Construction and Building Materials 248 (2020) 118558

Contents lists available at ScienceDirect

Construction and Building Materials

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

On the mechanical performance of K- and Na-based strain-hardening

geopolymer composites (SHGC) reinforced with PVA fibers
Ana Carolina Constâncio Trindade a,⇑, Iurie Curosu b, Marco Liebscher b, Viktor Mechtcherine b,
Flávio de Andrade Silva a
a
Department of Civil and Environmental Engineering, Pontifícia Universidade Católica do Rio de Janeiro (PUC-Rio), Rio de Janeiro, RJ 22451-900, Brazil
b
TU Dresden, Institute of Construction Materials, 01062 Dresden, Germany

h i g h l i g h t s

Both K- and Na-based formulations are used for SHGC manufacturing.

Na-based SHGC achieves improved mechanical responses, with an enhanced strain capacity.

K-based SHGC do not present a typical strain-hardening curve when in presence of fine-grained sand.

When compared to typical SHCC, the developed SHGC shows a mechanically improved cracking behavior.

SHGC is expected to be used in a wide range of applications.

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

Article history: Strain-hardening cementitious composites (SHCC) are known by their high deformation capacity due to a
Received 19 December 2019 multiple cracking tolerance under increasing tensile loading. While SHCC have been well investigated,
Received in revised form 18 February 2020 the design of geopolymer (GP) composites in the same context is not sufficiently understood, mainly
Accepted 25 February 2020
due to varying precursor materials used in previous studies. In this work, two metakaolin-based GP mix-
tures were produced, using K and Na as alkali sources. The incorporation of fine-grained sand was exten-
sively investigated to evaluate its compatibility with both strain-hardening geopolymer composites
Keywords:
(SHGC) reinforced with short polyvinyl-alcohol (PVA) fibers. Their mechanical properties were analyzed
Geopolymer
Fiber reinforcement
by means of compression, bending, and tension tests. Digital Image Correlation (DIC) was used to evalu-
PVA fiber ate the fracture mechanisms under uniaxial tension. Particle size distribution (PSD), X-ray diffraction
Strain-hardening, ductility (XRD), Thermogravimetric analysis (TGA) and mercury intrusion porosimetry (MIP) were conducted.
Aggregates Additionally, single-fiber pullout tests with subsequent Environmental Scanning Electron Microscopy
Pullout test (ESEM) were carried out. NaGP composites, reinforced with fine-grained sand, yielded improved mechan-
ical responses, reaching a strain capacity of approximately 4.5%, with enhanced multiple cracking, as
opposed to the lower performance observed for KGP composites, that did not present a typical strain-
hardening curve when in presence of fine-grained sand, but a strain-softening one. When compared to
typical SHCC responses, Na-based SHGC showed a mechanically improved multiple cracking behavior,
indicating the potential of this material for a wide range of applications, such as dynamic loading and
requirements of long-term durability.
Ó 2020 Elsevier Ltd. All rights reserved.

1. Introduction capacity under tensile loading. SHCC is developed based on theo-

ries of micromechanical and fracture mechanics [1–4]. Such com-
Strain-hardening cement-based composites (SHCC) [1] also posites are commonly obtained by incorporating moderate
called engineered cementitious composites (ECC) [2], represent a amounts (1–3%) of short, synthetic micro-fibers into purposefully
class of materials that were introduced in the early 90s [2] and designed, fine-grained cementitious matrices [2,4]. With their
which yield pronounced multiple cracking and hence, high strain favorable mechanical properties, such as tensile ductility, excellent
crack control, damage tolerance, and high energy dissipation
capacity, SHCC is suitable as the chief material both for new con-
⇑ Corresponding author.
struction and for retrofitting applications. In both cases, the use
E-mail address: [email protected] (A.C. Constâncio Trindade).

https://doi.org/10.1016/j.conbuildmat.2020.118558
0950-0618/Ó 2020 Elsevier Ltd. All rights reserved.
2 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558

of SHCC enhances the durability and mechanical response of comprehensive analysis of the matrix incorporated with sand as
structures exposed to severe loading and environmental conditions aggregate [38], optimal volume fractions of PE and PVA reinforce-
[5–12]. ments [37], and by considering micromechanics-based material
Although the benefits of SHCC are evident [1–11], it is impor- design [3,4], fly-ash-based SHGCs were obtained with strain capac-
tant to address their sustainability when choosing this type of ities up to 4.5% [34]. Batista et al. [39] evaluated the gradual
material for large-scale applications [12] since SHCC contain a rel- replacement of metakaolin by silica fume as a raw material in a
atively high amount of cement, due to strict limitations regarding GP also containing sand as fine aggregate. However, despite reach-
aggregates’ content [13]. These issues are associated with high CO2 ing reasonable tensile strength values with this substitution (on
emissions [12,14] as well as water demand, which is typical for the order of 4.5 MPa), the strain capacity of the composite
cement-based binders [15]. Thus, the material design concept remained in the range of 1%, possibly resulting from an enhanced
[13] should be extended to use feasible, more sustainable alterna- matrix toughness due to aggregate incorporation. It is interesting
tives to cement, when necessary, but still delivering equivalent to notice a lack of deep understanding of the effects of aggregate
mechanical performance [12]. The so-called geopolymers [16] addition on varied SHGC formulations, specially metakaolin-
appear to be such an alternative [12]. based. Also, there is a need for a proper correlation with SHCC per-
Initial studies on geopolymers date back to the 1930s, where formances under similar circumstances, since both fiber-matrix
Kuhl [15] evaluated the reaction between slag, a form of industrial interfaces are expected to play a different role in the composites
waste, and potassium hydroxide (KOH). In the 1940s, Purdon [17] strengthening mechanisms.
presented considerations on mixing the aforementioned residue Therefore, this study aims at a comprehensive evaluation of
with sodium-based solutions (NaOH). In 1967, Glukhovsky [18] metakaolin-based geopolymers with respect to their use as a SHGC
provided complete reports describing the production of alumi- material. Additionally, the effect of distinct alkali solutions (Na-
nosilicate binders with low calcium content, designating them as and K-based), the use, or not, of fine-grained aggregates into both
‘‘alkaline binding systems”. The term ‘‘geopolymers” was first SHGC materials reinforced with PVA fibers and its effect on the
introduced in the early 70s by Davidovits [16], who created a bin- composites’ cracking mechanisms are extensively analyzed. The
der based on aluminosilicate materials in order to enhance the resulting mechanical performance characteristics are discussed
thermal resistance of structural elements [19]. His team developed not only against the background of fracture mechanics theories
a fabrication method based on geosynthesis to obtain high- (used to validate the proposed SHGC design), but also with respect
performance materials by the reaction of pure aluminosilicates in to image observations (DIC and ESEM), analytics on the
a stable alkaline environment [16,20]. The polymer network con- microstructures (XRD, TGA, and MIP), as well as their behavior in
sists of SiO4 and AlO4 tetrahedra linked alternately in a range from the fresh state. Additionally, an efficiency comparison with SHCC
amorphous to semi-crystalline [21], requiring positive ions to bal- materials is made, showing the SHGC full potential.
ance the negativity of Al3+ in IV-fold coordination [16,21]. Its
empirical formula can be described as M2OAl2O3xSiO211H2O
(where M = Na, K, Cs; and x represents the Si/Al ratio used) [21]. 2. Experimental program
Although generally specified as a subclass of alkali-activated mate-
rials due to their similarities, geopolymers require distinct mixing 2.1. Materials
and reaction processes [21–24].
The reaction and hardening mechanisms, as well as the produc- Two GP mixtures were produced through the combination of an
tion particularities of geopolymers, aside from the benefits of their aluminosilicate source (metakaolin, MK) and an alkali-based solu-
sustainability, may present appealing advantages in their use as an tion (Na- or K-based water glass, WG), representing a final compo-
alternative to the common cementitious matrices. Such benefits sition of X 2 O  Al2 O3  4SiO2  11H2 O; where X is replaced by Na or K
range among fast setting time [22,25], enhanced thermal resis- elements. Sodium (Na) and potassium (K) hydroxide in pellets
tance [19,20,26], high chemical [26,27] and long-term [28] durabil- (Sigma-AldrichÒ, reagent grade > 90%) were dissolved in deionized
ity, low Young’s modulus (positive in repair applications [29]) with water, where hydrophilic fumed silica (CAB-O-SILÒ M-5) was
low variation determined mainly by its characteristic microstruc- added and mixed for 24 h using a magnetic stirrer, forming the
ture and aluminosilicate [28,30], and low porosity [31]. The pri- stable water-glass solution (X 2 O  2SiO2  11H2 O). Metamax MK
mary source of alumina and silica influences the mechanical (Al2 O3  2SiO2 ) from BASF (Ludwigshafen, Germany) was used as
properties of the final material since its chemical composition the aluminosilicate source due to its high level of purity and small
impacts the amount of alkali solution required to stabilize the reac- particle size, enabling adequate reactivity, as presented and dis-
tion [26,30]. Hence, geopolymers based on fly ashes and metakao- cussed in previous studies [22,40]. The MK chemical composition
lin present very distinct microstructures and, consequently, is presented in Table 1.
different properties and behaviors [22,26,30]. Thus, the choice of An LS 13-320 Beckman Coulter apparatus was used to verify the
raw materials must be considered depending on their availability particle size distribution of metakaolin. A small amount corre-
in the region of interest and particular applications [31]. sponding to 0.5 g of the material was prior mixed with isopropanol
In addition to extensive studies on plain geopolymer matrices, in a magnetic stirrer and then placed inside the device. Fig. 1 con-
various composites have been investigated containing: firms the small particle size range of Metamax MK, between 0.5 mm
and 15 mm.

sand and chamotte-based aggregates [22,27,28]; The WG-to-MK ratios obtained from the chemical formulations

synthetic (carbon) [32], mineral (basalt) [33] and natural fabrics showed previously are 1.71 and 1.85 for NaGP and KGP, respec-
(jute, curauá, sisal) [25]; tively. Fine-grained quartz sand, with a maximum diameter of

and short polymer micro-fibers [34,35]. 0.2 mm and density equal to 2.62 g/cm3, was used in a 50 wt%
(to the total MK amount) in order to evaluate its strain-
The latter approach was explored by Ohno et al. [34,36] and hardening effectiveness in various fiber-matrix combinations, as
Nematollahi et al. [35,37] in studies on the production of fly ash- presented in Table 2.
based engineered geopolymer composites (EGC) and strain- PVA short fibers produced by Kuraray under the brand name
hardening geopolymer composites (SHGC), respectively. Note KuralonÒ K-II REC 15 were used in a 2% content by volume for both
that both names stand for the same material group. Through a mixtures. This amount was established in order to allow a proper
 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558 3

Table 1
Metamax MK chemical composition.

Composition SiO2 Al2O3 TiO2 Fe2O3 K2O MgO CaO Na2O LOI
wt% 53.0 43.8 1.70 0.43 0.19 0.03 0.02 0.23 0.46

its full mechanical capacity. It is important to note that the molds

for flexural and tensile specimens used in this study were made of
steel, which in the case of geopolymers usually results in difficul-
ties on mold release. For this reason, semitransparent adhesive
tapes were used to cover and protect the steel parts of the molds,
allowing the material removal after 48 h of curing.
Prismatic specimens with dimensions of 160 mm  40 mm 
40 mm were produced for the bending tests. Additional specimens
with comparable dimensions were produced with a centered
notch, by positioning an aluminum plate in the middle of the mold,
creating a 1.5 mm thick and 12 mm high opening. Dumbbell-
shaped specimens were produced for uniaxial tension tests, with
a middle cross-section of 40 mm  24 mm, in accordance with
pre-established dimensions showed in previous studies; see e.g.
[9].
The manufacturing process of single-fiber pullout specimens
Fig. 1. Particle size distribution of metakaolin.
followed a distinct methodology, where a special rectangular
polymer-made mold was used; see Fig. 2. Three long polymeric
plates (two sturdy and one thinner and lower) were screwed
comparison with well-known SHCC results. The fiber’s physical together, creating a longitudinal channel in the middle of the mold
and mechanical properties are as follows: length equal to (Fig. 2a). The width of the channel corresponds to the desired fiber
12 mm, 40 mm diameter, and density 1.26 g/cm3, while tensile embedment length of 2 mm. The fibers were then transversely
strength and Young’s modulus are 1600 MPa and 40 GPa, respec- positioned over the channel with a spacing of 10 mm between each
tively. Since PVA fibers are highly hydrophilic, during production other and carefully fixed with wax at their ends. Two long plates
they are coated with an oil agent (1.2 wt%) to limit the chemical were then positioned and screwed onto the first two sturdy plates,
bond strength in cementitious matrices. Therefore, a similar rein- ensuring a 6 mm total channel depth, which was subsequently
forcement was used in this study, with the aim to observe possible filled completely with matrix material; see Fig. 2b. The molding
modifications caused by the matrix-oil interaction, and its influ- process and curing regime followed the same procedure presented
ence on the mechanical properties of SHGC. previously.
Demolding consisted of extracting the ‘‘long” beams from the
2.2. Production pullout mold with laterally protruding fibers. For single specimen
preparation, the beams were cut between the fibers, resulting in
A planetary Hobart mixer with a 10-liter capacity was used as at least 10 single specimens for each variation, with
follows: (i) addition of metakaolin and water glass in proper 6 mm  6 mm approximate dimensions, and 2 mm thickness (em-
amounts inside the mixer; (ii) mixing for 3 min at intermediate bedded length). The subsequent preparation of the samples con-
speed (198 rpm) to ensure adequate homogeneity and reactivity; sisted of cutting the fibers on one side of the specimen and
(iii) addition of aggregates; (iii) mixing for 1 min at intermediate isolating the remaining spot with wax to avoid any contact of the
speed (198 rpm); (iv) addition of fibers; (v) mixing for 3 min at fiber with glue in the later steps prior to testing.
higher speed (365 rpm) allowing a proper fiber distribution in
the geopolymer binder. The fresh mix was then poured into the
molds, which required a vibration step of 1 min both for consolida- 3. Testing methods
tion and removal of voids. The molds were sealed in plastic bags for
48 h at room temperature to prevent early dehydration. After this 3.1. Fresh-State properties of the geopolymer mixtures
period, the samples were removed from the molds and kept inside
dry plastic bags for 2 weeks, which is the necessary curing regime The fresh-state properties of the geopolymer mixtures with and
to stabilize the geopolymer mass (water loss). Failure to follow without aggregates were evaluated by means of flow table tests
these steps can lead to premature water loss, resulting in the for- according to ASTM C1437 [41] and ASTM C230 [42]. The cone
mation of unwanted cracks in the material, preventing the use of dimensions were 80 mm top diameter, 125 mm bottom diameter

Table 2
Content of incorporated materials for each composite’s variation (considering 1 kg of WG).

GP material NaGP KGP

Plain matrix Plain matrix + PVA Fine sand Fine sand + PVA Plain matrix Plain matrix + PVA Fine sand Fine sand + PVA
XOH/H2O ratio 0.44 0.62
WG/MK ratio 1.7123 1.8572
WG [g] 1000 1000
MK [g] 584.01 538.44
Fine sand [g] – 500 – 500
PVA [%] – 2.00 – 2.00 – – 2.00
4 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558

Fig. 2. Schematic view of the mold for single-fiber pullout specimens and its casting process: (a) fibers’ positioning; (b) final assembly after casting.

and 65 mm height. The flow diameter was measured after cone 3.3. Mechanical testing setups
removal, followed by 25 vertical jolts applied to the table for
15 s. The standard consistency was equal to the average of the A servo-hydraulic MTS universal testing system with a load
diameter measured after the test in two orthogonal directions. capacity of 200 kN was used to perform three-point bending tests
Four tests were performed for each mixture. The viscosity of the with all material variations presented previously. The tests were
mixtures was assessed following ASTM D6910 [43]: the fresh carried out based on the BS EN 196-1 [45]; 3 specimens were
mix was completely poured into a funnel and the time for all its tested for each variation under a load-controlled rate of 50 N/s.
content to pass through its 4.75 mm square opening was recorded. The span between the specimen supports was 100 mm. Their com-
Additionally, the fresh density was obtained according to ASTM pressive stress-deformation responses were obtained in the same
C138 [44], for which the fresh mixtures were placed, with subse- testing system using 40 mm cubical samples, at a 2400 N/s loading
quent consolidation, inside cylindrical containers with known rate.
dimensions. Three-point bending tests were also performed on notched
specimens in a servo-hydraulic MTS testing system with closed-
3.2. Analytical investigations loop control and a load cell of 100 kN, for all plain GP variations.
The specimens were tested at a constant crack mouth opening dis-
An XRD 3003 TT diffractometer system (model 7000XRD) was placement (CMOD) rate of 0.008 mm/min. The span distance
used to investigate the raw material (MK) and the synthesized between end supports was equal to 120 mm.
geopolymers in amounts of 0.22 g. The results of the X-ray diffrac- Hydraulic Instron testing equipment (model 8501) with closed-
tions were obtained by using copper radiation (Cu-Ka, k = 1.5418 loop control and a load capacity of 100 kN was used to perform the
Å) operating at 40 kV and 30 mA. Scans were performed using uniaxial tension tests on the SHGC under a displacement rate of
an angular velocity of 0.02° per 6 s, measuring the intervals 0.04 mm/s. Three dumbbell-shaped specimens were tested for
between (2h) 5° and 70° Bragg angles (2h). The GP samples were each composite material. The specimens were glued at their ends
ground in a McCrone Micronizing mill with isopropanol and mixed in 20 mm thick, steel rings bolted to the testing machine, ensuring
with 10 wt% of ZnO as the internal standard for quantification. MK non-rotatable boundary conditions. The specimen was first glued
was also analyzed with the same amount of ZnO but X-rayed with- in one ring outside of the machine, while after hardening the other
out isopropanol treatment. end could be attached and glued to the machine. Two Linear Vari-
TGA analyses were conducted on an STA 409 DC thermal ana- able Differential Transducers (LVDTs) were placed on each side of
lyzer from Netzsch Company (Germany. Both NaGP and KGP plain the specimen using a steel frame to measure the deformation in
mixtures were evaluated. All samples were mill grounded, depos- a 100 mm gauge length. Fig. 3 presents the tensile testing
ited on a TGA Al2O3 crucible-type and heated from 20 °C to configuration.
1000 °C at a heating rate of 10 K/min using 60 ml/min of O2 as During the tension tests, optical measurements were per-
the purge gas. formed, in order to monitor and quantify the specimen deforma-
Untested GP and SHGC specimens were cut into tion and crack formation with the help of Digital Image
10 mm  10 mm  10 mm cubes for porosity evaluations. Before Correlation (DIC). A black and white speckle pattern was sprayed
testing the samples were immersed in isopropanol for drying them onto the specimens for this purpose. The optical sampling rate
via solvent exchange. After drying and evaporation the samples was 1 frame/5 s. The frames were processed with a commercial
were measured with a Porotec Porosimeter PASCAL 140/440, with software ARAMIS 5 M, developed by GOM GmbH.
a mercury surface tension of 0.48 N/m, a contact angle of 140° and Single-fiber pullout tests were performed for all fiber-matrix
maximum testing pressure of 400 MPa. The procedure was per- combinations in a Zwick-Roell testing machine (model Z 1445)
formed in a temperature range of 23–23.6 °C with increased and with a 0.05 mm/s displacement rate using a 10 N capacity load cell.
decreased speeds of 6–19 MPa/min and 28–7 MPa/min, respec- It was only possible to use the displacement recorded from the
tively. The pore quantity and characteristics in the reinforced transverse actuator. The fiber elongation and displacement were
matrices were analyzed in order to evaluate the impact of fibers not measured directly. Fig. 4 presents the testing configuration
and aggregates on the composites’ microstructure. for single-fiber pullout experiments. The specimens were glued
An ESEM Quanta 250 FEG (FEI, Eindhoven/ The Netherlands) on a flat aluminum plate, which was screwed to the lower part
was used for microscopic analysis of the fracture surfaces of the of the machine. The free fiber end was glued to an upper plate,
GP matrices and fiber-reinforced composites. which was attached to the force sensor.
 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558 5

Fig. 3. Uniaxial tension test configuration.

Fundamental studies have fully demonstrated the crack-

bridging importance of fibers extending across a crack in the
matrix, usually referred to as the bridging stress (r) versus crack
opening (d) curve [36,48–50]. The bridging performance necessary
to reach a pseudo-strain-hardening (PSH) behavior can be esti-
mated through pullout and bending test responses on notched
specimens [1,37]. Fracture toughness is a measure of the magni-
tude of the stress concentration occurring in the crack tip when
the crack starts to propagate [36,50]. According to the crack model
presented in previous studies [51], the material’s Young’s modulus
(Em) and toughness (Km) can be obtained by the following
equations:
2 3
 3
0:413Pi 4 l 1 þ 8Pi
5xl
1:17l
Em ¼ 3 þ  5 ð1Þ
di 3 1:68bd 1  da
4bd 1  a d

pffiffiffiffiffi
K m ¼ rn ae YðaÞ ð2Þ

1:99  að1  aÞð2:15  3:93a þ 2:70a2 Þ

YðaÞ ¼ ð3Þ
ð1 þ 2aÞð1  aÞ1:5

where Pi corresponds to the load level in the initial portion of the

load-deflection plot, di is the deflection value; l, b, and d, are the
Fig. 4. Single-fiber pullout test configuration. span, width, and depth of the specimen, respectively; a and x are
the initial notch depth and self-weight of the specimen per unit
length, respectively. rn is the flexural strength of the specimen;
ae is the effective notch depth that can be obtained from Eq. (2)
3.4. Fracture mechanics requirements
and Y(a) is the correction factor; a = ae =d:
As shown in previous SHCC designs, two criteria must be satis-
Previous SHCC and SHGC studies used the fracture mechanics
fied to obtain a pseudo-strain hardening behavior with multiple
requirements to validate and tailor the fiber, matrix and interface
cracking formations: (i) an energy-based condition; and (ii) a
in order to manufacture a material with adequate performance
stress-based condition [1,35,39,48]. The first one can be expressed
with the minimum amount of fibers [46]. This validation is consid-
as [1,35,48]:
ered by many experts as a main design basis, allowing the evalua-
0
tion of several fibers and matrices [46–48]. Despite the challenges J tip  J b ð4Þ
in ensuring a reliable result, mainly due to difficulties in obtaining
accurate pullout responses, the fracture mechanics requirements where J0 b is the complementary energy, and Jtip is the composite
are used in this study to verify the performance of the MK-based crack tip toughness, obtained from the r(d) curve [1,35,39]. Jtip
SHGC manufactured. and J’b can be expressed as:
6 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558

ðK m Þ2 respectively. This incorporation had also a similar reduction effect

J tip ¼ ð5Þ on the viscosity of the fresh mixtures. Plain NaGP and KGP matrices
Em
required 102 and 78 s to pass through the funnel opening com-
! pletely, while on the addition of aggregates their duration
0 Lf s20 L2f
Jb ¼ V f  2Gd ð6Þ increased to 145 and 108 s, respectively. The influence of
df 6df Ef
aggregates on the reduction of flowability and viscosity can be
Eq. (6) corresponds to the energy equilibrium during steady- attributed to the increase in internal friction [52] associated with
state [1,35,39]. If J’b is greater than Jtip, both steady-state and the the great difference between MK and sand grain sizes. The fine
occurrence of a multiple cracking behavior are assured [1,4,48]. aggregates also influenced the fresh density results, where
The shape of the r(d) curve and the rising branch associated with sand-reinforced mixtures presented 6.9–9.1% greater values when
J’b are also related to the fiber-matrix interactions [4,49]. When compared to the plain matrices.
fibers are only in frictional contact with the matrix, the stiffness
is mainly controlled by the fiber content (Vf), diameter (df), length 4.2. Analytical investigations
(Lf), stiffness (Ef), and frictional bond (s0) [4,49]. When a chemical
bond (Gd) occurs, the starting point of the curve is not at the origin X-ray diffraction (XRD) is used to characterize the precursor and
but is shifted upwards [4,49], resulting in the need for interfacial quality of the geopolymerization process. As evidenced in previous
mechanisms to occur before the chemical adhesion is lost, in order studies [53,54], the use of high reactive MK results in unique XRD
to guarantee an improved loading distribution. The chemical adhe- profiles. Its verification is necessary in order to monitor the quality
sion energy (Gd) is equivalent to [35]: and type of material produced, validating the use of well-
 2 established amorphous GP formulations. The XRD patterns of MK
2 P ad  Pfr and the two GP formulations are presented in Fig. 5. Metamax
Gd ¼ ð7Þ
p2 Ef d3f MK is known as an almost completely amorphous material, charac-
terized by a wired spread hump centered at 22° 2h [22], which is
where Ef is the fiber’s Young’s, and df its diameter. also found in Fig. 5. Reduced evidence of crystallinity can be seen
Thus, the presence of Gd diminishes the complementary energy at the peaks shown at 25.3° 2h, from zincite (ZnO), and at several
J’b. The stress-based condition corresponds to [1,35,48]: points subsequently to 32.5° 2h, from anatase (TiO2). The first is
believed to have appeared due to the sample’s treatment used,
rfc  r0 ð8Þ
while the second results from the presence of impurities in the
where r0 is the ultimate tensile strength of the composite, and rfc MK material [53]. When evaluating the MK reactivity in GP sam-
is the stress at the formation of the first crack in the composite. If ples, it is worth noticing the shift in the broad amorphous hump,
r0 is greater than rfc, there will be multiple crack formation; if which is usually considered as fully reacted when centered at
rfc is greater than r0, the composite will fail after the formation 28° 2h [54]. This occurrence is also attested by the NaGP and
of its first crack [1,35]. In literature [1] it is also suggested to use KGP profiles, presented in Fig. 5.
PSH performance indices to evaluate the behavior of the composites Fig. 6 presents the TGA responses of both NaGP and KGP in an
[1,39], as follows: O2 gas condition. It is known by previous researches [40,55,56] that
0
hardened GP materials result in three types of water-release dur-
Jb ing heating: (i) physically bonded H2O; (ii) chemically bonded
Energy  based condition : PSHð1Þ ¼ ð9Þ
J tip H2O; and (iii) hydroxyl groups OH. From the TGA curves, it can
be observed that both materials were completely dried up to
r0 100 °C, due to the loss of physically bonded H2O [56], with 6.0%
Stress based condition : PSHð2Þ ¼ ð10Þ
rfc and 8.5% weight reductions recorded for NaGP and KGP, respec-
tively. The water loss in this range of temperature leaves empty
Indices higher than 1 indicate the likelihood of strain- spaces inside the GP materials and is directly correlated to the
hardening. material’s porosity. TGA curves also indicate a distinctive weight
loss in the range between 100 °C and 250 °C, due to the chemically
4. Results and discussion bonded H2O loss phase [56], with a weight reduction of 17.5% in
total for the Na-based GP and 13.0% for the K-based sample. It is
4.1. Fresh-State properties of the geopolymer mixtures worth noticing that this type of chemically bounded water may
also be referenced as zeolitic water [55,56] since it corresponds
The results with the fresh mixtures are summarized in Table 3. to a part of the reaction water generated during polycondensation
The average flowability values obtained for NaGP and KGP were that remains within the 3D GP framework. Therefore, the TGA
241 mm and 284 mm, respectively. This was expected, since the results indicate an initially more established geopolymer network
WG-to-MK ratios used in the mixing process were higher for KGP for the NaGP material, since it presented a greater zeolitic water
(1.85) when compared to NaGP (1.71), resulting in a greater KGP reduction up to 250 °C, as opposed to the previous degradation
workability. The incorporation of aggregates led to a reduction of phase up to 100 °C [56]. This effect is even more visible by the con-
the flow diameter to 193 mm and 226 mm for both mixtures, sidering DTG curves, showing distinctive peaks centered around

Table 3
Fresh properties of Na- and K-based GP mixtures (average values).

Mixture NaGP KGP

Reinforcement Plain matrix Fine sand Plain matrix Fine sand
Flowability [mm] 241 193 284 226
Funnel-passing time [s] 102 145 78 108
Fresh density [g/cm3] 1.69 1.81 1.61 1.76
 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558 7

Fig. 5. Diffractograms of metakaolin (MK), KGP and NaGP: (A) ZnO and (B) TiO2.

Fig. 6. TGA curves of plain NaGP and KGP in an oxygen environment.

Fig. 7. Results of MIP analysis for (i) Na and K-based plain matrix; (ii) enriched with
aggregates; and (iii) containing both aggregates and PVA fibers.

135 °C. Here a maximum water loss rate occurs of 2.1%/min and
1.1%/min, for NaGP and KGP materials, respectively. results, since a suitable aggregate-matrix adhesion is expected to
After this temperature range, the lower amount of remaining occur, reducing their brittleness due to a greater ability to with-
water may be associated with hydroxyl groups OH [55,56], where stand the applied stresses.
the materials’ gradual decomposition occurs due to further shrink- The larger increase in porosity occurred for KGP enriched with
age associated with dehydroxylation up to 1000 °C [40]. Total aggregates and fibers. Regarding bulk density, NaGP-based materi-
weight losses of 21.5% and 17.5% were recorded up to 1000 °C als yielded 1.71 and 1.77 g/cm3, respectively, while KGP compos-
for NaGP and KGP materials, respectively. ites showed 1.54 and 1.59 g/cm3. Additionally, the bulk density
The results obtained from MIP measurements are presented in values measured on the hardened materials were smaller than
Fig. 7 and Table 4 and corroborate the TGA analysis. The NaGP the ones found for the fresh state, which can be traced back to
and KGP plain matrices (blue and black curves) show a distinguish- the water loss during storage and treatments adopted for this anal-
able difference in their porous microstructure. NaGP exhibits a ysis; see Table 3.
lower total porosity of 16.0%, while KGP reaches a slightly larger Fig. 8 shows the scanning electron microscopy results for both
value of 18.5%, indicating again a higher degree of geopolymeriza- NaGP and KGP. A much denser microstructure was observed for
tion for the Na-based GPs. However, it is interesting to note that NaGP materials, while KGP samples presented a greater number
the average pore diameter found for both plain matrices was prac- of larger and randomly dispersed pores, confirming the previous
tically the same, showing regular nano-porosity for GPs in general, findings by MIP and TGA analysis.
despite their alkali precursors, as indicated in previous studies
[22,53].
The addition of fibers and aggregates increases the porosity of 4.3. Mechanical properties of the geopolymers
both GP materials. When only fine-grained sand was added to
the plain matrices, small increases in total porosity were recorded. Fig. 9 and Table 5 present the results of the three-point bending
KGP showed a larger modification, reaching now 20.3%, possibly tests. In general, both GP with and without aggregates exhibited a
due to the inadequacy of such grains in a high fluidity environment very brittle behavior, which can be laid to their fine-grained nature
when compared with NaGP. Also, both matrices demonstrated a and the intrinsic elastic behavior of ceramic materials [57]. The
small increase in average pore diameter, which may be related to plain GP yielded a lower load-bearing capacity and considerably
the high disparity between grain sizes (MK and sand). However, lower fracture toughness, the former being also dependent on
this difference should not influence negatively the mechanical the latter.
8 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558

Table 4
General results of MIP analysis (average values).

Mixture NaGP KGP

Reinforcement Plain matrix Fine sand Fine sand + PVA Plain matrix Fine sand Fine sand + PVA
Porosity [%] 16.04 16.46 16.86 18.49 20.32 21.02
Pore diameter [mm] 0.0067 0.0074 0.0084 0.0067 0.0079 0.0087
Bulk density [g/cm3] 1.65 1.71 1.77 1.49 1.54 1.59

Fig. 8. Microstructure of the GP matrices in distinct magnifications: (a) NaGP and (b) KGP.

to present an increased crack tip toughness (Jtip). This was verified

by calculating the Km and Jtip factors, which will be discussed later.
Indeed, KGP yielded higher Jtip values, 6.32 N/m on average, in
comparison to the 2.63 N/m measured for NaGP, mostly due to
the intrinsic toughness of the GP materials, where a more porous
microstructure leads to crack meandering effects [58]. As expected,
Young’s modulus was higher for materials enriched with aggre-
gates, being more evident in the case of NaGP, where increased
suitability of components has been observed in all analyses so
far. However, the Jtip values found for both GP containing sand
were not as distinct from the plain matrix, showing even slightly
lower values, which indicates a low variability in this parameter
of GP with aggregate inclusion, possibly reducing its cracking
capacities, which will be later on verified.

4.4. Mechanical properties of the fiber-reinforced composites

The representative flexural stress-deflection curves obtained for

Fig. 9. Effects of matrix composition and addition of aggregates on force vs. CMOD both GP composites reinforced with PVA fibers are shown in
curves. Fig. 10, while the corresponding average mechanical parameters
are summarized in Table 6. The fracture energy values were
obtained from the area below the load  displacement curves.
NaGP exhibited higher flexural strength when compared to the The flexural stress was calculated by using the general
KGP variations. These results are supported by the TGA, MIP and formulation:
ESEM investigations, where NaGP mixtures exhibited an amount
3Pl
of bound water, higher density and lower porosity when compared r¼ 2
ð11Þ
to KGP. A material with lower Young’s modulus, such as KGP, tend 2bd

Table 5
Results of crack-mouth opening displacement (CMOD) measurements (average values).

Mixture NaGP KGP

Reinforcement Plain matrix Fine sand Plain matrix Fine sand
Max. Load [N] 368.77 506.78 235.18 490.45
Em [GPa] 8.76 10.35 7.02 9.76
pffiffiffiffiffi
Km [MPa m] 0.15 0.15 0.25 0.25
Jtip [N/m] 2.63 2.12 9.21 8.32
 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558 9

the stress-strain curves. For NaGP, a pronounced increase in first-

crack stress can be observed for the material containing aggregates
in comparison to the plain matrix, from 2.0 MPa to 2.9 MPa. Also,
the composite without aggregates exhibited higher tensile
strength, reaching 4.2 MPa, as opposed to the 3.8 MPa measured
for the same material containing sand. This behavior was expected
from the previous fracture evaluations.
Nevertheless, both NaGP composites showed pronounced
strain-hardening and strain capacities of 3.5–4.8%. Obviously, an
optimal balance between the cracking strength of the matrix and
the crack-bridging capacity of the fibers was attained. The differ-
ences in fiber-matrix interactions for various types of matrices will
be discussed when presenting the results of pullout tests.
The tensile behavior of the KGP composites was even more
affected by the aggregate addition; the composites containing sand
did not yield a typical strain-hardening curve, but a strain-
softening one. In contrast, plain KGP composites demonstrated a
stress enhancement with increasing strain, presenting themselves
Fig. 10. Bending tests results on (a) Na- and (b) K-based geopolymers incorporated as a much more feasible option for SHGC application. For all SHGC
with different types of reinforcements. variations, the values of Young’s modulus resulted in similar
ranges, with ~9.5 GPa for plain mixtures, while a slight enhance-
ment for GP containing sand was found, ~10.5 GPa.
where P is the applied load; l is the span between supports; b and d Multiple cracking and fracture occurrence were evaluated
are the width and depth of the specimen. through digital image correlation (DIC), allowing the determina-
The results of the bending tests indicate the superior mechani- tion of the number of cracks, average crack width, and crack spac-
cal performance of NaGP fiber-reinforced composites when com- ing; as presented in Table 9 and Fig. 13. Fig. 13 also demonstrates a
pared to the same KGP variation; reaching an average equivalent typical crack pattern analysis at different multiple cracking stages
flexural stress of 19.7 MPa, opposed to 14.9 MPa found for the lat- of loading: 0.2%, 1.0% and 2.5% strains.
ter. Since the flexural stress of strain-hardening composites is A higher crack density was observed for NaGP composites, in all
determined by the crack bridging action of the fibers to a great strain levels, with and without aggregates, when compared to KGP,
extent, the results also indicate distinct fiber-matrix interactions. suggesting enhanced fiber-matrix adequacy. The addition of aggre-
The incorporation of aggregates had a negative effect on their flex- gates presented a lower crack formation, as expected and demon-
ural stress, which can be attributed to the negatively altered fiber strated on deformation values. In combination with a less
distribution. To better identify and quantify the effects of matrix pronounced multiple cracking, K-based SHGC yielded lower strain
composition and aggregates on the fiber-matrix interaction, capacities when compared to Na-based composites. It is interesting
single-fiber pullout tests appear to be instrumental. when one notices the slightly better crack control in comparison to
The stress-displacement curves obtained from compression common PVA-SHCC, which typically yield crack widths of about
tests are presented in Fig. 11, their evaluation summarized in 70 mm and crack spacing of 2.5 mm [59].
Table 7. An increase in compressive strength due to the use of Fig. 14 presents the microstructure of the fracture surfaces of
aggregates is observed with a stiffness enhancement of both types both composites, after complete failure in the uniaxial tension
of GP. The incorporation of fibers had a positive effect on compres- tests. Fig. 14a and b show the embedment of the fibers in the
sive strength, mainly for Na-based GP materials: 77.6 MPa was matrices, demonstrating adequate interaction for both K-based
reached, as opposed to 44.1 MPa measured for plain NaGP. This and Na-based materials, without any evident fiber surface damage.
behavior may be associated with the reduction of spalling effects Fig. 14c and d show the external surfaces of the partially pulled out
that occur during the loading in the fragile unconfined geopoly- fibers; once again, no deterioration can be observed. Micrographs
mers [57] by efficient crack bridging and force redistribution. also revealed some regions of fiber agglomerations for all mixtures
Aggregates and especially fibers act as internal micro- under investigation, exemplarily shown here in Fig. 14c. This find-
confinements, enhancing the toughness and damage tolerance of ing points out a need for enhancing the rheology of the fresh mix-
the material. This improvement in strength was less pronounced ture as well as mixing procedure. Fig. 14e and f depict the channels
in the case of KGP materials, arising from 37.4 MPa (plain KGP), left in the matrix after fiber pullout, confirming the occurrence of
to 59.4 MPa (with both aggregate and fiber). Obviously, the fibers partial fiber debonding. The images clearly show the high homo-
improved also the post-peak behavior of the specimens, leading geneity of the NaGP matrix and its pronounced brittleness, marked
to a shallower softening branch. by the fine loose fragments. The microstructures of the same com-
The tensile behavior of both composites reinforced with and posites enriched with sand were also investigated in ESEM, but no
without aggregates and incorporated with PVA fibers are presented clear distinctions were observed, evidencing a need for a deeper
in Fig. 12; Table 8 provides the characteristic values obtained with microscopical evaluation.

Table 6
Results of bending tests on Na- and K-based geopolymers incorporated with different types of reinforcements (standard deviation values in parenthesis).

Mixture NaGP KGP

Reinforcement Plain matrix Fine sand PVA Fine sand + PVA Plain matrix Fine sand PVA Fine sand + PVA
Equivalent flexural stress [MPa] 1.82 (0.17) 3.41 (0.21) 19.74 (0.18) 17.32 (0.94) 0.91 (0.12) 1.93 (0.35) 14.92 (1.82) 13.15 (1.08)
Deflection [mm] – – 2.74 (0.19) 2.31 (0.38) – – 2.30 (0.29) 1.95 (0.14)
Fracture energy [N.mm] – – 24,133 (2420) 20,153 (1792) – – 13,800 (2570) 11,805 (2784)
10 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558

Fig. 11. Results of the compression tests on (a) Na- and (b) K-based geopolymers incorporated with different types of reinforcements.

Table 7
Results of compression tests on Na- and K-based geopolymers incorporated with different types of reinforcements (standard deviation values in parenthesis).

Mixture NaGP KGP

Reinforcement Plain matrix Fine sand PVA Fine Plain matrix Fine sand PVA Fine
sand + PVA sand + PVA
Compressive strength [MPa] 44.09 (4.92) 56.83 (8.27) 65.48 (5.41) 77.56 (11.48) 37.35 (3.81) 53.94 (4.67) 47.09 (4.71) 59.43 (3.77)
Displacement at max. strength 0.775 0.647 1.185 1.054 (0.039) 0.682 0.661 0.889 0.991 (0.073)
[mm] (0.022) (0.032) (0.027) (0.024) (0.025) (0.044)

Fig. 12. Results of tension tests on Na and K-based geopolymers: (a) plain matrix with PVA fibers; and (b) with the addition of both aggregates and PVA fibers.

Table 8
Results of tension tests on Na- and K-based geopolymers: (i) plain matrix reinforced with PVA fibers; and (ii) containing both aggregates and PVA fibers (standard deviation values
in parenthesis).

SHGC NaGP KGP

Reinforcement Plain matrix Fine sand Plain matrix Fine sand
First crack stress [MPa] 2.01 (0.07) 2.87 (0.18) 1.86 (0.12) 2.73 (0.06)
Deformation at first crack stress [%] 0.17 (0.02) 0.16 (0.02) 0.20 (0.01) 0.18 (0.01)
Young’s modulus [GPa] 9.82 (0.99) 10.94 (1.13) 9.30 (0.14) 10.17 (0.51)
Tensile stress [MPa] 4.17 (0.47) 3.71 (0.38) 2.57 (0.29) 2.73 (0.06)
Deformation at max. stress [%] 3.53 (0.79) 4.76 (0.34) 2.68 (0.28) –
 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558 11

Table 9
Evaluation of cracks by means of DIC analysis (average values).

SHGC NaGP KGP

Reinforcement Plain matrix Fine sand Plain matrix Fine sand
Number of cracks 52.5 66.25 38.5 43.5
Crack width [mm] 64.08 62.45 62.37 59.37
Crack spacing [mm] 1.34 1.13 1.57 1.85

Fig. 15 presents a comparison of the stress-strain behavior of abrasion resistance of the fibers [62,63]. The interfacial damage
the Na-SHGC investigated in this study with a typical SHCC mechanisms result in slip-hardening pullout behavior, leading to
response [60]. By plotting the representative curves together, it fiber rupture after just short pullout phases. On the contrary, the
can be affirmed that similar overall mechanical performance was anchorage of the PVA fibers in the NaGP matrix results in a slip-
obtained for Na-based SHGC, with flexural strength close to softening behavior, indicating a dramatic reduction of the chemical
20 MPa both for SHCC and SHGC. At the same time, individual adhesion. This behavior may be related to (i) the hydrophilic nat-
properties of these mixtures differ more evidently: cement-based ure of the fibers, enhancing the chemical adhesion in mixtures
materials exhibit a higher Young’s modulus of 13.6 GPa and higher with great water contents; (ii) inadequate polarity between the
tensile strength of approximately 5 MPa [60]. Worth noting is also oil coating and the fresh GP mixtures; and (iii) difference in the
the explicit indication of crack formation in SHCC by sudden stress chemical composition and morphology of the matrix resulting
drops in the stress-strain curves, while the corresponding curves from geopolymerization. Moreover, the slip-softening pullout
for SHGC are much smoother. While lower crack widths and high behavior shows that the fibers do not suffer surface damage during
crack density in SHGC certainly contribute to this type of response, pullout, which is likely due to high matrix homogeneity at the
we need to look in detail at possible distinctions in fiber-matrix microscopic level. The addition of sand seems to roughen the walls
adhesive and frictional bonds as well. These features may also be of the fiber channels, leading to slight slip-hardening, as shown in
interesting for applications with dynamic loading and require- Fig. 14b.
ments of long-term durability. Furthermore, the little difference The K-based compositions yielded pullout behavior similar to
in the density of the matrix and the fibers in the case of geopoly- that observed in cementitious matrices [59,60], i.e. strong chemical
mers may contribute to better fiber dispersion and consequently adhesion (fiber anchorage) and slip-hardening with complete fiber
to improved crack control. Note that the cement-based materials rupture in 2 mm embedded lengths. From the KGP curves it is not
have a density of 2.01 g/cm3 as opposed to those made of GP with possible to accurately distinguish the debonding and pullout
1.49–1.77 g/cm3; the density of PVA fibers is 1.26 g/cm3. stages, due to the occurrence of several force drops, thus indicating
The pullout test results are presented in Fig. 16 and Table 10. a stepwise debonding process. The values used in Table 10 are
The adhesive and frictional stresses were evaluated using the fol- approximated and mostly used to provide a general comparison
lowing equations [35,36]: in the fracture mechanics requirements for both matrices, as also
performed in previous studies [37,46,49]. Given the flexural and
P ad
sadhesional ¼ ð12Þ tensile properties of the corresponding composites, the importance
2prL of a balanced fiber-matrix interaction, the negative effect of the
strong slip-hardening and the challenges in accurately measuring
Pfr
sfrictional ¼ ð13Þ the pullout responses must be explicitly highlighted. It is also
2prL
worth noticing the difficulties in providing visual evidences for
where r is the radius of the fiber and L is the embedded fiber length. such mechanisms, since the ESEM images acquired in Fig. 14 did
In ideal pullout curves, as shown in previous studies [61], the adhe- not provide enough insight on such parameters, evidencing a need
sional (Pad) force is obtained from the debonding point in the for further improvements in microscopical evaluations.
ascending part of the curve, characterized by the loss of linearity. Regarding fractural mechanics: Since there is no occurrence of
Then, there is the occurrence of a maximum peak load (Pmax), in frictional loads for KGP-based materials, the complementary
conditions of partial/critical debonding. Following the peak, the energy values Jb’ cannot be accurately measured, they depend fully
pullout response is governed by frictional forces (Pfr) until the fiber on the difference between frictional bond strength Pfr and chemical
is completely detached or broken. Approximated values of Pad and adhesion Gd. Previous studies have also highlighted the issues on
Pfr were selected in distinguishable phases of the curves presented the occurrence of high Gd values, which was found to mostly inval-
in Fig. 16, following the same procedures used in several previous idate the quality of a strain-hardening material [49]. In contrast, in
works on strain hardening cement-based composites [46–48]. the case of NaGP mixtures, it was possible to calculate the
Fig. 16 shows the pullout curves for GP materials with and complementary energy, which resulted in values of 3.52 N/m and
without aggregates. For NaGP compositions, a complete pullout 9.56 N/m for compositions without and with aggregates, respec-
of the fibers could be observed, which is indicated by the relatively tively. Herewith, NaGP-based SHGC containing fine sand exhibited
steady and long descending branches of force-slip displacement similar adequacy, as found for SHCC, for which typical Jb values
curves. NaGP without sand demonstrated lower average force val- range between 9.6 and 10.7 N/m [59].
ues at adhesive failure and friction activation, 0.186 and 0.141 N, This brings us to the design criteria for the occurrence of
respectively, with the maximum force recorded in the initial part pseudo-strain-hardening. The PSHtension value of 2.7 obtained for
of the curve. In contrast, the mixture incorporated with fine- NaGP composites without aggregates is higher than that calcu-
grained sand demonstrated increases in the slip hardening coeffi- lated for the same parameter variations with KGP, where 1.38
cient (the curves become convex upwards), with greater adhesive was observed. PSHenergy appears as the main variable in the esti-
and frictional forces, with the maximum occurring in the frictional mation of composite behavior. The value obtained for KGP-based
phase (average value of 0.361 N). materials is not representative due to the dominance of chemical
Note that in common PVA-SHCC matrices, the pullout of the adhesion between fiber and matrix, as mentioned previously. As
PVA fibers is accompanied by pronounced surface damage, partly for NaGP composites, a large value variation can be observed with
caused by the cohesive interfacial failure and partly by the low the addition of aggregates, mainly influenced by the frictional
12 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558

Fig. 13. Evaluation of cracking by means of DIC analysis for all composites in 3 different stages of loading.

adhesion, resulting in a PSHenergy value of 4.51, as opposed to 1.33 would theoretically result in micro-crack formation saturation
found for plain NaGP matrix. In the literature, it is possible to find in the composite. This is exactly what occurs in the case of NaGP
indications for better efficiency with PSHenergy > 3 [56], which with aggregates.
 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558 13

Fig. 14. Condition of the fibers and matrix on the fracture surfaces of (a, c, e) NaGP and (b, d, f) KGP geopolymer composites.

Fig. 15. Comparison of the mechanical behavior of SHGC and SHCC (SHCC data from [60]).
14 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558

Fig. 16. Behavior of PVA fibers when pulled out from (a) NaGP and KGP matrices and (b) NaGP and KGP matrices enriched with aggregates.

Table 10
Pullout test results based on approximated debonding phases in the curves (standard deviation values in parenthesis).

SHGC NaGP KGP

Reinforcement Plain matrix Fine sand Plain matrix Fine sand
Pad load [N] 0.186 (0.032) 0.341 (0.122) 0.595 (0.133) 0.574 (0.249)
Pfr load [N] 0.141 (0.031) 0.235 (0.056) – –
Max. load [N] 0.218 (0.058) 0.361 (0.113) 0.595 (0.133) 0.574 (0.249)
rad stress [MPa] 0.741 (0.131) 1.358 (0.486) 2.214 (0.525) 2.269 (0.992)
rfr stress [MPa] 0.557 (0.124) 0.935 (0.224) – –
Max. stress [MPa] 0.871 (0.233) 1.438 (0.451) 2.214 (0.525) 2.269 (0.992)
Gd [J/m2] 0.019 (0.016) 0.129 (0.153) 2.57 (1.232) 2.88 (1.395)

5. Conclusions crack spacing of around 1.4 mm. When compared to the typical
behavior of strain-hardening cement-based composites (SHCC), it
The results of the experimental investigations showed that was found that Na-based SHGC showed smaller crack widths and
NaGP mixtures exhibited lower flowability and workability in the a larger number of cracks. As a result, improved strain capacity
fresh state while achieving higher density and lower porosity in and smoother shape of the stress-strain curves were observed.
comparison with K-based GP, due to a higher degree of geopoly- The findings indicate the great potential of this material for a wide
merization. This resulted in higher tensile, flexural and compres- range of applications, including structural elements subject to
sive strengths as well as in higher Young’s moduli of the NaGP dynamic loading and requirements of long-term durability or
compositions. harsh chemical environments.
The addition of fine-grained sand caused a small decrease in the
workability of the fresh mixtures, for both NaGP and KGP, but
CRediT authorship contribution statement
enhanced their Young’s moduli and first-crack stress level. The
use of PVA fibers improved not only the tensile strength and duc-
Ana Carolina Constâncio Trindade: Conceptualization, Mate-
tility of the composites but also enhanced their compressive
rial preparation, Testing, Data curation, Writing. Iurie Curosu:
behavior, acting as an internal micro-confinement.
Investigation, Supervision, Validation, Writing - review & editing.
The fiber reinforcement was effective in ensuring typical strain-
Marco Liebscher: Investigation, Supervision, Validation, Writing
hardening behavior under tensile loading, accompanied by pro-
- review & editing. Viktor Mechtcherine: Funding acquisition, Pro-
nounced multiple cracking. NaGP strain-hardening geopolymer
ject administration, Supervision. Flávio de Andrade Silva: Concep-
composites (SHGC) with and without fine sand yielded strain
tualization, Validation, Funding acquisition, Project administration,
capacities of 3.5% and 4.7%, respectively. K-based SHGC without
Supervision.
aggregates achieved only 2.6% on average, while K-based compos-
ites containing both fiber and aggregates exhibited a strain-
softening behavior. Declaration of Competing Interest
Pullout tests made possible an approximated comparative anal-
ysis of the fiber-matrix interactions, evidencing a more balanced The authors declare that they have no known competing finan-
fiber anchorage for NaGP composites, with improvements in the cial interests or personal relationships that could have appeared
frictional adhesion supported by the addition of aggregates. In con- to influence the work reported in this paper.
trast, KGP composites showed a strong fiber-matrix chemical
adhesion, which led to fiber failure rather than proper debonding. Acknowledgments
The calculated pseudo-strain-hardening criteria based on the con-
cepts of fracture mechanics confirmed the better standing of Na- The authors acknowledge the financial support provided by the
based composites with respect to tensile ductility. Brazilian organizations CNPq and CAPES (Probral project number
DIC evaluations showed that both SHGC compositions exhibited 8887.144079/2017-00) and by the DAAD program (Probral project
full crack saturation with crack widths of 60 mm and an average number 8887.144079/2017-00) in Germany. Furthermore, the
 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558 15

authors express their gratitude to the German Research Founda- [29] D.R. Morgan, Compatibility of concrete repair materials and systems, Constr.
Build. Mater. 10 (1) (1996) 57–67.
tion (Deutsche Forschungsgemeinschaft – DFG) for the financial
[30] P. Duxson, J.L. Provis, G.C. Lukey, S.W. Mallicoat, W.M. Kriven, J.S. Van
support provided within the framework of the Research Training Deventer, Understanding the relationship between geopolymer composition,
Group GRK 2250/1, Project number 287321140. Also, the support microstructure and mechanical properties, Colloids Surf., A 269 (1–3) (2005)
of the laboratory staff at the Institute of Construction Materials, 47–58.
[31] X.Y. Zhuang, L. Chen, S. Komarneni, C.H. Zhou, D.S. Tong, H.M. Yang, H. Wang,
TU Dresden is greatly appreciated. Fly ash-based geopolymer: clean production, properties and applications, J.
Cleaner Prod. 125 (2016) 253–267.
[32] C. Menna, D. Asprone, C. Ferone, F. Colangelo, A. Balsamo, A. Prota, G. Manfredi,
References Use of geopolymers for composite external reinforcement of RC members,
Compos. B Eng. 45 (1) (2013) 1667–1676.
[1] V.C. Li, On engineered cementitious composites (ECC), J. Adv. Concr. Technol. 1 [33] D.P. Dias, C. Thaumaturgo, Fracture toughness of geopolymeric concretes
(3) (2003) 215–230. reinforced with basalt fibers, Cem. Concr. Compos. 27 (1) (2005) 49–54.
[2] V. Mechtcherine, Novel cement-based composites for the strengthening and [34] M. Ohno, V.C. Li, A feasibility study of strain hardening fiber reinforced fly ash-
repair of concrete structures, Constr. Build. Mater. 41 (2013) 365–373. based geopolymer composites, Constr. Build. Mater. 57 (2014) 163–168.
[3] J. Aveston, A. Kelly, Theory of multiple fracture of fibrous composites, J. Mater. [35] B. Nematollahi, J. Sanjayan, F.U. Ahmed Shaikh, Tensile strain hardening
Sci. 8 (3) (1973) 352–362. behavior of PVA fiber-reinforced engineered geopolymer composite, J. Mater.
[4] G.P. van Zijl, F.H. Wittmann, B.H. Oh, P. Kabele, R.D. Toledo Filho, E.M. Civ. Eng. 27 (10) (2015) 04015001.
Fairbairn, F. Altmann, Durability of strain-hardening cement-based composites [36] M. Ohno, V.C. Li, An integrated design method of engineered geopolymer
(SHCC), Mater. Struct. 45 (10) (2012) 1447–1463. composite, Cem. Concr. Compos. 88 (2018) 73–85.
[5] V. Mechtcherine, F. de Andrade Silva, S. Müller, P. Jun, R.D. Toledo Filho, [37] B. Nematollahi, J. Sanjayan, J. Qiu, E.H. Yang, Micromechanics-based
Coupled strain rate and temperature effects on the tensile behavior of strain- investigation of a sustainable ambient temperature cured one-part strain
hardening cement-based composites (SHCC) with PVA fibers, Cem. Concr. Res. hardening geopolymer composite, Constr. Build. Mater. 131 (2017) 552–
42 (11) (2012) 1417–1427. 563.
[6] H.D. Yun, S.W. Kim, Y.O. Lee, K. Rokugo, Tensile behavior of synthetic fiber- [38] B. Nematollahi, J. Sanjayan, F.U.A. Shaikh, Matrix design of strain hardening
reinforced strain-hardening cement-based composite (SHCC) after freezing fiber reinforced engineered geopolymer composite, Compos. B Eng. 89 (2016)
and thawing exposure, Cold Reg. Sci. Technol. 67 (1–2) (2011) 49–57. 253–265.
[7] S. Müller, V. Mechtcherine, Fatigue behaviour of strain-hardening cement- [39] R.P. Batista, A.C.C. Trindade, P.H. Ribeiro Borges, F.A. Silva, Silica fume as
based composites (SHCC), Cem. Concr. Res. 92 (2017) 75–83. precursor in the development of sustainable and high-performance MK-based
[8] S. Qian, V.C. Li, Simplified inverse method for determining the tensile alkali-activated materials reinforced with short PVA fibres, Front. Mater. 6
properties of strain hardening cementitious composites (SHCC), J. Adv. Concr. (2019) 77.
Technol. 6 (2) (2008) 353–363. [40] J.L. Bell, P.E. Driemeyer, W.M. Kriven, Formation of ceramics from metakaolin-
[9] I. Curosu, M. Liebscher, V. Mechtcherine, C. Bellmann, S. Michel, Tensile based geopolymers. Part II: K-based geopolymer, J. Am. Ceram. Soc. 92 (3)
behavior of high-strength strain-hardening cement-based composites (HS- (2009) 607–615.
SHCC) made with high-performance polyethylene, aramid and PBO fibers, [41] ASTM C1437-15, Standard Test Method for Flow of Hydraulic Cement Mortar,
Cem. Concr. Res. 98 (2017) 71–81. ASTM International, West Conshohocken, PA, 2015, www.astm.org.
[10] I. Curosu, V. Mechtcherine, D. Forni, E. Cadoni, Performance of various strain- [42] ASTM C230/C230M-14, Standard Specification for Flow Table for Use in Tests
hardening cement-based composites (SHCC) subject to uniaxial impact tensile of Hydraulic Cement, ASTM International, West Conshohocken, PA, 2014,
loading, Cem. Concr. Res. 102 (2017) 16–28. www.astm.org.
[11] M.D. Lepech, V.C. Li, Application of ECC for bridge deck link slabs, Mater. Struct. [43] ASTM D6910/D6910M-09, Standard Test Method for Marsh Funnel Viscosity of
42 (9) (2009) 1185. Clay Construction Slurries (Withdrawn 2018), ASTM International, West
[12] V.C. Li, Sustainability of Engineered Cementitious Composites (ECC) Conshohocken, PA, 2009, www.astm.org.
Infrastructure, in: Engineered Cementitious Composites (ECC), Springer, [44] ASTM C138/C138M-17a, Standard Test Method for Density (Unit Weight),
Berlin, Heidelberg, 2019, pp. 261–312. Yield, and Air Content (Gravimetric) of Concrete, ASTM International, West
[13] W.C. Choi, H.D. Yun, J.W. Kang, S.W. Kim, Development of recycled strain- Conshohocken, PA, 2017, www.astm.org.
hardening cement-based composite (SHCC) for sustainable infrastructures, [45] BS EN 196-1: 2016, Methods of testing cement. Part 1: Determination of
Compos. B Eng. 43 (2) (2012) 627–635. strength.
[14] M. Schneider, M. Romer, M. Tschudin, H. Bolio, Sustainable cement [46] V.C. Li, Advances in ECC research. ACI Special Publications, 206 (2002) 373-
production—present and future, Cem. Concr. Res. 41 (7) (2011) 642–650. 400.
[15] C. Shi, A.F. Jiménez, A. Palomo, New cements for the 21st century: the pursuit [47] V.C. Li, D.K. Mishra, H.C. Wu, Matrix design for pseudo-strain-hardening fibre
of an alternative to Portland cement, Cem. Concr. Res. 41 (7) (2011) 750–763. reinforced cementitious composites, Mater. Struct. 28 (10) (1995) 586–595.
[16] J. Davidovits, Geopolymers: inorganic polymeric new materials, J. Therm. Anal. [48] V.C. Li, S. Wang, C. Wu, Tensile strain-hardening behavior of polyvinyl alcohol
Calorim. 37 (8) (1991) 1633–1656. engineered cementitious composite (PVA-ECC), ACI Mater. J.-Am. Concr. Inst.
[17] A.O. Purdon, The action of alkalis on blast-furnace slag, J. Soc. Chem. Ind. 59 (9) 98 (6) (2001) 483–492.
(1940) 191–202. [49] V.C. Li, C. Wu, S. Wang, A. Ogawa, T. Saito, Interface tailoring for strain-
[18] V.D. Glukhovsky, Soil Silicate Articles and Structures, Russian Budivel’nyk hardening polyvinyl alcohol-engineered cementitious composite (PVA-ECC),
Publish, Kiev, 1967. Mater. J. 99 (5) (2002) 463–472.
[19] L. Vickers, A. Van Riessen, W.D. Rickard, Fire-resistant Geopolymers: Role of [50] P.E. Peterson, Fracture energy of concrete: method of determination, Cem.
Fibres and Fillers to Enhance Thermal Properties, Springer, Singapore, 2015. Concr. Res. 10 (1) (1980) 79–89.
[20] R.E. Lyon, P.N. Balaguru, A. Foden, U. Sorathia, J. Davidovits, M. Davidovics, [51] B.L. Karihaloo, P. Nallathambi, Effective crack model for the determination of
Fire-resistant aluminosilicate composites, Fire Mater. 21 (2) (1997) 67–73. fracture toughness (KICe) of concrete, Eng. Fract. Mech. 35 (4–5) (1990) 637–
[21] J. Davidovits, Geopolymers: man-made rock geosynthesis and the resulting 645.
development of very early high strength cement, J. Mater. Educ. 16 (1994). 91 [52] Mariana Braga, Jorge de Brito, Rosário Veiga, Incorporation of fine concrete
91. aggregates in mortars, Constr. Build. Mater. 36 (2012) 960–968, https://doi.
[22] W.M. Kriven, J.L. Bell, M. Gordon, Microstructure and microchemistry of fully- org/10.1016/j.conbuildmat.2012.06.031.
reacted geopolymers and geopolymer matrix composites, Ceram. Trans. 153 [53] P. He, M. Wang, S. Fu, D. Jia, S. Yan, J. Yuan, et al., Effects of Si/Al ratio on the
(1994) (2003). structure and properties of metakaolin based geopolymer, Ceramics
[23] K. Sankar, P. Stynoski, G.K. Al-Chaar, W.M. Kriven, Sodium silicate activated International 42 (13) (2016) 14416–14422.
slag-fly ash binders: Part I-Processing, microstructure, and mechanical [54] A.S. Rahman, Nanofiber reinforcement of a geopolymer matrix for improved
properties, J. Am. Ceram. Soc. 101 (6) (2018) 2228–2244. composite materials mechanical performance, Colorado State University, 2015
[24] R.A. Fletcher, K.J. MacKenzie, C.L. Nicholson, S. Shimada, The composition (Doctoral dissertation).
range of aluminosilicate geopolymers, J. Eur. Ceram. Soc. 25 (9) (2005) 1471– [55] V.F. Barbosa, K.J. MacKenzie, Thermal behaviour of inorganic geopolymers and
1477. composites derived from sodium polysialate, Mater. Res. Bull. 38 (2) (2003)
[25] F. de Andrade Silva, P.H.R. Borges, A.C.C. Trindade, Evaluation of fiber-matrix 319–331.
bond in the mechanical behavior of geopolymer composites reinforced with [56] J. Davidovits, Geopolymer Chemistry and Applications, Geopolymer Institute,
natural fibers, Adv. Civil Eng. Mater. 8 (3) (2019). 2008.
[26] J.L. Provis, J.S.J. Van Deventer, Geopolymers: Structures, Processing, Properties [57] S. Cho, Geopolymer Composites and their Applications in Stress Wave
and Industrial Applications, Elsevier, 2009. Mitigation, University of Illinois at Urbana-Champaign, 2015 (Doctoral
[27] H.A. Alcamand, P.H. Borges, F.A. Silva, A.C.C. Trindade, The effect of matrix dissertation).
composition and calcium content on the sulfate durability of metakaolin and [58] E.F.C. Pires, C.M.C.D. Azevedo, A.R. Pimenta, F.J.D. Silva, F.A.I. Darwish, Fracture
metakaolin/slag alkali-activated mortars, Ceram. Int. 44 (5) (2018) 5037– properties of geopolymer concrete based on metakaolin, fly ash and rice rusk
5044. ash, Mater. Res. 20 (2017) 630–636.
[28] A.C.C. Trindade, H.A. Alcamand, P.H.R. Borges, F.A. Silva, On the durability [59] G.P.A.G. Van Zijl, F.H. Wittmann (Eds.), Durability of strain-hardening fibre-
behavior of natural fiber reinforced geopolymers, Ceram. Sci. Proc. 38 (3) reinforced cement-based composites (SHCC), Springer Science & Business
(2017) 215–228. Media, 2010. vol. 4.
16 A.C. Constâncio Trindade et al. / Construction and Building Materials 248 (2020) 118558

[60] I. Curosu, Influence of fiber type and matrix composition on the tensile fiber-matrix interaction in composites, Colloid Polym. Sci. 297 (7–8) (2019)
behavior of strain-hardening cement-based composites (SHCC) under impact 1079–1093.
loading, TU Dresden, 2018. Doctoral thesis, ISBN 978-3-86780-555-1. [63] M. Ranjbarian, V. Mechtcherine, Z. Zhang, I. Curosu, J. Storm, M. Kaliske,
[61] A.E. Naaman, G.G. Namur, J.M. Alwan, H.S. Najm, Fiber pullout and bond slip. Locking Front Model for pull-out behaviour of PVA microfibre embedded in
II: Experimental validation, J. Struct. Eng. 117 (9) (1991) 2791–2800. cementitious matrix, Cem. Concr. Compos. 103 (2019) 318–330.
[62] A. Drechsler, R. Frenzel, A. Caspari, S. Michel, M. Holzschuh, A. Synytska, V.
Mechtcherine, Surface modification of poly (vinyl alcohol) fibers to control the