Materials and mix proportions
Samples were produced using mortar or concrete at a water/cement ratio of 0.4. The mix proportions (Table 1) were calculated using the absolute volume method. These were chosen after several trials to ensure that the mixes achieve satisfactory workability and that the cast spacers meet the 28-day compressive strength requirement of 50 MPa in accordance with BS 7973-1 . Portland cement CEM 1 52.5 N complying with BS EN 197-1:2011  with specific gravity of 3.15, and fineness of 291 m2/kg was used. The oxide composition was 63.4% CaO, 20.8% SiO2, 5.4% Al2O3, 2.4% Fe2O3, 1.5% MgO, 0.3% Na2O, 0.7% K2O, 2.9% SO3 and < 0.01% Cl−, with 2.1% loss on ignition. Its Bogue composition was 53.1% C3S, 19.1% C2S, 10.8% C3A and 7.2% C4AF.
Thames Valley gravel (< 10 mm) and sand (< 5 mm) complying with BS EN 12,620:2002 + A1:2008  medium grading were used as coarse and fine aggregates respectively. The specific gravity, moisture content (BS 812–109:1990 ) and 24 h absorption from 105 °C oven dry to saturated surface dry condition (BS 812–2: 1995 ) of the gravel were 2.75, 0.71% and 0.62% respectively. The corresponding values for the sand were 2.51, 2.45% and 0.60% respectively. The total aggregate volume fraction in the mortar and concrete were 50% and 70% respectively. A superplasticiser (Master Glenium SKY 920) was used at 0.75% wt. binder to improve workability. Finally, tap water was used as batching water.
Spacers with surface texture
Several approaches to produce cementitious spacers with surface texture were trialled. The most successful was the method using silicone rubber moulds made from low shrinkage platinum cured silicone. These come in a variety of small-scale texture shapes, orientations, profile depths, surface roughness and densities. Eight textures representing a range of surface characteristics were selected for detailed testing. These are summarised in Table 2. Three have regular convex protruding features (i.e. squares, rectangle-small and rectangle-large). The others are concave with indented features (i.e. pyramids, hemisphere, grooves-vertical and grooves-horizontal). A flat interface was used to represent conventional spacers as control reference.
The maximum profile h, defined as the distance between highest and lowest points of the profile, ranged from 0 (flat) to 9 mm (pyramid). Surface roughness ranged from 1 (flat) to 2.0 (rectangle), while texture density ranged from 0% (flat) to 55% (rectangle). Surface roughness (r) is defined according to BS 1134:2010  as r = AT/A0, where AT (mm2) is the total surface area and A0 (= L × W mm2) is the projected surface area (see Table 2). Texture density (Td) is calculated as Td = At/A0 × 100%, where At (= l × w × N mm2) is the area covered by the convex or concave features, l and w are the length and width of each feature and N is the total number of features. Figure 1 shows the spacers cast with various surface textures.
Figure 2 shows the preparation of cylindrical samples (100∅ × 50 mm) that consist of half spacer-concrete composites, cast in steel moulds and cured in two stages. The spacers were first produced by casting against textured silicone rubber sheets attached to wooden blocks in the 100∅ mm steel mould (Fig. 2a). Each wooden block is custom made such that its dimension fits the steel ring and its volume (including the silicon rubber) occupies exactly half of the sample volume. This is an important aspect of the mould assembly because it ensures that the resulting spacer-concrete interface passes through the cylindrical axis and is aligned to the load direction during strength testing. The blocks were tightly clamped from the top of the mould assembly to prevent movements during concrete placement and compaction.
Batching was done by weight. Cement and aggregates were firstly dry mixed for 30 s in a 30-L capacity pan mixer. Batch water premixed with superplasticiser was then added, and wet mixing was carried out for a further 3 min. A vibrating table with adjustable intensity was used for compaction. Samples were compacted in two equal-depth layers until no significant entrapped air escaped. During compaction, the fresh mix fills the crevices of the silicone mould to produce the desired texture as shown in Fig. 1.
The cast spacers were covered with a polyethylene sheet and wet hessian at room temperature for the first 24 h, then demoulded and transferred to a fog room at 100% RH, 21 °C where curing continued for 28 days. Subsequently, the spacers were returned to the steel mould assembly (Fig. 2b) and fresh concrete was cast against the spacers (Fig. 2c) following the procedures described above. The spacer-concrete composite samples (Fig. 2d) were then cured in the fog room (100% RH, 21 °C) for either 1, 7 or 28 days. In total, 224 composite cylindrical samples were tested. In addition, 100 mm monolithic cube samples (no interface) were prepared for density and compressive strength measurements in three replicates.
Splitting tensile strength
Bond strength of the spacer-concrete interface was determined by splitting tensile testing in accordance with BS EN 12,390-6:2009 . This is based on applying a compressive load perpendicularly to the cylindrical axis to induce a uniform tensile stress over the spacer-concrete interface. The sample was positioned using a centering device (Fig. 3) to ensure alignment between the spacer-concrete interface and applied load. Plywood strips were inserted between the sample and loading platen to reduce local compressive stress concentrations and to ensure that the sample remains aligned during testing . Loading was then applied at a constant rate of 0.03 MPa/s until failure. The splitting tensile strength ft was calculated as 2F/πLd, where F (N) is the maximum applied load, L (mm) and d (mm) are the length and diameter of the sample respectively. Two replicates were tested and averaged. In addition, tests were carried out on monolithic samples (no interface) of the mortar and concrete for comparison.
Oxygen diffusion and oxygen permeation
Samples for mass transport testing were prepared with concrete spacers and conditioned at 50 °C, 10% RH to constant weight to ensure that the measurements are not influenced by moisture content. The samples were monitored using an electronic balance accurate to 0.01 g for about 3.5 months when the mass loss was less than 0.01% per day. The conditioned samples were cooled to room temperature in a vacuum desiccator for 24 h, and then tested in three replicates following the sequence of oxygen diffusivity, oxygen permeability, water absorption and accessible porosity.
Oxygen diffusivity was measured by placing the sample in a transport cell sealed with silicone rubber ring. Load was applied to the rubber ring and the sample confined at 0.57 MPa to ensure that the gasses flowed through the sample only without causing further cracking or closure of existing cracks . Opposite faces were subjected to oxygen and nitrogen gasses at equal pressure, which counter diffused through the sample. The oxygen concentration in the outflow stream was measured at steady-state condition with a zirconia analyser (0.25 ppm to 25%) and the average diffusion coefficient D (m2/s) was calculated according to Fick’s first law.
Oxygen permeability was determined in a similar transport cell by subjecting the sample to three inflow pressures of approximately 0.5, 1.5 and 2.5 bar above atmospheric pressure, and measuring the steady-state outflow rates. Each applied inflow pressure takes approximately 1 h to reach steady-state. The outflow rate was determined using a series of bubble flowmeters, by measuring the time required for the bubble to move through a fixed volume using a digital stopwatch. Permeability at each pressure was then calculated by following Darcy's law for compressible fluids and the intrinsic permeability kint (m2) was determined by applying Klinkenberg's correction for gas slippage.
Water absorption and accessible porosity
Water sorptivity was measured by immersing the sample in a tray containing water to a depth of about 3 mm above the bottom surface and then measuring the mass gain with time using an electronic balance accurate to 0.01 g until saturation was achieved. The tray was kept covered with a loosely fitting lid all the times to prevent the sample from drying and to avoid condensation. The sorptivity coefficient S (g/m2min0.5) was determined from the slope of the regression line (R2 > 0.99) of cumulative absorption per unit flow area against square root of time t according to the classical unsaturated flow theory.
Immediately after water absorption, the samples were fully immersed in water and vacuum saturated for approximately 4 h. The samples were left immersed in water for 24 h, then surface dried and weighed. The sample accessible porosity ϕ (%) was determined from the mass difference between the vacuum saturated-surface dry condition and preconditioned state (50 °C oven) and divided by the sample volume. Full details of the transport tests are given in .
Fluorescent epoxy impregnation
Samples were impregnated with a fluorescein-dyed epoxy as another method to characterise microstructure and transport resistance. The resin was prepared by mixing low-viscosity epoxy with fluorescein dye at 0.05 wt% using a magnetic stirrer for 24 h and then heated to 40 °C. The fluorescein-dyed resin was then mixed with hardener at 25:3 mass ratio and 5% wt. toluene to reduce viscosity further . The sample was sealed in a pressure cell similar to the one used for permeability testing. Its flat surface was covered with the epoxy mix and 10-bar compressed air was applied overnight to force the epoxy into the sample. After 2 days of curing at room temperature, the impregnated samples were sectioned with a slow speed diamond abrasive cutter to expose the spacer-concrete interface. The cross-section was ground with silicon carbide at 120-grit and then imaged with a digital camera under UV illumination to induce fluorescence. The depth and amount of epoxy impregnation were then measured using image analysis. Further details are provided in [41, 42].