda, E.; Braga Marcazzan, G. M.;

Valkovi�c, V. Nucl. Instrum. Methods Phys. Res., Sect. B 1991, 56-57, 699.

11898 DOI: 10.1021/la901560h Langmuir 2009, 25(19), 11897–11904

Article Provis et al.

according to their composition and/ormicroscopic appearance.15,16

Such analysis will not be conducted in this paper, because the veryfine spatial resolution of the Nanoprobe instrument (∼30 nm) isideally suited to detailed analysis of small regions rather thanexhaustive samplingof large numbers ofmicrometer-sizedparticles.

SynchrotronX-ray fluorescence (μ-XRF and tomography) haspreviously been applied to the study of individual coal fly ashparticles.17-19 However, the products formed by the alkalineactivation of fly ash have not previously been subjected to thistype of analysis. Geopolymer gels are known to be heterogeneouson a length scale of nanometers to micrometers,9,20-23 but thenature of this heterogeneity in fly ash-derived geopolymers is atpresent poorly understood.24-26 Therefore, the primary aim ofthis paper is to analyze the structures of these materials on asubmicrometer length scale. One complicating factor in thegeneration of construction materials from wastes is that wastematerials must be either used “as is” or preprocessed; this meansthat the ability to understand the chemistry of fly ash and itsinteraction with alkaline solutions is critical to the successfulvalorization of a wide range of fly ash sources worldwide.

Additionally, a key question in the reuse or valorization of anywaste material is whether it contains any toxic or otherwise hazar-dous components which will impact its users. In the case of fly ash,the element of primary concern is usually chromium. Chromium isboth highlymobile and toxic in its hexavalent form,which is releasedinpreference to thepoorly solubleCr(III) upon exposureof fly ash toalkaline conditions.27 The fraction of chromium which is present asCr(VI) in fly asheshas been shown tovarywidely, froma fewpercentin most ashes up to a few rare instances in which 40% of the Crpresent is hexavalent.28,29 IfCr is being released fromtheashparticlesduringgeopolymerization, itwould thereforebe expected tobe in thisproblematic form, and this behavior must be understood to ensurethe safety of geopolymers in general construction applications.

Materials and Methods

The ash studied here was taken fromGladstone power station,Queensland,Australia, derived fromthe combustionof black coal

for electricity generation, and is Class F according to ASTMC618. It is usually sold for use as a cement additive and is arelatively fine ash,with a d50 of 10μmand a d80 of 20μm.The bulkchemical compositionofGladstone fly ash is given inTable 1, andpowder X-ray diffraction data are given in Figure 1.

It should also be noted that in Figure 1, the iron oxide phasescannot be specifically identified because of severe peak overlapbetween maghemite and a substituted ferrite spinel phase similartomagnetite.30Ashes that are low inunburnt carbon, as is the casehere, have previously been observed to have similar amounts ofFe2O3- and Fe3O4-type phases,31 so it is likely that both arepresent here. The importanceofX-raydiffraction is that it enablesthe identification of the ferrite spinel, which is able to host Cr3þ asan isomorphous substituent on the Fe3þ sites, as an importantphase in the fly ash studied.

We prepared fly ash samples for analysis by embedding a smallamount of fly ash in a commercial epoxy binder (Araldite, Selleys,Australia). We prepared alkali-activated fly ash samples byreacting the fly ash sample with an 11 M KOH solution(“hydroxide-activated sample”) at a mass ratio of 0.41, and withapotassiumsilicate solution [11MKOHand11MSiO2 (“silicate-activated sample”)] at amass ratio of 0.56. Thesemass ratios gaveidentical molar ratios of alkali to fly ash, to enable directcomparison of the two samples. Samples were cured in sealedmolds at 40 �C for 7 days and then demolded and maintained atroom temperature in sealed plastic bags. Analysis was conductedon specimens which had been polished to a thickness of 100-200 μm using successively finer grades of abrasive and withethanol as a lubricant to minimize leaching of soluble compo-nents, based on the procedure used by Lloyd et al.22

Samples were analyzed using the Center for NanoscaleMateri-als Hard X-ray Nanoprobe instrument, which operates on Sec-tor ID-26 at the Advanced Photon Source, Argonne NationalLaboratory.6,7 Briefly, this instrument uses two undulators as a

Table 1. Oxide Composition of Gladstone Fly Ash As Given by Bulk X-ray Fluorescence Analysis

SiO2 Al2O3 Fe2O3 CaO MgO SO3 Na2O K2O Cl Cr2O3 Mn3O4 P2O5 SrO TiO2 ZnO LOIa

wt % 47.68 30.28 11.26 1.84 1.41 0.14 0.33 0.47 0.362 0.0138 0.153 0.932 0.146 1.67 0.0225 3.7aLOI is the loss on ignition at 1000 �C.

Figure 1. PowderX-raydiffractiondata (CuKR radiation,PhilipsPW1800, 0.02� 2θ step size, 4 s/step) for the fly ash used in thisinvestigation. M is mullite, Q quartz, and F a combination of ironoxides,mostly a “ferrite spinel” (Fe3O4with substitutionof variouselements ontobothFe2þ andFe3þ sites) but alsopossibly includingmaghemite and other iron oxides with peaks that overlap at theresolution of the instrument used.

(15) Pietersen, H. S.; Vriend, S. P.; Poorter, R. E. P.; Bijen, J. M. In Fly Ash andCoal Conversion By-Products: Characterization, Utilization, and Disposal VI; Day,R. L., Glasser, F. P., Eds.; Materials Research Society: Warrendale, PA, 1990;pp 115-126.(16) Gier�e, R.; Carleton, L. E.; Lumpkin, G. R. Am. Mineral. 2003, 88, 1853.(17) Somogyi, A.; Janssens, K.; Vincze, L.; Vekemans, B.; Rindby, A.; Adams,

F. Spectrochim. Acta, Part B 2000, 55, 1039.(18) Golosio, B.; Simionovici, A.; Somogyi, A.; Camerani, C.; Steenari, B. M. J.

Phys. IV 2003, 104, 647.(19) Vincze, L.; Somogyi, A.; Os�an, J.; Vekemans, B.; T€or€ok, S.; Janssens, K.;

Adams, F. Anal. Chem. 2002, 74, 1128.(20) Provis, J. L.; Lukey, G. C.; van Deventer, J. S. J. Chem. Mater. 2005, 17,

3075.(21) Duxson, P.; Provis, J. L.; Lukey, G. C.; Mallicoat, S. W.; Kriven, W. M.;

van Deventer, J. S. J. Colloids Surf., A 2005, 269, 47.(22) Lloyd, R. R.; Provis, J. L.; vanDeventer, J. S. J. J.Mater. Sci. 2009, 44, 608.(23) Lloyd, R. R.; Provis, J. L.; vanDeventer, J. S. J. J.Mater. Sci. 2009, 44, 620.(24) Rees, C. A.; Provis, J. L.; Lukey, G. C.; van Deventer, J. S. J. Langmuir

2007, 23, 8170.(25) Rees, C. A.; Provis, J. L.; Lukey, G. C.; van Deventer, J. S. J. Langmuir

2007, 23, 9076.(26) van Deventer, J. S. J.; Provis, J. L.; Duxson, P.; Lukey, G. C. J. Hazard.

Mater. 2007, A139, 506.(27) Narukawa, T.; Riley, K. W.; French, D. H.; Chiba, K. Talanta 2007, 73,

178.(28) Huggins, F. E.; Najih, M.; Huffman, G. P. Fuel 1999, 78, 233.(29) Goodarzi, F.; Huggins, F. E. Energy Fuels 2005, 19, 2500.

(30) Winburn, R. S.; Lerach, S. L.; McCarthy, G. J.; Grier, D. G.; Cathcart, J.D. Adv. X-Ray Anal. 2000, 43, 350.

(31) Chaddha, G.; Seehra, M. S. J. Phys. D: Appl. Phys. 1983, 16, 1767.

DOI: 10.1021/la901560h 11899Langmuir 2009, 25(19), 11897–11904

Provis et al. Article

hard X-ray source, a Si(111) double-crystal monochromator toselect the photon energy for X-ray excitation of the sample andFresnel zone plate optics for focusing ofX-rays on the sample. Anactive vibration control system based on laser interferometersprovides accurate positioning. The selected photon energy was11.3 keV, which is to some extent limited by the available zoneplateX-ray optics but did not restrict the utility of the informationobtained from this study as the elements heavier than Fe that arepresent in fly ash are not generally of significant interest ingeopolymer chemistry. The photon flux of the focused beamwas ∼5 � 108 photons/s. Emitted characteristic X-ray fluores-cence radiation was detected with a four-element silicon driftenergy dispersive detector (Vortex ME4). Scanning of the zoneplate provided high-resolutionmaps of the elemental distributionof selected regions of the samples.

The experimental setup used, operating in air, was capable ofdetecting fluorescence from elements with Z g 18. Particularattention was paid to K, Ca, Fe, and Cr in the samples studiedhere. Data for Ti were also collected; however, the levels of Tiobserved at most points in the samples were very low, so thiselement was not subjected to such detailed analysis. Four fullspectra are obtained at each point, one for each element of theVortex detector. From these spectra, regions of interest (ROIs) ineach spectrum are identified (see the Supporting Information),and the correspondingROIs for all detector elements are summedto provide the total counts for Fe, Ca, Cr,K, andTi at each point.To achieve a high energy resolution, the full width at half-maximum of each ROI was selected to be ∼150 eV.

Datawere collected for 2.0 s per point, with a step size of 80 nmselected to enable utilization of the high spatial resolution of theinstrument while mapping an area which is large enough to beconsidered to some extent representative of the sample. Thenarrow (11 μm) depth of focus of the Nanoprobe instrumentwas utilized to obtain accurate and high-spatial resolution data

from the relatively thick samples studied here. Element-specificfocusing on individual particles within the sample allows unam-biguous determination of the depth of constituent parts within thesample. Preliminary test work, presented as Supporting Informa-tion, using these samples has shown that variation in the focaldepth provides the ability to separately identify particles locatedat different depths within the sample, proving that the resultspresented here are unlikely to be significantly affected by theoverlap of multiple particles in the z-direction. The sizes of thesquare regionsmapped varied from 5.6 to 9.6 μmand are noted inthe captions of the respective figures.

Results and Discussion

Fly Ash. Figures 2 and 3 show the elemental maps of K, Ca,Fe, and Cr in two different regions of the fly ash/epoxy sample.Both regions show the presence of multiple fly ash particles, withsignificant differences in composition, and morphologies that areapparently close to spherical. Hollow particles (Cenospheres) arecommonly observed in fly ashes but were not seen in any of thesamples (either fly ash/epoxy or alkali-activated fly ash) investi-gated here. A cenosphere-type morphology would be expected toresult in a more uniform distribution of the fluorescence signalacross the particle, as the shells of these particles are generallyquite uniform in thickness.32 It is not possible to tell from the dataobtained here whether the particles are solid spheres or plero-spheres (Cenospheres filled with smaller particles), although therelative uniformity of the regions in the centers of the particlessuggests that they are most likely solid. The morphology of theiron-rich areas within the particles, i.e., whether the iron contentis uniform throughout the particle or present in a dendritic

Figure 2. Elementalmaps of (a) calcium, (b) chromium, (c) iron, and (d) potassium in a 5.6μm� 5.6 μmregion (71 points� 71 points) of thefly ash/epoxy sample.Data are presented in units of total fluorescence counts per point in the spectral region of interest, and the scale used foreach element is shown separately.

(32) Ngu, L.-N.; Wu, H.; Zhang, D.-K. Energy Fuels 2007, 21, 3437.

11900 DOI: 10.1021/la901560h Langmuir 2009, 25(19), 11897–11904

Article Provis et al.

morphology intermixed with another phase, cannot be deter-mined from these data because the hard X-ray beam used has adepth of focus of several micrometers as mentioned above. Thecomposition presented at each point is therefore an averagedetermined across an interaction volume which is much deeperthan it is wide, which precludes identification of the dendriticstructures sometimes observed in fly ash particles by back-scattered electron imaging.33

Figures 2 and 3 each show part of an ∼10 μm fly ash particlecontaining all the Ca, Cr, and Fe. Such particles are well-known tocomprise a significant fraction of most Class F fly ashes, andsubstitution of Ca2þ and Cr3þ onto the Fe2þ/3þ sites in the ferritespinel phase is expected. The main Fe-rich particle in Figure 2 ispoor inK,while theparticle inFigure 3 is rich inK.Such variabilityin alkali content between particles is also well-known in fly ashes.13

There is no apparent surface enrichment of any of the elementsstudied; in a cross section takenacross the scan region starting fromthe point of highest Fe intensity in Figure 3, the fluorescenceintensities of all elements decrease according to the same trend(to within experimental uncertainty). It is also of interest thatthe chromium and iron sites seem to correspond very closelyto each other in both scans; this will be discussed in more detailbelow.

However, there are additional particles observed in each scanwhich are of particular interest in the context of the interactionsbetween fly ash and alkaline solutions to form geopolymers. Therole of iron in geopolymers has been the subject of some discus-

sion recently,22,23,26,34,35 with some studies showing the presenceof iron to have a negative impact on the formation of reactionproducts but others showing little or no effect. The localization ofiron into specific particles within the fly ash but not othersis therefore of potential importance, particularly given that theseother particles identified are sometimes very rich in eitherpotassium (top right of Figure 2d) or calcium (bottom left ofFigure 3a), and these elements have been identified as beingimportant in determining fly ash reactivity during geopolymerformation.36

Fly Ash Alkali Activation Products. To study the effect ofalkali activation on the different elements present within fly ash,the ash was reacted with a potassium hydroxide activating solutionand with a potassium silicate activating solution to form twogeopolymer samples, as outlined in Materials and Methods.Nanoprobe data for these two samples are shown in Figures 4and 5.

An important point to note in Figures 4 and 5 is that theconcentration of potassium is much higher in the geopolymer thanin the fly ash/epoxy sample, because of the high concentration ofpotassium in the activating solutions used. This means that theregions of newly formed geopolymer binder can be identified asbeing the regions which are enriched in potassium alone(i.e., potassium aluminosilicate). Solvated potassium present inopen pores, which is known to comprise some proportion ofthe potassium present in many geopolymers,37 will most likely beremoved during polishing of the samples to a thickness of muchless than a millimeter, and it is likely that this is responsible for the

Figure 3. Elemental maps of (a) calcium, (b) chromium, (c) iron, and (d) potassium in a 9.6 μm� 9.6 μm region (121 points� 121 points) ofthe fly ash/epoxy sample.

(33) Ramsden, A. R.; Shibaoka, M. Atmos. Environ. 1982, 16, 2191.(34) Fern�andez-Jim�enez, A.; Palomo, A.; Macphee, D. E.; Lachowski, E. E. J.

Am. Ceram. Soc. 2005, 88, 1122.(35) Keyte, L. M. Ph.D. Thesis, University ofMelbourne,Melbourne, Australia,

2008.

(36) Duxson, P.; Provis, J. L. J. Am. Ceram. Soc. 2008, 91, 3864.(37) Duxson, P.; Provis, J. L.; Lukey, G. C.; vanDeventer, J. S. J.; Separovic, F.;

Gan, Z. H. Ind. Eng. Chem. Res. 2006, 45, 9208.

DOI: 10.1021/la901560h 11901Langmuir 2009, 25(19), 11897–11904

Provis et al. Article

low-potassium region in the bottom right-hand corner ofFigure 5d.

Figure 4 can therefore be identified as depicting a large fly ashparticle centered beyond the bottom right-hand corner of the fieldof view, as shown by the Fe and Cr maps, with a region ofgeopolymer binder in the left-hand side of the image identifiableby its elevated potassium content. The calcium appears to havebecome significantly dispersed from its original location withinthe fly ash particle and has intermingled with the geopolymerbinder. In contrast, the iron appears to have remained essentiallyunaffected by the geopolymerization process and is restricted tothe area within the remnant fly ash particle. Similar observationshave been made using elemental mapping in an electron micro-scope, although with a spatial resolution approximately 10 timeslower.23While it is not currently possible to compare “before and

after” images of the exact same fly ash particle undergoinggeopolymerization to confirm that nomovement has taken place,the fact that the iron concentration falls away smoothly at theparticle edge suggests that it is unlikely that it has migratedsignificantly, if at all, into the geopolymer binder region.

Figure 5 also depicts the edge of an apparently large iron-richparticle at the top of the field of view, with a remnant calcium-richfly ash particle at the bottom left-hand corner and what appearsto be a pore in the right corner, as mentioned above. The calciumin this sample is also clearly more dispersed than the iron orchromium. Like the hydroxide-activated case in Figure 4, calciumappears to have been released from the fly ash particle andbecome somehow incorporated into the binder phase. Exactlywhat structural role is played by the calcium in the geopolymerbinder is not yet well understood, but it is believed to play an

Figure 4. Elementalmaps of (a) calcium, (b) chromium, (c) iron, and (d) potassium in a 5.6μm� 5.6 μmregion (71 points� 71 points) of thefly ash/KOH sample. (e) Expanded view of part of panel a (in the geopolymer binder region, bottom left corner), showing localized regions ofhigh Ca concentrations (circled), with color scaling changed to highlight this.

11902 DOI: 10.1021/la901560h Langmuir 2009, 25(19), 11897–11904

Article Provis et al.

important role in determining both the kinetics of geopolymerformation and the performance of the final product.23,38,39

However, the difference in binder homogeneity between thehydroxide-activated and silicate-activated samples can be clearlyobserved by comparing Figures 4 and 5, and this provides acritical demonstration of the unique capabilities of the Nano-probe instrument. Panels a and d of Figure 4 in particular show adistribution of calcium and potassium, respectively, that is highlyinhomogeneous on a length scale of tens to hundreds of nano-meters, while the corresponding panels a and d of Figure 5 showmuch less local variability in composition. The narrow depth offocus of the Nanoprobe instrument, as discussed above, providesconfidence that these are in fact isolated particles and not artifactsdue to overlapping particles within a large probe volume.

This observation of inhomogeneous calcium distribution is inagreement with the known differences in aluminosilicate frame-work microstructure between silicate-activated and hyd-roxide-activated geopolymermaterials as observedunder electronmicroscopy.21,23,40,41 However, direct comparison of elementaldistributionswithin thesemicrostructures has not previously beenpresented, and in particular, little emphasis has been placed uponthe distribution of nonframework species as depicted here.

Figure 4a shows that there are very localized regions of highCacontent within the region of the sample that can be identified as

containing geopolymer binder. While some of these high-Caregions overlap with the fly ash particle visible in the Fe map(Figure 4c), theK concentration gradient suggests that the centralpart of the field of viewcontains both geopolymer and remnant flyash phases overlaid on each other,with both being sampled due tothe penetration depth of the hardX-rays.Considering the absenceof small discrete high-Ca regions within any other fly ash particlesobserved here, these can be confidently identified as being locatedwithin the geopolymer binder. The K is also somewhat inhomo-geneously distributed, but not to the same extent as Ca.

A physical interpretation of this may therefore be proposed:calcium released from the fly ash particles into the very highlyalkaline (pH>14) solution reacts rapidly with the hydroxide ionspresent, nucleating as particles of Ca(OH)2 which are thenenclosed by the binder as it forms. This has been proposed as anexplanation for the behavior and structure ofmetakaolin geopoly-mers containing calcium silicate minerals,38 but the formation ofsuch discrete nanoscale particles has not been observed beforenow. The relative rates of release of calcium, silicon, and alumi-num will also play some role here, with the formation of calcium(alumino)silicate hydrate and calcium aluminate hydrate phasesalso possible under some circumstances,42 but the chemistry of thehydroxide-activated system studied here suggests that Ca(OH)2 isthe phase most likely to form. No newly formed crystallinecalcium-containing phases are able to be identified in the sampleby X-ray diffraction, which may mean that the high-calciumprecipitates are crystallographically disordered or may simply bebecause they are small (a few tens of nanometers in size) andcomprise only a very minor fraction of the total sample volume.

Figure 5. Elemental maps of (a) calcium, (b) chromium, (c) iron, and (d) potassium in a 9.6 μm� 9.1 μm region (121 points� 115 points) ofthe fly ash/potassium silicate sample.

(38) Yip, C. K.; Lukey, G. C.; Provis, J. L.; van Deventer, J. S. J. Cem. Concr.Res. 2008, 38, 554.(39) Provis, J. L.; Yong, S. L.; Duxson, P.; van Deventer, J. S. J. In 3rd

International Symposium on Non-Traditional Cement and Concrete; Bı́lek, V.,Ker�sner, Z., Eds.; Brno University of Technology: Czech Republic, 2008;pp 589-597.(40) Steveson, M.; Sagoe-Crentsil, K. J. Mater. Sci. 2005, 40, 4247.(41) Fern�andez-Jim�enez, A.; Palomo, A. Cem. Concr. Res. 2005, 35, 1984.

(42) Yong, S. L. Ph.D. Thesis, University of Melbourne, Melbourne, Australia,2009 (in progress).

DOI: 10.1021/la901560h 11903Langmuir 2009, 25(19), 11897–11904

Provis et al. Article

By comparison, Figure 5a, consistent with the scans of otherregions of the silicate-activated sample (data not shown), does notshow localized high-Ca regions. Instead, the Ca appears to be“smeared” around the edges of the fly ash particles, suggestingthat it has been released from the particles and has been able todiffuse more uniformly into the surrounding region before thesystem hardened. It cannot be stated conclusively whether thisregion contains any discrete calcium silicate hydrate phases thathave separated from the alkali aluminosilicate geopolymer bin-der, but the homogeneity of the elemental distributions of bothCaand K in Figure 5 suggests that this has not occurred. Previouswork has shown the absence of such phase separation on a lengthscale of several micrometers in a similar system,23 and the resultspresented here enable a further reduction in the upper bound onthe size of any segregated calcium silicate hydrate regions in thismaterial to less than 100 nm. Given that X-ray photoelectronspectroscopic analysis ofmetakaolin-derived geopolymer systemsdoes indicate nanoscale phase segregation in these materials,39,42

but extensive X-ray diffractometry studies have not shown anyindication of crystalline hydrate formation, it appears increas-ingly likely that any segregation present is taking place on a lengthscale of at most a few nanometers.Disposition of Chromium in Fly Ash Geopolymers. As

outlined in Introduction, a key concern in the use of fly ash inconstruction materials is the potential availability of chromiumfrom the hardened materials. This is particularly the case forgeopolymers, where the fly ash content is much higher than inPortland cement-based materials. The identification of a strongcorrelation between unreacted fly ash particles and chromiumin the geopolymer samples studied here is therefore significantin determining whether these materials will meet widespreadapproval in themarket. The importance of this particular correla-tion is that if the chromium is restricted to the unreacted (andgenerally not very porous) ash particles embedded within thegeopolymer binder, it is much less likely to be accessible to thepore networkof the geopolymer and is therefore less susceptible toleaching.Also, if the chemical state of the chromiumwithin fly ashparticles is resistant to leaching under the highly alkaline condi-tions prevailing during geopolymerization, it is likely to be capableof resisting removal under less aggressive service conditions.

It has previously been observed27,43 that the level of availablechromium inAustralian fly ashes is in general low. Themaximumleaching takes place in alkaline environments, and more alkalineashes are usually more susceptible to Cr leaching. The Gladstonefly ash used here is considered quite an alkaline ash,44 meaningthat its exposure to highly alkaline conditions during geopoly-merization is likely to represent close to a worst-case scenario interms of release of Cr from an Australian fly ash. The Gladstonefly ash used here contains 0.0138 wt % (138 ppm) chromium onan oxide basis according to bulk X-ray fluorescence analysis(Table 1), but Narukawa et al.27 have observed that the concen-trations of total Cr and available Cr in Australian fly ashes showat most a weak correlation.

Figure 6 shows the correlation between Fe and Cr concentra-tions in the unreacted fly ash and in the two geopolymer samples.The data are normalized to remove background effects, with thelowest concentration of each element in each sample regionplotted as 0 and the highest as 1. The diagonal line on eachplot therefore represents an exact correspondence between Cr

concentrations and Fe concentrations. It is apparent fromFigure 6 that the correlation between the locations of thesetwo elements in all samples is very strong, particularly in theintermediate concentration region.

The data presented in Figure 6 were obtained from two to threeregions on each sample, the regions shown in Figures 2-5 andadditional regions on each sample whose full fluorescence mapsare not presented in this paper, but the trends across regionsappear to be very consistent. It is obviously somewhat perilous toattempt to represent a heterogeneous sample by analyzing a smallnumber of small regions, but this consistency in the observedtrends brings some measure of confidence in the analysis pre-sented here. Similar consistency was not always observed inthe correlations between other element pairs, in particular, the

Figure 6. Correlation between Fe and Cr concentrations in(a)Gladstone fly ash, (b) hydroxide-activated fly ash geopolymer,and (c) silicate-activated fly ash geopolymer. Different symbols ineach panel correspond to the different regions mapped in eachsample. The region shown in Figure 2 is region A2; Figure 3corresponds to region A3, Figure 4 to region B2, and Figure 5 toregion C2. Full elemental maps for the other regions (A1, B1, andC1) were collected but are not presented in this paper. Data werenormalized so that the highest concentration of each element ineach region is displayed as 1 and the lowest as 0.

(43) Jankowski, J.; Ward, C. R.; French, D. Preliminary assessment of traceelement mobilisation from Australian fly ashes, Research Report 45, CooperativeResearch Centre for Coal in Sustainable Development, 2004.(44) Pathan, S.M.; Aylmore, L. A. G.; Colmer, T. D. J. Environ. Qual. 2003, 32,

687.

11904 DOI: 10.1021/la901560h Langmuir 2009, 25(19), 11897–11904

Article Provis et al.

Fe-Ca and Fe-K pairs, reflecting the highly heterogeneousdistribution of Ca and K across the ash particles.

Interestingly, both panels a and b of Figure 6 (the fly ash andthe hydroxide-activated sample, respectively) appear to show anumber of data points falling in a region where the Fe concentra-tion is very low but the Cr concentration is significant (up to 25%of the highest Cr fluorescence intensity recorded), and thismay beidentified as being of some concern in terms of the chromiumbeingmobile. However, Figure 6c does not show any such points.Thismay simply be due to having selected twomapping regions inthe silicate-activated sample which happened not to contain anyregions of excess chromium content, or it may be due to the lessaggressive silicate activating solution used in this instance. Giventhat such Cr-enriched regions were observed in the ash itself andthat the chemistry of the other samples tested was consistentacross the different regions analyzed, the former explanationseems more likely, but further work is necessary to providedecisive evidence one way or the other.

However, regardless of the exact situation in the silicate-activated sample presented in Figure 6c, some important informa-tion can be obtained from Figure 6. In particular, it is observedthat even a geopolymer formed by activation of an alkaline fly ashby a hydroxide solution, which should be considered close to aworst-case scenario as far as chromium release is concerned, doesnot contain significantly more mobile chromium (defined here aschromium which is not associated closely with iron) than theoriginal fly ash. Given that the use of fly ash activated by an alkalihydroxide solution is becoming increasingly widespread in variousapplications in Europe and elsewhere,45-47 this observation is ofsome significance. The data presented here provide some insightinto this behavior, but it would be necessary to confirm theseobservations by a technique which is sensitive to Cr oxidationstates, such as spatially resolved X-ray absorption near-edgespectroscopy (μ-XANES), to fully confirm these suggestions.

However, it should also be noted that many commercialincarnations of geopolymer concrete incorporate at least amoderate level of blast furnace slag, which accelerates the settingprocess and also gives added performance benefits in the finalhardened product.10,36,38 Blast furnace slag contains sufficientsulfide to generate a reducing environment, which is capable ofreducing mobile Cr(VI) to relatively immobile Cr(III). It hasrecently been shown that the addition of 0.5 wt % Na2S(simulating the reducing environment generated by slag) to a flyash geopolymermatrix containing elevated levels ofCr(VI) addedasNa2CrO4 reduces the leaching rate of chromiumverymarkedlyunder both acidic and alkaline conditions.48,49 It may thereforebe expected that, even if Cr is released from the fly ash intothe binder region of these materials, its reduction to Cr(III)and consequent immobilization will minimize the potential for

harm to the biosphere during the use of slag-containing geo-polymer concrete.

Conclusions

Elemental mapping of fly ash and fly ash-derived geopolymersusing hard X-ray fluorescence mapping with nominal 30 nmspatial resolution has been shown to provide valuable informationregarding the structures of these materials. The calcium distribu-tion in a hydroxide-activated fly ash geopolymer gel is highlyheterogeneous, with discrete high-calciumparticles observedwith-in the binder structure. The exact chemical identity of these regionscannot be confirmed; however, the fact that they are at most a fewtens of nanometers in size and highly enriched in calcium suggeststhat they may be Ca(OH)2 precipitates which nucleate as Ca2þ isreleased from the fly ash particles into the activating solutionwhich contains a very high concentration of free OH-. Theexistence of such structures within geopolymers has previouslybeen the subject of much speculation, but this work provides thefirst direct evidence confirming their existence. Calcium in thesilicate-activated fly ash geopolymer is much more dispersedwithin the gel structure andappears to have spreadmore graduallyfrom the fly ash particles during geopolymer formation. Calciumis also seen to have been released in preference to iron from the flyash particles which are rich in both elements.

The locations of chromium and iron are very closely correlatedwithin the structures of both fly ash and the geopolymer product,suggesting that the chromium is present as Cr(III) within a ferritespinel phase which is relatively resistant to alkali attack. This isimportant because chromiumcontamination is a potential issue inthe valorization of fly ash, and the data indicate that the level ofreadily available and toxic chromium in a fly ash-derived geopo-lymer will be significantly lower than the total chromium contentof the precursor fly ash.

Acknowledgment. Use of the Center for Nanoscale Materialswas supported by theU.S.Department of Energy,Office of Science,Office of Basic Energy Sciences, under Contract DE-AC02-06CH11357. Use of the Advanced Photon Source was supportedby theU.S.Department of Energy,Office of Science,Office of BasicEnergy Sciences, under ContractDE-AC02-06CH11357. This workwas funded by the Australian Research Council (ARC), includingpartial funding from the Particulate Fluids Processing Centre, aSpecialResearchCentreof theARC,and throughDiscoveryProjectgrants. Thework of S.A.B.was supported by travelling scholarshipsfrom Colciencias and from the Walter Mangold Trust. Travelfunding for J.L.P. was supplied by the Australian SynchrotronResearchProgram.We thankDr. J€orgMaser,Dr.RobertWinarski,Dr. Martin Holt, and Ms. Claire White for assistance with experi-ments on the Nanoprobe instrument.

Supporting Information Available: An example of one ofthe spectra obtained from the Vortex detector elements withthe ROIs identified and a more detailed discussion andvalidation of the probe volume of the Nanoprobe instru-ment. Thismaterial is available free of charge via the Internetat http://pubs.acs.org.

(45) Kovalchuk, G.; Fern�andez-Jim�enez, A.; Palomo, A. Fuel 2007, 86, 315.(46) Palomo, A.; Fern�andez-Jim�enez, A.; L�opez-Hombrados, C.; Lleyda, J. L.

Rev. Ing. Constr. 2007, 22, 75.(47) Fern�andez-Jim�enez, A.; Palomo, A.; Pastor, J. Y.; Martı́n, A. J. Am.

Ceram. Soc. 2008, 91, 3308.(48) Zhang, J.; Provis, J. L.; Feng, D.; van Deventer, J. S. J. J. Hazard. Mater.

2008, 157, 587.(49) Zhang, J.; Provis, J. L.; Feng, D.; van Deventer, J. S. J. Cem. Concr. Res.

2008, 38, 681.