Cement & Concrete Composites 28 (2006) 427431

www.elsevier.com/locate/cemconcomp

Inuence of alkalis on porosity percolation

in hydrating cement pastes
D.P. Bentz

Materials and Construction Research Division, National Institute of Standards and Technology, 100 Bureau Drive Stop 8615,
Gaithersburg, MD 20899-8615, USA
Received 17 October 2005; received in revised form 9 January 2006; accepted 15 January 2006
Available online 28 February 2006

Abstract
Loss-on-ignition (LOI) measurements and low temperature calorimetry (LTC) are used to study the properties of hydrating cement
pastes with various quantities of alkalis. In addition to the well-known acceleration of early age hydration and retardation of later age
hydration, the alkalis are observed to have a signicant eect on the percolation of the porosity in the hydrating systems, as assessed
using the LTC technique. At equivalent degrees of hydration, the capillary pores in cement pastes with sucient added alkalis may depercolate while those in lower alkali cement pastes remain percolated. A simple dissolution/precipitation three-dimensional microstructural
model is applied to examine the potential eects of hydration product morphology (random, needles, and plates or laths) on pore space
percolation. The model suggests that the observed experimental results could be consistent with the higher alkali levels modifying the
morphology of the CSH gel to produce more lath-like hydration products, as has been observed by others previously using electron
microscopy. Potential implications for the transport properties and durability of these materials are discussed.
2006 Elsevier Ltd. All rights reserved.
Keywords: Acceleration; Alkalis; Hydration; Low temperature calorimetry; Percolation

1. Introduction
The alkali content of cements has long been a subject of
interest, due both to the role of alkalis in deleterious alkalisilica reactions (ASR) and also to their inuence on the
hydration and microstructure of cement paste. It is generally accepted that potassium and sodium ions (present
along with either sulfates or hydroxides) accelerate the
early hydration of cement, while resulting in reduced
hydration and strength at later ages [1,2]. Electron microscopy observations have suggested that the alkalis also
modify the morphology of the calcium silicate hydrate gel
(CSH) formed during hydration, leading to the formation of plate (lath)-like gel hydration products [3,4]. The
inuences of alkalis and the altered morphology of the
CSH gel on transport properties and durability have
*

Tel.: +1 301 975 5865; fax: +1 301 990 6891.

E-mail address: dale.bentz@nist.gov

0958-9465/$ - see front matter 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cemconcomp.2006.01.003

been studied much less. Recently [5], low temperature calorimetry (LTC) results have indicated that after about
100 d of saturated curing, cement pastes with a water-tocement ratio (w/c) of 0.40 by mass fraction, containing signicant additional alkalis, exhibited a depercolated (disconnected) capillary pore structure, while the capillary
pores in the original (low alkali) cement paste were still
highly percolated. This paper presents further studies that
have been conducted to verify this observation, using a
combination of LTC and loss-on-ignition (LOI) experimental techniques.
2. Experimental
Cement pastes with a w/c of 0.40 were prepared by mixing
cement and concrete reference laboratory (CCRL) prociency cement sample 140 [6] with water at 20 C, using
a high speed blender. The mixing water was either distilled water or a solution of alkalis (sulfates or hydroxides),

428

D.P. Bentz / Cement & Concrete Composites 28 (2006) 427431

prepared by adding the appropriate sodium and potassium

compounds to distilled water and stirring with a glass rod
until complete dissolution. Cement 140 is a low-alkali
cement, containing only 0.093%Na2O and 0.186%K2O per
unit mass of cement [6]. For preparing the cement paste with
additional alkali in the form of sulfates, 0.76%Na2SO4 and
0.93%K2SO4 per unit mass of cement were added to the mixing water. This provides a total alkali content of 0.42%Na2O
and 0.69%K2O per unit mass of cement. These contents, and
all others prepared in this study, are thus well within the
range observed in commercially available portland cements,
as for example CCRL prociency cement samples 141 and
142 contained Na2O contents of 0.4% and 0.31%, and
K2O contents of 1.04% and 1.08%, respectively [7].
Because the additional sulfates will also modify the
phase assemblage produced during cement hydration, additional alkalis were also introduced to the cement paste in
the form of hydroxides. For the alkali hydroxides, to maintain the same molar addition rates of the sodium and
potassium, 0.43%NaOH and 0.60%KOH per unit mass of
cement were added, being sure to account for the quoted
89% purity of the commercially available KOH pellets.
To further investigate the inuence of the level of alkali
addition on the observed results, for the hydroxide additions, cement pastes with 0.14%NaOH/0.20%KOH (low
level of additions) and 0.28%NaOH/0.40%KOH (medium
level of additions) were also prepared. After mixing, cast
wafers (5 g) of the paste were placed in sealed plastic
vials. A small quantity of distilled water was added to the
top of the wafers to maintain saturated curing conditions.
The capped vials were placed in a walk-in environmental
chamber maintained at 20 C. At various ages, specimens
of the pastes were removed from the vials for further
analysis.
Degrees of hydration of the cement pastes, dened as the
mass fraction of the original cement that has participated in
hydration reactions, were assessed using LOI analysis to
measure the non-evaporable water content (WN) as that
quantity removed from the specimens between 105 C and
1000 C, corrected for the LOI of the initial cement powder
[8]. Previously, the expanded uncertainty in the calculated
WN had been estimated to be 0.001 g/g cement [8]. WN values were converted to estimated degrees of hydration based
on the phase composition of the cement and published coefcients for the non-evaporable water contents of the various
hydrated cement clinker phases [9]. Based on a propagation
of error analysis, the estimated uncertainty in the calculated
degree of hydration was 0.004. For a smaller subset of the
specimens, thermogravimetric analysis (TGA) was conducted between 100 C and 1000 C at a scan rate of
10 C/min, with a sample size of about 30 mg.
Small pieces of the hydrated cement pastes were also utilized in the LTC experiments. Sample mass was typically
between 30 mg and 90 mg. For each LTC experiment,
one small piece of the relevant cement paste was surface
dried and then placed in a small open stainless steel pan.
The pan with the sample, along with an empty reference

pan of similar mass to the empty sample pan, was placed

in the calorimeter cell. Using a protocol developed previously [10], a freezing scan was conducted between 5 C
and 55 C at a scan rate of 0.5 C/min. For temperatures between 100 C and 500 C, the equipment manufacturer has specied a constant calorimetric sensitivity of
2.5% and a root-mean-square baseline noise of 1.5 lW.
In this study, the peaks observed in a plot of heat ow
(normalized to the mass of the sample) versus temperature
are assumed to correspond to water freezing in pores with
various size entryways (pore necks). The smaller the pore
entryway, the more the freezing peak is depressed. Thus,
the presence of, absence of, or change in peaks can be used
to infer critical information concerning the characteristic
sizes of the percolated (connected) water-lled pores in
the microstructure of the hydrating cement pastes. Specically, the presence or absence of a capillary pore freezing
peak near 15 C is assumed to indicate the percolation or
depercolation, respectively, of the capillary pores in a
hydrating cement paste [5,10].
The peak at 15 C is taken to indicate the percolation of
the capillary pore network and not simply the water freezing
in the capillary pores located throughout the cement paste
microstructure for the following reasons. In [5], SEM images
obtained after 90 d hydration were segmented into four
phases and the areas occupied by individual two-dimensional capillary pores (the darkest of the four phases) were
determined. The SEM results indicated the presence of a
set of large capillary pores (P7.5 lm2 in area) in a
w/c = 0.35 hydrated cement paste, even though no peak
was present in the range of 15 C in the LTC scans
obtained on equivalent age samples, implying that while
these pores are physically present, they are isolated and
not part of a percolated (capillary) network. In [10], it was
indicated that the disappearance of this peak for cement
pastes with w/c = 0.3 and w/c = 0.4 corresponded to the
expected times of depercolation from previous measurements [8,11]. Finally, Villadsen has shown that LTC produces a pore size distribution that is basically equivalent to
that obtained with mercury intrusion porosimetry (MIP),
when specimens are prepared (dried) equivalently [12]. However, a critical advantage of LTC over MIP and other techniques for assessing pore size and connectivity is that the
specimens may be evaluated without any drying that might
damage the pore structure. LTC studies with variable alkali
content are complicated somewhat by the change in freezing
point depression due to the variable ionic concentration of
the (freezing) pore solution, and this eect must be considered as well. The maximum initial dosages of alkali sulfates
or hydroxides in the cement pastes with additional alkalis
prepared in this study would be expected to depress the
freezing point of bulk water between 1 C and 2 C [13].
3. Results and discussion
The measured LOI-based degree of hydration results for
the ve cement pastes are provided in Fig. 1. As has been

D.P. Bentz / Cement & Concrete Composites 28 (2006) 427431

Temperature (o C)

0.9
0.8

-50

-45

-40

-35

-30

-25

2d
-20

-15

0.7

no alkalis
alkali sulfates
low alkali hydroxides
medium alkali hydroxides
high alkali hydroxides

0.6
0.5

no alkalis

0.4

Low NaOH+KOH

0.3

Medium NaOH+KOH

0.2

High NaOH+KOH

0.1

0.07
0.05
0.03

Na2SO4+K2SO4

0.0
1

10

100

1000

10000

0.01

Time (h)

Fig. 1. Degree of hydration via LOI technique for w/c = 0.40 cement
pastes with and without alkali additions. Each data point is the average of
two specimens taken from a single cast wafer at the indicated age.

Fig. 3. LTC results for CCRL 140 w/c = 0.40 cement pastes with and
without alkali additions cured under saturated conditions for 2 d.

8d

Temperature (o C)

Relative mass (%)

120

115
No alkalis
Alkali sulfates
Alkali hydroxides
105

100
100

200

300

400

500

600

700

800

900

1000

Temperature (oC)

Fig. 2. Relative mass of w/c = 0.40 cement pastes versus temperature for
specimens with the highest level and without alkali additions hydrated
under saturated conditions for 8 d at 20 C.

-50

-45

-40

-35

no alkalis
low alkali hydroxides
high alkali hydroxides

-30

-25

-20

alkali sulfates
medium alkali hydroxides

-15

-10
0.035
0.030
0.025
0.020

Heat flow (W/g)

observed previously [1,2], the addition of alkalis at high

levels accelerates the early age (<7 d) hydration, while
retarding the hydration achieved at later ages (28 d and
beyond). Interestingly, at an age of around 7 d (150
200 h), the ve pastes have similar degrees of hydration
as measured by non-evaporable water content. Specically,
the degrees of hydration of the no additional alkali cement
pastes and those with the highest level of alkali addition are
basically equivalent at 8 d, as are the two values for the low
and medium level additions. Furthermore, as shown in
Fig. 2, the TGA scans for 8 d old specimens of the paste
with no alkali addition and the two pastes with the highest
addition levels are basically identical, suggesting that the
same amounts of bound water exist in the pastes, both in
the CSH gel and other hydration products (removed in
the temperature range of 100400 C) and in the calcium
hydroxide (removed in the temperature range of 400
550 C). The TGA-measured mass loss between 105 C
and 1000 C was essentially identical (0.176 g/g ignited
cement) for the three specimens examined by TGA.
LTC results for the ve pastes after 2 d and 8 d of saturated curing are presented in Figs. 3 and 4, respectively.
The peaks near 15 C indicate the presence of a connected or percolated capillary pore system within the
three-dimensional paste microstructure [10]. For the 2 d
specimens, a prominent peak near this temperature is

110

-10
0.09

Heat flow (W/g)

Degree of hydration

429

0.015
0.010

Fig. 4. LTC results for CCRL 140 w/c = 0.40 cement pastes with and
without alkali additions cured under saturated conditions for 8 d.

observed for all ve pastes. The two pastes with the highest
levels of additional alkalis do indicate some shifting of this
peak to lower temperatures by as much as 23 C, as would
be expected due to the higher ionic concentrations in their
pore solutions. In Fig. 3, the peaks at 40 C to 45 C
indicate dense gel pores (entryways) within the CSH
hydration product [10]. Here, the pastes with additional
alkalis all exhibit larger peaks, consistent with their higher
degree of hydration after 13 d (Fig. 1). The more surprising LTC results are to be found in Fig. 4, where only dense
gel pores are clearly detected for the three specimens with
medium and high levels of alkali additions, while the original paste and the paste with the low addition level of alkali
hydroxides both exhibit three percolated pore structures
corresponding to entryways composed of capillary pores,
open gel pores (peak at 25 C), and dense gel pores
[10]. These results imply that the larger open gel and capillary pores in the pastes with higher alkali levels have been
depercolated by the hydration products, and at a degree of
hydration similar to that achieved in the original paste
without additional alkalis and the one with a low level of
addition, both of which did not yet achieve depercolation.
This dierence in pore space percolation at 8 d for specimens with equal w/c ratios and equal degrees of hydration
suggests a dierence in either the specic volume or the
morphology of the CSH gel hydration products. One possibility would be that the presence of substantial numbers of
alkali ions in the pore solution and their incorporation into

D.P. Bentz / Cement & Concrete Composites 28 (2006) 427431

the CSH gel nanostructure leads to a gel with a higher

total water content (greater degree of swelling). For example, it is well known that the alkali silica gel (similar in
chemical composition to CSH [14]) formed in concretes
undergoing alkali-silica reaction can be quite expansive.
However, in contradiction to this hypothesis, the measured
mass losses between 60 C and 100 C in the TGA measurements (which can provide some indication of the evaporable
gel water) were quite similar for the various pastes, as were
their non-evaporable water contents as indicated previously
in Figs. 1 and 2. The second possibility would be that the
presence of substantial alkalis results in the formation of
a CSH hydration product with a dierent morphology
than that formed in low alkali cement pastes. Previous electron microscopy observations on tricalcium silicate (C3S)
and portland cement pastes [3,4] have indeed indicated a
greater tendency towards plate or lath-like hydration products with a higher degree of crystallinity in the presence of
alkalis. Such a product morphology could also reduce
long-term hydration, due to either a less permeable, more
crystalline layer of CSH forming on the remaining unhydrated cement clinker particles or reduced diusion rates
within the pore space. The issue of whether product morphology could indeed aect the percolation of the capillary
pores is considered next.
The inuence of product morphology on porosity percolation was preliminarily examined using a simple dissolution/precipitation microstructural model. The model was
developed only to consider the possible inuence of reaction product morphology on the percolation/depercolation
of the capillary porosity between the original solid particles (modeled as digitized spheres) in a system that reacts
with water via a dissolution/precipitation mechanism.
The model is similar to the original pixel-based C3S hydration microstructural model developed at NIST [15], but
with the creation of only one type of reaction product that
nucleates and grows as crystals in the available waterlled capillary pore space. Portions of the original particles
dissolve, diuse, and precipitate and grow this reaction
product. In the model, the three-dimensional morphology
of this single reaction product is controlled to be exclusively either one-dimensional needles (one pixel by one
pixel across), two-dimensional plates (one pixel thick), or
random agglomerates. For computational simplicity, the
needles and plates were always oriented along one of the
three principal axes (chosen at random), because the model
is being executed on a three-dimensional volume that consists of cubic elements (voxels). Here, the model was run
with an initial water-to-solid mass ratio of 0.40 and an
expansion factor of 2.15. The expansion factor is dened
as the ratio of the volume of the reaction products to that
of the solid reactants; for a typical portland cement, it has a
value of about 2.15 [8,10,15]. As the reaction proceeds, the
percolation of the pore space is periodically monitored
using a simple burning algorithm [15]. Typical results for
the connected fraction of the capillary porosity as a function of total capillary porosity are provided in Fig. 5.

1.0

Connected fraction

430

Plates

0.8

Needles
0.6

Random

0.4
0.2
0.0
0.20

0.25

0.30

0.35

0.40

0.45

0.50

Capillary porosity

Fig. 5. Dissolution/precipitation model results for capillary porosity

percolation (connected fraction) as a function of capillary porosity for
dierent product morphologies (plates, needles, or random) as indicated in
the legend. Results shown are the average of 10 separate simulations for
each morphology at a w/c = 0.40, with an expansion factor of 2.15.

Clearly, in the systems where the reaction product morphology is limited to either needles or plates, the capillary
porosity is depercolated at a signicantly higher total
porosity (equivalent to a lower degree of hydration). The
formation of either needle or plate structures is seen to
be more ecient at depercolating the capillary pore space
than the formation of a random reaction product. This
simple model thus supports the experimental observation
that the capillary pore space depercolates at a lower degree
of hydration in the pastes with sucient additional alkalis
where plate-like hydration products have been observed to
be preferable [3,4].
This dierence in percolation could potentially have
some inuence on the transport properties and durability
of the cement pastes. For instance, as directly indicated
by the LTC results in Fig. 4, at an age of 8 d, the pastes with
sucient (medium or high) additional alkalis will contain
little if any freezable water for temperatures down to about
40 C, which should result in an improvement in freeze/
thaw durability relative to the lower alkali cement pastes
that contain a measurable amount of water that is freezable
at 20 C and above. Also, as rst indicated by Powers
et al. [11], there is typically a dramatic decrease in measured
uid permeability as the capillary porosity depercolates.
While higher potassium and sodium contents in cements
may be unfavorable due to ASR and other concerns, similar
eects on the depercolation of the capillary porosity might
be achievable by the addition of the lithium admixtures that
are conventionally used to mitigate ASR, as lithium ions
have been observed to be the alkali most preferentially
incorporated into the cement hydrates [16]. Current studies at NIST are being directed at examining the inuence of
lithium additions on achieved hydration and porosity percolation using the techniques presented in this paper.
4. Conclusions
Low temperature calorimetry measurements have indicated that alkalis (present as either sulfates or hydroxides)
can have a signicant inuence on the percolation of the

D.P. Bentz / Cement & Concrete Composites 28 (2006) 427431

capillary porosity in hydrating cement paste, possibly via

their inuence on the morphology of the gel hydration
products, increasing the tendency for the formation of lath
or plate-like hydration products. At equivalent degrees of
hydration (and equivalent ages), pastes with sucient
additional alkalis may exhibit a depercolated capillary pore
network, while those present in equivalent w/c, lower alkali
cement pastes may remain percolated.

[7]

[8]

[9]

References
[10]
[1] Jawed I, Skalny JP. Alkalies in cement: A review II. Eects of alkalies
on hydration and performance of Portland cement. Cem Concr Res
1978;8:3752.
[2] Garci Juenger MC, Jennings HM. Eects of high alkalinity on cement
pastes. ACI Mater J 2001;98(3):2515.
[3] Mori H, Sudoh G, Minegishi K, Ohta T. Some properties of CSH
gel formed by C3S hydration in the presence of alkali. In: Proc of 6th
Int Congress on the Chemistry of Cement, Moscow, vol. 2,
Supplementary Paper, section 2, 1974, p. 212.
[4] Richardson IG. Tobermorite/jennite- and tobermorite/calcium
hydroxide-based models for the structure of CSH: Applicability
to hardened pastes of tricalcium silicate, b-dicalcium silicate, Portland
cement, and blends of Portland cement with blast-furnace slag,
metakaolin, or silica fume. Cem Concr Res 2004;34:173377.
[5] Bentz DP, Stutzman PE. Curing, hydration, and microstructure of
cement paste. ACI Mater J, in press.
[6] Cement and Concrete Reference Laboratory. Cement and concrete
reference laboratory prociency sample program: Final report on

[11]
[12]

[13]
[14]
[15]

[16]

431

Portland cement prociency samples number 139 and 140. Gaithersburg, MD, March 2001.
Cement and Concrete Reference Laboratory. Cement and concrete
reference laboratory prociency sample program: Final report on
Portland cement prociency samples number 141 and 142. Gaithersburg, MD, September 2001.
Bentz DP. Three-dimensional computer simulation of cement hydration and microstructure development. J Am Ceram Soc 1997;80(1):
321.
Molina L. On predicting the inuence of curing conditions on the
degree of hydration. CBI Report 5:92, Swedish Cement and Concrete
Research Institute, Stockholm, 1992.
Snyder KA, Bentz DP. Suspended hydration and loss of freezable
water in cement pastes exposed to 90% relative humidity. Cem Concr
Res 2004;34:204556.
Powers TC, Copeland LE, Mann HM. Capillary continuity or
discontinuity in cement paste. PCA Bull 1959;10:212.
Villadsen J. Pore structure in hardened cement paste measured by
three dierent methods. 3rd Nordic SympBuilding Physics 1993;93:
74552.
CRC Handbook of Chemistry and Physics. 68th ed. Boca Raton
(FL): CRC Press; 1987.
Taylor HFW. Cement Chemistry. 2nd ed. London: Thomas Telford;
1997.
Bentz DP, Garboczi EJ. Percolation of phases in a three-dimensional
cement paste microstructural model. Cem Concr Res 1991;21(2):
32544.
Berube MA, Tremblay C, Fournier B, Thomas MD, Stokes DB.
Inuence of lithium-based products proposed for counteracting ASR
on the chemistry of pore solution and cement hydrates. Cem Concr
Res 2004;34:164560.