Construction and Building Materials 66 (2014) 275–285

Contents lists available at ScienceDirect

Construction and Building Materials

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

Effect of particle size on alkali–silica reaction in recycled glass mortars

Hongjian Du ⇑, Kiang Hwee Tan
Department of Civil and Environmental Engineering, National University of Singapore, Singapore 117576, Singapore

h i g h l i g h t s g r a p h i c a l a b s t r a c t

 ASR preferentially occurs in internal

cracks of glass particles.
 ASR expansion increases with more
inherent cracks in glass particles.
 ASR expansion decreases with
smaller particle size regardless of
glass color.
 The optimal contents for various ASR
mitigation methods are determined.
 Fly ash and slag show the highest
suppression efficiency on ASR
expansion.

a r t i c l e i n f o a b s t r a c t

Article history: The effect of particle size on alkali–silica reaction (ASR) was investigated to clarify the difference in long-
Received 11 December 2013 term ASR expansions in mortars using recycled green and brown glass as fine aggregates. Test results
Received in revised form 27 May 2014 revealed that green particles of 1.18 and 2.36 mm showed the highest reactivity while the other sizes
Accepted 28 May 2014
resulted in relatively low ASR expansions. Brown glass less than 2.36 mm did not result in large ASR
Available online 17 June 2014
expansion. Furthermore, different ASR suppressors were explored to suppress ASR expansion in mortar
using 1.18 mm green glass particles. These included supplementary cementitious materials, steel fiber
Keywords:
reinforcement and lithium compounds.
Aggregates
Chemical compounds
Ó 2014 Elsevier Ltd. All rights reserved.
Cracking
Durability
Fibers
Fly ash
Glass
Microstructure
Slag

1. Introduction
Waste glass has been explored as a substitute for sand in
making concrete in view of sustainability issues related to
production and use of concrete [1–3]. For certain type of natural
and manufactured materials, such as opal, shale, and silica glass,
however, alkali–silica reaction (ASR) has been widely reported to
occur [4–6]. The amorphous silica in glass is dissolved under alkali
attack to form ASR gel, which can imbibe water and subsequently
⇑ Corresponding author. Tel.: +65 81226102; fax: +65 67791635.
E-mail addresses: ceedh@nus.edu.sg (H. Du), tankh@nus.edu.sg (K.H. Tan).
expand, providing tensile stresses. Once the tensile stress exceeds
the tensile strength of the material, cracks will occur. As a result of
cracking, corrosive alkalis will be in contact with internal glass
structure more easily, accelerating ASR, widening cracks and
degrading concrete quality.
ASR is of concern in the re-utilization of waste glass in cemen-
titious mortar and concrete. This study focuses on soda-lime glass
since it is the most commonly used and disposed in urban environ-
ment. Some researchers have investigated the alkali reactivity of
glass particles with different types, colors, sizes and contents
[1,7–10]. It is commonly recognized that clear glass shows a higher
reactivity than green and brown glass [1,8,10]. Green glass has

http://dx.doi.org/10.1016/j.conbuildmat.2014.05.092
0950-0618/Ó 2014 Elsevier Ltd. All rights reserved.
276 H. Du, K.H. Tan / Construction and Building Materials 66 (2014) 275–285

been reported to perform better under alkali attack than brown Table 2
and clear glass due to the presence of chromium ions, as explained ASTM standard and modified mortar mixture proportion.

by double-layer theory [1,7,11]. However, there are still debates on Mix no. Water Cement Sand Coarse aggregate
the relative reactivity of glass particle with different colors. To C30 0.49 1 1.96 2.77
name a few, an opposite trend has been reported [12], with flint C45 0.38 1 1.33 2.15
(clear) glass producing the smallest expansion while green glass C60 0.32 1 0.99 1.81
producing the largest. It has also been found that mortar bars with ASTM C1260 0.47 1 2.25 –
Modified mix 0.47 1 1.6 –
brown glass sand expanded less than green glass sand mortar, par-
ticularly in the long term [13]. This was attributed to other factors
including the nature of the glass product manufacturing processes
[12], the availability of alkalis from the glass [9] and the existence Table 3
Grading requirement of sand as per ASTM C 1260.
of micro-cracks inside the glass particles [14].
Effective ASR mitigation methods have been well documented. Sieve size Mass (%)
They include cementitious material replacement, use of mineral Passing Retained on
and chemical admixtures, as well as additional reinforcing fibers
4.75 mm 2.36 mm 10
[7,8,15,16]. The effect of ASR suppressor also depends on the dos- 2.36 mm 1.18 mm 25
age quantity. However, studies reported in the literature are con- 1.18 mm 600 lm 25
cluded at a certain amount of ASR suppressor. Therefore, the 600 lm 300 lm 25
300 lm 150 lm 15
study was aimed at: (a) comparing alkali reactivity of green and
brown glass particles; (b) examining the effect of green and brown
glass particle size on ASR; and (c) investigating the optimal amount
for each mitigation method. 60 MPa and were mix-proportioned according to ACI 211 [19]. For each mortar mix-
ture, natural sand was replaced by glass sand by 0%, 25%, 50%, 75% and 100%. There-
fore, there were in total 15 mixes each for green and brown glass mortars. The mix
2. Materials and methods proportions are shown in Table 2. The size distribution of both natural and glass
sand followed the requirement of ASTM C 1260 [20] as shown in Table 3. The nat-
2.1. Materials ural sand used in this study exhibited an ASR expansion less than 0.2% at 28 days
[13], and is classified as non-reactive aggregate by ASTM C 1260.
Type I Ordinary Portland cement (OPC I) with chemical composition shown in The effect of glass particle size was investigated using 100% single-graded glass
Table 1 was used in this study. It had a sodium oxide equivalent alkali content of sand with two different mix ratios, that is, an ASTM standard mix and a slightly
0.6% (determined as Na2Oeq = %Na2O + 0.658  %K2O). Green and brown glasses, modified mix proportion (see Table 2). In this study, the single-graded glasses
mainly from beer bottles, were collected from a local recycler and processed in include 0.15 mm (that is, those passing/retained on 0.15/0.30 mm sieves),
the laboratory. The processing of recycled waste glass involved: (1) pre-soaking 0.30 mm (0.30/0.6 mm sieves), 0.6 mm (0.6/1.18 mm sieves), 1.18 mm (1.18/
and washing by tap water; (2) drying and separating by color; (3) crushing by a 2.26 mm sieves) and 2.36 mm (2.36/4.75 mm sieves). It is well known that mono-
jaw crusher; and (4) sieving according to ASTM standard sieve size [17]. With size particles would result in larger void content than well-graded particles of the
regard to the crusher, its bottom opening can be adjusted, depending on the target same shape and surface textures. Therefore, more cement paste will be required
size of glass sand. The chemical compositions of green and brown glass are shown to fill the increased voids. To produce good quality mortar, the sand/cement ratio
in Table 1. The glass sand was determined as fine aggregates as per ASTM C 128 was reduced from the standard value of 2.25 to the modified value of 1.6 in mod-
[18]. They had a specific gravity of 2.53 and water absorption of 0.07%, respectively. ified mix, while the water-to-cement (w/c) ratio was kept unchanged as 0.47. All
Compared to natural fine aggregates, the crushed glass sand exhibited sharper edge, mortar mixtures showed sufficient paste to fill aggregates voids and provide lubri-
more angular shape and smoother surface texture. cation layer for sufficient workability. For comparison purpose, the effect of particle
Among the ASR suppressors investigated, fly ash, ground granulated blast-fur- size for natural sand was also determined using the modified mix proportion.
nace slag (GGBS) and undensified silica fume, were used as supplementary cemen- The ASR mitigation methods were carried out for the modified mix proportion
titious materials (SCM) to substitute cement, and their chemical compositions are with 100% of 1.18 mm green glass sand for which ASR was found to be relatively
shown in Table 1. Smooth steel fibers, with an average diameter of 0.16 mm and prominent. Fly ash, GGBS, and silica fume was used to substitute cement from
length of 5 mm (that is, aspect ratio of 31), were added as fiber reinforcements. 10% to 50% (at 10% increments); 15% to 60% (at 15% increments), and 5.0% to
Lithium chloride and lithium carbonate were added as chemical admixtures first 12.5% (at 2.5% increments), respectively by mass. Steel fibers were added, by vol-
by dissolving the solid powder in the mixing water and then pouring it into the ume of the mortar mixture, at 0.5–2.0% (in steps of 0.5%). The content of lithium
dry mixture of cement and sand. It should be noted that the solid lithium carbonate chloride and lithium carbonate was 0.5–2.0% (in steps of 0.5%) by mass of cement.
cannot be completely dissolved due to its low solubility. Therefore, all the solid par-
ticles should be poured into the mixer, without residue attached on the container
wall. 2.3. Test methods

2.2. Mix proportions
To compare the alkali reactivity of green and brown glass sand, three mortar
mixtures were obtained from three different concrete mixtures by excluding the
coarse aggregates. The three concrete mixtures had design strength of 30, 45 and
Accelerated mortar-bar test (AMBT) was carried out as per ASTM C 1260 [20].
Three mortar bars, 25  25  285 mm, were prepared for each mix. The mixing
and molding procedure was in accordance with ASTM C 305 [21] and C 1260
[20], respectively. The mortar specimens were demolded 24 h after casting and
immersed in water at room temperature in a container, which was transferred into
an oven at a temperature of 80 °C for the next 24 h. The initial lengths were then

Table 1
Chemical compositions of materials.

Composition (%) Cement Fly ash GGBS Silica fume Natural sand Green glass Brown glass
SiO2 20.8 38.9 32.15 95.95 88.54 71.22 72.08
Al2O3 4.6 29.15 12.87 0.28 1.21 1.63 1.21
Fe2O3 2.8 19.64 0.36 0.32 0.76 0.32 0.76
CaO 65.4 2.5 40.67 0.16 5.33 10.79 10.45
MgO 1.3 2.1 6.05 0.37 0.42 1.57 0.72
SO3 2.2 0.19 4.95 0.18 – – –
Na2O 0.31 0.26 0.28 0.05 0.33 13.12 13.71
K2O 0.44 0.48 0.51 0.57 0.31 0.64 0.16
TiO2 – – – – 0.05 0.07 0.1
Cr2O3 – – – – – 0.22 0.01
 H. Du, K.H. Tan / Construction and Building Materials 66 (2014) 275–285 277

recorded before they were immersed in 1 N 80 °C NaOH solution. The expanded
lengths were subsequently measured every week. According to ASTM C 1260,
expansion higher than 0.2% at 14 days is considered potentially deleterious while
less than 0.1% is innocuous. However, for mortars containing pozzolans such as
fly ash, ASTM C 1567 [22], which specifies 0.1% as the threshold value for poten-
tially deleterious expansion, should be followed. For each mortar mix containing
different amount of ASR suppressors, three 50 mm cubes were prepared for com-
pressive strength following ASTM C 109 [23].
3. Test results and discussion
3.1. ASR expansion of mortar with green and brown glass as fine
aggregates
The ASR expansions of brown glass sand mortars sieved from
C30, C45 and C60 mixes are plotted in Fig. 1(a)–(c). All the mixes
showed ASR expansion less than 0.2% at 28 days, suggesting alkali
non-reactivity. The test results indicate that ASR expansion
consistently decreased with higher brown glass sand content at
each mix. At 100% replacement, the ASR expansion and develop-
ment rate were extremely low; at the end of test period of
140 days, the expansion was less than 0.05%, regardless of the mor-
tar mix. Also, no crack was observed on surface of specimens with
100% brown glass sand. This finding confirms the previous study by
the authors [13] that brown soda-lime glass sand might not exhibit
alkali reactivity as per ASTM C1260. It can be attributed to the poz-
zolanic reaction of fine glass sand (<0.6 mm), which has been con-
firmed by previous literature [24–26] and the calcium ions from
the soda-lime glass, which could lead to non-expansive ASR gel.
The ASR expansions of green glass sand mortar of C30, C45 and
C60 mixes are displayed in Fig. 1(d)–(f). The ASR behavior of green
glass sand mortar was quite different from brown glass sand mor-
tar, especially after 28 days. Within the first 21 days, the green
glass sand mortar showed similar trend as brown glass sand mor-
tar, that is, reduced expansion with increasing glass sand content

1.0 0.4
(a) C30-Brown (d) C30-Green
0.8
0 0.3 0
ASR expansion, %

ASR expansion, %
25% 25%
0.6 50% 50%
75% 0.2 75%
100% 100%
0.4

0.1
0.2

0.0 0.0
0 14 28 42 56 70 84 98 112 126 140 0 7 14 21 28 35 42 49
Age, days Age, days
1.0 0.8
(b) C45-Brown (e) C45-Green
0.8
0 0.6 0
ASR expansion, %

ASR expansion, %

25% 25%
0.6 50% 50%
75% 0.4 75%
100% 100%
0.4

0.2
0.2

0.0 0.0
0 14 28 42 56 70 84 98 112 126 140 0 7 14 21 28 35 42 49
Age, days Age, days

1.0 2.0
(c) C60-Brown (f) C60-Green
0.8 1.6
0 0
ASR expansion, %

ASR expansion, %

25% 25%
0.6 50% 1.2 50%
75% 75%
100% 100%
0.4 0.8

0.2 0.4

0.0 0.0
0 14 28 42 56 70 84 98 112 126 140 0 14 28 42 56 70 84 98 112 126 140
Age, days Age, days

Fig. 1. ASR expansion of mortar containing brown and green glass sand.
278 H. Du, K.H. Tan / Construction and Building Materials 66 (2014) 275–285

(see Fig. 2(a)). However, thereafter, the ASR expansion showed a
reverse trend, with higher ASR expansion occurring at higher glass
sand content. The ASR expansion rate of 75% green glass sand mor-
tar increased noticeably faster than 25% and 50% green glass sand
mortar. At 49 days, mortar with 75% green glass sand showed the
highest expansion, followed by 50% and 25%, for both C45 and
C60 (see Fig. 2(b)). Mortar with 100% green glass sand did not
expand faster before 35 days (Fig. 1(d)–(f)). Although at 49 days,
the 100% glass mortar of all mixes did not show the highest ASR
expansion, they exhibited the highest expansion rate and eventu-
ally exhibited the largest ASR expansion, as shown in the case of
C60 (Fig. 1(f)).
Although green glass sand exhibit comparable alkali reactivity
as brown glass sand within 28 days as previously reported [3,31],
green glass sand appears to be more ASR reactive in the long term.
Idir et al. also found that green glass sand showed much higher ASR
expansion than other colored glasses [10]. The reaction can be as
follows. ASR expansion of mortar is caused by the internal cracks
in large glass particles. ASR gel in these cracks is more expansive
due to the high content of sodium. The expansive gel will cause
tensile stress to the glass particle and its surrounding cement
paste, rendering higher expansion. However, it takes time to exhi-
bit the noticeable expansion, beyond 28 days in general.
Regarding the effect of w/c ratio on ASR, it is inferred from Fig. 3
that C45 mix (w/c = 0.38) exhibited the highest expansion and C30
mix (w/c = 0.49) showed the lowest for both brown and green glass
sand regardless of glass sand content. The C60 mix (w/c = 0.32)
showed intermediate values. The pessimum effect with the highest
ASR expansion at w/c = 0.38 was due to two opposing effects of
increasing w/c ratio on ASR; first, there is high susceptibility to
tensile cracking and increased expansion due to reduced tensile
strength; and second, there is decreased expansion because of a
more porous microstructure, relieving the pressure due to gel
expansion [1].
3.2. Effect of glass particle size on ASR expansion
The effect of green and brown glass particle size on ASR
expansion is shown in Fig. 4. For green glass sand mortar, the
ASR expansion increased with the glass sand size, and peaked at
a particle size of 1.18 mm (Fig. 4(a) and (b)), regardless of mix
ratio. With a larger size of 2.36 mm, the ASR expansion reduced
slightly.
Similar results have been reported by some researchers [27].
Smaller glass sand should have larger surface area, which allows
ASR to readily occur, causing more expansion. However, with a size
smaller than the critical value, pozzolanic reaction would occur,
which may form non-swelling CSH gel instead. Some models have
also been proposed to explain this pessimum effect [27–29]. How-
ever, the difference in the particle size effect for green and brown
glass remains unsolved. Recently, a different explanation has been
proposed for the pessimum effect of glass sand size on ASR expan-
sion [14,30]. That is, swelling ASR gel would occur inside inherent
micro-cracks of glass particles, rather than at surface. Therefore,
inherent micro-cracks in larger glass particles would render more
reaction and higher expansion.
It is interesting to note that such pessimum was not observed
for brown glass sand; as particles of all sizes except 2.36 mm

2.0 3.0
(a) C30-Brown
C30-Green
C45-Brown
C45-Green
C60-Brown
C60-Green
(b) C30-Brown
C30-Green
C45-Brown
C45-Green
C60-Brown
C60-Green
2.5
1.5
Relative expansion

Relative expansion

2.0

1.0 1.5

1.0
0.5
0.5

0.0 0.0
0 25 50 75 100 0 25 50 75 100
Glass sand content, % Glass sand content, %

Fig. 2. Relative expansion of mortar with glass sand compared to the reference mortar at (a) 28 days and (b) 49 days.

0.4 0.4
(a) Brown (b) Green
0 0
0.3 25% 0.3 25%
ASR expansion, %

50% 50%
ASR expansion, %

75% 75%
100% 100%
0.2 0.2

0.1 0.1

0.0 0.0
0.30 0.35 0.40 0.45 0.50 0.30 0.35 0.40 0.45 0.50
w/c ratio w/c ratio

Fig. 3. Effect of w/c ratio on glass sand mortar ASR expansion at 28 days.
 H. Du, K.H. Tan / Construction and Building Materials 66 (2014) 275–285 279

2.0 2.0
(a) ASTM standard mix-Green (b) Modified mix-Green
7-day 7-day
14-day 14-day
1.5 21-day 1.5 21-day
ASR expansion, %

ASR expansion, %
28-day 28-day
35-day 35-day
42-day 42-day
1.0 49-day 1.0 49-day
56-day 56-day

0.5 0.5

0.0 0.0
0.1 0.3 0.6 1 4 0.1 0.3 0.6 1 4
Glass sand size, mm Glass sand size, mm

0.5 0.5
(c) ASTM standard mix-Brown (d) Modified mix-Brown
7-day 7-day
0.4 14-day 0.4 14-day
21-day 21-day
ASR expansion, %

ASR expansion, %
28-day 28-day
35-day 35-day
0.3 0.3
42-day 42-day
49-day 49-day
56-day 56-day
0.2 0.2

0.1 0.1

0.0 0.0
0.1 0.3 0.6 1 4 0.1 0.3 0.6 1 4
Glass sand size, mm Glass sand size, mm

0.5
(e) Modified mix-Natural sand
7-day
0.4 14-day
21-day
ASR expansion, %

28-day
35-day
0.3
42-day
49-day
56-day
0.2

0.1

0.0
0.1 0.3 0.6 1 4
Sand size, mm

Fig. 4. Effect of particle size on ASR expansion.

showed low expansions after 49 days regardless of mix ratio
(Fig. 4(c) and (d)). The result provides a plausible explanation on
the different ASR expansion of green and brown glass mortar in
the long term. The extremely high alkali reactivity of 1.18 and
2.36 mm green glass particles might cause increased expansion
and rapid expansion rate. Thus, the following section will discuss
the various ASR mitigation methods based on the modified mix
with 1.18 mm green glass particles as fine aggregates in mortar.
For natural sand, ASR expansion was largest with 0.3 and
0.6 mm size particle (Fig. 4(e)). The results confirmed that tradi-
tional ASR is a surface reaction, increasing with larger surface area.
When a large amount of reactive material is present in a finely
divided form (i.e. under 75 lm), there may be considerable petro-
graphic evidence of ASR without significant expansion [6], giving
weight to the observations of low ASR expansion in glass particles
with smaller size.
Fig. 5 displays the microstructure of mortar with 1.18-mm green
glass sand with some representative cracks after ASR test. It is obvi-
ously noted that there are cracks inside of the glass particles
(Fig. 5(a) and (b)) and in the matrix (Fig. 5(b)). Fig. 5(c) shows that
the crack in the glass sand which extend into cement paste being
filled with ASR gel. From Fig. 5(d), the crack in the cement paste
was around 15 lm wide. It is postulated that the alkalis, such as
Na+ and K+ penetrate into the glass, particularly for glass particles
with inherent micro-cracks, and form ASR gel inside the cracks. This
ASR gel would expand after absorbing water. This expansion results
in tensile stress inside glass particles. Once the tensile stress
exceeds the tensile strength of glass, cracks would propagate in
the glass particles (Fig. 5(e) and (f)). Subsequently, such cracks
may extend to cement paste matrix if the tensile stress is higher
than the tensile strength of cement paste (Fig. 5(b)). Therefore,
the ASR reactivity of glass would depend on the existence of cracks,
280 H. Du, K.H. Tan / Construction and Building Materials 66 (2014) 275–285

(a) (b) Cement (c)

paste

Glass
Paste
crack
Glass

Internal Cement
crack paste
ASR gel

(d) (e) (f)

To-crack
glass

Cracked
glass

(g) (h) (i)

(j) (k) (l)

Steel fiber

ITZ

Cement
paste

Fig. 5. SEM pictures. (a) Green glass sand mortar with internal cracks inside glass particles; (b) cracks in green glass particle and cement paste matrix; (c) cracks extending
from inside of green glass particle to cement paste; (d) ASR gel in the crack of cement paste; (e) cracked green glass particle; (f) about-to-crack green glass particle; (g) green
glass particle after crushing; (h) micro-cracks inside green glass particle; (f) brown glass particle after crushing; (i) brown glass sand mortar without internal cracks; (k) ITZ
between steel fiber and cement paste; and (l) silica fume agglomerates before casting.

and this also explains the results of Fig. 1(d)–(f). In this study, there
were some micro-cracks in glass particles after crushing, especially
for those with size of 1.18-mm, as shown in Fig. 5(g) and (h). How-
ever, no crack was found in brown glass particles and in mortar con-
taining brown glass, as displayed in Fig. 5(i) and (j) respectively.
This might explain the different alkali reactivity of green and brown
glass particles, as observed in this study.
Internal micro-cracks are likely to occur in larger glass particles
due to the crushing process. Even if there are cracks in smaller par-
ticles, they should be much narrower and difficult for ions to pen-
etrate into. The absence of large internal cracks in brown glass
particles might be due to the different manufacturing processes
that can affect the internal stress level of glass products, as men-
tioned by other researchers [12,31].
The chemical composition of the formed ASR products inside
the micro-cracks of green glass particles was determined by using
energy-dispersive X-ray spectroscopy (EDS). The location and
microstructure for each EDS testing points are displayed in Fig. 6
and the analysis results are shown in Table 4. The ASR gel inside
the micro-crack has higher sodium content and CaO/(SiO2 + Na2O)
ratio, which could absorb water and expand, resulting in tensile
stresses in the glass particle [32,33]. It is also noted that the
sodium content increases and CaO/(SiO2 + Na2O) ratio decrease
with deeper distance from the crack mouth (points 1–4 in Table 4).
This is because the dissolved sodium and silica are not able to
escape from the crack and remain in the dissolved places to form
ASR gel products. However, the sodium at the glass particle surface
can more easily diffuse into the surrounding cement paste, leading
to a product with low sodium content and low potential expansion.
This chemical composition can support the ASR mechanism
observed from the microstructure, as discussed above.
In brown glass mortar, no cracks were observed. It is therefore
believed that the ratio of CaO/(SiO2 + Na2O) is also higher, indicat-
ing a lower level of ASR.
3.3. Optimal content of ASR mitigation methods
The ASR expansion results of the ASR mitigation methods inves-
tigated are shown in Figs. 7, 9 and 10. The efficiency of each miti-
gation method is discussed herein.
3.3.1. Supplementary cementitious materials
As expected, the replacement of cement by SCM, that is, fly ash,
GGBS and silica fume in this study, could reduce ASR, because of
 H. Du, K.H. Tan / Construction and Building Materials 66 (2014) 275–285 281

2.0
(a) Fly ash
1.6 7-day
14-day
1.2 21-day
28-day
0.8

ASR expansion, %
35-day
0.4 42-day
0.2 49-day
56-day
0.04

0.02

0.00

-0.02
0 10% 20% 30% 40% 50%
Content, %

2.0
(b) GGBS 7-day
Fig. 6. SEM picture for C45-75 green glass mortar after mortar-bar test.
14-day
21-day
1.5 28-day

ASR expansion, %
Table 4 35-day
Chemical analysis of ASR gel products in C45-75 green glass mortar by EDS (by 42-day
49-day
weight).
1.0 56-day
Location CaO Na2O SiO2 CaO/(SiO2 + Na2O) Description
1 13.3 15.5 66.4 0.16 Gel at internal crack
2 16.1 13.0 66.5 0.20 Gel at internal crack
0.5
3 25.6 8.8 60.3 0.37 Gel at internal crack
4 39.9 6.1 49.8 0.71 Gel at crack mouth
5 43.0 5.7 46.5 0.82 Glass-paste interface
6 52.0 5.8 36.7 1.22 Cement paste 0.0
7 13.9 8.1 76.5 0.16 Glass 0 15% 30% 45% 60%
Content, %

2.0
their pozzolanic reactivity. A number of studies [12,34–39] have (c) Silica fume 7-day
14-day
established the ASR suppressing mechanism of pozzolanic reac-
21-day
tion, which results in: (a) enhanced impermeability of cement 1.5 28-day
ASR expansion, %

paste and hence reduced mobility of ions; (b) reduced alkalinity 35-day
42-day
of the pore solutions as a result of depletion of cement; (c) reduced 49-day
amount of Ca(OH)2 which formed into secondary CSH; and (d) 1.0 56-day
increased strength of cement paste provided by mineral admix-
tures, leading to higher resistance to the expansive stress as devel-
oped by ASR.
0.5
In the case of fly ash, the ASR restraining effect was clearly sig-
nificant, both in terms of absolute expansion and expansion rate
(Fig. 7(a)). All mortar specimens with fly ash showed significantly
0.0
low expansion values, less than 0.02% at all the ages, up to 56 days.
0 5% 7.5% 10% 12.5%
This effect was prominent even with 10% of fly ash, and higher fly
Content, %
ash content did not change the results. It is also noted that some
early-age values were negative, indicative of autogenous shrinkage Fig. 7. Mitigation of ASR expansion of mortar using supplementary cementitious
instead of expansion. Although ASR expansion generally increased materials.
with time, a decrease in expansion was also observed from 14 to
21 days and from 49 to 56 days. Some researchers have reported
the self-healing of micro-cracks in the cement system due to fly than 0.2% with 45% GGBS, while with 60% GGBS it was only
ash [40], and it has been specifically reported that cracks with 0.02%. Based on the test results, GGBS was effective in reducing
widths below 0.1 mm can be closed by a self-healing process ASR expansion if it replaced at least 30% of cement content.
[41,42]. The self-healing capacity due to fly ash may originate from With silica fume, ASR expansion was effectively suppressed
its pozzolanic characteristics, which can occur at a later age, thus when it replaced 12.5% of cement. The expansion was higher than
filling the pores and micro-cracks. As a result, the resistance of 0.2% at 14 days, with 5%, 7.5% and 10% silica fume, although they
paste to alkali diffusion from external NaOH solution would all exhibited less expansion than the reference mortar specimen.
increase. Compared with the first 14 days, the ASR expansion increased very
With cement partially replaced by GGBS, the ASR expansion of slowly thereafter. Some cracks showed up on the surface of mortar
glass mortar decreased with higher content of GGBS (Fig. 7(b)). bars with silica fume at low silica fume contents. There were only a
Up to 28 days, mixes with 30% or more GGBS exhibited less than few major cracks which widened with time, while the remaining
0.2% ASR expansion. Also, at 56 days, the ASR expansion was less surface area remained uncracked, as shown in Fig. 8.
282 H. Du, K.H. Tan / Construction and Building Materials 66 (2014) 275–285

2.0
7-day
14-day
21-day
1.5 28-day

ASR expansion, %
35-day
42-day
49-day
1.0 56-day

Fig. 8. Appearances of mortar bars with varying silica fume contents after 56 days 0.5
of testing (5–12.5% from front to back).

In addition, mortar with 10% silica fume showed the highest 0.0
0 0.5 % 1.0 % 1.5% 2.0%
expansion, indicating a pessimum effect. A pessimum replacement
for densified silica fume in suppressing ASR expansion has been Content, %
reported as 15% [43]. According to Diamond [44], a pessimum
Fig. 9. Mitigation of ASR expansion of mortar using steel fibers.
occurred for silica fume at 4% replacement when dealing with
the ASR of quartzite aggregate, while 12% replacement would pro-
duce innocuous behavior. This critical content agrees well with the 3.3.3. Lithium compounds
result in this study. However, no plausible reason has been pro- The effectiveness of lithium compounds were first extensively
vided to explain this pessimum study. investigated by McCoy and Caldwell [50] and confirmed as effec-
The substitution of cement by silica fume beyond 5% can signif- tive in ASR mitigation [51,52]. However, until now, the underlying
icantly reduce the porosity of cement paste and lead to pozzolanic mechanisms are not well understood and different explanations
reaction, especially at the ITZ, because of its ultra-fine particle size have been provided. It was reported by Collins et al. [53] and Feng
(around 0.1 lm compared with 10 lm in the case of cement parti- et al. [54] that lithium ions easily bound in the ASR products due to
cle) [45,46]. Compared with the reduction in permeability which the small ionic radius and high surface charge density, compared to
reduces the mobility of alkali ions, the alkalis depletion due to poz- the Na+ and K+ ions in concrete. Consequently, the formed ASR gel
zolanic reaction might play a secondary role in restraining the ASR would be changed in chemical composition. Also, lithium com-
expansion [45,46]. The restrained ASR gel increase the volume and pounds might reduce the solubility of the reactive silica and pre-
stress in the vicinity of reaction site and cause crack to occur once
the expansive stress exceeds the tensile strength of cement paste,
as shown in Fig. 8. However, if cement is substituted by 12.5% silica 2.0
fume, the pozzolanic reaction would deplete the alkalis and hence (a) LiCl
1.6
7-day
ASR is less. It should be noted that silica fume in concrete does not
1.2 14-day
always prevent ASR distress [44]; sometimes it can induce ASR dis- 21-day
tress particularly as oversized or un-dispersed grains could react to
ASR expansion, %

0.8 28-day
35-day
generate expansive ASR gel when the alkali hydroxide concentra-
0.4 42-day
tion is high enough. 0.2 49-day
56-day
0.04
3.3.2. Fiber reinforcement 0.03
The use of fibers to control ASR is traditionally considered as a 0.02
mechanical confinement, which could contribute to higher
0.01
strength and resistance to crack growth [4,16,47,48]. According
to a chemo-mechanical confinement model proposed by Ostertag 0.00
0 0.5% 1.0% 1.5% 2.0%
et al. [49], the resistance in crack propagation and crack opening
displacement by steel fibers not only imposes compressive stress Content, %
on the expanding ASR gel but also prevents the ASR gel from leav-
2.0
ing the reaction site.
(b) Li 2CO3
The ASR expansion of mortar bars with varying amount of steel 1.6
7-day
fibers is shown in Fig. 9. The overall ASR restraining efficiency of 14-day
1.2
steel fiber was less than SCM. Also, this restraining efficiency gen- 21-day
ASR expansion, %

0.8 28-day
erally became more distinct with higher fiber content. Adding 35-day
more than 1% of steel fibers could effectively reduce the expansion 0.4 42-day
to below 0.1% at 14 days. However, all mortars with steel fibers had 0.2 49-day
56-day
expansion exceeding 0.2% at 28 days and 0.7% at 56 days. At 0.04
28 days, the ASR expansion of mortar with a steel fiber content
0.03
of 1% was higher than that with 0.5%. Turanli et al. [47] have sim-
ilarly reported that adding 1% of steel microfiber would result in 0.02
increased the ASR expansion, but no reason was provided. The pos- 0.01
sible reason could be the relatively weak bond at fiber–matrix 0.00
interface (as shown in Fig. 5(k)), through which the alkali ions 0 0.5% 1.0% 1.5% 2.0%
could penetrate to the glass particles more easily. With higher steel Content, %
fiber content, however, the crack growth would be more effectively
restrained. Fig. 10. Mitigation of ASR expansion of mortar using lithium compounds.
 H. Du, K.H. Tan / Construction and Building Materials 66 (2014) 275–285 283

vent dissolved silica to form ASR gel particles [55]. At the same 0.10
time, it was reported that Li+ ions added could form a protective (a) Green C45-Fly Ash
layer (non-expansive lithium-bearing silicate) on the surface of
0.08
the reactive aggregate particle [56–58]. 0

ASR expansion, %
The test results of LiCl and Li2CO3 as ASR suppressors are shown 25%
in Fig. 10. Both lithium salts can reduce the ASR expansion remark- 0.06 50%
ably, to far below 0.1%. The difference between 0.5% and 2% of lith- 75%
ium compound was quite small. The efficiency of LiCl and Li2CO3 0.04 100%
was comparable. Some literature reported the minimum value of
molar ratio of [Li]/[Na] [50] to effectively reduce ASR expansion,
such as 0.9 for LiCl, which was not observed in this study. The test 0.02
results suggest that using LiCl and Li2CO3 with an amount of 0.5%
by mass of cement should be enough to control ASR. 0.00
0 7 14 21 28 35 42 49
3.3.4. Comparison of different ASR mitigation methods Age, days
From Figs. 7, 9 and 10, the optimum content for each method is
determined as follows: 10–50% for fly ash; 45–60% for GGBS; 12.5% 0.10
for silica fume; 1.5–2.0% for steel fiber; and 0.5–2.0% for lithium (b) Green C45-GGBS
chloride or lithium carbonate. Moreover, the effectiveness of differ- 0.08
ent methods, at their optimum contents, generally decreases in the 0

ASR expansion, %
sequence of: fly ash, lithium compounds, GGBS, silica fume and 25%
steel fiber. 0.06 50%
The type and amount of ASR suppressor to be used should be 75%
selected according to other requirements, such as compressive 0.04 100%
strength. The compressive strengths of mortar with different
amounts of ASR suppressor at 28 days are shown in Fig. 11. More
than 20% strength will be lost if fly ash content exceeds 40%. 0.02
Although this strength loss can be offset by pozzolanic reaction
at a later age, caution should be exercised if early strength of con- 0.00
crete is of main concern. In contrast, even at 60% replacement ratio 0 7 14 21 28 35 42 49
of cement by GGBS, compressive strength would not be affected. Age, days
This is due to the high CaO content of 40%, which could be con-
verted to Ca(OH)2 and result in pozzolanic reaction at an early Fig. 12. ASR expansions of green glass mortar C45 with (a) 30% fly ash and (b) 60%
GGBS.
age. This is also the reason for the use of high GGBS content as
cement replacement.
However, silica fume appeared to actually reduce the compres- agglomerates might create relatively weak zones which resulted
sive strength rather than increase it. The effect of silica fume in in stress concentrations and strength reduction.
pore structure refinement may be more obvious with w/c ratio The deformability (or strain at peak stress) of mortar with steel
lower than 0.47, which is used in this study. However, the primary fiber under compression was significantly increased, although the
reason may be the un-dispersed grains of silica fume (see Fig. 5(l)). compressive strength was actually reduced. The benefit of fiber
It should be mentioned that in this study no superplasticizer was reinforcement for cement-based composites is in the toughness,
used for the dispersion of silica fume to avoid introducing factors ductility, impact resistance, fatigue endurance and impermeability,
affecting the ASR expansion. All mortars containing silica fume rather than on compressive strength.
exhibited sufficient flowability during casting. However, the With regard to the addition of lithium compounds, particularly
agglomerates of silica fume may merely be dispersed partially dur- Li2CO3, the compressive strength of mortar was also reduced. It is
ing the mixing process while a considerable portion of those postulated that the carbonate retards the hydration of Portland
agglomerates was kept intact [59–61]. Thus, those unreacted cement and thus delays the ASR. Since lithium treatment is

60
Fly ash GGBS Silica fume Steel fiber LiCl Li2CO3
Compressive strength, MPa

50

40

30

20

10

0
0 10 20 30 40 50 15 30 45 60 5 7.5 10 12.5 0.5 1 1.5 2 0.5 1 1.5 2 0.5 1 1.52

Amount of suppressor, %

Fig. 11. Compressive strength of mortar with different amount of ASR suppressors.
284 H. Du, K.H. Tan / Construction and Building Materials 66 (2014) 275–285
expensive, it would presumably be applied only to concretes at
high risk of ASR [43,47].
3.3.5. Practical use of fly ash and GGBS as ASR suppressor
For practical ASR mitigation, fly ash and GGBS are recom-
mended, especially if sustainability and economy concerns are
taken into account. It is well known that fly ash and slag, as indus-
try by-products, are disposed in landfills in massive quantities over
the world since they are not yet re-utilized in sufficiently large
amount. If they can be used as cement replacement, not only neg-
ative environmental impacts are reduced but also better concrete
durability (i.e. the reduced ASR expansion for glass mortar in this
study) can be achieved. With respect to only ASR restraining, steel
fiber, silica fume and lithium compound may be less competitive
due to their higher prices, since concrete is a cost-sensitive
industry.
As a final stage of the study, 30% fly ash and 60% GGBS were
selected as ASR reducing methods to examine the expansion of
C45 green glass sand mortar (with sand grading following ASTM
C 1260). The results are shown in Fig. 12. Compared with the ref-
erence mortar with green glass sand, all mortars with 30% fly ash
or 60% GGBS exhibited substantially smaller expansions, less than
0.1% even at 49 days, regardless of glass sand content. With the use
of fly ash or GGBS, the alkali reactivity of green glass sand was
found to diminish since mortar with higher content of green glass
sand showed less expansion. In addition, the suppressing effect of
fly ash and GGBS was more pronounced at later age.
4. Conclusions
From the experimental investigations in ASR expansions of
green and brown glass sand mortars, the following conclusions
can be drawn:
(1) Brown glass sand showed better ASR resistance than green
glass sand in the long term. The reason is attributed to the
extreme alkali reactivity of 1.18 mm green glass particles,
which have internal cracks where ASR preferentially occurs.
(2) Higher strength (or lower w/c ratio) does not necessarily
mean better durability of concrete with respect to ASR,
due to the reduced porosity which caused slower release
of formed expansive ASR gel.
(3) Not all the mitigation methods showed increasing efficiency
with higher amount. The optimal content for each mitigation
method was determined as 10–50% for fly ash, 45–60% for
GGBS, 12.5% for silica fume, 1.5–2.0% for steel fibers and
0.5–2.0% for lithium compounds.
(4) Fly ash and GGBS could be used in practice due to their
remarkable suppression on ASR expansion.
References
[1] Jin W, Meyer C, Baxter S. ‘‘Glascrete’’ – concrete with glass aggregate. ACI
Mater J 2000;97:208–13.
[2] Lee G, Ling TC, Wong YL, Poon CS. Effects of crushed glass cullet sizes, casting
methods and pozzolanic materials on ASR of concrete blocks. Constr Build
Mater 2011;25:2611–8.
[3] Tan KH, Du H. Use of waste glass as sand in mortar: Part I – fresh, mechanical
and durability properties. Cem Concr Compos 2013;35:109–17.
[4] Hobbs W. Alkali–silica reaction in concrete. London: Telford; 1988.
[5] Polley C, Cramer SM, de la Cruz RV. Potential for using waste glass in Portland
cement concrete. J Mater Civ Eng 1998;10:210–9.
[6] Mehta PK, Monteiro PJM. Concrete: microstructure, properties, and materials.
3rd ed. New York: McGraw-Hill; 2006.
[7] Park S, Lee B. Studies on expansion properties in mortar containing with waste
glass and fibers. Cem Concr Res 2004;34:1145–52.
[8] Topcu IB, Boga AR, Bilir T. Alkali–silica reactions of mortars produced by using
waste glass as fine aggregate and admixtures such as fly ash and Li2CO3. Waste
Manage 2008;28:878–84.
[9] Saccani A, Bignozzi MC. ASR expansion behavior of recycled glass fine
aggregates in concrete. Cem Concr Res 2010;40:531–6.
[10] Idir R, Cyr M, Tagnit-Hamou A. Use of fine glass as ASR inhibitor in glass
aggregate mortars. Constr Build Mater 2010;24:1309–12.
[11] Zhu H, Chen W, Zhou W, Byars EA. Expansion behavior of glass aggregates in
different testing for alkali–silica reactivity. Mater Struct 2009;42:485–94.
[12] Dhir RK, Dyer TD, Tang MC. Alkali–silica reaction in concrete containing glass.
Mater Struct 2009;42:1451–62.
[13] Du H, Tan KH. Use of waste glass as sand in mortar: Part II – alkali–silica
reaction and mitigation methods. Cem Concr Compos 2013;35:118–26.
[14] Rajabipour F, Maraghechi H, Fischer G. Investigating the alkali silica reaction of
recycled glass aggregates in concrete materials. J Mater Civ Eng
2010;22:1201–8.
[15] Taha B, Nounu G. Utilizing waste recycled glass as sand/cement replacement
in concrete. J Mater Civ Eng 2009;21:709–21.
[16] Meyer C, Baxter S. Use of recycled glass for concrete masonry blocks. Final
report for the New York state energy research and development authority 97-
15; 1997.
[17] ASTM E 11. Standard specification for wire cloth and sieves for testing
purposes. Philadelphia: American Society for Testing and Materials; 2009.
[18] ASTM C 128. Standard test method for density, relative density (specific
gravity), and absorption of fine aggregate. Philadelphia: American Society for
Testing and Materials; 2007.
[19] Du H, Tan KH. Concrete with recycled glass as fine aggregates. ACI Mater J
2014;111:47–58.
[20] ASTM C 1260. Standard test method for potential alkali reactivity of aggregates
(mortar-bar method). Philadelphia: American Society for Testing and
Materials; 2007.
[21] ASTM C 305. Standard practice for mechanical mixing of hydraulic cement
pastes and mortars of plasticity consistency. Philadelphia: American Society
for Testing and Materials; 2006.
[22] ASTM C 1567. Standard test method for determining the potential alkali–silica
reactivity of combinations of cementitious materials and aggregate
(accelerated mortar-bar method). Philadelphia: American Society for Testing
and Materials; 2008.
[23] ASTM C 109. Standard test method for compressive strength of hydraulic
cement mortars using 2-in. or [50-mm] cube
specimens. Philadelphia: American Society for Testing and Materials; 2005.
[24] Dyer TD, Dhir RK. Chemical reactions of glass cullet used as cement
component. J Mater Civ Eng 2001;13:412–7.
[25] Xie Z, Xi Y. Use of recycled glass as a raw material in the manufacture of
Portland cement. Mater Struct 2002;35:510–5.
[26] Shi C, Wu Y, Riefler C, Wang H. Characteristics and pozzolanic reactivity of
glass powders. Cem Concr Res 2005;35:987–93.
[27] Idir R, Cyr M, Tagnit-Hamou A. Pozzolanic properties of fine and coarse color-
mixed glass cullet. Cem Concr Compos 2011;33:19–29.
[28] Bazant ZP, Steffens A. Mathematical model for kinetics of alkali–silica reaction
in concrete. Cem Concr Res 2000;30:419–28.
[29] Bazant ZP, Zi G, Meyer C. Fracture mechanics of ASR in concretes with waste
glass particles of different sizes. J Eng Mech 2000;126:226–32.
[30] Maraghechi H, Shafaatian SMH, Fischer G, Rajabipour F. The role of residual
cracks on alkali silica reactivity of recycled glass aggregates. Cem Concr
Compos 2012;34:41–7.
[31] Ballauff M, Brader JM, Egelhaaf SU, Fuchs M, Horbach J, Koumakis N, et al.
Residual stresses in glasses. Phys Rev Lett 2013;110:215701.
[32] Helmuth R, Stark D. Alkali–silica reactivity mechanism. In: Skalny J, editor.
Materials science of concrete. Ohio: American Ceramics Society; 1992. p.
131–208.
[33] Shi C. Corrosion of glasses and expansion mechanism of concrete waste
containing glasses as aggregates. J Mater Civ Eng 2009;21:529–34.
[34] Diamond S. Effect of two Danish flyashes on alkali contents of pore solutions of
cement–fly ash pastes. Cem Concr Res 1981;11:383–94.
[35] Helmuth R. Alkali–silica reactivity: an overview of research. SHRP-C-
342. Washington: National Research Council; 1993.
[36] Xu GJZ, Watt DF, Hudec PP. Effectiveness of mineral admixtures in reducing
ASR expansion. Cem Concr Res 1995;25:1225–36.
[37] Monteiro PJM, Wang K, Sposito G, dos Santos MC, de Andrade WP. Influence of
mineral admixtures on the alkali–aggregate reaction. Cem Concr Res
1997;27:1899–909.
[38] Hasparyk NP, Monteiro PJM, Carasek H. Effect of silica fume and rice husk ash
on alkali–silica reaction. ACI Mater J 2000;97:486–92.
[39] Turanli L, Bektas F, Monteiro PJM. Use of ground clay brick as a pozzolanic
material to reduce the alkali–silica reaction. Cem Concr Res 2003;33:1539–42.
[40] Termkhajornkit P, Nawa T, Yamashiro Y, Saito T. Self-healing ability of fly ash–
cement systems. Cem Concr Compos 2009;31:195–203.
[41] Reinhardt HW, Jooss M. Permeability and self-healing of cracked concrete as a
function of temperature and crack width. Cem Concr Res 2003;33:981–5.
[42] Sahmaran M, Li VC, Andrade C. Corrosion resistance performance of steel-
reinforced engineered cementitious composite beams. ACI Mater J
2008;105:243–50.
[43] Chen H, Soles JA, Malhotra VM. Investigations of supplementary cementing
materials for reducing alkali–aggregate reactions. Cem Concr Compos
1993;15:75–84.
 H. Du, K.H. Tan / Construction and Building Materials 66 (2014) 275–285
[44] Diamond S. Alkali silica reactions – some paradoxes. Cem Concr Compos
1997;19:391–401.
[45] Berube MA, Duchesne J. Does silica fume merely postpone expansion due to
alkali–aggregate reactivity? Constr Build Mater 1993;7:137–43.
[46] Boddy AM, Hooton RD, Thomas MDA. The effect of the silica content of silica
fume on its ability to control alkali–silica reaction. Cem Concr Res
2003;33:1263–8.
[47] Turanli L, Shomglin K, Ostertag CP, Monteiro PJM. Reduction in alkali–silica
expansion due to steel microfibers. Cem Concr Res 2001;31:825–7.
[48] Yi CK, Ostertag CP. Mechanical approach in mitigating alkali–silica reaction.
Cem Concr Res 2005;35:67–75.
[49] Ostertag CP, Yi CK, Monteiro PJM. Effect of confinement on properties and
characteristics of alkali–silica reaction gel. ACI Mater J 2007;104:276–82.
[50] McCoy WJ, Caldwell AG. New approach to inhibiting alkali–aggregate
expansion. ACI J Proc 1995;47:693–706.
[51] Stark DC. Lithium salt admixtures – an alternative method to prevent
expansive alkali–silica reactivity. In: Proc of the 9th int conf on alkali–
aggregate reaction in concrete, London; 1992. p. 1017–25.
[52] Hudec PP, Banahene NK. Chemical treatments and additives for controlling
alkali reactivity. Cem Concr Compos 1993;15:21–6.
285
[53] Collins CI, Ideker JH, Willis GS, Kurtis KE. Examination of the effects of LiOH,
LiCl, and LiNO3 on alkali–silica reaction. Cem Concr Res 2004;34:1403–15.
[54] Feng X, Thomas MDA, Bremner TW, Balcom BJ, Folliard KJ. Studies on lithium
salts to mitigate ASR-induced expansion in new concrete: a critical review.
Cem Concr Res 2005;35:1789–96.
[55] Mitchell LD, Beaudoin JJ, Grattan-Bellew P. The effects of lithium hydroxide
solution on alkali silica reaction gels created with opal. Cem Concr Res
2004;34:641–9.
[56] Prezzi M, Monteiro PJM, Sposito G. Alkali–silica reaction, Part 1: use for the
double-layer theory to explain the behavior of reaction-product gels. ACI
Mater J 1997;94:10–7.
[57] Prezzi M, Monteiro PJM, Sposito G. Alkali–silica reaction, Part 2: the effect of
chemical admixtures. ACI Mater J 1998;95:3–10.
[58] Stark D, Morgan B, Okamoto P. Eliminating or minimizing alkali–silica
reactivity. Washington: National Research Council; 1993. 75–93.
[59] Juenger MCG, Ostertag CP. Alkali–silica reactivity of large silica fume-derived
particles. Cem Concr Res 2004;34:1389–402.
[60] Maas AJ, Ideker JH, Juenger MCG. Alkali silica reactivity of agglomerated silica
fume. Cem Concr Res 2007;37:166–74.
[61] Diamond S, Sahu S, Thaulow N. Reaction products of densified silica fume
agglomerates in concrete. Cem Concr Res 2004;34:1625–32.