7/24/2019 Leaching Phung
1/18
7/24/2019 Leaching Phung
2/18
much higher than in a portlandite-controlled system (e.g. 20 mM vs.2730 mM[11]).
Most studies on Ca-leaching quantify the propagation of degradeddepth and the consequences for the mechanical behavior [12,13]. How-
ever, only a few studies discuss microstructural alterations afterleaching [14,15], typically leading to a materialwith coarsermicrostruc-ture in case of accelerated leaching in ammonium nitrate solution, and
the effects of leaching on transport properties [16,17]. However, to
what extent the microstructure and transport properties are modi
edis still questionable. Alterations depend, beside on cement type andportlandite content, strongly on pozzolanic activity and other additives
which may eventually even lead to porosity decrease by secondary pre-cipitation [1820]. Generally, additives such as y ash and blast furnaceslag or glass powder and silica fume[5]improve the resistance againstleaching. From a mineralogical point of view, Ca-leaching mainly affectsportlandite, but several studies show also important decalcication of
the C-S-H phases, especially in long-lasting leaching experiments [5,21]. As llers in general, and also limestone llers, specically inuenceprocesses as carbonation[22], microstructure[23], hydration[24,25],and mechanical properties[26,27], it is expected that they also have
an effect on Ca-leaching of cementitious materials. However, informa-tion is still missing in the literature.
A fundamental question raised in this study is how these microstruc-tural and mineralogical changes during Ca-leaching alter effective trans-port properties as diffusivity and permeability. Alterations observed in
reported experiments are in most studies in line with the expectations,which is an increase with increasing degradation [16,17,28], but thelink with the underlying micro-scale changes is mostly not studied. Astudy of Bernard et al. [51]showed the importance of mineralogical
changes on the interpretation of measured HTO diffusivity on degradedCEM I cement paste and mortar samples. The effective diffusivity in-creases from a 31-time increase, when only portlandite is completely dis-solved, to a 118-time increase, when also hydrated aluminates and
sulfoaluminates are dissolved.As cement additives andllers are expected to have consequences on
Ca-leaching, this study presents an in-depth investigation of mineralogi-cal and microstructural alterations and changes in diffusivity and perme-
ability after Ca-leaching of cement pastes with limestone ller within afactorial design with limestone ller replacement and water/powderratio as experimental variables. The extensive experimental ow
chart consists of leaching cement pastes in aggressive environment(NH4NO3), during whichCa-leaching wasfollowedup by ionchromatog-raphy (IC), followed by a series of quantitative and qualitative analyses(scanning electron microscopy (SEM) in combination with energy dis-
persive X-ray microanalysis (SEM-EDX), mercury intrusion porosimetry(MIP), (quantitative) X-ray diffraction (XRD/QXRD), N2-adsorption, phe-nolphthalein spraying. Recent developed experimental setups were usedto measure water permeability[29] and diffusivity [30] of dissolved gases
in reference and leached samples.
2. Accelerated Ca-leaching in ammonium nitrate solution
2.1. Materials
Experiments were performed on cement pastes made from cement
Type I ordinary Portland cement (CEM I 52.5 N,Table 1), tap water,and limestone ller. The limestone ller (Calcitec 2001S) has a Blainespecic surface of 3500 cm2/g and a CaCO3 content of 98.30% (see
Table 2for some other properties). Superplasticizer Glenium 27 wasadded to the mix with content of 0.5% with respect to mass of cement.
Cement paste compositionswere made on the basisof a factorialex-perimental design with 22 factorial points and 1 center point along the
experimental variables water/(limestone ller + cement) ratio (w/p)and the cement replacement by limestone (ls/p = limestone ller/(limestone ller + cement)). w/p ratio and ls/p ratio varied,respective-
ly, from 0.325 to 0.425 and from 0 to 20% (Table 3). The material com-
positions used in this study simulate (but without coarse aggregates)the composition for the BelgianSupercontainer in the newreference de-sign for disposal of high-level waste and spent fuel[31], in which lime-
stone has been used as ller and aggregates because it contains lowreactive constituents (silicon, aluminium, magnesium), which limitsthe alkali-aggregate reaction resulting in expansion and cracking[32].S1 to S5 and S1L to S5L are herein referred to the reference and leachedsamples, respectively. The run order was randomly generated to elimi-
nate bias. Note that in cement industry, Portland limestone cementshave been produced by inter-grinding clinker and limestone (and gyp-sum) in which different limestone/(limestone + clinker) ratios are ini-tially chosen. Therefore, the water/cementitious material ratio for
Portland limestone cement is actually the water/powderratio. Forconsistency, w/p ratio rather than w/c ratiowas chosen as oneof the ex-perimental variables. However, dilution effects due to limestone lleraddition can be easily obtained by converting w/p ratio to w/c ratio.
2.2. Experimental procedure and test setup
Cement pasteswere poured and cured in a cylindrical PVC tube with
an inner diameter of 97.5 mm. The samples were then rotated during24 hours to prevent segregation [29] and subsequently cured for28 days under sealed conditions in a temperature-controlled room(21 1 C). The cured cement pastes were sawn into 25 mm thick
slices. To impose only one-dimensional leaching at the bottom and topof the slice, the PVC cover in the axial direction was not removed andany gap between the PVC cover and cement paste was lled withepoxy resin (Fig. 1a).
In orderto avoid initial leaching, the cement slices were saturated ina saturated lime solutionfollowing the procedure described in [29]. Thesaturated cement slices were immersed in ammonium nitrate solution
6 M chambers. The water-contact surface area ratio was 8 cm3/cm2.This factor ensured that the pH of the solution remained below 9.25 be-cause this pH is critical to maintain leaching acceleration[33]. As such,solution renewal was avoided which facilitated quantitative analysis
of the Ca-leached amounts.A picture of the test setup is given inFig. 1b. Nitrogen was bubbled
through the system to prevent carbonation during leaching and to re-move the formed NH3gas via the bubbler (water lock). The solution
was homogenized by a magnetic stirrer. There was an extraction lineto extract solution for further analysis and follow up the pH. The setupconsisted of three NH4NO3chambers which allowed simultaneously
leaching of three samples. Reference samples were prepared from thesame batchof cement paste and kept under the same conditions exceptfor immersion in ammonium nitrate solution.
Table 1
Chemical properties (wt. %) of the cement (from manufacture fact sheet).
CaO 63.0% Chromium(VI) b2.104%
SiO2 20.0% Cl 0.06%
Fe2O3 3.0% Na2O eq. 0.85%
Al2O3 5.0% Loss on ignition 1.60%Sulphate SO3 2.9% Insoluble residue 0.50%
Table 2
Chemical and physical properties of limestone ller (from manufacture fact sheet).
Density 2.7 g/cm3 CaCO3 98.30%
Blaine-specic surface 3500 cm2/g MgO 0.36%
Particle size distribution: SiO2 0.75%
Passing 2 mm 100.0% Al2O3 0.25%
Passing 500m 99.9% Fe2O3 0.10%
Passing 125m 97.0% Na2O eq. 0.05%
Passing 63m 77.0% pH 9.5D50(50% passing) 10.1m
2 Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxxxxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
3/18
3. Characterization of mineralogical and microstructural alteration
3.1. Phenolphthalein spraying, ion chromatography, and mass loss
The leaching tests were stopped after 7, 14, 21, and 28 days to exam-ine different leaching grades (sample S1 only after 7, 21, and 28 days).The leached samples were cut into hemispherical parts and sprayedby a phenolphthalein solution to determine the leached depth[35]bya color difference between a zone above and below a pH of about 910. Note that the phenolphthalein indicator cannot detect exactly theportlandite dissolutionfront as the portlandite dissolution front extendsbeyond a pH zone of about 910 because of buffering by other cementhydrates[36].
The amount of leached calcium was measured by ion chromatogra-
phy (IC) in solutions extracted on a weekly basis from thesetup. Prior tothe measurement, the solution was ltered through a 0.45m syringe
lter, and then diluted with milli-Q water to a (1:1000) solution. TheIC analysis was performed using a Dionex CS12A column and a 20 mM
MSA isocratic elution. Detection was performed by a Dionex EC50electro conductivity detector.
In addition, sample masses before and after leaching were deter-mined under saturated condition by a precise balance (103 g). This is
a direct method to quantify the mass loss due to leaching.
3.2. Mercury intrusion porosimetry and N2-adsorption
The combinationof mercury intrusion porosimetryandN2-adsorptionenables to characterize pore structural changes over a broad pore size
range where the former gives reasonable results for larger pore sizes,while the later provides information on the smaller pores. The deforma-tion of leached samples is very large at the high applied pressures of mer-cury during an MIPexperimentwhich is not the case foran N2-adsorption
experiment. For this reason, combining MIP and nitrogen adsorption
measurements will give a better understanding of the pore structure ofcementitious materials, especially for leached paste.
Several leached paste pieces were taken within 3 mm from thereac-tive surface and referred as leached sample.The mass of each sample
was about 34 g. Freeze drying was chosen to prepare the samples forMIP and nitrogen adsorption measurements. By freeze drying, watercrystals sublimate, which prevents micro crack formation because cap-
illary stresses are created if drying passes through the liquid state[37].
This method also gives a dried sample in a relatively short period. Sam-ples were directly immersed in liquid nitrogen until the escape of gasbubbles stops. Subsequently, the samples were transferred to a vacuumchamber where a vacuum pressure of 2.5 102 mbar was applied for24 hours. Dried samples were kept in closed glass bottles untilexamination.
MIP experiments were performed on the PASCAL 140/440
porosimeter. The pressurization was done in low and high pressureparts. In the low pressure part, after evacuation and mercury lling, thepressure was continuously increased up to 0.2 MPa then reduced to at-mospheric pressure before moving to the high pressure part. Both thesample volume and information on the larger pores were determined
from thelow pressure part. In thehighpressure part, thepressureof mer-cury was continuously increased up to a maximum pressure of 200 MPafor reference samples and of 30 MPa for the leached material because ofits high ductility. The compression of the leached sample was also takenintoaccount in the porestructural calculations. Nitrogenadsorption mea-
surements were done on a TriStar II 3020 Micromeritics. Dried sampleswere weighed and degassed under vacuum at 35 C for 24 hours. Theamount of nitrogen adsorbed during the test was calculated by the volu-metric method (based on gas laws). About 80 equilibrium points were
collected during adsorption and desorption. More data points were col-lected at the low relative pressure part in order to derive the microporevolume.
3.3. Scanningelectron microscope and energy dispersive X-ray microanalysis
Scanning electron microscope combined with energy dispersive X-ray microanalysis (SEM-EDX) was used to identify the pore structure
andmineralogicalchanges. EDX was usedto obtainatomic Ca/Si prolesand elemental mapping over the transition zone of leached materials.
Combination with backscattered electron (BSE) images allowed for bet-ter separation of the phases of reference/degraded cement pastes.
Polished sections are required for BSE images. The solvent replace-ment method was chosen to dry the samples. This method is found to
be a good technique to preserve the microstructure for SEM analysis[38]. The samples were immersed in 2-propanol 99.5% for 2 weeks
Table 3
Detail of factorial experiment design.
Run Cement LS Powder Water w/c w/p ls/p SP Sample
order kg kg kg kg % l
1 450 0 450 146 0.325 0.325 0 2.3 S1
2 360 90 450 191 0.531 0.425 20 1.8 S2
3 450 0 450 191 0.425 0.425 0 2.3 S3
4 405 45 450 169 0.417 0.375 10 2.0 S4
5 360 90 450 146 0.406 0.325 20 1.8 S5
Fig. 1. Preparation of cementpaste slice (a) and accelerated leaching setup using NH4NO3 solution(b): 1magnetic stirrer; 2NH4NO3 vessel;3nitrogen line; 4extraction line; 5bubbler
(adapted from[34]).
3Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxx xxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
4/18
and then placed in a vacuum chamber under 2.5 102 mbar for
3 days. The dried samples were impregnated in a high strength and ex-tremelylow viscosity resin before grinding andpolishing.Resin impreg-nation helps to ll the voids and pores, preserve the originalmicrostructure, enhance contrast between solid phases and pores, and
prevent cracking during grinding/polishing. The polished sampleswere then coated by a thingoldlayer (few m) to preventcharging dur-ing SEM examination. Even with gold coating, the samples might becharged because of the non-conductive resin. Therefore, the top andbottom of the sample were connected by a conductive copper tape.
SEM measurements of samples were done using a JEOL JSM 6610 scan-
ning electron microscope. EDX analysis was carried out with the aid ofESPRIT Software. ImageJ software was used for separation of poresand solid phases.
3.4. Qualitative and quantitative X-ray diffraction
XRD was also used to identify mineralogical changes after leaching.Powder was collected from reference and degraded samples by drillinglongitudinally in the samples. Drilling was halted every 3 mm to collect
Fig. 2.Increase of phenolphthalein degraded depth (a) and normalized Ca-leached amount to cement over immersed time (b).
Fig. 3.Main effects plot (a) and interaction plot (b) for degradation rate (top) and Ca-leached rate (bottom).
4 Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxxxxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
5/18
the dust. To preventcross contamination, the hole was carefully cleanedby compressed air before continuing the drilling. For the reference sam-ple, the surfacelayer was removedbefore collectingthe powder. Prior toanalysis, the sampleswere vacuum-dried. Forsamples on which a quan-
titative analysis was performed, 10% of ZnO internal standard materialwas added to the mixture. The classical grinding procedure (by handwith pestle and porcelain mortar) was used to reduce the size of thegrains. This method is time consuming compared to using micronizing
mill but avoids reaction of portlandite with alcohol (needed in the mi-cronizing mill) forming a carbonate-like product (Ca(OCH3)2)[39].Powder XRD measurements were done using a Philips X'Pert Pro Dif-fractometer. The samples were scanned over the 2range of 570o
using CuKradiation at 45 kV and 40 mA.The Rietveld method was used to rene the crystal structure model
of a material, allowing determination of the mass fraction at differentphases. A least squared error tting of the full prole tting was per-
formed using the X'Pert Highscore Plus software. The crystal structuredatabase was downloaded from the American Mineralogist Crystal
Structure Database[40]. The following parameters were adjusted in a
Fig. 4.Comparisons of mass loss of leached samples during leaching determined by weighing and ion chromatography methods with assumption of no C-S-H decalci cation.
Fig. 5.Adsorption/desorption isotherms of reference and 28-day leached sample S3L.
5Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxx xxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
6/18
semi-automatic mode one-at-a-time: background, scale factor, zeroshift, lattice cell, Caglioti parameters (W, U, V), peak shape, asymmetry,
and preferred orientation.
4. Determination of the changes in transport properties
After the leaching tests, the samples were stored in water until the
water permeability and diffusivity were measured. This was to preventdrying of the samples and migration of calcium nitro-aluminate (a reac-tion product of calcium nitrate and hydrated aluminates) to the dry sur-face which can generate micro cracks [9]. Prior to permeability/
diffusivity measurements, the samples were embedded in polycarbonaterings by a resin under an atmosphere of 100% relative humidity. Waterpermeability was determined using a controlled constant ow methodas described in[29]. The diffusivities of dissolved gases (He and Xe)
were determined using a through-diffusion methodology which allowssimultaneous determination of diffusivities of two dissolved gases as de-
scribed in[30]. Permeability measurements were performed on all refer-ence and leached samples after 28days of leaching. Furthermore, thetimeevolution of permeability was investigated by measuring the permeabili-ty of sample S3 after 1, 2, 4, and 8 weeks of immersion in an ammoniumnitrate solution. Diffusivity measurements were only performed on refer-
ence and leached samples S3 after 28 days of leaching.
5. Results and discussion
5.1. Effects of w/p ratio and limestoneller replacement on degraded depth
and Ca-leached amount
The degraded depth measured by phenolphthalein linearly in-creased with the square root of immersion (reaction) time in NH 4NO3
(Fig. 2a), which indicates a diffusion-controlled process.Fig. 2b showsthe amountof calcium in the surrounding solution normalized to the ce-
ment content (limestoneller is not leached because of extremely low
solubility), which equals the amount of calcium leached. Again, a linearrelation of leached amount over square root of time was observed.
Degradation rates and Ca-leached rates estimated as the slope of de-
graded depths and Ca-leached amounts, respectively, versus t0.5 areshown inFig. 3. It is clear that the degradation and Ca-leached ratesstrongly increase with the increase of w/p ratio. The degradation and
Ca-leached rates were increased by 70% and 50%, respectively, when w/p ratio increased from 0.325 to 0.425. Adding LS also induced faster deg-radation and Ca-leached rates but had a smaller effect than the w/p ratio.To improve the resistance to leaching, one should choose a mix design
with a low w/p ratio and low LS replacement. The regression equations(rst-order) of degradation and Ca-leached rates are obtained as follows:
Ratedep 1:3970:538ls=p 7:3 w=p3:5 ls=p w=p 1
RateCa 0:715 0:168ls=p 6:1 w=p 1:1 ls=p w=p 2
whereRatedep[mm/day0.5] andRateCa[%/day
0.5] are the degradation andCa-leached rates, respectively, ls/p is the ratio of limestone ller overpowder (cement + limestone ller), and w/p is the ratio of water over
powder. The correlation coefcients (r2) are 0.999 and 0.998 forRatedepandRateCa, repectively, indicating that the models t the experimentaldata well.
The interaction plots (Fig. 3b) enable to visualize the interaction ef-
fects of w/p ratio and LS replacement on the degradation and Ca-leached rates. As the connected lines of factor levels are almost parallelto each other, there might not be interaction present (or minorcontribution).
Table 4
Summary of N2-adsorption results of leached and reference samplesLdenotes leached sample.
Sample
w/p
ls/p
S1
0.325
0
S1L
0.325
0
S2
0.425
0.2
S2L
0.425
0.2
S3
0.425
0
S3L
0.425
0
S4
0.375
0.1
S4L
0.375
0.1
S5
0.325
0.2
S5L
0.325
0.2
BET surface area, m2/g 9.9 73.1 33.3 130.9 30.6 145.8 34.8 182.3 5.5 86.7
Porosity (BJH), % 5.0 16.2 15.2 21.9 13.0 23.8 9.2 26.1 5.4 21.2
BJH adsorption average pore
diameter (4 V/A), nm
13.4 8.7 12.5 8.6 11.6 7.8 9.5 7.3 15.8 6.8
Table 5
Micropore information of leached and reference samples determined using DubininAstakhov model[45]Ldenotes leached sample.
Sample
w/p
ls/p
S1
0.325
0
S1L
0.325
0
S2
0.425
0.2
S2L
0.425
0.2
S3
0.425
0
S3L
0.425
0
S4
0.375
0.1
S4L
0.375
0.1
S5
0.325
0.2
S5L
0.325
0.2
Micropore spec. surface, m2/g 8.1 58.6 27.6 95.7 23.7 131.6 26.4 165.4 4.2 67.9Micropore volume, mm3/g 4.0 27.9 13.5 47.5 11.6 65.0 12.8 82.2 2.1 32.5
Meso/microvolume ratio 6.5 4.2 6.5 4.2 6.9 3.3 3.9 2.8 12.1 4.6
Fig. 6.Inuence of deformation of leached sample on MIP results: sample S1L
the blue curves are the corrected curves.
6 Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxxxxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
7/18
5.2. Evidence of C-S-H leaching from mass loss
This section aims at giving indirect evidence that C-S-H decalcica-
tion contributesto themass loss of the leached sample. When assuming
no C-S-H leaching, IC-measuredCa can be used to calculate the decreaseof the sample weight by making the assumption that the mass loss due
to the leaching of other ions (K+, Na+, and Mg++ ions) is negligible. IC
results of sample S1L (data not shown) with concentrations of Ca much
Fig. 7.Changes in pore structure due to leachingassessed by combining MIP and N2-adsorption:porosity vs. pore diameter (left) and differential pore size distribution (right) of leached
and reference samples.
7Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxx xxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
8/18
largerthan the concentrations of other ions conrmed this assumption.The amount of dissolved portlandite is then obtained as:
CH Ca VaMCHMCa
3
where CHis the dissolved portlandite content [g], [Ca] is Ca ion concen-tration determined by IC [g/l],Vais the volume of the NH4NO3solution[l],MCHis molar mass of Ca(OH)2, 74 [g/mol], andMCais molar mass ofCa, 40 [g/mol]. To link the change in portlandite content to the mea-
sured sample mass loss, it is further assumed that portlandite dissolu-tion creates capillary pores which are quickly lled by water. Thechange in porosity due to portlandite dissolution is
CH
MCHVCH
Vs100%; 4
whereVCHis molar volume of Ca(OH)2, 0.033 [l/mol] andVsis samplevolume [l]. Consequently, the decrease in sample mass is expressed as
mCHCH
MCH
VCHw
Ms100%; 5
where mis percentage of mass loss, [%],Msis sample mass, [g] andwis density of water, 1000 [g/l].Fig. 4shows a comparison between thedirect measured and the calculated mass loss during leaching. The sam-ple masses linearly decreased as a function of square root of immersedtime. The mass loss obtained from direct measurements was smaller
than the mass loss obtained by the IC-calculated method. It is hypothe-sized here that the difference is attributedC-S-H leaching. Leaching of C-S-H creates mainly gel pores and small capillary pores of which the for-mer is hardly accessible by water under normal condition. By excluding
the possible not-relling of pores in the IC method, mass loss is poten-tially overestimated. The mass loss of sample S3L obtained by eitherweighing or IC method was similar. This inconsistent observation is dif-
cult to explain but should not be interpreted as no C-S-H leaching as
EDX results clearly indicated C-S-H dissolution (seeSection 5.5.2).
5.3. Changes in porosity and pore size distribution
5.3.1. Nitrogen gas adsorption results
Fig. 5compares the nitrogen adsorption/desorption isotherms of
leached and reference samples. The shape and size of the hysteresisloopof the leached sample are characteristic for large ink-bottle-shapedpores, whereas the hysteresis loop of the reference sample is more char-acteristic for a slit-shaped or plate-like pore system[41]. Such a plate-
like pore system corresponds to C-S-H gel which consists of very thinsheets[42]. The alteration toward ink-bottle-shaped pores might indi-cate that part of the C-S-H has been dissolved during leaching. The spe-cic surface areas (SSA) determined by the BET method of the leached
samples signicantly increased (Table 4). The SSA of samples withhigh w/p ratio (S2L, S3L, S4L) increased 4 5 times after 28-day ofleaching in ammonium nitrate solution. For the samples with a loww/p ratio, the SSA increased up to 7 and 15 times for S1L and S5L, re-spectively. Such an increase cannot only be attributed to portlandite
dissolution but also to C-S-H dissolution. The gel pores of C-S-H whichhave a higher SSA compared to micro-/mesopores are more easily ac-cessible by nitrogen after leaching as also observed by Thomas et al.[43]and Trapote-Barreira et al.[44].
Leaching seriously increased the porosity in the mesopore region inall leached samples as shown in Table 4. The largest increase in porositywas for sample S4L which showed the largest porosity increase with afactor of almost three. The samples with 20% limestone ller replace-
ment (S2L, S5L) had similar porosity after leaching despite different ini-tial porosities. Interestingly, the average pore diameter determined byBJH method decreased after leaching for all compositions. Rather thana renement of the pore structure of the leached materials (a rene-
ment is inconsistent with the MIP results, see further), this reductionis an additional indication of opening of initially inaccessible gel pores.Sample S1Land S5L (low w/p ratio) exhibited the largestdecrease in av-erage pore diameter, which could be the reason for obtaining the
highest SSA increase (relative values) after leaching.Table 5presents the change in micropore structure of the leached
materials. The micropores of all leached samples were strongly affectedby leaching in ammonium nitrate solution; among them, the highest al-
teration was again from the samples S1L and S5L. Both micropore sur-face area and micropore volume were signicantly increased duringaccelerated leaching. The meso/micro volume ratio was reduced after
leaching which can serve as an evidence for C-S-H decalcicationwhich resulted in higher accessibility of gel pores.
5.3.2. Combined MIP and N2-adsorption results
MIP with leached samples is a delicate measurement becauseleaching decreases elasticity and increases ductility. The leaching re-duces the Ca/Si ratio of C-S-H (seeSection 5.5.2), which decreases the
stiffness of C-S-H evidenced by nanoindentation [46]. Furthermore,portlandite leaching leads to a more homogeneous matrix with lessstress resulting in a lower brittleness of thecement matrix [47]. The de-
formation at failure of leached samples can increase with a factor of intension and 2 in uniaxial compression [47]. As a sample is under isotro-pic compression during MIP, deformation must be higher than that re-ported in [47]. No samples were broken after the MIP, but thecompression effect must still be taken into account in the interpretation
Fig. 8. Comparison of porosity increases due to leaching of samples with different w/p ra-
tios and limestone ller replacements.
Table 6
Changes in porosity, critical and threshold pore diameters and bulk density due to leaching determined by N2-adsorption and MIP combinationLdenotes leached sample.
Sample
w/p
ls/p
S1
0.325
0
S1L
0.325
0
S2
0.425
0.2
S2L
0.425
0.2
S3
0.425
0
S3L
0.425
0
S4
0.375
0.1
S4L
0.375
0.1
S5
0.325
0.2
S5L
0.325
0.2
Accessible porosity, % 11.2 27.8 26.7 44.8 19.7 38.4 16.8 40.5 13.8 35.8
Critical pore diameter, nm 25.7 180 49.9 150 32.0 110 36.4 170 29.0 180
Threshold pore diameter, nm 60 800 300 2000 1000 900 200 800 300 450Bulk density, g/cm3 1.89 1.39 1.72 1.10 1.78 1.11 1.85 1.14 2.06 1.43
8 Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxxxxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
9/18
because the porosimeter interprets sample compression as occurrenceof pores (applied correction method is described inAppendix 1).
An example of the correction procedure in (Fig. 6) shows the cumu-lative intrusion curves and differential pore size distributions (PSD) of
leached sample S1L before and after correcting for the compression ef-fect. It can be seen that the pore volume is compressed by 33% whichcannot be disregarded. However, even with compression correction,the MIP corrected results cannot be considered as the complete pore
structure of the tested sample because the pore volume of the sampleis always higher than the mercury intruded volume due to pore volumecompression under pressure. The difference is larger when the pressureincreases. In the context of this study, the corrected and uncorrected
curves started diverging more and more at pressure larger than approx-
imately 30 MPa (corresponding to a pore diameter of 0.05 m).
Furthermore, N2-adsorption enables to measure the pores up to0.07m. For these reasons, the combination of MIP and gas adsorptionmeasurements will give a better understanding of the pore structureof cementitious materials, especially for leached pastes. The MIP gives
reasonable results for pore size larger than 0.05 m, while N2-adsorption provides information on the smaller pores. In this way,pore structure information is obtained in a broader pore size rangeand compression effect is reduced. For reference samples, the pores
with diameter larger than 0.007 m was determined by MIP and smallerpores was determined by N2-adsorption. Notethat the pore informationobtained by MIP and N2-adsorption is not identical because differentmethods/models are applied to calculate the porosity and PSD. The po-
rosity calculated by N2-adsorption was shifted at the combined pore di-
ameter (i.e. 0.05m) in order to obtain a continuous curve.
Fig. 9.XRD patterns of reference and leached samples taken in the rst 3 mm from the reactive surface.
9Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxx xxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
10/18
Fig. 7compares the porosity and PSD of leached and reference sam-ples determined by the combined MIP and N2-adsorption method. Inthe range of 0.003100m, the porosity of leached samples increasedfrom 68% up to 159% compared to the porosity of reference samples.
Such a porosity increase is larger than the initial volume of portlanditewhich has a maximum value of about 12.8% for sample S1[48]. Afterleaching, sample S1L had the smallest porosity, while sample S2L had
the largest porosity. This may be attributed to different initial porosities
because of different w/p ratio and LS replacements. However, samplesS3L, S4L, and S5L exhibited a similarnal porosity after leaching despitetheir differences in w/p ratio and LS replacement (Fig. 8).
Despite the difference in w/p ratios in composition (with/withoutlimestone llers), the cumulative pore volume curves of all leachedsamples exhibit two stages of signicant porosity increase. This is alsoindicated in the differential PSD plots (Fig. 7, right) with two(S1L, S5L) or three (S2L, S3L, S4L) peaks. The rst stage was in the
range of about 60500 nm pore sizes, which corresponds to portlanditedissolution. The second stage was for pore sizes smaller than 10 nmwhich corresponds to C-S-H (andother phases) dissolution. In contrast,the reference samples exhibited only one signicant pore volume in-
crease stage. The critical pore size (which was determined from the av-erage of two peaks with large pore sizes forS2L, S3L, and S4L; while thepeak with large pore size for S1L and S5L) was signicantly shifted tobigger pore size range. The threshold pore diameter tended to increasein most leached samples, exceptfor S3Land S5Lof which the threshold
pore diameters remained almost unchanged. In most cases, the bulkdensities of the leached samples largely decreased after leaching(Table 6). Sample S4L had the largest decrease in bulk density causedby the largest Ca removal as shown inTable 6.
5.4. Phase changes via XRD/QXRD
Fig. 9presents the XRD patterns for leached and reference samples.Note that leached sample refers to a subsample taken from theleached zone (pH b 9) of the sample. Portlandite was completely dis-
solved in all leached samples as the two intense peaks at 2= 17.9and 34.2o for the reference samples were absent in the leachedsamples.
On the other hand, the most visible phases in the XRD patterns of theleached samples are calcite, ettringite, and unhydrated cement, which
indicate that they are hardly degraded by the NH4NO3solution. Notethat the intense peaks for calcite in reference and leached samplesS2L, S4L, and S5L are due to the limestone ller addition. Similar toleaching in deionized water [49,50], no new crystalline phases were ob-
served. This underpins the statement that leaching in ammonium ni-trate results in same end-products (solid phases), as obtained for
naturalleaching processes.The quantitative XRD results of 28-day leached and reference sam-
ples S3, S4, and S5 are shown inTable 7.Among 15 quantied phases,portlandite and calcite were the most occurring phases in the mixtures.The results showed that unhydrated phases, mainly alite, were stillpresent in both leached and intact materials. There was no portlandite
detected within 6 mm depth from the reactive surface, which is consis-tentwith the degraded depth determined by phenolphthalein spraying.The quantitative XRD analysis (Fig. 10) alsoshows a gradualportlanditeleaching front beyond the phenolphthalein degraded depth (at 9.25,
7.5, and 5.9 mm for S3L, S4L, and S5L, respectively).Although it is expected that the absolute amount of calcite should
remain constant (or slightly decreased) because calcite leaching is neg-ligible, the relative amount increased toward the exposed surface in
Fig. 10as leaching decreased the bulk density. The relative amount ofcalcite provides information to calculate the relative change in bulkdensity along the sample as described inAppendix 2.Fig. 11presentsthechanges in bulk density with depth in samples S4Land S5L. Thecal-
cite amount in sample S3 (without limestonellers) was too small toallow for reliable estimations of bulk density. The estimated reductions
in bulk density in the 0
3 mm interval (43% and 30% for samples S4L Table
7
Q
uantitativeXRDresultsofleachedandreferencesam
plesS3,S4,andS510%internalstandardZnOwas
addedtothemixtures.
Sample
Depthinterval,mm
Portlandite
Calcite
Ettringite
Belite
Alite
Tricalcium
aluminate
Brownmillerite
Calcium
monocarboaluminate
Tober
morite
Clinotobermorite
Jennite
Stratlingite
Thaumasite
Anhydrite
S3
Ref.
16.7
2.4
0.2
0.0
1.4
0.0
0.4
0.8
0.1
0.9
0.2
0.0
0.5
0.1
S3L
03
0.0
5.0
0.5
0.6
1.2
0.0
0.5
0.5
0.1
1.3
0.2
0.0
0.9
0.1
36
0.0
2.6
0.8
0.8
1.2
0.0
0.3
0.3
0.2
1.2
0.3
0.0
1.1
0.2
69
0.6
1.9
0.6
2.0
2.3
0.0
0.3
0.4
0.1
0.9
0.1
0.0
2.8
0.7
912
12.3
1.4
0.5
0.9
1.5
0.0
1.5
0.7
0.2
0.8
0.3
0.0
2.2
0.5
S4
Ref.
17.7
6.7
0.3
0.0
1.4
0.4
0.4
1.2
0.1
1.4
0.2
0.0
0.4
0.0
S4L
03
0.0
17.1
0.7
0.0
1.2
0.0
0.4
0.0
0.1
0.7
0.2
0.0
0.3
0.0
36
0.0
13.4
1.0
0.5
1.4
0.0
0.2
0.2
0.1
1.2
0.5
0.0
0.6
0.0
69
1.0
11.0
0.6
0.0
2.7
0.0
0.2
1.3
0.0
0.9
0.1
0.1
3.1
0.0
912
9.9
8.2
0.5
0.5
3.1
0.0
0.3
1.1
0.2
1.0
0.3
0.2
1.5
0.4
S5
Ref.
15.3
13.1
0.5
0.2
1.0
0.0
1.0
0.0
0.2
0.1
0.2
0.0
0.1
0.0
S5L
03
0.0
21.8
0.6
0.8
1.1
0.1
0.5
0.7
0.1
0.9
0.2
0.0
0.2
0.4
36
0.0
19.9
0.7
0.4
0.6
0.0
0.2
0.2
0.1
0.7
0.1
0.0
0.2
0.1
69
4.7
15.6
0.5
0.4
2.3
0.0
1.2
0.7
0.2
1.3
0.2
0.0
0.5
0.1
912
13.8
12.2
0.8
0.0
3.4
0.0
0.8
0.7
0.1
1.5
0.1
0.0
0.7
0.0
10 Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxxxxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
11/18
and S5L, respectively) were similar to the MIP estimations (38% and 31%for S4L and S5L, respectively). The bulk densitygradually increased over
the depth of the sample. Beyond the phenolphthalein degraded depththe bulk density of S5L still decreased 5%, while S4L exhibited a largerreduction (13%).
5.5. SEM results
5.5.1. Microstructural alterations
SEM images (Fig. 12) show, similar to the MIP and N2-adsorption
measurements, an increase in pore sizes and porosity after leaching.Some big pores with diameter up to about 10 m were observed in
most of the leached samples which can be attributed to the portlanditedissolution. The dissolution of C-S-H mainly contributes to the increaseof pore volume in the micropore to lowmesopore range whichis difcultto detect by SEM. Micro cracks were observed in all leached samples,
especially around the limestone llers (S2L, S4L, S5L) and unhydrated ce-mentparticles.Decalcication of C-S-H inducespolymerization shrinkage,especially when the Ca/Siratio is lower than 1.2 [10]. In addition, leachinginduces a signicant loss of mechanical properties. As a consequence, de-
calcication shrinkage could generate tensile stresses around rigid parti-cles (e.g. limestone llers, unhydrated cements) which results incracking. The decalcication shrinkage even can generate differentialstresses in the region where a gradient of Ca/Si ratio exists[10].
A third element in the microstructural changes after leaching is anincreased connectivity in the XY direction (examined on 2-D SEM
images). Creating connected pathways during leaching of Ca and otherleachable elements can increase the percolation of the pore systemand, consequently, transport properties.
5.5.2. Modication in atomic Ca/Si ratio
An EDX measurementcrossing the transition zone of leached sampleS1L as indicated by the red arrow inFig. 13(top-left) quantied the CaandSi amount. The atomic Ca/Si ratio was obtained fromthe normalizedmass percentage of Si and Ca. It was found that the average Ca/Si ratio
was approximately double in the intact area compared to the degraded
area. The sudden jump in Ca/Si ratio at distance of about 220 m wasdue to scanning through a Ca-rich region.
Fig. 13 (top-right) shows the spatial distribution of Ca, whileFig. 13(bottom-right) shows the combined spatial distribution of fourmain el-ements (Si, Ca, Al, and Fe) in the transition zone. The intact area clearlyhad a higher amount of Ca indicated by the more intense green color.
Other elements (Si, Al, Fe) seemed to be equally distributed, which indi-cates that those elements did not undergo signicant leaching. Microcracking was also observed in the degraded part in both SEM image(top-left) and mapping picture (bottom-right). A similar change in ele-
ment distribution was also found for S4 and S5 (results not shown).Detailed proles of the Ca/Si ratio were obtained from element line
scans on 28-day leached samples with scan intervals of 9.9, 6.1, and0.5m for S3L, S4L, and S5L, respectively. Smaller scan interval gave nois-
ier results as seen in Fig. 14. However, thescan results may be better withlowscan interval, especially in the transition zones. It is conrmed that C-
Fig. 10.Portlandite (CH) and calcite (CC) proles in the leached samples S3L, S4L, and S5L after 28 days of leaching in ammonium nitrate solution.
Fig. 11.Changes in bulk density over the depth of leached samples S4L and S5L after 28 days of leaching in ammonium nitrate solution.
11Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxx xxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
12/18
S-H is dissolved in part of the samples as its Ca/Si ratios are below the Ca/Si ratio in initial C-S-H (roughly 1.7dotted line in Fig. 14).In front ofthephenolphthalein-identied degraded depth, the atomic Ca/Si ratiosslightly increased over the depth with lowest average atomic Ca/Si ratio
of about 1, 1.1, and 1.2 for S3L, S4L, and S5L, respectively, close to the re-active surface. These Ca/Si ratios dropped in the range where high decal-cication shrinkage occurs. As a result, micro cracking was observed in
SEMimages. In thetransitionzones of about1 mm,Ca/Si ratios rapidlyin-
creased toward the intact area. Beyond the transition zones, the Ca/Siratio still increases slightly which is in line with the quantitative XRD re-sults which showed a portlandite decrease beyond the degraded zone.
5.6. Change in water permeability
The permeability of sample S3L increased as a function of immersiontime in ammonium nitrate solution (Fig. 15). The permeability in-
creased with proceeding leaching time, hence progressing leachingdepth. After 1 week of NH4NO3immersion, the permeability coefcientincreased 6 times, from 2.3 1020 to 1.4 1019 m2. It continued in-
creasing up to 3.2 1019 m2 and 1.4 1018 m2 after 2 weeks and
8 weeks NH4NO3immersion, respectively.It is important to note that the decalcication of the sample was notuniform with depth, and thus, the permeability represents a composite
Fig. 12.SEM backscattered electrons images of reference (top-left) and leached (top-right) samples; and increases in porosity (in black), pore size, and percolation due to leaching
(bottom)eld width of 128m.
12 Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxxxxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
13/18
permeability of different permeabilities in series corresponding to dif-ferent degrees of degradation. The intrinsic permeability of samplesleached during 28 days increased one to two orders of magnitude de-pending on w/p ratio and composition (Table 8). Sample S1L with the
lowest w/c ratio (0.325) exhibited an increase in permeability by a fac-tor of 6, while sample S2L with the highest w/c ratio (0.531) increasedby a factor of 235. The extent of permeability increase is correlatedwith the porosity and PSD after leaching. As seen inFig. 8, the porosity
is the largest and the PSD is the coarsest for sample S2L and vice versafor sample S1L. Because the samples S3L, S4L, and S5L differed in de-
graded depth rather than in nal porosity and PSD, the differences in
permeability increase were due to the former. An increase with a factorof 8, 18, 25 was obtained after 28 days of leaching for samples S5L, S4L,and S3L, respectively, with respective degraded depths of 5.9, 7.5, and9.25 mm. For a full leached sample (S5L after 142 days of leaching),
the permeability increased by a factor of 592.The permeability change is not limited to the zone where the pH is
larger than 9 (phenolphthalein indicator). A model with two permeabil-ities in series (unknown permeability in degraded zone and known per-
meability in intact zone) estimated a negative value for the intrinsicpermeability of the leached zone. Therefore, it is assumed that there ex-
ists a relatively thick transition zone over which the permeability
Fig. 12 (continued).
13Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxx xxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
14/18
increased from a value representative for fully leached material to avalue representative for intact hardened cement paste. Because theCa/Si ratio still decreases beyond the degraded depth (Fig. 14), micro-structure and thereby permeability are still expected to be altered inthat zone. However, a numerical model is needed to conrm this as-
sumption and to quantify the depth of the transition zone [48].In general, mass transport properties are mainly affectedby capillary
and large pores[16]. As such, an increase in permeability after leachingis linked to an increase in capillary pores (see MIP results and SEM im-
ages) resulting from portlandite dissolution. In addition, it is
hypothesisedhere that C-S-H decalcication under accelerated leaching
conditions is alsoan important factor because it increases the connectiv-ity of different types of pores (gel, micro- and mesopores). Furthermore,the decalcication shrinkage originating from a polymerization of C-S-Hdueto lower Ca/Siratio inducesmicro cracks as seen inSEM images. The
study of Agostini et al.[28]also conrmed the contribution of C-S-Hleaching to permeability increase. When the sample is completely de-graded (complete dissolution of portlandite) and further immersed inNH4NO3 solution (foran extra 8 days), the intrinsic permeability slightly
increases.
Fig. 12 (continued).
Fig. 13. SEMBSE image of transition zone of leachedsampleS1L:CH = portlandite,CSH = C-S-H phase,C = residual cementclinkers(top-left); Ca/Si ratio along thered arrow (bottom-
left); spatial distribution of Ca (right-top) and element (Si, Ca, Al, Fe) mapping generated by x-ray imagingeld width of 500m.
14 Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxxxxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
15/18
5.7. Change in diffusivity
The diffusion experiments were performed on a 28-day leachedsample S3L. A 1-D diffusive transport model was used to interpret theexperimental data[30]. Like permeability, the effective diffusion coef-
cient of the leached sample should be treated as the composite (overall)effective diffusion coefcient. The porosity of the leached sample, whichis one of inputs of the transport model, is not constant over the samplelength. The average porosity of the leached sample was calculated by
weighted average method as follows:
aldl re f ddl
d 6
wherea,l, andref[] denote the average porosity of the leachedsample, the porosityof the leached zone, and the porosities of therefer-
ence sample, respectively; dl and d [m] denote the thicknesses ofleached zone and the sample, respectively, as determined by phenol-
phthalein. Slow hydration processes which may decrease porosity dur-ing long-time experiments were not taken into account in thecalculation. Fig. 16 shows the tted and measured partial pressure pro-
les of He and Xe at the outlet for sample S3L. The experimental diffu-
sion proles were well described by the diffusion model. The Heeffective diffusion coefcient of S3L was 9.48 1011 m2/s, which is4.1 times higher than the effective diffusion coefcient of the referencesample S3. Compared to the increase of permeability, the increase in dif-
fusion due to leaching is much lower because permeability is much
more sensitive to porosity change and the development of percolatedpore networks, and also to cracking compared to diffusivity.
The Xe effective diffusion coefcient of the leached sample S3L was
8.58 1012 m2/s, which is much smaller thanthe one of He. The effec-tive diffusion coefcient of Xe for S3L is almost double compared to theestimated effective diffusivity[48]of intact material (the measured ef-fective Xe diffusivity for the reference sample S3 was not obtained due
to very long time experiment). It is important to note that the changein He diffusivity is different of that of Xe. One possible explanation forthis observation is the difference in molecular size of He (0.26 nm)and Xe (0.46 nm). Leaching modied the microstructure of cement
paste from micrometer scale down to subnanometer scale as observedby MIP and N2-adsorption experiments. Therefore, the change of poreswith smaller diameter likely affects the diffusion of He (with smaller
molecular size) more than the diffusion of Xe.
6. Conclusions and remarks
In this study, anammonium nitrate solutionof 6 M was used to accel-erate the decalcication kinetics of hardened cementpaste samples. A va-riety of techniques including IC, XRD and quantitative XRD, SEM/SEM-
EDX, MIP, and N2-adsorption were used to characterize the microstruc-tural and mineralogical changes of cement pastes with different mixturesafter leaching. The effect of accelerated leaching on transport behaviorwas studied by measuring changes in the water permeabilityand diffusiv-
ity of dissolved gases.Results showed that NH4NO3solution was a reactive agent which
can be used to accelerate leaching kinetics while still keeping the na-ture of the leaching process. The mineralogical changes under leaching
in ammonium nitrate solution were quite similar to leaching in purewater as no new phase was formed during leaching in ammonium ni-trate solution. In this way, we can imitate long-term naturally degradedcementitious materials by accelerated leaching using ammonium ni-
trate solution. However, deeper investigations in the changes of miner-alogy are still needed especially when Ca/Siis lower than 1. The square-root-time law of propagation of the leaching front and leached Caamount was applicable under accelerated conditions indicating diffu-
sive transport conditions. Cement mixture had a signicant impact onleaching; a higher w/pratio or limestone llerreplacement gave a largerrate of leaching propagation (degraded depth). The effects of w/p ratiowere more signicant than limestoneller replacement. The compari-
son of mass change calculated from IC results and direct measurements(weighing) proved that C-S-H leaching contributes to the mass loss of
the sample.
Fig. 14.Decrease inatomic Ca/Si ratio of sample: (a) S3L due to leachingscan interval was 9.9 m; (b) S4L due to leachingscan interval was 6.1m; and (c) S5L due to leachingscan
interval was 0.5m; the dotted line denotes the Ca/Si ratio in initial C-S-H.
Fig. 15.Change in intrinsic permeability of leached sample S3L due to leaching in 6 mol/l
NH4NO3as a function of immersed time.
15Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxx xxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
16/18
A combination of MIPnitrogen adsorption analysis was used to ob-tain information on the porosity and pore size distribution changes. The
accelerated leaching highly altered the microstructure of the cementpaste to a more porous material which is evidenced by the increasesof specic surface area and total porosity andby thecreation of a coarserpore size. The total accessible porosity of the leached samples highly
depended on w/c ratio. Beside the increase in meso volumes, it is inter-esting to observe that the micropore volume was altered much morethan the mesopore volume.Probably, the leaching of C-S-H with a larger
volume fraction than portlandite results in a signicant modication ofthe micropore structure. The bulk density was signicantly reduced dueto Ca removal (up to 42%). It was not uniform but increased over thedepth of leached zone, which indicates a gradual alteration in mineral-
ogy. Quantitative XRD indicates that portlandite was completely de-
graded in the leached zone while C-S-H was partially dissolved asevidenced from microstructural analysis. Beyond the leached front,part of the portlandite was also dissolved as conrmed by quantitative
XRD and SEM-EDX.SEM image analysis showed a signicant increase in porosity and
pore size for the leached materials. Some big pores (up to 10 m)were observed. The decalcication shrinkage resulted in formation of
micro cracks. More micro cracks were observed in samples with lime-stone ller. The Ca-leaching created connected pathways resulting inan increase of the percolation of the pore system. In the leached zone,the atomic Ca/Si ratio was gradually increased toward the degraded
front with a jump in Ca/Si ratio indicative for the transition zone.The changes in microstructures and mineralogy led to a signicant
increase in transport properties. The permeability increased by one to
two orders of magnitude depending on the immersion time inNH4NO3and w/c ratio. For similar w/c ratio (S3, S4), the addition oflimestone helps to reduce the permeability alteration. The extent of per-meability increase is correlated to the porosity and pore size distribu-tion after leaching. The permeability of the non-invasion zone
(determined by phenolphthalein) was probably altered as the Ca/Siratio still decreased beyond the degraded depth. The diffusivity alsosig-nicantly increased after leaching, but with smaller magnitude
compared to the change in permeability (e.g. factor of 24 for water per-meability compared to factor 4 for He diffusivity of sample S3L).
Acknowledgements
Thenancial support by Belgian Nuclear Research Centre (SCKCEN)
is gratefully acknowledged.
Appendix 1. Correction procedure for compressioneffectduring MIP
measurements.
In order to account for the deformation of the sample, the measure-ment needs to be corrected by subtracting the volume of sample com-
pression as:
vc vi 1
b
1
a
7
wherevcis specic compression volume [m3/kg],viis specic intruded
volume [m3/kg],bis bulk density [kg/m3], andais the apparent den-
sity which is dened as unit weight of sample at the end of the lling
process [kg/m3]:
a ms
VbVcori
8
where msis sample mass [kg], Vb is sample volume (including pores)[m3],Vi
cor is corrected intruded volume [m3] and can be calculated asfollows:
Vcori vivc ms 9
Appendix 2. Change in bulk density due to leaching
The following derivations enable to understand the variation of bulkdensity of leached sample over the depth. The mass fractions of calcite
Fig. 16. Comparisonof theHe (a) andXe (b)partial pressure at theoutlet obtained from experiment andmodelfor carbonatedsampleS3L after 28 days of leaching in ammonium nitrate
solution.
Table 8
Intrinsic permeability of cement pastes before and after leaching sample S5L-full was measured after 142-day leaching when the sample was fully decalcied (determined by phenol-
phthalein), the others were measured after 28-day leaching.
Sample S1 S1L S2 S2L S3 S3L
Permeability, m2 1.5 1021 9.1 1021 1.1 1019 2.6 1017 2.3 1020 5.7 1019
Sample S4 S4L S5 S5L S5L-full
Permeability, m2 1.4 1020 2.5 1019 2.0 1020 1.6 1019 1.2 1017
16 Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxxxxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.0177/24/2019 Leaching Phung
17/18
in reference,CCref[], and in leached,CCl[], materials are expressedas in Eqs.(10) and (11):
CCre f mCC
msolidre f ma
re f
10
CCl mCC
msolidl ma
l
11
wheremCCis the absolute mass of calcite [kg]; mrefsolid andml
solid are thetotal mass of solid phases in reference and leached materials, respec-tively [kg]; andmref
a andmla are the mass of amorphous phases in refer-
ence and leached materials, respectively [kg]. Note that XRD onlydetects crystalline phases; therefore, the denominators of Eqs.(10)and (11)are the mass of crystalline phases instead of the total mass ofsolid phases. With the assumption that there is no shrinkage during
leaching, the total volumes of material before and after leaching areidentical. Thus, the ratio between the mass fractions of calcite in refer-ence and leached materials is related to the bulk densities and amor-phous fractions of the reference and leached materials as expressed in
Eq.(12):
CCre fCCl
msolid
l ma
l
msolidre f ma
re f
1 mal
msolidl
1mare f
msolidre f
msolid
l
msolidre f
CCre fCCl
1Al1Are f
l V
solidl V
porel
re f Vsolidre f V
porere f
1Al1Are f
lre f
12
whereAl andArefare themass fractions of amorphous phases in leachedand reference materials, respectively [];Vref
solid andVlsolid are the vol-
umes of solid phases in reference and leached materials, respectively[m3]; Vref
pore and Vlpore are thepore volumes in reference and leached ma-
terials, respectively [m3]; andrefandlare the bulk densities of refer-ence and leached materials, respectively [kg/m3]. The reduction inbulk density due to leaching can be written as follows:
re fl
re f100% 1
CCre f 1Are f
CCl 1Al
100% 13
References
[1] D. Jacques, L. Wang, E. Martens, D. Mallants, Modellingchemicaldegradationof con-crete during leaching with rain and soil water types, Cem. Concr. Res. 40 (2010)13061313.
[2] K. Yokozeki, K. Watanabe, N. Sakata,N. Otsuki, Modeling of leaching fromcementitiousmaterials used in underground environment, Appl. Clay Sci. 26 (2004) 293308.
[3] H. Saito, A. Deguchi, Leaching tests on different mortars using accelerated electro-chemical method, Cem. Concr. Res. 30 (2000) 1815 1825.
[4] K. Haga, S. Sutou, M. Hironaga, S. Tanaka, S. Nagasaki, Effects of porosity on leachingof Ca from hardened ordinary Portland cement paste, Cem. Concr. Res. 35 (2005)17641775.
[5] J. Jain, N. Neithalath, Analysis of calcium leaching behavior of plain and modied ce-ment pastes in pure water, Cem. Concr. Compos. 31 (2009) 176 185.
[6] A. Bertron, J. Duchesne, G. Escadeillas, Accelerated tests of hardened cement pastesalteration by organic acids: analysis of the pH effect, Cem. Concr. Res. 35 (2005)155166.
[7] L. De Windt, P. Devillers, Modeling the degradation of Portland cement pastes bybiogenic organic acids, Cem. Concr. Res. 40 (2010) 1165 1174.
[8] E.J. Butcher, J. Borwick, N. Collier, S.J. Williams, Long term leachate evolution duringow-throughleaching of a vault backll (NRVB), Mineral. Mag. 76 (2012) 30233031.
[9] C. Carde, G. Escadeillas, A.H. Franois, Use of ammonium nitrate solution to simulateand accelerate the leaching of cement pastes due to deionized water, Mag. Concr.Res. 49 (1997) 295301.
[10] J .J. Chen, J.J. Thomas, H.A. Jennings, Decalcication shrinkage of cement paste, Cem.Concr. Res. 36 (2006) 801809.
[11] K. Wan, L. Li, W. Sun, Solidliquid equilibrium curve of calcium in 6 mol/L ammoni-um nitrate solution, Cem. Concr. Res. 53 (2013) 4450.
[12] T. de Larrard, S. Poyet, M. Pierre, F. Benboudjema, P. Le Bescop, J.-B. Colliat, J.-M.Torrenti, Modelling the inuence of temperature on accelerated leaching in ammo-nium nitrate, Eur. J. Environ. Civ. Eng. 16 (2012) 322335.
[13] V.H. Nguyen, H. Colina, J.M. Torrenti, C. Boulay, B. Nedjar, Chemo-mechanical cou-pling behaviour of leached concrete: Part I: Experimental results, Nucl. Eng. Des.237 (2007) 20832089.
[14] H. Yang, L. Jiang, Y. Zhang, Q. Pu, Y. Xu, Predicting the calcium leaching behaviorof cement pastes in aggressive environments, Constr. Build. Mater. 29 (2012)8896.
[15] S. Poyet, P. Le Bescop, M. Pierre, L. Chomat, C. Blanc, Accelerated leaching of cemen-titious materials using ammonium nitrate (6 M): inuence of test conditions, Eur. J.
Environ. Civ. Eng. 16 (2012) 336
351.[16] C. Perlot, J. Verdier, M. Carcasses, Inuence of cement type on transport propertiesand chemical degradation: application to nuclear waste storage, Mater. Struct. 39(2006) 511523.
[17] C. Gall, H. Peycelon, P.L. Bescop, Effect of an accelerated chemical degradation onwater permeability and pore structure of cementbased materials, Adv. Cem. Res.16 (2004) 105114.
[18] M. Berra, F. Carassiti, T. Mangialardi, A.E. Paolini, M. Sebastiani, Leaching behaviourof cement pastes containing nanosilica, Adv. Cem. Res. (2013) 352 361.
[19] I. Segura, M. Molero, S. Aparicio, J.J. Anaya, A. Moragues, Decalcication of cementmortars: characterisation and modelling, Cem. Concr. Compos. 35 (2013) 136 150.
[20] A. Cheng,S.J. Chao, W.T. Lin, Effects of leaching behavior of calciumions on compres-sion and durability of cement-based materials with mineral admixtures, Materials 6(2013) 18511872.
[21] K. Haga, M. Shibata, M. Hironaga, S. Tanaka, S. Nagasaki, Change in pore structureand composition of hardened cement paste during the process of dissolution,Cem. Concr. Res. 35 (2005) 943950.
[22] Q.T. Phung, N. Maes, D. Jacques, E. Bruneel, I. Van Driessche, G. Ye, G. De Schutter,Effect of limestonellers on microstructure and permeability due to carbonation
of cement pastes under controlled CO2 pressure conditions, Constr. Build. Mater.82 (2015) 376390.
[23] S. Liu, P. Yan, Effect of limestone powder on microstructure of concrete, J. WuhanUniv. Technol.-Mater. Sci. Ed. 25 (2010) 328331.
[24] B. Lothenbach, G. Le Saout,E. Gallucci, K. Scrivener, Inuenceof limestone on thehy-dration of Portland cements, Cem. Concr. Res. 38 (2008) 848 860.
[25] G. Ye, X. Liu, G. De Schutter, A.M. Poppe, L. Taerwe, Inuence of limestone powderused as ller in SCC on hydration and microstructure of cement pastes, Cem.Concr. Compos. 29 (2007) 94102.
[26] M. Ghrici, S. Kenai,M. Said-Mansour,Mechanicalproperties anddurability ofmortarand concrete containing natural pozzolana and limestone blended cements, Cem.Concr. Compos. 29 (2007) 542549.
[27] M.E.-S.I. Saraya, Study physico-chemical properties of blended cements containingxed amount of silica fume, blast furnace slag, basalt and limestone, a comparativestudy, Constr. Build. Mater. 72 (2014) 104112.
[28] F. Agostini, Z. Lafhaj, F. Skoczylas, H. Loodsveldt, Experimental study of acceleratedleaching on hollow cylinders of mortar, Cem. Concr. Res. 37 (2007) 71 78.
[29] Q.T. Phung, N. Maes, G. De Schutter, D. Jacques, G. Ye, Determination of water per-meability of cementitious materials using a controlled constant ow method,Constr. Build. Mater. 47 (2013) 14881496.
[30] Q.T. Phung, N. Maes, D. Jacques, E. Jacop, A. Grade, G.D. Schutter, G. Ye, Determina-tion of diffusivities of dissolved gases in saturated cement-based materials, Interna-tional Conference on Concrete Repair, Rehabilitation and Retrotting, ICCRRR 2015,Leipzig, Germany, 2015.
[31] J.J.P. Bel, S.M. Wickham, R.M.F. Gens, Development of the supercontainer design fordeep geological disposal of high-level heat emitting radioactive waste in belgium,MRS Online Proc. Libr. 932 (2006) 23 32.
[32] B. Craeye, Early-age thermo-mechanical behaviour of concrete Supercontainers forradwaste disposal, Department of Structural engineering, Ghent University, Ghent,2010 376.
[33] F.H. Heukamp, F.J. Ulm, J.T. Germaine, Mechanical properties of calcium-leached ce-ment pastes: triaxial stress states and the inuence of the pore pressures, Cem.Concr. Res. 31 (2001) 767774.
[34] Q.T. Phung, Effect of Chemical Degradation on Microstructural Changes and Trans-port Properties of Concrete Phenomenological Study, 14th FEA PhD Symposium,Ghent University, Belgium, 2013.
[35] RILEM, CPC-18 Measurement of hardened concrete carbonation depth, Mater.
Struct. 21 (1988) 453
455.[36] T. de Larrard, F. Benboudjema, J.B. Colliat, J.M. Torrenti, F. Deleruyelle, Concrete cal-
cium leaching at variable temperature: experimental data and numerical model in-verse identication, Comput. Mater. Sci. 49 (2010) 3545.
[37] C. Gall, Effect of drying on cement-based materials pore structure as identied bymercury intrusion porosimetry: a comparative study between oven-, vacuum-,and freeze-drying, Cem. Concr. Res. 31 (2001) 14671477.
[38] N.C. Collier, J.H. Sharp, N.B. Milestone, J. Hill, I.H. Godfrey, The inuence of water re-moval techniques on the composition and microstructure of hardened cementpastes, Cem. Concr. Res. 38 (2008) 737744.
[39] J. Zhang, G.W. Scherer, Comparison of methods for arresting hydration of cement,Cem. Concr. Res. 41 (2011) 10241036.
[40] R.T. Downs, M. Hall-Wallace, The American mineralogist crystal structure database,Am. Mineral. 88 (2003) 247250.
[41] S.J. Gregg, K.S.W. Sing, Adsorption, Surface Area and Porosity, 2nd ed. AcademicPress, London, 1982.
[42] T.C. Powers, T.L. Brownyard, Studies of the physical properties of hardenedportland cement paste, Research Bulletin, 22, Portland Cement Association,1948.
17Q.T. Phung et al. / Cement and Concrete Research xxx (2015) xxx xxx
Please cite this article as: Q.T. Phung, et al., Investigation of the changes in microstructure and transport properties of leached cement pastesaccounting for mix composition, Cem. Concr. Res. (2015),http://dx.doi.org/10.1016/j.cemconres.2015.09.017
http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0005http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0005http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0005http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0005http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0005http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0010http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0010http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0010http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0010http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0015http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0015http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0015http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0015http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0020http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0020http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0020http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0020http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0020http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0025http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0025http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0025http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0025http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0025http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0025http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0030http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0030http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0030http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0030http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0030http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0035http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0035http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0035http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0035http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0040http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0040http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0040http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0040http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0040http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0040http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0040http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0045http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0045http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0045http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0045http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0045http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0050http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0050http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0050http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0050http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0050http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0050http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0055http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0055http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0055http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0055http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0055http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0055http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0060http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0060http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0060http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0060http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0060http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0060http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0060http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0065http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0065http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0065http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0065http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0065http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0070http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0070http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0070http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0070http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0070http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0075http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0075http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0075http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0075http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0075http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0075http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0075http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0080http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0080http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0080http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0080http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0080http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0080http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0080http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0085http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0085http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0085http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0085http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0085http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0245http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0245http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0245http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0245http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0090http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0090http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0090http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0090http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0090http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0090http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0095http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0095http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0095http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0095http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0095http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0100http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0100http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0100http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0100http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0100http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0110http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0110http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0110http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0110http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0115http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0115http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0115http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0115http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0115http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0115http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0125http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0125http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0125http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0125http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0125http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0130http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0130http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0130http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0130http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0130http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0130http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0135http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0135http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0135http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0135http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0140http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0140http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0140http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0140http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0140http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0140http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0140http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0250http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0250http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0250http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0250http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0250http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0250http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0150http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0150http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0150http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0150http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0150http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0255http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0255http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0255http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0160http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0160http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0160http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0160http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0160http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0160http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0160http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0260http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0260http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0260http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0260http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0260http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0265http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0265http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0265http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0265http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0170http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0170http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0170http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0170http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0170http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0170http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0170http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0175http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0175http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0175http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0175http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0175http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0175http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0175http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0180http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0180http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0180http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0180http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0180http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0180http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0180http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0185http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0185http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0185http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0185http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0190http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0190http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0190http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0190http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0195http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0195http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0270http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0270http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0270http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://dx.doi.org/10.1016/j.cemconres.2015.09.017http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0270http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0270http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0270http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0195http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0195http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0190http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0190http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0185http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0185http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0180http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0180http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0180http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0175http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0175http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0175http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0170http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0170http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0170http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0265http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0265http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0260http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0260http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0260http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0160http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0160http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0160http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0255http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0255http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0255http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0150http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0150http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0150http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0250http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0250http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0250http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0250http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0140http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0140http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0140http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0135http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0135http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0130http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0130http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0130http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0125http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0125http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0125http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0120http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0115http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0115http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0110http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0110http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0105http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0100http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0100http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0100http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0095http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0095http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0095http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0090http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0090http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0245http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0245http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0085http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0085http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0085http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0080http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0080http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0080http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0075http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0075http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0075http://refhub.elsevier.com/S0008-8846(15)00255-0/rf0070http://refhub.elsevier.com/S0008-8846(Top Related