An Electronic Textile Embedded Smart Cementitious Composite

Structural health monitoring (SHM) using self-sensing cement-based materials has been reported before, where nano-fillers have been incorporated in cementitious matrices as functional sensing elements. A percolation threshold is always required in order for conductive nano-fillers modified concrete to be useful for SHM. Nonetheless, the best pressure/strain sensitivity results achieved for any self-sensing cementitious matrix are<0.01 MPa-1. In this work, we introduce novel reduced graphene oxide (RGO) based electronic textile (e-textile) embedded in plain and polymer-binder-modified cementitious matrix for SHM applications. As a proof of concept, it was demonstrated that these coated fabric-based sensors can be successfully embedded within the cement-based structures, which are independent of any percolation threshold due to the interconnected fabric inside the host matrix. The piezo-resistive response was measured by applying direct and cyclic compressive loads (0.1 to 3.9 MPa). A pressure sensitivity of 1.5 MPa-1 and an ultra-high gauge factor of 2000 was obtained for the system of the self-sensing cementitious structure with embedded e-textiles. The sensitivity of this new system with embedded e-textile is many orders of magnitude higher than nanoparticle based self-sensing of cementitious composites. The manufactured e-textile sensors showed mechanical stability and functional durability over long-term cyclic compression tests of 1000 cycles.


Introduction
Civil infrastructure (dams, bridges, tunnels, road network etc.) is the backbone of our societal and economic growth. Concrete and the related cement-based materials are the construction industry's favourites for a variety of reasons: (i) ease and cost of construction compared to alternatives (e.g. steel); (ii) robustness for a variety of exposure scenarios; (iii) ability to construct a large variety of complex geometries and (iv) excellent mechanical performance [1].
These materials have excellent response in compressive loads and typically display quasi-brittle behavior. To be used for infrastructure projects the presence of steel reinforcement is essential.
Steel alleviates the quasi-brittle behavior of concrete and the new composite (reinforced concrete) can sustain well both compressive and flexural loads. The exposure of cement-based infrastructure to a vast number of degrading environments throughout their service life cannot be prevented. The integrity of concrete and cement related materials largely depend on their ability to withstand mechanical and environmental weathering. It is the coupled effect of phenomena such as impact, service loading, chloride/CO 2 concentration, pressure/thermal differentials, freeze-thaw and sulfates that can cause severe damage, hence reducing functionality [2]. Damage can manifest itself by localised weakening of the material and can progress in the form of microcracks that gradually reduce the integrity of structures and elements. Under continuous mechanical and environmental stress these microdefects coalesce and expand compromising the mechanical properties of materials. In order to monitor the in-service integrity of the cementbased structures, structural health monitoring (SHM) is deemed vital especially for the safety critical components [3]. SHM can provide real-time data about the condition of a structure by using suitable sensors. This will allow for timely intervention in critical situations minimizing considerably the maintenance regimes and extending the service life of an infrastructure asset.
Conventionally, these sensors are either embedded or applied on the external surfaces of the structure [4]. A number of sensing techniques including, foil strain gauge [5], optical fiber [5,6], piezoelectric ceramic [7,8] and shape memory alloys [7] have been studied for SHM of cementbased constructions. Such sensors are usually incompatible with cementitious materials and reduce the strength and durability of the structure [9].
Various polymers such as urea-formaldehyde resin, unsaturated polyester resin, methylmethacrylate, epoxy resin, furan resins, polyurethane resins and waste tire rubber have been used by researchers to create polymer-modified mortar (PMM)/polymer-modified concrete (PMC) [43][44][45][46][47]. There are two major reasons for adding polymeric materials to the cementitious matrix: (i) to re-use the waste polymer; and (ii) to change specific properties of the cement-based materials e.g. density, fatigue life, toughness, brittleness, and moisture absorbance [44]. The effect of polymer modification on the piezo-resistivity of self-sensing polymers has not been reported.
In this work, we present the application of RGO-coated Nylon textile as the piezo-resistive strain-sensing element in a cement-based system. A comparison was also made between matrices with and without polymer modification. It was shown that the RGO-coated piezo-resistive fabric provides a novel route for making self-sensing cement-based materials and the addition of polymer in the matrix improves the sensing and bonding capabilities of the structure to a great extent.

Preparation of RGO coated Nylon fabric
The e-textile was prepared by coating graphene oxide (GO) on a commercially available Nylon fabric, supplied by Gurit ® , as a substrate. The Nylon has been chosen as an example in this study, however, any textile fabric compatible with GO can be used as demonstrated in our earlier studies (Ali et al. 2017). For the coating process, the pristine fabric was soaked in a GO solution.
The GO, supplied in the form of aqueous paste, was obtained from Abalonyx AS, Norway. The aqueous acidic GO paste contains 25 % GO, 74 % water and 1-1.5 % HCl by weight. The GO solution was prepared by diluting aqueous acidic GO paste of concentration about 100 mg/ml to a concentration of 2.5 mg/ml. The solution was sonicated in a bath sonicator at a frequency of 50 Hz for 30 minutes with the bath temperature maintained at 47°C. A sheet of the Nylon ply was soaked in the solution for 24 hours and then dried in a controlled environment at 80 °C for 5 hours. Once the GO was deposited on the samples, it was reduced by heating under the same controlled environment at 170 °C for 24 hours. The coating process is repeated until uniform coating throughout the fabric was achieved. The reduced graphene oxide (RGO) coated fabrics exhibit a smooth coating that wraps every fiber. The appearance of the coated fabrics is like that of a shiny grey fabric (see Fig. 1). Under a scanning electron microscope (SEM), the coating looks very smooth, adhering well to the fibers' surface (see Fig. 1).

Compression Testing Sample Preparation
The prepared e-textiles were subsequently embedded in plain and polymer modified cementitious matrices. For the preparation of the matrices a slurry was prepared using plain white cement obtained from a local supplier (UltraTech®). The cement was thoroughly mixed with water in in a water to cement ratio of 0.67 (water to cement ration of 2:3) to form the slurry.
The samples were prepared by pouring the slurry in metallic molds (200x10x5 mm) in two stages. Initially, the slurry was poured to the middle of the mold, then a small strip of the RGO coated fabric was placed on it and subsequently the mold was filled up to the top. The resulting product was a sandwich structure. The molds were then placed in an oven at 50 C and ambient humidity (20%) to accelerate the setting of the cementitious matrix. For the polymer-modified samples, 2% by weight Sil-Poxy™ Silicone Rubber Adhesive (Smooth-On) was mixed with the cement and water. The samples were prepared in the same exact way as described for above. The process is shown schematically in Fig. 1b & S2b.

Compression Testing
An Instron 5969 universal testing machine with a 2 kN static loading capacity, was used for the mechanical compression tests. Each sample was placed at the bottom stationary platen of the testing frame. The top section was an indenter with rectangular cross-section of 50x10 mm. Two electrical connections were taken along the shorter side of the sample by applying copper strips using silver paste (PECLO® conductive silver paint from TedPell®). A linear force profile with an increment of 10 N/s was applied to the samples to a maximum load of 1.95 kN. After reaching the target maximum load, the unloading cycle was initiated. The electrical current was recorded using an electrochemical workstation (Autolab 302 N), and the load vs. deflection curve was recorded using the Instron data acquisition system. The current values were later converted to resistance using Ohm's law and subsequently fractional change in resistance was obtained. A fixed voltage of 1V was applied and the corresponding current was recorded during loading and unloading periods. The experimental set-up is shown schematically in Fig. 1c & S3.

Piezoresistive Response
The e-textile embedded pure and polymer-modified cementitious samples were subjected to a loading-unloading cycle at a maximum compressive stress of 3.9 MPa (1950 N). The fractional change in resistance (FCR) as a function of stress for loading and unloading regime is shown in Figure 2a. An FCR of ~35% is recorded with an application of a compressive stress of 3.9 MPa for the plain sample. For the polymer modified matrix, the FCR is ~60% for the same stress. This

°C
Sil-Poxy is attributed to better transfer of load to the e-textile in the polymer-modified composite. The pressure sensitivity and the corresponding strain sensitivities (gauge factors) for these measurements can be estimated by the following equations [48]: Pressure sensitivity (PS): Gauge factor (GF): The pressure sensitivity (PS), estimated using equation (1), for a compressive strength of 3.9 MPa for the plain matrix is about 0.1 MPa -1 and for that of polymer-modified specimen is 0.15 MPa -1 . This pressure sensitivity is at least 25 orders of magnitude better than state-of-the-art particulate matter reinforced smart concrete [49]. The maximum compressive strength of 3.9 MPa corresponds to a maximum compressive strain of approximately 0.03% [50,51]. The gauge factor (GF), as estimated using equation (2), for the plain cementitious matrix is > 1100, which is at least two orders of magnitude and 30 times better than best reports on self-sensing concrete based on conductive particulate matter reinforced cement-based composites [52].
The plain cementitious sandwich structure with embedded e-textile performs better in terms of pressure and strain sensitivity than the polymer-modified one. Nonetheless, the response for the polymer-modified samples is smooth and the curve follows an elliptical locus during the loading and unloading. The corresponding behavior of the non-modified matrix possesses a relatively linear response. Cement is a material rich in minerals such as calcium silicates, aluminates and aluminoferrites. These minerals when in contact with water they undergo a chemical reaction, known as hydration, and with the process of time they transform from a powder form to a network of solid fibrous crystals [53]. The polymer modification of the cementitious matrix impacts the evolution of this cementitious crystalline network. The viscoelastic nature of polymer chains would reduce the brittleness of the bulk structure and would increase the toughness of the bulk material. A schematic representation of the material during loadingunloading cycle is shown in Figure 2b. It is envisaged that the presence of the polymeric chains in the matrix would affect the load transfer to the fabric and hence, the piezo-resistive response.
The corresponding to a pressure sensitivity of 1.5 MPa -1 for the polymer-modified composite. The compressive stress of 0.1 MPa in the plain cementitious matrix causes a change in resistance of ~2% corresponding to a pressure sensitivity of 0.2 MPa -1 but a gauge factor of ~2000 as the strain in such low pressures for such matrix is a maximum of 0.001%. This is due to the new system of embedding the e-textile directly inside the cementitious matrix causing a maximum load transfer to the e-textile at smallest overall strain in the host matrix. To an extent the embedded textile operates like a spinal cord within the composite, sensing the stress and feeding the information to a processing unit.
The changes in resistance as a result of the applied load can be explained via two main mechanisms, namely percolation theory and quantum tunneling effect theory. When subjected to compressive strains, the number of RGO-to-RGO contacts increases to form more conductive paths, and the gaps between the RGO particles decrease, leading to the manifestation of the tunneling effect, thereby causing a decrease in electrical resistance leading to the piezo-resistive phenomenon [1,54]. Also, there is variation in the shape of the peaks of FCR between plain and polymer-modified matrices highlighting possible differences in the way the electrical current is passing through the material. The FCR response as a result of applied compressive stress is shown in Figure 2a

Modeling cyclic response of piezoresistive materials
In this study, we borrowed the concept of circuit theory and employed the analogy of fault This resistance change expression is a generalized expression to represent the decay of resistor network with multiple time decay constant subjected to any periodic cyclic loading. The first term represents the resistance decay in a piezoresistive material where multitude decay processes due to decreasing contact resistance give rise to dispersion of the decay time. Where ! is the initial resistance, ! is the dimensionless material constant representing the weighted contribution of each parallel resistor branch towards the total resistance decay and τ ! represents

Conclusion
The use of RGO based e-textile embedded in cementitious matrices has been successfully demonstrated. It was shown that the novel e-textiles possess sensitivities as high as 1. Raman spectra for GO and partially reduced RGO are shown in Figure S1a. Two fundamental vibrations can be observed in the range of 500 and 2500cm -1 . From Lorentzian fitting of the curves it can be seen that the D vibration band, which is formed from a breathing mode of j-point photons of A 1g symmetry is at 1356 and 1358 cm -1 for GO and RGO, respectively. G vibration band appeared at 1592 for GO and 1589 cm -1 for RGO from first-order scattering of E 2g phonons by sp 2 carbon. The D and G bands in the Raman spectrum depict the disorder bands and the tangential bands, respectively [ i , ii ]. The roughness measure ( Figure S1b) from atomic force microscopy topographical image ( Figure S1c) shows a thickness of about 3nm corresponding to about 4-5 layers of GO sheets with the edges folded for up to 8nm. The folding at the edges is due to the large size of the sheets. Figure S1: Raman spectra of graphene oxide (GO) coated and partially reduced graphene oxide (RGO) coated fabric

D Band G Band
Fabrication process of RGO coated e textile: For the fabrication of RGO coated e-textile, the graphene oxide (GO) solution was prepared by using aqueous GO paste (Abalonyx AS, Norway). The aqueous acidic GO paste contains 25 % GO, 74 % water and 1-1.5 % HCl by weight. The GO solution with a concentration of 2.5 mg/ml was prepared by diluting aqueous acidic GO paste to a concentration of 10 % by weight which corresponds to a GO. The solution was sonicated in a bath sonicator at a frequency of 50 Hz for 30 minutes with the bath temperature maintained at 47 o C. A sheet of the Nylon fabric (Gurit®) was soaked in the solution for 24 hours and then dried in a controlled environment at 80 °C for 5 hours. Once the GO was deposited on the samples, it was thermally reduced by heating the textile fabric at 170 °C for 24 hours in an oven. The process is shown schematically in Fig. S2a.
Manufacturing of e-textile embedded cement structure: For manufacturing the e-textile embedded cement samples for compression testing, first the strips were cut from RGO coated e textile and electrical connections were made on edges of the strips by attaching copper strips using silver paste (Product 16062, PELCO® Conductive Silver Paint, TedPella). Once the etextile strips were prepared a slurry of cement and water as prepared by mixing. Initially, the slurry was poured in the mold (200×10 mm) and a small strip of the RGO coated fabric was placed on the wet cement mixture. Additional slurry was poured on the top surface to form a sandwich structure. The mold was then placed in an oven for fast curing of the cement. In the second set of samples 2% by weight Sil-Poxy™ Silicone Rubber Adhesive (Smooth-On) was mixed with the cement and the samples were casted in the same way as described for the plain cement. The process is shown schematically in Fig. S2b.  Piezo-resistive response of the e-textile embedded cement samples was obtained by applying the compression loading using an Instron 5969 universal testing machine ( Figure S3a) with a 2 kN static loading capacity and recording current using Autolab 302 N electrochemical workstation ( Figure S3b). A fixed voltage of 1V was applied during the experiments. The current values were later converted to resistance using Ohm's law and subsequently fractional change in resistance was obtained. Figure S3 Experimental setup used to obtain the piezo-resistive response.
These graphs shown in Figure S4  Here, instead of compressive stress (MPa), the compressive force (N) is shown in these graphs to show the magnitude of compressive forces used in this work.
The fractional change in resistance (FCR) as a function of force in Newton (N) for loading and unloading regime is shown in Figure S4a. An FCR of ~35% is recorded with an application of a compressive force of ~1800N for the plain cement. For the polymer modified cement, the FCR is ~60% for the same stress. This is due to better transfer of load to the e-textile in the polymer modified cement composite than that of the plain cement. The force sensitivity (N -1 ) can be calculated with the following expression: Force sensitivity (FS):

=
The force sensitivity (FS), estimated using equation (i), for a compressive load of 18 N for plain white cement embedded e-textile is about 0.20 mN -1 and for that of polymer-modified e-textile embedded cement is 0.33 mN -1 . The FCR response of the non-modified e-textile embedded cementitious matrix to applied load is about linear. However, the FCR response of the polymermodified e-textile embedded cementitious matrix is non-linear due to the presence of a viscoelastic polymer. The FCR response of polymer-modified structure is more prominent due to a more pronounced transfer of load enabled by polymeric chains. The current vs voltage (I/V) curves correspond well with the results of Figure S4a as the change in current with applied load is also more prominent for polymer-modified structure ( Figure S4c) than that of the nonmodified ( Figure S4b). Cyclic FCR responses to different loads (50-500N) are shown in Figure  S4d & e for polymer-modified and non-modified structures respectively. Figure S4 (a) FCR response to direct loading and unloading of e-textile embedded pure cement and polymermodified cement. (b) IV curves of e-textile embedded pure cement (b) and polymer-modified cement (c) as a function of different cyclic compressive loads. FCR response of e-textile embedded (d) polymer-modified cement (e) and non-modified cement as a function of different cyclic compressive loads.