Assessment of the Permeability to Aggressive Agents of Concrete with Recycled Cement and Mixed Recycled Aggregate

: Acceptance by the construction industry of recycled concrete as a sustainable alternative material is contingent upon a reliable assessment of its permeability to corrosive agents. This study analyses the transport mechanisms associated with chloride (Cl − ), oxygen (O 2 ) and carbon dioxide (CO 2 ) ions in concrete with cement made with 10% or 25% ground recycled concrete (GRC) separately or in combination with 50% mixed recycled aggregate (MRA). The ﬁndings show that, irrespective of aggregate type, concrete with GRC exhibited lower resistance to ingress than conventional concrete due to its greater porosity. Nonetheless, O 2 permeability was consistently below 4.5 × 10 − 17 m 2 and CO 2 penetration, under 4 mm/year 0.5 , indicative of concrete with high quality. Resistance to CO 2 and Cl − penetration in the materials with 10% GRC was similar to the values observed in conventional concrete. On the other hand, the incorporation of 25% GRC increased the penetration of CO 2 and Cl - by 106% and 38%, respectively. Further to those ﬁndings in normal carbonation environments, reinforcement passivity would be guaranteed in such recycled materials over a 100 year service life.


Introduction
The evaluation of the mechanical behaviour of concrete is always considered essential, while the evaluation of the durability of this construction material is sometimes overlooked. Concrete with high porosity allows the entry of external agents, which can lead to durability problems; hence, it is essential to study in detail several properties that are related to the durability of concrete.
The permeability to aggressive agents in concrete can be measured through several tests, such as evaluating the transport mechanisms associated with chloride ions, oxygen and carbon dioxide.
The analysis of the permeability of concrete to gases concerns the entire porous structure of this material, i.e., both small and large pores, due to the reduced size of the oxygen molecules. Therefore, the oxygen permeability test has a higher sensitivity than the water absorption tests, since the water molecules are larger than the oxygen ones [1].
The transport mechanism of chloride ions in concrete is somewhat complex, and may involve diffusion, impregnation and capillary water absorption processes. Transport mechanisms vary widely with the microenvironment in which the structural elements are inserted. The penetration of chlorides into ordinary concrete is usually carried out through the continuous pore structure of the cementitious paste, the interface between the aggregates, and the paste (ITZ) and micro-cracks [2]. The penetration of chlorides is, together with carbonation, the main factor responsible for the depassivation and corrosion of steel reinforcement in reinforced concrete. stoneware), there was even a slight improvement in this property (lower than 10%). In turn, the authors observed an improvement of 20% and 30% in oxygen permeability with the total replacement of coarse and fine aggregates, respectively. The authors explain this result with the highest degree of hydration in the cementitious paste of concrete with RA.
Thomas et al. [9] evaluated some properties of precast concrete elements in which coarse RA and recycled cement were used. In this experimental campaign, six concrete mixes were characterised: a reference mix; two mixes with 25% and 50% (by weight) RA from mixed CDW; three mixes similar to the previous ones but with recycled cement with low clinker content (cement produced with 25% of ceramic waste). As in other studies [6], it was found that the oxygen permeability coefficient increases with the replacement of natural aggregates with RA. The authors [9] also concluded that this increase is higher in concrete with recycled cement and with the use of 50% of RA (higher than 300%) than in concrete with Portland cement and with the use of 50% of RA (higher than 40%).
Bravo et al. [6] studied the replacement of natural aggregates (fine and coarse) by RA from mixed CDW from four recycling plants. To this end, the authors analysed the oxygen permeability of concrete produced with 0%, 10%, 50% and 100% of fine or coarse RA. Total replacement of the fine and coarse aggregates caused increases in this property of more than 43% and 91%, respectively. Thomas et al. [9] also obtained significant increases in this property with the incorporation of coarse RA from CDW. The authors obtained increases of 50% in the oxygen permeability test performed at 28 days.
Torgal et al. [8] also evaluated the resistance to the penetration of chloride ions in concrete with RA (ceramic) and recycled cement (from four types of ceramic waste). The authors concluded that the use of RA and recycled cement significantly improved this property, obtaining decreases in the diffusion of chloride ions between 12% and 70%, and between 23% and 29%, respectively.
Qin and Gao [10] analysed the influence of the use of recycled cement produced from concrete waste (0%, 10%, 20%, 30% and 50% of the total weight of the binders) on the resistance to the penetration of chloride ions of cementitious composites. The authors concluded that the use of 50% recycled cement increases the permeability to chloride ions by more than 300%. However, the authors found that this increase is reduced to around 200% when the concrete is subjected to accelerated carbonation curing.
Some studies have also analysed the carbonation resistance of concrete with recycled cement. Kim [11] evaluated the behaviour of self-consolidating concrete with recycled cement from concrete (0%, 15%, 30% and 45% of the total mass of the binders). The author found that the use of this recycled cement causes a significant increase in the carbonation depth of the concrete (between 2.3 and 6.9 times).
Sáez del Bosque et al. [12] also evaluated the carbonation resistance of concrete with coarse RA from CDW (25% and 50% of the total coarse aggregates) and recycled cement (with 25% of ceramic wastes). Regardless of the type of cement, the average carbonation depth was slightly higher in materials with 25% or 50% recycled aggregate than in the reference concrete. The use of this partially recycled cement produced from ceramic waste led to an increase in the carbonation depth at 28 days from 4.2 mm to 5.0 mm (increase of 19%).
The present investigation follows previous ones that intended to analyse the mechanical behaviour, water absorption, shrinkage and thermal performance of concrete with RA from CDW and recycled cement [13][14][15]. In these previous investigations, the authors observed that the use of GRC enhanced concrete environmental performance. At 10% replacement, it lowered the CO 2 emitted in concrete manufacture by 7.5%, and at 25% GRC by 18.7%, relative to concrete made with ordinary Portland cement (OPC).
There are already some studies that address the permeability of concrete with recycled cement (mainly from ceramic wastes) or with RA from CDW. However, no previous study has evaluated the permeability to aggressive agents (chloride ions, carbon dioxide and oxygen) of concrete with simultaneous replacement of the two elements evaluated in this investigation. To fill this scientific and technical knowledge gap, this study analyses the Appl. Sci. 2021, 11, 3856 4 of 17 effect of replacing Portland cement with ground recycled cement (GRC), at 10% (R10) or 25% (R25). This assessment was carried out on concrete with 100% NA, 50% NA and 50% mixed recycled aggregates (MRA). This experimental campaign involved the evaluation of the permeability to the aggressive agents of these mixes through the performance of oxygen permeability tests, resistance to the penetration of chloride ions and resistance to carbonation. Finally, the evolution of the carbonation front in these mixes was studied with the increase in the exposure time for mixes in different exposure classes, according to the forecasting model proposed by EHE-08 [16]. This assessment in this type of concrete has not yet been carried out in previous investigations.

Experimental Programme
The three types of binders used in this study were as follows: (i) type I 42.5 R (CEM I 42.5 R) ordinary Portland cement (OPC); (ii) a binder with 90% OPC and 10% GRC; and (iii) a binder with 75% OPC and 25% GRC. The GRC was obtained by crushing and grinding laboratory-prepared concrete specimens (stored in a wet chamber at 20 ± 2 • C and relative humidity of 55 ± 5% for 90 days), batched as per the Faury method with 300 kg/m 3 of cement and a water/binder ratio of 0.55. The GRC processing is illustrated in Figure 1: (i) in Stage 1, the specimens were mixed, moulded, cured and subsequently tested to breakage; (ii) in Stage 2, the cubic and prismatic specimens were jaw-crushed to a particle size of <16 mm; (iii) in Stage 3, the particles generated in Stage 2 were roll-crushed to <5 mm; and (iv) in Stage 4, the crushed concrete was ball-ground for 2 h at 60 rpm (ratio between abrasive load and material of 3/1) to the target fineness.
There are already some studies that address the permeability of concrete with recycled cement (mainly from ceramic wastes) or with RA from CDW. However, no previous study has evaluated the permeability to aggressive agents (chloride ions, carbon dioxide and oxygen) of concrete with simultaneous replacement of the two elements evaluated in this investigation. To fill this scientific and technical knowledge gap, this study analyses the effect of replacing Portland cement with ground recycled cement (GRC), at 10% (R10) or 25% (R25). This assessment was carried out on concrete with 100% NA, 50% NA and 50% mixed recycled aggregates (MRA). This experimental campaign involved the evaluation of the permeability to the aggressive agents of these mixes through the performance of oxygen permeability tests, resistance to the penetration of chloride ions and resistance to carbonation. Finally, the evolution of the carbonation front in these mixes was studied with the increase in the exposure time for mixes in different exposure classes, according to the forecasting model proposed by EHE-08 [16]. This assessment in this type of concrete has not yet been carried out in previous investigations.

Binders
The three types of binders used in this study were as follows: (i) type I 42.5 R (CEM I 42.5 R) ordinary Portland cement (OPC); (ii) a binder with 90% OPC and 10% GRC; and (iii) a binder with 75% OPC and 25% GRC. The GRC was obtained by crushing and grinding laboratory-prepared concrete specimens (stored in a wet chamber at 20 ± 2 °C and relative humidity of 55 ± 5% for 90 days), batched as per the Faury method with 300 kg/m 3 of cement and a water/binder ratio of 0.55. The GRC processing is illustrated in Figure 1: (i) in Stage 1, the specimens were mixed, moulded, cured and subsequently tested to breakage; (ii) in Stage 2, the cubic and prismatic specimens were jaw-crushed to a particle size of <16 mm; (iii) in Stage 3, the particles generated in Stage 2 were roll-crushed to <5 mm; and (iv) in Stage 4, the crushed concrete was ball-ground for 2 h at 60 rpm (ratio between abrasive load and material of 3/1) to the target fineness. OPC's and GRC's chemical compositions, particle size distributions and densities are given in Table 1. The main phases in GRC are SiO2 (46.1 wt%); CaO (40.0 wt%); and Al2O3 (5.3 wt%). OPC's and GRC's chemical compositions, particle size distributions and densities are given in Table 1. The main phases in GRC are SiO 2 (46.1 wt%); CaO (40.0 wt%); and Al 2 O 3 (5.3 wt%).
D 90 (90th percentile value for particle diameter) in GRC was 147 µm. The difference from the 46 µm recorded for OPC is associated with the natural aggregate present in the source recycled concrete. In contrast, the lower value of D 10 in the GRC (1.58 µm) than for OPC may be attributed to the presence of a cementitious matrix in the recycled Appl. Sci. 2021, 11, 3856 5 of 17 aggregate used to prepare the former. The GRC used was less dense (2.54 g/cm 3 ) than OPC (3.11 g/cm 3 ).

Aggregates
The coarse natural aggregate (NA) used contained two sizes of limestone gravel, 4/12 mm (NG-M) and 12/22 mm (NG-G), and two sizes of siliceous river sand, 0/2 mm (NS-F) and 2/4 mm (NS-C). The single fraction (0/32 mm) MRA supplied by a CDW recycling plant at Lisbon (Portugal) was sieved and classified at the laboratory, where sizes <4 mm and >22 mm were discarded. The particle size distributions of all the NA and of the 4/22 mm coarse MRA aggregate ultimately used complied with the recommendations of European standard EN 933-1 [17] and are given in Figure 2.
source recycled concrete. In contrast, the lower value of D10 in the GRC (1.58 µm) t OPC may be attributed to the presence of a cementitious matrix in the recycled ag used to prepare the former. The GRC used was less dense (2.54 g/cm 3 ) than OP g/cm 3 ).

Aggregates
The coarse natural aggregate (NA) used contained two sizes of limestone grav mm (NG-M) and 12/22 mm (NG-G), and two sizes of siliceous river sand, 0/2 mm and 2/4 mm (NS-C). The single fraction (0/32 mm) MRA supplied by a CDW re plant at Lisbon (Portugal) was sieved and classified at the laboratory, where sizes and >22 mm were discarded. The particle size distributions of all the NA and of t mm coarse MRA aggregate ultimately used complied with the recommendations ropean standard EN 933-1 [17] and are given in Figure 2. The composition of the 4/22 mm coarse MRA aggregate, likewise determined standard EN 933-11 [18], is listed in Table 2, according to which the main comp were Rc (47.1 wt%), Rb (22.3 wt%) and Ru (25.2 wt%), with glass and plaster (<2 w minor constituents. On the grounds of those data and the criteria set out in Spain' tural concrete code (EHE-08) [16], with its <95 wt% Rcu (Rcu=Rc + Ru) content and Rb content, MRA qualified for classification as a mixed recycled aggregate. The p The composition of the 4/22 mm coarse MRA aggregate, likewise determined as per standard EN 933-11 [18], is listed in Table 2, according to which the main components were Rc (47.1 wt%), Rb (22.3 wt%) and Ru (25.2 wt%), with glass and plaster (<2 wt%) as minor constituents. On the grounds of those data and the criteria set out in Spain's structural concrete code (EHE-08) [16], with its <95 wt% Rcu (Rcu = Rc + Ru) content and >5 wt% Rb content, MRA qualified for classification as a mixed recycled aggregate. The physical and mechanical properties recommended in standard EN 12620 [19] for coarse aggregates to be used in structural concrete (f ck ≤ 30 MPa) are listed in Table 3.  The specimens were prepared and cured for the O 2 permeability and CO 2 and Cl − penetration tests as per European standard EN 12390-2 [24]. The size of the prismatic specimens for O 2 permeability testing was 150 × 300 mm (∅ × height) and that of the samples for CO 2 and Clpenetration was 100 × 200 mm. All of the specimens were cured for 28 d or 90 d in a wet chamber at a relative humidity of 95 ± 5% and a temperature of 20 ± 2 • C. Concrete disks 50 mm thick (measured with a digital calliper with a precision of 0.1 mm) were cut out of the mid-section of each specimen with a water-cooled diamond saw. The samples for O 2 permeability and Clpenetration testing were subsequently sealed around the outer rims and the disks for the CO 2 penetration tests at the top and bottom with three layers of epoxy resin to ensure one-dimensional flow.
The samples for the O 2 permeability and CO 2 penetration tests were stored in a wet chamber at 20 ± 2 • C and relative humidity of 55 ± 5% for 3 weeks, in keeping with standardised procedures. The samples for the Cl − penetration tests were pre-saturated for 4 h in a vacuum container with the top and bottom exposed to the ultra-low pressure (<1 mm Hg). Water was then added to the container, where the vacuum was released after 1 h and the samples left to soak in the water for a further 18 ± 2 h prior to testing.

Oxygen Permeability
O 2 permeability was assessed with a Cembureau apparatus in a room at the controlled temperature (20 ± 2 • C) and relative humidity (50 ± 5%) as specified in Spanish standard UNE 83981 [25]. The O 2 permeability coefficient, defined in terms of Darcy's law, was calculated here with the Hagen-Poiseuille relationship (Equation (1)) for the laminar flow of a compressible fluid across a porous medium under steady flow conditions [26]: where K (m 2 ) is the O 2 permeability coefficient; Q s the outlet flow (m 3 ·s −1 ); P s , the absolute pressure at the sample outlet (Nm −2 ); P o , the absolute pressure at the sample inlet (Nm −2 ); S, the cross-sectional area of the sample (m 2 ); L, the sample thickness in the direction of the flow (m); and µ, the O 2 dynamic viscosity at 20 • C (2.02 × 10 −5 Nsm −2 ).

Rapid Chloride Permeability (RCP) Test
Cl − penetration was determined using the electrical procedure described in standard ASTM 1202-97 [27] in a room at a controlled temperature (20 ± 5 • C). The total electric current passing through the samples in 6 h was calculated by integrating current over time (RCP test finding), as in Equation (2): where RCP is total charge (coulombs); I o , the electric current immediately after applying voltage (Ampere); and I t , the electric current at time t (Ampere), with subscripts denoting time in minutes. The empirical formula proposed by Berke and Hicks [28] (Equation (3)) was applied to correlate RCP to the diffusion coefficient (D) in m 2 /s × 10 −12 : where the unit for RCP is Coulomb.

Accelerated Carbonation Test
The resistance to CO 2 penetration test was conducted as described in Portuguese standard LNEC E391 [29]. The samples were stored for 7 d, 28 d, 56 d or 90 d in an accelerated carbonation chamber with a CO 2 concentration of 3 ± 0.1%, relative humidity of 60 ± 5% and temperature of 23 ± 3 • C. The carbonation depth was found by spraying the two sides of the sample resulting from diametric breakage with a phenolphthalein solution (0.8 g of the indicator in powder form dissolved in a solution containing 70 mL of ethanol and 30 mL of deionised water). As phenolphthalein is purple at pH > 10, the colourless concrete was determined to have been carbonated (pH < 8). Measurements were taken at a total of 12 points in each sample and recorded along with the maximum and minimum CO 2 penetration depths.

Service Life Prediction
Concrete carbonation affects the durability limit states of concrete structures, insofar as it deteriorates the passive cover that protects the reinforcing steel [30]. In the semiprobabilistic approach adopted by Spain's structural concrete code EHE-08, the condition to be met is as follows: where t s is estimated service life and t d service life calculated for the concrete structure. Calculated service life (t d ) is defined (Equation (5)) as the product of the design service life (t SL ) for the type of structure times a service life safety coefficient (γ i ), quantified in the aforementioned code as 1.10. A 50 year design service life is assumed for residential or office buildings, bridges and flyovers less than 10 m long and low-or medium-order engineering structures, whereas monumental buildings or high-order structures are designed for a 100 year service life: Further to the CO 2 penetration model described in code EHE-08 [16], characteristic depth (X c ) can be expressed as in Equation (6): where X c is the characteristic depth (mm) for the design service life; K n is the natural carbonation coefficient listed in Table 8 (mm/year 0.5 ) and t d , the calculated service life.

Concrete Design
Batching for all the concrete mixes is given in Table 4. In all mixes, irrespective of the MRA and/or NA content, the particle size distribution was fitted to Faury's [31] theoretical curve for a maximum aggregate size of 22 mm, maximum compactness and the same volume of aggregate in all mixes. That called for separating the MRA into sub-fractions: 16/22 mm, 11.2/16 mm, 8/11.2 mm, 5.6/8 mm, 4/5.6 mm. All the mixes were designed to meet the requirements set out in European standard EN-206-1 [32] for durability class XC2 and strength class C25/30. Cement was added at a rate of 300 kg/m 3 and all the mixes were prepared to S2 workability, defined in European standard EN-206-01 [32] as equivalent to a slump of 70 ± 20 mm [33]. Consequently, a higher water/binder ratio was required in the recycled materials than in the conventional materials due to the lower density of the GRC than OPC, and MRA than NA [15,34].

Oxygen Permeability
The graphs in Figure 3 of the 28 d and 90 d O 2 permeability coefficient and standard deviation (error bars) values denote a decline in permeability with curing age, irrespective of the GRC and MRA content. At the same time, the incorporation of the new recycled components induced a rise in O 2 permeability due to (i) the higher porosity of the new cement matrices, attributable to the dilution due to the presence of non-reactive particles in GRC (see item 2.1.1) and (ii) the greater porosity of MRA than NA, induced by the bound mortar and masonry materials found in the former (Table 2). Those findings were consistent with, and within, the range (1.0 × 10 −17 to 8.0 × 10 −17 ) reported for mixes prepared with 50% RCA [35], with 25% masonry aggregate [36], and mixes with 20% masonry waste (in the form of fired clay brick, gloss-or bisque-fired white stoneware or sanitary ware) used as a cement substitute [8].
Analogously, the use of 50% MRA (mix R0/50) raised 28 d O 2 permeability by 92.7% and the 90 d value by 67.5% relative to the conventional mix with NA. That rise was primarily associated with the effect of the presence of bound mortar and masonry material in MRA (more porous than NA) on O 2 permeability and the more readily accessible ingress pathways generated by microcracks in the MRA microstructure [37,38]. In addition, all the increases lay within the range (18% to 98%) observed in mixes with up to 50% RCA [35,39], with 25% masonry material [36], or prepared with both 25% masonry waste as an OPC replacement and 25% to 50% MRA [9]. cement matrices, attributable to the dilution due to the presence of non-reactive particles in GRC (see item 2.1.1) and (ii) the greater porosity of MRA than NA, induced by the bound mortar and masonry materials found in the former ( Table 2). Those findings were consistent with, and within, the range (1.0 × 10 −17 to 8.0 × 10 −17 ) reported for mixes prepared with 50% RCA [35], with 25% masonry aggregate [36], and mixes with 20% masonry waste (in the form of fired clay brick, gloss-or bisque-fired white stoneware or sanitary ware) used as a cement substitute [8].
Analogously, the use of 50% MRA (mix R0/50) raised 28 d O2 permeability by 92.7% and the 90 d value by 67.5% relative to the conventional mix with NA. That rise was primarily associated with the effect of the presence of bound mortar and masonry material in MRA (more porous than NA) on O2 permeability and the more readily accessible ingress pathways generated by microcracks in the MRA microstructure [37,38]. In addition, all the increases lay within the range (18% to 98%) observed in mixes with up to 50% RCA [35,39], with 25% masonry material [36], or prepared with both 25% masonry waste as an OPC replacement and 25% to 50% MRA [9]. Further to the plots in Figure 3b, the O2 permeability coefficient and open porosity were linearly and closely (R 2 >0.91) correlated in the mixes studied and the former was observed to increase proportionally to the latter. Similar findings were reported for mixes prepared with other additions (fly ash [40], rice husk ash [41] and masonry dust [42]) to replace cement; mixes containing RCA fines and coarse aggregates [43]; and mixes made with masonry aggregate [36]. The inference drawn from such a close correlation indicates that this property may serve as an index for assessing the material's porosity and, consequently, its durability. Further to the plots in Figure 3b, the O 2 permeability coefficient and open porosity were linearly and closely (R 2 > 0.91) correlated in the mixes studied and the former was observed to increase proportionally to the latter. Similar findings were reported for mixes prepared with other additions (fly ash [40], rice husk ash [41] and masonry dust [42]) to replace cement; mixes containing RCA fines and coarse aggregates [43]; and mixes made with masonry aggregate [36]. The inference drawn from such a close correlation indicates that this property may serve as an index for assessing the material's porosity and, consequently, its durability.

Resistence to Chloride Ion Penetration
The results of the rapid chloride penetration (RCP) test, Berke and Hicks [28] equationcalculated chloride diffusion (D) and the respective standard deviations for the 28 d and 90 d mixes are given in Table 5. The data show a steeper decline in Cl − penetration with curing age (∆RCP 28d→90d ) in the mixes containing recycled materials (GRC and/or MRA) than in the OPC/NA concrete. Similarly, the absolute value of ∆RCP 28d→90d was greater in both the GRC-NA (−13.7% in N10/0 and −14.3% in N25/0) and GRC-MRA (−14.9% in R10/50 and −16.5% in R25/50) families than the −9.3% in the reference NAC. According to Figure 4a, Cl − penetration rose linearly (R 2 ≥ 0.949) with the GRC replacement ratio, irrespective of curing age (28 d or 90 d) and family (GRC-NA or GRC-MRA), with values 8.0% higher in mix N10/0 and 38.1% in N25/10 than in NAC, and 2.7% higher in R10/50 and 16.3% in R25/50 than in R0/50. The incorporation of 50% MRA (mix R0/50) also induced increases of 14.7% to 24.6%, relative to mix NAC. That behaviour can be directly associated with the higher porosity of the new cementitious matrices, given the presence of new recycled materials (GRC and/or MRA). It is also consistent with earlier observations for mixes with additions, such as 10% to 40% cement powder [44], 10% to 50% concrete powder [10], or 5% to 15% cement kiln dust [45] as OPC replacements, and materials containing 15% to 30% biomass bottom ash (BBA) to replace OPC, separately or in combination with 30% MRA as an NA substitute [46].  The 90 d mixes with 10% GRC, separately or in combination with 50% MRA, qualified for the same 'moderate' corrosion risk class as mix NAC, given that all the rises were observed to lie within the 8% to 30% observed in earlier studies in mixes with 10% to 40% of other inert additions (marble powder and granite dust) [47].
The close linear relationship (R 2 = 0.861 > 0.85 [48,49]) between O2 permeability and the chloride diffusion coefficient in the mixes studied (Figure 4b) is clear proof that both transport mechanisms depend on pore structure and pore interconnectivity alike [50]. Similar findings were reported earlier for mixes made with additions such as marble powder or granite dust [47], ground granulated blast furnace slag (GGBFS) [51,52] or BBA [46]  The 90 d mixes with 10% GRC, separately or in combination with 50% MRA, qualified for the same 'moderate' corrosion risk class as mix NAC, given that all the rises were observed to lie within the 8% to 30% observed in earlier studies in mixes with 10% to 40% of other inert additions (marble powder and granite dust) [47].
The close linear relationship (R 2 = 0.861 > 0.85 [48,49]) between O 2 permeability and the chloride diffusion coefficient in the mixes studied (Figure 4b) is clear proof that both transport mechanisms depend on pore structure and pore interconnectivity alike [50]. Similar findings were reported earlier for mixes made with additions such as marble powder or granite dust [47], ground granulated blast furnace slag (GGBFS) [51,52] or BBA [46] as cement replacements, mixes with 20% to 100% RCA [43], and mixes where 5% to 10% of the OPC was replaced with silica fume and 50% to 100% of NA with RCA [39].

Carbonation Resistance
The mean carbonation depths (Cd) of all the mixes after 7 d, 28 d, 56 d or 90 d in an accelerated carbonation chamber and associated standard deviations are listed in Table 6. Carbonation resistance was intensely affected by the porous microstructure of the new recycled concrete mixes with GRC and/or MRA. The use of GRC raised 28 d Cd by +0.8 mm in mix N10/0 and 5.7 mm in mix N25/0 relative to NAC and by 1.0 mm in mix R10/50 and 3.0 mm in mix R25/50 relative to R0/50. Similar findings were observed in mixes prepared with additions such as 5% to 25% marble slurry [53], 25% masonry CDW [12], and 25% to 50% fly ash from biomass-fired power plants [54]. The data in Table 7, in turn, denote the very close (correlation coefficients, R 2 , of over 0.89) linear relationship between the Cd and GRC content, irrespective of exposure time and family (GRC-NA or GRC-MRA). The incorporation of 50% MRA (R0/50), in turn, induced an increase in Cd of 3.0 mm in the 28 d results, relative to the NAC mix. Such deeper carbonation penetration was associated with higher water absorption and porosity in MRA than in NA [6,49]. This was the same pattern as observed in O 2 penetration discussed in Section 3.1, depicted in Figure 3b and observed by other authors [55,56] in mixes with RA.
The linear relationship in the 28 d and 90 d results between carbonation depth and oxygen permeability is depicted in Figure 5a, and between Cd and the chloride diffusion coefficient in Figure 5b. The high R 2 values (all lying between 0.895 and 0.965) denoted a close correlation between those properties and showed that both the O 2 permeability coefficient and the Cldiffusion coefficient may serve to predict the carbonation depth [57,58].
Earlier studies with mixes with additions such as fly ash, GGBFS [40,41] or BBA [46] as cement replacements [51,52] together with 25% to 100% RCA [59] yielded similar findings. The accelerated carbonation coefficients (Kac) given in Table 6 were calculated from linear regression of the carbonation depth/square root of time curve and Fick's first law (Equation (7)): where Cd is the mean carbonation depth at time t (mm); Kac, the carbonation rate under the experimental conditions (mm/year 0.5 ); d0, the carbonation depth at time = 0 (mm); and t, the exposure time (years). The reliability of the carbonation coefficients calculated was over 98% in all mixes (see R 2 in Table 6). Kac was 14.5% higher in mix N10/0, 106.6% in N25/0, 54.1% in R0/50, 66.9% in R10/50 and 117.0% in R25/50 than in mix NAC. Those findings denoted a higher rate of carbonation spread in recycled concrete mixes with GRC and/or MRA than in the conventional material. A similar behaviour was reported for mixes with 25% to 50% MRA in combination with 25% fired clay powder processed from CDW and used as a cement substitute [12], and for mixes containing from 50% to 100% RCA, as well as 25% to 55% fly ash [60]. Table 8 lists the carbonation coefficients found for real or natural conditions (Kn) with Equation (8), proposed by Sisomphon and Franke [61] to express the relationship between accelerated and natural carbonation in terms of CO2 concentration: where Kn is the natural carbonation coefficient; Kac the accelerated carbonation coefficient; cn the CO2 concentration under natural conditions (∼0.04%) [62]; and cac the CO2 concentration in the accelerated carbonation test (3%). All the mixes studied exhibited values lower than or similar to the 4 mm/year 0.5 indicative of quality concrete [63] and none exceeded the critical value (6 mm/year 0.5 ), defined to mixes with low carbonation resistance [59]. Further to the classification proposed The accelerated carbonation coefficients (K ac ) given in Table 6 were calculated from linear regression of the carbonation depth/square root of time curve and Fick's first law (Equation (7)): where Cd is the mean carbonation depth at time t (mm); K ac , the carbonation rate under the experimental conditions (mm/year 0.5 ); d 0 , the carbonation depth at time = 0 (mm); and t, the exposure time (years). The reliability of the carbonation coefficients calculated was over 98% in all mixes (see R 2 in Table 6). K ac was 14.5% higher in mix N10/0, 106.6% in N25/0, 54.1% in R0/50, 66.9% in R10/50 and 117.0% in R25/50 than in mix NAC. Those findings denoted a higher rate of carbonation spread in recycled concrete mixes with GRC and/or MRA than in the conventional material. A similar behaviour was reported for mixes with 25% to 50% MRA in combination with 25% fired clay powder processed from CDW and used as a cement substitute [12], and for mixes containing from 50% to 100% RCA, as well as 25% to 55% fly ash [60]. Table 8 lists the carbonation coefficients found for real or natural conditions (K n ) with Equation (8), proposed by Sisomphon and Franke [61] to express the relationship between accelerated and natural carbonation in terms of CO 2 concentration: where K n is the natural carbonation coefficient; K ac the accelerated carbonation coefficient; c n the CO 2 concentration under natural conditions (~0.04%) [62]; and c ac the CO 2 concentration in the accelerated carbonation test (3%). All the mixes studied exhibited values lower than or similar to the 4 mm/year 0.5 indicative of quality concrete [63] and none exceeded the critical value (6 mm/year 0.5 ), defined to mixes with low carbonation resistance [59]. Further to the classification proposed by Greve-Dierfeld and Gehlen [64], N10/0 would lie within the same carbonation resistance range (RC3) as NAC, whilst mixes R0/50 and R10/50 would be in class RC4, and mixes N25/0 and R25/50 in class RC5.

Service Life Prediction
Carbonation spread across concrete exposed to different environments is plotted against exposure time in Figure 6, in keeping with the model set out in Spanish structural concrete code EHE-08 [16]. The dashed horizontal lines show the minimum rated cover for each exposure environment, cement type and design service life. The code defines rated cover as the distance between the outer surface of the reinforcement and the closest concrete surface. In light of the resulting curves, mixes in which 10% GRC or 50% MRA were included separately as well as jointly (R10/50) would be at no risk whatsoever of carbonation-induced reinforcement depassivation in structures with a service life of up to 100 years and design the characteristic strength as listed in Table 4. Mixes with 25% GRC, with or without recycled aggregate, would not be apt for reinforced concrete exposed to carbonation. by Greve-Dierfeld and Gehlen [64], N10/0 would lie within the same carbonation resistance range (RC3) as NAC, whilst mixes R0/50 and R10/50 would be in class RC4, and mixes N25/0 and R25/50 in class RC5.

Service Life Prediction
Carbonation spread across concrete exposed to different environments is plotted against exposure time in Figure 6, in keeping with the model set out in Spanish structural concrete code EHE-08 [16]. The dashed horizontal lines show the minimum rated cover for each exposure environment, cement type and design service life. The code defines rated cover as the distance between the outer surface of the reinforcement and the closest concrete surface. In light of the resulting curves, mixes in which 10% GRC or 50% MRA were included separately as well as jointly (R10/50) would be at no risk whatsoever of carbonation-induced reinforcement depassivation in structures with a service life of up to 100 years and design the characteristic strength as listed in Table 4. Mixes with 25% GRC, with or without recycled aggregate, would not be apt for reinforced concrete exposed to carbonation.

Conclusions
The conclusions that may be drawn from this study are set out below:

Conclusions
The conclusions that may be drawn from this study are set out below: • Incorporating GRC and/or MRA induces an increase in O 2 permeability associated with the greater porosity of these new materials than that found in their conventional counterparts, OPC and NA. Nonetheless, all the O 2 permeability coefficients were below the 4.5 × 10 −17 m 2 ceiling for quality concrete; • Clpermeability was not significantly affected (<8%) by the replacement of 10% OPC with GRC, irrespective of the aggregate type present (NA or MRA). The resulting 90 d mixes exhibited the same 'moderate' risk of corrosion as conventional concrete of the same age; • The high linear correlation between the O 2 permeability and Cldiffusion coefficients, irrespective of cement and aggregate type, can be interpreted as proof that these transport mechanisms are governed by both pore structure and interconnectivity; • The mean carbonation depth in mixes with 10% GRC, separately or jointly with 50% MRA, was 15% to 75% greater than that in mixes prepared with conventional cement and natural aggregate; • All mixes, irrespective of the GRC replacement ratio and aggregate type, exhibited CO 2 penetration coefficients lower than or similar to 4 mm/year 0.5 ; • The high correlation between O 2 permeability and carbonation depth suggests that the former may be a good indicator for predicting the latter; • According to the service life prediction model proposed in Spanish structural concrete code EHE-08, incorporating up to 10% GRC as a cement replacement and/or 50% MRA as an NA substitute does not compromise the reinforcement's passivity. Acknowledgments: This study benefitted from the Spanish Ministry of Science, Innovation and Universities mobility grant ETS18/00313, awarded in connection with pre-doctoral university professor scholarship 16/02693, under which Blas Cantero conducted research at the Lisbon University Instituto Superior Técnico construction laboratory. European Regional Development Fund (ERDF) funding was also received for an Interreg-POCTEP research grant (0008_ECO2CIR_4_E, 'Project for cross-border cooperation to introduce the eco-and circular economy by reducing waste generation and enhancing waste recycling, management and valorisation in Central Spain, Extremadura and Alentejo) and, in conjunction with the Government of Extremadura, under grant GR 18122, awarded to the MATERIA research group. The support of the Foundation for Science and Technology, Civil Engineering Research and Innovation for Sustainability (CERIS) research centre and Instituto Superior Técnico is also acknowledged.

Conflicts of Interest:
The authors have no conflict of interest that may have influenced the research described in this paper.