Adhesion studies in view of automated repair using 3D concrete printing

. Similar to dental cavities, in that the damaged area must be filled, concrete also suffers from spall damage during long-term service. Inspired by 3D printing a resin patch that fits the shape of the damaged area in the tooth, an automated approach for concrete repair is proposed, where the damaged area of a concrete structure is filled layer-by-layer. It is therefore essential to ensure the adhesion of the printed materials. In this study, printable mixtures were formulated and the effect of adding redispersible polymer powder and cellulose ether on the adhesion of 3D printed materials was investigated. The adhesion of the material in the fresh and hardened stage was analyzed using a tack test and a pull-off test, respectively.


Introduction
Spalling is a term used to describe delamination of areas of concrete from the substrate, which can be attributed to a handful of reasons such as freeze-thaw cycling, alkali-silica reaction, exposure to fire, and the expansion of corroding steel reinforcement [1][2][3]. Spalling can be hazardous in terms of falling debris or trip hazards. More seriously, spalling will tend to spread if remained unrepaired, so that the structure could become unstable eventually. Therefore, diligent maintenance towards spalling is of great importance to ensure the good performance of concrete structures.
Similar to treating a dental cavity where the affected area is removed and the space is filled with resin or amalgam, repair mortar is manually placed in damaged areas [4,5]. However, this process is time-consuming and labor-intensive. Alternatively, there has been a significant improvement in tooth repair that involves laser scanners and 3D printers [1]. It has been demonstrated that using photogrammetry and 3D printing could repair tooth damage quicker and more accurately than traditional procedures. Potentially, 3D printing using concrete provides an automatic approach to repair spall damage of concrete [6][7][8].
Different from common 3D printed structures that are printed layer by layer from the ground surface, fresh materials should be placed against the deteriorated surface of cavities. It is therefore essential to ensure the adhesion of freshly printed materials, as well as the bond strength after hardening. To enhance the adhesion performance of repair mortars, different kinds of polymers have been used such as redispersible polymer powder (RDP) and cellulose ether (CE). RDP is used as a powder component in dry-mix mortar formulations. When mixed with water, the powder redisperses and forms films in the cement matrix [9,10]. CE is another * Corresponding author: Yaxin.Tao@UGent.be commonly-used polymer in repair materials to obtain some of the required properties such as reducing the absorption of water into the porous substrate and increasing the cement hydration and mechanical strength of the mortar [11,12].
However, the effect of RDP and CE on the adhesion of 3D printable concrete for automatic repair has not been studied yet. Therefore, in this study, the effect of RDP and CE on the adhesion performance of 3D printable concrete in the fresh and hardened stage was evaluated using a tack test and a pull-off test, respectively.
Five mixtures were formulated including one reference mixture (REF), two RDP-modified mixtures (RDP-0.2 and RDP-0.6), and two CE-modified mixtures (CE-0.2 and CE-0.6). The sand-to-binder ratio was fixed as 1 and the water-to-binder ratio was fixed as 0.35 for all mixtures. The dosage of superplasticizer was 0.1% by mass of binder for all mixtures. The addition levels of RDP were 0.2% and 0.6% by mass of binder for mixtures RDP-0.2 and RDP-0.6. The addition levels of CE were 0.2% and 0.6% by mass of binder for mixtures CE-0.2 and CE-0.6. Small mortar batches (1 liter) were prepared in a planetary mixer according to the following protocol. First, superplasticizer was added to the water and manually mixed for 10 s. Next, dry materials including sand and cement were then mixed at 140 rpm for 10 s. Afterwards, water with superplasticizer inside was added to the dry materials and mixed at 140 rpm for 30 s. This was followed by a mixing period of 30 s at 285 rpm. After that, the materials were scraped and allowed to rest for 90 s. Next, the materials were mixed at 285 rpm for 60 s. Finally, RDP or CE was added and mixed at 285 for 60 s. The total mixing period for the small batches was around 290 s.
In addition, big mortar batches (30 liters) were prepared in a pan mixer for 3D printing experiments according to the following protocol. Similar to preparing the small mortar batches, superplasticizer was initially added to the water and manually mixed for 10 s, dry materials including sand and cement were then mixed at 60 rpm for 30 s. Subsequently, water with dissolved superplasticizer was added to the dry materials and mixed at 60 rpm for 180 s. Next, the materials were scraped and allowed to rest for 60 s. Last, RDP or CE was added and mixed at 60 rpm for 60 s. The total mixing period for the big batches was around 340 s.
A tack test was used to analyze the adhesion performance of fresh mixtures (see Fig. 1). A rheometer provided by Anton Paar (MCR 102) equipped with a parallel-plate geometry (diameter 50 mm) was used. A concrete plate with the same diameter as the parallelplate geometry (50 mm) was drilled from a grit-blasted concrete slab and glued to the top plate geometry. In addition, sandpaper (diameter 50 mm, root mean square roughness value 0.18 mm) was fixed to the bottom plate to avoid slippage. The test protocol of the tack test was as follows: first, a plastic circular mold with a height of 20 mm and a diameter of 50 mm was placed on the bottom plate. Following that, the fresh materials from the small mortar batches were placed and compacted in the mold and then demolded. Next, the top plate was moved downwards until a gap of 10 mm was obtained, which was equal to the thickness of one printed layer. After that, the concrete plate was pulled off and the normal force versus displacement curves were recorded. In this study, the test was carried out in a load control mode, where linearly increasing loads were applied to simulate the stepwise increasing loads of the upsidedown 3D printing process. The loading rate was controlled at 0.04 N/s. More details about the test protocol of the tack test can be found in previous work of the authors [13,14]. The test was repeated three times for each mixture.
A pull-off test was performed to evaluate the bond strength of the mixtures in the hardened stage (see Fig.  2). Sand-blasted concrete slabs with a surface-saturateddry state were used as the substrate. Three layers were printed with a speed of 100 mm/s using the large mortar batches. The samples were cured in a controlled environment (temperature 20 ℃, relative humidity 60%) until the age of 7 days. The pull-off test was performed with an automatic bond strength device (Proceq DY-2) and the maximum force required to pull off the core was measured. The test was repeated three times for each mixture.

Results and discussion
Normal stress versus strain curves were obtained from the tack test, as shown in Fig. 3. It can be observed that the fresh materials mainly followed two stages. In the first stage, the normal stress increased from zero to a maximum value, where an inward flow with a small strain was observed. The second stage was characterized by a debonding behavior where the normal stress dropped sharply from the maximum value to zero and the strain suddenly increased. The failure mode of the fresh samples during the tack test was recorded, see two representative failure modes (RDP-0.2 and CE-0.2) in Fig. 4. On the one hand, the adhesive failure, where the failure occurred at the interface between the fresh sample and the concrete substrate, was observed when we tested the reference mixture (REF) and the two RDP-modified mixtures including RDP-0.2 and RDP-0.6. On the other hand, the cohesive failure, where the failure occurred in the bulk of the fresh material, was observed when we tested the two CE-modified mixtures including CE-0.2 and CE-0.6. In previous studies, it was stated that the physical origin of the effort to lift the plates relates to the force to overcome the Laplace depression (or capillary pressure) in the soft material layer, which is generated by the curvature of the meniscus at the material-plate interface [15,16]. Alternatively, another explanation was that the effort required to separate the parallel plates relates to the inwards flow of the soft material and the flow is a kind of entirely radial shear flow in the early stages of measurements [17]. In this study, the same phenomenon for which a small increase in plate separation induced a large inward motion of the fresh material was also observed during the tack test. In total four types of failures of fresh cementitious materials were described including cohesive rupture, adhesive rupture, liquid behavior, and mortar behavior [18]. For paste samples, the failure mode was always found to be cohesive, that is, occurring within the layer of paste, regardless of the type of paste or the type of debonding experiment [19]. In this study, the failure occurred at the interface between the grit-blasted concrete substrate and the fresh mortar sample for the reference mixture (REF) and the two RDP-modified mixtures (RDP-0.2 and RDP-0.6), rather than in the bulk of fresh mortar samples. This indicated that the interface region was a weak point in the chain regarding the adhesion of the fresh mortar mixtures in the fresh stage. The result can be further explained in terms of the presence of sand in the fresh sample which increased the overall shear stress and contributed a lot to the shear resistance [20]. It was pointed out that shear stress in mortar is more complicated than shear stress in paste or mud. The overall shear stress of the fresh mortar can be considered as the sum of the shear stress originating from the yield stress of the cement paste, the flow of the cement paste, the interaction between cement paste and aggregates, and the shear stress resulting from the aggregate movement [20]. Therefore, it seems that the aggregate content has a significant effect on the inward flow of the tested samples, the failure mode including adhesive failure or cohesive failure, and the adhesive strength in the fresh stage. It can be concluded that the flow resistance of the sample took the leading role during the increasing stage. The maximum normal stress was dominated by the adhesion properties at the interlayer position when an adhesive failure occurred, while the maximum value was determined by the shear resistance of the fresh material when a cohesive failure occurred. The CE-modified mixtures presented different normal stress versus strain curves and this can be attributed to the fact that a high dosage of cellulose ether enhanced the adhesion at the interface while it decreased the shear resistance [21][22][23][24].
The maximum normal stress (the adhesive strength in the fresh stage) is summarized in Fig. 5. According to the results, the peak adhesive strength was slightly decreased for the mixtures with the addition of RDP, as compared to the reference mixture (REF). On the other hand, the CE-modified mixtures presented a higher adhesive strength, as compared to the reference mixture (REF). However, for CE-based mixtures, the increase in the CE dosage from 0.2% to 0.6% did not lead to a further improvement of the adhesive strength.
In previous studies, the addition of RDP was pointed out to enhance the bond strength due to polymer film formation in the hardened stage. Normally, it would take a long time for polymer particles to flocculate and form polymer films with the drainage of water among polymer particles [25][26][27]. However, for the adhesive strength in the fresh stage, the period was too short to allow the formation of polymer films.
Different from RDP-modified mixtures, results indicated that CE increased the peak normal stress, which can be attributed to the adsorption effect [28]. After contacting the grit-blasted concrete substrate, CE can adsorb on the substrate surface by hydrogen bonding and specific interactions involving the hydroxyl groups of CE and the concrete surface. This would result in an improved adhesion performance in the fresh stage. The bond strength at the age of 7 days is shown in Fig. 6. It is evident that the addition of RDP enhanced the bond strength, while CE reduced the bond strength in the hardened stage. The bond strength of the mixture RDP-0.6 (2.88 MPa) was increased by 52% compared to the value of the reference mixture RDF (1.89 MPa). Previous studies exhibited that the RDP modification of cement-based mortar and concrete was governed by both the processes of cement hydration and polymer film formation. When RDP was mixed with fresh cement-based materials, the polymer particles can be uniformly distributed. Due to the water drainage along with the development of the cement gel structure, polymer particles were gradually confined and the formed films bound the RDP-modified sample and the concrete substrate with an additional connection [29]. A nearly 10 times increase in bond strength of RDPmodified mortar with a polymer-cement ratio of 20% was reported, compared to that without RDP [30].
Concerning the effect of CE, it was demonstrated that CE is highly hydrophilic and has a high capacity to bind water molecules, increasing their effective volume in solution. This water-capturing function increased the dynamic viscosity of the interstitial solution. As a result, a high dosage of CE would prevent water to move into the porous substrate and the formation of hydration products at the interface region [31].

Conclusions
The effect of redispersible polymer powder and cellulose ether on the adhesion performance of printable concrete for automated repair was studied. A tack test and a pull-off test were performed. According to the results and the discussion, re-dispersible polymer powder presented no significant effect on the adhesion in the fresh stage, while improving the bond strength in the hardened stage due to film formation. In addition, cellulose ether enhanced adhesion at the interface in the fresh stage due to its adsorption effect, while limiting the bond strength in the hardened stage with a dehydrated interface structure.