Construction and Building Materials 67 (2014) 344–352

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier .com/locate /conbui ldmat

Characterization of sustainable bio-based mortar for concrete repair

0950-0618/$ - see front matter � 2014 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.conbuildmat.2014.01.012

⇑ Corresponding author. Address: Delft University of Technology, Stevinweg 1,Room 6.19, 2628 CN Delft, The Netherlands. Tel.: +31 (0)152789405; fax: +31(0)152786383.

E-mail addresses: m.g.sierrabeltran@tudelft.nl (M.G. Sierra-Beltran),H.M.Jonkers@tudelft.nl (H.M. Jonkers), h.e.j.g.schlangen@tudelft.nl (E. Schlangen).

M. Guadalupe Sierra-Beltran ⇑, H.M. Jonkers, E. SchlangenDelft University of Technology, Faculty of Civil Engineering & Geosciences, Department of Materials & Environment, Stevinweg 1, 2628 CN Delft, The Netherlands

h i g h l i g h t s

� O2 consumption was used to characterized bacterial activity in bio-based mortar.� Bio-based agents used in SHCC enhanced its self-healing capacity.� Bio-based SHCC exhibits good mechanical and bonding behaviour as a repair material.

a r t i c l e i n f o

Article history:Available online 5 February 2014

Keywords:Concrete repairBio-based mortarBacteriaCrack healingBond stressRestrained shrinkage

a b s t r a c t

The paper describes mechanical properties, self-healing capacity and bonding behaviour of a sustainablebio-based mortar repair system for concrete. Two different mixes of strain-hardening cement-based com-posites (SHCC) have been used. The bio-based agent added to the SHCC consists of both bacteria and foodfor the bacteria. The metabolic activity of bacteria was monitored by oxygen profile measurements,which reveals O2 consumption by bacteria-based samples, but not by control samples. The mechanicalproperties of the mortar (flexural behaviour, compression strength and drying shrinkage) are evaluated.The bonding behaviour with the concrete substrate is evaluated based on pullout tests and restrainedshrinkage.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Currently available concrete repair systems face durability-re-lated problems that have a huge impact on national economies.In the United States of America, for instance, the annual directcosts for maintenance and repair of concrete highway bridgesdue to corrosion of the reinforcement sum up to 4 billion dollars[1]. Most of the durability-related problems of repair systems aredue to the lack of compatibility with the concrete substrate. A com-bination of physical, chemical and mechanical processes results inthe failure of concrete repair [2,3]. Restrained volume changes dueto drying shrinkage or differential thermal expansion induce sur-face cracking in the repair system and interface delamination be-tween repair and concrete substrate. Additionally, currentlyavailable repair systems are largely based on environmental un-friendly materials such as epoxy systems, acrylic resins or sili-cone-based polymers. This paper focus on the characterization ofa concrete compatible and sustainable bio-based repair system

that features better bonding and improved durability and sustain-ability characteristics compared to existing repair systems.

The European Standards describe the principles and methodsfor protection and repair of concrete structures [4]. The repair sys-tem studied in this paper was designed as a product for concreterestoration by applying or spraying mortar (repair methods 3.1and 3.3), structural strengthening by adding mortar (repair method4.4) or replacing contaminated or carbonated concrete (repairmethod 7.2) [4].

A special type of strain-hardening cement-based composite(SHCC), called Engineered Cementitious Composite (ECC) has beenstudied as repair material for concrete structures for overlayapplications and patch repair [5–7]. ECC was micro-mechanicallydesigned to have a large strain capacity with a low percentage ofrandomly distributed polymer fibres [8]. Because of the presenceof fibres the material develops multiple micro-cracks prior tofailure. The crack width remains below 0.1 mm [9]. In concrete re-pair materials fibres are also used to control drying shrinkage andservice load-related cracking [5]. Repair materials crack whensubjected to differential shrinkage: the early-age shrinkagedeformation of the new repair material is restrained by the oldconcrete substrate that has already undergone shrinkage. Tensilestress is developed in the repair layer and a combination of tensileand shear stresses built up along the interface between the repairlayer and the concrete substrate [5]. When applied as a repair

Table 1Chemical composition (weight%) of the powders materials.

Compound CEM I 42.5 BFS FA LP

CaO 63.3 40.8 7.1 –SiO2 19.5 35.4 48.4 0.3Al2O3 5.6 13 31.4 0.1Fe2O3 2.3 0.5 4.4 0.1MgO 1.1 8.0 1.4 0.2K2O 0.9 0.5 1.6 –Na2O 0.3 0.2 0.7 –SO3 2.7 0.1 1.2 –CaCO3 – – – 98.8

M.G. Sierra-Beltran et al. / Construction and Building Materials 67 (2014) 344–352 345

material, ECC is capable of carrying more tensile load and accom-modate larger tensile strain than other repair systems [6]. Underdifferential shrinkage ECC will suppress localized brittle fracturein favour of distributed micro-crack damage [5,6].

Autogenous healing can be understood as a natural process offilling and sealing cracks without any external operations andworks [11]. Li and Yang [10] investigated autogenous crack healingin ECC. They reinforced ECC with only 0.5% per volume of PVA fi-bres in order to achieve wide cracks. They reported that ECC witha maximum crack width of 50 lm achieved full recovery ofmechanical and transport properties due to autogenous healing.Between 50 and 150 lm, partial recovery could be attained. Autog-enous healing behaviour in SHCC heavily depends on the availabil-ity of unhydrated cement and other supplementary cementitiousmaterials, such as blast furnace slag or fly ash. Low water/bindermaterial ratio and high percentage of binder material and a controlsmall crack width appear to promote autogenous healing in ECC[12,13].

Conventional ECC is designed without coarse aggregates andwith only a small amount of fine sand in order to control the frac-ture toughness of the matrix [8]. This characteristic leads to higherwater to cement/binder ratio and eventually to a high value ofshrinkage [14]. Li [5] and Yang et al. [15] reported drying shrinkagestrain values of 1200 � 10�6 to 1800 � 10�6 for conventional ECC,and Zhou [6] reported a drying shrinkage strain value of2900 � 10�6 for ECC designed using limestone powder instead ofsand as filler. In similar drying conditions of 20 �C and 60% relativehumidity, normal concrete has a drying shrinkage strain of400 � 10�6 to 600 � 10�6 [16]. Conventional ECC uses micro-silicawith maximum grain size of 200 lm [10,17]. In this way the fibresin ECC are only separated by fine aggregate particles which are al-lowed to move freely between the fibres. Particles that are biggerthan the average distance between fibres will cause the fibres toconcentrate in balls and lead to an irregular distribution of the fi-bres [18]. As mentioned above, the particle size is limited in ECCin order to limit the fracture toughness to a level in which the crackinitiation could occur before the tensile load reaches the maximumfibre-bridging stress that will result in failure of the fibre bridges[17].

To improve the durability of the concrete repair system aswell as to improve the bonding with the concrete substrate thispaper proposes a bio-based agent to be included in the ECC mix.The bio-based agent consists of alkali-resistant bacteria and afood source for those bacteria. When applied in concrete, thisbio-based agent has the capacity to produce calcite-based miner-als inside cracks thus reducing the permeability of the concrete[19,20]. The precipitated minerals lead to the closure of crackswith a maximum crack width of 460 lm [20]. In case of surfacecracking in the repair mortar due to restrained shrinkage orother mechanical processes the bio-based agent is capable ofhealing these cracks to ensure the durability of the repairproduct.

Repair materials are frequently subject to limited evaluationsdriven by manufacturers rather than by users [21]. This is due toa lack of widely accepted testing methods. If only the isolatedproperties of repair materials are studied the more importantproperties of the composite system are neglected. In this research,the performance of the composite repair system as well as theinteraction between the repair and concrete substrate have beeninvestigated. Restrained shrinkage tests have been conducted ona simulated repair system, that contains one layer of old concretesubstrate and one layer of new repair material to simulate real re-pair conditions.

For this study, two ECC-type mortars with particles containing abacterial-based agent have been studied as a sustainable bio-basedrepair system for concrete. The mechanical properties of the ECC

materials are evaluated by means of flexural and compression testsand free drying shrinkage measurements. The behaviour of the ECCmaterial as a repair mortar is evaluated by means of restrainedshrinkage measurements and the bonding behaviour with the con-crete substrate is evaluated based on pullout.

2. Experimental investigation

2.1. Materials

Ordinary Portland cement CEM I 42.5N, fly ash (FA), blast furnace slag (BFS),limestone powder (LP), sand (S) have been used. The chemical composition of thepowder materials are given in Table 1 and the mix designs in Table 2. The sandhas an average and maximum grain size of 250 lm and 500 lm respectively. Inthe mix design, BFS, FA and Portland cement are considered as binder materials,and the limestone powder is considered as inert filler material. Polyvinyl alcohol(PVA) fibres with a length of 8 mm and diameter of 40 lm have been used in thecontent at 2% by total volume. These PVA fibres have a tensile strength of1600 MPa. The surface of the fibres is coated with 1.2% oil by weight to optimizethe fibre-cement matrix bond [22].

The bio-based agent considered for this research is an alkaliphilic (alkali-resis-tant) spore-forming bacteria and calcium lactate as a nutrient source for the bacte-ria. Both the bacteria and the food source are embedded in lightweight aggregates.Bacteria from genus Bacillus and more specifically related to the species B. cohnii[23], originally isolated from alkaline soil samples were chosen for this research[24]. Previous studies show that these bacteria are capable of healing cracks by di-rect and indirect calcium carbonate (CaCO3) formation [19,20]. The direct CaCO3

precipitation is due to the bacterial metabolic conversion of calcium lactate accord-ing to the following reaction:

CaC6H10O6 þ 6O2 ! CaCO3 þ 5CO2 þ 5H2O ð1Þ

The indirect formation is due to reaction of metabolically produced CO2 moleculeswith Ca(OH)2 minerals (portlandite) present in the concrete matrix, according to:

5CO2 þ 5CaðOHÞ2 ! 5CaCO3 þ 5H2O ð2Þ

The latter reaction is homologous to carbonation, a slow process that naturallyoccurs in concrete due to inward diffusion of atmospheric carbon dioxide, whoserate is however substantially enhanced by the metabolic conversion of calcium lac-tate [24]. The production of in total six calcium carbonate equivalents is possibledue to the process of bacterial calcium carbonate conversion, which will result inefficient crack sealing.

To prepare the bio-based agent, lightweight aggregates (LWA) were firstimpregnated with calcium lactate (150 g/L) and yeast extract (7.5 g/L) solution fol-lowed by a second impregnation with a bacterial spore solution. Between impreg-nations the LWA were dried in an oven at 37 �C for 5 days. The obtainedimpregnated LWA contains 15% (by weight) calcium lactate and 1.2 � 107 bacterialspores per gram of particles. The LWA particles have sizes ranging between 0.25and 2 mm.

Bacterial spores are known to be able to withstand high mechanical forces andare characterized by a long-term viability of up to 200 years under dry conditions[25]. When mixed directly into concrete the number of viable spore cells in con-crete decrease in time [24]. Later studies by the same author established that bac-teria immobilized in porous expanded clay particles prior to concrete mix additionsubstantially increased bacterially-mediated self-healing in comparison to unpro-tected addition of bacteria [19,20]. Furthermore bacterial spore viability increasedto more than 6 months when added protected inside porous particles to a con-crete mixture [19]. Oxygen consumption measurements provided evidence of bac-terial activity in concrete specimens up to several months after concrete casting[20].

Table 2Mix design of mortars by weight (in kg/m3).

Type Mix 1 control Mix 1BBio-based

Mix 2control

Mix 2BBio-based

Cement 493 461 440 405BFS – – 131 121FA 592 553 558 516LP 394 267 – –Sand – – 440 308LWA – 101 – 97Water 414 385 379 364SP 7 16 7 16PVA 26 26 26 26w/b* 0.38 0.38 0.34 0.35

* Water-to-binder ratio.

346 M.G. Sierra-Beltran et al. / Construction and Building Materials 67 (2014) 344–352

Superplasticizer (SP), Cretoplast SL01 from CUGLA (Breda, The Netherlands),was used to achieve a workable condition of the mixtures. The specimens were pre-pared in a Howard mixer following different mixing sequences for each mix. ForMix 1 first all the powders were pre-mixed for 1 min. Then the water and superp-lasticizers were added and mixed. Later the PVA fibres were slowly added andmixed. For Mix 1B the mixing continued by adding the particles impregnatedwith the bio-based agent and mixing at low speed. Due to the presence of sandin Mix 2 the mixing sequence was modified to assure the proper consistence ofthe paste at the moment the fibres were added. First, 90% of the powders and75% of the sand were mixed for 1 min. Water and superplasticizers were then addedand mixed followed by the PVA fibres. Then the remaining powders and sand wereadded and mixed. The mixing sequence for Mix 2B continued with addition of theparticles with bio-based agent.

The fresh mixtures were cast into moulds with dimensions of240 mm � 60 mm � 10 mm for flexural tests and dimensions of160 mm � 40 mm � 40 mm for compression tests and free drying shrinkage mea-surements. After 24 h the specimens for flexural and compression tests weredemoulded and moist cured in plastic bags at 95% relative humidity (RH),25 ± 2 �C for 7, 28 and 56 days. The specimens for free drying shrinkage were placedin a climate controlled room under conditions of 50 ± 2% RH and 20 ± 2 �C.

2.2. Testing methods

2.2.1. Compression testThe compressive strength of the mortars was determined by testing at least four

40 mm cubic specimens at different ages. The cubic specimens were cut out ofbeams of 160 mm � 40 mm � 40 mm.

2.2.2. Flexural testA four-point bending test was performed under displacement control at a load-

ing rate of 0.01 mm/s. The samples for flexural tests have dimension120 mm � 30 mm � 10 mm. They were cut out of specimens casted with dimen-sions 240 mm � 60 mm � 10 mm. The support span of the flexural loading was110 mm and the loading span was 30 mm. During the flexural tests load andmid-span deflection were recorded. To investigate the self-healing capacity sevenand 28 days old specimens were loaded to a mid-span-deflection of 2.5 ± 0.2 mm.When this deflection was reached the load was released, after which the specimenswere removed and cured in water until age 56 days. At this age the surface crackswere observed by means of a stereomicroscope. Afterwards, the specimens weretested to failure to evaluate the mechanical properties. Specimens with and withoutbio-based agent were kept in separate water containers to avoid cross contamina-tion. Reference specimens from each mixture were cured under the same condi-tions as the pre-loaded specimens and were tested at 56 days.

2.2.3. Oxygen consumption and optical observationsOxygen is consumed by aerobic bacterial metabolic conversion of calcium lac-

tate [20]. In this research optical oxygen microsensors (micro-optodes) have beenused for the quantification of oxygen consumption of samples containing the bio-based agent and for control samples. 3-month old specimens that have been air-cured were placed in water 48 h prior to testing and cut just before the test. Themeasurements were performed on the fresh fracture surface of these samples in or-der to monitor the O2 consumption and there for the bacterial activity. The cut sam-ples were submerged in a solution with pH 11. Oxygen microprofiles weremeasured in vertical steps of 20 lm, from 2.5 mm above towards the surface ofthe specimens in the pH 11-solution at room temperature.

Microanalysis and electron imaging was done with an Environmental ScanningElectron Microscope (ESEM, Philips XL30 Series) equipped with an Energy Disper-sive X-ray (EDAX) element analysing system under hi-vac condition. The integrityof the samples was maintained by impregnation using a low viscosity epoxy undervacuum.

2.2.4. Interface bonding testThe tensile bond strength between the bio-based mortars and a concrete sub-

strate was determined according to EN 12636 [26]. The surface of a 28-day old con-crete samples was first sand-blasted and then a primer was applied (HechtprimerCement from CUGLA). Immediately after, a 12 mm layer of Mix 1B was casted ontop. After 28 days curing a ring groove was drilled through the mortar layer andabout 12 mm into the concrete substrate. Cylindrical steel dollies were then gluedto the mortar. The bonding tests were performed at age 35 days.

2.2.5. Free drying shrinkageDrying shrinkage measurements were performed for all mortars on three

40 mm � 40 mm � 160 mm specimens up to 120 days after an initial curing of1 day in the mould. The drying shrinkage specimens were stored in a drying roomat 20 ± 2 �C and 50 ± 2% RH.

2.2.6. Restrained shrinkageThe restrained shrinkage of the two repair materials were studied by means of a

layered system consisting of a concrete substrate and a repair mortar, as can beseen in Fig. 1. The dimensions of the concrete substrates are 500 � 100 � 100 mmand the thickness of the repair layer casted on-top is 12 mm. The mix designs forall the materials are reported in Table 3. The concrete samples were demouldedafter 24 h and then sealed with plastic and aluminium foil. After 7 days the coverwas removed exposing the samples to RH 50 ± 5% and 20 ± 2 �C. The samples werethen cured under these conditions for 2 years. During this curing time any potentialshrinkage in the substrates occurs. To prepare the concrete samples for the re-strained shrinkage test one side of the samples was roughened with a chisel andsteel brush to remove slurry cement from the surface of coarse aggregates, and laterthis surface was cleaned with high-pressure air. The concrete substrates were thenmoist-cured for 3 days after which the roughened side was dried. After casting a12 mm thick repair layer on top of this contact surface the whole samples were cov-ered with plastic and cured for 24 h. After demoulding, the layered specimens wereplaced in a room with ambient conditions RH 50 ± 5% and 20 ± 2 �C. For each spec-imen two dial gauges were used to record the interface delamination in terms ofinterface vertical separation distance at end locations of the specimens as a functionof drying time. For practical applications such as the one represented in this test,the author found that 2% per volume of fibres could be reduced to 1.7% with onlya minor reduction in the deflection capacity. The lower amount of fibres will im-prove the mixing and ensure a uniform distribution of the fibres in the mortar whenbeing mix at a construction site.

3. Results and discussion

3.1. Compressive strength

The average compressive strength of the four mixes is shownin Fig. 2. The compressive strength of Mixes 2 and 2B are higherthan the strength of Mixes 1 and 1B. This is due to: A lowerwater-to-binder ratio, higher binder content and the presence ofblast furnace slag. Mixes 1 and 1B fulfil the compressive strengthrequirement for repair material of concrete structures Class R3(P25 MPa) according to the standards NEN 1504-3 [27]. Mixes 2and 2B have compressive strengths higher than 45 MPa, fulfillingthe requirement for Class R4 in the same standard. In spite of thefact that the mixes with bio-based agents have a lower amountof cement than the control mixes and that the LWA containingthe bio-based agent are much weaker than the other filler compo-nents the average compressive strength at 7, 28 and 56 days of themixes with bio-based agent is higher than the average compressivestrength of the control mixes. This may be an effect of the presenceof calcium lactate in the bio-based agent. Jonkers and colleagues[24] reported the same effect. Mix 2B has a lower amount of bio-based agent compared to Mix 1B, and the increase of compressivestrength is less noticeable for Mixes 2B compared to Mix 2, than forMix 1B compared to Mix 1. For Mixes 1 and 1B the water-to-binderratio is 0.38, while for Mix 2 it is 0.34 and for Mix 2B it is 0.35. Anincreased water-to-binder ratio may have influenced the averagecompressive strength.

3.2. Flexural behaviour

The ductility and deformation capacity of ECC can be investi-gated either directly by a uni-axial tensile test or indirectly by a

Fig. 1. Layered repair system setup under restrained shrinkage.

Table 3Mix design for layered system by weight (in kg/m3).

Repair mortars Concrete substrate

Type Control Bio-based

CemI 42.5N 483 449 CemIII 42.5 LH HS 340FA 580 538 Sand 836LP 387 260 Gravel 1017LWA – 123Water 420 390 Water 169SP 21 20 SP 0.7PVA 22 22w/b* 0.4 0.4 w/b* 0.5

* Water-to-binder ratio.

Fig. 2. Average compressive strength.

Fig. 3. Typical flexural stress–deflection curves for M1B.

M.G. Sierra-Beltran et al. / Construction and Building Materials 67 (2014) 344–352 347

flexural test. Qian and Li [28] reported that the deflection capacityis linearly correlated with tensile strain capacity for truly strain-hardening materials. Hence, the deflection capacity can be usedfor quality control of SHCC type material due to the simplicity offour point bending test compared to uni-axial tensile test.Therefore the deflection capacity is of major concern for the ECCmortar repair system since its structural application will requirehigh deformation and energy dissipation capacity. The deflectioncapacity herein is defined as the deflection that corresponds tothe maximal flexural stress, i.e. flexural strength in flexuralstress–deflection curve (Fig. 3).

The flexural behaviour of the mortars is illustrated in Fig. 3 andsummarized in Tables 4 and 5. All mixtures show high deflectioncapacity which is typical for SHCC. The addition of bio-based agentresults in a slight decrease of the average flexural strength mea-sured at 56 days (Table 4) from 11.6 MPa for Mix 1 to 11.1 MPafor Mix 1B, and from 12.7 MPa for Mix 2 to 10.4 MPa for Mix 2B.The deflection capacity, on the other hand, increased due to thepresence of particles with bio-based agent (Table 5).

Fig. 3 shows the representative flexural stress–deflection curveof the reloading tests, along with a preloading test curve and a ref-erence test to failure at 56 days for Mix 1B. Note that in this paperfor the reloading curve the permanent residual deflection left from

the preloading stage is taken into account. Reloaded specimens forboth Mix 1 and Mix 1B have a higher strength than the referencespecimens. Different results were obtained with the reloaded spec-imens of Mix 2 that had lower flexural strength than the referencesamples. The results for Mix 2B reloaded specimens were about thesame as the reference samples. The results are slightly better formixes with bio-based agent, but in order to know if this is due tothe healing process it is necessary to study the crack healing bymeans of microscopic observation and oxygen consumption mea-surements in order to understand the healing mechanisms withand without the bio-based agents.

No significant difference was observed regarding the flexuralstrength, deflection capacity, number of cracks or decrease of stiff-ness for samples pre-loaded at an early age (7 days) compared tosamples pre-loaded at 28 days. Stiffness is defined as the secantof the initial rising branch of the stress–deflection curve. For thedata in this study the first point was chosen at 1.5 MPa and the sec-ond point at 3.5 MPa. The decrease of stiffness during the reload-ing, to about half of the initial stiffness for all mixes andpreloading ages, show no influence from the age or preloading.Apparently if the cracking behaviour is important for the self-heal-ing capacity of the material it is the cracking behaviour in the firstseven days that makes the differences as the samples pre-loadedafter either 7 or 28 days do not show a significant difference.

By pre-loading the samples of mixes 1 and 1B the deflectioncapacity decreases, as can be seen in Table 5. A special situationarises with reference samples of Mix 2 tested at 56 days. Thedeflection capacity for all 4 samples tested was considerably lowerthan the pre-loaded samples, and fewer cracks developed prior tofailure. This was not the expected behaviour for a SHCC material.Since the four tested samples were initially 1 specimen, it is possi-ble that for instance this single specimen was not casted properly.A second sample was then cut and tested at age 90 days. Thesespecimens have a similar flexural strength than the specimenstested at 56 days, and a higher deflection capacity along with ahigher number of cracks. The deflection capacity of reloaded sam-ples is then compared to the reference at 90 days in Table 5. Thereloaded specimens have a higher deflection capacity. For mix 2B

Table 4Average flexural stress at pre-load level and flexural strength after healing (standard deviations are indicated between brackets).

Pre-load level After healing

7 days 28 days 56 days Heal/Ref.

Mix 1 Pre-loaded 9.11 [0.58] 11.63 [2.30] 1.0010.6 [0.67] 11.68 [0.82] 1.01

Reference 11.59 [1.28]Mix 1B Pre-loaded 7.74 [0.21] 11.85 [0.21] 1.07

9.13 [0.60] 11.28 [1.19] 1.01Reference 11.12 [1.73]

Mix 2 Pre-loaded 10.21 [0.40] 11.73 [1.47] 0.9211.47 [0.42] 11.12 [0.71] 0.87

Reference 12.74 [1.50]Reference 90d 12.61 [0.99]

Mix 2B Preloaded 7.69 [0.52] 10.44 [0.11] 1.0110.21 [0.64] 10.19 [1.49] 0.98

Reference 10.39 [0.20]

Table 5Mid-span deflection at pre-load level and after healing (standard deviations areindicated between brackets).

Pre-load level After healing

7 days 28 days 56 days Heal/Ref.

Mix 1 Pre-loaded 2.41 [0.07] 4.66 [1.28] 0.642.61 [0.14] 5.69 [1.03] 0.78

Reference 7.30 [1.58]Mix 1B Pre-loaded 2.35 [0.02] 6.61 [0.81] 0.86

2.69 [0.09] 5.42 [0.70] 0.71Reference 7.66 [1.88]

Mix 2 Pre-loaded 2.43 [0.07] 4.81 [1.46] 1.302.44 [0.05] 4.87 [1.25] 1.31

Reference 2.73 [0.15]Reference 90d 3.71 [1.01]

Mix 2B Preloaded 2.28 [0.11] 5.64 [0.21] 0.992.35 [0.16] 7.28 [1.07] 1.28

Reference 5.67 [0.45]

Fig. 4. Healed specimen after reloading.

348 M.G. Sierra-Beltran et al. / Construction and Building Materials 67 (2014) 344–352

the deflection capacity is fully recovered by the reloadedspecimens.

Fig. 5. Measured oxygen concentration microprofiles towards surfaces of sub-merged control and bio-based samples.

3.3. Oxygen consumption and optical observations

Due to the flexural testing method multiple cracks with differ-ent widths were generated. To investigate the surface crack closurethe surface cracks were evaluated with a stereomicroscope. Fig. 4shows typical cracking pattern in a sample of Mix 1B subjectedto 2.5 mm deflection, healing and reloading. The appearance ofcrack closing would be mostly due to the precipitation of calciumcarbonate on the surface of the specimens. Whereas in controlspecimens most of the cracks under reloading pass through thepre-existing cracks, in specimens with bio-based agent somecracks deviate from the healed pre-existing crack and form newcracks. This shows the potential of the bio-based agent to improvethe mechanical properties of healed specimens. To confirm this po-tential oxygen concentration profile measurements were done to-wards the surface of submerged Mix 1 and Mix 1B samples.

Fig. 5 shows the profile measurement for a control Mix 1 and aMix 1B sample after 18 h submerged in solution with pH 11. The O2

profiles of control samples show an almost homogeneous and con-stant O2 concentration in the 2.5-mm water column above thesample surface. In contrast, the O2 concentration of the bio-basedsample progressively decreases in the diffuse boundary layer(0.6-mm water layer overlaying the sample surface). This oxygenconsumption measured above the bio-based agent is due to bacte-rial metabolic activity. This means that 3 months after casting bac-teria impregnated in LWA are viable and can, in the presence of

water, metabolize the organic food source and consume oxygen.Similar behaviour was observed with bacteria immobilized in ex-panded clays [20].

More detailed information of the crack healing can be obtainedby observing preloaded specimens in the Environmental ScanningElectron Microscope (ESEM). The selected samples were not re-loaded. As can be seen in Fig. 6 the LWA particles have a very soundbonding with the cement matrix. The crack breaks the particle andthis will expose the bacteria impregnated in the LWA aggregate tothe water present in the crack in the same way as the sample used

Fig. 7. Photomicrograph and EDS analysis of superficial healing product (crackmouth) in M1B sample.

M.G. Sierra-Beltran et al. / Construction and Building Materials 67 (2014) 344–352 349

for oxygen profiles measurements. Therefore oxygen consumptionand precipitation of minerals is expected in the cracks in speci-mens with bio-based agents.

Figs. 7 and 8 show the microstructure and the chemical compo-sition of healing products found in different cracks in M1B samples.Surface crack sealing took place in some cracks (Fig. 7). It is clearfrom the chemical composition that this healing product formedin the crack mouth is based on calcium but is not only composedof calcium carbonate (CaCO3). Small silica and alumina concentra-tions are present too. Other cracks that were only partially sealedor whose mouths remained opened showed healing products withdifferent morphology inside the internal crack (Fig. 8). Loose formsof healing products (Fig. 8a and b) could be seen in cracks withcrack width about 30 lm, and EDS results revealed a high concen-tration of calcium, silica and alumina. Other cracks were partiallyfilled with needle-like healing product (Fig. 8c) formed with highconcentrations of sulphur, silica and alumina (crack width35 lm). Deeper into the crack, where the crack widths are 20 lmor less, a dense healing product is found (Fig. 8d) with similar com-position to the loose particles seen in Fig. 8a and b. To investigatethe variability of the healing minerals the concentrations weredetermined at different randomly selected spots of multiple spec-imens for which the semi-quantitative results are demonstrated inFigs. 9 and 10. Based on the Si, Ca and S content the healing prod-ucts inside the cracks (similar to those shown in Fig. 8a, b and d)can be considered a mixture between CH (calcium hydroxide)and CSH (calcium silicate hydrate) (Fig. 9). But the high aluminacontent of these products reveals a combination of AFt and Afmphases with a sulphur deficiency (Fig. 10). AFm phases are typifiedby monosulfate [(Ca2Al(OH)6)2SO4 6H2O] and AFt phases by etting-rite [(Ca3Al(OH)6)2(SO4)3 26H2O] [29]. The S deficiency of the AFmphases can be explained if the sulphate was replaced by carbonate

CO2�3

� �or hydroxide (OH�) [29]. On the other hand, the needle-

like minerals seen in Fig. 8c shows strong characteristics of AFtphases and ettingrite combined with CSH (Figs. 9 and 10). The re-sults also reveal that the healing minerals in the mouth of thecracks are a mixture of CaCO3 and CH (Figs. 9 and 10).

In M1 control samples only cracks with a crack width below20 lm exhibited crack healing (Fig. 11). The chemical compositionof healing products inside these cracks is similar to the chemicalcomposition of healing products in samples with bio-based healingagent as can be seen in Figs. 9 and 10.

3.4. Interface bonding strength

The pullout tests were carried out on five samples of Mix 1Bcasted on top of a concrete substrate. The average bonding strength

Fig. 6. LWA inside mortar samples M1B.

at 35 days is 2.89 N/mm2. The failure occurred in all cases in themortar. This bonding strength fulfilled the bonding requirementfor repair material of concrete structures Class R3 (P1.5 MPa)according to the standards NEN 1504-3 [27]. These tests were doneaccording to EN 12636 [26]. The samples were cured under waterprior to testing.

3.5. Free drying shrinkage

The free drying shrinkage measurements for all mortars up to120 days are shown in Fig. 12. Each value in this figure representsthe average of three specimens. The shrinkage strain for Mix 2 islower than for Mix 1. This is due to the higher cement content(Table 2) as cement paste is the source of shrinkage [30]. Fineaggregates like sand and filler materials like limestone powderare inert and therefore help to control the deformations due toshrinkage depending on their particle sizes. Mix 2 has sand as filler,which has a larger particle size than the limestone powder inMix 1.

The mixes with LWA with bio-based agent have a higher dryingshrinkage. It is possible that in Mixes 1 and 2 the absence of LWAlead to the densification of the cement matrix, which preventsinternal moisture evaporation [31]. Nonetheless, Fig. 12 shows thatthe shrinkage strain of mixes Mix 1 and Mix 2 is less than 0.25%after 120 days and below 0.35% for Mix 1B and Mix 2B. This isone order of magnitude lower than its tensile strain capacity(3–6.5%) estimated for these ECC materials based on its mid-spandeflection capacity [28,32]. This implies that any potentialrestrained shrinkage will lead to a strain-hardening stage of theECC material. During this stage because of the material’s ductility,stress caused by shrinkage deformation will form multiplemicro-cracks that do not progressively increase but maintain a

Fig. 8. Photomicrographs and EDS analyses of internal healing products in M1B sample. (a) Crack 2 internal. (b) Crack 3 internal. (c) Crack 4 internal. (d) Crack 5 internal.

350 M.G. Sierra-Beltran et al. / Construction and Building Materials 67 (2014) 344–352

certain maximum crack width. There will be no localized fracturefailure. Once micro-cracked, the effective modulus of the ECCrepair materials will be substantially reduced, as seen in the re-duced slope of the bending stress–deflection curve in the deflec-tion-hardening stage (Fig. 3), thus further suppressing tensilestress build-up in the repair material due to continued restrainedshrinkage [5].

3.6. Restrained shrinkage

The average values of interfacial delamination height at bothends as a function of time are shown in Fig. 13. The control repairmortar completed most of its interface delamination at early ages –within 14 days it reached 90% of the delamination registered at200 days. The bio-based repair mortar continues to delaminate

Fig. 9. Chemical analysis of healing products by EDS. Atomic ratio plot of S/Ca vs Si/Ca. ETT = ettingrite.

Fig. 10. Atomic ratio plot of healing products; the limits of composition used tocharacterized the phases are shown.

Fig. 11. Photomicrograph of internal healing products in M1 sample.

Fig. 12. Free shrinkage strain of repair mortars at different ages.

Fig. 13. Specimen-end delamination heights at different ages.

M.G. Sierra-Beltran et al. / Construction and Building Materials 67 (2014) 344–352 351

up to 60 days. At 28 days the repair registered 90% of the totaldelamination and at 60 days 97%. As can be seen, the control mor-tar repair exhibited interface delamination height 4.5 times higherthan the bio-based mortar. The bio-based mortar has a signifi-cantly greater resistance to delamination due to shrinkage thanthe control mortar, despite its higher free drying shrinkage values.It is possible that the bio-based mortar has a higher creep/stressrelaxation, therefore drying shrinkage does not result in stressand thus not in delamination. Mortar M1B, with a slightly higherbending capacity than control mortar M1 (Table 5), is able to re-lease tensile and shear stresses in the interface so that delamina-tion is lower. This higher capacity may be due to a change in thefibre–matrix interface and a possible coating effect of inert parti-cles, LWA, on the fibre surface. The LWA with the bio-based agentmay be introducing initial flaws that assist in triggering multiplemicro-cracks during tensioning of the composites [33].

4. Conclusions

The main objective of this study was to clarify whether bio-based self-healing via bacteria-mediated calcium carbonate pro-duction will result in improved mechanical properties and bondingstrength of a concrete-compatible fibre-reinforced repair system.Results show that a SHCC type material with bio-based agent fulfilthe requirements for a structural repair material in terms of com-pressive and bonding strength. When applied as a repair materialthe mortar with bio-based agent shows reduced delamination fromthe concrete substrate compared to mortar without the bio-basedagent. Furthermore, after cracking and healing the mixtures withbio-based healing agent show a slightly better recovery of bothflexural strength and deflection capacity from control mixtureswithout bio-based healing agent. Although oxygen measurementsindicate that bacteria were metabolically active in the bio-basedspecimens, observed amounts of calcium carbonate precipitatedid not appear to substantially differ from control specimens.The reason for the apparent uncoupling between metabolic activityand lack of enhanced calcium carbonate precipitation can possiblybe attributed to limited amounts of feed applied and this remainsto be clarified in pending studies.

Acknowledgements

Arjan Thijssen and Ger Nagtegaal are acknowledged for helpwith ESEM analysis and laboratory support, Virginie Wiktor andJure Zlopasa for valuable discussion on the subject. The authorswould like to acknowledge the financial support from STW, Project11342 ‘‘Bio-based repair and performance improvements of agedconcrete structures, Bio-Retrofit’’. The authors would like to

352 M.G. Sierra-Beltran et al. / Construction and Building Materials 67 (2014) 344–352

express appreciation for materials supplied and laboratory testsdone by Cugla B.V. and BAS Research & Technology.

References

[1] FHWA-RD-01-156. Corrosion cost and preventive strategies in the UnitedStates. Report by CC Technologies Laboratories, Inc. to Federal HighwayAdministration (FHWA), Office of Infrastructure Research and Development.2001.

[2] Emberson NK, Mays GC. Significance of property, mismatch in the patch repairof structural concrete, Part 1: properties of repair systems. Mag Concr Res1990;42(152):147–60.

[3] Emmons PH. Concrete repair and maintenance illustrated. WestConshohocken, PA: ASTM International; 1994. pp. 155–164.

[4] NEN-EN 1504-9. Products and systems for the protection and repair ofconcrete structures – Part 9. 2008.

[5] Li M. Multi-scale design for durable repair of concrete structures. PhD thesis.USA: University of Michigan at Ann Arbor; 2009.

[6] Zhou J. Performance of engineered cementitious composites for concreterepairs. PhD thesis. Netherlands: Delft University of Technology; 2011.

[7] Sierra-Beltran MG, Jonkers HM. Bio-based mortar for concrete repair. In: FordeMC, editors. Proceedings of 14th int. conf. structural faults and repair, 2012.

[8] Li VC. From micromechanics to structural engineering – the design ofcementitious composites for civil engineering applications. J Struct MechEarthquake Eng 1993;10(2):37–48.

[9] Li VC. Advances in ECC research. ACI Special publication on concrete: MaterialScience to Applications. SP206-23; 2002. p. 373–400.

[10] Li VC, Yang EH. Self healing in concrete materials. In: Self healing materials: analternative approach to 20 centuries of materials sciences; 2007. p. 161–93.

[11] de Rooij M, Van Tittelboom K, De Belie N, Schlangen E, editors. Self-healingphenomena in cement-based materials. Springer; 2013. p. 1–17.

[12] Qian S, Zhou J, de Rooij MR, Schlangen E, Ye G, van Breugel K. Self-healingbehavior of strain hardening cementitious composites incorporating localwaste materials. Cem Concr Compos 2009;31(9):613–21.

[13] Yang Y, Yang EH, Li VC. Autogenous healing of engineered cementitiouscomposites at early age. Cem Concr Res 2011;41(2):176–83.

[14] Buffenbarger J, Hondo F, Nmai C, Tomita R. Shrinkage-reducing admixtures forconcrete. ACI Concr Int 1998;20(4):31–7.

[15] Yang EH, Yang Y, Li VC. Use of high volumes of fly ash to improve ECCmechanical properties and material greenness. ACI Mater J2007;104(6):303–11.

[16] Neville AM. Properties of concrete. 4th ed. Harlow: Longman; 1995.

[17] Sahmaran M, Lachemi M, Hossain KMA, Ranade R, Li VC. Influence of aggregatetype and size on ductility and mechanical properties of engineeredcementitious composites. ACI Mater J 2009;106(3):308–16.

[18] de Koker D, van Zijl GPAG. Extrusion of engineered cement-based compositematerial. In: Proceedings of 6th Rilem Symposium on Fiber-ReinforcedConcretes (FRC)-BEFIB; 2004.

[19] Jonkers HM. Bacteria-based self-healing concrete. Heron – Self-healing Mater2011;56(1/2):1–12.

[20] Wiktor V, Jonkers HM. Quantification of crack-healing in novel bacteria-basedself-healing concrete. Cem Concr Compos 2011;33:763–70.

[21] Vaysburd AM, Brown CD, Bissonnette B, Emmons PH. ‘‘Realcrete’’ versus‘‘Labcrete’’ – searching for tests that give reliable results. Concr Int2004;26(2):90–4.

[22] Li VC, Wu C, Wang S, Ogawa A, Saito T. Interface tailoring for strain-hardeningpolyvinyl alcohol-engineered cementitious composite (PVA-ECC). ACI Mater J2002;99(5):463–72.

[23] Spanka R, Fritze D. Bacillus cohnii sp. nov., a new, obligately alkaliphilic, oval-spore-forming bacillus species with ornithine and aspartic acid instead ofdiaminopimelic acid in the cell wall. Int J Syst Bacteriol 1993;43(1):150–6.

[24] Jonkers HM, Thijssen A, Muyzer G, Copuroglu O, Schlangen E. Application ofbacteria as self-healing agent for the development of sustainable concrete.Ecol Eng 2010;36:230–5.

[25] Schlegel HG. General Microbiology. 7th ed. Cambridge: University Press; 1993.[26] NEN-EN 12636. Products and systems for the protection and repair of concrete

structures – test methods – determination of the adhesion concrete toconcrete; 1999.

[27] NEN-EN 1504-3. Products and systems for the protection and repair ofconcrete structures – Part 3; 2005.

[28] Qian S, Li VC. Simplified inverse method for determining the tensile straincapacity of strain hardening cementitious composites. J Adv Concr Technol2007;5(2):235–46.

[29] Lachowski EE, Mohan K, Taylor HFW, Moore AE. Analytical electronmicroscopy of cement pastes: II, pastes of portland cements and clinkers. JAm Ceram Soc 1980;63(7,8):447–52.

[30] Brandt AM. Cement-based composites. Routledge Taylor and Francis Group;2009.

[31] Ozbay E, Karahan O, Lachemi M, Hossain KMA, Duran Atis C. Investigation ofproperties of engineered cementitious composites incorporating high volumesof fly ash and metakaolin. ACI Mater J 2012;109(5):565–71.

[32] Zhou J, Qian S, Sierra Beltran MGM, Ye G, van Breugel K, Li VC. Development ofengineered cementitious composites with limestone powder and blast furnaceslag. Mater Struct 2010;43(6):803–14.

[33] Lawler JS, Zampini D, Shah SP. Permeability of cracked hybrid fiber-reinforcedmortar under load. ACI Mater J 2002;99(4):379–85.