**1. Introduction**

Cement composites, including pastes, mortars, and concrete are the most widely used construction materials; however, cementitious structures often suffer from poor durability that necessitates repair and maintenance (R&M) to be carried out. From 2011 to 2015, approximately a fifth of the civil engineering works in the UK were due to R&M [1], while in the US, Americans undertake over 200 million trips a day across deficient bridges [2]. This inadequate durability has resulted in a need for structural inspections which are usually carried out visually [3,4]; however, due to the inherent uncertainties and risks with visual surveys, the use of sensors has started to flourish. Despite the increasing use of structural health monitoring systems, the use of external sensors has often resulted in high costs, low sensitivity, need for frequent calibration, and incompatibility with structural materials [5–7].

Recent advances in material science have generated increasing interest in the use of smart, biomimetic materials that mimic natural systems and could monitor their own condition [4]. One such example, is self-sensing concrete, which refers to a material that can sense its condition and identify any damage, whilst maintaining or improving the structural performance [8]. A self-sensing mechanism can be achieved by passing electric current through the structure and monitoring changes in the electrical conductivity. The non-reversible change in conductivity can be used to monitor damage, while the reversible change can be used to monitor strain (loading) [8–10]. Nevertheless, cementitious composites behave as insulators to electricity and do not allow the passage of electric current [11]. Thus, the use of a conductive additive is necessary to ensure that an electrically conductive network can be formed within the cementitious structure.

In terms of conductive additives, at least 10 different types, along with hybrid combinations, have been investigated, such as steel fibers, carbon fibers, graphite powder, carbon nanotubes, and graphene nanoplatelets [8,12,13]. The minimum additive dosage that is needed to form continuous electrical paths inside the composite is known as the percolation threshold, which depends on many parameters, including additive composition (size and shape), concentration, and degree of aggregation [12]. For example, fibrous additives with a high aspect ratio, reach a percolation threshold at a lower dosage (~1.5 wt %), compared to particle ones (>5 wt %) [8].

One such functional additive is natural graphite, which due to its wide availability and low cost, could have promising applications in cementitious composites for self-sensing structures. Graphite powder has a layered planar structure, rendering it relatively soft due to its anisotropy and weak inter-planar forces, ability to conduct electricity and heat well, resistance to chemical attack, and stability under standard conditions [8,14]. Graphite has been used as a conductive filler in some studies, and it was found that it could improve the electrical conductivity performance [15–18]. However, graphite size can vary, and very often, this property is not reported, while at the same time, the percolation threshold could depend on the size of the conductive additive. In one study, it was found that when graphite was added in cement, the composite became a conductor; however, the mechanical and electrical properties depended on the water content and the setting process [15]. When dry-mixing graphite and cement powder, a minimum threshold of 2 wt % graphite was needed, below which the insulating cement prohibited the formation of conductive graphite pathways and the conductivity leveled off at around 10 wt % graphite [16]. Similarly, the DC electrical resistivity decreased rapidly between the percolation threshold at 2 and 10 wt %, where it plateaued with a graphite that had a particle size of 10–20 μm [17]; however, another study found a conductivity threshold at ~30 to 40 wt % of graphite with a *d*<sup>50</sup> size of 4.5 μm [18].

Despite some promising early findings in the literature around the improvement in electrical conductivity with graphite, the effect of its fineness has not been investigated in detail. Most studies do not report the inherent material properties, such as the size; thus, it is not possible to understand the effect of the additive on the properties of cement composites and use the graphite material at its full potential. Furthermore, there is a lack of a holistic assessment of the effect of natural graphite, not only on the electrical conductivity, but also on other properties that are essential for cementitious composites, such as rheology, hydration, and mechanical performance. Therefore, this study experimentally investigated three natural graphite products of varying sizes to understand the effect of graphite fineness on the electrical conductivity, microstructure, hydration, and mechanical performance of cementitious pastes that could then be used for self-sensing applications.

## **2. Materials and Methods**

#### *2.1. Materials*

Portland Cement CEMI 52.5N, supplied by Hanson and conforming to BS EN 197-1:2011 [19] was used and its chemical composition is tabulated in Table 1. The cement had a size distribution of 5–30 μm, a surface area of 0.3–0.4 m2/g, a density of 2.7–3.2 g/cm3 and a loss on ignition (LOI%) of 2.2, as provided by the supplier in the material safety data sheet. Three commercial products of natural graphite were used to create the graphite-cement paste. A coarse graphite powder (−10 mesh, 2 mm) was supplied by AlfaAesar (Haverhill, MA, USA), and two finer graphite powders (−100 mesh, 0.150 mm and −325 mesh, 44 μm) were supplied by Sigma-Aldrich (St. Louis, MO, USA). The suppliers only provided the mesh size of the graphite rather than the particle size distribution and hence only the mean size is used here. These graphite products are referred to as coarse (2 mm), medium (0.15 mm) and fine (44 μm). The graphite powders were used at varying concentrations needed to reach the percolation threshold (from 0 to 40 wt %). The cement paste had a constant *w*/*c* = 0.45. It should be noted that the graphite powders were added as additions rather than cement replacement; therefore, the cement content and *w*/*c* ratio are the same in all mixes.


**Table 1.** Chemical composition of the CEMI.
