Contents lists available at ScienceDirect

Composites Part B

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

Effect of slag on the mechanical properties and bond strength of fly ash-based engineered geopolymer composites

Yifeng Linga, Kejin Wanga,∗, Wengui Lib, Guyu Shia, Ping Luc

a Department of Civil, Construction and Environmental Engineering, Iowa State University, Ames, IA, 50011, USAb Center for Built Infrastructure Research (CBIR), School of Civil and Environmental Engineering, University of Technology Sydney, NSW, 2007, Australiac Bridge Engineering Center, Institute for Transportation, Iowa State University, Ames, IA, 50010, USA

A R T I C L E I N F O

Keywords:Engineered geopolymer composite (EGC)Fly ashSlagMechanical property

A B S T R A C T

Recently, the concept of engineered cementitious composites (ECCs) has been extended to the creation of en-gineered geopolymer composites (EGCs). Although showing similar mechanical characteristics (e. g., strainhardening and multiple cracking) to conventional ECC, the strength of existing EGC is generally low, and thissometimes restrains its applications. In the present study, a low-calcium (Class F) fly ash-based, polyvinyl alcohol(PVA) fiber reinforced EGC was developed and further modified by a ground-granulated blast-furnace slag (slag).The slag was used to replace the fly ash at content of 0%, 10%, 20%, and 30% (by weight). The effects of the slagon the mechanical properties (e.g., compressive strength, modulus of elasticity, uniaxial tensile behavior, flex-ural bending strength, and pullout bond strength) of the EGCs were investigated. The results revealed that allEGCs studied exhibited a strain/deflection hardening behavior under tension/flexure, and all slag replacementsfor fly ash enhanced strength-related properties but reduced ductility-related properties of the EGCs. The EGCmix with 20% slag replacement for fly ash (FA-20%S) had 102.3 MPa compressive strength, 6.8MPa tensilestrength, and 6.2MPa bond strength, while the EGC mix with no slag (FA-0%S) had 72.6MPa compressivestrength, 4.7MPa tensile strength, and 3.5MPa bond strength at 28 days. These strength enhancements weremainly attributed to the improved density of the EGC matrix and the bond between the matrix and fiber. Thereare close relationships between the bond strength and other strengths, especially the tensile and flexuralstrengths, of the EGCs.

1. Introduction

Engineered cementitious composite (ECC) is a mortar compositereinforced with a small amount of discontinuous fibers (typically≤ 2%by volume). Having a characteristic of strain hardening, ECC oftenexhibits ductility as approximately 600 times as conventional Portlandcement concrete (PCC) [1] and a tensile strain capacity up to 6% [2].One feature of ECC is its strain hardening behavior, especially undertension and bending, accompanied by multiple fine cracks, which leadsto improved ductility, toughness, fracture energy [3]. To achieve thesesuperior properties, a typical ECC mix often needs cement content 2–3times as high as a conventional PCC mix [4]. As the manufacture ofPortland cement is an energy intensive and high CO2 emission process,use of non-clinker cement, e.g., geopolymer, to produce ECC is an at-tractive sustainable alternative. Research has shown that the produc-tion of fly ash based geopolymer consumes approximately 60% lessenergy and generates about 80% less CO2 emission when it is compared

to the manufacture of Portland cement [5,6].Recently, a feasibility study was conducted to develop geopolymer-

based ECC, known as engineered geopolymer composite (EGC), wherePortland cement was completely replaced by a fly-ash-based geopo-lymer [7]. Fly ash is extensively used in geopolymers because of itsaluminosilicate composition, low water demand for high workability,and worldwide availability. In an alkali solution, fly ash undergoesgeopolymerization, forming structurally disordered, highly cross-linkedalkaline aluminosilicate-hydrate (N-A-S-H) gels that bonds loss parti-cles/fibers through solidification [8]. The previous research has in-dicated that the EGC displayed similar mechanical characteristics as aconventional ECC [7]. Under uniaxial tension or bending, the EGC alsoshowed strain and deflection hardening accompanied by multiple finecracks. However, due to the low reactivity of fly ash, the fly-ash-basedEGC mixes had relatively low strengths 28 (17.4–27.6MPa in com-pression and 2.9–3.4MPa in tension at 28 days), while a typical ECCgenerally has much higher strengths (50–60MPa in compression and

https://doi.org/10.1016/j.compositesb.2019.01.092Received 8 July 2018; Received in revised form 4 January 2019; Accepted 18 January 2019

∗ Corresponding author.E-mail address: kejinw@iastate.edu (K. Wang).

Composites Part B 164 (2019) 747–757

Available online 01 February 20191359-8368/ © 2019 Published by Elsevier Ltd.

T

4–5MPa in tension at 28 days). The low strength of existing EGC hasrestrained its applications in the construction industry.

Lately, significant advances have been made to improve propertiesof conventional fly ash-based geopolymers, and one of promising ap-proaches is to use reactive powders, such as ground-granulated blast-furnace slag (GGBFS or slag), to replace fly ash [9]. Compared with flyash, GGBFS generally has much higher reactivity and better cementingproperties, contributed mainly by its high CaO content, thus increasingearly age strength of the geopolymer composite [10]. Moreover, CaOcan also participate in fly ash activation process and produce C-S-H/C-A-S-H gel along with fly ash-based geopolymer gel (N-A-S-H gel), thusfurther improving mechanical properties and altering microstructure ofthe geopolymer [11]. Deb et al. [12] studied the effects of slag andactivator content on the workability and strength properties of fly ash-based geopolymer concrete, and they found that the compressivestrength of slag-fly ash geopolymer increased with increasing slagcontent at all ages up to 180 days. Nath and Kumar studied varioustypes of slags blended with fly ash for geopolymers, and they concludedthat the strength improvements are mainly attributed to higher degreeof slag reaction, formation of more gel phases and development of morecompact microstructure [11,13,14]. Nevertheless, very limited studyhas been performed exploring the effects of slags on mechanical prop-erties of fly ash-based EGCs since the development of EGC is still in itsinfancy.

The present study aims at investigating the mechanical properties,including compressive, flexural, tensile and geopolymer-reinforcingsteel bond strengths, of newly developed fly ash-based EGC mixes witha partial slag replacement. In this study, the control fly ash-based EGCmix (no slag) was developed from a conventional fly ash-based geo-polymer mix based on a previous study [15]. The geopolymer mix wastailored for improved workability by using a low calcium (Class F) flyash. In addition to slag replacement, a reduced activator-to-binder ratio(by mass) and increased activator concentration were used to enhancethe strength of EGC. The fiber commonly used in conventional ECC wasused for EGC. Besides mechanical properties, the microstructure of theEGC mixes, especially the geopolymer-fiber bond, was also evaluatedunder a scanning electron microscope (SEM).

2. Experimental program

2.1. Raw materials

The fly ash used in this study was class F fly ash with a specificgravity of 2.61 g/cm3 and fineness of 426m2/kg, and it had CaO con-tent of 11.8% (<15%), SiO2+Al2O3+Fe2O3 of 81.3% (> 70%) andSO3 of 0.45% as in compliance with ASTM C618 [16]. The slag usedwas GGBFS with a specific gravity of 2.50 g/cm3 and a fineness of455m2/kg, and it had a CaO/SiO2 ratio of 1.13 (between 0.5 and 2.0)and an Al2O3/SiO2 ratio of 0.23 (between 0.1 and 0.6). The hydrationmodulus, defined as the ratio of CaO + MgO + Al2O3 to SiO2, was 1.62(> 1.4), indicating an appropriate reactivity [16]. The major chemicalcomponents of the fly ash and slag are provided in Table 1. The acti-vator used was made from NaOH (solid) and Na2SiO3 (water glass, li-quid), and their specifications are listed in Table 2. The fiber used was apolyvinyl alcohol (PVA) fiber that is commonly used for ECC, and itscharacteristics are presented in Table 3.

2.2. Mix proportion

A liquid activator, consisting of solid NaOH, Na2SiO3 solution, andtap water, was used for the EGC. A SiO2/Na2O mole ratio (Module) of1.0 and solute (solid NaOH and Na2SiO3) concentration (Concentration)of 30% (by mass) were selected to prepare the activator solution for theEGC based on the authors’ previous experience in fly ash-based geo-polymer [15]. Four fly ash-based EGC mixes were prepared with 0%,10%, 20%, and 30% (by weight) slag replacements for fly ash. Thesemixes are denoted as (1) FA-0%S, (2) FA-10%S (3) FA-20%S, and (4)FA-30%S. An activator solution-to-binder (FA and slag) ratio of 0.27was selected after some trial tests to achieve desirable strength as wellas appropriate workability. Similar to conventional ECC, 2% PVA fiber(by volume of matrix) was added to the EGC mixes. Table 4 presents themix proportions of the slag blended, fly ash-based EGC mixes reportedin this paper.

2.3. Specimen preparation

The EGC mixing started with a dry mixing of the binder materials(i.e., FA and slag) for 2min in a laboratory Hobart mixer at a low speed(107 rpm). The activator solution, which was prepared 24 h prior to theEGC mixing to allow the dissipation of the heat attributed to the exo-thermic chemical reaction between NaOH and Na2SiO3, was then gra-dually added to the running mixer, and the mixture was mixed foranother 3min. Next, PVA fibers were added in the running mixerprogressively to obtain a uniform fiber dispersion. After all fibers wereadded, the mixture was mixed at a high speed (198 rpm). During themixing, a small amount of the mixture was taken out periodically tocheck fiber dispersion by hand. The mixing ended when no fiber ag-glomeration was felt. The entire mixing procedure for each compositegenerally took 8–10min.

Table 1Chemical composition of fly ash and slag used.

SiO2 Al2O3 Fe2O3 SO3 CaO MgO Na2O K2O Others LOI

Fly ash 57.06 18.82 5.43 0.45 11.8 2.89 0.64 1.12 1.74 0.03Slag 36.5 8.54 0.83 0.6 41.1 9.63 0.29 0.44 2.07 2.46

Note: All values in mass %, expressed on an oven-dry basis; LOI: loss on ignitionat 1000 °C.

Table 2Specification of sodium silicate solution and sodium hydroxide.

Products Sodium silicate solution (Na2SiO3) Sodium hydroxide (NaOH)

Company Sigma-Aldrich Fisher ScientificGrade Reagent Certified ACSComposition Na2O: 10%, SiO2: 27%, H2O: 63% NaOH Solid (≥97%)Density 1.39 g/ml 2.13 g/mlFormula (NaOH)x (Na2SiO3)yzH2O NaOH

Table 3Physical properties of PVA fiber.

Parameter Value

Fiber label RECS 15Diameter (μm) 40Length (mm) 12Young's modulus (GPa) 41Elongation (%) 6.7Density (g/cm3) 1.3Tensile strength (MPa) 1586

Table 4Mix proportions of EGC mixes.

Mix designation Fly ash Slag Activator PVA fiber

FA-0%S 1.0 – 0.27 0.02FA-10%S 0.9 0.1 0.27 0.02FA-20%S 0.8 0.2 0.27 0.02FA-30%S 0.7 0.3 0.27 0.02

Note: All numbers in the table are mass ratios of fly ash, except for the fibercontent, which is a volume fraction.

Y. Ling et al. Composites Part B 164 (2019) 747–757

748

2.4. Compressive strength

Compressive strength of the EGC mixes was tested according toASTM C109 [17]. The fresh EGC was placed and consolidated in twolayers in a set of three 50.8 mm cubic molds, and the specimens werethen further compacted using a vibrating table at a frequency of 3600vpm for 10 s. After casting, all samples were sealed to minimizemoisture loss and placed in an oven at 50 °C for 24 h. After the moldswere removed, the specimens were placed in sealed jars, which werethen placed in an oven at 50 °C until testing. For each of the EGC mixes,a set of three specimens was tested at 3, 7, and 28 days after casting.

2.5. Tensile strength

Uniaxial tension tests were conducted to evaluate the tensile be-havior of the slag blended, fly ash-based EGCs. The test setup is shownin Fig. 1. For each mix, two dog-bone EGC samples were cast, and theywere cured in the same environmental conditions as the specimensprepared for the above-mentioned compressive strength.

After 28 days of curing, the specimens were tested in uniaxial ten-sion under a displacement control condition using a mechanical testingsystem (MTS) testing machine. The displacement rate used was0.5 mm/min.

Right before the testing, the ends of the dog-bone specimens werepolished to facilitate gripping. The tested specimens were then placedin an alignment with the hydraulic grips of the MTS machine. The MTSmachine had a digital control panel and software to run the test andrecord the load automatically. Two linear variable differential trans-formers (LVDTs) were installed on each side of the specimen to measurethe displacements between two points on the specimen in a gaugelength of 110mm. Based on the recorded load and average displace-ment data, the tensile stress - strain curve of the tested specimen wasplotted.

According to previous studies on micromechanical design of con-ventional ECC [18], the tested composite must satisfy a stress-basedcondition in order to achieve a pseudo-strain hardening (PSH) beha-vior, accompanied by multiple fine cracks. Such a stress-based condi-tion, or criterion, can be expressed as

≤σ σfc 0 (1)

where σ0 is the maximum fiber bridging stress (i.e., ultimate tensilestrength of the composite) and σfc is the tensile first-crack strength ofthe composite.

Fig. 2 illustrates a typical stress (σ) - strain (ε) curve of a materialwith a strain hardening behavior, where stress increases with increasingstrain after the first cracking. According to the above-mentioned stress-based condition (Eq. (1)), multiple cracking could occur with an in-creasing load as long as σ0 exceedsσfc. As a result, the stress-strain curveof ECC has a shape similar to that of a ductile metal [1]. In accordancewith the stress condition for PSH behavior, Kanda and Li proposed aperformance index σ σ/ fc0 [19]. Theoretically, this index must exceedunity to achieve PSH behavior in a fiber-reinforced composite. Thehigher the performance index value, the greater possibility for saturatedmultiple cracking (or saturated PSH behavior) and the higher tensilestrain capacity the composite will have. In addition, as shown in Fig. 2,the area under the stress-strain curve up to failure can be derived as thetensile toughness of the tested composite, which indicates the totalenergy absorption capacity of the material. The tensile elasticity, theslope of elastic portion (σ ε/fc fc), of the tested composite could also bederived from the stress-strain curve.

2.6. Flexural bending test

A four-point flexural bending test was carried out on a slab

Fig. 1. Uniaxial tension test setup for GGC: (a) EGC specimen under tensile test; (b) schematic illustration of tensile test.

Fig. 2. Typical stress/strain curve of strain hardening composites.

Y. Ling et al. Composites Part B 164 (2019) 747–757

749

specimen to evaluate flexural bending behavior of the EGC compositeas illustrated in Fig. 3. For each EGC mix, two slabs with a dimension of711mm×108mm×13mm were cast, and the slabs were cured undera condition similar to the cube specimens for 28 days. To perform thetest, the tested slab was simply supported with a support span of457mm as illustrated in Fig. 3(b). Two flexural loads were appliedsymmetrically at 152mm from the supports using an MTS testing ma-chine. The loading span was one-third of the support span. Two LVDTswere installed at the bottom edges of the slab to monitor the mid-spandeflections of the slab. In accordance with Martin et al. [20] and Sarkeret al. [21], the loads were applied at a displacement control rate of0.2 cm/min until the slab failed.

The flexural stress in the tested slab was calculated according to Eq.(2):

=σ FLbd2 (2)

where σ is flexural stress; F is the load at the fracture point; L is thelength of the support span; b is the width of slab and d is thickness ofslab.

2.7. Pullout bond strength test

A direct tension pullout bond test was conducted to investigate thebond strength between EGC and an embedded steel. The test setup isillustrated in Fig. 4.

For each EGC mix, two cylinder specimens were prepared. Eachspecimen was made with a Փ152mm×L152 mm EGC cylinder and aՓ13mm×L609 mm smooth steel bar that was embedded in the centerof the EGC cylinder. The specimens were cast and cured in a similarway to the cube specimens used for compressive strength tests. Table 5lists the properties of the steel bar used. Before EGC casting, each steelbar was covered with a 76mm long polyvinyl chloride (PVC) tube sothat it would not bond with the top half of the EGC cylinder. As a result,only the bottom half of the EGC cylinder bounded with the embeddedsteel bar. To ensure a proper bond during EGC casting and avoid stressconcentration on EGC during pullout tests, the steel bar went through

the EGC cylinder with 19mm out of specimen bottom as shown inFig. 4(b). Thus, the effective bond length between EGC and steel barwas 76mm.

At the age of 28 days, pullout tests were conducted for all thespecimens. During the test, the top of steel bar was gripped on the MTSmachine. An alloy plate was placed on the top surface of EGC cylinderby four bolts tightened on four threaded rods to fix the EGC cylinder.The steel bar was then subjected to a tensile force that is transferred tothe EGC as tensile stresses throughout the bond between the EGC andthe steel bar. All pullout tests were performed under a displacementcontrol condition, and the displacement rate applied was 0.3 mm/minin accordance with Qian and Li [22].

For each EGC mix studied, an average result of two specimens wasderived. The bar slip was recorded by the two LVDTs installed on thetwo sides of cylinder until the tested specimen failed. In order tocompensate the displacement on the steel bar itself caused by tension, amicro-strain gauge was pre-installed on the steel bar surface. During thepullout test, the strain of the steel bar was recorded by a data logger.The bond stress was computed according to Eq. (3):

=τ P πL d/( )e b (3)

where τ is the bond stress; P is the load; Le is the contacted length of barin EGC; and db is the bar diameter. The slip of bar (s) was calculatedaccording to Eq. (4)

= − ×s d ε LL e (4)

where dL is the displacement of bar from LVDT; and ε is the microstrainof bar from the strain gauge.

2.8. Microstructure characterization

A microscopic study was carried out using an environmental scan-ning electron microscope (ESEM), the FEI Quanta 250 FE instrument.Attention was paid on the bond conditions between the geopolymermatrix and fibers of selected two specimens (FA-0%S and FA-20%S)that were previously ruptured during their tensile strength tests. Theobservation was performed using backscattered electron mode with alow vacuum of 80 Pa and high voltage of 15 Kv from 50 to 500 mag-nifications.

3. Results and discussion

3.1. Compressive strength

The compressive strength of each EGC mix with standard error barsis presented in Fig. 5. As seen in the figure, the basic fly ash-based EGC(FA-0%S) developed in the present study reached compressive strengthover 70MPa at the age of 28 days. When compared with mix FA-0%S,the EGC mix with 10% slag replacement for fly ash (FA-10%S) onlyincreased the 28 days compressive strength very slightly. However, theEGC mix with 20% slag replacement for fly ash (FA-20%S) exhibitedthe highest compressive strength among all EGC mixes studied at alltesting ages (102MPa at 28 days). Based on Palomo et al. [23] and Chi[24], this is likely due to the optimal production of aluminosilicatehydrate and C-S-H gels. However, further increase in slag replacementfor fly ash to 30% (FA-30%S) led to a decrease in compressive strength.This is also consistent with a previous study conducted by Wardhonoet al. [25], where they reported that excessive calcium contributed byslag in geopolymer could undergo a hydration reaction to form Ca(OH)2and cause expansion and cracking, thus decreasing strength. A similartrend of the compressive strength with different percentage of slag re-placement in a fly ash-based geopolymer was also recently reported byHassan and Ismail [26].

Fig. 3. Four-point flexural bending test setup for EGC: (a) EGC specimen underbending test; (b) schematic illustration of bending test.

Y. Ling et al. Composites Part B 164 (2019) 747–757

750

3.2. Tensile strength

Tensile stress-strain behaviors of the EGC mixes studied are pre-sented in Fig. 6. As it can be seen from the figure, all EGCs exhibitedstrain-hardening behavior, accompanied by multiple cracking. This isattributed to the bridging mechanism of the PVA fibers. The EGC mixwithout slag replacement (FA-0%S) had the lowest tensile strength(4.7 MPa), while the EGC mixes with 10–30% slag replacement for flyash all had higher tensile strength (5.7–6.8MPa). The highest tensilestrength was 6.8 MPa for EGC mix with 20% slag replacement (FA-20%S), while it was below 4MPa for fly ash-based EGCs and around 5MPafor slag-fly ash blended EGCs reported by previous researches [7,27].As mentioned previously, the enhanced strength was a result of thesynergy of the optimal amount of slag and fly ash in the geopolymersystem that produced C-S-H and C-A-S-H gel from slag hydration andpolymerization combined with the C-A-S-H gel yielded from fly ashpolymerization [11]. Such a combination enhanced the bond betweenthe geopolymer matrix-PVA fibers used.

Table 6 lists the first-crack strength (σfc), tensile elasticity (σ ε/fc fc),ultimate tensile strength (σfc), tensile strain capacity (ε max), toughness

and stress index (σ σ/ fc0 ) of each mix.The following observations can be made from the Table:

(1) The ultimate tensile strengths (σ0) of all the EGC mixes were sig-nificantly higher than the first crack strength (σfc), indicating thatall the mixes possessed strain-hardening behavior.

(2) All slag blended EGC mixes had higher first crack strength, highertensile elasticity, and higher ultimate tensile strength but lowertensile strain capacity, lower toughness, and lower stress index thanthe EGC mix without slag (Mix FA-0%S).

(3) Among all mixes studied, Mix FA-20%S (20% slag replacement)exhibited the highest first-crack strength and highest ultimate

Fig. 4. Pullout bond strength test setup for EGC: (a) EGC specimen under pullout test; (b) schematic illustration of pullout test.

Table 5Physical properties of used steel bar.

Diameter(mm)

Length(mm)

Yield strength(MPa)

Ultimatestrength (MPa)

Elongation (%)

12.6 609.6 531 680 16

Fig. 5. Compressive strength of EGCs with different amounts of slag.

Fig. 6. Tensile stress-strain responses of EGCs: (a) full stress-strain curves; (b)Initial stress-strain curves: strain from 0 to 0.5% in (a).

Y. Ling et al. Composites Part B 164 (2019) 747–757

751

tensile strength. The first-crack strength of FA-30%S, FA-10%S, andFA-0%S were 16.1%, 32.3%, and 45.2% lower, respectively. Theultimate tensile strength of FA-30%S, FA-10%S, and FA-0%S were10.3%, 25.0%, and 30.9% lower, respectively.

(4) The tensile elasticity increased while tensile strain capacity,toughness, and stress index decreased with increasing slag contentin the EGC mixes. This suggests that the EGC composite becamemore brittle as more slag was used to replace fly ash.

The first crack strength (σfc) reflects the properties of the geopo-lymer matrix. Increased σfcvalue resulting from a slag replacement isonce again due to the slag modification of the chemistry of the geo-polymer gels and densification of the geoplymer microstructure [9*].The ultimate tensile strength of a composite material is closely relatedto its interfacial properties. In other words, the chemical bonding en-ergy and the frictional bond strength of FA-20%S might have increasedmore than the cracking strength when compared with the other EGCcomposites studied, resulting in higher fiber-bridging strength [27].The tensile elasticity increased with the increment of slag content from1.58 GPa for FA-20%S to 2.98 GPa for FA-30%S was attributed to the C-S-H and C-A-S-H gel formation resulting from the slag hydration andpolymerization, leading to a reduced porosity or increased density [28].One of the underlying reasons for considerable reduction in the tensilestrain capacities with slag EGCs lies in their different stress indices. Thestress index of FA-0%S was the highest among the EGCs, accompaniedby the highest tensile strain capacity. In addition, the stress indices ofFA-10%S, FA-20%S, and FA-30%S were 12.3%, 13.8%, and 15.2%lower than that of FA-0%S, respectively. As mentioned previously, thehigher the stress index value, the greater possibility for saturated PSHbehavior the composite will have, which leads to a higher tensile straincapacity or higher ductility of the composite. Table 6 shows that theductility of the fly ash-based EGC reduced as the slag content in the EGCincreased. The toughness results also revealed that as slag content in-creased from 0% to 30%, toughness of the fly ash-based EGC reducedfrom 12.6 J/cm3 to 3.7 J/cm3, indicating that the energy absorption ofthe EGC decreased. This trend corresponds to that of tensile strain ca-pacity. In general, the slag replacement for fly ash had led a low duc-tility and toughness but high ultimate tensile strength EGC.

The multiple cracking pattern of each EGC mix studied is presentedin Fig. 7. After unloading, a clear trace of all visible cracks was obtainedfrom the wetted specimen surfaces. As it is seen from the figure, the mixwithout slag replacement (FA-0%S) exhibited massive and uniformlydistributed microcracks with tightly controlled crack widths (i.e., asaturated cracking behavior), which corresponds to its significantlyhigh tensile strain capacity and ductility. However, as slag replacementlevel increased from 10% to 30%, the crack spacing became larger, andthe crack distribution became less uniform (i.e., an unsaturatedcracking behavior), which confirms that the EGC tended to be morebrittle as slag content increased.

3.3. Flexural bending strength

The stress-deflection curves of all EGC mixes under four pointbending tests are shown in Fig. 8. As it can be observed from the figure,the EGC with 20% and 30% slag replacement demonstrated a relativelybrittle behavior - the stress decreased rapidly when the mid-span

deflection reached about 10mm. However, for the FA-0%S and FA-10%S mixes, the stress-deflection curves were quite flat and the maximummid-span deflection reached 39.81mm and 35.32mm, respectively.These results suggest that the pseudo-hardening responses were largelydependent on the slag content.

Table 7 summaries the modulus of rupture (MOR), deflection at firstcrack (Dfc), deflection at failure (Dfl) and ductility index (D D/fl fc) of theEGC mixes studied. The ductility index (D D/fl fc) is the ratio of mid-spandeflection at failure to that at the first crack [29]. The greater theductility index value, the higher ductility the tested material will have.

It can be seen from Table 6 that the flexural strength or MOR of theEGC increased with increasing slag content. Among the three percen-tages of slag content evaluated, the mix with 20% slag (FA-20%S) hadthe highest MOR, while the mix with 0% slag had the lowest MOR. Thetrend is similar to that of compressive strength test results.

Table 6 also shows that the mid-span deflection at the first crack(Dfc) increased with MOR. The maximum mid-span deflection at failure(Dfl) reached 39.81mm for the mix without slag (FA-0%S) but it de-clined as slag content increased. When slag content increased from 10%to 20%, the mid-span deflection of the specimens at failure decreased68.8%.

Regarding ductility index, Mix FA-0%S displayed the highest value.The ductility index value was dramatically reduced when slag contentincreased from 10% to 20%. Mix FA-20%S showed the lowest ductilityindex value among all four mixes studied. This result repeatedly sug-gests that Mix FA-20%S is an optimal EGC mix for strength but not thebest for ductility. Slag replacement for fly ash reduced the ductility ofthe EGC. Such a reduction in ductility is also attributed to the densifiedmicrostructure provided by the slag hydration and polymerization asdiscussed previously.

Fig. 9 shows the crack patterns on the wetted surfaces of EGC spe-cimens after flexural tests. Multiple cracks, which were uniformly dis-tributed in the area with pure bending moment, were observed for allEGC mixes studied. Similar to what was observed from the specimenssubjected to tension tests, the number of cracks reduced and theirspacing increased in the specimens subjected to flexural tests as slagcontent increased. These crack patterns once again signify that the EGCmixes had a very good deflection hardening property.

3.4. Pullout bond strength

One of advantages of geopolymer over Portland cement-based ma-terials is its bound with steel reinforcement because of the strongeradhesion of the geopolymer gel to steel and the higher splitting tensilestrength [30]. The bond strength between conventional geopolymerand steel reinforcement has been studied by various investigators[31–34]. The factors that affect the bond strength include the quality(e.g., gel compositions and strength) of the surrounding geopolymer,the bar size of steel reinforcement, the embedded length of the steel bar,the thickness of the rebar cover, etc. However, the present study focuseson the effect of slag replacement on the bond strength of EGC only.Fig. 10 shows the bond stress-slip relationship of the EGC mixes studied.It can be seen from the figure that all EGC mixes exhibited a pulloutfailure mode with a post-peak slip behavior. That is, after the peak load,the pullout stress had a little rapid drop but still maintained constant fora significant time until the complete failure. These little drops in the

Table 6Tensile strength test results of EGCs.

Mix σfc(MPa) Tensile elasticity, σ ε/fc fc (GPa) σ0 (MPa) Tensile strain capacity, (%) Toughness, (J/cm3) Stress index, (σ σ/ fc0 )

FA-0%S 3.4 1.18 4.7 3.14 12.6 1.38FA-10%S 4.2 1.41 5.1 1.62 6.9 1.21FA-20%S 5.7 1.58 6.8 1.04 5.5 1.19FA-30%S 5.2 2.98 6.1 0.74 3.7 1.17

Y. Ling et al. Composites Part B 164 (2019) 747–757

752

pullout stress represent a low bond stress reduction due to slip, evi-denced by a high ductility measurement after the peak load. Such apost-peak behavior implies enhanced bond strength between the EGCand steel bar due to the fiber confinement, which also resulted in animproved ductility of the steel reinforced EGC.

As the failure of geopolymer in a pull-out test is closely related tothe strength of the surrounding geopolymer [31], Table 8 lists the ul-timate bond strength (τu) and ultimate slip (su), along with compressivestrength ( fcu), tensile strength (σ0), and flexural strength (MOR) of theEGCs studied. It shows that all EGC mixes with slag replacement for flyash had higher bond strength than the mix with no slag (FA-0%S). Thebond strength between the mix with 20% slag content (FA-20%S) andthe steel bar is the highest one, while the bond strength between the

mix with no slag (FA-20%S) and steel bar FA-0%S is the lowest one, thetrend of which is also coherent with that of the compressive strength ofthese mixes. The bond strength improvement provided by slag re-placement for fly ash in these EGC mixes was associated with the de-creased porosity in the EGC-steel interfacial zone, which could in turnprovide an increased EGC-steel contact surface and result in a higherfrictional bond [35]. The decrease in porosity provided by slag re-placement in EGC was also observed (Fig. 12), which will be discussedlater. Regardless of the improved bond strength, the ultimate slip of theEGCs reduced significantly as the slag content increased. This alsoevidenced that the slag replacement for fly ash could decrease ductilityof the EGC composite.

Fig. 11 illustrates the relationships between the rebar-EGC bondstrength and other strengths (compressive, tensile, and flexure strength)of the surrounding EGC. The figure indicates the relationships betweenthe bond strength and tensile/flexure strengths are better than thatbetween the bond strength and compressive strength of the EGC. Thismanifested that most of the tested samples were failed under thesplitting failure modes.

In the reinforced concrete design practice, bond strength is oftenpredicted from compressive strength of concrete. Based on BS 8110[36], the relationship between bond strength (τu) and compressivestrength ( fcu) of concrete can be expressed as

=τ β fu cu (5)

where τu is bond strength; β is a coefficient dependent on specimen; andfcu is compressive strength.

Based on Eq. (5), the following equation (Eq. (6)) was proposed topredict bond strength between the slag blended, fly ash-based EGCs andsteel bar in the present study:

=<

≥τ

f f MPa

f f MPa{0.4 , 80

0.6 , 80u

cu cu

cu cu (6)

The comparison of the predicted bond strength and experimentallymeasured bond strength are presented in Table 9. It can be observedfrom the table that the predicted bond strength obtained from theproposed equation (Eq. (6)) agrees well with the values obtained fromexperiments. That is, for the slag blended, fly ash-based EGCs, the BS

Fig. 7. Multiple cracking behavior of EGCs after tension tests: (a) FA-0%S; (b) FA-10%S; (c) FA-20%S; (d) FA-30%S.

Fig. 8. Flexural stress-deflection responses of EGCs.

Table 7Flexural strength test results of EGCs.

Mix MOR (MPa) Dfc (mm) Dfl (mm) Ductility index

FA-0%S 9.3 1.26 39.81 31.6FA-10%S 10.4 1.41 35.32 25.0FA-20%S 18.5 2.07 11.03 5.3FA-30%S 15.3 1.60 9.46 5.9

Y. Ling et al. Composites Part B 164 (2019) 747–757

753

equation is still valid asβ =0.4 for EGCs with compressive strength lessthan 80MPa or β =0.6 for EGCs with compressive strength greaterthan or equal to 80MPa.

3.5. Microstructure

A microscopic study was conducted to provide additional informa-tion on the explanation of the mechanical properties of EGCs. Thesamples examined under ESEM were small pieces obtained from thespecimens after their tensile strength tests. The examination was fo-cused on the bond between PVA fibers and the geopolymer matrix interms of slag replacement as well as the fiber distribution in the EGCs.

Fig. 12 presents ESEM-BEI of FA-0%S and FA-20%S samples at amagnification of 50× . It can be seen from the figure that although carehad been taken, some agglomerations of PVA fiber still occurred in theboth EGC samples. The areas with fiber agglomerations were commonlyassociated with entrapped voids and cracks. Such weak spots couldsignificantly weaken mechanical properties of the EGCs. The techniquefor fiber distributions in EGCs shall be further studied, and the methodand procedures for fiber mixing shall be further improved.

In addition to the fiber distribution, it can also be observed from thefigure that some fibers in EGC matrixes were pulled out during tensiontests, evidenced by the partially embed fibers and the well-visibleempty voids that were occupied by fibers before the tension tests.Previous research has indicated that while many fibers were pulled outduring tension, a considerable amount of fibers could be rupturedduring the tensile tests, thus significantly increasing the tensile strengthand stress propagation of the composites [37]. However, fracture offibers was not clearly identified in the samples examined in the presentstudy.

Moreover, it shall be noted that fewer voids were captured in the

Fig. 9. Crack patterns of EGCs after flexural tests: (a) FA-0%S; (b) FA-10%S; (c) FA-20%S; (d) FA-30%S.

Fig. 10. Bond stress-slip relationships for EGCs.

Table 8Pullout bond strength between EGCs and steel rebar.

Mix τu (MPa) Su (mm) ƒcu (MPa) σo (MPa) MOR (MPa)

FA-0%S 3.5 16.3 72.6 4.7 9.3FA-10%S 3.7 11.9 73.9 5.1 10.4FA-20%S 6.2 9.4 102.3 6.8 18.5FA-30%S 5.9 7.4 82.3 6.1 15.3

Fig. 11. Relationships between bond strength and other strengths of EGC.

Table 9Experimental (τu) and predicted (τp) bond strength between EGCs and steelrebar.

Mix τu, (MPa) τp, (MPa) τ τ/u p

FA-0%S 3.5 3.41 1.03FA-10%S 3.7 3.44 1.08FA-20%S 6.2 6.07 1.02FA-30%S 5.9 5.44 1.08

Y. Ling et al. Composites Part B 164 (2019) 747–757

754

FA-20%S sample than in the FA-0%S sample. As mentioned before, theimproved density resulting from the optimal slag hydration and poly-merization was one of the major contributors to the superior strengthsof the FA-20%S mix.

Fig. 13 provides more detailed information on the fiber-EGC matrixbond. As seen in Fig. 13 (a) and (b), relatively cleaning, smooth surfacesof fibers were observed in the FA-0%S sample, which indicates that thebond between fiber and geopolymer might not be very strong, and theweak mechanical properties of the FA-0%S mix was due to the weakfiber bridging effect. Differently, in Fig. 12 (c) and (d), much moregeopolymerization products can be observed on the fiber surfaces of theFA-20%S sample, indicating a good cohesion between the fiber and theEGC matrix. Such well-bounded geopolymerization products on thefiber surfaces could help increase the friction between the fiber and theEGC matrix during the fiber pullout, thus enhancing the bond strengthbetween the PVA fiber and geopolymer as well as other strength

properties of the EGC composites.It shall also be noted that there are some thread-like features clearly

seen on the surfaces of the fibers in the FA-20%S sample (Fig. 13(d)),which was also frequently observed in other images of the same sampleand was possibly a sign of the beginning of the fiber fracture. Conse-quently, one can infer that use of 20% slag replacement for fly ashreplacement had converted the failure mode of the fly ash-based EGC(FA-0%S) composite from a fiber pullout dominated failure to a fiberpullout/fracture failure due to its improvement in the fiber-geopo-lymer. Single fiber pullout tests may be conducted in the future to helpverify the fiber fracture behavior in EGCs.

4. Conclusions

The effects of slag replacement on mechanical properties of low-calcium (Class F) fly ash-based engineered geopolymer composite

Fig. 12. ESEM images at the magnification of 50× : (a)FA-0%S; (b)FA-20%S.

Fig. 13. SEM images of FA-0%S and FA-20%S at different magnifications: (a) FA-0%S (150× ); (b) FA-0%S in area A (500× ); (c) FA-20%S (150× ); (d) FA-20%Sin area A (500× ).

Y. Ling et al. Composites Part B 164 (2019) 747–757

755

(EGC) were investigated in the present study. The following conclusionscan be drawn from the experimental results:

(1) Using a low liquid-to-binder (fly ash and slag) ratio of 0.27, a SiO2/Na2O mole ratio of 1.0, solute (NaOH and Na2SiO3) mass con-centration of 30%, and polyvinyl alcohol (PVA) fibers of 2% (byvolume), a fly ash-based EGC (FA-0%S) was developed in the pre-sent study. This EGC mix reached 73MPa at the age of 28 days anddisplayed significant strain and deflection hardening behaviorunder a uniaxial tension and flexure, respectively.

(2) Compressive strength of the fly ash-based EGC can be further en-hanced by using slag (GGBFS) to replace fly ash. Although all slagcontent used (10–30% by weight of the binder) had improved theEGC strength, the highest strength was achieved by the mix with20% slag replacement (FA-20%S), which had compressive strengthof 102MPa at 28 days.

(3) All EGCs studied (with and without slag) exhibited strain-hardeningbehavior in tension. All slag blended EGC mixes had higher firstcrack strength, higher tensile elasticity, and higher ultimate tensilestrength but lower tensile strain capacity, lower toughness, andlower stress index than the EGC mix without slag (FA-0%S). Theresults suggested that the EGC composite became stronger but morebrittle as more slag was used to replace fly ash. This inference wasalso evidenced by the flexural and pullout test results.

(4) A mix with 20% slag replacement (FA-20%S) appeared to be themost optimal EGC mix for all strengths tested (compressive, flex-ural, and bond strengths), however, it is not the best mix for duc-tility related properties (e.g., tensile strain capacity, maximumflexural deformation, ultimate slip, and toughness).

(5) Multiple fine cracks were observed in all EGC specimens undertension and flexure. The number of cracks reduced and their spa-cing increased with increasing slag in the specimens. These crackpatterns also signified that the fly ash-based EGC (FA-0%S) hadvery good strain/deflection hardening properties and showed ex-cellent ductility, but it became more brittle as the slag content (orslag replacement level) increased.

(6) An excellent correlation was found between compressive strengthand bond strength of the EGCs studied. Based on the BS predictionequation for bond strength, a new equation (Eq. (6)) was proposedto assess the EGC's bond strength based on their compressivestrength. The predicted strength had a good agreement with ex-perimental results, indicating that the BS equation is still suitablefor the bond strength prediction of EGCs.

(7) Results from the microscopic study indicated that 20% slag re-placement for fly ash reduced pores and densified the EGC matrix,and it improved the adhesion of geopolymization products withPVA fibers, which was primarily responsible for strength enhance-ment of the EGC. Although fiber fracture might have happened,fiber pullout was commonly observed in both samples with 0% and20% slag replacement for fly ash. Fiber agglomerations were ob-served in these samples. Further study is necessary to improve thefiber distribution and fiber-steel bar bond.

Declarations of interest

None.

Acknowledgements

The present study is a part of a research project, InitialCharacterization of Geopolymer Based UHPC Material Properties,sponsored by the Midwest Transportation Center, Iowa State University(ISU). Significant help was provided by Dr. Chuanqing Fu during hisvisit at ISU and by Mr. Douglas Wood at ISU on the mechanical tests.Prompt assistance was received from Mr. Warren Straszheim on theESEM studies. Valuable inputs were given by Mr. Zhuo Tang, University

of Technology Sydney, on discussions of the test results.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.compositesb.2019.01.092.

References

[1] Li VC, Kanda T. Engineered cementitious composites for structural applications. JMater Civ Eng 1998;10(2):66–9.

[2] Kong HJ, Bike SG, Li VC. Development of a self-consolidating engineered ce-mentitious composite employing electrosteric dispersion/stabilization. CementConcr Compos 2003;25(3):301–9.

[3] Nematollahi B, Sanjayan J, Shaikh F. Tensile strain hardening behavior of PVAfiber-reinforced engineered geopolymer composite. J Mater Civ Eng2005;27(10):04015001.

[4] Wang S, Li VC. Engineered cementitious composites with high-volume fly ash. ACIMater J 2007;104(3):233–41.

[5] Shi C, Palomo A. New cements for the 21st century: the pursuit of an alternative toPortland cement. Cement Concr Res 2011;41(7):750–63.

[6] Juenger M, Winnefeld F, Provis J, Ideker J. Advances in alternative cementitiousbinders. Cement Concr Res 2011;41(12):1232–43.

[7] Ohno M, Li VC. A feasibility study of strain hardening fiber reinforced fly ash-basedgeopolymer composites. Constr Build Mater 2014;57:163–8.

[8] Nath SK, Maitra S, Mukherjee S, Kumar S. Microstructural and morphologicalevolution of fly ash based geopolymers. Constr Build Mater 2016;111:758–65.

[9] Li Z, Liu S. Influence of slag as additive on compressive strength of fly-ash-basedgeopolymer. J Mater Civ Eng 2007;19(6):470–4.

[10] Chang J. A study on the setting characteristics of sodium silicate-activated slagpastes. Cement Concr Res 2003;33(7):1005–11.

[11] Nath SK, Kumar S. Influence of iron making slags on strength and microstructure offly ash geopolymer. Constr Build Mater 2013;38:924–30.

[12] Deb P, Nath P, Sarker P. The effects of ground granulated blast-furnace slagblending with fly ash and activator content on the workability and strength prop-erties of geopolymer concrete cured at ambient temperature. Mater Des2007;62:32–9.

[13] Nath SK, Kumar S. Influence of granulated silico-manganese slag on compressivestrength and microstructure of ambient cured alkali-activated fly ash binder. WasteBiomass Valor 2018https://doi.org/10.1007/s12649-018-0213-1.

[14] Nath SK. Geopolymerization behavior of ferrochrome slag and fly ash blends.Constr Build Mater 2018;181:487–94.

[15] Ling Y. Proportion and performance evaluation of fly ash-based geopolymer and itsapplication in engineered composites Graduate Theses and Dissertations vol 16398.2018.

[16] ASTM C618. Standard specification for coal fly ash and raw or calcined naturalpozzolan for use in concrete. Philadelphia: ASTM International; 2017.

[17] ASTM C109/C109M. Standard test method for compressive strength of hydrauliccement mortars. West Conshohocken (PA): ASTM International; 2016.

[18] Kanda T, Li VC. New micromechanics design theory for pseudo-strain hardeningcementitious composite. J Eng Mech 1999;125(4):373–81.

[19] Kanda T, Li VC. Practical design criteria for saturated pseudo strain hardeningbehavior in ECC. J Adv Concr Technol 2006;4(1):59–72.

[20] Martin J, Stanton J, Mitra N, Lowes LN. Experimental testing to determine concretefracture energy using simple laboratory test setup. ACI Mater J2007;104(6):575–84.

[21] Sarker PK, Haque R, Ramgolam KV. Fracture behavior of heat cured fly-ash-basedgeopolymer concrete. Mater Des 2013;44:580–6.

[22] Qian S, Li VC. Headed anchor/engineered cementitious composites (ECC) pulloutbehavior. J Adv Concr Technol 2001;9(3):339–51.

[23] Palomo A, Grutzeck MW, Blanco MT. Alkali-activated fly ashes: a cement for thefuture. Cement Concr Res 1999;29(8):1323–9.

[24] Chi M. Effects of dosage of alkali-activated solution and curing conditions on theproperties and durability of alkali-activated slag concrete. Constr Build Mater2012;35:240–5.

[25] Wardhono A, Law DW, Sutikno, Dani H. The effect of slag addition on strengthdevelopment of Class C fly ash geopolymer concrete at normal temperature. AIPConference Proceedings 1887 2017; 020030. https://doi.org/10.1063/1.5003513.

[26] Hassan H, Ismail N. Effect of process parameters on the performance of fly ash/GGBS blended geopolymer composites. J Sustainable Cement-Based Mater2018;7(2):122–40.

[27] Lee BY, Cho CG, Lim HJ, Song JK, Yang KH, Li VC. Strain hardening fiber reinforcedalkali-activated mortar — a feasibility study. Constr Build Mater 2012;37:15–20.

[28] Criado M, Aperador W, Sobrados I. Microstructural and mechanical properties ofalkali activated colombian raw materials. Materials 2016;9(3):158.

[29] Jaejer LG, Mufti AA, Tadros G. The concept of the overall performance factor inrectangular- section reinforced concrete beams. Proceedings of the 3rd interna-tional symposium on non-metallic (FRP) reinforcement for concrete structure(FRPRCS-3). Sapporo, Japan: Japan Concrete Institute; 1997. p. 551–8.

[30] Sarker PK. Bond strength of reinforcing steel embedded in fly ash-based geopolymerconcrete. Mater Struct 2011;44:1021.

[31] Al-Azzawi M, Yu T, Hadi MNS. Factors affecting the bond strength between the flyash-based geopolymer concrete and steel reinforcement. Structure 2018;14:262–72.

Y. Ling et al. Composites Part B 164 (2019) 747–757

756

[32] Castel A, Foster SJ. Bond strength between blended slag and Class F fly ash geo-polymer concrete with steel reinforcement. Cement Concr Res 2015;72:48–53.

[33] Topark-Ngarm P, Chindaprasirt P, Sata V. Setting time, strength, and bond of highcalcium fly ash geopolymer concrete. J Mater Civ Eng 2015;27(7):04014198.

[34] Dahou Z, Castel A, Noushini A. Prediction of the steel-concrete bond strength fromthe compressive strength of Portland cement and geopolymer concretes. ConstrBuild Mater 2016;119:329–42.

[35] Kim J, Kim J, Ha G, Kim Y. Tensile and fiber dispersion performance of ECC (en-gineered cementitious composites) produced with ground granulated blast furnaceslag. Cement Concr Res 2007;37(7):1096–105.

[36] BS 8110-1. Structural use of concrete - Part 1: code of practice for design andconstruction. 1997. p. 89–90.

[37] Wang S, Li VC. Polyvinyl alcohol fiber reinforced engineered cementitious com-posites: material design and performances. RILEM SARL 2006:65–73.

Y. Ling et al. Composites Part B 164 (2019) 747–757

757