Cement and Concrete Research 123 (2019) 105796

Contents lists available at ScienceDirect

Cement and Concrete Research

journal homepage: www.elsevier.com/locate/cemconres

Alkali-silica reactivity of alkali-activated concrete subjected to ASTM C 1293 T

and 1567 alkali-silica reactivity tests
⁎
Zihui Li, Robert J. Thomas, Sulapha Peethamparan
Department of Civil and Environmental Engineering, Clarkson University, 8 Clarkson Avenue, Box 5710, Potsdam, NY 13699, USA

A R T I C LE I N FO A B S T R A C T

Keywords: This paper presents an experimental investigation of the alkali-silica reactivity of alkali-activated concrete
Alkali-activated concrete (AAC). Some researchers suggest that the high alkalinity of the pore solution could make AAC more susceptible
Alkali-silica reaction to alkali-silica reaction (ASR) than comparable ordinary Portland cement (OPC) concrete, perhaps even with
Fly ash aggregates that are normally considered non-reactive. Many have further questioned the suitability of stan-
Slag, petrography
dardized ASR test methods for use with AAC. In an attempt to resolve this controversy, the authors subjected
OPC, alkali-activated fly ash (AAF), and alkali-activated slag (AAS) concrete made with both reactive and non-
reactive aggregates to accelerated and long-term ASR test protocols (ASTM C1567 and C1293). Paradoxically,
some AAS mixtures with non-reactive aggregates showed significant expansion while AAF mixtures with reactive
aggregates expanded minimally. AAS mixtures with reactive aggregates underwent extremely high expansions.
Microstructural evaluation using scanning electron microscopy revealed significant cracking but did not identify
ASR gel in the majority of specimens.

1. Introduction features of the alkali-silica reaction, potential mechanisms, and miti-

Alkali-silica reaction (ASR) is a deleterious reaction in concrete
between highly alkaline pore fluid and reactive amorphous or crypto-
crystalline silica in aggregate that occurs over time. Reactive ag-
gregates, or those whose use may result in ASR, include those con-
taining amorphous silica (silicate glasses and opal), unstable crystalline
polymorphs of silica (cristobalite and tridymite), poorly formed crys-
talline silica, quartz with defects in its lattice structure and micro-
crystalline quartz-bearing rocks. In the presence of moisture, the silic-
eous aggregates react with alkaline pore solution to form an expansive
alkali-silica gel. Swelling of the alkali-silica gel generates pressure
within concrete and manifests physically as expansion, map-cracking,
spalling, or pop-outs [1–14]. The composition of the alkali-silica gel
depends on the porosity, particle size, degree of crystallinity of the
aggregate, and the location of the gel formed. The chemical composi-
tion of gel mainly consists of silica, calcium, potassium and sodium
[15–18]. The ability of the ASR gel to absorb moisture and swell is
determined by the Ca/Si ratio; ASR gels with a Ca/Si-ratio of about 0.20
show a maximum in moisture adsorption and consequently have the
highest swelling potential [19–21].
Since its discovery by Thomas Stanton in the 1930s [22], re-
searchers have studied ASR in OPC concrete extensively. The general
gation methods are well-documented in the literature [1–12]. It is clear
that a sufficient quantity of reactive silica, alkalinity, and moisture must
be present before ASR will occur. Some evidence [5,7] suggests that the
presence of calcium hydroxide is also required before ASR will occur,
although its exact role in the formation of ASR gel has not been iden-
tified. Chappex and Scrivener [23] showed that concrete made with
class F fly ash was more resistant to ASR; they attributed this to the
reduction in the amount of calcium hydroxide and the presence of more
dissolved alumina ions in the pore solution.
Recently, alkali-activated slag (AAS) cement or fly ash concrete
(AAF) has been proposed as a promising alternative to OPC. A strong
alkali activates slag cement or fly ash, producing a cementitious binder.
Although ASR in OPC concrete is well understood [24], very little is
known about silica dissolution in alkali-activated fly ash and slag
binder concretes. On one hand, the alkalinity of the pore solution in
alkali-activated concrete is much higher than that of an OPC concrete,
which suggests increased susceptibility to ASR. On the other hand, the
reaction products in AAF and AAS concrete contain only traces of cal-
cium hydroxide and have more dissolved alumina in the pore solution,
suggesting ASR may be unlikely to occur. Experimental investigations
by several researchers have produced conflicting conclusions, as sum-
marized in part by Pacheco-Torgal [25]. Furthermore, at least one study

⁎
Corresponding author.
E-mail addresses: zihli@clarkson.edu (Z. Li), rthomas@clarkson.edu (R.J. Thomas), speetham@clarkson.edu (S. Peethamparan).

https://doi.org/10.1016/j.cemconres.2019.105796
Received 6 August 2018; Received in revised form 4 June 2019; Accepted 19 June 2019
Available online 09 July 2019
0008-8846/ © 2019 Elsevier Ltd. All rights reserved.
Z. Li, et al. Cement and Concrete Research 123 (2019) 105796

suggests that existing test methods for alkali-silica reactivity are in- premixed reagent grade aqueous solution of 26.5% SiO2 and 10.6%
effective when used for alkali-activated binders [26]. Na2O by mass. Sodium hydroxide was in the form of food grade pellets
ASTM standard test methods for alkali-silica reactivity of aggregates of 99% pure NaOH. The activator concentration was specified by the
and cement (i.e., ASTM C 227, C 1293, C 1260 and C 1567) rely on silica modulus (Ms, the mass ratio of silica to sodium oxide in the ac-
expansion as an indicator of ASR. Davidovits [27] reported shrinkage in tivator) and the equivalent sodium oxide dosage (%Na2O, by mass of
alkali-activated class C fly ash mortars under ASTM C 227, while OPC binder). Activator solutions were prepared at least 24 h in advance of
mortars under the same conditions expanded significantly. Jimenez and use to allow for adequate dissolution and thermal stability.
Puerta [26] suggested that some expansion could occur in AAS mortars The coarse aggregate was a quarried crushed limestone with a bulk
exposed to these test methods, but would be much smaller in magnitude unit weight of 1560 kg/m3, a specific gravity of 2.46, and a nominal
than that in OPC binders. Puertas et al. [28] confirmed that ASR could maximum particle size of 19 mm. The coarse aggregate met the gra-
occur in AAS concrete containing reactive siliceous aggregates. Gifford dation requirements of ASTM C1293. Two types of fine aggregate were
and Gillott [29] did the standard dimensional change tests for OPC and used: reactive (R) and non-reactive (N). Both fine aggregates met the
AAS concrete using aggregate from six sources in Canada, and they gradation requirements of ASTM C 1567. The non-reactive fine ag-
suggested that AAS concrete is less susceptible to deleterious expansion gregate was natural sand with a specific gravity of 2.65 and fineness
due to ASR compared to OPC concrete. Wang et al. [30] concluded that modulus of 2.54. The reactive aggregate was Spratt siliceous limestone
AAS concrete suffers no ASR because hydration products bind 80% of sourced in Ottawa, ON. Spratt limestone is known to be highly alkali-
the alkali at one year. Additionally, Garcia-Lodeiro et al. [31] found silica reactive when used in OPC concrete [36]. The reactive aggregate
that AAF concrete has smaller ASR expansion compared to that of OPC was mechanically crushed and fractionated to provide the required
under test method ASTM C1260, and they also proved that calcium gradation.
plays a vital role during the ASR gel expansion process. Conversely,
Bakharev et al. [32] found that AAS concrete exhibited higher expan- 2.2. Experimental methods
sion compared to OPC under the long-term test method ASTM C 1293,
and concluded that AAC is more vulnerable to ASR compared to OPC. The experiments were performed in three stages. The authors first
With respect to effects of aggregate type, Puertas et al. [28] also in- performed standardized accelerated and long-term ASR tests, which use
dicated that, compared to calcareous aggregates, siliceous aggregates expansion as an indicator of ASR. Next, the authors used backscattered
are more prone to ASR in AAS mixtures. Thomas [33] and Li et al. scanning electron microscopy (BSEM) and energy dispersive X-ray
[34,35] observed mixed behavior in sodium silicate activated class C fly spectroscopy (EDS) to look for cracking and ASR gel within the mi-
ash and slag mortar and concrete made with reactive Spratt aggregate crostructure to determine if any observed expansion was due to ASR. In
and identified the need for more advanced analyses. cases where ASR was suspected but no ASR gel was found, the authors
The lack of consensus in the literature leaves several important further used secondary mode scanning electron microscopy (SSEM) to
questions related to the alkali-silica reactivity of AAF and AAS binders look for the gel without the influence of any preparation method that
unanswered. This research seeks to address these gaps in knowledge by: could potentially obscure or destroy the gel.
Alkali-silica reactivity was evaluated by the long-term concrete
1. Evaluating the susceptibility of AAS and AAF concrete mixtures prism test method in accordance with the specification of ASTM C1293
prepared with non-reactive and reactive aggregates to deleterious and by the accelerated mortar bar method in accordance with the
ASR; specification of ASTM C1567. For the long-term method,
2. Investigating the effectiveness of ASTM standard ASR test methods 50.8 × 50.8 × 254-mm concrete prisms were stored vertically over
C1293 and ASTM C1567 in identifying the occurrence of ASR in AAS water at 100% RH and 38 °C for one year and the change in length and
and AAF concrete; and mass were recorded periodically during that time. For the accelerated
3. Validating the ASTM C1293 and ASTM C1567 expansion test results method, 25.4 × 25.4 × 254-mm mortar bars were stored submerged in
with microstructural analyses in cases where ASR is suspected. 1 N NaOH at 80 °C for 14 d and the change in length and mass were
recorded. Both test methods use expansion as an indicator of ASR.
2. Materials and methods Under the long-term test method, expansion in excess of 400 μɛ (0.04%)
after one year suggests that ASR may have occurred. Under the ac-
2.1. Materials celerated test method, expansion in excess of 1000 μɛ (0.1%) after 14 d
suggests that ASR may have occurred. However, neither provides con-
Binders included ASTM C150 Type I OPC, ASTM C989 Grade 100 clusive evidence.
ground granulated blast furnace slag, and ASTM C618 Class C coal fly BSEM/SSEM observations were made using a field emission SEM
ash. Table 1 lists the oxide composition of the binders as determined by under 15-kV acceleration voltage, 12-μA probe current, and 5 × 10−4
X-ray fluorescence spectrometry. Pa chamber pressure. Microscopy specimens were sawn from dummy
The alkali activator used was a compound solution of sodium sili- ASR specimens that had been subject to the long-term ASR test condi-
cate and sodium hydroxide. Sodium silicate was in the form of a tion described above. The sawn samples were soaked in isopropanol for
three days followed by drying in an oven at 50 °C for 48 h. These spe-
Table 1 cimens were epoxy-impregnated, polished, and sputter-coated for BSEM
Oxide composition of binders. analysis following the procedure described by Deir et al. [37]. For the
Oxide Composition (mass-%) secondary mode (SSEM) analysis, the dried specimens were fractured
and the freshly fractured surface was sputter coated and imaged. The
OPC Slag Fly ash elemental composition was determined by energy dispersive X-ray
spectroscopy (EDS) using NORAN System Six X-ray Microanalysis
CaO 65.00 39.10 21.1
SiO2 21.10 37.80 40.7 software package.
Al2O3 6.20 7.91 21.2
MgO 1.60 10.30 4.0 2.3. Mixture proportions
Na2O 0.09 0.30 1.5
Fe2O3 2.90 0.43 5.4
Table 2 lists the mixture proportions, which included six alkali-ac-
SO3 2.00 2.60 2.4
K2O 0.30 0.47 0.6 tivated slag (AAS), five alkali-activated fly ash (AAF) mixtures and one
OPC mixture with non-reactive (N) aggregate and one each AAS, AAF,

2
Z. Li, et al. Cement and Concrete Research 123 (2019) 105796

Table 2
Mixture proportions.
Mixture Binder Aggregate %Na2O Ms s/b ASTM C1293 ASTM C1567

3 3
Binder (kg/m ) Solution (kg/m ) Binder (kg/m3) Solution (kg/m3)

S1 Slag N 4.0 0.75 0.40 420 168 693 277

S2 N 5.0 0.40 420 168 693 277
S2-R R 5.0 0.40 420 168 693 277
S3 N 4.0 1.5 0.40 420 168 693 277
S4 N 5.0 0.40 420 168 693 277
S5 N 5.0 0.45 420 189 693 312
S6 N 5.0 0.50 420 210 693 347
F1 Fly ash N 4.0 1.5 0.40 420 168 693 277
F2 N 5.0 0.40 420 168 693 277
F2-R R 5.0 0.40 420 168 693 277
F3 N 6.0 0.40 420 168 693 277
F4 N 5.0 0.45 420 189 693 312
F5 N 5.0 0.50 420 210 693 347
OPC OPC N – – 0.40 420 168 693 277
OPC-R R 420 168 693 277

Note: The aggregate quantity and gradation comply with the requirements in ASTM C1293 and ASTM C 1567, respectively.

and OPC mixture with reactive (R) aggregate. The table lists the sodium the expansion in AAS concrete approached the limit of 400 μɛ after one
oxide equivalent dosage (%Na2O, by mass of binder), the silica modulus year. While expansion was highest in AAS the expansion in AAF con-
(Ms, the mass ratio of silica to sodium oxide in the activator solution), crete was generally much lower than that in OPC concrete. As expected,
and the solution-to-binder ratio (s/b). Most mixtures used s/b = 0.40, regardless of binder type (AAS, AAF, or OPC), expansion was several
except those with high activator concentrations that required additional times higher for concretes with reactive aggregate than for those with
water to retain adequate workability. Specimens were moist-cured at non-reactive aggregate. This difference became apparent after about
23 ± 2 °C. These mixtures were selected because they have been used 28 d of exposure; before this, the expansion trends were similar be-
extensively for other research [38–40]. tween reactive and non-reactive aggregates. The expansion in AAF
concrete with reactive aggregate did not reach the 400 μɛ limit even
though it was several times higher than the expansion observed with
3. Results and discussion non-reactive aggregate.
The change in mass in alkali-activated concretes (AAC) with non-
3.1. Expansion reactive aggregate was much lower than in OPC concrete with non-
reactive aggregate, but the trend over time was similar. Beginning
Table 3 summarizes the expansion and mass change results from around 100 d, the change in mass in OPC concrete was higher with
accelerated and long-term ASR testing of AAS, AAF, and OPC concrete reactive aggregate, but only by a small margin (10%). In comparison,
and mortar with normal and non-reactive aggregate. The reader will the change in mass in AAC was 50–100% higher with reactive ag-
recall that expansions in excess of 1000 μɛ for the accelerated test gregate than with non-reactive aggregate. Furthermore, the trend in
method (ASTM C1567) and 400 μɛ for the long-term test method (ASTM mass change over time was quite different with AAC containing reactive
C1293) suggest the potential for deleterious ASR. Expansion values in aggregate. Instead of reaching a plateau after 180 d (as in AAC with
excess of these limits are italicized in the table. non-reactive aggregate and Portland cement concrete with both ag-
Fig. 1 shows the expansion and mass change of AAS, AAF, and OPC gregate types), the mass increase in AAC with reactive aggregate began
concrete with non-reactive and reactive fine aggregates under ASTM to rise sharply, reaching or exceeding 1.0% after one year. Increased
C1293 test method. In general, the expansion in AAS concrete was expansion (Fig. 1(a)) in AAS, AAF, and OPC concretes made with re-
higher than in OPC or AAF concrete. Even with non-reactive aggregate, active fine aggregate suggest the occurrence of ASR. However, mass
change (Fig. 1(b)) was not always correlated with the expansion trend.
Table 3 For example, in Fig. 1(b) the mass change for S2 is almost stabilized
Summary of results for accelerated and long-term ASR testing.
after 200–300 days but the expansion continues and even accelerated.
Mixture ASTM C1567 (14 d) ASTM C1293 (365 d) This means that simply using the internal water, the ASR gel may ex-
pand without changing the mass of the samples.
Expansion (με) Mass change (%) Expansion (με) Mass change (%)
The expansion data from ASTM C1293 suggest that ASR is likely to
S1 304 0.97 334 0.76 have occurred in AAS, AAF, and OPC concrete made with reactive fine
S2 837 0.68 394 0.70 aggregate and is not likely to have occurred in any of these binders with
S2-R 1723 1.64 2470 1.12 non-reactive aggregates. In presence of reactive aggregate, AAS ex-
S3 827 1.52 337 0.83
panded the most and AAF expanded the least.
S4 890 0.62 330 1.01
S5 814 0.74 387 1.28 Fig. 2 and Fig. 3 show the expansion and change in mass of AAS
S6 804 0.84 326 1.27 concrete with varying activator concentration and solution-to-binder
F1 1057 1.48 84 0.57 ratio, respectively, under the long-term ASR test method ASTM C1293.
F2 460 1.08 54 0.51 Both figures reveal minimal differences, which are echoed by the ex-
F2-R 2954 1.38 390 0.96
F3 1564 0.88 40 0.84
pansion reported for the accelerated test method in Table 3. Since these
F4 430 1.40 137 1.31 specimens were made with non-reactive aggregates and the expansions
F5 377 1.78 120 1.18 are all relatively small, there is little reason to suspect ASR. Further-
OPC 624 0.54 180 0.91 more, there is no evidence that the activator concentration or the so-
OPC-R 2670 2.02 1040 1.05
lution-to-binder ratio has any major influence on the resulting expan-
Note: Expansion values in excess of ASTM limits are italicized in the table. sion.

3
Z. Li, et al.
Fig. 1. (a) Expansion and (b) change in mass of AAS, AAF, and OPC concrete with non-reactive and reactive (R) aggregate under ASTM C1293.
Fig. 4 and Fig. 5 show the expansion and change in mass of AAF
concrete with varying activator concentration and solution-to-binder
ratio, respectively, under the long-term ASR test method ASTM C1293.
The figures show some major differences in the long-term expansion
measurements, but the small magnitude of the measurements them-
selves makes it difficult to draw any significant conclusions. The most
noteworthy observation is that the short-term expansion measurement
reported in Table 3 are starkly different from those reported for the
long-term method. For example, the long-term expansion in mixture F3
was minimal, but the short-term expansion was 50% more than the
threshold for that test. The same is true for AAF mixture F1.
3.2. Microstructure and chemical composition
The microstructural investigation included a general examination of
several locations within the paste matrix, the aggregates, and the in-
terface between the two. Each location was inspected to look for
Fig. 2. (a) Expansion and (b) change in mass of AAS concrete with varying activator concentration under ASTM C1293 (non-reactive aggregates).
4
Cement and Concrete Research 123 (2019) 105796
cracking or gels within cracks, both of which could indicate the oc-
currence of ASR. EDS was used to determine the composition of any
gels observed within cracks.
The micrographs in Fig. 6(a) and (b) are representative of the mi-
crostructure in OPC concrete with reactive aggregate after one year
under the long-term ASR test condition. The figure shows a crack in a
reactive aggregate extending through the interfacial transition zone and
into the paste matrix. This type of cracking was present throughout the
specimen and is consistent with the type of cracking typically observed
with ASR. A spongy gel filled the cracks, as observed in Fig. 6(b).
Fig. 6(c) shows the typical composition of that gel, as determined by
EDS. While the relative concentrations of the elements varied de-
pending on the location of the gel (i.e., in the aggregate vs. in the paste
matrix or air voids), the composition was similar throughout. The pri-
mary constituents were silicon, calcium, and potassium. This is typical
of ASR gels observed elsewhere [11,12].
Similarly, the micrograph in Fig. 7 is representative of the
Z. Li, et al.
Fig. 3. (a) Expansion and (b) change in mass change of AAS concrete with varying s/b under ASTM C1293 (non-reactive aggregates).
Fig. 4. (a) Expansion and (b) change in mass of AAF concrete with varying activator concentration under ASTM C1293 (non-reactive aggregates).
microstructure in AAS concrete with non-reactive aggregate (specifi-
cally, mixture S3) after one year under the long-term ASR test method.
Despite this mixture including innocuous aggregate, the observed ex-
pansions under accelerated and long-term ASR testing were about 85%
of the corresponding thresholds. Widespread cracking was evident
within the microstructure but it was atypical of ASR cracking. Where
the cracks in the OPC microstructure (Fig. 6) extended through the
aggregates and into the paste matrix, those in the S3 microstructure
were confined to the paste matrix. Close observation of the cracks, as in
Fig. 7(b), showed them to be devoid of any gel products. This provides
conclusive evidence that the observed expansion in AAS with non-re-
active aggregate is not the result of ASR.
Similarly, the micrographs in Fig. 8(a) and (b) are representative of
the microstructure in AAS concrete with reactive aggregate (specifi-
cally, mixture S2-R) after one year under the long-term ASR test
method. The microstructure was extensively cracked. Like those in
5
Cement and Concrete Research 123 (2019) 105796
Fig. 7, most of the cracks were contained to the paste matrix and did not
extend through the aggregate. These cracks are inconsistent with those
normally resulting from ASR, which tend to extend through the ag-
gregates. A minority of the cracks did pass through the aggregate, as
shown in Fig. 8(b), but these cracks were devoid of the gel product
typical of ASR.
Expecting but failing to find conclusive evidence of ASR within the
S2-R microstructure, the authors hypothesized that the BSEM specimen
preparation procedure destroyed or removed the ASR gel, possibly
during polishing. SSEM specimens require minimal preparation, leaving
any ASR gel undisturbed. Still, most cracks in the S2-R microstructure
did not contain any gel products. A small number of cracks near the
aggregate-paste interface contained the rosette-shaped gel product
shown in Fig. 9(a). This product is consistent with the ASR gel product
observed elsewhere [26]. Fig. 9(b) shows the EDS spectrum of this
product. Despite the absence of potassium in this spectrum, it is
Z. Li, et al. Cement and Concrete Research 123 (2019) 105796

Fig. 5. (a) Expansion and (b) change in mass of AAF concrete with varying s/b under ASTM C 1293 (non-reactive aggregates).

Fig. 6. (a) and (b) Microstructure of OPC concrete with reactive aggregate (OPC-R) after one year under long-term ASR test conditions, (c) EDS spectrum of gel.

6
Z. Li, et al.
Fig. 7. (a) and (b) Microstructure of AAS concrete with non-reactive aggregate (S3) after one year under long-term ASR test condition.
consistent with the typical composition of ASR gel. The absence of
aluminum and the small fraction of calcium present in the spectrum
confirm that it is not N-A-S-H or C-A-S-H reaction product. The presence
of this gel confirms the occurrence of ASR in AAS with reactive Spratt
limestone, but the small quantity of gel found in the microstructure is
inconsistent with the very large expansion observed during ASR testing.
Short-term ASR testing of AAF mortars showed extremely large
expansions with both reactive and innocuous aggregates (Table 3). The
expansions in specimens with non-reactive aggregates were relatively
small under the long-term test method, which is generally considered
more reliable. With reactive aggregates (i.e., F2-R), the expansion after
one year was within 2% of the threshold value. The micrographs in
Fig. 10(a) and (b) are representative of the microstructure in AAF
concrete with reactive aggregate (specifically, mixture F2-R) after one
year under the long-term ASR test condition. Fig. 10(a) shows that the
cracking was primarily contained to the paste matrix and again did not
extend into the aggregates. Fig. 10(c) shows the EDS spectrum of the
product found within these cracks. The composition is high in silicon,
calcium, and aluminum, and is typical of the C-A-S-H AAF reaction
product. It is therefore evident that the observed expansion is not the
result of ASR.
3.3. Discussion
The results presented here indicate that, as expected, Spratt silic-
eous limestone is highly alkali-silica reactive in OPC concrete. This is
evidenced primarily by expansions in great excess of the thresholds for
both the accelerated and long-term expansion tests. Microstructural
observations confirmed that this expansion was the result of ASR; gel-
filled cracks were visible in the aggregate and in the paste-aggregate
Fig. 8. (a) and (b) Microstructure of AAS concrete with reactive aggregate (S2-R) after one year under long-term ASR test condition.
7
Cement and Concrete Research 123 (2019) 105796
interface. The composition of this gel was high in silicon, calcium, and
potassium, which is typical of ASR gel.
Accelerated testing of AAF concrete with both reactive and non-
reactive aggregate resulted in extremely high expansions. Long-term
testing of the same mixtures resulted in much smaller expansions, al-
though the expansion for AAF concrete with reactive aggregate was
near the 400 μɛ threshold. Microstructural investigations revealed sig-
nificant cracking within the paste matrix but did not show the through-
aggregate cracks typical in cases of ASR. EDS investigations showed
that the composition of the gel product within these cracks was con-
sistent with C-A-S-H, the primary reaction product of AAF, and not with
ASR gel. This provides relatively conclusive evidence that ASR did not
occur in AAF concrete with reactive Spratt limestone. Since Spratt
limestone is known to be highly reactive, this suggests that AAF con-
crete may be resistant to ASR, or at least much more resistant than OPC
concrete. This result should of course be verified by additional experi-
mentation. Further, these results suggest that the accelerated ASR test
method gives spuriously high expansion results when used with AAF
concrete, and that these expansions are not indicative of ASR. The long-
term test method should be used with this type of binder.
Accelerated and long-term ASR testing of AAS concrete with non-
reactive aggregate resulted in expansions that approached but did not
exceed the corresponding thresholds. Again, microstructural in-
vestigations showed significant cracking in the paste matrix but not in
the aggregates. Furthermore, the cracks did not include any gel re-
sembling the ASR product. It is therefore clear that no ASR occurred in
these mixtures despite relatively large expansions. The expansion in
AAS concrete with reactive Spratt limestone was significantly higher
than with innocuous aggregate. Similar cracking patterns were ob-
served in the microstructure under BSEM, but the cracks were again
Z. Li, et al. Cement and Concrete Research 123 (2019) 105796

Fig. 9. (a) Secondary mode image of ASR gel in air void of AAS concrete with reactive aggregate (S2-R) after one year under long-term ASR test condition and (b)
EDS spectrum of gel.

devoid of ASR gel. SSEM investigations also failed to find ASR gel in
most of the cracks. A few locations did contain a gel with morphology
and composition consistent with ASR gel, but these were the extreme
minority of cases.
Two realities are possible given the conflicting experimental evi-
dence regarding the alkali-silica reactivity of AAS concrete with
reactive Spratt limestone:
1. The observed expansion was the result of ASR, but the gel was either
destroyed or otherwise not visible in the microstructure under BSEM
or SSEM investigations; or
2. The observed expansion and cracking was caused by some other

Fig. 10. (a) and (b) Microstructure of AAF concrete with reactive aggregate (F2-R) after one year under long-term ASR test condition; (c) EDS spectrum of gel.

8
Z. Li, et al.
phenomenon.
The likelihood of these scenarios is discussed below.
If the expansion is the result of ASR, there must be some reaction
product. This product was not visible under BSEM. SSEM investigations
were performed in case the BSEM preparation procedure destroyed or
obscured the ASR gel. BSEM specimen preparation includes quenching
in isopropanol, drying, epoxy-impregnating, polishing with a diamond
paste, and sputter coating. It is possible that this procedure could re-
move the ASR gel from the microstructure in AAC. However, the ASR
gel was undisturbed in the OPC-R micrographs (Fig. 6). It is unlikely
that the ASR gel could survive the preparation procedure in OPC but
not in AAC, unless the chemistry or morphology of the gel is different.
There is some support for this conclusion in the literature. A recent
paper by Williamson and Juenger [41] obtained similar results, ob-
serving cracking but no ASR gel within the cracks. That paper suggested
that this phenomenon was caused by insufficient calcium, which re-
sulted in the growth of a low-viscosity gel. However, it was suggested
that the formation of such a gel would result in low expansions, which
were not observed here.
This leaves the second scenario, which implies that an unrelated
phenomenon caused all or most of the expansion observed during ASR
testing. The unusually high expansions observed in AAS specimens
made with innocuous aggregates tend to support this hypothesis.
Unfortunately, this does not provide any clues about the mechanism
behind said expansion. It could relate to continued product formation in
the alkali-activated binder systems, or to swelling of the C-A-S-H/N-A-
S-H gel. However, this would likely be accompanied by debonding of
the aggregates. The observed cracking could be explained in part by
drying as part of the SEM specimen preparation, but this does not ex-
plain the expansion. Additional work is necessary to identify the phe-
nomena at play.
Further research in this area should focus on microstructural in-
vestigations of ASR specimens at very late ages (i.e., in specimens
subject to ASR test conditions for a very long time). This will increase
the chances of finding any ASR gel and will help the researchers
identify the physical morphology and chemical characteristics of that
gel. Alternatively, this may reveal an alternative mechanism that may
be responsible for the observed expansion.
The apparent resistance of AAF concrete may be attributable to the
reduced calcium content in the reaction products. It is generally ac-
cepted that calcium hydroxide is essential to the ASR reaction in OPC
binders [4,7]. The reaction products in alkali-activated binders are
mainly N-A-S-H, C-A-S-H, and C-S-H with Ca/Si < 1.0 [42–47].
Meanwhile, hydrated OPC includes residual alite and belite, CH, and C-
S-H with Ca/Si near 1.7 [48]. Without the presence of a threshold
calcium hydroxide, as in AAF, the reaction may not be initiated.
The reader should note one confounding factor and two short-
comings in this work. First, AAC is generally prone to both autogenous
and drying shrinkage; this shrinkage could either temper the expansions
during the accelerated and long-term ASR tests or contribute to the
observed cracking. Second, the study only considered two types of ag-
gregate: one highly reactive and another innocuous. Third, the study
considered a relatively narrow subsection of AAC mixtures.
With respect to the first point: The authors and their colleagues have
conducted extensive analyses of shrinkage in AAC and suggest that
shrinkage is not a major factor in interpreting the results presented
here. Previous work has shown that most of the shrinkage in AAC oc-
curs at very early age [49,50], and is therefore complete before the
specimens enter the ASR exposure condition.
With respect to the second: This work does not address the spectrum
of reactivity that exists within the realm of alkali-reactive aggregates. In
particular, the conclusion regarding the resistance of AAF concrete to
ASR is only valid for this aggregate (and, as discussed below, for this
combination of binder and activator). Spratt limestone is known to be
highly reactive, but there exist a spectrum of reactive aggregates that
9
Cement and Concrete Research 123 (2019) 105796
react more weakly or more slowly. Given that Spratt limestone is so
highly reactive, it is likely that it represents a worst-case scenario and
that AAF concrete would experience even less deformation and
cracking with a more slowly reactive aggregate. Additional experi-
mentation is necessary to confirm.
Finally, with respect to the third point: Although this study con-
sidered a broad spectrum of activator concentrations, it only considered
a single activator type. Further, it only considered one type of fly ash
and one type of slag. It is possible that different binders and activators
could produce different results. Again, additional experimentation is
necessary.
4. Conclusions
This paper presents the results of an experimental investigation of
the alkali-silica reactivity of AAF and AAS concrete. The authors sought
to (1) evaluate the suitability of existing long-term and accelerated
standardized ASR test methods for prediction of the alkali-silica re-
activity of AAC and (2) determine if AAF and AAS concrete is suscep-
tible to ASR when made with reactive aggregates. The findings are
summarized below:
1. The Spratt siliceous limestone used in this study is highly reactive in
OPC systems. This is primarily evidenced by large expansions during
accelerated and long-term ASR tests. An extensive crack network
filled with a typical ASR gel product supports this evidence. This is
not a new revelation but provides context for the rest of the results
and conclusions presented here.
2. ASR does not occur in the AAF and AAS concrete mixtures tested
here when made with non-reactive limestone aggregates. Despite
some very large expansions in accelerated (ASTM C1567) and long-
term (ASTM C1293) ASR tests with reactive aggregate, the cracking
pattern observed within the microstructure was limited to the paste
matrix and was therefore inconsistent with the through-aggregate
cracks typically found in cases of ASR. Furthermore, the cracks did
not contain any products resembling ASR gel in either physical
morphology or chemical composition.
3. ASR does not occur in the AAF concrete mixtures tested here with
reactive Spratt limestone. Although extremely large expansions
were observed in accelerated ASR tests, the expansions during long-
term tests were comparatively small. Furthermore, the cracking
patterns observed in the microstructure were again confined to the
paste matrix and were filled with a C-A-S-H reaction product rather
than ASR gel.
4. ASR may or may not occur in the AAS concrete mixtures tested here
with reactive Spratt limestone. Large expansions were observed
under both the accelerated and long-term ASR tests. Furthermore,
cracks were observed in the microstructure, but these were again
confined to the paste matrix. Small infrequent deposits of ASR gel
were found in the microstructure, but only in a few locations and
only in secondary mode images. This means that either the observed
expansion is due to ASR but the gel differs in morphology or com-
position, making it difficult to observe, or the expansion is related to
some other phenomenon.
5. The apparent resistance of AAC to ASR is likely related to the re-
duced calcium content of both the unreacted binder phases and the
reaction products. The C-A-S-H, N-A-S-H, and C-S-H products in
AAC are all lower in calcium than the C-S-H and CH reaction pro-
duct in OPC systems, and it is generally agreed that CH contributes
to the initiation of ASR in the presence of moisture.
6. The accelerated ASR test is not appropriate for use with AAF con-
crete mixtures. The accelerated test data presented here show ex-
tremely large expansions in several AAF mixtures with and without
reactive aggregate. Meanwhile, long-term test data and micro-
structural analyses refute this evidence, showing that no ASR oc-
curred in these specimens.
Z. Li, et al. Cement and Concrete Research 123 (2019) 105796

Acknowledgment [24] A.E.K. Jones, L.A. Clark, The effects of ASR on the properties of concrete and the
implications for assessment, Eng. Struct. 20 (1998) 785–791.
[25] F. Pacheco-Torgal, Z. Abdollahnejad, A.F. Camoes, M. Jamshidi, Y. Ding, Durability
The authors gratefully acknowledge the financial support from The of alkali-activated binders: a clear advantage over Portland cement or an unproven
University Transportation Research Center, Region 2 (UTRC2 Award issue, Constr. Build. Mater. 30 (2012) 400–405.
No. 49198-35-27) through the University Transportation Centers pro- [26] A. Fernández-Jiménez, F. Puertas, The alkali–silica reaction in alkali-activated
granulated slag mortars with reactive aggregate, Cem. Concr. Res. 32 (2002)
gram, U.S.A. Any opinions, findings, and conclusions or recommenda- 1019–1024.
tions expressed in this article are those of the authors and do not ne- [27] J. Davidovits, Geopolymers: inorganic polymeric new materials, J. Therm. Anal.
cessarily reflect the views of the UTRC2. Calorim. 37 (1991) 1633–1656.
[28] F. Puertas, M. Palacios, A. Gil-Maroto, T. Vázquez, Alkali-aggregate behaviour of
alkali-activated slag mortars: effect of aggregate type, Cem. Concr. Compos. 31
References (2009) 277–284.
[29] P.M. Gifford, J.E. Gillott, Alkali-silica reaction (ASR) and alkali-carbonate reaction
(ACR) in activated blast furnace slag cement (ABFSC) concrete, Cem. Concr. Res. 26
[1] M.D. Thomas, B. Fournier, K.J. Folliard, Alkali-Aggregate Reactivity (AAR) Facts
(1996) 21–26.
Book, (2013).
[30] S.D. Wang, K.L. Scrivener, P.L. Pratt, Factors affecting the strength of alkali-acti-
[2] Z.P. Baz̆ant, A. Steffens, Mathematical model for kinetics of alkali–silica reaction in
vated slag, Cem. Concr. Res. 24 (1994) 1033–1043.
concrete, Cem. Concr. Res. 30 (2000) 419–428.
[31] I. García-Lodeiro, A. Palomo, A. Fernández-Jiménez, Alkali–aggregate reaction in
[3] S. Chatterji, The role of Ca (OH)2 in the breakdown of Portland cement concrete due
activated fly ash systems, Cem. Concr. Res. 37 (2007) 175–183.
to alkali-silica reaction, Cem. Concr. Res. 9 (1979) 185–188.
[32] T. Bakharev, J.G. Sanjayan, Y.-B. Cheng, Resistance of alkali-activated slag concrete
[4] Z. Li, Acid Leaching Resistance and Alkali Silica Reaction (ASR) of Alkali Activated
to alkali–aggregate reaction, Cem. Concr. Res. 31 (2001) 331–334.
Cement Free Binders, Clarkson University, 2017 (Ph.D. Thesis).
[33] R.J. Thomas, Properties and Performance of Alkali-Activated Concrete, Clarkson
[5] M. Thomas, The effect of supplementary cementing materials on alkali-silica reac-
University, 2016 (Ph.D. Thesis).
tion: a review, Cem. Concr. Res. 41 (2011) 1224–1231.
[34] Z. Li, R.J. Thomas, D. Lazama, S. Peethamparan, Alkali-silica reaction (ASR) in
[6] P.R. Rangaraju, J. Desai, Effectiveness of fly ash and slag in mitigating alkali–silica
cement free alkali-activated sustainable concrete, Final Repoert for Department of
reaction induced by deicing chemicals, J. Mater. Civ. Eng. 21 (2009) 19–31.
Transportation, 2016.
[7] H. Wang, J.E. Gillott, Mechanism of alkali-silica reaction and the significance of
[35] Z. Li, R.J. Thomas, S. Peethamparan, Evaluation of ASTM Methods for Detection of
calcium hydroxide, Cem. Concr. Res. 21 (1991) 647–654.
Alkali-Silica Reaction in Alkali-Activated Concrete. Proceedings of the 96Th Annual
[8] F.P. Glasser, Chemistry of the alkali-aggregate reaction, Alkali-Silica React. Concr.
Meeting of the Transportation Research Board, Number: 17–04375, (2017).
(1992) 30–53.
[36] C. Rogers, P.E. Grattan-Bellew, R.D. Hooton, J. Ryell, M.D. Thomas, Alkali-ag-
[9] P.J. Nixon, C.L. Page, Pore solution chemistry and alkali aggregate reaction, Spec.
gregate reactins in Ontario, Can. J. Civ. Eng. 27 (2000) 246–260.
Publ. 100 (1987) 1833–1862.
[37] E. Deir, B.S. Gebregziabiher, S. Peethamparan, Influence of starting material on the
[10] T. Miyagawa, Durability design and repair of concrete structures: chloride corrosion
early age hydration kinetics, microstructure and composition of binding gel in al-
of reinforcing steel and alkali–aggregate reaction, Mag. Concr. Res. 43 (1991)
kali-activated binder systems, Cem. Concr. Compos. 48 (2014) 108–117.
155–170.
[38] R.J. Thomas, E. Ariyachandra, D. Lezama, S. Peethamparan, Comparison of chloride
[11] M.D.A. Thomas, F.A. Innis, Effect of slag on expansion due to alkali aggregate re-
permeability methods for alkali-activated concrete, Constr. Build. Mater. 165
action in concrete, Mater. J. 95 (1998) 716–724.
(2018) 104–111.
[12] T.C. Powers, H.H. Steinour, An interpretation of some published researches on the
[39] R.J. Thomas, S. Peethamparan, Stepwise regression modeling for compressive
alkali-aggregate reaction part 2-a hypothesis concerning safe and unsafe reactions
strength of alkali-activated concrete, Constr. Build. Mater. 141 (2017) 315–324.
with reactive silica in concrete, in: J. Proc., (1955), pp. 785–812.
[40] R.J. Thomas, S. Peethamparan, Alkali-activated concrete: engineering properties
[13] Z. Li, S. Peethamparan, Leaching resistance of alkali activated slag and fly ash
and stress-strain behavior, Constr. Build. Mater. 93 (2015) 49–56.
mortar exposed to organic acid, Gre Mater. (2018) 1–14.
[41] T. Williamson, M.C.G. Juenger, The role of activating solution concentration on
[14] F. Naiqian, J. Hongwei, C. Enyi, Study on the suppression effect of natural zeolite on
alkali-silica reaction in alkali-activated fly ash concrete, Cem. Concr. Res. 83 (2016)
expansion of concrete due to alkali-aggregate reaction, Mag. Concr. Res. 50 (1998)
124–130.
17–24.
[42] J.C. Petermann, A. Saeed, M.I. Hammons, Alkali-Activated Geopolymers: A
[15] N. Thaulow, U.H. Jakobsen, B. Clark, Composition of alkali silica gel and ettringite
Literature Review, Applied research associates Inc., Panama City, FL, 2010.
in concrete railroad ties: SEM-EDX and X-ray diffraction analysis, Cem. Concr. Res.
[43] S. Song, D. Sohn, H.M. Jennings, T.O. Mason, Hydration of alkali-activated ground
26 (1996) 309–318.
granulated blast furnace slag, J. Mater. Sci. 35 (2000) 249–257.
[16] I. Fernandes, Composition of alkali-silica reaction products at different locations
[44] S.D. Wang, K.L. Scrivener, Hydration products of alkali-activated slag cement, Cem.
within concrete structures, Matr. Char. 60 (2009) 655–668.
Concr. Res. 25 (1995) 561–571.
[17] T. Katayama, Petrographic Study of the Alkali-Aggregate Reactions in Concrete,
[45] B.S. Gebregziabiher, R.J. Thomas, S. Peethamparan, Temperature and activator
Doctoral thesis (Science) Department of Earth and Planetary Science, Graduate
influence on early age hydration kinetics, Constr. Build. Mater. 113 (2016)
school of Science, University of Tokyo, 2012.
783–793.
[18] L.S. Dent Glasser, N. Kataoka, The chemistry of ‘alkali-aggregate’ reaction, Cem.
[46] B.S. Gebregziabiher, R.J. Thomas, S. Peethamparan, Very early-age reaction ki-
Concr. Res. 11 (1981) 1–9.
netics and microstructural development in akali-activated slag, Cem. Concr.
[19] A. Leemann, C. Merz, An attempt to validate the ultra- accelerated microbar and the
Compos. 55 (2015) 91–102.
concrete performance test with the degree of AAR-induced damage observed in
[47] C. Li, H. Sun, L. Li, A review: the comparison between alkali-activated slag
concrete structures, Cem. Concr. Res. 49 (2013) 29–37.
(Si + Ca) and metakaolin (Si + Al) cements, Cem. Concr. Res. 40 (2010)
[20] A. Leemann, S.G. Le, F. Winnefeld, D. Rentsch, B. Lothenbach, Alkali–silica reac-
1341–1349.
tion: the influence of calcium on silica dissolution and the formation of reaction
[48] E.M. Gartner, J.F. Young, D.A. Damidot, I. Jawed, Hydration of Portland cement,
products, J. Am. Ceram. Soc. 94 (2011) 1243–1249.
Struct. Perform. Cem. 2 (2002) 57–108.
[21] A. Gholizadeh Vayghan, F. Rajabipour, Quantifying the swelling properties of al-
[49] D.B. Kumarappa, S. Peethamparan, M. Ngami, Autogenous shrinkage of alkali ac-
kali-silica reaction (ASR) gels as a function of their composition, Cem. Concr. Res.
tivated slag mortars: basic mechanisms and mitigation methods, Cem. Concr. Res.
94 (2017) 49–58.
109 (2018) 1–9.
[22] S.E. Thomas, A study of alkali-aggregate reactivity, Proc. Amer. Soc. Civ. Eng. 66
[50] R.J. Thomas, D. Lezama, S. Peethamparan, On drying shrinkage in alkali-activated
(1940) 1781–1792.
concrete: improving dimensional stability by aging or heat-curing, Cem. Concr. Res.
[23] T. Chappex, K.L. Scrivener, The influence of aluminium on the dissolution of
91 (2017) 13–23.
amorphous silica and its relation to alkali-silica reaction, Cem. Concr. Res. 42
(2012) 1645–1649.

10