Keywords

1 Introduction

Aggressive ions such as sulfates (\({\mathrm{SO}}_{4}^{2-}\)) are naturally available in the surrounding aquatic media of concrete infrastructures. Moreover, these anionic species can be liberated from the underground environments or internal sources by moisture ingress (chemical origin) or can be produced through the metabolism of microorganisms (biogenic origin). The main sources of \({\mathrm{SO}}_{4}^{2-}\) anions are sulfuric acid (\({\mathrm{H}}_{2}{\mathrm{SO}}_{4}\)), magnesium sulfate (\({\mathrm{MgSO}}_{4}\)), and sodium sulfate (\({\mathrm{Na}}_{2}{\mathrm{SO}}_{4}\)), which cause severe degradation of concrete material over long-term exposure [1,2,3,4]. An innovative idea regarding prevention of concrete material deterioration is the development of a sensor to distinguish \({\mathrm{SO}}_{4}^{2-}\) species coming from identical [5] or non-identical sources or having different concentrations/quantities.

Therefore, in the present study, carbon nanotube (CNT) alkali-activated sensors are proposed for \({\mathrm{SO}}_{4}^{2-}\) species sensing and discriminating released from \({\mathrm{H}}_{2}{\mathrm{SO}}_{4} \& {\mathrm{MgSO}}_{4}\), for the first time to the best of our knowledge. The sensors were fabricated from a sodium-based fly ash ground granulated blast-furnace slag (GGBS) alkali-activated material and CNTs. As the first step, the percolation threshold of the sensors was determined by measuring the electrical resistance (converted to conductivity) of CNT alkali-activated nanocomposites, incorporating different content of CNTs. To study the \({\mathrm{SO}}_{4}^{2-}\) sensing potential and transitional behavior of the nanocomposites from insulating to conducting mode, the sensors were fabricated to incorporate different CNT concentrations. The assessments were carried out by introducing \({\mathrm{H}}_{2}{\mathrm{SO}}_{4} \&\mathrm{ Mg}{\mathrm{SO}}_{4}\) with identical concentration. Finally, the sensors were fabricated with the percolated CNT concentration and evaluated by introducing different volumetric quantities of \({\mathrm{H}}_{2}{\mathrm{SO}}_{4}\).

2 Methods

2.1 Materials

The applied CNTs were TUBALL™, supplied by OCSIAL Europe and their properties can be found in Davoodabadi et al. [5, 6]. The surfactant was a technical grade of sodium dodecylbenzenesulfonate (SDBS) produced by Merck KGaA. The utilized precursors were fly ash (Steag Power Minerals GmbH) and GGBS (Opterra GmbH); their properties are described in Davoodabadi et al. [7]. Sodium disilicate powder (Sikalon; Wöllner GmbH) was used to activate and geopolymerize the blend. Sulfuric acid and magnesium sulfate for the sensing measurements were diluted to 0.1 M from a stock solution of 98% ACS reagent sulfuric acid (Merck KGaA), and anhydrous magnesium sulfate powder (Merck KGaA), respectively.

2.2 Methods

The nanofluids were prepared by ultrasonication of the CNTs and SDBS in ultrapure water according to the procedures in Davoodabadi et al. [5,6,7]. The CNT concentration range spanned from 0.010 to 1.000 wt% of oxide mass and SDBS was added with the same mass as CNTs. The nanocomposites were fabricated with the formulation in Davoodabadi et al. [5,6,7]. The mixed slurries were cast into plastic molds with slot dimensions of 60 × 10 × 10 mm3 for 24 h. After demolding, the nanocomposites were cured in the chemical laboratory ambient conditions. Thereafter, the nanocomposites were heat treated (at 105 °C for 24 h) to eliminate any negative impact of water on the CNTs’ conductive network.

2.3 Characterizations

A programmed setup was used for the electrical properties and sensing measurements composed of a KEYSIGHT B2912A precision source/measure unit (SMU), a computer, and Grafana time-series database [5]. The received data in the form of the resistance \((\mathrm{R})\) in Ω were converted to resistivity (\(\uprho =\mathrm{R}.\mathrm{A}.{\mathrm{L}}^{-1}\)) in Ω.m and conductivity (\(\upsigma ={\uprho }^{-1}\)) in S.m−1 by applying the sensors’ cross-sectional area (A) and length (L). For further analysis, relative resistance \((\mathrm{RR}=100. \left({\mathrm{R}}_{1}-{\mathrm{R}}_{0}\right).{\mathrm{R}}_{0}^{-1})\) was used for sensor evaluations. A GeminiSEM 500 (Carl Zeiss QEC GmbH) was used for scanning electron microscopy imaging of the specimens’ cross section. The FEI Tecnai F30 (ThermoFisher Scientific) was used to conduct high resolution transmission electron microscopy (HRTEM imaging) of the samples.

3 Results and Discussion

3.1 Percolation Threshold

In the available literature, the sensing properties of geopolymers exploited the electrolytic (ionic) conductance of the composites because of their ion-rich structure and pore system [8,9,10,11,12]. However, investigated alkali-activated composites in this study are insulators and exhibit insufficient conductive (electronic) properties to act as a sensor. The measured inherent resistances of these composites were unsteady, in the range of mega ohm (≥10 MΩ). Because measurements are conducted on heat-treated specimens, the electrolytic conductivity of the ion-rich framework is assumed to be negligible. Sufficient inclusion of CNTs causes the resistance range of nanocomposites to descend abruptly from mega ohm (highly insulator character) to hundreds of ohms (highly conductive behavior). This proves that the electronic conductive network of CNTs has been generated, and the nanocomposites are percolated [13].

The correlation curve of the nanocomposites’ conductivity and CNTs’ concentration is depicted in Fig. 1. The observed increasing trend of conductivity corresponded to the required quantity of CNTs for establishing a functional conductive network, mostly in tube-contacting mode rather than the tunneling or hopping mode [13]. This conductive network of CNTs provides the alkali-activated matrix with a percolating transition zone between 0.070 wt% and 0.200 wt% as indicated by the inset in Fig. 1. With respect to percolation theory, such curves can be fitted with power regression models [14,15,16,17]. In addition to the investigated nanocomposites, CNT Portland cement-based nanocomposites have a relatively similar percolation trend, but their documented thresholds exhibit, naturally, non-conclusive values. Some of the documented percolating transition zones of CNT Portland cement-based nanocomposites are shown in Table 1 for comparison.

Fig. 1
A line graph of conductivity, siemens per meter, versus single-walled carbon nanotubes, weight percent, depicts an increasing curve. A dashed rectangle is drawn around the plots slightly above and below (0.1, 10 to the power of 1).

Percolation diagram of CNT alkali-activated nanocomposites (concentrations (wt%) between 0 and 0.25 are 0.010, 0.025, 0.050, 0.075, & 0.100)

Full size image
Table 1 Reported percolating transition zones of CNT cementitious nanocomposites
Full size table

The most influential parameters on the percolation of nanocomposites are the CNTs’ structure (i.e., chirality, aspect ratio, waviness, and the number of shells) and CNTs’ tunneling resistance/distance. The fabrication methodology (applied surfactant, ultrasonication energy, mixing, and curing), which determines the 3D orientation, configuration, interconnection of CNTs, and geometry of agglomerates, further affects the percolating character of the nanocomposite [15, 16, 18,19,20,21]. Considering the percolation threshold as an outset, the preference of composite researchers is to maintain the nano-additives’ concentration as low as possible (i.e., low or ultra-low percolation thresholds), because high concentrations of nano-additives have a destructive impact on the microstructure and consequently on the mechanical properties of the nanocomposite [14, 22, 23] (Table 1).

Based on these arguments, to achieve a firm conductive network of CNTs, in which the nanocomposite is not so overloaded as to be destroyed by the additives and not too under-loaded to be an insulator, a concentration of 0.1 wt% was considered for the sensor fabrication. In Sect. 3.2. Sulfate discrimination, the selection of CNT concentration of 0.1 wt% as the percolation threshold is further validated. In comparison, a wide concentration range of 0.05–2 wt% was reported in the literature for sensing and smart applications of the Portland cement-based nanocomposites [27,28,29,30,31,32,33,34].

3.2 Sulfate Discrimination

The CNT alkali-activated nanocomposites sensing responses upon exposure to 90 µL (30 µL in each cycle) of 0.1 M \({H}_{2}{SO}_{4} \&\) \({MgSO}_{4}\) with CNT concentration ranging from 0.0 to 1.0 wt% are illustrated in Fig. 2. The response of the sensors upon exposure to different sulfate containing regimens was explored and compared. A range of CNT concentrations were considered to confirm the influence of the generating conductive percolated network of CNTs on the sensing behavior of the nanocomposites with respect to percolation analysis.

Fig. 2
Nine graphs of relative resistance percent versus time seconds depict 2 fluctuating curves. The upper curve represents sulfuric acid, and the lower curve represents magnesium sulfate. The single-walled carbon nanotube concentrations are 0, 0.01, 0.025, 0.05, 0.075, 0.1, 0.25, 0.5, and 0.75.

The sensing responses of CNT alkali-activated nanocomposites exposed to 90 µL (30 µL in each cycle) of 0.1 M \({H}_{2}{SO}_{4} \& Mg{SO}_{4}\); CNT concentrations (wt%) in the nanocomposites are a 0.000; b 0.010; c 0.025; d 0.050; e 0.075; f 0.100; g 0.250; h 0.500; i 0.750

Full size image

The nanocomposites with incorporation of 0.0 and 0.010 wt% exhibited fully insulating behavior (Fig. 2a, b). The responses were highly noisy and stochastic as expected from the ultra-low conductivity of these nanocomposites; the approximate magnitude was 3E-5 S·m–1 as shown in Fig. 1. With CNT incorporation of 0.025 wt% and 0.050 wt%, the conductivity rose sharply to almost 3E-2 and 1 S·m–1, respectively, and the responses were slightly functional but still not accurate (Fig. 2c, d). Considering the onset of the percolating area, at CNT concentration of 0.070 wt% the functional response of the sensors began at 0.075 wt% with sensor conductivity of ≈6 S·m–1 (Fig. 2e). From this point onwards, the CNT signals showed a regular configuration, which corresponded to the normal behavior of pristine p-type CNTs (Fig. 2e–j). The \({\mathrm{H}}_{2}{\mathrm{SO}}_{4} \&\) \({\mathrm{MgSO}}_{4}\) discrimination mechanism in the percolated area was mostly based on the difference in the signal shape and magnitude.

The correlations of maximum relative resistance and CNT concentration are plotted in Fig. 3. The main results to note were (i) the positive relationship of the relative resistance and CNT concentration, and (ii) the higher sensitivity of the nanocomposites towards \({\mathrm{H}}_{2}{\mathrm{SO}}_{4}\) exposure than \({\mathrm{MgSO}}_{4}\). Result (ii) means the sensors can discriminate \({\mathrm{SO}}_{4}^{2-}\) species introduced from two different sources (\({\mathrm{H}}_{2}{\mathrm{SO}}_{4} \&\) \({\mathrm{MgSO}}_{4}\)), by means of relative resistance as demonstrated in Fig. 3, and by signal configuration and shape as illustrated in Fig. 2. The CNTs’ signal differentiation was particularly recognizable in the percolating transition zone of the nanocomposites (between 0.075 and 0.250 wt%) shown in Fig. 2e–g.

Fig. 3
A scatterplot of maximum relative resistance percent versus single-walled carbon nanotube concentration weight percent depicts 2 plots with positive correlations. The upper plot represents sulfuric acid, and the lower plot represents magnesium sulfate. It includes the equation, intercept, and slope.

Maximum relative resistance variations of CNT alkali-activated nanocomposites based on CNT concentrations upon exposure to 90 µL of 0.1 M \({H}_{2}{SO}_{4} \& Mg{SO}_{4}\)

Full size image

3.3 Quantity Differentiation

The \({\mathrm{H}}_{2}{\mathrm{SO}}_{4}\) quantity differentiation potential of the sensors is shown in Fig. 4. For this purpose, the nanocomposites were fabricated with CNT inclusion of 0.1 wt% based on the percolation and sensing analyses. The same concentration regimen of \({\mathrm{H}}_{2}{\mathrm{SO}}_{4}\) (i.e., 0.1 M) was introduced in volumetric quantities of 90, 180, 270 µL (30, 60, and 90 µL, respectively in each cycle) to the sensors to evaluate the responses. The increments of relative resistance were approximately + 140% and + 60% with the volume increasing from 90 µL to 180 µL and afterwards to 270 µL, exhibiting a linear behavior.

Fig. 4
Two graphs. a. Relative resistance percent versus time seconds with 3 increasing curves with fluctuations labeled 90, 180, and 270 microliters from bottom to top. b. Relative resistance versus volume microliters with an increasing line passing through (90, 2.5), (180, 6.25), and (270, 9.75).

Quantity differentiation of CNT alkali-activated sensors (CNT conc. 0.1 wt%). a Exposure to different volumes of 0.1 M \({H}_{2}{SO}_{4}\); b regression of relative resistance vs. volumetric quantity

Full size image

3.4 CNTs’ Conductive Network

The CNTs’ network distribution and expansion are shown in Fig. 5. The addition of SDBS as surfactant for the dispersion of CNTs, entrained air bubbles into the alkali-activated microstructure. In addition, SBDS probably had a negative impact on the activating reactions and created a weaker microstructure. Nevertheless, the alkali-activated microstructure exhibited sufficient strength (compressive strength of 40 ± 2.48 MPa for 28-day nanocomposite) and mechanical performance. The main mechanism was the reinforcing effect of the CNTs because of their distribution (Fig. 5a), and crack covering and bridging abilities (Fig. 5b). Furthermore, this mode of interaction by CNTs has a secondary function, which is the formation of a conductive network (Fig. 5c). This network endows the nanocomposites with high electrical conductivity and consequently a sensing ability as explained in previously. The percolated conductive network of CNTs can be seen in the HRTEM images of Fig. 6. The network comprised overlapping CNT agglomerations, which formed directional pathways.

Fig. 5
Three microscopic images demonstrate the network of carbon nanotubes. It includes spongy and root-like structures. The scale bars are as follows. a. 2 micrometers. b. 1 micrometer. c. 500 nanometers.

The reinforcing and conductive network of CNTs in the microstructure of the alkali-activated matrix: a CNT distribution, b crack covering and bridging by CNTs, c CNTs’ network expansion

Full size image
Fig. 6
Two microscopic images demonstrate the network of carbon nanotubes. They include a rough and patchy surface with yellow pathways. The scale bars are 20 nanometers. The measurements are displayed in the top left corner.

Percolated conductive network of CNTs in the nanostructure of the alkali-activated matrix at the atomic scale of HRTEM. The percolated network of overlapped CNTs is shown as yellow pathways

Full size image

4 Conclusions

The present conceptual study has proposed a structural sensor for assessing the \({\mathrm{SO}}_{4}^{2-}\) sensing potential of CNT alkali-activated nanocomposites. The investigated discrimination criteria were CNT concentration, \({\mathrm{SO}}_{4}^{2-}\) bearing media (\({\mathrm{H}}_{2}{\mathrm{SO}}_{4}\mathrm{ vs}.\mathrm{ Mg}{\mathrm{SO}}_{4}\), and analyte volumetric quantity \({\mathrm{H}}_{2}{\mathrm{SO}}_{4}\). The obtained results can be summarized as follows.

  • The percolating zone of the nanocomposites was between 0.07 and 0.20 wt% of CNT content.

  • The sensors can be fabricated by incorporation of 0.1 wt% of CNT into the alkali-activated matrix material based on the percolation threshold study of the nanocomposites.

  • The sensors exhibit differentiation behavior by variation of shape and magnitude of the obtained relative resistance.

  • There was a linear correlation between relative resistance and CNT concentration when the sensors were exposed to 0.1 M \({\mathrm{H}}_{2}{\mathrm{SO}}_{4} \&\mathrm{ Mg}{\mathrm{SO}}_{4}\). Similarly, the relationship between the relative resistance of the sensors and volumetric quantity of 0.1 M \({\mathrm{H}}_{2}{\mathrm{SO}}_{4}\) was linear.

  • The sensors exhibited a higher magnitude of relative resistance when exposed to 0.1 M \({\mathrm{H}}_{2}{\mathrm{SO}}_{4}\) compared with 0.1 M \({\mathrm{MgSO}}_{4}\).

  • The percolated sensors presented a response curvilinear shape upon \({\mathrm{H}}_{2}{\mathrm{SO}}_{4}\) exposure and a rectangular shape upon \({\mathrm{MgSO}}_{4}\) exposure.