QT

Za

b

c

ARRAA

KGIKS

1

sfirIewpaaic

0h

Thermochimica Acta 565 (2013) 163– 171

Contents lists available at SciVerse ScienceDirect

Thermochimica Acta

jo ur nal ho me page: www.elsev ier .com/ locate / tca

uantitative kinetic and structural analysis of geopolymers. Part 2.hermodynamics of sodium silicate activation of metakaolin

uhua Zhanga,∗, John L. Provisb, Hao Wanga, Frank Bullena, Andrew Reidc

Faculty of Engineering and Surveying, University of Southern Queensland, Toowoomba, QLD 4350, AustraliaDepartment of Materials Science and Engineering, The University of Sheffield, Mappin St, Sheffield S1 3JD, United KingdomHalok Pty Ltd. West End, QLD 4101, Australia

a r t i c l e i n f o

rticle history:eceived 11 July 2012eceived in revised form 25 January 2013ccepted 26 January 2013vailable online 14 May 2013

eywords:eopolymer

sothermal conduction calorimetryineticsodium silicate

a b s t r a c t

The paper describes the outcomes of a study using isothermal conduction calorimetry (ICC) to character-ize the geopolymerization kinetics of metakaolin activated with sodium silicate. Two exothermic peaksare observed in the calorimetric curves for all systems reacting within the temperature range 20–40 ◦C.The peaks are assigned to the dissolution of metakaolin, and the formation of geopolymeric gels withdisordered structure, respectively. Compared with the use of NaOH solution to activate metakaolin, thepresence of soluble silicate in the activator hinders the reorganization of the local structure of geopoly-meric gels and also suppresses the formation of zeolites or zeolite precursors. The ICC data are used viaa thermochemical model to quantify the reaction kinetics of geopolymerization, by assuming that thegeopolymeric gels have an analcime-like local structure and taking into account the speciation of thesilicate monomers and dimers in the activator. Decreasing the modulus from 1.6 to 1.0 increases thefractional reaction extent from 0.12 to 0.26 after 72 h at 25 ◦C. When the modulus is 1.2, increasing thereaction temperature from 20 ◦C to 35 ◦C results in an improved reaction extent from 0.24 to 0.35. Therapid polymerization that occurs at 40 ◦C appears to hinder the further reaction of MK and consequentlyresults in a lower reaction extent than at 35 ◦C. Combined with the findings from previous analysis ofsystems where NaOH was used to activate MK, the concentration of available Na+ appears to have a

more pronounced influence on the extent of geopolymerization than temperature and the concentra-tion of soluble Si. The higher reaction extent of the solid precursor particles with the soluble Si from theactivator results in binders with more compact microstructure and higher mechanical strength. Consid-ering the longer-term utilization of geopolymers, the addition of soluble silicate to the activator willdelay the transformation of amorphous raw materials to locally ordered materials, potentially providing

icros

advantages in terms of m

. Introduction

The mechanism and kinetics of geopolymerization have beentudied by using isothermal conduction calorimetry (ICC) in therst part of this study [1]. With the assumption that geopolymericeaction products have an analcime-like local structure, the use ofCC enables reliable quantification of the thermochemical reactionxtent for the relatively simple case of activation of metakaolinith sodium hydroxide solution. However, in most geopolymerreparation and production, activators including dissolved silicatere preferred, as their products tend to develop higher strength

nd denser microstructures [2,3]. Compared to the alkali hydrox-de type activators, alkali silicate type activators are much moreomplex in terms of chemistry. The distribution of silicate species

∗ Corresponding author. Tel.: +61 7 46311335.E-mail address: zuhua.zhang2@usq.edu.au (Z. Zhang).

040-6031/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.tca.2013.01.040

tructural stability in service.© 2013 Elsevier B.V. All rights reserved.

varies in aqueous sodium silicate solutions according to the sil-ica concentration, and the concentration and nature of the alkalimetal(s) present [4].

The degree of polymerization of soluble silicates in highly alka-line solutions is usually low to moderate, mainly in the form ofmonomers, dimers and trimers, depending on the modulus Ms(the molar ratio of SiO2 to Na2O in the activator) [5]. The mod-ulus Ms will thus play a very important role in determining thekinetics and products of geopolymerization, since the distributionof poly-silicate anions in solution can affect the precipitation of sil-icates to a significant extent [6]. Variation in Ms has been foundto significantly affect the mechanical properties of geopolymers,with an optimum strength often observed at moderate values ofMs, between 1 and 2 in most cases, consistent with the develop-

ment of a compact microstructure [2,7–10]. The setting behaviorand rheological properties of geopolymer pastes are also largelydetermined by Ms [7,10–12], which is in part because the poly-merization state of soluble silicate affects the viscosity remarkably,

164 Z. Zhang et al. / Thermochimica Acta 565 (2013) 163– 171

Table 1Mixture composition and reaction temperature of geopolymerization systems for ICC measurement.

Ms MK (Al2O3·2SiO2) (mmol) Na2O·nSiO2 (mmol)a H2O (mmol) T (◦C)

1.0 8.15 6.84 83.4 20, 25, 30, 35, 401.2 8.15 6.20 83.1 20, 25, 30, 35, 401.4 8.15 5.68 82.9 20, 25, 30, 35, 40

aded M

pisatrp

mmatTm

2

2

haa4

spi9pcN(vt

2

AadrfstbdsTt

2

r

1.6 8.15 5.24

a In each mix, n is the modulus of activator, i.e., the value given in the column he

articularly at the high concentrations (up to 35–45%) often usedn geopolymer synthesis [13]. These experimental results havehown the need to develop a better understanding of the kineticsnd mechanisms involved in the transformation from precursorso geopolymeric gels, so as to gain a better understanding of theelationship between the activator and the properties of the finalroducts.

The primary aim of this paper is to extend the study of geopoly-erization kinetics by the ICC method to sodium silicate-activatedetakaolin systems. The influence of the effects of soluble silicate in

ctivating solution and reaction temperature on the extent of reac-ion of metakaolin will be studied as a function of reaction time.he kinetic quantification results will be further linked with theechanical and microstructural properties of resultant binders.

. Experimental

.1. Materials

The metakaolin (MK) used, and the process of its production,ave been described elsewhere [1]. Briefly, the material possesses

specific surface area of 14.1 m2/g as determined by N2 sorptionnd the Brunauer–Emmett–Teller (BET) method, 55.57 wt.% SiO2,1.55 wt.% Al2O3, and 0.91% loss on ignition at 1000 ◦C.

To study the effects of the addition of soluble silicate, the sodiumilicate activating solutions with Ms of 1.0, 1.2, 1.4 and 1.6 were pre-ared by adding chemical grade NaOH pellets and distilled water

nto a commercial sodium water glass (supplied with Ms = 3.33,.28 wt.% Na2O, 29.91 wt.% SiO2). The mass concentration of therepared activators, defined in terms of Na2O + SiO2, was keptonstant at 35%, and correspondingly the molar concentration ofa2O·nSiO2 was in the range 4.27 mol/L (at Ms = 1.0) to 3.27 mol/L

at Ms = 1.6), where n is equivalent to Ms. The sodium silicate acti-ating solutions were prepared at least 24 h before use to ensurehat they had cooled to room temperature.

.2. Calorimetry

The ICC measurements were performed in a 3114/3236 TAM 83ir isothermal conduction calorimeter (Thermometric AB) usingn internal mixing procedure [1]. It should be noted that the pow-er and liquid (loaded in an injector) were held separately in theeaction container in the calorimeter at the reaction temperatureor 4–6 h to equilibrate prior to the experiment. Preliminary testshowed that when the liquid/solid ratio was higher than 0.6 mL/g,he mix could be mixed easily. Thus, to ensure successful mixingy the mini-blender in the instrument, the pastes studied wereesigned with a liquid/solid ratio of 0.8 mL/g. The molar compo-itions of the mixtures, and reaction temperatures, are provided inable 1. The heat evolution rate data were calibrated by subtractinghe heat evolution of MK mixed with water as a baseline [1].

.3. Characterization and mechanical testing of geopolymers

To analyze the Si and Al environments before and aftereaction, FTIR spectra were collected using a Nexus 670 FTIR

82.8 20, 25, 30, 35, 40

s.

spectrometer (Nicolet) in transmittance mode using the KBr pel-let technique, with a resolution of 2 cm−1. The spectra in the range1300–800 cm−1 were subjected to deconvolution analysis using thePeakfit software (Version 4.12) as described in [1]. XRD analysisof the hardened binders was conducted using an ARL 9900 SeriesX-ray workstation with Co K� radiation, operated at 40 kV and40 mA at a scanning speed of 2.4◦/min. A JEOL scanning electronmicroscope (SEM) was used to analyze the morphology of fracturesurfaces of selected specimens, which were sprayed with a thinAu coating to enhance conductivity, and imaged at an acceleratingvoltage of 15 kV.

To observe the effects of soluble silicate on the mechanicalproperties of geopolymers, cylindrical specimens with size of Ø25 × 37.5 mm were prepared at a liquid/solid ratio of 0.65 mL/g andcured at 25 ± 2 ◦C for 24 and 72 h, for compressive strength testing.Six replicate specimens of each mix were tested by using a WHY-200 Auto Compressive Resistant tester (Shanghai Hualong, China)at a loading speed of 0.5 mm/min.

3. Results

3.1. ICC testing

Fig. 1 shows the heat evolution rates of MK activated withsodium silicate solutions at 20–40 ◦C, in the first 72 h of reaction.Two distinguishable peaks are observed in all of the calorimetrycurves during the recording period. A first sharp peak (denotedI) appears immediately after mixing, followed by a second broadexothermic peak (denoted II). Increasing the liquid/solid ratioseems not to change the number of exothermic peaks [14,15],which indicates that the basic geopolymerization paths are con-stant as liquid/solid ratio varies.

The maximum heat evolution rate in peak I increases signifi-cantly as the reaction temperature increases. At 20 ◦C, all the curvesshow a maximum rate of around 24 mW/g MK, and no significantdifference in heat evolution rate is found between the four systems.As the temperature increases, the differences between the systemswith different Ms become more notable; at 40 ◦C, the heat evolu-tion rates are 45 and 70 mW/g MK for Ms = 1.2 and 1.6 respectively.As Ms increases (i.e., as alkali concentration decreases in the sam-ple formulations used here), the heat evolution rates in both peaksI and II decrease consistently, at each temperature above 20 ◦C.This indicates that more rapid reactions occur in these two stageswhen the activator alkali concentration is higher. Peak II (some-times overlapping with peak I) also indicates the reaction duration.At a given temperature, the heat evolution duration is prolongedas Ms increases, particularly for the system with Ms = 1.6 at 20, 25and 30 ◦C. This means that introducing more soluble silicate in theactivator, or decreasing the amount of alkali cations available, willmake the reaction slower in the later stages.

The cumulative heat released by all systems in these two stagesis summarized in Table 2. The total heat release in each case is

in the range 230–490 J/g MK for the first 72 h, which is consistentwith the results obtained by Rahier et al. for comparable systemsusing differential scanning calorimetry (DSC) [16]. As tempera-ture increases, the total heat release generally increases, except

Z. Zhang et al. / Thermochimica Acta 565 (2013) 163– 171 165

0 12 24 36 48 60 720

5

10

15

20

25a b

c d

e

0.0 0.2 0.4 0.6 0.80

5

10

15

20

25

1.61.4

1.2

Reaction time (h)

Hea

t ev

olu

tio

n (

mW

/g M

K)

1.0

Peak I

1.61.4

1.2

Peak II

Reaction time (h)

Hea

t ev

olu

tio

n (

mW

/g M

K)

Ms=1. 0

Ms=1. 2

Ms=1. 4

Ms=1.6Peak I

1.0

T=20oC

0 12 24 36 48 60 720

5

10

15

20

25

30

35

1.61.4

1.2

Peak II

Reaction time (h)

Hea

t ev

olu

tio

n (

mW

/g M

K)

Ms=1. 0

Ms=1.2

Ms=1.4

Ms=1.6

Peak I

1.0

T=25oC

0 12 24 36 48 60 720

10

20

30

40

0.0 0.2 0.4 0.6 0.80

15

30

45

1.6

1.4

1.2

Reaction time (h)

Heat

ev

olu

tio

n (

mW

/g M

K)

1.0

Peak I

1.6

1.4

1.2 Peak II

Reaction ti me (h)

Heat

evo

luti

on

(m

W/g

MK

)

Ms=1. 0

Ms=1. 2

Ms=1. 4

Ms=1. 6

Peak I

1.0

T=30oC

0 12 24 36 48 60 720

10

20

30

40

50

1.6

1.4

1.2Peak II

Reaction time (h)

Hea

t ev

olu

tio

n (

mW

/g M

K)

Ms=1. 0

Ms=1.2

Ms=1. 4

Ms=1. 6

Peak I

1.0

T=3 5oC

0 12 24 36 48 60 720

10

20

30

40

50

60

70

0.0 0.2 0. 4 0.6 0. 80

20

40

60

80

1.6

1.4

1.2

React ion time (h )

He

at

ev

olu

tio

n (

mW

/g M

K)

1.0

Peak I

1.6

1.4

1.2

Peak II

ction

Hea

t ev

olu

tio

n (

mW

/g M

K)

Ms=1.0

Ms=1.2

Ms=1.4

Ms=1.6

Peak I

1.0

T=40oC

tion r

wipho

Rea

Fig. 1. Effects of sodium silicate modulus and reaction temperature on heat evolu

hen Ms = 1.0, where the slow initial dissolution at 20 ◦C results

n a higher total heat release than at 30 ◦C and 40 ◦C. A similarhenomenon, where slow reaction at low temperature results inigher heat release, was also noted in sodium hydroxide activationf metakaolin [1].

ti me (h )

ate of MK geopolymerization: (a) 20 ◦C, (b) 25 ◦C, (c) 30 ◦C, (d) 35 ◦C and (e) 40 ◦C.

The variation in total heat release is mainly due to the differ-

ences in the dissolution extent of metakaolin, and the differentpolymerization reactions which take place as a function of Si/Alratio. The reaction process involved in the dissolution of metakaolinis complicated, involving several basic steps, including absorption

166 Z. Zhang et al. / Thermochimica Acta 565 (2013) 163– 171

Table 2Cumulative heat release in the reaction systems, J/g MK.

Ms 20 ◦C 25 ◦C 30 ◦C 35 ◦C 40 ◦C

1.0 387.1 392.0 369.6 486.4 325.81.2 332.5 377.0 382.2 470.6 389.6

o(srbmt

A

((SieogeTcspdf

afudaofimlpao‘ap[tfppbvsi[

3

otf

10 20 30 40 50 60 70 80

Ms=1.6

Ms=1.4

Ms=1.2

Ms=1.0

2 (o)

hall oysit equartz

kaoli nit e

MK

10 20 30 40 50 60 70 80

Ms=1.0

Ms=1.2

Ms=1.4

Ms=1.6quartzhall oysit ekaoli nit e

2 (o)

a

b

Fig. 2. Co K� radiation XRD patterns of MK and geopolymerization productsafter activation with sodium silicate solution at (a) 20 ◦C and (b) 30 ◦C for 72 h

1.4 275.6 312.4 332.2 326.3 340.31.6 228.6 286.8 288.5 276.3 309.8

f aqueous components on and in MK particles, breaking of bondsAl O, Si O), and formation and exchange of Si OH and Al OHpecies. The heat release due to water absorption has been cor-ected in the ICC data presented here by use of an MK-water paste asaseline comparison, as noted in Section 2.2. Thus, the first exother-ic peak indicates the enthalpy of the initial chemical reactions, i.e.,

he dissolution of MK, as shown in Eq. (1):

l2O3 · 2SiO2 + 5H2O + 4 OH−�H1−→2Al(OH)−4 + 2SiO(OH)−

3 (1)

Using the standard enthalpies of formation of MK−3213.4 kJ/mol [17]), H2O (−285.8 kJ/mol [18]), OH−

−230.0 kJ/mol [18]), Al(OH)4− (−1500.8 kJ/mol [18]) and

iO(OH)3− (−1436.1 kJ/mol [18]), the dissolution enthalpy �H1

s calculated to be −208.3 kJ/mol. This explains the first strongxothermic peak in the ICC data. At lower Ms, the concentrationf the NaOH component of the sodium silicate solutions is higher,iving an increased alkalinity. This leads to a higher dissolutionxtent, as indicated by a more intense first exothermic peak.he nature of the second exothermic peak is explained by theontinuous polymerization between the dissolved species andilica oligomers present in activation solution. The polymerizationroducts are too complex to enable us to perform exact thermo-ynamic calculations, however, as a whole system, this will beurther discussed in the following sections.

Therefore, with reference to Part 1 of this study, it is reason-ble to ascribe the nature of the reaction processes responsibleor these two peaks to two basic stages: (I) dissolution of MKnder alkaline conditions, and (II) polymerization between theissolved Al and Si and the silicate monomers present in thectivator. In Stage I, MK particles dissolve to release Al and Since in contact with activator solutions, as indicated by therst exothermic peak. When the concentrations of Al and Sionomers and other small dissolved species reach a critical

evel exceeding saturation conditions in the alkaline environmentrevailing here, polymerization becomes the dominant reactionnd the second exothermic peak appears. The aluminosilicateligomers will further polymerize into amorphous polymers or

proto-zeolitic nuclei’, which are thermodynamically metastable,nd will transform into aluminosilicate gels or semicrystallizedhases [19,20]. Unlike the NaOH-activated systems studied in1], there is no third exothermic peak appearing in the sys-ems reacting at higher temperatures here, meaning that theast nucleation of aluminosilicate gels promotes the setting theaste and hinders their further reorganization, which is therocess that leading to the third exotherm in NaOH activatedinders. This implies that the products in sodium silicate acti-ation systems require higher energy to reorganize the localtructure than in sodium hydroxide activation systems, as proposedn the literature for other aluminosilicate geopolymer systems7].

.2. XRD study of the geopolymerization products

Fig. 2 shows XRD data for MK and the geopolymer samplesbtained after 72 h of reaction at 20 ◦C and 30 ◦C. After reac-ion with sodium silicate activators, the center of the broadeature observed, often described as an ‘amorphous hump’, has

(Kaolinite, Al2Si2O5(OH)4, Powder Diffraction File (PDF) # 01-0527; Halloysite,Al2Si2O5(OH)4·2H2O, PDF # 09-0451; Quartz, SiO2, PDF # 05-0490).

shifted from 25◦ 2� in MK to 32◦ 2�. This shift to higher 2�angle upon reaction is also found in NaOH-activated systems [1],and is usually attributed to the formation of geopolymeric gels[21,22].

The XRD patterns of geopolymers formed at 30 ◦C (Fig. 2b) aresimilar to the patterns of products activated with NaOH at thesame temperature [1], and there are no detectable crystallized zeo-lite phases in the systems studied here after 72 h of reaction. Onenotable feature in the patterns is the increased diffraction inten-sity at ∼32◦ 2� with lower Ms, i.e., with higher alkali concentrationas the samples are formulated here. This trend was also noted inthe XRD patterns of products activated with NaOH [1]. Combinedwith the results related to total heat release, it indicates that theamount of geopolymeric gel formed increases with the alkali con-centration, which is consistent with the current understanding ofthe alkali activation mechanism. According to this mechanism, at

−

a given silica concentration, a higher available OH concentrationresults in more extensive dissolution of MK and consequently moreAl and Si monomers become available to participate in polymeriza-tion.

Z. Zhang et al. / Thermochimica Acta 565 (2013) 163– 171 167

Table 3Reaction equations and reaction enthalpies of geopolymerization at 25 ◦C.

Ms Reaction equation �H298r0 (kJ/mol)

1.0 Al2O3·2SiO2 + 2Na+ + 0.5SiO(OH)3− + 0.5SiO2(OH)2

2− + 0.5OH− + 0.5H2O → 2Na(AlO2·1.5SiO2)·H2O −338.9+ − 2−

3(OH) 2− −

3(OH)3(OH)

4

4

tcCogwtpofupttwoitoesade

sg1seaep

�

w−

wtaSvprNcttt

i

= Si2O3(OH)42− + SiO2(OH)2

2− (�Hro = −134.6 kJ/mol)

(3)

1.0 1.2 1.4 1.60.0

0.1

0.2

0.3

0.4

0.5

Reacti

on

ex

ten

t

Ms

calculated p art

total

1.2 Al2O3·2SiO2 + 2Na + 0.4SiO(OH)3 + 0.4SiO2(OH)2 + 0.2Si2O1.4 Al2O3·2SiO2 + 2Na+ + 0.3SiO(OH)3

− + 0.3SiO2(OH)22− + 0.4Si2O

1.6 Al2O3·2SiO2 + 2Na+ + 0.2SiO(OH)3− + 0.2SiO2(OH)2

2− + 0.6Si2O

. Quantification of geopolymerization kinetics

.1. Thermochemical parameters

The definition of precursor and reaction product species, andhus the calculation of their thermochemical parameters, is criti-al in quantifying the extent of geopolymerization from ICC data.omparing these geopolymer materials with the products in Part 1f this study [1] when NaOH solution was used as the activator, theeopolymers produced in the current study are less ordered andithout detectable crystalline zeolitic phases. This suggests that

he soluble silicate in the activator hinders the formation of zeoliterecursors and their later crystallization. This trend has also beenbserved in fly ash-based geopolymer synthesis [8], and the reasonor this behavior is related to the Si/Al ratio in the initial gels and liq-ids. Zeolite nuclei are more favored, relative to the amorphous gelroduct, at Si/Al ratios close to 1 in alkali activated metakaolin sys-ems [23]. Introducing soluble silicate into the activator increaseshe Si/Al ratio above 1, particularly in the early stages of reactionhen the MK particles are only partially converted, meaning that

nly a fraction of the Al has been released from the particle into thenitially Si-rich activating solution, and so the effective Si/Al ratio ofhe gel is high. This does not preclude the possibility of formationf zeolites with high Si/Al ratio, such as zeolites NaY and NaP, how-ver, the observation of these crystalline structures in geopolymersimilar to those studied here has only been noted for systems curednd/or aged at high temperature, and the development of a higheregree of crystallinity requires months or longer at temperaturesxceeding 85 ◦C [24].

Although no well-crystallized zeolite phases are found in theodium silicate-activated systems here, the local structure of theeopolymer gels is proposed to be analcime-like as discussed in Part

of this study [1], and consistent with local structural analysis ofilicate-activated metakaolin geopolymers [25–27]. The standardnthalpies of geopolymer gels of composition NaAlSimO4+2m·H2Os a function of Si/Al ratio can thus be calculated from the standardnthalpies of formation of analcime ([NaAlSi2O6]·H2O) and amor-hous silica glass (SiO2), according to Eq. (2):

H0f(geopolymer) = �H0

f(analcime) + (m − 2) · �H0f(am-SiO2) (2)

here �H0f(analcime) = −3303.2 kJ/mol [28] and �H0

f(am-SiO2) =904.2 kJ/mol [18].

It can be seen in Table 2 that the Na/Al molar ratio is <1,hich means that the Na+ present is stoichiometrically not enough

o balance the negative charges if all Al atoms from the MKre converted to geopolymer gel and a tetrahedral environment.o, the reactions are based on the available Na+ in the acti-ator as a limiting reagent. With these assumptions and theroposed parameters, the geopolymers derived from the cur-ent four systems are NaAlO2·1.5SiO2·H2O, NaAlO2·1.6SiO2·H2O,aAlO2·1.7SiO2·H2O and NaAlO2·1.8SiO2·H2O according to the stoi-hiometric compositions of the four reaction systems from Ms = 1.0o 1.6. Their standard enthalpies of formation are thus calculated

o be −2851.1, −2941.52, −3031.94 and −3122.36 kJ/mol respec-ively.

The other required thermochemical parameters relate to the sil-cates supplied by the activator. Although the geopolymeric gels

4 + 0.4OH + 0.4H2O → 2Na(AlO2·1.6SiO2)·H2O −399.64

2− + 0.3OH− + 0.3H2O → 2Na(AlO2·1.7SiO2)·H2O −460.34

2− + 0.2OH− + 0.2H2O → 2Na(AlO2·1.8SiO2)·H2O −521.1

are assumed to be similar in local structure regardless of the acti-vator type, the detailed reaction paths leading to gel formationwill be different on a chemical level. The soluble silicate speciespresent in the activator make the reaction much more compli-cated than that in NaOH activated systems, because the silicatemonomers have a wide distribution of Qn environments. The exactdistribution of these species in each specific solution can be deter-mined from NMR spectra or by other chemical methods, but forthe purposes of the calculations presented in this study, to sim-plify the calculation, it is assumed that the silicates are presentmainly in Q0 (SiO(OH)3

− and SiO2(OH)22−) and Q1 (Si2O3(OH)4

2−)environments. This is a reasonable assumption to make in this con-text because Q0 and Q1 are the two most abundant environmentswhen 1.0 < Ms < 1.5 [29], and when the pH is 13–14 the silicateions in sodium silicate solution are mainly deprotonated eitheronce or twice [16]. Using the standard enthalpies of formationof MK (Al2O3·2SiO2) (−3213.4 kJ/mol) [17], OH− (−230.0 kJ/mol)[18], Na+ (−240.3 kJ/mol) [18], H2O (−285.8 kJ/mol) [18], SiO(OH)3

−

(1436.1 kJ/mol) [18] and SiO2(OH)22− (1386.7 kJ/mol) [18], the the-

oretical reaction enthalpy of each systems can then be estimated.Table 3 describes the reactions that are used here to represent

the process of geopolymerization in each of the systems. With Msdecreasing from 1.6 to 1.0, the amount of Si2O4(OH)4

2− is assumedto decrease linearly. This is because when NaOH was added intothe sodium water glass to decrease its modulus, depolymerizationreactions occurred among the polymeric silicate tetrahedral speciesas sketched in Eqs. (3) and (4), which show examples of this processfor trimeric and dimeric species respectively:

Si3O5(OH)53− + OH−

Fig. 3. The geopolymerization extent (in terms of the conversion of metakaolin) as afunction of the modulus of sodium silicate activator at 25 ◦C. The calculated amountrefers to the amount in the theoretical calculation based on the quantity of availablesodium silicate.

168 Z. Zhang et al. / Thermochimica Acta 565 (2013) 163– 171

1300 1200 1100 1000 900 800

959

868

1008

Wavenumber ( cm-1

)

Ms=1.0, 20oC

1208

1040 10

83

1300 1200 1100 1000 900 800

952

870

1009

Wavenu mber ( cm-1

)

Ms=1.2, 20oC

1217

1151

1092

1300 120 0 1100 1000 900 800

960

876

1017

Wavenu mber ( cm-1

)

1214

1149

1091

Ms=1.4, 20oC

1300 120 0 1100 1000 900 800

966

876

1017

Wavenu mber ( cm-1

)1209

1140 1084

Ms=1.6, 20oC

1300 120 0 1100 1000 900 800

957

867

1005

Wavenu mber ( cm-1

)

1210

1148

1090

Ms=1.0, 30oC

1300 120 0 1100 1000 900 800

959

870

1009

Wavenu mber ( cm-1

)

1206

1140

1085

Ms=1.2, 30oC

a b

c d

e f

1300 120 0 1100 1000 900 800

956

872

1008

-1

1225

1158 1094

Ms=1.4, 30oC

1300 120 0 1100 1000 900 800

967

874

1015

-1

1219

1155

1095

Ms=1.6, 30oCg h

F zation( 1.0, 30

waSO

Wavenu mber ( cm )

ig. 4. FTIR spectra, and deconvolutions into component peaks, of MK geopolymeria) Ms = 1.0, 20 ◦C; (b) Ms = 1.2, 20 ◦C; (c) Ms = 1.4, 20 ◦C; (d) Ms = 1.6, 20 ◦C; (e) Ms =

Si2O3(OH)42− + 2OH−

= 2SiO2(OH)22− + H2O (�Hr

o = −329.4 kJ/mol) (4)

here the standard formation enthalpies �Hfo of the species

re obtained from [18]: Si3O5(OH)53− = −3292.0 kJ/mol,

i2O3(OH)42− = −2269.9 kJ/mol, SiO2(OH)2

2− = −1386.7 kJ/mol,H− = −230.0 kJ/mol, and H2O = −285.8 kJ/mol.

Wavenu mber ( cm )

products after activation with sodium silicate solutions at 20 ◦C and 30 ◦C for 72 h:◦C; (f) Ms = 1.2, 30 ◦C; (g) Ms = 1.4, 30 ◦C; (h) Ms = 1.6, 30 ◦C.

These depolymerization reactions result in a reduction in thedegree of polymerization of silicate tetrahedra. A strong heatrelease during preparation of the sodium silicate solutions withlow modulus was noted, consistent with the exothermic nature of

reactions (3) and (4), and also the exothermic dissolution of solidNaOH into an aqueous environment. However, all solutions werecooled to room temperature before use in ICC experiments, so thisdid not influence the observed heat flow data.

Z. Zhang et al. / Thermochimica Acta 565 (2013) 163– 171 169

1.0 1.2 1.4 1.60

20

40

60

80

100

Rela

tiv

e a

reas

of

reso

lved

peak

s (%

)

Ms

Wavenumbe r /cm-1

1208-1217

1140-1151

1083-1092

1008-1017

952-966

868-876

1.0 1.2 1.4 1.60

20

40

60

80

100

Rela

tiv

e a

reas

of

reso

lved

peak

s (%

)

Ms

Wavenumbe r /cm-1

1206-1225

1140-1158

1085-1095

1005-1015

956-967

867-874

a

b

Fv

dttrdt−ecef

4e

itefifMossb

t

Fig. 6. SEM images of the fracture surfaces of the geopolymerization products by◦

ig. 5. Relative areas of the deconvoluted component peaks within the main Si-O-Tibration region of 800–1300 cm−1: (a) 20 ◦C and (b) 30 ◦C.

The actual balance between SiO(OH)3− and SiO2(OH)2

2−

epends on pH, and will also vary according to Ms. However, hereheir concentration ratio is further assumed to be 1 to simplifyhe calculation, as a higher or lower ratio does not change theeaction enthalpy �H298

0 significantly due to the small enthalpyifference associated with deprotonation. For example, for Ms = 1.0,he calculated overall enthalpy of reaction is −343.1 kJ/mol or336.7 kJ/mol if the silicate present in Q0 sites is assumed to beither all SiO(OH)3

− or all SiO2(OH)22−, respectively. With the

alculated reaction enthalpies, an overall reaction extent can bevaluated. This overall “reaction extent” is an index showing howar the geopolymerization proceeds, at a basic chemical level.

.2. Effects of polymerization state of soluble silicate on thextent of geopolymerization

As a first approximation, the overall extent of geopolymerizations defined as the ratio of the measured cumulative reaction heat tohe calculated reaction enthalpy. Fig. 3 plots two types of reactionxtents defined in terms of MK: one is based on the overall (super-cial) reaction extent at 25 ◦C, and the other is adjusted to account

or the excess MK in the mixture by considering that only enoughK to correspond to a 1:1 Na/Al ratio will react. The reaction extent

f MK activated with sodium silicate solution is much lower than inystems activated with NaOH solutions [1], but appears to be rea-

onable based on additional experimental evidence as discussedelow.

Figs. 4 and 5 provide supporting evidence for this calcula-ion, obtained by deconvolution analysis of the FTIR spectra of

(a) Ms = 1.2 sodium silicate activator at 30 C, where Na/Al = 0.76; and (b) 6 M NaOHat 30 ◦C, where Na/Al = 0.74.

geopolymerization products obtained at 20 and 30 ◦C. The bandsat 867–876 cm−1 are due to OH bending in Si OH groups [30],probably associated with Q3 sites and/or unreacted activating solu-tion [31], while the new bands at 1005–1017 cm−1 are assignedto T O Si (T: tetrahedral Si or Al) asymmetric stretching in thegeopolymer networks, and change substantially both in shape andin intensity as a function of Ms. The band at 952–967 cm−1 corre-sponds to Si O stretching in Si O Na structures [30] which form inthese systems by association of the alkalis from the activating solu-tion with the deprotonated under-coordinated (Q3 or lower) sitesin the relatively immature aluminosilicate framework. The bandsat 1082–1225 cm−1 are consistent with the T O Si asymmetricstretching in the unreacted MK. If it is assumed that the relativeareas of the resolved bands of unreacted MK are proportional totheir concentrations, a general trend is that the quantities of newlyformed products increases with a decreasing in Ms.

Evidence from consideration of the mechanism whereby Na+

balances the charge on Al tetrahedra also supports this conclusion.In Part 1 of this study [1], when the MK was activated with 6 mol/LNaOH solution at an overall Na/Al ratio of 0.74, the fractional reac-tion extent after 72 h was 0.20. The system with Ms = 1.2 here hasan Na/Al ratio of 0.76, which is similar to the 6 M NaOH activa-tion system. At the same reaction temperature, 25 ◦C, the adjustedextent of reaction value is also around 0.20. This finding furthersuggests that the Na/Al ratio could be the most important factor indetermining the reaction extent of raw materials. The soluble sil-ica present in the activator can change the overall Si/Al ratio of thesystem, and also affects the microstructural properties of hardened

products, as shown in Fig. 6 and consistent with the literature [2];however, soluble silica seems overall to be less important than theavailable Na+ in controlling the final extent of reaction reached.

170 Z. Zhang et al. / Thermochimica A

0 12 24 36 48 60 720.0

0.1

0.2

0.3

0.4

0.5

20oC

25oC

35oC

30oC

Reacti

on

ex

ten

t

Reaction time (h)

40oC

Ms=1.2

Fgc

4

otgoart(asrfMvwrert

Fr

ig. 7. Effects of reaction temperature on the superficial reaction extent ofeopolymerization of MK activated by sodium silicate solution with Ms = 1.2 andoncentration 35 wt.%.

.3. Effect of reaction temperature on geopolymerization extent

Fig. 7 presents the effects of reaction temperature on the extentf geopolymerization for the system with Ms = 1.2, calculated fromhe ICC data based on the assumption that the reaction enthalpyiven in Table 3 is constant within the 20–40 ◦C temperature rangef interest. As indicated by the slopes of the curves, the reactionccelerates significantly with increasing temperature. The rate ofeaction at 40 ◦C is highest in the first 5 h, and the ICC data showhat the second peak in system at 40 ◦C appears after only 30 minsee Fig. 1e), meaning that the polymerization in this system occurst very early time. However, the continuous reaction after 5 h in theystem at 35 ◦C releases more heat overall, and results in a highereaction extent than is observed at 40 ◦C. This indicates that theast formation of geopolymeric gels hinders the further reaction of

K and consequently decreases the reaction extent; similar obser-ations have been developed from synthetic geopolymer systemsith tailored rates of Al release, where a rapid early reaction led to a

educed overall extent of reaction due to hindered mass transportffects [32]. This result suggests that there could be an optimumeaction temperature for a given MK, reacted with a given activator,o obtain the highest reaction extent. The use of ICC to determine

1.0 1.2 1.4 1.60

10

20

30

40

50

Ms

Co

mp

ress

ive

stre

ng

th (

MP

a)

24 h

72 h

ig. 8. Compressive strengths of hardened binders synthesized with liquid/solidatio of 0.65 mL/g at 25 ◦C.

cta 565 (2013) 163– 171

this optimum temperature appears to be a good alternative to othermechanical and chemical methods.

4.4. Relationship between reaction extent and mechanicalproperties of hardened binders

Fig. 8 plots the compressive strengths evolution of hardenedbinders at 25 ◦C. The fast setting and hardening behavior of theMK geopolymer pastes results in rapid strength development.After 24 h of reaction, the compressive strength of the binderwith Ms = 1.2 reaches 22 MPa, which increases to 40 MPa when thecuring time is prolonged to 72 h. With increasing Ms (and thusdecreasing alkali content), the compressive strength decreases.The compressive strength depends on many factors, including thegel/residual MK ratio, the nature of gels and their interaction withresidual particles, as well as the hardness of the gel and the resid-ual precursor [33]. Regardless of the effects of the hardness of theresidual phase for the systems studied here (as metakaolin particlesare weak and unlikely to contribute much to the total strength), thequantity and nature of the amorphous phase could be expected tobe major contributors to the strength of the material as a whole,as the cross-linking and compaction provided by this phase willgive strength to the material. Comparing Figs. 3 and 8, the trend incompressive strength of samples fits well with the trend in reac-tion extents, although the liquid/solid ratio in the compressivestrength samples is different from the systems studied by calorime-try. Considering that the quantity of amorphous phase that couldpotentially be generated through solidification of the activator isequivalent for the four systems, as the total dissolved solids con-centration of the activators is constant, the mechanical strengthappears to be approximately proportional to the reaction extent ofMK and the resultant quantity of gel phases.

4.5. Limitations of using ICC in studying geopolymerizationkinetics

It has been demonstrated in this study that ICC is a useful tech-nique for analysis of the kinetics of MK-based geopolymerization.However, there are also limitations associated with this technique.First, to determine the general reaction extent, the thermochem-ical parameters of the reactants and products must be obtainedor estimated. For metakaolin and some other relatively pure pre-cursors, the thermodynamic parameters are available; however,for fly ash which is composed of various glasses and crystallinephases, the thermodynamic parameters are not available and noteasy to estimate. Second, this technique can only provide a generalgeopolymerization rate and extent, which is defined in thermo-chemical (as opposed to microstructural or geometric) terms. Theextent of reaction can only realistically be measured by this tech-nique at relatively early age, but is difficult to compare directly withdata obtained from SEM image analysis or other microstructure-based techniques, because these are insensitive to the degree oflocal (chemical/nanostructural) evolution taking place within thebinder gel itself. The reaction extent as quantified from ICC datahere seems useful as a general index describing how far the systemevolves toward a final ideal structure, although this does differ fromthe extent of transformation of metakaolin on a microstructurallevel.

Thus, for tasks including comparing the reactivity of raw materi-als derived from different sources, after different types of thermalor mechanical pretreatment, or to find the temperature depend-ence or activating solution dependence of the rate and extent of

geopolymerization of a particular precursor, ICC is certainly a valu-able technique. The study of geopolymerization of fly ash and slagis useful for the large-scale industrial application of geopolymerproducts, since these seem to be the most promising precursors at

mica A

pi

5

uhwsatfsp(mn

tpmq1a(gtatoosp

A

s

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

Z. Zhang et al. / Thermochi

resent, and further research into the application of ICC techniquesn this area will be undertaken and reported in the future.

. Conclusions

The geopolymerization process and early-age reaction prod-cts of sodium silicate-activated metakaolin, reacted at 20–40 ◦C,ave been studied using isothermal calorimetry in combinationith structural and microscopic analysis. The calorimetric curve

hows two distinguishable exothermic peaks in the first 3 daysfter metakaolin is brought into contact with sodium silicate solu-ion. The first peak corresponds to the dissolution of metakaolin,ollowed by a broad exothermic peak which indicates the multi-tep polymerization process of Al and Si monomers in the solutionhase. Unlike in NaOH-activated systems, no third exothermic peakwhich was attributed to gel reorganization in NaOH activation of

etakaolin) is observed in sodium silicate-activated systems, ando crystalline phases are found in any of the systems studied here.

By the application of a simplified description of silicate specia-ion in the activator, and by using analcime as a model structure toredict the thermochemistry of sodium aluminosilicate geopoly-er gels, the reaction extent of geopolymerization has been

uantified from the ICC data. Increasing the activator modulus from.0 to 1.6 decreases the fractional reaction extent from 0.26 to 0.12fter 72 h at 25 ◦C. The most important conclusion of this researchcombining the outcomes of Parts 1 and 2) is that the Na/Al in theeopolymerization process has a more pronounced influence onhe geopolymerization extent of metakaolin than do temperaturend Si/Al ratio. Since the early-age reaction extent is limited (lesshan 40% at 72 h) at temperatures <40 ◦C, in either NaOH-activatedr sodium silicate-activated metakaolin, it seems that benefits arebtained from the use of silicate-containing activators, in which theoluble silicate will polymerize and help to make the binder com-act, and thus provide better mechanical and transport properties.

cknowledgements

ZZ thanks Halok Pty Ltd. for the geopolymer research projectupport for part of the work conducted in this study.

eferences

[1] Z. Zhang, H. Wang, J.L. Provis, X. Yao, F. Bullen, Y. Zhu, Quantitative kinetic andstructural analysis of geopolymers. Part 1. The activation of metakaolin withsodium hydroxide, Thermochim. Acta 539 (2012) 23–33.

[2] P. Duxson, J.L. Provis, G.C. Lukey, S. Mallicoat, W.M. Kriven, J.S.J. van Deventer,Understanding the relationship between geopolymer composition, microstruc-ture, and mechanical properties, Colloids Surf. A 269 (2005) 47–58.

[3] A. Fernández-Jiménez, A. Palomo, Composition and microstructure of alkaliactivated fly ash binder: effect of the activator, Cem. Concr. Res. 35 (2005)1984–1992.

[4] P.W.J.G. Wijnen, T.P.M. Beelen, J.W. de Haan, C.P.J. Rummens, L.J.M. van de Ven,R.A. van Santen, Silia gel dissolution in aqueous alkali metal hydroxides studiedby 29Si NMR, J. Non-Cryst. Solids 109 (1989) 85–94.

[5] J.L. Provis, P. Duxson, G.C. Lukey, F. Separovic, W.M. Kriven, J.S.J. van Deventer,

Modeling speciation in highly concentrated alkaline silicate solutions, Ind. Eng.Chem. Res. 44 (2005) 8899–8908.

[6] D. Dimas, I. Giannopoulou, D. Panias, Polymerization in sodium silicate solu-tions: a fundamental process in geopolymerization technology, J. Mater. Sci. 44(2009) 3719–3730.

[

cta 565 (2013) 163– 171 171

[7] M. Criado, A. Fernández-Jiménez, A.G. de la Torre, M.A.G. Aranda, A. Palomo,An XRD study of the effect of the SiO2/Na2O ratio on the alkali activation of flyash, Cem. Concr. Res. 37 (2007) 671–679.

[8] S.A. Bernal, J.L. Provis, V. Rose, R. Mejía de Gutiérrez, Evolution of binder struc-ture in sodium silicate-activated slag-metakaolin blends, Cem. Concr. Compos.33 (2011) 46–54.

[9] A.S. de Vargas, D.C.C. Dal Molin, A.C.F. Vilela, F.J. Da Silva, B. Pavão, H.Veit, The effects of Na2O/SiO2 molar ratio, curing temperature and age oncompressive strength, morphology and microstructure of alkali-activated flyash-based geopolymers, Cem. Concr. Compos. 33 (2011) 653–660.

10] A. Hajimohammadi, J.L. Provis, J.S.J. van Deventer, The effect of silica availabilityon the mechanism of geopolymerisation, Cem. Concr. Res. 41 (2011) 210–216.

11] J.L. Provis, J.S.J. van Deventer, Geopolymerisation kinetics. 1. In situ energydispersive X-ray diffractometry, Chem. Eng. Sci. 62 (2007) 2309–2317.

12] F. Pacheco-Torgal, D. Moura, Y. Ding, S. Jalali, Composition, strength and work-ability of alkali-activated metakaolin based mortars, Constr. Build. Mater. 25(2011) 3732–3745.

13] X. Yang, W. Zhu, Q. Yang, The viscosity properties of sodium silicate solutions,J. Solut. Chem. 37 (2008) 73–83.

14] X. Yao, Z. Zhang, H. Zhu, Y. Chen, Geopolymerization process of alkalimetakaoli-nite characterized by isothermal calorimetry, Thermochim. Acta 493 (2009)49–54.

15] K. Sagoe-Crentsil, L. Weng, Dissolution processes, hydrolysis and condensationreactions during geopolymer synthesis. Part II. High Si/Al ratio systems, J. Mater.Sci. 42 (2007) 3007–3014.

16] H. Rahier, W. Simons, B. Van Mele, M. Biesemans, Low-temperature synthesizedaluminosilicate glasses. Part III. Influence of the composition of the silicate solu-tion on production, structure and properties, J. Mater. Sci. 32 (1997) 2237–2247.

17] N.C. Schieltz, M.R. Soliman, Thermodynamics of the various high temperaturetransformations of kaolinite, in: Proceedings of the 13th National Conferenceon Clays and Clay Minerals, Pergamon Press, New York, 1966, pp. 419–428.

18] Common Thermodynamic Database Project, 2013, http://www.ctdp.org/#http%3A//www.ctdp.org/ctd import/species list/species 3785/view (25.08.11).

19] J.L. Provis, J.S.J. van Deventer, Geopolymerisation kinetics. 2. Reaction kineticmodeling, Chem. Eng. Sci. 62 (2007) 2318–2329.

20] J.L. Provis, P.A. Walls, J.S.J. van Deventer, Geopolymerisation kinetics. 3. Effectsof Cs and Sr salts, Chem. Eng. Sci. 63 (2008) 4480–4489.

21] I. Lapides, L. Heller-Kallai, Reactions of metakaolinite with NaOH and colloidalsilica – comparison of different samples (part 2), Appl. Clay Sci. 35 (2007) 94–98.

22] F. Zibouche, H. Kerdjoudj, J.B. d’Espinose de Lacaillerie, H. Van Damme,Geopolymers from Algerian metakaolin, influence of secondary minerals, Appl.Clay Sci. 43 (2009) 453–458.

23] C. Sathy, P.N. Pramada, Microwave assisted synthesis of zeolite A frommetakaolin, Microporous Mesoporous Mater. 108 (2008) 152–161.

24] R.R. Lloyd, The durability of inorganic polymer cements, University of Mel-bourne, 2008 (Ph.D. Thesis).

25] J.L. Bell, P. Sarin, J.L. Provis, R.P. Haggerty, P.E. Driemeyer, P.J. Chupas, J.S.J. vanDeventer, W.M. Kriven, Atomic structure of a cesium aluminosilicate geopoly-mer: a pair distribution function study, Chem. Mater. 20 (2008) 4768–4776.

26] C.E. White, J.L. Provis, A. Llobet, T. Proffen, J.S.J. van Deventer, Evolution oflocal structure in geopolymer gels: an in situ neutron pair distribution functionanalysis, J. Am. Ceram. Soc. 94 (2011) 3532–3539.

27] J.L. Bell, P. Sarin, P.E. Driemeyer, R.P. Haggerty, P.J. Chupas, W.M.Kriven, X-ray pair distribution function analysis of a metakaolin-based,KAlSi2O6·5.5H2O inorganic polymer (geopolymer), J. Mater. Chem. 18 (2008)5974–5981.

28] P.S. Neuhoff, G.L. Hovis, G. Balassone, J.F. Stebbins, Thermodynamic propertiesof analcime solid solutions, Am. J. Sci. 304 (2004) 21–66.

29] D. Cao, D. Su, G. Song, Geopolymeric behavior and structure of lower modulussodium silicate solutions, J. Chin. Ceram. Soc. 32 (2004) 1036–1039 (in Chinesewith English abstract).

30] W.R. Taylor, Application of infrared spectroscopy to studies of silicate glassstructure: examples from the melilite glasses and the systems Na2O–SiO2

and Na2O–Al2O3–SiO2, Proc. Indian Acad. Sci. (Earth Planet. Sci.) 99 (1990)99–117.

31] I. Giannopoulou, D. Panias, Hydrolytic stability of sodium silicate gels in thepresence of aluminum, J. Mater. Sci. 45 (2010) 537–5377.

32] A. Hajimohammadi, J.L. Provis, J.S.J. van Deventer, Effect of alumina release

rate on the mechanism of geopolymer gel formation, Chem. Mater. 22 (2010)5199–5208.

33] H. Xu, J.S.J. van Deventer, Microstructural characterisation of geopolymerssynthesised from kaolinite/stilbite mixtures using XRD, MAS-NMR, SEM/EDX,TEM/EDX, and HREM, Cem. Concr. Res. 32 (2002) 1705–1716.