*3.4. Mechanical Performance of Graphite-Cement Pastes*

Figure 8 shows the effect of graphite fineness on the compressive strength of cement paste. The three graphite materials were used at 10 and 20 wt % concentrations. At all test ages, the compressive strength reduced with graphite, irrespective of its fineness. Furthermore, the higher the graphite dosage, the lower the compressive strength, with the 20 wt % concentration samples always resulting in a lower compressive strength than the 10 wt % samples. The summary of the % reduction in strength comparing to the control is presented in Table 2. The compressive strength had an inverse relationship with graphite size; the coarse graphite produced the lowest strength, whilst the fine graphite had the least reduction, as compared to the control.

**Figure 8.** Effect of graphite fineness on the compressive strength of cement paste (*w*/*c* = 0.45) at 2, 7, and 28 days.

**Table 2.** Reduction in compressive strength (%) with the three different graphite materials.


Micro-indentation testing was also carried out to assess the effect of graphite fineness on the mechanical performance of cement paste, by assessing the hardness of the specimens. Graphite–cement paste samples were tested at 56 days, with a 20 wt % graphite concentration, and the error bars show the average of 15 measurements. The compressive strength results would suggest that the hardness would reduce with graphite addition and the effects would be more pronounced for the coarse graphite. Indeed, as seen in Figure 9a, the coarser the graphite, the lower the hardness. The fine graphite almost maintained the hardness of the control specimen, while, on the contrary, the coarse graphite reduced the hardness by 28%. The results can be explained by the graphite softness, which was expected to reduce the hardness of the paste, and also because of packing density, where the finer graphite resulted in a more compact mix, which improved the overall hardness of the sample. The hardness results also provide a further indication of sufficient dispersion of graphite. The error bars were small, and the variance was similar to the control, further supporting the SEM and μCT scan findings. Young's modulus in Figure 9b reduced for the coarse graphite by 15% but increased for the two finer graphite materials, as compared to the control (14% improvement for fine graphite). The modulus of elasticity is expected to increase with increasing compressive strength [11]; therefore, the finer the graphite, the higher the Young's modulus (as strength was also higher). The increase in stiffness with the fine graphite could also be explained by changes in porosity, where due to better packing density, the fine graphite-cement paste had a lower porosity, and therefore it was stiffer compared to the control.

**Figure 9.** Micro-indentation results of the effect of 20 wt % graphite addition on the cement paste, at 56 days, in terms of (**a**) hardness (**b**) and Young's modulus.

Mechanical performance testing indicated that increasing graphite fineness was beneficial for mechanical performance. The fine graphite was more effective in maintaining the compressive strength and hardness of the specimens, while the use of the coarse graphite led to significant reductions in compressive strength and hardness. Therefore, when using coarse graphite materials, the mechanical performance could be significantly compromised, and this is a key limitation in their use as conductive fillers for self-sensing application. From a practical viewpoint, it would not be feasible to use higher than 10 wt % graphite additions as the compressive strength is significantly compromised. This is particularly the case for the coarse graphite powders, that more than halved the compressive strength when added as a 20 wt % addition. To mitigate the impacts on mechanical performance, a more realistic perspective would be the targeted use of the material in locations more prone to damage. Potentially, the graphite-cement paste could be locally used as a coating of the structure rather than in the bulk of the matrix, ensuring, in this way, that the structural performance is retained, while the graphite-cement paste coating could yield the sensing capabilities, assisting with maintenance and ensuring the resilience of the structure.

#### *3.5. Electrical Conductivity of Graphite-Cement Pastes*

The effect of graphite fineness on the electrical conductivity of cement paste was investigated to establish whether graphite could be used for self-sensing applications. Tests were carried out at 2, 7, and 28 days, and the results in terms of electrical conductivity vs. graphite content are illustrated in Figure 10. Irrespective of graphite fineness and test age, before the sudden increase in electrical conductivity, the conductivities of all samples were less than 2 S/m. Moreover, the electrical conductivity clearly reduced as curing progressed. The two-day samples (solid lines) had higher conductivities than those tested at 7 or 28 days (dashed lines). The reason that the conductivity of the samples reduced over time was due to the free water content available in the mix. Electric current can travel both through the conductive additive (termed as electronic conduction) and through the free water available, which is termed as electrolytic conduction. As the hydration progressed from 2 to 28 days, the free water in the mix reduced; thus, it was more difficult for the electric current to pass the matrix. At the same time, at low graphite dosages, the conductive filler content was not sufficient for electronic conduction to take place and form uninterrupted travel paths. This means that, at low graphite dosages, the specimens were acting as insulators, and, over time, the electrical conductivity would diminish. Furthermore, the percolation threshold was not affected by curing age.

By observing the effect of graphite fineness, the coarse graphite (blue lines) had a percolation threshold between 30 and 40 wt % dosages, and, at 28 days, the conductivity at 40 wt % dosage was over nine times higher than the control. For the coarse graphite, it was also found that the conductivity was compromised at low graphite concentrations, and this can be explained by changes in porosity and water content in the mix. Even though the *w*/*c* remained constant, as graphite was added in the mix, the effective water/solids ratio was reduced, which resulted in less water available for electrolytic conduction and at the same time, the graphite concentration was not sufficient to create electric conduction paths through the specimen. Hence, the conductivity of the sample was compromised due to a reduction in the effective water in the mix. By observing the medium graphite (green lines), the percolation threshold was found at a lower concentration, compared to the coarse graphite, and at 30 wt %, the electrical conductivity was ~17 times higher, compared to the control. Therefore, the medium graphite resulted in much higher conductivity at a lower concentration, compared to the coarse graphite. For the fine graphite (red lines), the percolation threshold could be found between 20% and 30%. At 30 wt % concentration, the fine graphite had a conductivity twenty-eight times higher than the control and a 37% higher conductivity compared to the medium graphite at the same dosage.

**Figure 10.** Effect of graphite fineness and dosage on the electrical conductivity of CEMI pastes (*w*/*c* = 0.45).

To better understand the effect of graphite fineness on the electrical behavior of the graphite-cement pastes, electrical impedance spectroscopy (EIS) testing was undertaken. The control mix refers to a cement paste with *w*/*c* = 0.45, and it is the same in all cases. Coarse graphite was tested at 10 and 40 wt %, with the former being below and the latter above the percolation threshold. The Nyquist plots for 7 and 28 days are shown in Figure 11a,b, whilst the resistance vs. frequency plot is illustrated in Figure 11c. As the hydration progressed, the Nyquist plots shifted to the right, at higher true resistance values on the *x*-axis, which was expected as electrical resistance increases with curing age. The incomplete electrode arc on the right side of the plot, was very clear for the 10 wt % graphite, meaning that the resistance measurement comprised of both the inherent electrical conductivity of the material and that of the electrode. However, at 40 wt % coarse graphite (above the percolation threshold), the Nyquist plot was obviously different, and the electrode arc was not present. The measured electrical response corresponded only to the bulk response of the material, meaning that a fully conductive path was formed through the cement composites. The arc of the 40 wt % graphite remained almost unchanged as curing progressed from 7 to 28 days and no increase in resistance over time was observed. Therefore,

when graphite was added at a concentration below the percolation threshold, the electrical conductivity depended on the water content, and therefore the resistivity increased over time as the hydration progressed. When the percolation threshold was exceeded, the electrolytic conduction mechanism became irrelevant and electric current traveled primarily due to the conductive network that was formed with the graphite particles. In this case, the continuous cement hydration, which reduces the free water, had an insignificant effect on the electrical resistivity.

**Figure 11.** Cement paste with coarse graphite: (**a**) 7-day Nyquist plot, (**b**) 28-day Nyquist plot, and (**c**) frequency-dependent resistance.

The cement pastes with the medium graphite (0.150 mm), at 10 and 20 wt %, were examined. As shown in Figure 12, the Nyquist plots at 7 and 28 days showed similar arcs, with the electrode effect present in all cases and illustrated by the incomplete rightmost arc. The resistance increased with hydration age, due to the consumption of free water, which meant that less water was available for electrolytic conduction. The fact that the arcs showed both the electrode effect and the bulk material response meant that the conductive network was not fully formed at 20 wt % and that the percolation threshold was slightly higher for this medium graphite. From Figure 12c, it can be observed that the total resistance reduced with increasing graphite content, irrespective of the frequency.

**Figure 12.** Cement paste with medium graphite: (**a**) 7-day Nyquist plot, (**b**) 28-day Nyquist plot, and (**c**) frequency-dependent resistance.

Figure 13 illustrates the findings for the fine graphite (44 μm) at 10 and 20 wt %, where the Nyquist plots between the two graphite dosages were different. The 10 wt % graphite was characterized by two arcs, a full semicircle representing the bulk material response, and an incomplete arc on the right size, which showed the electrode response to the electric current. For the 10 wt % fine graphite, the resistance increased with age due to cement hydration; therefore, the electrical conduction was due to both the presence of water (electrolytic) and conductive filler (electronic). Instead, at 20 wt %, only a semicircle arc was found at the Nyquist plots, which remained almost unchanged with age, meaning that the percolation threshold was reached. Comparing to the medium graphite, the percolation threshold was reached at 20 wt % concentration, which was not the case for the medium graphite where a higher graphite dosage was needed. Similar to the coarse and medium graphite materials, the electrical response was frequency dependent, especially below the percolation threshold. The response stabilized at ~1000 Hz for 20 wt % fine graphite but only after ~10,000 Hz for the lower graphite concentration.

The effect of graphite fineness on electrical conductivity was investigated, with an outlook that the graphite-cement pastes could be used for self-sensing applications. It was found that the coarse graphite had a percolation threshold at ~30 to 40 wt %, which reduced to between 20 and 30 wt % when a medium and a fine graphite were used. Therefore, increasing graphite fineness leads to a lower percolation threshold, which is beneficial in terms of material usage and in maintaining the

mechanical performance. It should be noted that these filler dosages are higher, compared to the dosages traditionally used for inert fillers in cementitious composites. The reason that high filler dosages were selected was to ensure that an uninterrupted electrically conductive path was successfully formed in the cementitious matrix. Contrary to other inert fillers that are primarily used to improve the packing density of the mix, the purpose here was to reach the percolation threshold, which necessitated the higher additive concentrations. EIS testing showed that a combination of an electrolytic and electronic conduction mechanism was present when the conductive additive was used below the percolation threshold. Instead, when the percolation threshold was reached and an uninterrupted conductive path was formed, the electronic conduction became the dominant mechanism, and the conductivity did not depend on the presence of water. The main finding from the electrical conductivity testing was that, the finer the graphite, the lower the dosage that is needed to establish a percolation threshold, and the higher the conductivity at that dosage, as compared to a coarser graphite. These results can be explained by packing density principles. The formation of conduction paths relates to the particle size and aspect ratio of the conducting graphite filler [32]. The finer graphite particles tend to stabilize in dense configurations, resulting in more inter-particle contacts, as compared to coarser materials, creating inevitably more paths for current to pass through [33].

**Figure 13.** Cement paste with fine graphite: (**a**) 7-day Nyquist plot, (**b**) 28-day Nyquist plot, and (**c**) frequency-dependent resistance.

#### **4. Conclusions**

This study examined the effect of graphite fineness on the performance of cement paste to be used in self-sensing applications. Three graphite products of varying fineness were tested, and it was found that graphite fineness can greatly affect the performance of cementitious composites. In terms of early age performance, increasing graphite fineness led to a dramatic reduction in the fluidity of the paste, which could introduce practical limitations. On the other hand, graphite was not found to affect the hydration of the cement paste, meaning that it acts as an inert conductive filler. Compressive strength testing and micro-indentation showed that, the finer the graphite, the lesser the effect on mechanical performance. Electrical conductivity testing then showed that increasing graphite fineness reduces the percolation threshold for electrical conductivity. The finer graphite had a percolation threshold at 20 wt % concentration, which increased to 30–40 wt % for the coarse graphite. However, the use of graphite as a conductive additive in cement paste introduces some practical limitations. The reduction in fluidity could be a significant barrier for using this composite material in situ, while the reductions in compressive strength could make its use prohibiting for structural applications. Overall, it is recommended to use a graphite of greater fineness, as this leads to better maintenance of mechanical performance, and it reduces the dosage required to reach an electrical percolation threshold. Practical limitations around the fluidity of the paste could be overcome by adjusting the mix design and dispersion protocol.

**Author Contributions:** Experimental investigation, data processing and analysis, and writing original draft, I.P.; data analysis and writing—review and editing, C.L. and A.Z.; conceptualization and supervision, A.A.-T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Engineering and Physical Sciences Research Council (Grant No. EP/L016095/1-EPSRC Centre for Doctoral Training in Future Infrastructure and Built Environment and Grant No. EP/P02081X/1-Resilient Materials for Life (RM4L)) and Costain Group.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.
