This article was downloaded by: 10.3.98.

104
On: 06 Dec 2019
Access details: subscription number
Publisher: CRC Press
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place, London SW1P 1WG, UK

Alkali-Aggregate Reaction in Concrete: A World Review

Ian Sims, Alan Poole

So-Called Alkali-Carbonate Reaction (ACR)

Publication details
https://www.routledgehandbooks.com/doi/10.1201/9781315708959-3
Paddy E. Grattan-Bellew, Tetsuya Katayama
Published online on: 13 Jul 2017

How to cite :- Paddy E. Grattan-Bellew, Tetsuya Katayama. 13 Jul 2017, So-Called Alkali-
Carbonate Reaction (ACR) from: Alkali-Aggregate Reaction in Concrete: A World Review CRC Press
Accessed on: 06 Dec 2019
https://www.routledgehandbooks.com/doi/10.1201/9781315708959-3

PLEASE SCROLL DOWN FOR DOCUMENT

Full terms and conditions of use: https://www.routledgehandbooks.com/legal-notices/terms

This Document PDF may be used for research, teaching and private study purposes. Any substantial or systematic reproductions,
re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents will be complete or
accurate or up to date. The publisher shall not be liable for an loss, actions, claims, proceedings, demand or costs or damages
whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.
 Chapter 3
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

So-Called Alkali-Carbonate Reaction

(ACR)
Paddy E. Grattan-Bellew & Tetsuya Katayama

3.1 INTRODUCTION

The term ‘alkali-carbonate reaction’ (ACR) would appear to apply to a reaction

between the alkaline pore solution in concrete and any aggregate made from carbonate
rocks such as limestones or dolostones (dolomite rocks). However, the term alkali-
carbonate reaction is really applied only to the reaction between certain argillaceous
dolomitic limestones and the alkaline pore solution in concrete that gives rise to rapid
expansion and cracking of concrete. In North America the ACR dolomitic limestones
are of Ordovician age, but in the Sichuan Province of China, some are of Triassic age.
RILEM considered the use of carbonate-aggregate reaction (CAR) instead of ACR, but
the latter was too established.
The argillaceous dolomitic limestones, that are quite rare, form in shallow intertidal
or lagoon environments adjacent to the edge of an ancient continent (Tang et al., 2000).
The Pittsburg quarry in Kingston, Ontario, Canada, from where the ACR aggregate
was first encountered, is in the Ordovician Gull River formation that lies unconform-
ably upon the Pre-Cambrian Canadian Shield and is overlain by limestones formed
deeper in the basin. The presence of gypsum in the rock of the Pittsburg quarry is
consistent with the formation of the rock in a shallow marine environment. Evidence
from thin sections indicates that the deposit was initially laid down as limestone and
only later became dolomitized. Fossils are generally rare or absent from the reactive
argillaceous dolomitic limestones. In the Pittsburg quarry, the expansivity of aggregates
in concrete varies considerably from layer to layer, but the pit-run aggregate from the
entire first lift of the quarry is deleteriously expansive.
The term ‘alkali-carbonate reaction’ (ACR) has been in use since the 1960s, but recent
research indicates that the expansion and cracking of concrete caused by this reaction is
actually due to the formation of alkali-silica gel by reaction of the alkaline pore solution
and cryptocrystalline quartz in the argillaceous dolomitic limestone (Katayama, 1992,
2004, 2006, 2010a, 2011; Katayama & Sommer, 2008; Grattan-Bellew et al., 2009;
Katayama & Grattan-Bellew, 2012; Grattan-Bellew & Chan, 2013; Katayama et al.,
2016). In 2004 and 2006, Katayama coined the term “so-called ACR” to take into
account that the expansion was due to an alkali-silica reaction and not, as originally
thought, some other type of reaction involving dolomite, as envisaged by Swenson (1957).
Retention of the term “so-called ACR” to distinguish it from typical alkali-silica
reactive limestones may be justified, because of its different behaviour in the mortar-bar
test and also because the proportions of supplementary cementitious materials (SCMs)
 64 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

0.6

0.5

0.4
Expansion %

0.3
K OPC…
K HAC…
0.2

0.1

0
0 200 400 600
Days @ 38 Deg, C, & 100% R.H.

Figure 3.1 Expansion of concrete prisms made with ACR aggregate and a high alkali Portland cement
(OPC) and high-alumina cement (HAC).

that are effective in reducing expansion in alkali-silica reactive limestones are not
effective in concrete made with the “so-called ACR” dolomitic limestones. This may,
in part at least, be due to the formation of additional alkaline hydroxide in the
dedolomitization reaction, equations (i, ii).

Reaction rim within aggregate: CaMg(CO3)2 + 2MOH → CaCO3 +Mg(OH)2 + M2CO3

(i)

Carbonate halo within surrounding cement paste: M2CO3 +Ca(OH)2 → 2MOH +

CaCO3 (ii)

Expansion of typical ASR aggregates in concrete can be prevented by the use of a low
alkali-cement, (ASTM 0.60% Na2O equivalent), but in the case of ACR aggregates
such as those from the Pittsburg quarry, the alkali content of the cement has to be less
than 0.31% to reduce the expansion of concrete prisms below 0.04%, Swenson (1957).
Grattan-Bellew, unpublished results, has shown that the expansion of concrete prisms
made with ACR aggregate (Pittsburg) can be significantly reduced by use of high-
alumina cement (HAC), Figure 3.1. In these concrete prisms, dedolomitization occurs
on the coarse aggregate mixed with HAC (Figure 3.2a), but ASR gel forms when used
with high alkali Portland cement (Figure 3.2b).

3.2 HISTORICAL PERSPECTIVES

ACR was first described by Swenson (1957) in a paper entitled ‘A Reactive Aggregate
Undetected by ASTM Tests’. On the basis of a different response in the mortar-bar test
 So-Called Alkali-Carbonate Reaction (ACR) 65
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

(a) (b)

10 µm 50 µm

Figure 3.2 SEM micrographs of the concrete prisms depicted in Figure 3.1: a) [left] dedolomitization of
the aggregate mixed with high-alumina cement (Katayama & Grattan-Bellew, 2012), but b)
[right] ASR gel forming when mixed with a high-alkali Portland cement.

method ASTM C227, Swenson apparently assumed that the reaction was different
from that of the classical alkali-silica reaction as described by Stanton (1940). The term
ACR was not used by Swenson in his original paper. The ACR aggregate evaluated by
Swenson came from the Pittsburg quarry in Kingston, Ontario, Canada which is of
Ordovician age. The term ‘alkali-carbonate rock reaction’ was first used by Swenson
and Gillott (1964).
There was considerable research on the problem of the reactivity of carbonate
rocks in the U.S.A. in the 1960s. In 1962, Newlon and Sherwood described the
occurrence of Alkali-Reactive Carbonate Rock in Virginia, USA. In 1964, Hadley
reported on Alkali-Reactive Carbonate Rocks in Indiana and Lemish and Moore,
(1964) reported on alkali-carbonate reactions in Iowa, USA. In 1969, Buck reported
on the potential alkali reactivity of carbonate rock in Virginia and Kentucky.
Subsequently, the ACR problem in the U.S.A., was, for the most part, forgotten
apart from a few papers by Ozol (1974, 1994), until a large number of cases of
concrete deterioration were investigated by the present authors in Kentucky during
the period 2006 to 2008 (Katayama & Grattan-Bellew, 2012). The petrology and
expansivity of the Kentucky argillaceous dolomitic limestone is almost identical to
that of the Kingston aggregate (Figure 3.3).
Katayama (1992), based on literature reviews, predicted that ACR is a combined
phenomenon of harmless dedolomitization and expansive ASR of cryptocrystalline
quartz hidden in the dolomitic aggregate. Since then, this hypothesis has been examined
in detail by various petrographic methods (polarizing microscopy, SEM observation
and quantitative EDS analysis), using i) field ACR concretes from the type locality
(Kingston) of ACR (Katayama, 2004, 2006, 2010a, 2011; Gratten-Bellew et al., 2009),
ii) field reproduced ACR concretes containing the Kingston ACR aggregate
(Katayama & Gratten-Bellew, 2012; Grattan-Bellew & Chan, 2013) and iii) labora-
tory produced ACR concrete specimens (Katayama, 2004). After all these efforts,
however, a paradox was encountered, that the concrete deterioration meeting the
original definition of ACR by Swenson & Gillott, that is, deleterious expansion other
than typical ASR but associated with dedolomitization, had not been identified, even
when typical ACR concretes from the type localities Kingston and Cornwall in Ontario
 66 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

0.18

0.16

0.14

Expansion % 0.12

0.10

0.08 Kingston
0.06 Kentucky

0.04

0.02

0.00
0 50 100
Days @ 38 Degrees C.

Figure 3.3 Expansion of aggregates from Kingston and Kentucky in the CSA concrete prism test A23.2-14A.

were examined. Consequently, this led to re-defining the classical term ACR as
“so-called ACR” in 2004 & 2006 by Katayama (2004, 2006) (Katayama et al., 2016).
ACR was reported by Milanesi et al., in Argentina in 1996. Qian et al. (2001)
reported the occurrence of ACR aggregate from Jixian, Tianjin, China in 2001.
However, before 2004, the mechanism of dedolomitization and expansion were not
yet clearly distinguished based on detailed petrographic examination, hence early
reports on ACR should be reviewed critically.

3.3 MECHANISM OF SO-CALLED ACR REACTION

3.3.1 General
The deleterious alkali-carbonate reaction occurs in argillaceous dolomitic limestones
with porphyrotopic or mosaic textures (Swenson and Gillott, 1960; Buck, 1969;
Melanesi et al., 1996). The rocks consist essentially of a matrix of fine grained micritic
calcite crystals (2μm), intermixed with clay minerals, mostly illite with minor chlorite
and cryptocrystalline quartz (chert), in which are embedded large dolomite rhombohe-
dra (50 μm). According to Hadley (1964) the deleteriously expansive rocks generally
contain over 7% clay minerals. Figure 3.4a shows a photomicrograph of a typical thin
section of the Kingston dolomitic limestone taken through a petrographic microscope.
A photograph taken of a polished section of the Kingston dolomitic limestone taken
using SEM is shown in Figure 3.4b. Dolostone (dolomite rock) and dolomitic limestone
with a mosaic texture produce marked dedolomitization but are not always expansive.
Hence, it is important to distinguish two distinctive phenomena, i.e., reaction (dedo-
lomitization) and expansion, Katayama (2004, 2006, 2010a, 2011).
Hadley (1961) showed that the dolomite content of the deleteriously expansive
dolomitic limestones was about 50%. Lemish and Moore (1964) found that expansion
 So-Called Alkali-Carbonate Reaction (ACR) 67
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

(a) (b)

20 µm 10 µm
IRC 20.0kV 8.9mm x2.51k SE(M) 2/12/2013 20.0 µm

Figure 3.4 Kingston dolomitic limestone: a) [left]: Thin section photomicrograph showing a large
dolomite rhomb in a matrix of fine grained calcite. b) [right]: SEM micrograph of dolomite
rhombs in a matrix of fine grained calcite.

1.4

1.2
Expansion % @ 150 Days

0.8

0.6

0.4

0.2

0
30 50 70 90
Dolomote % in Rock

Figure 3.5 Dolomite content of aggregate vs expansion of concrete, adapted from Lemish and Moore
(1964).

was greatest at dolomite contents of 60% (Figure 3.5). However, Dolar-Mantuani

(1964), using the rock prism test, found most expansive rocks in Ontario contain less
than 40% dolomite. Katayama, (1992) reviewed the North American carbonate rocks
and indicated that insoluble residue at least 5-10% is the sole common characteristic
among expansive carbonate rocks, irrespective of whether it is a limestone or a
dolostone. Katayama & Sommer (2008) found that the argillaceous dolomitic lime-
stone in the most expansive layer of the Pittsburg quarry in Kingston (Pt-16) contained
16.8% of insoluble residue and 4.8% of mostly cryptocrystalline quartz by H3PO4
extraction (Figure 3.10a).

3.3.2 Dedolomitization
It was initially noticed that the ACR involved dedolomitization of the aggregates in an
alkaline medium. Hadley (1961) reported that a large single crystal of dolomite
 68 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

expanded by 0.15% after 100 days in 3M NaOH solution. Hadley concluded that “the
expansion of the de-dolomitizing crystals themselves would be sufficient to account for
the expansion of the reactive carbonate rocks”. However, Gillott and Swenson (1969)
failed to reproduce the expansion of a dolomite single crystal and noted that the molar
volumes of the products of dedolomitization were less than that of the reactants and
hence they concluded that dedolomitization could not account for the expansion of
dolomitic limestone aggregates in concrete. Deng and Tang (1993) calculated that the
solid volume change on dedolomitization was - 4.32 % (less than zero), however they
noted that the products of dedolomitization are fine and enclose many voids. They
calculated that the void space was 29.21 % and from this they calculated that the
volume change for the dedolomitization reaction was 36.56 %. Walker (1978) con-
cluded that “Dedolomitization can be an innocuous reaction but in specific kinds of
carbonate rocks it can result in expansion of the aggregate particles and destructive
expansion of the concrete”.
In his critical review of the published references on carbonate rock reactions before
1992, Katayama concluded that dedolomitization is an innocuous reaction, and that
deleteriously expansive ACR had been confused with dedolomitization. In 2000,
Grattan-Bellew and Rogers (2000) found that, after 13 years outdoor exposure, most
of the dolomite in the Kingston ACR aggregate was unreacted and so it was most
unlikely that dedolomitization could be the underlying mechanism of expansion.

3.3.2.1 Reaction rims

How, then, does dedolomitization proceed in concrete? Decomposition of dolomite
crystals, equation (i), forms a reaction rim within the dolomitic aggregate. Figure 3.6a
shows a conspicuous dark rim of dedolomitization developed within a reacted dolo-
mitic limestone aggregate in the 6-month old field concrete in Saudi Arabia. At the
interface between the dolomite crystals and cement paste, a thin rim of Mg-Al-bearing
hydrate resembling hydrotalcite may be formed on the dolomite absorbing Al ions from
the cement paste, instead of brucite (Katayama et al., 2008; Katayama, 2010a, 2012a).

(a) (b)

100 µm 10 µm

Figure 3.6 Dedolomitization rims around dolomitic limestone aggregate, a) [left] showing a brucite-rich
dark zone with an invisibly narrow hydrotalcite layer at the very interface with the cement
matrix, Saudi Arabia (Katayama, 2012a), b) [right] showing minute spots of brucite (dark) and
calcite (bright) formed by dedolomitization of Austrian dolostone in a RILEM AAR-5 con-
crete microbar.
 So-Called Alkali-Carbonate Reaction (ACR) 69
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

By contrast, within the reaction rim of the dolomitic aggregate, or along the cracks
extending from the periphery of the reacted aggregate into the interior of the aggregate,
minute spots (<3 μm) of brucite and calcite are formed as a result of dedolomitization
(Figure 3.6b) (Katayama, 2004, 2010a, 2011; Katayama & Sommer, 2008).

3.3.2.2 Carbonate haloes

In the dedolomitization, secondary calcite forms a carbonate halo, equation (ii), within
the cement paste that surrounds the reacted dolomitic aggregate. It should be noted that
no radiating crack suggestive of expansion is formed in the cement paste. In sufficiently
thin polished thin sections (15μm), under the crossed polars, the carbonate halo has a
high birefringence showing a bright yellow colour, characteristic of calcite
(Figure 3.7a). This carbonate halo is a porous crust composed mainly of calcite that
had precipitated from the interstitial solution and replaces CSH gel in the cement paste
(Figure 3.7b). Elemental mapping of the carbonate halo by EDS revealed a zonal
distribution of alkalis corresponding to the formula (ii) (Katayama, 2004, 2010a).

3.3.3 Evidence of expansion

Previous research on ACR was likely to be confused by believing that dedolomitization
is evidence of expansive ACR. However, according to Katayama (2004, 2006, 2010a,
2011), wherever expansion cracks exist in concrete, ASR gel is always present.
Katayama found that evidence of expansion in concrete is demonstrated by the forma-
tion of radiating cracks from the reacted particles of the aggregate, irrespective of
whether it is typical ACR of argillaceous dolomitic limestone or typical ASR of andesite
or another reactive siliceous rock. It has been shown, aided by SEM observations on
polished thin sections, that there is no difference in the process of crack formation in
concretes between ASR and ACR. Katayama and Katayama and Grattan-Bellew
(2012), revealed that the sequence of the crack formation in ACR concretes is: 1) rim
formation, 2) internal cracking with ASR gel within the aggregate, 3) cracking from

(a) (b)

200 µm 100 µm

Figure 3.7 Carbonate halo in the cement paste surrounding the dedolomitized aggregate particle: a)
[left] showing the distinctive colour indicative of calcite, RILEM AAR-5 concrete microbar
with Austrian dolostone, b) [right] showing that the carbonate halo replaces CSH gel in the
cement paste in a concrete building in Saudi Arabia.
 70 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

(a) (b)

100 µm 20 µm

Figure 3.8 Photomicrgraphs from expanded AAR-5 concrete microbars containing dolomitic aggre-
gate: a) [left] ASR gel lining a crack emanating from the expanding particle and into the
surrounding cement matrix, b) [right] veins of ASR gel within the reacted dolomitic lime-
stone from a concrete pavement (sidewalk) in Kingston, Ontario, wherein the dolomite
rhombs are only slightly decomposed (from Katayama & Grattan-Bellew, 2012).

aggregate into the cement paste exuding ASR gel, and 4) ASR gel lining air voids and
cracks distant from the reacting aggregate.
With RILEM AAR-5 (in Nixon & Sims, 2016) concrete microbars, expansion
continues even when brucite disappears, transforming into non-expansive magnesium-
-silicate gel, and exudation of ASR gel becomes prominent along radial expansion
cracks (Katayama, 2006, 2010a; Katayama & Grattan-Bellew, 2011) (see Figure 3.8a).

3.3.4 Formation of alkali-silica gel in the concrete

In 1986, Katayama (1992) collected a piece of concrete from Cornwall, Ontario,
Canada and examined ultra-thin sections (15 μm thick) made from it. He observed
alkali-silica gel filling cracks in the concrete suggesting that ASR had at least a role in
the deterioration of the concrete made with an ACR aggregate. During 2004 to 2006,
Katayama, after examining a number of samples of concrete affected by the ‘so-called
alkali-carbonate reaction’, including the typical ACR concrete from near Kingston
constructed in the late 1950s, concluded that the “ACR could be a mixture of deleter-
iously expansive ASR of cryptocrystalline quartz, and non-expansive dedolomitization
of dolomitic aggregates”, and that “ACR is believed to be ASR, influenced more (fine
aggregate) or less (coarse aggregate) by dedolomitization” (Katayama, 2004).
Katayama identified ASR of crypto- to microcrystalline quartz in the dolomitic
aggregates in several structures from Montreal to Quebec, Canada. He confirmed
that ACR is a combined effect of deleteriously expansive ASR of cryptocrystalline
quartz, and harmless dedolomitization, which produces brucite and carbonate haloes
without accompanying expansion cracks (Katayama, 2004, 2006, 2010a; Katayama &
Sommer, 2008). What is important is that dolomitic aggregates do not develop expan-
sion cracks in the embedding cement paste, unless ASR is involved. Figure.3.8a shows
the typical expansion crack filled or lined with ASR gel in the ACR aggregate in an
AAR-5 concrete microbar (Katayama & Grattan-Bellew, 2012). In the Kingston
experimental pedestrian pavement (sidewalk) at age 22 years, radial expansion crack
is filled with veins of ASR gel within the reacted dolomitic limestone aggregate
 So-Called Alkali-Carbonate Reaction (ACR) 71
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

3.5 3.5 CSH gel (alite)

CSH gel (alite)
3.0 CSH gel (belite) 3.0 CSH gel (belite)
ASR gel (void) ASR gel (void)
Ca/Si atomic rao

Ca/Si atomic rao

2.5 ASR gel (paste) 2.5 ASR gel (paste)
ASR gel (aggregate) ASR gel (aggregate)
2.0 2.0
1.5 1.5 Convergent
1.0 1.0 point

0.5 0.5
0.0 0.0
0.1 1 10 100 1000 10000 0.1 1 10 100 1000 10000
Ca/(Na+K) atomic rao Ca/(Na+K) atomic rao

Figure 3.9 Comparison of ASR gel compositions: a) [left] from a typical ACR bridge near Kingston,
Ontario (Katayama, 2010a), and b) [right] from a typical example of ASR caused by andesite
in Japan (Katayama, 2010b).

(Figure.3.8b). Note the dolomite rhombs are only slightly decomposed into a thin rim,
consisting of brucite or magnesium silicate gel, whereas ASR gel is formed abundantly.
Lu et al. (2006) also observed gel in samples of the ACR aggregate from the Pittsburg
quarry autoclaved at 150ºC in KOH solution.

3.3.4.1 Compositional trends of ASR gel in ACR concrete

According to Katayama (2010b, 2008) and Katayama et al. (2008), ASR gel in typical
ASR concretes migrates along cracks from the reacted aggregate into the cement paste,
absorbs calcium from cement paste, and approaches the composition of CSH gel. In
principle, on the [Ca/Si]-[Ca]/[Na+K] diagram, alkali-silica reaction terminates at a
convergent point of CSH gel at around Ca/Si=1.5, Ca/(Na+K)=100-200. It should be
noted that there is no difference in the compositional trends of ASR gel between ACR
concretes and ASR concretes. Figure 3.9a shows the data of typical ACR bridge near
Kingston, while Figure 3.9b is typical ASR of andesite in Japan. In both cases, ASR gel
crystallizes into the rosette-like aggregation at Ca/(Na+K) <1. Where leaching of alkalis
occurs along cracks in concrete, a short parallel compositional line appears, irrespec-
tive of whether it is ACR as shown here (Katayama, 2010a), or ASR (Katayama,
2010b, 2008).

3.3.5 Identification of cryptocrystalline quartz

In the ACR reactive argillaceous dolomitic limestone from Kingston, cryptocrystalline
quartz is contained in an isolated form, and not as an aggregation like a chert. This
explains why this form of reactive silica is difficult to identify in thin section, and had
long been missed in the optical microscopy of the ACR aggregate from Kingston. This
cryptocrystalline quartz was extracted from the Kingston rock by the phosphoric acid
treatment, Katayama (2004, 2010a) (Figure.3.10a).
In field concretes undergoing ACR, cryptocrystalline quartz in the Kingston rock is
difficult to identify, because this quartz reacts very fast (Katayama, 2004, 2010a, 2011;
 72 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

(a) (b)

10 µm 10 µm

Figure 3.10 Cryptocrystalline quartz from the Kingston dolomitic limestone aggregate: a) [left]
extracted by phosphoric acid treatment, b) [right] in some AAR-5 concrete microbars,
small amounts of cryptocrystalline quartz remain in isolated form, but most has converted
to ASR gel.

Katayama & Sommer, 2008). However, in RILEM AAR-5 concrete microbars, some
cryptocrystalline quartz occasionally remains, but mostly converts to ASR gel (Figure
3.10b) or Mg-silicate gel (Katayama, 2004, 2010a; Katayama & Sommer, 2008). In the
siliceous argillaceous dolomitic aggregates undergoing ASR in Quebec, crypto- to
microcrystalline quartz is abundant and is partly converting to ASR gel as revealed
by SEM observation on the polished thin section, Katayama (2006).

3.3.6 Formation of magnesium silicate gel

The magnesium silicate gel (Mg-silicate gel) occurs as an intermediate product between
ASR gel and magnesium hydroxide gel or brucite in the reacted dolomitic aggregate
that contains reactive silica (Katayama, 2004, 2006, 2010a, 2011, 2012a). It is found
within a dedolomitized rim or along cracks in the interior of the aggregate. Mg-silicate
gel forms a thin inner rim of the reacted dolomite crystals (Figure 3.11a), or a

(a) (b)

10 µm 10 µm

Figure 3.11 Magnesium silicate gel: a) [left] forming a thin inner rim around reacting dolomite particles
(adapted from Katayama, 2006), b) [right] forming a pseudomorph after the original
dolomite crystals and showing shrinkage cracks.
 So-Called Alkali-Carbonate Reaction (ACR) 73
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

pseudomorph after the dolomite crystals keeping their original rhombohedral outlines
(Figure 3.11b).
To identify the Mg-silicate gel, it is essential to perform a quantitative SEM-EDS
analysis on the polished thin section at higher magnifications (e.g., > 1500 x) to obtain a
sharp image to distinguish the minute textures composed of dolomite, Mg-silicate gel,
ASR gel, etc. In the aged concrete constructed in the 1950s near Kingston containing
the ACR reactive Pittsburg aggregate, Mg-silicate gel became more or less crystalline,
showing the transformation into sepiolite-like material or flakes of antigorite-like
material as estimated by their composition, (Katayama, 2006, 2010a, 2011). Where
dolomitic rock is argillaceous (Pittsburg aggregate), Al-Mg silicate gel forms
(Katayama, 2006), and on aging this gel converts to a chlorite-like material.
There has been no evidence that Mg-silicate gel generates expansion, because no
radiating cracks are found to accompany its formation, except for shrinkage cracks (see
Figure.3.11b). The occurrence of the Mg-silicate gel in field concretes has been docu-
mented from variously reactive dolomitic aggregates in Ontario (deleterious), Austria
(harmless) (Katayama, 2004, 2006, 2010a; Katayama & Sommer, 2008) and Saudi
Arabia (harmless) (Katayama, 2012a), as well as in the ASR-reactive aggregates in
Quebec and Japan (harmless) (Katayama et al., 2008) based on SEM-EDS analysis, but
has not been identified by other researchers in a number of works before around 2000.

3.4 COUNTER ARGUMENTS AGAINST GEL FORMATION AS THE

CAUSE OF EXPANSION DUE TO ACR

The arguments on this subject have been reviewed critically by Katayama (2010a,
2011) and Katayama et al., (2016). The main points of the problem are that 1) the
conventional methods for preparation and examination of concrete thin sections are
inappropriate, 2) general understanding on the expansion behaviours and remedial
measures for a wide range of ASR remains stereotypical and has been reproduced by the
papers without critical reviewing, and 3) interpretations on the expansion mechanisms
related to dedolomitization lack support by petrographic evidence.

3.4.1 Why is gel not observed in thin sections of ACR concrete?

3.4.1.1 Thickness of thin section

The standard thickness of rock thin sections is generally accepted by geologists to be
from around 30 μm in Europe and North America to 20 μm in Japan, and published
optical properties for minerals usually relate to this range of thickness (although 30 μm
is too thick for quartz, because it presents a pale yellow interference colour). However,
whilst traditional 30 μm sections are usable for identification of the main mineralogy
and types of coarse-grained rocks, they are often too thick for concrete petrography,
particularly for the detailed study of AAR. For the petrographic examination of
Portland cement clinker, thin sections with thickness of <15 μm have been accepted
as standard (Insley & Fréchette, 1955). Similarly, the ideal thin section thickness for
fine-grained alkali-reactive rock types (e.g., volcanic rocks, cherts, and carbonate
rocks) is 15 μm, because most of the reactive constituents contained in them
 74 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

(a) (b)

300 µm 300 µm

Figure 3.12 Comparison of clarity between traditional 30 and 15 µm thick thin sections: a) [left] near-
opaque appearance of cement paste matrix with a traditional 30 µm thick section, and b)
[right] clearer resolution of both aggregate and matrix using a 15 µm thick section (adapted
from Katayama, 2012b).

(cristobalite, tridymite, volcanic interstitial glass, cryptocrystalline and microcrystal-

line quartz) are small (mainly <10μm) (Katayama, 2010a, 2012b; Katayama et al.,
2016).
The calcite crystals that comprise the matrix of the ACR aggregate are about 2 μm in
size (see Figure 3.14b, Grattan-Bellew et al., 2009) and hence, in 30 μm thick thin
sections, a petrographer is looking through about 15 layers of calcite that effectively
obstructs the view, making it almost impossible to observe fine gel filled cracks. The 30
μm thin section of concrete gives an almost opaque appearance of cement paste
(Figure 3.12a), whereas 15 μm section permits detailed observation by giving a trans-
parent appearance of cement paste and argillaceous carbonate matrix of the dolomitic
aggregate (Figure 3.12b).
Many papers on ACR, both published and submitted manuscripts to journals, are
based on observations of either very thick thin sections, justifying the thickness of 30
μm in the references on petrography of natural rocks, or on the SEM observations on
the fracture surface of concrete. In addition, enthusiasm for producing large-area
thin sections (e.g., 12 cm by 18 cm, 30 μm thick) prevented preparation of suffi-
ciently thin sections suitable for detailed petrographic examination. This is one of
the main reasons why the presence of crypto- to microcrystalline quartz, which is
responsible for the deleterious expansion due to ASR in the ACR (‘so-called ACR’ by
the present definition), has long been missed in the petrography of concrete
(Katayama, 2010a).

3.4.1.2 Carbonation of ASR gel

In the ACR concretes undergoing dedolomitization, both cement paste and ASR gel are
most likely subject to secondary carbonation by the carbonate ions that are liberated
from the reacted dolomite through dedolomitization (Katayama, 2010a). Because the
carbonate minerals calcite and dolomite constituting the dolomitic aggregates have a
high-birefringence, they overlap with fine cracks and ASR gel that lines or fills these
cracks, as well as crypto- to microcrystalline quartz in the aggregate, thereby hindering
 So-Called Alkali-Carbonate Reaction (ACR) 75
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

(a) (b)

20 µm 100 µm

Figure 3.13 Clarity enabled by polished thin (15 µm thick) thin sections: a) [left] vein of combined ASR
gel and calcite formed through carbonation during storage of a concrete prism (450 days),
b) [right] ASR gel mixed with secondary calcite (from Katayama & Grattan-Bellew, 2012).

the details of the microscopic texture in the transmitted light. In the concretes affected
by more or less atmospheric weathering, such as field structures and concrete prisms
stored for a long time, secondary calcite forms along the cracks often replacing veins of
ettringite and ASR gel (Katayama, (2010a). As a result, ASR gel is likely to be missed by
the thin section microscopy.
Although the above alteration processes obscure the presence of ASR gel in the
transmitted polarizing microscopy, SEM observation on the polished thin section is
able to reveal clearly the presence of the composite vein in which ASR gel is
enveloped by its carbonation product calcite vein (Figure 3.13a, Katayama &
Grattan-Bellew, 2012), as well as thin films of ASR gel intermixed with secondary
calcite (Figure 3.13b). In the thick thin section (~30 μm) with a cover glass, none of
fine cracks, ASR gel or cryptocrystalline quartz is detectable. To identify these fine
textures and reaction products, combined petrographic examination is recom-
mended, which consists of polarizing microscopy in both transmitted and reflected
lights, SEM observation to characterize the reaction sites, and quantitative EDS
analysis, using the same polished thin section (15 μm thick) (Katayama, 2004, 2006,
2010a, 2011).

3.4.1.3 Fluorescence method to identify thin films of ASR gel

Subsequent to Katayama, Grattan-Bellew et al. (2010) concluded on the basis of
laboratory experiments that the formation of alkali-silica gel was the cause of the
expansion and deterioration of concrete made with ACR aggregates. Data from
Gillott (1961) and Swenson and Gillott, (1964) show that there is a direct correlation
between the amount of chert in the insoluble residue of the Kingston aggregate and the
expansion of concrete prisms. The fluorescence of alkali-silica gel on the fractured
surface of a core, treated with uranyl acetate, from the 23 year old experimental
sidewalk in Kingston made with the argillaceous dolomitic limestone and high alkali
cement is shown in Figure 3.14a.
Further confirmation for the occurrence of ASR in concrete made with so-called
ACR aggregates was presented by Katayama and Grattan-Bellew (2012) and by
 76 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

(a) (b)

Figure 3.14 Detection of ASR gel in concretes made with ACR aggregate: a) [left] fluorescence of gel
after treatment with uranyl acetate, b) [right] thin films of ASR gel between micro-crystals
of calcite.

Grattan-Bellew and Chan (2013), showing the presence of the thin films of alkali-
silica gel between the small crystals of calcite in the matrix of the rock,
Figure 3.14b.

3.4.2 Why does coarse aggregate expand more?

Why do ACR aggregates not cause expansion in mortar-bars if the mechanism of
expansion is ASR? This subject has been discussed in detail by Katayama (2010a,
2011). According to Katayama and Sommer (2008) and Katayama (2010a), in
mortar-bars the -4 mm size of the aggregate results in more or less complete
dedolomitization of the finer particles that produce brucite Mg(OH)2, which
reacts with the silica gel, produced by reaction of the crypto-crystalline quartz
with the alkaline pore solution, to produce a magnesium silicate gel that is either
less expansive or non-expansive and so expansion of mortar-bars is much
reduced.
In general, typical ASR proceeds by the expansion generated in the interior of the
aggregate. With the coarse aggregate, expansive alkali-rich ASR gel tends to remain
within the interior of the aggregate particle, continues expansion, and generates wide
cracks filled with ASR gel. By contrast, fine aggregate particles tend to lose the
expansive nature rapidly through the replacement of alkali ions in the ASR gel by
calcium ions from the surrounding cement paste, thus producing small cracks. In the
alkali-immersion tests, the size of the aggregate is important. In the accelerated
mortar-bar test (ASTM C1260: 2004), chert particles do not expand significantly
due to dissolution of ASR sol into NaOH solution, whereas in the accelerated
concrete microbar test [AAR-5 (in Nixon & Sims, 2016), Sommer et al., 2004] they
produce expansion due to formation of ASR gel within the coarse aggregate.
Nowadays, ASR gel is considered to be a universal product for generating expansion
of all the traditionally known varieties of alkali-aggregate reaction, i.e., alkali-silica
 So-Called Alkali-Carbonate Reaction (ACR) 77
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

(a) (b)

20 µm 10 µm

Figure 3.15 a&b Cracking filled or lined with ASR gel in a RILEM AAR-5 concrete microbar made using
50% Portland cement replacement by Japanese ground granulated blastfurnace slag, in
which expansion was reduced by 30 % but cracking not prevented (from Katayama &
Grattan-Bellew, 2012).

reaction, alkali-silicate reaction, and alkali-carbonate reaction (Katayama, 2012b,

2012c).

3.4.3 Why are supplementary cementitious materials not

effective?

3.4.3.1 Blastfurnace slag

Supplementary cementitious materials that are usually effective in minimizing expan-
sion due to ASR are not effective when used in concrete made with ACR aggregates,
which may be due to a number of causes. The crystallinity of the cryptocrystalline
quartz in the ACR aggregate is lower than in typical ASR limestones, resulting in it
being much more reactive and hence more difficult to suppress (Grattan-Bellew &
Rogers, 2000). Katayama (2008) reported that replacement of 50% of Portland cement
by fine grained Japanese blastfurnace slag reduced expansion by 30%, but that it failed
to prevent the formation of cracks filled with ASR gel at later ages (Figure 3.15a,b).
Because cryptocrystalline quartz in the Pittsburg aggregate occurs as isolated forms,
rather than a massive aggregation like a chert, this quartz is considered to be more
reactive than the particles of ground granulated slag and fly ash (Katayama, 2012b).
Generally, blastfurnace slag is not always effective in preventing ASR: replacement of
40% of cement by slag could not suppress cracking due to late-expansive ASR in a mass
concrete structure in Japan.

3.4.3.2 Fly ash

Fly ash is not effective in suppressing ACR in concrete. In Kentucky, high-calcium fly ash
(40% CaO in glass, content 40%) was not effective in preventing cracking by the ACR
aggregate and ASR of chert particles (Figure 3.16a, Katayama & Grattan-Bellew, 2012). In
the reacted dolomitic aggregate, expansion cracks originated from the pool of ASR gel
that had formed surrounding the calcite particles (Figure 3.16b). With typical ASR, even
 78 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

(a) (b)

10 µm 5 µm

Figure 3.16 Cracking of concrete in Kentucky caused by ACR, a) [left] not prevented by the use of 40%
cement replacement by high-calcium fly ash, b) [right] cracks originating from pools of ASR
gel surround the calcite particles (from Katayama & Grattan-Bellew, 2012).

low-calcium fly ash is not effective when the aggregate is highly reactive. In a Japanese case,
low-calcium fly ash (<4% CaO in glass, replacement of 18%) was effective for the late-
expansive ASR of the sandstone pebbles in concrete, but was not effective for the early
expansive ASR due to the glassy bronzite andesite aggregate. The average SiO2 content of
the fly ash glass was 58%, while that of the rhyolitic interstitial glass in the andesite was
75%, which explains the low pozzolanic reactivity of this fly ash (Katayama, 2010b).

3.4.3.3 Lithium salt

The addition of lithium salts to concrete made with ACR aggregates is not effective in
suppressing expansion (Tang & Deng, 2004). However, this is also the case with late-
expansive ASR of orthoquartzite and greenschist in Canada (Durand, 2000). All these
may be due to the presence of a dense matrix surrounding the reactive particles that
either prevents the lithium ions from reaching them or the lithium is absorbed by clay
minerals (Katayama, 2010a).

3.4.4 Dedolomitization-related mechanisms

3.4.4.1 Hypothetical volume-increase dedolomitization

López-Buendia et al. (2006, 2008) claimed that dedolomitization itself is a volume-
increase reaction, because calcite has a larger d-spacing (d104 = 3.03Å) than that of
dolomite (d104 = 2.89Å). To compare the unit cell volumes of dolomite and calcite,
mass balance should be considered, then one molar dolomite is equivalent to two molar
calcites, expressed as equation (iii) (Katayama, 2011, 2010b). This is a hypothetical
dedolomitization, in which complete ion-exchange takes place between Mg in the
dolomite and Ca in the cement paste. Mg ions liberated from dolomite into solution
do not precipitate as brucite, and more than 10% of volume increase may result
(Katayama, 2010b). However, in concrete, such an entire replacement of dolomite by
calcite in equation (iii) does not occur, as is evident from the microtexture of Figure
 So-Called Alkali-Carbonate Reaction (ACR) 79
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

(a) (b)

200 µm 50 µm

Figure 3.17a) & b) Natural dedolomitization, in which calcite pseudomorphs replace dolomite rhom-
bohedra without producing brucite.

3.6b which follows the equation (i). Actually, concrete specimens of López-Buendía
et al. (2008) made with a Spanish dolomitic limestone aggregate presented contraction.
For comparison, natural dedolomitization occurs in geological time by the near-
surface weathering of dolomitic formations (Katayama, 2010b). In this process,
calcite replaces the dolomite crystals, leaving a pseudomorphic texture of the origi-
nal outlines of the dolomite rhombs (Figure 3.17a,b). This dedolomitization is a
dissolution/precipitation process in the open system, and there is no volume change
after the precipitation of calcite. Dolomites are dissolved into interstitial solution
and secondary calcites are precipitated into the dissolution voids formed after the
dolomite crystals. Mg ions liberated from the dolomite are carried away and do not
precipitate as brucite, which is different from the dedolomitization in concrete. This
reaction without volume change could be written as equation (iv) (Katayama, 2011,
2010b).

Hypothetical dedolomitization: CaMg(CO3)2 + Ca2+ → 2CaCO3 + Mg2+ (iii)

Natural dedolomitization: CaMg(CO3)2 + 0.75Ca2+ → 1.75CaCO3 + Mg2++ 0.25CO32–

(iv)

3.4.4.2 Uptake of moisture and swelling of dry clay minerals in the

dolostone
Gillott and Swenson (1969) concluded that dry clay minerals in dolomite crystals in the
aggregate would become exposed to alkalis and moisture due to dedolomitization and
the swelling of previously un-wetted, non-swelling clays (illite) could account for the
expansion of the aggregates. However, if a swelling clay, typically smectite, is con-
tained in the aggregate, this would produce distinctive shrinkage cracks of cement paste
concentrically surrounding the aggregate through the repeated drying and wetting
seasonally encountered in the outdoor conditions. Katayama (2011) reported that no
such shrinkage cracks have been observed in the Kingston dolomitic aggregate in the
typical ACR concretes from Ontario.
 80 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

3.4.4.3 Expansion due to dedolomitization and formation of brucite in

confined space
Tong and Tang (1999) concluded that “brucite formed in a confined space and
generated the force to cause the expansion of rock prisms and aggregate particles in
concrete microbars”. This hypothesis appears rather unlikely as only a small amount of
the dolomite limestone de-dolomitizes in the field deteriorated concretes.
The accelerated condition (150ºC) of concrete microbars and compacted mortars
produces portlandite and brucite in the particles of pure limestone and magnesite
(Tong, 1994; Tang & Deng, 2000). They believed that the expansion was caused by
a “topochemical reaction” that took place within a confined space to generate a
pressure of crystallization of portlandite and brucite, claiming that the expansion of
the ACR is due to the formation of brucite through the dedolomitization. However, if
this mechanism is accepted, then pure limestone should also be regarded as deleter-
iously expansive, which contradicts the reality, since the expansion of this limestone
aggregate in the autoclave is attributed to the thermal expansion of coarsely crystalline
calcite (Mu et al., 1996). Hence, it is unlikely that brucite causes expansion, Katayama
(2011).

3.5 TEST METHODS

3.5.1 Rock Cylinder Test (ASTM C 586)

The first reported use of the rock prism test is by Hadley (1961). It was also used by
Swenson and Gillott (2009), Dolar-Mantuani (1964), Walker (1978) and intensively
by Rogers (1985). ASTM C586 (2006) specifies that the cylinders or prisms be stored in
1N NaOH at a temperature range of 20 to 27.5ºC. The overall size of the specimens
specified to be 9 mm diameter by 35±5 mm in length. However, Deng et al. (1993)
stored the prisms at 80ºC in 1N KOH. The relatively short length of the cylinders or
prisms permits evaluation of individual layers in a quarry either by taking rock hand
samples, or by evaluating cores, something that is difficult to achieve using other test
methods. Hadley found good correlation between expansion of rock prisms after 14
days in 1M NaOH and the expansion of concrete prisms stored at 38 ºC.
A modified miniature rock prism test was developed by Grattan-Bellew (1981) and
this method was also used by Tang and Lu (1986), who tested prisms made from the
Kingston Rock. They showed that expansion increased with temperatures from 20 to
80ºC. It should be noted that laminated argillaceous carbonate rocks have a strong
anisotropy in expansion by the direction of bedding. Rock cylinders cut perpendicular
to the bedding plane produce expansion larger than those parallel to the bedding: e.g.,
20% to 40% larger at age 28 days to 19 months with the Kingston ACR rock, whereas
7 times larger was found with a Japanese dolomitic limestone (Katayama et al., 1996).
This test does not necessarily reproduce ACR as recognized in concrete, because
cement is not involved and interaction between the dolomitic rocks and calcium in the
cement paste, e.g., formation of the carbonate halo, is thus not generated. Swelling of
the clay minerals by water and alkali solution, may occur, but this is not relevant to
ACR. In addition, NaOH solution used as the immersion medium tends to dissolve
ASR sol before solidifying into ASR gel. On the basis of these differences from the tests
 So-Called Alkali-Carbonate Reaction (ACR) 81
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

using concrete specimens, rock cylinder/prism test is not necessarily suitable for the
study of the mechanism of ACR.

3.5.2 Concrete Prism Testing

The concrete prism test [ASTM C1105 (2006), CSA A23.2-14A (2014), French
Standard NFP 18-454 (2004), RILEM AAR-3 (in Nixon & Sims, 2016)] is the most
widely used method for the evaluation of the potential expansivity of argillaceous
dolomitic limestones and probably comes closest to replicating the performance of
concrete made with ACR aggregates in the field. However, the use of a specified
mixture design of the concrete places some limitations on the use of this test method,
for example, to predict the field performance of mixtures with lower cement contents
and those using lower alkali cements. The CSA standard specifies the use of a non-
reactive fine aggregate in the concrete prism test, however, Fournier et al. (2000),
showed the importance of evaluating the potential expansivity of combinations of
coarse and fine aggregates as the reactivity of the fine aggregate would likely have an
effect of the expansion of the prisms. The Canadian Standard CSA A23.2-14A (2014)
was developed from the work of Swenson (1957). The specified size of the prisms in the
current standard is 75 x 75 x 275 mm or a maximum length of 405 mm (coarse
aggregate: max.19 mm). The prisms are stored at 38 ºC as for testing for ASR. The
specified cement content for the concrete prisms is 420 kg/m3 and the alkali content of
the cement is increased to 1.25% Na2O equivalent by addition of NaOH.
In ASTM C1105 (2006), the specified cross section of the prisms is the same as in
CSA, but the storage temperature is 23ºC. In the French standard NF P 18-454 (2004)
the prisms are 70 x 70 x 282 mm. The prisms are stored in a “reactor” at 60 ºC to
accelerate the reaction. In RILEM AAR-3 (in Nixon & Sims, 2016), the prisms are 75 x
75 x 250mm, and the storage temperature is 38ºC. An accelerated version, RILEM
AAR-4, (in Nixon & Sims, 2016) was developed from the French method, which uses
the prisms of 75 x 75 x 250mm stored at 60 ºC in a double container to avoid drying up
and leaching of alkalis from the prisms.
Expansion curves of concrete prism tests often present a plateau at later ages,
forming asymptote towards a certain expansion value, but this does not necessarily
mean that reactive constituents in the aggregate were entirely consumed. Leaching of
alkalis from the concrete prisms occurs during the long-term storage, perhaps in the
order of 30-40% in some cases, irrespective of the temperature applied, 23ºC or 60 ºC.
For this reason, late-expansive ASR aggregates present a sigmoidal expansion curve with
apparent asymptote, whereas early-expansive ACR aggregates tend to give a simple
convex curve with asymptote, rather than a sigmoid with an inflection point. At the
same time, secondary carbonation proceeds along the cracks that were formed on the
concrete prisms and calcite replaces ASR gel and ettringite in varying degrees (Katayama,
2010a).

3.5.2.1 Correlation Between Expansions in the Concrete Prism Test & in

the Field
There are only two papers on the comparison between expansions in the field and in
concrete prisms made with ACR aggregates (Grattan-Bellew, 2000; Lu et al., 2008).
 82 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

0.5 % Expansion Field

% Exp 23 Deg. C
0.4
% Exp. 38 Deg. C
Expansion %
0.3

0.2

0.1

0
0 20 40 60
Storage Time in Months

Figure 3.18 Comparison between actual field expansion of the experimental concrete sidewalk in
Kingston and that predicted by concrete prism tests at 23 °C and 38 °C.

Both used data from the experimental sidewalk in Kingston, Ontario, Canada. Rogers
and Hooton, (1992) noted that after one year the expansion of concrete prisms stored
at 23 ºC was about the same as concrete slabs exposed outdoors under ambient
conditions. However, after 5 years, leaching of alkalis from the concrete prisms stored
at 23 ºC had reduced expansion significantly below that of the slab exposed outdoors,
but in the case of one slab, the expansion of the prisms stored at 38 ºC was similar to
that of the slab exposed outdoors, Figure 3.18. It is concluded from these results that
the expansion of concrete prisms stored at 38 ºC and ~100% humidity provides a better
prediction of the long term expansion of concrete in the field. The main disadvantage of
the concrete prism test is that it takes up to one year to obtain results.

3.5.3 Concrete Microbar Test

In this test procedure [RILEM AAR-5 (in Nixon & Sims, 2016), Tang & Deng, 2004],
concrete microbars, 40 by 40 by 160 mm are made with a single size fraction of
aggregate, 4.0 to 8.0 mm. The water to cement ratio is 0.33 and the aggregate to
cement ratio is 1:1. The length change of the bars is monitored for 28 days in 1 M
NaOH @ 80ºC. In Tang and Deng (2004), expansions of bars made with ACR
aggregates in excess of 0.1% at 28 days are considered to be indicative of potentially
deleterious expansion in concrete. Good correlation was found between the expansions
of concrete microbars and concrete prisms (Lu et al., 2008). The similar procedure in
RILEM AAR-5 is used to compare behaviour with the conventional accelerated
mortar-bar test for ASR [AAR-2 (in Nixon & Sims, 2016)], on the basis that reactive
carbonates will give greater expansion with the coarser aggregate particle size (4 to 8
mm) used in AAR-5.
During the storage, concrete microbar specimens are not affected by atmospheric
carbonation. In this test, all the important features and reaction products characteristic
 So-Called Alkali-Carbonate Reaction (ACR) 83
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

of the so-called ACR, i.e., dedolomitization rims, carbonate haloes, brucite, Mg-silicate
gel and ASR gel, can be reproduced. Besides, owing to the larger size of the aggregate
than that in the mortar-bar tests (ASTM C1260, RILEM AAR-2), chert aggregates
produce deleterious expansion without undergoing excessive dissolution of ASR sol/gel
into the NaOH solution (Yamada et al., 2015).

3.5.4 Chinese Autoclave Test

Concrete microbars 20 by 20 by 60 mm are used in this test (Tang et al., 1994). The
aggregate size is 5-10 mm, the aggregate:cement ratio is 1 and the water/cement ratio
of 0.3. The alkali content of the cement is adjusted to 1.5% Na2O equivalent by
adding KOH to the mixing water. After demolding the bars are steamed at 100ºC for
4 hours and then autoclaved at 150 ºC. Expansions in excess of 0.1% after 6 hours of
autoclaving are considered to be indicative of potentially deleterious expansion of
ACR aggregates, (Guangren, 2001). However, this condition (150ºC) is too vigorous
for well-crystallized carbonate aggregates with distinctive cleavage planes within
them because it produces irreversible thermal expansion (Katayama, 2011).
Coarsely crystalline, non-reactive limestone from Nanjing produced abnormal
expansion, independent of the alkali level of the cement and curing conditions (Mu
et al., 1996), that is, coarser aggregate (5-10mm) gave larger expansion (0.09%),
comparable to the threshold for deleterious expansion in this test (0.10%). Frequent
heat treatment for measuring the length changes at various storage ages was respon-
sible for the larger expansion.
According to Tong (1994), compacted mortar specimens (9 mm in diameter by 30
mm in length) containing the limestone or coarsely crystalline magnesite aggregate also
produced significant expansion with the same autoclaving condition (150ºC). Calcite
and magnesite were decomposed into portlandite and brucite, although the total solid
molar volumes should decrease after the reactions (Tang & Deng, 2004). Because the
aggregates tested were coarsely crystalline rocks and the heat cycles applied were
frequent, irreversible thermal expansion must have occurred with their specimens.
Hence, care should be taken in interpreting the expansion data of the autoclave
method.

3.5.5 Chemical Screening Test

This test procedure (CSA A23.2-26A: 2014) consists of determining the aluminum
oxide, the calcium oxide and the magnesium oxide contents of pulverized carbonate
rock samples and plotting a graph of Al2O3 versus the CaO:MgO ratio. The potential
expansivity of the aggregate is determined from the position of the plotted point on the
graph (Figure 3.19). This test procedure was developed by analysis of a large suite of
carbonate rocks from Ontario for the Canadian Standards Association by Rogers
(personal communication, 2000). This method was also used successfully by the
authors to determine the potential expansivity of argillaceous dolomitic limestone
from Kentucky, U.S.A.
According to Jensen (2009), about 6% of Norwegian carbonate aggregates tested
was “considered potentially expansive” and therefore potentially ACR-reactive on
this diagram, but ACR has not been diagnosed in Norway. Considering that ASR of
 84 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

100 AGGREGATES CONSIDERED

NON-EXPANSIVE
CaO : Mgo rao 50

20
AGGREGATES CONSIDERED
10
POTENTIALLY EXPANSIVE

2 AGGREGATES CONSIDERED NON-EXPANSIVE

0 1 2 3 4 5 6
Al2O3

Figure 3.19 Graph showing the chemical composition of expansive and non-expansive carbonate rocks
after CSA A23.2 26A (2014).

the cryptocrystalline quartz is the real cause of the deleterious expansion of the so-
called ACR aggregates, and that metamorphic carbonate rocks common in Norway
contain coarsely crystalline quartz with lower ASR reactivity, this diagram should
better be applied to the non-metamorphosed carbonate rocks in North America.
Another version of the screening diagram (CaO:MgO and insoluble residue) was
proposed by Rogers (1986), which is equivalent to the ratio of calcite-dolomite-
insoluble residue on the triangular diagram (Katayama, 1992), meaning that the
dolomitic limestones in Ontario with a certain range of impurity are deleteriously
expansive.

3.6 CONCLUSION

In 1992, Katayama reported on the occurrence of alkali-silica gel in concrete made with
ACR argillaceous dolomitic limestone indicating that the formation of gel was at least a
contributing factor in the expansion of the concrete. Up until 1992 ACR was generally
considered to have a different mechanism of expansion from that of concrete made
with classical ASR aggregates, and a number of mechanisms were put forward to
explain the mechanism of expansion, particularly in Canada, the U.S.A. and China.
Subsequently, Katayama (2004, 2006, 2010a) obtained more data to support the
formation of gel as the underlying mechanism of expansion of concrete containing
ACR aggregates, and Katayama’s work has been supported by observations of
Grattan-Bellew (2009, 2013), Katayama and Grattan-Bellew (2012) and Katayama
et al., 2016. It is concluded that the so-called alkali-carbonate reaction (ACR) is a
special case of alkali-silica reaction similar to that occurring in siliceous limestones.
The long-standing confusion about ACR is due to the inappropriate methods applied
in the sample preparation, observation and analysis, as well as to the misunderstanding
that the distinctive chemical reaction (dedolomitization) is directly or indirectly respon-
sible for the deleterious expansion.
 So-Called Alkali-Carbonate Reaction (ACR) 85
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

REFERENCES

ASTM C586. (2006) Standard Test Method for Potential Alkali Reactivity of Carbonate
Rocks as Concrete Aggregates (Rock Cylinder Method), Annual Book of ASTM
Standards, 2006 Section 4 Construction, Volume 04.02 Concrete and Aggregates.
ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA
19428–2959, pp. 312–316.
ASTM C1105. (2006) Standard Test Method for Length Change of Concrete Due to Alkali-
Carbonate Rock. Annual Book of ASTM Standards, 2006 Section 4 Construction, Volume
04.02 Concrete and Aggregates, ASTM International, 100 Barr Harbor Drive, PO Box C700,
West Conshohocken, PA 19428–2959, pp. 600–603.
ASTM C1260. (2004) Standard Test Method for Potential Alkali Reactivity of Aggregates
(Mortar-Bar Method), Annual Book of ASTM Standards, 2004 Section 4 Construction,
Volume 04.02 Concrete and Aggregates. ASTM International, 100 Barr Harbor Drive, PO
Box C700, West Conshohocken, PA 19428–2959, pp. 680–684.
Buck, A.D. (1969) Potential Alkali Reactivity of Carbonate Rock from Six Quarries.
Miscellaneous Paper C-69–15, U.S. Army Engineer Waterways Experiment Station
Vicksburg, Mississippi, U.S.A.
CSA A23.2 14A. (2014) Potential Expansivity of Aggregates (procedure for length change due to
alkali-aggregate reaction in concrete prisms at 38ºC. CSA Standards A23.1–09/A23.2–09
Concrete materials and methods of concrete construction/Test methods and standard prac-
tices for concrete. Canadian Standards Association, 5060 Spectrum Way Suite 100,
Mississauga, Ontario, Canada L4W 5N6, pp. 285–296.
CSA A23.2–26A. (2014) Determination of potential alkali-carbonate reactivity of quarried
carbonate rocks by chemical composition. CSA Standards A23.1–09/A23.2–09 Concrete
materials and methods of concrete construction/Test methods and standard practices for
concrete. Canadian Standards Association, 5060 Spectrum Way Suite 100, Mississauga,
Ontario, Canada L4W 5N6, pp. 366–370.
Deng, M. & Tang, M. (1993) Mechanism of dedolomitization and expansion of dolomitic rocks.
Cement Concrete Res., 33, 1397–1408.
Deng, M., Han, S.F., Lu, Y.N., Lan, X.H., Hu, Y.L. & Tang M.S. (1993) Deterioration of
concrete structures due to alkali-dolomite reaction in China. Cement Concrete Res., 23,
1040–1046.
Dolar-Mantuani, L. (1964) Expansion of Gull River Carbonate Rocks in Sodium Hydroxide.”
Highway Research Record, No. 45, pp.178–195.
Durand, B. (2000) More results about the use of lithium salts and mineral admixtures to inhibit
ASR in concrete. In: Bérubé, M.A., Fournier, B. & Durand, B. (eds.) Proceedings of the 11th
International Conference on Alkali-Aggregate Reaction in Concrete (ICAAR), Quebec,
Canada, pp. 623–632.
Fournier, B., Bérubé, M.A., & Frenette, J. (2000) Laboratory investigations for evaluating
potential alkali-reactivity of aggregates and selecting preventive measures against AAR;
What do they really mean? In: Fournier, B. & Durand, B. (eds.) Proceedings of the 11th
International Conference on Alkali-Aggregate Reaction in Concrete, Quebec City (Canada),
June 2000, CRIB, Université Laval, Québec, pp. 287–296.
Gillott, J.E. (1961) Chemistry of the Kingston Dolomitic Limestone. DBR Internal Report No.
186, National Research Council of Canada, 8p.
Gillott, J.E. & Swenson, E.G. (1969) Mechanism of the alkali-carbonate rock reaction. J Eng
Geol., 2, 7–23.
Grattan-Bellew, P.E. (1981) Evaluation of miniature rock prism test for determining the poten-
tial alkali-expansivity of aggregates. Cement Concrete Res., 11, 699–711.
 86 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

Grattan-Bellew, P.E. & Chan, G. (2013) Comparison of the morphology of alkali-silica gel
formed in limestones in concrete affected by the so-called Alkali-Carbonate Reaction (ACR)
and Alkali-Silica Reaction (ASR). Cement Concrete Res., 47, 51–54.
Grattan-Bellew, P.E. & Rogers, C.A. (2000) Expansion Due to Alkali-Carbonate Reaction:
Laboratory Prognosis Versus Field Experience. In: Fournier, B. & Durand, B. (eds.)
Proceedings of the 11th International Conference on Alkali-Aggregate Reaction in
Concrete, Quebec City, CRIB, Université Laval, Québec, pp. 41–49.
Grattan-Bellew, P.E., Mitchell, L.D., Margeson, J. & Deng, M. (2009) Is Alkali-Carbonate
Reaction just a Variant of Alkali-Silica Reaction, ACR = ASR? Cement Concrete Res., 40,
556–562.
Guangren, Q., Min, D. & Tang, M. (2001) ACR expansion of dolomites with mosaic textures.
Mag. Concrete Res., 53, (5), 327–336.
Hadley, D.W. (1961) Alkali Reactivity of Carbonate Rocks – Expansion and Dedolomitization.
Highway Research Board Proceedings, 40, pp. 462–474.
Hadley, D.W. (1964) Alkali-Reactive Carbonate Rocks in Indiana – A Pilot Regional
Investigation. Highway Research Record, No. 45, pp. 196–221.
Insley, H. & Fréchette, van D. (1955) Microscopy of Ceramics and Cements. New York,
Academic Press Inc. 286p.
Jensen, V. (2009) Alkali Carbonate Reaction (ACR) and RILEM AAR-0 Annex A: Assessment of
Potentially Reactivity of Carbonate Rocks. In: Proceedings of the 12th Euroseminar on
Microscopy Applied to Building Materials, Dortmund, Germany. pp. 15–19.
Katayama, T. (1992) A Critical Review of Carbonate Rock Reactions – Is their Reactivity Useful
or Harmful? In: Proceedings of the 9th International Conference on Alkali-Aggregate
Reaction in Concrete, London England, July 1992. pp.508–518.
Katayama, T. (2004) How to identify carbonate rock reactions in concrete. Materials
Characterization, 53 (2–4), 85–104.
Katayama, T. (2006) Modern Petrography of Carbonate Aggregates in Concrete – Diagnosis
of So-Called Alkali-Carbonate Reaction and Alkali-Silica Reaction. In: Proceedings of
the Professor Marc-André Bérubé Symposium, Oral Presentation, 22p.
Katayama, T. (2008) ASR Gel in Concrete Subject to Freeze/thaw Cycles – Composition
between Laboratory and Field Concretes from Newfoundland, Canada. In: Proceedings
of the 13th International Conference on Alkali-Aggregate Reaction, Trondheim, Norway.
pp. 174–183.
Katayama, T. (2010a) The so-called Alkali-Carbonate Reaction (ACR) – Its Mineralogical
and Geochemical details, with special reference to ASR. Cement Concrete Res., 40,
643–675.
Katayama, T. (2010b) Diagnosis of Alkali-Aggregate Reaction – Polarizing microscopy and
SEM-EDS analysis. In: Castro-Borges, P., Moreno, E.I., Sakai, K., Gjorv, O.E. &
Banthia, N. (eds.) Proceedings of the 6th International Conference on Concrete under
Severe Conditions, Environment and Loading (CONSEC’10). Merida, Yukatan, Mexico.
Concrete under Severe Conditions. London, Taylor & Francis Group. pp. 19–34.
Katayama, T. (2011) So-called Alkali-Carbonate Reaction – Petrographic Details of Field
Concretes in Ontario. In: Proceedings of the 13th Euroseminar on Microscopy Applied to
Building Materials, Ljubljana, Slovenia. 15p.
Katayama, T. (2012a) Rim-Forming Dolomitic Aggregate in Concrete Structures in Saudi
Arabia – Is Dedolomitization Equal to the So-Called Alkali-Carbonate Reaction? In:
Proceedings of the 14th International Conference on Alkali-Aggregate Reaction in
Concrete. Austin, USA, 10p. Paper 030411-KATA-01.
Katayama, T. (2012b) Petrographic Study of Alkali-Aggregate Reactions in Concrete. Doctoral
Thesis (Science), The University of Tokyo, Japan.
 So-Called Alkali-Carbonate Reaction (ACR) 87
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

Katayama, T. (2012c) ASR Gels and Their Crystalline Phases in Concrete – Universal Products in
Alkali-Silica, Alkali-Silicate and Alkali-Carbonate Reactions. In: Proceedings of the 14th
International Conference on Alkali-Aggregate Reaction in Concrete, Austin, USA. 12p.
Paper 030411-KATA-03.
Katayama, T. & Grattan-Bellew, P.E. (2012) Petrography of the Kingston Experimental
Sidewalk at Age of 22 Years – ASR as the Cause of Deleterious Expansive, So-Called
Alkali-Carbonate Reaction. In: Proceedings of the 14th International Conference on Alkali-
Aggregate Reaction in Concrete. Austin, USA. 10p. Paper 030411-KATA-06.
Katayama, T. & Sommer, H. (2008) Further Investigation of the Mechanism of So-Called Alkali-
Carbonate Reaction Based on Modern Petrographic Techniques. In: Proceedings of the 13th
International Conference on Alkali-Aggregate Reaction in Concrete, Trondheim, Norway.
pp. 850–860.
Katayama, T., Jensen, V., & Rogers, C. (2016) The Enigma of the ‘so-called’ Alkali-Carbonate
Reaction. In: Proceedings of the Institution of Civil Engineers: Construction Materials, 169
(CM4). pp. 223–232.
Katayama, T., Ochiai, M., & Kondo, H. (1996). Alkali-Reactivity of Some Japanese Carbonate
Rocks Based on Standard Tests. Proceedings of the 10th International Conference on Alkali-
Aggregate Reaction in Concrete, Melbourne, Australia. pp. 294–301.
Katayama, T., Oshiro, T., Sarai, Y., Zaha, K., & Yamato, T. (2008) Late-expansive ASR Due to
Imported Sand and Local Aggregates in Okinawa Island, Southwestern Japan. In:
Proceedings of the 13th International Conference on Alkali-Aggregate Reaction in
Concrete, Trondheim, Norway. pp. 862–873.
Lemish, J. & Moore, W.J. (1964) Carbonate Aggregate Reactions: Recent Studies and an
Approach to the Problem. Highway Research Record, No. 45, pp. 57–71.
López-Buendía, A.M., Climent, V., & Verdu, P. (2006) Lithological influence of aggregate in the
alkali-carbonate reaction. Cement Concrete Res., 36, 1490–1500.
López-Buendía, A.M., Climent, V., Mar Urquiola, M., & Bastida, J. (2008) Influence of
Dolomite Stability on Alkali-Carbonate Reaction. In: Proceedings of the 13th International
Conference on Alkali-Aggregate Reaction in Concrete, Trondheim, Norway. pp. 233–242.
López-Buendía, A.M., Climent, V., & Verdu, P. (2006) Lithological influence of aggregate in the
alkali-carbonate reaction. Cement Concrete Res., 36, 1490–1500.
Lu, D., Fournier, B., Grattan-Bellew, P.E., Zhongzi, X., & Tang, M. (2008) Development of a
universal accelerated test for alkali-silica and alkali-carbonate reactivity of concrete aggre-
gates. Mater. Struct, 41, 235–246.
Lu, D., Mei, L., Xu, Z., Tang, M., & Fournier, B. (2006) Alteration of alkali reactive aggregates
autoclaved in different alkali solutions and application to alkali-aggregate reaction in con-
crete (I) alteration of alkali reactive aggregates in alkali solutions. CementConcrete Res., 36,
1176–1190.
Melanesi, C.A., Marfil, S.A., Batic, O.R., & Maiza, P.J. (1996) The Alkali-Carbonate Reaction
and Its Reaction Products – An Example with Argentinean Dolomite Rocks. Cement Concrete
Res., 26, 1579–1591.
Mu, X., Xu, Z., Deng, M. & Tang, M. (1996) Abnormal Expansion of Coarse-Grained Calcite in
the Autoclave Method. In: Proceedings of the 10th International Conference on Alkali-
Aggregate Reaction in Concrete, Melbourne, Australia. pp. 310–315.
NF P 18-454 (2004) Reactivity of a concrete formula with regard to the alkali-aggregate
reaction, performance test, French Association of National Standards, Paris (in French).
Newlon, H.H., Jr. & Sherwood, C. (1962) An Occurrence of Alkali-Reactive Carbonate Rock in
Virginia. Highway Research Board Bull, No. 355, pp. 27–43.
Nixon, P.J. & Sims, I. (eds.) (2016) RILEM Recommendations for the Prevention of Damage by
Alkali-Aggregate Reactions in New Concrete Structures, RILEM State-of-the-Art Reports,
Volume 17, Springer, Dordrecht, RILEM, Paris, 168p.
 88 Grattan-Bellew & Katayama
Downloaded By: 10.3.98.104 At: 07:39 06 Dec 2019; For: 9781315708959, chapter3, 10.1201/9781315708959-3

Ozol, M.A. (1994) Alkali-Carbonate Rock Reaction. ASTM STP 169C, pp. 372–387.
Ozol, M.A. & Newlon, H.H. (1974) Bridge Deck Deterioration Promoted by Alkali-Carbonate
Reaction: A Documented Example. Transportation Research Record, No. 525, pp. 55–63.
Qian, G., Deng, M. & Tang M. (2001) ACR expansion of dolomites with mosaic textures.
Magazine of Concrete Research, 53 (5), 327–336.
Rogers, C.A. (1985) Alkali Aggregate Reactions, Concrete Aggregate Testing and Problem
Aggregates in Ontario – A Review. Ontario Ministry of Transportation and
Communications, Engieering Materials Office, Report EM-41, 5th Revised Edition, 44p.
Rogers, C.A. (1986) Evaluation of the potential for expansion and cracking of concrete caused
by the alkali-carbonate reaction. Cement Concrete Aggr., 8, 13–23.
Rogers, C.A. (2000) Personal Communication.
Rogers, C.A. & Hooton, R.D. (1992) Comparison between Laboratory and Field Expansion of
Alkali-Carbonate Reactive Concrete. In: Proceedings of the 9th International Conference on
Alkali-Aggregate Reaction in Concrete, London, pp. 877–884.
Sommer, H., Grattan-Bellew, P.E., Katayama, T. & Tang, M. (2004) Development and Inter-
Laboratory Trial of the RILEM AAR-5: Rapid Preliminary Screening Test for Carbonate
Aggregates. In: Proceedings of the 12th International Conference on Alkali-Aggregate
Reaction, Vol. 1, Beijing, pp. 407–412.
Stanton, T.E. (1940) Expansion of Concrete Through Reaction Between Cement and Aggregate.
In: Proceedings of the American Society of Civil Engineers, December 1940, pp. 1781–1811.
Swenson, E.G. (1957) A Canadian Reactive Aggregate Undetected by ASTM Tests. ASTM
Bulletin No. 226, December 1957, pp. 48–51.
Swenson, E.G. & Gillott, J.E. (1960) Characteristics of Kingston Carbonate Rock Reaction.
Highway Research Board Bull. No. 275, pp. 18–31.
Swenson, E.G. & Gillott, J.E. (1964) Alkali-Carbonate Rock Reaction. Highway Research
Record, No. 45, pp. 21–40.
Tang, M. & Lu, Y. (1986) Rapid Method for Determining the Alkali-Reactivity of Carbonate
Rock. In: Grattan-Bellew, P.E. (ed.) Concrete Alkali-Aggregate Reactions, Noyes
Publications, Park Ridge, New Jersey, U.S.A., pp. 286–287.
Tang, M. & Deng, M. (2004). Progress on the Studies of Alkali-Carbonate Reaction. In:
Proceedings 12th International Conference on Alkali-Aggregate Reaction in Concrete,
Beijing, China, pp. 51–59.
Tang, M., Deng, M. & Zhongzi, X. (2000) Comparison between Alkali-Silica Reaction and
Alkali-Carbonate Reaction. In: Bérubé, M.A., Fournier, B. & Durand, B. (eds.) Proceedings
11th International Conference on Alkali-Aggregate Reaction in Concrete, Québec City,
pp.109–117.
Tang, M.S., Lan, X. & Han, S.F. (1994) Autoclave method for identification of Alkali Reactive
Carbonate Rocks. Cement Concrete Comp., 16, 163–167.
Tong, L. 1994. “Alkali-Carbonate Rock Reaction,” Ph.D thesis, Nanjing Institute of Chemical
Technology, Nanjing, China.
Tong, L. & Tang, M. (1999) Expansion mechanism of alkali-dolomite and alkali-magnesite
reaction. Cement Concrete Comp., 21, 361–373.
Walker, H.N. (1978) Chemical Reactions of Carbonate Aggregates in Cement Paste, ASTM
Special Technical Publication 169B, Chapter 41, pp.722–743.
Yamada, K., Kawabata, Y., Ogawa, S., & Maruyama, I. (2015). A study on alkali-aggregate
reaction in nuclear power related facilities. Cement Science & Concrete Technology., 68,
457–464, see Figure 2 (in Japanese).