mTec

Keywords:Class F y ashGeopolymerX-ray diffraction29Si MAS NMR27Al MAS NMR

olo. Trep

compressive strength. Chabazite-Na and Al-rich chabazite-Na are identied as major crystalline phases

Korea g

rable mechanical properties with a considerable reduction of CO2emission [3]. Geopolymer is synthesized by alkali-activation ofan amorphous aluminosilicate material, and the term was rstlygiven to the alkali-activated metakaolin by Davidovits [4,5]. Typi-cal alkaline activators are sodium hydroxide (NaOH) or sodiumsilicate [6]. The metakaolin-based geopolymer reaction product isan aluminosilicate gel and often called, zeolitic precursor because

n of y ashes, thenderstoodd the amoaction pro

believed to be geopolymer [3,79].The inconsistent reactivity is an inevitable outcome of

material characteristics of y ashes such as in oxide compparticle size distribution, content of alkaline metals, content ofglass phase, which signicantly differ between y ashes. Accord-ingly, y ashes with different reactivities are expected to producediverse reaction products, resulting in different compressivestrengths [9]. The reactivity of coal y ash for geopolymerizationvaries a lot from source to source or even in different batches fromthe same source [9,10]. Such a complex heterogeneity of y ash

Corresponding author. Tel.: +82 52 217 2815; fax: +82 52 217 2819.E-mail address: ohjaeeun@unist.ac.kr (J.E. Oh).

Cement & Concrete Composites 50 (2014) 1626

Contents lists availab

re

evocean, and thus the re-utilization of y ashes is one of the top envi-ronmental issues [2].

Geopolymer has shown a superior potential as an environmen-tally alternative to portland cement since the binder gains compa-

Despite the importance of the alkali-activatioreaction mechanism and products are not well umainly because of the complexity of y ashes annature of geopolymer to X-ray although the rehttp://dx.doi.org/10.1016/j.cemconcomp.2013.10.0190958-9465/ 2014 Elsevier Ltd. All rights reserved.as yetrphousduct is

variousosition,5.29 million tons of y ash a year. Despite the large consumption ofy ash in cement and concrete industries, a considerable portion ofthe ash is still buried in landlls in South Korea [1]. The y ash dis-posal is potentially seriously harmful to the environment because asignicant level of leachable toxic trace elements (e.g., Cd, Cu, As,etc.) in y ashes may contaminate the water reservoir, soil or

than y ash, the alkali-activation of metakaolin has yielded morereliable results than that of y ash. However, the geopolymerproduction using coal y ash has gained an elevated interest fromconstruction elds because this can have a strong competitiveprice over portland cement in a market of concrete productiondue to the very low supply cost for y ash [3].1. Introduction

Coal-red power plants in Southin the high strength samples, supporting the favoring formation of ABC-6 family of zeolitic precursorsfor the higher mechanical strength while the CSH formation from the high CaO content (or crystallineCaO) is not a major source of the strength. The XRD analysis shows that the presence of amorphoushumps located at 2729 2h is not a sufcient indicator of geopolymeric gel formation. In the 29Si MASNMR, some portion of 108 ppm Q4(0Al) peak is not related to quartz, implying that this portion of Siatoms actively participate in geopolymerization. The 27Al MAS NMR spectra exhibit more conversionof Al(V) and Al(VI) aluminum atoms into Al(IV) units in the higher strength sample, which can be anindication of more geopolymeric reaction.

2014 Elsevier Ltd. All rights reserved.

enerate approximately

the product has many structural and chemical similarities withsynthesized zeolites [5].

Since metakaolin is chemically simpler and more amorphousAvailable online 6 March 2014 compositional and physical characteristics. The analysis conrms that differences in the network modi-fying element content, the amorphous phase content, and the particle size lead to large differences inCharacterization of geopolymers from codifferent Class F y ashes

Jae Eun Oh , Yubin Jun, Yeonung JeongSchool of Urban and Environmental Engineering, Ulsan National Institute of Science and

a r t i c l e i n f o

Article history:Received 15 June 2013Accepted 20 October 2013

a b s t r a c t

The alkali-activation technharmful disposal of y ashreaction products for two

Cement & Conc

journal homepage: www.elspositionally and physically

hnology, Ulsan Metropolitan City 689-798, Republic of Korea

gy of coal y ash is one of several potential solutions to minimize thehis study reports high-resolution characterization of the alkali-activatedresentative Korean Class F y ashes, which are signicantly different in

le at ScienceDirect

te Composites

ier .com/locate /cemconcomp

leading to difculties of quality control has been one of obstacles tocommercial implementation of alkali-activated y ash binders.

This study selected commercially representative two Class F yash sources in South Korea to explore the effects of the composi-tional and physical differences of the y ashes on alkali-activatedreaction products using powder X-ray diffraction (XRD), compres-

2. Experimental

The chemical compositions of the raw coal y ashes are given inTable 1, determined by X-ray uorescence spectrometer (Bruker S8Tiger). The equivalent elemental compositions for major elementswere graphically summarized in Fig. 1.

Although the use of sodium silicate as an activator generallyproduces much stronger matrices, it generally causes the reaction

Table 1Chemical composition of original y ashes.

Oxide Weight %

FA1 raw y ash (%) FA2 raw y ash (%)

SiO2 65.2 51.8Al2O3 23.9 20.0CaO 2.0 10.1Fe2O3 4.2 10.3BaO 0.2 0.2K2O 1.4 1.0MgO 0.7 2.0MnO 0.0 0.1TiO2 1.0 1.2P2O5 0.4 1.4Na2O 0.1 0.6SO3 0.3 0.9SrO 0.1 0.1ZrO2 0.1 0.1Tb4O7 0.0 0.2MoO3 0.3 0.0

Element FA1 raw FA2 raw

Fig. 2. Density distribution of particle size for y ashes with median sizes.

J.E. Oh et al. / Cement & Concrete Composites 50 (2014) 1626 17Two Class F coal y ashes (denoted FA1 and FA2) used in thepresent study were supplied from the largest coal-red powerplants, which are about 220 km apart each other, in South Korea.

Si (atomic %)

90 10

80 20

70 30sive strength testing, particle size distribution analysis, X-ray uo-rescence (XRF), solid-state magic-angle-spinning nuclear magneticresonance (MAS NMR) and synchrotron transmissive X-raymicroscopy.Al (atomic %) 2Ca + Na + K + 2Mg (atomic %)90 80 70 60

60 40

50 50

50

40 60

40

30 70

30

20 80

20

10 90

10

FA1 raw fly ash

FA2 raw fly ash

fly ash Fly ashSi 58.6% 40.9%Al 19.4% 15.7%Ca 3.9% 15.7%Fe 9.0% 18.7%Mg 0.5% 1.7%

Na+K 3.2% 2.3%2Ca+Na+K+2Mg 12.1% 37.3%

Si/Al (atomic ratio) 3.0 2.6 Ca/Si (atomic ratio) 0.1 0.4

Fig. 1. Triangular compositional plot for original y ashes used in this investigation.

Table 2Compressive strengths of hardened samples.

Flyash

Sample label (y ash-solution-temp.)

Solution/y ash (wt./wt.)

NaOH solution(M)

Curing temperature(C)

1-day Curing 28-day Curing

Strength(MPa)

Stdeva

(MPa)Strength(MPa)

Stdeva

(MPa)

FA1 FA1-5M-60C 0.6 5.0 60 1.3 0.3 8.7 2.2FA2 FA2-5M-60C 0.6 5.0 60 6.6 0.9 41.6 12

a Stdev: standard deviation of strength test results.products to be more amorphous to X-ray, which makes it difcultto analyze the XRD results [11,12]; thus, this study activated theashes using only NaOH solution with a relatively high curing tem-perature of 60 C to increase the crystallinity of the hardenedashes.

The y ashes were alkali-activated with 5M (denoted 5M inthe sample label) NaOH solution made from a reagent grade ofNaOH pellets (SigmaAldrich). The solution to solid binder weightratio (s/b) of samples was xed to 0.6 to attain suitable consistency

Atomic %In-let table of composition

for all the pastes. Three equivalent paste mixtures were cast into2.54 2.54-cm cylinder molds for axial compressive strengthtesting. The cast samples were cured at 60 C (denoted 60C inthe sample label) with a constant 99% relative humidity untiltested. Details of mixture proportions of the samples are presentedin Table 2 with compressive strength testing results, measured at28 days of curing.

X-ray diffraction (XRD) was carried out using a Rigakuhigh power X-ray diffractometer, employing Cu Ka radiation(k = 1.5418 ) with a 2h scanning range of 560 at a roomtemperature. The weight contents of crystalline and amorphous(glassy) phases of the y ashes were estimated by internalstandard quantitative XRD analysis. Accurately weighedcrystalline corundum (Al2O3) (NIST RMS 676a, crystalline alumina99.02% 1.11%) was added into each y ash powder at a concen-tration of 10.0% as an internal standard [13] to calculate the con-tent of glassy phase in each y ash.

Particle size distributions of the original y ashes were esti-mated by Sympatec HELOS laser diffraction particle size analyzerequipped with RODOS dispersing unit, and the result is shown inFig. 2.

MAS-NMR (solid-state magic-angle spinning nuclear magneticresonance) spectroscopic experiments were performed at KBSIDaegu Center in Korea using an AVANCE II 400 MHz Bruker NMRinstrument with magnetic eld strength of 9.4 T and ZrO2 4 mmrotors at ambient temperature. 29Si MAS-NMR spectra were mea-sured under conditions of proton decoupling with a 30 pulse, apulse repetition delay of 100 s for quantitative study and a spin-ning rate of 10 kHz; 27Al MAS-NMR spectra were obtained with a30 pulse, a repetition delay time of 3 s for quantitative analysisand a sample spinning rate of 14 kHz. The recorded chemical shiftspectra of 29Si and 27Al were referenced relative to tetramethylsil-ane (TMS) at 0.0 ppm and 1 M AlCl3(aq.) at 0.0 ppm respectively.

Transmissive synchrotron X-ray microscopic images were re-corded for the original y ashes and the 28-day cured hardenedsamples to observe the geopolymer reaction products with a high

0

10

20

30

40

50

60

FA1-5M-60C FA2-5M-60C

Com

pres

sive

Str

engt

h (M

Pa)

Labels of Hardened Samples

1-day 28-day

Fig. 3. Compressive strengths of alkali-activated y ashes at 28 days of curing. Errorbars indicate the standard deviation of test results of triplicate samples.

18 J.E. Oh et al. / Cement & Concrete Composites 50 (2014) 1626FA1-5M-60C, 1-day

FA1raw fly ashZeolite Na-A (LTA) (01-089-8015)

Quartz (01-085-0794)

Mullite (01-074-4143)

Hematite (01-087-1165)

10 20 30 40 50 60Position [2, ] [CuK]

5 5

(a)Fig. 4. Integrated powder X-ray diffraction patterns for original y ashes (FA1 raw y ash5M-60C, 1-day) and (b) 28-day cured alkali-activated FA1 sample (FA1-5M-60C, 28-day)to have the same signal intensity seen in the 1-day and 28-day cured samples.Quartz (01-085-0794)

Hematite (01-087-1165)

Mullite (01-074-4143)

Zeolite Na-A (LTA) (01-089-8015)

FA1-5M-60C, 28-day

FA1raw fly ash

10 20 30 40 50 60Position [2, ] [CuK]

(b)

) and alkali-activated samples for (a) 1-day cured alkali-activated FA1 sample (FA1-. The simulated reference diffraction patterns (e.g., quartz, zeolite NaNa) are scaled

resolution in the hard X-ray nano-imaging beamline 7C at PohangLight Source in South Korea [14]. Due to the transmissive nature,this beamline allows us to observe the inner structure of thegeopolymeric products. The beamline 7C employs a full-eld Zern-like phase-contrast microscopy using high ux photon energy of6.9 keV. The hard X-ray is monochromated via silicon (111) doublecrystals and beamed into a sample placed on a Kapton lm posi-tioned at the middle hole of a sample plate. The beam size wasapproximately 1.5 mm 0.7 mm (FWHM) at the sample location.The penetrating X-ray through a sample is projected onto CCDdetector with 4096 4096 pixels of 9 lm size. Unlike electron-based microscopes (e.g., scanning electron microscope or transmis-sion electron microscope), the nano-imaging beamline 7C does notrequire any specic sample preparations (e.g., polishing or

thin-sectioning) or any necessary experimental conditions (e.g.,vacuum), and thus the original morphology of specimen can besafely observed without any sample destruction. For the samplepreparation, a few activated ne particles were collected from afractured surface of 28-day-cured sample and placed on a thinKapton lm with 25-lm thickness. To avoid any possiblecarbonation or other interaction with air, the hardened samplewas fractured on site at the beamline and the sample preparationwas taken only less than 5 minutes.

3. Results and discussion

In the XRD analysis, each phase was rstly identied usingXpert HighScore Plus software [15] with ICDD PDF-2 database[16]. The analysis results were corrected by manually scrutinizing:(1) reection positions; (2) peak intensities; and (3) reectionshapes for all XRD patterns.

Quartz, mullite, and hematite are found in common in theashes, but the crystalline CaO is seen only in the y ash FA2 (seeFig. 4).

The weight contents of glass and crystalline phases in each ashwere semi-quantitatively calculated by the normalized RIR (Refer-ence Intensity Ratio) method [13,17] with a help of the software.Because the RIR method is known to be relatively less accurate,in order to increase the RIR quantication accuracy, we used 15different ICDD patterns for each phase (e.g., quartz and mullite)with the highest matching scores, which are high only when reec-tion prole characteristics (i.e., peak positions, peak intensities andpeak shapes) are well matched between a referred ICDD pattern

Table 3Mineralogical composition of original y ashes.

Phase Content (%)

FA1 FA2

Quartz 13.5 9.9Mullite 13.9 7.8Hematite 1.2 1.2Crystalline CaO 0.0 0.3Glass 71.4 80.8Total sum 100.0 100.0

Note: referred ICDD PDF-2 card # for phase analysis: 01-085-0794, 01-083-0539 forquartz; 01-074-4143, 01-084-1205 for mullite; 01-087-1165, 01-079-1741 forhematite; 01-077-2010 for crystalline CaO; 01-082-1399 for corundum (note: theweight of corundum is excluded in table).

J.E. Oh et al. / Cement & Concrete Composites 50 (2014) 1626 19FA2-5M-60C, 1-day

FA2raw fly ash

Chabazite-Na (00-019-1178)C-S-H (00-033-0306)

Quartz (01-083-0539)

Mullite (01-084-1205)

Hematite (01-079-1741)

10 20 30 40 50 60Position [2, ] [CuK]

5

CaO (01-077-2010): shown only in FA2 raw fly ash

Al-rich Chabazite-Na (00-044-0248)

Quartz (01-083-0539)

5

(a)Fig. 5. Integrated powder X-ray diffraction patterns for original y ashes (FA2 raw y ash5M-60C, 1-day) and (b) 28-day cured alkali-activated FA2 sample (FA2-5M-60C, 28-day).have the same signal intensity seen in the 1-day and 28-day cured samples.Hematite (01-079-1741)

Mullite (01-084-1205)

10 20 30 40 50 60Position [2, ] [CuK]

(b)Al-rich Chabazite-Na (00-044-0248)

Chabazite-Na (00-019-1178)

FA2-5M-60C, 28-day

FA2raw fly ash

C-S-H (00-033-0306)

CaO (01-077-2010): shown only in FA2 raw fly ash ) and alkali-activated samples for (a) 1-day cured alkali-activated FA2 sample (FA2-The simulated reference diffraction patterns (e.g., quartz, chabazite-Na) are scaled to

ete20 J.E. Oh et al. / Cement & Concrand a measured XRD pattern. For instance, for the original y ashFA2, the use of 15 ICDD patterns produces the average content%with a standard deviation for quartz, 9.9% 0.1%; for mullite,7.8% 0.3%; consequently for glass phase, 80.8 0.7%. The resultfor the quantication is shown in Table 3.

The compressive strength testing results for the 1-day and 28-day cured samples are summarized in Table 2 and Fig. 3. The hard-ened samples made from the ash FA2 show about ve times higheraveraged mechanical strength than the samples of the ash FA1 atall the curing periods, displaying the higher reactivity, arising fromthe higher content of network modifying elements (Na, K, Ca, andMg), the smaller mean particle size, and the larger glass phase con-tent in the original y ash FA2 [6,8] (see Figs. 1 and 2 and Table 3).

The powder X-ray diffraction patterns taken after curing for 1and 28 days are given in Figs. 4 and 5 for the activated FA1 andFA2 samples respectively. The diagrams include the simulated X-ray diffraction patterns for the identied mineral phases fromthe ICDD PDF-2 database. The intensity scales of the simulated pat-terns are adjusted to have the same intensities appearing in themeasured X-ray diffraction patterns. Note that the X-ray diffrac-tion patterns of the samples can be constructed by superimposingthe simulated reference patterns.

Fig. 6. Evolution of diffraction patterns with curing time. The intensities of diffractioncomparison between patterns.Composites 50 (2014) 1626In Fig. 5(a), the crystalline CaO peaks disappear by hydration(see mark A in gure), and the small CSH peaks are formedwithin one day. Afterward, no change of the CSH peaks isobserved until 28 days, implying the CSH formation is not a ma-jor source of strength development after the rst day.

Major crystalline phases formed in the activated samples of thisstudy are zeolite Na-A(LTA) (Na12Al12Si12O4827H2O [18]), Al-richchabazite-Na (NaAlSiO4xH2O [19]) and chabazite-Na (=herschelite,Na4Al4Si8O2412H2O [20]).

In Fig. 5, the low angle peaks appearing at 9 2h (marked withB and C) are the reection peaks of (101) planes of chabazite-Naand Al-rich chabazite-Na, but these are much smaller (or evenbarely seen) than the simulated reference patterns. This is proba-bly because the development of long-range ordered structure ishindered due to the low solution/binder ratio and the high NaOHconcentration, resulting in a larger number of reaction nuclei witha smaller space for crystallization [3]. Also, the broadness of thediffraction peaks at 9 2h indicates the small crystalline sizes ofreaction products [13].

Provis et al. [21] suggested that the typical broad humps near at2729 2h in X-ray diffraction patterns of geopolymers are primar-ily resulted from the extensive presence of nano-crystal size

patterns are adjusted to have the same intensity scale and overlapped for a direct

-reteppm-130-120-110-100-90-80-70

No.12345678

FA1 raw fly ash

FA1-5M-28day-60C No. Pos.(ppm) %Area Qn1 -111.3 9 Q4(0Al)2 -107.4 21 Q4(0Al)3 -102.8 9 Q4(1Al)4 -98.6 25 Q4(2Al)5 -95.1 4 Q4(3Al)6 -92.5 10 Q4(3Al)7 -89.5 4 Q4(4Al)8 -86.6 18 Q4(4Al)

(a)

J.E. Oh et al. / Cement & Conczeolites with no more than four unit cells, corresponding to810 nm. Given that the nano-sized zeolites in geopolymericgel are closely related to crystalline zeolite phases observed inX-ray diffraction [3,1012,2123], careful inspection of the crystal-line zeolites may provide us useful insights on the geopolymericreaction products for y ashes [12].

Oh et al. claimed [12,22] that the geopolymeric gel is structur-ally associated with a specic type of zeolite framework, which iscalled ABC-6 family consisting of 6-membered rings. Typicalzeolites (or zeolitic material) possessing 6-membered rings arenepheline, chabazite-Na (=herschelite), hydroxysodalite andhydroxycancrinite. These zeolites have been often reported in alka-li-activated class F y ash samples under a relatively high NaOHconcentration (>5 M) [3], whereas use of lower concentration ofNaOH (0.53 M) solution generally produces much lower strengthof matrices [6] with non-ABC-6 family zeolites such as zeoliteNa-P1 or zeolite Na-A.

In this study, note that the high strength samples (i.e., FA2-5M-60C) contain chabazite-Na and Al-rich chabazite-Na, whichbelong to ABC-6 family of zeolite framework while the weaksamples (FA1-5M-60C) show only zeolite Na-A.

Fig. 6 illustrates that most of changes in the diffraction patternsoccur during the rst day for both FA1 and FA2. In Fig. 6(a and b),note that the elevation of the amorphous hump is noticeably low-ered in the region of 1030 in 2h, and the top point of the broadhump is shifted from 23 into 27 within one day. Earlier studieshave reported that the formation of the amorphous hump at 2729 2h is a strong evidence of aluminosilicate gel (or geopolymer)formation, thereby creating mechanical strength [3,12,21]. How-ever, in this study, the shift of the hump seems barely connected

ppm-130-120-110-100-90-80-70 -

(c)Fig. 7. Quantitatively deconvoluted 29Si MAS NMR spectra for original y ashes (a andspectra. Dashed lines are simulated deconvoluted spectra. In-let tables contain peak posppm-130-120-110-100-90-8070

FA2 raw fly ash No. Pos.(ppm) %Area Qn1 -120.2 14 Q4(0Al)2 -112.9 27 Q4(0Al)3 -108.3 36 Q4(0Al)4 -103.0 6 Q4(1Al)5 -98.7 10 Q4(2Al)6 -94.3 2 Q4(3Al)7 -91.1 4 Q4(4Al)8 -88.6 1 Q4(4Al)9 -86.1 0.5 Q4(4Al)

FA2-5M-28day-60C No. Pos.(ppm) %Area Qn1 -120.8 4 Q4(0Al)2 -110.4 4 Q4(0Al)3 -107.5 11 Q4(0Al)4 -103.6 21 Q4(1Al)5 -97.8 19 Q4(2Al)6 -92.6 18 Q4(3Al)7 -86.4 21 Q4(4Al)8 -80.3 2 Q19 -77.8 1 Q1 10 -75.6 0.5 Q1

(b)

Composites 50 (2014) 1626 21to a signicant formation of geopolymeric gel because the samplesat the rst day produced very low compressive strengths despitethe considerable shift. Besides, when the compressive strengthsare largely developed beyond one day to 28 days, only minorchanges are shown in the humps [see Fig. 6(c and d)]. Particularlyin Fig. 6(c), no change is seen during the signicant strength devel-opment from 1.3 MPa to 8.7 MPa. This indicates that other factorssuch as porosity, which cannot be seen in XRD, are likely to affectthe strength at a higher order. Thus, the typical change of the dif-fraction pattern (e.g., amorphous hump formation 2729 2h) maynot be a sufcient condition of geopolymeric gel formation formechanical strength development.

The solid-state MAS NMR spectra of 29Si are given in Fig. 7 forthe original y ashes and the activated ones with deconvolutiondata as in-let tables. The spectra were semi-quantitatively decon-voluted with the help of DMt software package [24]. In this study,the Qn(mAl) notation (n = 04, m = 0n) is used to describe thechemical bonds in the local structure of the resonating Si nuclei[25], where the subscript n indicates the number of adjacent tetra-hedral SiO4 linked to a specied SiO4 tetrahedron, and the m im-plies the number of substitution of Al to the neighboring Sitetrahedra.

The Qn(mAl) assignment to the measured 29Si NMR spectra isperformed by referring to the earlier studies [3,10,19,2628]. Be-cause the NMR spectra of y ashes are highly complex, we rstlyset the well reported y ash 29Si NMR peak positions from the lit-erature, and then adjusted the peak areas for prole-tting for thespectra to increase the reliability of the result. Note that weattempted to assign the same Qn(mAl) notations for multiplepeaks [e.g., Q4(0Al) peaks in Fig. 7], which is probably unusual in

ppm-130-120-110-100-90-8070

(d)b) and alkali-activated y ashes (c and d). Solid bold lines indicate experimentalitions and integrated peak areas (%) of deconvoluted resonance peaks.

vated samples contain relatively strong and sharp Al(IV) peaks at60 ppm [3].

Fig. 10 allows us directly to compare the 27Al NMR spectra be-tween before- and after-activation. A substantial amount of Al(V)and Al(VI) units disappear after activation with a notable growthof Al(IV) units, indicating the participation of aluminum atoms inAl(V) and Al(VI) environments in geopolymerization and conver-sion into Al(IV) units [3]. Note that the activated FA2 sample showsmore distinct and stronger Al(IV) peak, indicative of further zeoliticprecursor formation, possibly leading to the higher strength, thanthe activated FA1 sample.

Synchrotron transmissive X-ray microscopic images were taken

Fig. 8. Direct comparison between before and after activation of y ashes in 29SiMAS NMR spectra. Change of integrated peak area of 108 ppm is highlighted ingures.

Table 4Averaged number of Si atoms connected to a given tetrahedral Al atom [=(Si/Al)NMR]in SiOAl linkage in zeolitic aluminosilicate framework.

Sample (Si/Al)NMR (Si/Al)XRF

FA1 raw y ash 4.9 3.0FA2 raw y ash 7.7 2.6FA1-5M-28day-60C 2.1 FA2-5M-28day-60C 2.0

Note: (Si/Al)NMR takes into account only Si and Al atoms in Q4(mAl) environments,whereas (Si/Al)XRF includes all Si and Al atoms in the original y ashes.

etethe literature; however, considering that y ashes are highlyheterogeneous mixtures of various phases and even the activatedashes are not fully reacted [29], the component phases mayproduce slightly different peak positions for the same Qn(mAl)environment.

As seen in Fig. 7, the proles of 29Si NMR spectra are quite sim-ilar between the original ashes, but after activation become muchdifferent each other. Particularly, a signicant change is observedin the activated FA2 [see Fig. 7(d)], displaying evenly-distributedpeaks of the Q4(mAl) environments (m = 04), compared to thatof the activated FA1 sample in Fig. 7(c).

Generally, 29Si peak at 108 ppm is assigned to unreactivequartz in the literature [10,26,30]; however, in this study, despitethe small XRD intensity changes for the quartz after activation(see Fig. 6), the integrated area of 108 ppm peaks are consider-ably reduced by 24% for FA1 and 25% for FA2 (see Fig. 8), which im-plies that at least 2425% of the 108 ppm peaks are not directlyrelated to quartz, but to more reactive Si atoms.

In a zeolitic aluminosilicate consisting of exclusively SiOAland SiOSi linkages [i.e., only Q4(mAl) environment], the ratioof Si to Al [=(Si/Al)NMR] can be calculated using the following for-mula [27,31]:

Si=AlNMR P4

m0 IQ4 mAlP4

m0 0:25 m IQ4 mAl

IQ4 0Al IQ4 1Al IQ4 2Al IQ4 3Al IQ4 4Al0:25IQ4 1Al 0:5IQ4 2Al 0:75IQ4 3Al IQ4 4Al

1

where IQ4 mAl denotes the integrated peak area of 29Si NMR signalof Q4(mAl) units. Note that this formula (1) cannot count the num-ber of Al atoms in AlOAl linkage [31].

Given that the geopolymer product of coal y ash is structurallyanalogous to zeolites as stated earlier and most Si atoms of thisstudy have tetrahedral Q4 units (=basic units of zeolitic frame-work), this formula may apply with caution to the Si atoms exist-ing in the original and the activated y ashes [30], providing anaveraged number of Si atoms connected to a given tetrahedral Alatom forming zeolitic framework as tabulated in Table 4.

For the original ashes, the (Si/Al)NMR values exhibit much higher(4.9 for FA1 and 7.7 for FA2) than the overall (Si/Al)XRF ratios (3.0for FA1 and 2.6 for FA2), implying that: (1) a sizable amount ofAl atoms in the original y ashes are not in the zeolitic SiOAl linkage, but exist in the AlOAl linkage [27]; and (2) theoriginal ash FA2 contains more amount of Al atoms in the AlOAl linkage than FA1, which may be related to the geopolymericreactivity

After activation, the (Si/Al)NMR values become much lower (2.1for FA1 and 2.0 for FA2), implying that a large portion of theoriginal Al atoms in the AlOAl linkage convert intozeolitic SiOAl linkage.

Additionally note that the (Si/Al)NMR values of the activatedsamples are very similar each other regardless of their mechanicalstrengths, disagreeing with the previous nding in [30], statingthat the (Si/Al)NMR of activated y ash is proportional to themechanical strength.

Fig. 9 shows the 27Al MAS NMR spectra for the original y ashesand the activated samples. Aluminosilicate materials generallycontain three types of coordinated aluminum atoms (4-, 5- and6-fold coordination by oxygen), denoted Al(IV), Al(V) and Al(VI)respectively. The chemical shift ranges are illustrated by the graycolored shades in Figs. 9 and 10 for Al(IV) at 52 to 68 ppm; for

22 J.E. Oh et al. / Cement & ConcrAl(V) at 26 to 38 ppm; and for Al(VI) at 6 to 15 ppm [25]. Inthe original y ashes, the Al(IV) and Al(V) peaks overlap at50 ppm, and the Al(VI) is shown at 0 ppm, whereas the acti-Composites 50 (2014) 1626for the original y ashes and the 28-day cured hardened samples asshown in Figs. 1113. Although the microstructures in the imagesdo not reveal noticeable differences between the activated samples,

50.40 ppm, [Al(IV)+Al(V)]

-2.89 ppm, Al(VI)

ppm-40-20080 60 40 20120 100

FA1 raw fly ash

Al(IV) Al(V)Al(VI)

ppm-40-20040 206080120 100

Al(IV) Al(V)Al(VI)

49.56 ppm, [Al(IV)+Al(V)]

FA2 raw fly ash

-0.47 ppm, Al(VI)

ppm-40-2002060 40100 80120

Al(IV) Al(V)Al(VI)

58.24 ppm, Al(IV)[partially, Al(V) included]

-2.18 ppm, Al(VI)

FA1-5M-28day-60C

ppm-20 -40080 60 40 20100120

Al(IV) Al(V)Al(VI)

59.45 ppm, Al(IV)

-0.63 ppm, Al(VI)

FA2-5M-28day-60C

(a) (b)

(c) (d)Fig. 9. 27Al MAS NMR spectra for original y ashes (a) FA1, (b) FA2 and alkali-activated y ashes (c) activated FA1 and (d) activated FA2. Light-dark shaded areas indicatestable ranges of Al(IV), Al(V) and Al(VI) environments.

Fig. 10. Direct comparison between before and after activation of y ashes in 27Al MAS NMR. Light-dark shaded areas indicate typical chemical shift ranges of Al(IV), Al(V)and Al(VI) environments.

J.E. Oh et al. / Cement & Concrete Composites 50 (2014) 1626 23

and similarmorphologies are often observed in SEM (scanning elec-tron microscope) images in the earlier studies [5,11,29], some in-sights can be gained for better understanding of geopolymerreaction. Firstly, the transmissive characteristic of the beamlineallows us to see clearly the inner structures of ceno-sphere andplero-sphere y ash particles without destruction (see Fig. 11).Secondly, it tells that such hollow particles (i.e., ceno- and plero-spheres) are difcult to be found after activation, indicating morereactivity in geopolymerization. Thirdly, it identies a partially dis-solved y ash particle with an intrusion of a reaction product,where the boundary line of the original y ash particle is partially

vanished. See that Fig. 13(a) clearly show the intrusion of the reac-tion product into the partially dissolved area in the y ash particle.

4. Conclusions

The present work characterizes the alkali-activation of twoClass F y ashes with different reactivity due to different contentsof network modifying elements (e.g., Ca, Na, K and Na), particlesizes, and weight proportions of glassy phase of y ashes. The fol-lowing conclusions are made by using XRD, compressive strength

Fig. 11. Transmissive synchrotron X-ray nano-imaging for original y ash (a) FA1 and (b) FA2 particles.

24 J.E. Oh et al. / Cement & Concrete Composites 50 (2014) 1626Fig. 12. Transmissive synchrotron X-ray nano-imaging for alkali-activated y ash FA1.

reteJ.E. Oh et al. / Cement & Conctest, particle size distribution analysis, XRF, 29Si MAS NMR, 27AlMAS NMR and synchrotron X-ray nano-imaging microscopy.

(1) The y ash FA2 with the higher contents of network modify-ing elements and glass phase and the smaller mean particlesize yielded a much higher mechanical strength, indicating ahigher reactivity than the ash FA1.

(2) Although the crystalline CaO peaks disappear and the smallCSH peaks are formed within the rst day in the alkali-activation of FA2, the CSH formation is not a major sourceof strength development.

(3) The XRD result shows that each activated y ash samplecontains different major crystalline phases (i.e., zeoliteA-Na for FA1; chabazite-Na and Al-rich chabazite-Na forFA2). Given that the zeolitic crystalline phases possibly rep-resent the nano-scale sized zeolitic precursors (i.e., geopoly-mer products) formed in activated samples, the activatedFA2 may consist of mostly ABC-6 family of zeolitic precur-sors, leading to increase of mechanical strength.

(4) The presence of amorphous humps locating at 2729 2hmay not be a sufcient indicator of geopolymeric gel forma-tion for mechanical strength development from the interpre-tation for changes in the XRD patterns after activation.

(5) The 29Si MAS NMR spectra for the activated samples illus-trate signicant differences between FA1 and FA2 inQ4(mAl) environments. The calculation of (Si/Al)NMRsuggests that a sizable amount of Al atoms in the originaly ashes are not in the zeolitic SiOAl linkage, but exist

Fig. 13. Transmissive synchrotron X-ray nanoComposites 50 (2014) 1626 25in the AlOAl linkage; however, after activation, a largeportion of the original Al atoms in the AlOAl linkageconvert into zeolitic SiOAl linkage.

(6) The result suggests that some portion of 108 ppm peak isnot related to quartz, implying that a large proportion of Siatoms existing in Q4(0Al) at 108 ppm actively participatein geopolymerization considering XRD and NMR.

(7) The 27Al MAS NMR spectra exhibit more conversion of Al(V)and Al(VI) aluminum atoms into Al(IV) units in FA2, whichmay be an indication of more geopolymeric reaction.

(8) The transmissive X-ray nano-imaging microscopy imagesshow that hollow y ash particles (e.g., ceno- andplero-spheres) are more reactive than solid particles ingeopolymerization.

Acknowledgments

This research was supported by Basic Science Research Programthrough the National Research Foundation of Korea (NRF) fundedby the Ministry of Education, Science and Technology (2011-0014407) and also by a grant from (Future Challenge Project orCreativity and Innovation Project) funded by UNIST(Ulsan NationalInstitute of Science and Technology) (1.120018.01). Sunha Kim atthe KBSI Daegu Center is gratefully acknowledged for solid-stateNMR experiments.

-imaging for alkali-activated y ash FA2.

References

[1] Wee J-H. A review on carbon dioxide capture and storage technology usingcoal y ash. Appl Energy 2013;106:14351.

[2] Sharma S, Fulekar M, Jayalakshmi C, Straub CP. Fly ash dynamics in soilwatersystems. Crit Rev Environ Sci Technol 1989;19(3):25175.

[3] Duxson P, Fernndez-Jimnez A, Provis J, Lukey G, Palomo A, van Deventer JSJ.Geopolymer technology: the current state of the art. J Mater Sci2007;42(9):291733.

[4] Davidovits J, Sawyer JL. Early high-strength mineral polymer. US Patent4,509,985; 1985.

[5] Palomo A, Grutzeck M, Blanco M. Alkali-activated y ashes: a cement for thefuture. Cem Concr Res 1999;29(8):13239.

[6] Pacheco-Torgal F, Castro-Gomes J, Jalali S. Alkali-activated binders: a review.Part 2. About materials and binders manufacture. Constr Build Mater2008;22(7):131522.

[7] Pacheco-Torgal F, Castro-Gomes J, Jalali S. Alkali-activated binders: a review:Part 1. Historical background, terminology, reaction mechanisms andhydration products. Constr Build Mater 2008;22(7):130514.

[8] Duxson P, Provis JL. Designing precursors for geopolymer cements. J Am CeramSoc 2008;91(12):38649.

[9] Van Deventer J, Provis J, Duxson P, Lukey G. Reaction mechanisms in thegeopolymeric conversion of inorganic waste to useful products. J Hazard Mater2007;139(3):50613.

[10] Fernndez-Jimnez A, Palomo A. Characterisation of y ashes. Potentialreactivity as alkaline cements. Fuel 2003;82(18):225965.

[11] Zhang B, MacKenzie KJD, Brown IWM. Crystalline phase formation inmetakaolinite geopolymers activated with NaOH and sodium silicate. JMater Sci 2009;44(17):466876.

[12] Oh JE, Monteiro PJM, Jun SS, Choi S, Clark SM. The evolution of strength andcrystalline phases for alkali-activated ground blast furnace slag and y ash-based geopolymers. Cem Concr Res 2010;40(2):18996.

[13] Madsen IC, Scarlett NVY. Quantitative phase analysis, powder diffraction:theory and practice. Royal Society of Chemistry; 2008.

[14] Lim J, Park SY, Huang JY, Han SM, Kim H-T. Large-eld high-contrast hard X-ray Zernike phase-contrast nano-imaging beamline at Pohang Light Source.Rev Sci Instrum 2013;84(1):0137071137074.

[15] XPert HighScore Plus. 3 ed. Almelo, Netherlands: PANalytical; 2012.[16] PDF-2 Database. In: Data ICfD, editor. Newton Square, PA; 2000.

[17] Chung FH. Quantitative interpretation of X-ray diffraction patterns ofmixtures. I. Matrix-ushing method for quantitative multicomponentanalysis. J Appl Crystallogr 1974;7(6):51925.

[18] Hasegawa K, Nishibori E, Takata M, Sakata M, Togashi N, Yu J, et al. Asynchrotron powder diffraction study of Na-LTA. Jpn J Appl Phys1999;38:658.

[19] Thrush KA, Kuznicki SM. Characterization of chabazite and chabazite-likezeolites of unusual composition. J Chem Soc, Faraday Trans1991;87(7):10315.

[20] Gude A, Sheppard R. Silica-rich chabazite from the Barstow Formation, SanBernardino County, Southern California. Am Mineral 1966;51:90915.

[21] Provis JL, Lukey GC, van Deventer JSJ. Do geopolymers actually containnanocrystalline zeolites? A reexamination of existing results. Chem Mater2005;17(12):307585.

[22] Oh JE, Clark SM, Monteiro PJM. Determination of the bulk modulusof hydroxycancrinite, a possible zeolitic precursor in geopolymers, byhigh-pressure synchrotron X-ray diffraction. Cem Concr Compos2011;33(10):10149.

[23] Roy DM. Alkali-activated cements opportunities and challenges. Cem ConcrRes 1999;29(2):24954.

[24] Massiot D, Fayon F, Capron M, King I, Le Calve S, Alonso B, et al. Modelling one-and two-dimensional solid-state NMR spectra. Magn Reson Chem2001;40(1):706.

[25] MacKenzie KJD, Smith ME. Multinuclear solid-state nuclear magneticresonance of inorganic materials. 1st ed. Pergamon; 2002.

[26] Fernndez-Jimenez A, De La Torre A, Palomo A, Lpez-Olmo G, Alonso M,Aranda M. Quantitative determination of phases in the alkali activation of yash. Part I. Potential ash reactivity. Fuel 2006;85(5):62534.

[27] Engelhardt G. Multinuclear solid-state NMR in silicate and zeolite chemistry.TrAC Trends Anal Chem 1989;8(9):34367.

[28] Klinowski J. Solid-state NMR studies of molecular sieve catalysts. Chem Rev1991;91(7):145979.

[29] Fernndez-Jimnez A, Palomo A, Criado M. Microstructure development ofalkali-activated y ash cement: a descriptive model. Cem Concr Res2005;35(6):12049.

[30] Kovalchuk G, Fernndez-Jimnez A, Palomo A. Alkali-activated y ash: effect ofthermal curing conditions on mechanical and microstructural developmentPart II. Fuel 2007;86(3):31522.

[31] Hunger M. Solid-state NMR spectroscopy. Zeolite Charact Catal 2010:65105.

26 J.E. Oh et al. / Cement & Concrete Composites 50 (2014) 1626

Characterization of geopolymers from compositionally and physically different Class F fly ashes1 Introduction2 Experimental3 Results and discussion4 ConclusionsAcknowledgmentsReferences