Materials and Design 32 (2011) 2675–2684

Contents lists available at ScienceDirect

Materials and Design

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

Properties of architectural mortar prepared with recycled glasswith different particle sizes

Tung-Chai Ling 1, Chi-Sun Poon ⇑Faculty of Construction and Land Use, The Hong Kong Polytechnic University, Hong Kong

a r t i c l e i n f o

Article history:Received 28 August 2010Accepted 11 January 2011Available online 15 January 2011

Keywords:ConcreteGlassesReutilization

0261-3069/$ - see front matter � 2011 Elsevier Ltd Adoi:10.1016/j.matdes.2011.01.011

⇑ Corresponding author. Tel.: +852 2766 6024.E-mail addresses: cetcling@inet.polyu.edu.hk, tclin

cecspoon@polyu.edu.hk (C.-S. Poon).1 Tel.: +852 6697 5982.

a b s t r a c t

The recycling of glass waste as a source of aggregate for the production of concrete products has attractedincreasing interest from the construction industry. However, the use of recycled glass in architecturalmortar is still limited. This study attempts to develop a self-compacting based architectural mortar usingwhite cement and 100% recycled blue glass as key ingredients. To improve the aesthetic qualities, a cer-tain minimum quantity of glass cullets of larger particle size must be present. The influence of particlesize of the recycled glass on the engineering properties of fresh and hardened architectural mortar isinvestigated. The experimental results demonstrate that it is feasible to utilize 100% recycled glass asthe aggregate for the production of self-compacting based architectural mortar. These products havean average compressive strength of 40 MPa and flexural strength of 6 MPa at 28 days which are appro-priate for some architectural and building applications. Also, the overall performances of all the architec-tural mortars prepared with different particle sizes of glass aggregates are comparable to those of controlmortar mix prepared with river sand.

� 2011 Elsevier Ltd All rights reserved.

1. Introduction

A large volume of post-consumer beverage glass bottles is beingdisposed daily worldwide, only a small proportion is either washedfor reuse or re-melted to manufacture new glass. In Hong Kong, themajority of waste glass is used once or a few times before beingdiscarded into landfills [1]. According to government surveys, only1–2% of the waste glass in Hong Kong is recycled. The use of recy-cled glass as aggregate for the production of concrete blocks has re-ceived considerable application interest [2,3]. General resultsshowed that the recycled glass cullet can be used as a replacementfor natural aggregate in the production of concrete blocks withoutcompromising its mechanical properties [4]. Indeed, in many cases,the performance is enhanced [5].

There are a number of new applications of recycled waste glass,including the use of glass cullet in granular base/fill and asphaltpavement (Glassphlat) [6]. They have also been widely used asaggregates in cement mortar and concrete mixtures [7,8]. How-ever, most of the previous studies reported that the use of glassas a coarse aggregate has negative effects on bonding, adversealkali-silica reaction (ASR) and reduction in concrete strength [9].Thus, most of the recent works have concentrated on milling glasscullet into powder form (glass powder) to replace cement in

ll rights reserved.

g611@yahoo.com (T.-C. Ling),

concrete [10–13]. The implementation has gained wide acceptancedue to the innocuous behaviour of ASR in concrete [13]. Further-more, the use of supplementary cementing materials (SCM) suchas fly ash, metakaolin and slag are recommended as partialreplacements of cement for mechanical properties enhancementand mitigating ASR expansion [14,15].

Recently, the use of recycled glass cullets (due to their aestheticproperties) to produce decorative concrete has attracted muchinterest. In order to be aesthetically pleasing and visible, glassparticles of a certain minimum size must be present (for example,particle size ranging from 2.36 to 5 mm or 5 to 10 mm is needed).However, the use of glass aggregates of these sizes in concrete isalso the most detrimental to the concrete in terms of strengthreduction and ASR expansion [16].

This paper investigates the effect of different particle sizes ofglass aggregates on the engineering properties of fresh and hard-ened self-compacting based architectural mortar (SCAM). In addi-tion, the influence of different percentages of metakaolin aspartial replacements of white cement in the mortar is also reported.

2. Experimental programme

2.1. Materials

2.1.1. CementA white cement (WC) was chosen to be used in this study due to

aesthetic considerations for architectural mortar applications. It is

2676 T.-C. Ling, C.-S. Poon / Materials and Design 32 (2011) 2675–2684

particularly suitable for exposed decorative glass aggregate fin-ishes because it contains comparatively low alkali content whichcould reduce potential deterioration due to ASR expansion. Thechemical compositions of the cement are presented in Table 1.

2.1.2. Supplementary cementitious materialSupplementary cementitious materials such as fly ash and

metakaolin (MK) are well-known to be able to reduce the alkali-silica reaction (ASR) in concrete [16,17]. In order to achieve betteraesthetic value, white MK was chosen to be used in this study. Thechemical composition of MK is given in Table 1.

2.1.3. Fine aggregatesThe natural fine aggregate used in this study was river sand with

most of the particle sizes passing through a 2.36 mm sieve. Therecycled waste glass used was discarded post-consumer blue glassbottles sourced from a glass recycling company. The recycled blueglass bottles were washed and crushed in the laboratory. Thecrushed glass was sieved and sorted into three size classes (smallglass (SG), medium glass (MG) and large glass (LG)) according totheir gradations (<2.36 mm, 2.36–5 mm and 5–10 mm), respec-tively. The gradation curves of these glass aggregates and sand areshown in Fig. 1. Besides the gradation, the shape and texture of thesecrushed glass aggregates were different. As seen from Fig. 2, the SGaggregate appeared to be of irregular-rounded shape. The MGaggregate also had a similar shape but it had a rougher surface tex-ture and higher surface-contact angle. However, the LG aggregate isflat and angular in shape and it also has a smoother surface texture.

2.1.4. AdmixtureA superplasticizer GLENIUM SP8S based on a modified polycar-

boxylate was used to give satisfactory fluidity for the differentSCAM mixes. This product is free of chloride and complies withASTM C 494 [18] for Type A and F admixtures.

2.2. Mix proportions

In this study, the amount of cementitious materials was keptconstant at 706 kg/m3 and the cementitious to fine aggregate ratiowas fixed at 1:2. It aimed to provide a high volume fraction of finematerials as commonly used for self-compacting mortar design. Allthe mixtures were proportioned with a fixed water/cementitiousmaterials (w/c) ratio of 0.40 and the superplasticizer dosage of2.0–5.5% by weight of cementitious to obtain the targeted mini-slump flow values of 250 ± 10 mm. Two control mixtures namedcontrol-sand (CS) and control-glass (CG) were designed withcementitious content of (90%WC + 10%MK) and fine aggregate of100% sand (mix code: CS) and 100% crushed glass (60%SG +40%MG) (mix code: CG), respectively.

To enhance the aesthetic value of SCAM, it is necessary to in-clude glass of larger sizes in the mix proportions. To investigatethe influence of larger glass particles on the mechanical properties,the MG aggregate (2.36–5 mm) in the control-glass mix were sys-

Table 1Chemical composition of cementitious materials.

White cement Metakaolin

Chemical composition (%)Silicon dioxide (SiO2) 21.36 51.39Aluminum oxide (Al2O3) 5.27 32.91Ferric oxide (Fe2O3) 0.20 0.58Calcium oxide (CaO) 67.49 0.01Magnesium oxide (MgO) 1.14 0.01Sodium oxide (Na2O) 0.048 0.39Potassium oxide (K2O) 0.077 0.98Sulfur trioxide (SO3) 2.60 –Loss on ignition 1.58 13.57

temically replaced by LG aggregate (5–10 mm) at 25%, 50%, 75%and 100% by weight. The mix proportions of the two control mixesand the SCAM mixes are given in Table 2. For the CG-MK20 mix,the mix proportion was similar to that of the control-glass (CG) ex-cept that metakaolin was used to replace 20% of white cement.

2.3. Sample preparation

An appropriate mixing sequence and duration are essential forachieving good SCAM samples. The SCAM mixtures were preparedin a standard rotating drum-type mixer with a maximum capacityof 8 kg. Initially, fine aggregates (surface dried) and cementitiousmaterials were mixed for about 90 s to obtain a uniform mix indry conditions. Then, the superplasticizer (thoroughly mixed withwater) was added to the mix, and the mechanical mixing processwas resumed for another 90 s. To avoid the dry materials becomingstuck at the bottom part of the mixer, the mixture was mixed man-ually by turning it over twice or thrice using a steel trowel. Finally,the mixture was mechanically mixed for an additional 2 min tocomplete the whole mixing process. After the mixing was com-pleted, the fluidity of freshly prepared mortar was evaluated bymeasuring the spread diameter of the mortar in two perpendiculardirections specified by European specification EFNARC [19].

Twelve 40 � 40 � 160 mm prisms were prepared for flexuraland compressive strength tests. Six additional prism samples werecast for the determination of water absorption and resistance toacid attack. The drying shrinkage and expansion due to thealkali-silica reaction (ASR) were determined on prisms withdimensions of 25 � 25 � 285 mm. A 70 � 70 � 70 mm cube wascast to determine the abrasion resistance.

After casting, all the specimens were covered with a thin plasticsheet and left in the laboratory at a temperature of 23 ± 3 �C and75% relative humility. After demoulding for one day, three25 � 25 � 285 mm prisms were used to determine the initial dry-ing shrinkage values before they were transferred to a dryingchamber at a temperature of 23 �C and relative humidity of 50%.The other specimens were stored in a water tank at an averagetemperature of 23 ± 3 �C until further testing.

2.4. Testing details

2.4.1. Fresh propertiesA mini-slump flow cone with an internal diameter of 100 mm

was used to evaluate the fluidity of fresh mortar mixture as de-scribed by European specification EFNARC [19]. Before the test,the truncated cone mould was placed on a clean metal plate andthe freshly prepared mortar mixture was poured into the conewithout any compaction. Once the cone was fully filled with themortar, the cone was lifted vertically and the spread diameters ofthe freshly prepared mortar in two perpendicular directions weremeasured. Occurrence of segregation and/or bleeding, if any, wasvisually observed and noted during the mini-slump flow test.

2.4.2. Permeable voids and water absorptionThe permeable voids and water absorption test were conducted

to assess the water permeability characteristics of SCAM in accor-dance with ASTM 642 [20]. For this test, three 40 � 40 � 160 mmprism specimens were cured in water for 90 days. The surfaces ofthe saturated specimens were dried by removing surface moisturewith a towel and the weight was determined (W90). Afterwards, thesurface dried specimens were further dried in an oven at a constanttemperature of 105 ± 5 �C until a constant weight was achieved(Wod). The permeable voids in percentage are given below:

Permeable voids ¼ ½ðW90 �WodÞ=V � � 100 ð1Þ

where V is the volume of prism specimen.

0

10

20

30

40

50

60

70

80

90

100

10 5 2.36 1.18 0.6 0.3 0.15 0

Sieve size (mm)

Cum

ulat

ive

pass

ing

(%)

LG:5-10mm

MG:2.36-5mm

SG: 0-2.36mm

Sand

Fig. 1. Particle size distribution of different particle sizes of blue glass aggregates.

SG: <2.36mm MG: 2.36-5mm LG: 5-10mm

Irregular and rounded

flat and angular

Fig. 2. Shapes and textures of different particle sizes of blue glass aggregates (SG: <2.36 mm, MG: 2.36–5 mm and LG: 5–10 mm).

Table 2Mix proportions of SCAM mixtures (kg/m3).

Notation w/c Cementitious Sand Recycled blue glass cullet % of MG replaced by LG SPa Mini slump

WC MK SG MG LG (mm)

Control-sand (CS) 0.4 282 635 71 1412 0 0 0 – 5.5 243Control-glass (CG) 0.4 282 635 71 0 847 565 0 0 3.8 257CG-LG25 0.4 282 635 71 0 847 424 141 25 3.5 243CG-LG50 0.4 282 635 71 0 847 282 282 50 3.0 245CG-LG75 0.4 282 635 71 0 847 141 424 75 2.5 243CG-LG100 0.4 282 635 71 0 847 0 565 100 2.0 253CG-MK20 0.4 282 565 142 0 847 565 0 – 4.0 245

a % of SP dosage as cementitious weight.

T.-C. Ling, C.-S. Poon / Materials and Design 32 (2011) 2675–2684 2677

Subsequently, the initial surface absorption (ISA) and finalwater absorption (FWA) tests were carried out by immersing theoven dried specimens completely in water. The specimens were re-moved from the water immersion and weighed at 30 min and 96 hto evaluate the mass gained for ISA and FWA, respectively. The ISAand FWA values were determined by the following formulation:

ISA ¼ ½ðW30min �WodÞ=Wod� � 100 ð2ÞFWA ¼ ½ðW96h �WodÞ=Wod� � 100 ð3Þ

where W30min is the weight of surface dried specimen after 30 minof immersion, and W96h is the weight of surface dried specimenafter 96 h of immersion.

2.4.3. Flexural strengthA three-point flexural strength test in conformity with ASTM

C348 [21] was performed at 1, 7, 28 and 90 days after casting. Acentre line was marked at the top of the 40 � 40 � 160 mm prismspecimens, using a black felt-tip marker perpendicular to itslength. The SCAM specimens were tested under a central line loadwhile simply supported over a span of 120 mm. For this test, a uni-versal test machine with a load capacity of 50 kN was used with adisplacement rate of 0.10 mm/min set.

2.4.4. Equivalent compressive strengthThe equivalent compressive strength test was carried out

according to ASTM C349 [22]. The compressive strength was

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

Control-

Sand

Control-

Glass

CG-LG25 CG-LG50 CG-LG75 CG-LG100 CG-MK20

Supe

rpla

stic

izer

dos

age

(%)

Fig. 3. Fluidity (% of SP) of control and SCAM mixes.

2678 T.-C. Ling, C.-S. Poon / Materials and Design 32 (2011) 2675–2684

determined using a Denison compression machine with a loadcapacity of 3000 kN on the broken pieces (portions of the prismsbroken in flexure). The reported test results were the average ofsix measurements.

2.4.5. Drying shrinkageA modified British standard (BS ISO, Part 8: 1920) method was

used for the drying shrinkage test in this study [23]. After demoul-ding, the initial length of three 25 � 25 � 285 mm mortar bar spec-imens was measured. After the reading, the specimens wereconveyed to a drying chamber at a temperature of 23 �C and rela-tive humidity of 50% until further measurements at 1st, 4th, 7th,28th, 56th and 90th days, and the final length measurement wasrecorded at the 112th day. The length of each specimen was mea-sured within 15 min after it was removed from the dryingchamber.

2.4.6. Expansion due to alkali-silica reactionFor each mix, three 25 � 25 � 285 mm mortar bar specimens

were used for the ASR test in accordance with ASTM C1260 [24] –the accelerated mortar bar method. A zero reading was taken afterstoring the prisms in distilled water at 80 �C for 24 h. The mortarbars were then transferred and immersed in 1 N NaOH solution at80 �C until the testing time. The expansion of the mortar bars wasmeasured within 15 ± 5 s after they were removed from the 80 �Cwater or alkali storage condition by using a length comparator.The measurements were conducted at the 1st, 4th, 7th, 14th, 21stand 28th days.

2.4.7. Abrasion resistanceThe abrasion resistance test was determined by abrading the

surface of the 70 � 70 � 70 mm cube specimen after 28 days ofwater curing as specified by BS 6717 [25]. The test began by thetest specimen being placed in contact with an abrasion wheelrotating at the rate of 75 revolutions in 60 ± 3 s. The dimensionof the groove resulting from the abrasive action was measured toevaluate the abrasion resistance of the mortar. A smaller grooveindicated a better resistance to abrasion.

2.4.8. Chemical resistanceThe chemical resistance was studied by immersing the speci-

mens in a sulphuric acid solution in accordance with ASTM C267 [26]. After the 28 days of curing, three 40 � 40 � 160 mmprisms were removed from the water tank and each specimenwas marked and tied with a nylon string around them. After the

initial weight was recorded, the specimen was immersed in a 3%H2SO4 solution. The solution was replaced at 4-week regular inter-vals to ensure consistent acid concentration throughout the testperiod. The specimens were extracted from the solution and thesurfaces were cleaned with a soft nylon brush before their weightswere measured. The weight of specimens was measured weekly atthe 4th week and subsequently the reading was measured at the8th week and 12th week. The cumulative mass loss of each speci-men expressed as a percentage is given below:

Cumulative mass loss ¼ ½ðMt �MintÞ=Mint� � 100 ð4Þ

where Mt is the mass at time t, and Mint is the initial mass beforeimmersion in sulphuric acid.

3. Results and discussion

3.1. Fresh properties

The mini-slump flow diameters of control and SCAM fresh mix-tures were in the range of 240–260 mm specified by Europeanspecification EFNARC [19] and the amount of superplasticizer(SP) required for each mixture is presented in Fig. 3. A totalreplacement of sand with glass aggregates increased the workabil-ity due to the fact that glass aggregates have smoother surfacelayer and nearly zero water absorption properties. It is importantto note that for every 25% of MG aggregate replaced by LG aggre-gate, superplasticizer contents had to be reduced by about 0.5%in order to achieve the same targeted mini-slump flow diameter.A decrease in the superplasticizer dosage with increasing LG aggre-gate replacement level might be attributed to the larger particlesize of LG aggregate which led to a reduction in total surface areaper unit volume. As shown in Table 2 and Fig. 4, for a given w/c andmetakaolin content, an increase in LG aggregate led to a slight in-crease in bleeding and segregation. This effect was more pro-nounced when more LG aggregate was incorporated. This mightbe due to the fact that the LG aggregate had a flat and smoothshape which was prone to segregation. As seen in Fig. 4, thereplacement of cement by metakaolin strongly influenced the flu-idity. Increasing metakaolin from 10% to 20% enhanced the cohe-siveness of the mortar which in turn reduced the bleeding andsegregation. Even though a higher superplasticizer dosage of 4.0was required for the CS-MK20 mixture, no bleeding and segrega-tion could be seen.

Fig. 4. Mini-slump flow appearance of SCAM mixes.

T.-C. Ling, C.-S. Poon / Materials and Design 32 (2011) 2675–2684 2679

3.2. Surface appearance

The pictures of blue glass aggregates exposed in SCAM after cut-ting the samples by a diamond saw are given in Fig. 5. As thereplacement ratio of MG aggregate by LG aggregate increased,the aesthetics of the surface appearance improved although thereis no objective method of measuring aesthetic value.

3.3. Water absorption and permeable voids

The effects of LG aggregate and metakaolin content on the ini-tial surface absorption (ISA), final water absorption (FWA) and per-meable voids values of control and SCAM specimens are illustratedin Fig. 6. It can be seen that the ISA increased with an increase in LGaggregate content. As the LG was use to replace 100% MG, the ISAvalue was increased from 3.02% to 3.47%. This might be due to the

Fig. 5. Architectural mortar featuring differen

large amount of LG aggregate leading to an increase in bleedingand a more porous microstructure.

The FWA is considered to be the water that is continually ab-sorbed at the inert zone of the mortar after the ISA test. It can beseen from Fig. 6 that all the mixtures continued to absorb waterfor an extended period from 30 min (average of 2.4% for ISA) upto 96 h (average of 7.5% for FWA). An opposite trend of absorp-tion rate could be seen between the ISA (near surface) and theFWA (inert mortar). The water absorption ratio (FWA/ISA) de-creased with increase of LG aggregate content. The decrease inthe water absorption ratio with increasing LG aggregate replace-ment level might be attributed to the smooth surface of LGaggregate which causes lower air content in the produced mor-tar. On the contrary, the irregular shape of MG aggregate resultedin a larger relative surface area that trapped more air in the mor-tar matrix.

t particle sizes of blue glass aggregates.

2680 T.-C. Ling, C.-S. Poon / Materials and Design 32 (2011) 2675–2684

The FWA was reduced by the inclusion of recycled glass as totalreplacement of sand in the mortar. This could be due to the factthat glass by its nature is an impermeable material, resulting inlower air void contained in the mortars. It was also noted that anincrease in metakaolin content led to an increase in the FWA val-ues. Similar observations concerning the effect of metakaolin onthe water absorption characteristic of concrete was also reportedby Khatib and Clay [27]. It can be concluded that the FWA (after96 h) and permeable voids values were closely related, whereasas the FWA increased the permeable voids also increased.

3.4. Flexural strength

The results of flexural strength of the control and SCAM mixesincorporating different sizes of glass aggregates are given inFig. 7. The results revealed that the flexural strength of all theSCAM specimens increased when the curing age was extended.After 7 days of water curing, the flexural strength reached approx-imately 90% of the 28-day flexural strength. This indicated a rapidhydration during the early 7 days. It can be noted that at the age of1 and 7 days, the flexural strength of all the SCAM specimens wascomparable, while at the ages of 28 and 90 days, the flexuralstrength seemed to be affected more by the grading of the glass

0

2

4

6

8

10

12

14

16

18

20

Control-Sand

Control-Glass

CG-LG25 CG-

ISA

, FW

A a

nd p

erm

eabl

e vo

ids

(%)

ISA (30min) FWA (

Fig. 6. ISA, FWA and permeable voi

0.00

2.00

4.00

6.00

8.00

10.00

12.00

0 20 40D

Fle

xura

l str

engt

h (M

Pa)

Fig. 7. Flexural strength of c

aggregates than by the cement matrix characteristic. The flexuralstrength at 28 and 90 days decreased as the replacement ratio ofMG aggregate with LG aggregate increased. A possible reason forthis is that the LG aggregate had a smoother surface which couldsignificantly weaken the bonding strength between the LG aggre-gate and the surrounding cement mortar, thus giving a lowerstrength. On the contrary, the smaller particle size of MG aggregateenhanced the aggregate-cement matrix bonding strength.

Besides, when glass aggregates were used as a total replace-ment for sand, an average reduction of 36.7% and 33.6% in 28-day and 90-day flexural strength were observed, respectively. Thisis because the bonding strength of cement pastes with smooth sur-face glass aggregate was weaker than that with rough sand. It wasalso observed that with an increase in metakaolin from 10% to 20%,the flexural strength was decreased by about 7.5% at 90 days. Interms of strength properties, Li and Ding [28] suggested that thebest performance of metakaolin replacement was achieved at 10%.

3.5. Equivalent compressive strength

The compressive strength results of the control and SCAMmixes are presented in Fig. 8. The results show that the rate of gainin compressive strength was rapid up to 7 days and then slowed

LG50 CG-LG75 CG-LG100 CG-MK20

96h) permeable voids

ds of control and SCAM mixes.

60 80 100ay

Control-Sand Control-Glass

CG-LG25 CG-LG50

CG-LG75 CG-LG100

CG-MK20

ontrol and SCAM mixes.

T.-C. Ling, C.-S. Poon / Materials and Design 32 (2011) 2675–2684 2681

down afterward. At later ages of 28 and 90 days, it was observedthat the compressive strength decreased with increasing LG aggre-gate content in a similar manner with that observed in flexuralstrength. As mentioned earlier, this might be due to the fact thatthe LG aggregate had relatively smoother surfaces and often con-tained a series of stepped fractures which tended to decrease thebonding strength between the glass aggregate and the cementmatrix.

The reduction rate of SCAM incorporating 100% of glass aggre-gates as sand replacement was much lower in compressivestrength than in flexural strength. It was possibly because theinfluence of weaker bonding by the presence of glass aggregatesin cement mortar was much lower in compression failure than thaton bending. As the flexural strength was found to be about one-fourth to one-sixth of the compressive strength, it is always inter-esting to establish a relationship between the two parameters. Therelationship between the average flexural strength and the com-pressive strength is plotted in Fig. 9 which shows a strong linearcorrelation.

3.6. Drying shrinkage

The drying shrinkage results of the control and SCAM mixesat various ages up to 112 days are shown in Fig. 10. In general,

0.00

10.00

20.00

30.00

40.00

50.00

60.00

0 10 20 30 40D

Com

pres

sive

str

engt

h (M

Pa)

Fig. 8. Compressive strength of

0

2

4

6

8

10

12

0 10 20 30

Compressive

Fle

xura

l str

engt

h (M

Pa)

Fig. 9. Relationship between flexural and compr

the drying shrinkage values of all samples were similar up to4 days, while considerable differences were observed for the la-ter period. The results show that the particle sizes of glass aggre-gates had a significant effect on the drying shrinkage. Thereduction of drying shrinkage values by incorporating LG aggre-gate may be due to the lower absorption capacity of glassaggregates.

In comparison, a control-sand mix showed the highest value ofdrying shrinkage at 1 day up to 56 days. However, the dryingshrinkage was stable as the curing time continued to increase upto 112 days. Also, increasing the replacement of white cement bymetakaolin from 10% to 20% slightly reduced the drying shrinkageby 4%. This was because of the hydration of cement and secondarypozzolanic reaction of metakaolin which used up significantamounts of the free water in the cement mortar [29]. The test re-sults at 56 days for control and all the SCAM specimens satisfiedthe drying shrinkage requirement (<0.075%) of the Australian Stan-dard AS 3600 [30].

3.7. Alkali-silica reaction

The ASR results of the control and SCAM mixes are shown inFig. 11. As seen in the figure, the expansions due to ASR were lessthan 0.10% at 14 days. The use of metakaolin admixture (10–20%)

50 60 70 80 90 100ay

Control-Sand

Control-Glass

CG-LG25

CG-LG50

CG-LG75

CG-LG100

CG-MK20

control and SCAM mixes.

y = 0.118x + 1.36

R2 = 0.96

40 50 60 70

strength (MPa)

essive strength of control and SCAM mixes.

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0 20 40 60 80 100 120Day

Len

gth

chan

ge (

%)

Control-Sand Control-Glass

CG-LG25 CG-LG50

CG-LG75 CG-LG100

CG-MK20

Fig. 10. Drying shrinkage curve of control and SCAM mixes.

-0.02

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0 5 10 15 20 25 30

Day

ASR

exp

ansi

on (

%)

Control-Sand

Control-Glass

CG-LG25

CG-LG50

CG-LG75

CG-LG100

CG-MK20

Fig. 11. ASR reactivity of control and SCAM mixes.

2682 T.-C. Ling, C.-S. Poon / Materials and Design 32 (2011) 2675–2684

was found to be able to successfully mitigate ASR expansion. It wasalso noticed that when the particle size of blue glass aggregateswas in the range of 2.36–5 mm and 5–10 mm, the deleterious ef-fect on ASR expansion was negligible.

3.8. Abrasion resistance

The effect of LG aggregates and metakaolin replacement on theabrasion resistance of SCAM is shown in Fig. 12. As seen in the fig-ure, the abrasion resistance was reduced by the inclusion of glassaggregates as total sand replacement in the mortar. Also, the abra-sion resistance was reduced with an increase in the LG aggregatecontent. According to a previous study [31], the percentage wearof coarser glass aggregates was higher than that of finer glassaggregates. It was found that SCAM incorporating 20% of metakao-lin exhibited a weaker abrasion resistance than all other SCAMmixes. This is consistent with a previous study [4] which foundthat the ability of concrete to withstand abrasion decreased witha decrease in concrete strength.

3.9. Chemical resistance

The loss of mass of SCAM mixes upon immersing in a 3% solu-tion of H2SO4 up to 12 weeks is shown in Fig. 13. The control-sand

specimen showed the most significant deterioration with respectto loss of mass. The inclusion of glass aggregate in mortar signifi-cantly reduced the mass loss. Up to 12 weeks of immersion, theloss of mass of SCAM mixes incorporating 100% of glass aggregatesranged from 44.0% to 45.9%. When it was closely examined, the sig-nificant increase in the rate of mass loss was in the first 4 weeks.The loss of mass during this period was probably related to the re-moval of material from the surface of the specimens by degrada-tion of the cementitious matrix. Besides, the rate of mass lossdecreased with additional immersion time. This can be explainedby the fact that after the deterioration of the surface layer, a higherpercentage of the inert glass aggregates which had better acidresistance than the cement mortar were exposed to the sulphuricacid solution. However, due to the degradation of the cement ma-trix, some of the glass aggregates were dislodged from thespecimens.

It can be observed that the composition of cementitious mate-rials has a pronounced effect on the mass loss results. ForCG-MK20 specimens, the mass loss at 12 weeks was about 22.3%lower than that of the Control-Glass specimens. This can be ex-plained by the increase in metakaolin content in the cement matrixwhich reduced the amount of calcium hydroxide content availablefor the acid–base reaction, thus resulting in a lower rate of acid at-tack [32].

15.0

16.0

17.0

18.0

19.0

20.0

21.0

22.0

23.0

Control-

Sand

Control-

Glass

CG-LG25 CG-LG50 CG-LG75 CG-LG100 CG-MK20

Abr

asio

n re

sist

ance

(m

m)

Fig. 12. Abrasion resistance of control and SCAM mixes.

0.00

10.00

20.00

30.00

40.00

50.00

60.00

0 1 2 3 4 5 6 7 8 9 10 11 12 13

Week

Mas

s lo

ss (

%)

Control-Sand Control-Glass

CG-LG25 CG-LG50

CG-LG75 CG-LG100

CG-MK20

Fig. 13. Rate of mass loss over time of control and SCAM mixes.

T.-C. Ling, C.-S. Poon / Materials and Design 32 (2011) 2675–2684 2683

4. Conclusion

The results of this study concluded that it is feasible to utilize100% recycled glass as decorative aggregates in the production ofself-compacting based architectural mortar. Glass-modified cementmortar exhibited lower strength values and abrasion resistance ascompared to sand-controlled cement mortar. However, fullyreplacement of sand by recycled glass has improved workability,resistance in term of water absorption, drying shrinkage and acid at-tack of cement mortar. Also, it seems that SCAM products of variousappearances for diverse applications can be achieved by using com-binations of different particle sizes of glass aggregates. The overallperformance of the architectural mortar was studied. Based on thelaboratory results, the following conclusions may be drawn:

� Incorporating glass aggregates, particularly LG aggregate inSCAM increased the fluidity which in turn reduced the use ofsuperplasticizer dosage. However, incorporating metakaolin asa supplementary mineral admixture in the mortar mixturesincreased the superplasticizer dosage to maintain workability.� The presence of LG aggregate in the SCAM mixtures gradually

decreased the final water absorption; however, increasing thelevel of metakaolin admixture increased the water absorptionvalues. In all cases, the final water absorption and permeablevoids were directly related.

� The presence of glass aggregates caused a reduction in bothflexural and compressive strengths. Also, increasing the LGaggregate and metakaolin content led to a decrease in the flex-ural and compressive strength, particularly between the curingage of 28 and 90 days.� The drying shrinkage of SCAM decreased as the LG aggregate

increased. The use of metakaolin also marginally reduced thedrying shrinkage.� Using 10% of metakaolin successfully mitigated the ASR expan-

sion of mortar. Generally, the size of blue glass aggregates ran-ged from of 2.36 to 5 mm and 5 to 10 mm was found to havenegligible ASR expansion in SCAM.� The SCAM specimens showed lower abrasion resistance when

the LG aggregate and metakaolin content was increased from0% to 100% and 10% to 20%, respectively.� The use of glass aggregates and metakaolin in SCAM signifi-

cantly enhanced the ability to resist acid attack. However, theacid resistance of SCAM slightly decreased when LG aggregateswere used to replace MG aggregate.

Acknowledgements

The authors would like to thank The Hong Kong PolytechnicUniversity and SHK Properties for funding supports.

2684 T.-C. Ling, C.-S. Poon / Materials and Design 32 (2011) 2675–2684

References

[1] Environmental Protection Department. Future landfill development in HongKong. Hong Kong: The Government Printer; 2006.

[2] Lam CS, Poon CS, Chan D. Enhancing the performance of pre-cast concreteblocks by incorporating waste glass – ASR consideration. Cem Concr Compos2007;29(8):616–25.

[3] Chen J, Poon CS. Photocatalytic activity of titanium dioxide modified concretematerials – influence of utilizing recycled glass cullets as aggregates. J EnvironManage 2009;90(11):3436–42.

[4] Poon CS, Chan D. Effects of contaminants on the properties of concrete pavingblocks prepared with recycled concrete aggregate. Constr Build Mater2007;21(8):164–75.

[5] Chen J, Poon CS. Photocatalytic construction and building materials: fromfundamentals to applications. Build Environ 2009;44(9):1899–906.

[6] Siddique R. Waste materials and by-products in concrete. Berlin,Heidelberg: Springer-Verlag; 2008. p. 151.

[7] Corinaldesi V, Gnappi G, Moriconi G, Montenero A. Reuse of ground waste glassas aggregate for mortars. Waste Manage 2005;25(2):197–201.

[8] Ismail ZZ, AL-Hashmi EA. Recycling of waste glass as a partial replacement forfine aggregate in concrete. Waste Manage 2009;29(2):655–9.

[9] Kou SC, Poon CS. Properties of self-compacting concrete prepared withrecycled glass aggregate. Cem Concr Compos 2009;31(2):107–13.

[10] Topçu IB, Canbaz M. Properties of concrete containing waste glass. Cem ConcrRes 2004;34(2):267–74.

[11] Shao Y, Lefort T, Moras S. Damian Rodriguez studies on concrete containingground waste glass. Cem Concr Res 2000;30(1):91–100.

[12] Shi C, Wu Y, Riefler C, Wang H. Characteristics and pozzolanic reactivity ofglass powders. Cem Concr Res 2005;35(5):987–93.

[13] Schwarz N, Cam H, Neithalath N. Influence of a fine glass powder on thedurability characteristics of concrete and its comparison to fly ash. Cem ConcrCompos 2008;30(6):486–96.

[14] Shayan A, Xu A. Value-added utilisation of waste glass in concrete. Cem ConcrRes 2004;34(1):81–9.

[15] Topçu _IB, Boga AR, Bilir T. Alkali–silica reactions of mortars produced by usingwaste glass as fine aggregate and admixtures such as fly ash and Li2CO3. WasteManage 2008;28(5):878–84.

[16] Lee G, Ling TC, Wong YL, Poon CS. Effects of crushed glass cullet sizes, castingmethods and pozzolanic materials on ASR of concrete blocks. Constr BuildMater 2011. doi:10.1016/j.conbuildmat.2010.12.008.

[17] Taha B, Nounu G. Properties of concrete contains mixed colour waste recycledglass as sand and cement replacement. Constr Build Mater 2008;22(5):713–20.

[18] ASTM C494. Standard specification for chemical admixtures for concrete. US:American Society of Testing Materials; 2010.

[19] EFNARC. Specification and guidelines for self-compacting concrete. Norfolk,UK, English ed.: European Federation for Specialist Construction Chemicalsand Concrete System; February 2002.

[20] ASTM C 642. Standard test method for density, absorption, and voidsin hardened concrete. US: American Society of Testing Materials;2006.

[21] ASTM C348. Standard test method for flexural strength of hydraulic-cementmortars. US: American Society of Testing Materials; 2008.

[22] ASTM C349. Standard test method for compressive strength of hydraulic-cement mortars (using portions of prisms broken in flexure). US: AmericanSociety of Testing Materials; 2008.

[23] BS ISO 1920-8. Determination of drying shrinkage of concrete for samplesprepared in the field or in the laboratory. UK: British Standard Institution;2007.

[24] ASTM C 1260. Standard test method for potential alkali reactivity ofaggregates (mortar-bar method). US: American Society of Testing Materials;2007.

[25] BS 6717. Precast, unreinforced concrete paving blocks-method for measuringabrasion resistance. UK: British Standards Institution; 2001.

[26] ASTM C 267. Standard test methods for chemical resistance of mortars, grouts,and monolithic surfacings and polymer concretes. US: American Society ofTesting Materials; 2001.

[27] Khatib JM, Clay RM. Absorption characteristics of metakaolin concrete. CemConcr Res 2004;34(1):19–29.

[28] Li Z, Ding Z. Property improvement of Portland cement by incorporating withmetakaolin and slag. Cem Concr Res 2003;33(4):579–84.

[29] Brooks JJ, Johari MMA. Effect of metakaolin on creep and shrinkage of concrete.Cem Concr Compos 2001;23(6):495–502.

[30] AS 3600. Concrete structures-incorporating AMD 1: May 2002 and AMD 2.Australia: Australia Standard; October 2004.

[31] HDR Engineering, Inc. Glass cullet utilization in civil engineering application.Nebraska State Recycling Association; 1997.

[32] Al-Akhras NM. Durability of metakaolin concrete to sulfate attack. Cem ConcrRes 2006;36(9):1727–34.