Carbonation and Restrained Shrinkage Cracking as Coupled Service-Life Controls in Low-Carbon Building Concrete: A Critical Review

Abstract

Embodied-CO₂ savings in clinker-reduced concrete must be evaluated alongside cover-zone durability and service life. This review examines carbonation and shrinkage/restrained cracking as coupled, rather than independent, controls on atmospheric service life in reinforced concrete buildings. The synthesis links binder route to cover-zone transport and alkalinity buffering and explains how curing, preconditioning, relative humidity (RH), CO₂ concentration, exposure geometry, and crack state affect reported carbonation rankings. It develops a two-domain conceptual framework for interpreting the effective carbonation front in cracked cover, with a descriptive crack-amplification index for reporting localized penetration. Practical mitigation is organized around mix-design levers, curing and construction actions, and a three-tier performance-based assessment workflow with a coupled test matrix. The review closes with a reporting checklist and five research gaps. The central conclusion is conditional: clinker-reduced building concrete can meet specified carbonation-performance requirements when curing, RH history, paste volume, and crack connectivity are assessed and managed together.

1. Introduction

Embodied-carbon reduction has made concrete mix design a central concern in low-carbon building delivery. The dominant carbon-intensive constituent is Portland cement clinker, whose production releases CO₂ through limestone calcination and high-temperature fuel combustion, two distinct emission sources that cannot be eliminated by energy switching alone [1,2,3]. In practice, two complementary strategies address this at the mixture level: clinker reduction, through partial replacement with supplementary cementitious materials (SCMs), limestone-containing cements, or limestone–calcined clay cement (LC3-type) systems [4,5,6]; and binder efficiency, achieved by meeting structural performance requirements with less cementitious paste [7,8,9]. Low-carbon concrete is therefore treated here as a mix-design concept with measurable descriptors, i.e., clinker factor, SCM replacement level, limestone and calcined clay content, total binder content, paste volume, water-to-binder (w/b) ratio, and binder efficiency, which is recordable in mix submittals and environmental product declarations, rather than as a generic sustainability label. Lower embodied CO₂ in building concrete must consequently be evaluated alongside the protective performance of the concrete cover under atmospheric exposure. Here, near-surface performance refers to the cover’s resistance to CO₂ and moisture transport, characterized by gas diffusivity, sorptivity, and permeability, together with its alkalinity-buffering capacity and the width, depth, and surface connectivity of shrinkage-induced cracks.

That evaluation cannot assume durability follows automatically from clinker reduction. Replacing clinker modifies portlandite availability, alkalinity reserve, hydrate assemblage, pore solution chemistry, and the balance between pore refinement and pore connectivity [4,10,11]. Under adequate curing, clinker-reduced systems may develop a refined pore network with reduced gas diffusivity and sorptivity; under early drying, the near-surface zone may remain incompletely reacted, with connected porosity and elevated carbonation sensitivity [11,12]. Compressive strength is not a reliable proxy for cover quality: concrete can satisfy strength requirements while retaining elevated transport coefficients in the surface layer [13,14]. For reinforced concrete buildings where chloride exposure is not the governing mechanism, carbonation is the principal atmospheric service-life process: progressive CO₂ ingress lowers pore-solution alkalinity and may depassivate reinforcement once the carbonation front reaches bar depth [15,16,17]. The risk is most relevant in sheltered facades, walls, slabs, parking levels, and interior or semi-exposed elements characterized by thin cover, variable curing histories, and intermediate relative humidity (RH) [18,19].

Shrinkage connects mixture design, construction practice, and cracking risk. In clinker-reduced systems, autogenous and drying shrinkage do not follow a single trend: low w/b and refined pore systems can increase autogenous shrinkage through self-desiccation in some SCM and LC3-type binders, while drying shrinkage is sensitive to paste volume, aggregate skeleton, curing duration, moisture gradients, and ambient RH [20,21,22,23]. Real building members are rarely free to deform. Reinforcement, structural continuity, substrates, and one-sided drying impose restraint that converts shrinkage strain into tensile stress, particularly in thin walls, slabs, and elements with large exposed surfaces [24,25]. Where early-age tensile capacity and stress relaxation are insufficient, surface microcracks or wider restrained cracks may develop before the cover zone has reached adequate maturity. These are not merely serviceability defects: surface-connected cracks provide preferential pathways for CO₂ ingress, increase local moisture exchange, and reduce effective cover quality [26,27,28].

Carbonation and restrained cracking therefore need to be interpreted together. In intact concrete, carbonation is commonly modeled as a matrix-controlled front governed by pore saturation, gas diffusivity, and alkalinity buffering [15,29]. In cracked cover, surface-connected cracks and distributed microcracks can shift the controlling domain: crack-wall carbonation may advance ahead of the surrounding matrix, and crack-domain transport can move the effective carbonation front beyond what intact-matrix properties predict [28,30,31]. This coupling also explains much of the apparent inconsistency in published carbonation results for clinker-reduced binders. Measured carbonation is also sensitive to test boundary conditions, i.e., curing duration, age at first exposure, preconditioning dryness, CO₂ concentration and exposure RH, specimen geometry and face sealing, and the crack-generation method [11,32,33,34]. Binder rankings reported in the literature may therefore reflect boundary conditions and testing methodology as much as binder chemistry.

Existing reviews provide essential foundations on cement decarbonization [1,2], SCM chemistry and availability [4,5,35], LC3 systems [6,36], carbonation of SCM-containing concretes [11,16], and carbonation-induced corrosion [17,18]. These studies establish the main mechanisms, but they often treat carbonation resistance, transport, shrinkage, cracking, and testing boundary conditions as separate evidence streams. For building-focused decision-making, the practical question is not simply whether a given SCM or replacement level improves or impairs carbonation resistance. The more relevant question is: under what curing duration, RH history, restraint condition, and crack-connectivity state does a clinker-reduced cover zone remain matrix-controlled, and when does carbonation become governed by surface-connected cracking? The contribution of this critical review is therefore not to re-review carbonation, shrinkage, or SCM chemistry separately, but to translate these evidence streams into a building-oriented service-life interpretation in which binder route, cover-zone transport, alkalinity buffering, restrained shrinkage, crack connectivity, and testing boundary conditions are considered together.

Accordingly, the aim of this critical review is to synthesize how clinker-reduced binder strategies influence service-life performance in building concrete when carbonation and shrinkage/restrained cracking act as coupled controls. The objectives are to: (i) classify binder and mixture routes relevant to building concrete; (ii) link binder modification to near-surface transport, microstructure, and carbonation buffering; (iii) explain carbonation mechanisms and shrinkage-cracking risk in building members; (iv) interpret cracking-induced carbonation acceleration through a two-domain conceptual framework; and (v) summarize practical mitigation, performance-based testing, and reporting recommendations. The two-domain framework is used as an interpretive tool to distinguish matrix-controlled carbonation from crack-domain transport, rather than as a standardized service-life model. The scope is limited to atmospheric, non-chloride-reinforced concrete building environments; chloride ingress, sulfate attack, alkali–silica reaction, freeze–thaw damage, marine exposure, early-age thermal cracking (a hydration-heat rather than shrinkage-driven mechanism), and detailed corrosion-propagation modeling fall outside its boundaries.

This critical review focuses on cover-zone transport and microstructure, alkalinity buffering, carbonation depth and coefficients, shrinkage, restrained-cracking response, and crack geometry and connectivity in low-carbon building concrete. Published findings are interpreted with respect to binder composition, mixture proportions, material scale, curing and preconditioning, exposure conditions, and crack-generation methods, where reported. Particular attention is given to how these factors distinguish intact-matrix carbonation from crack-assisted penetration and affect comparisons between studies. The review follows a focused literature-selection approach rather than a formal systematic review or meta-analysis. Details of the literature-selection approach are provided in the Supplementary Material, and the associated workflow is shown in Figure S1.

2. Low-Carbon Binder Routes Used in Building Concrete

This section narrows the broad low-carbon landscape to the binder and mixture routes most relevant to building practice and to the carbonation-shrinkage focus of this review. The classification is organized around two specification-relevant levers, i.e., clinker reduction and binder efficiency, because both directly influence embodied CO₂ while modifying the cover-zone variables that govern service life: carbonation buffering, pore refinement versus connectivity, curing sensitivity, paste volume, w/b, and cracking tendency [1,2,7]. Table 1 summarizes these routes, their main CO₂-reduction pathway, and their expected carbonation and shrinkage-cracking tendencies.

Table 1. Low-carbon binder routes for building concrete and expected implications for carbonation and shrinkage cracking.

Binder Route/Strategy	Main CO2-Reduction Pathway	Carbonation-Relevant Tendency	Shrinkage/Cracking-Relevant Tendency	Building-Practice Sensitivity	Refs.
Slag/GGBS blended cements	Latent-hydraulic clinker replacement	↑: Later-age pore refinement possible; carbonation can increase under short curing or dry preconditioning	V: Depends on paste volume, w/b, curing, early maturity	Needs sufficient curing before drying/exposure	[4,11,37]
Fly ash blended cements	Pozzolanic clinker replacement	↑: Strongly age-dependent; early exposure tends to increase carbonation depth	V: Drying shrinkage varies; cracking depends on curing, restraint, strength development	Sensitive to age at exposure and curing duration	[11,38]
Calcined clay/metakaolin blends	Reactive aluminosilicate replacement	V: Earlier pore refinement possible; outcome depends on portlandite consumption, curing, RH	↑ (autogenous): Fineness/reactivity may raise water demand and autogenous shrinkage if unadjusted	Mix adjustment and curing continuity decisive	[6,36]
Limestone-containing cements	Clinker dilution, packing, nucleation	↑: Depends on limestone content, w/b, curing, residual buffer; higher susceptibility reported at reduced buffer	V: Mainly governed by paste volume, w/b, aggregate skeleton	Should not be judged by strength class alone	[11,39]
LC3-type limestone–calcined clay systems	High clinker reduction with carboaluminate formation	V: Dense microstructure possible; carbonation remains sensitive to curing, RH, reduced buffering	V: Sensitive to clay reactivity/fineness, paste volume, w/b	Curing control needed in thin/early-dried members	[12,22,40,41]
Silica-fume ternary/high-performance blends	Binder efficiency plus partial replacement	V: Matrix densification can reduce transport; cracking may dominate under high restraint	↑ (autogenous): Low w/b and fine particles raise autogenous-shrinkage risk	Requires internal/external curing and crack control	[10,20]
Natural pozzolan and emerging SCM blends	Pozzolanic replacement	V: Source-dependent; governed by reactivity, fineness, curing maturity	V: Varies with water demand, fineness, replacement level	Requires source-specific performance testing	[5,35]
Binder-efficient optimized mixtures	Less binder/paste for required performance	V: Can reduce transport when compaction, w/b, curing controlled	↓ (drying), ↑ (autogenous): Lower paste volume may reduce drying-shrinkage demand, while very low w/b may increase autogenous shrinkage.	Requires workability control and curing discipline	[7,8,9,20,23]
CO2 curing/mineralization (boundary; precast-oriented)	CO2 uptake during production	NC: Not comparable with service-life carbonation of reinforced cover	NC: Process- and product-dependent	Keep separate from atmospheric carbonation	[42]
Alkali-activated binders (boundary)	Low-clinker/clinker-free chemistry	NC: Comparison with Portland systems confounded by different pore-solution chemistry	NC: System- and activator-dependent	Treat with AAB-specific mechanisms and standards	[43]

Note: ↑ and ↓ indicate a potential increase or decrease in carbonation depth or the specified shrinkage component under the stated conditions. V indicates a variable response for which no single direction is assigned. NC denotes a boundary route not ranked within this review. Arrows summarize conditional effects, not a universal ranking or an unconditional comparison with Portland cement. Greater carbonation depth indicates lower carbonation resistance under comparable exposure conditions. Shrinkage arrows refer to strain components, not directly to restrained-cracking probability.

2.1. SCM-Based Clinker Reduction

SCM-based clinker reduction remains the most established route, through factory-blended cements or batched additions. Slag (GGBS), fly ash, calcined clays, silica fume, and natural pozzolans reduce clinker demand through dilution and, depending on their compositions, a latent hydraulic or pozzolanic reaction [4,5,10]. The shared durability trade-off is that additional hydrate formation can refine pores and reduce transport at later ages, while portlandite consumption and lower clinker content reduce the alkalinity reserve available to buffer carbonation [11,37]. The carbonation response is conditional: with adequate curing and maturity, pore refinement may dominate; with short curing, early drying, or high replacement levels, a connected near-surface network can raise carbonation sensitivity [11,38]. The SCM family should not be treated as a single material response. Slag-rich systems can develop dense later-age microstructures, yet near-surface carbonation may remain sensitive to early curing and demolding practice [37,44]. Fly ash systems tend to be more age-dependent because pozzolanic reaction develops gradually, so the age at first exposure strongly affects measured carbonation [11,38]. Calcined clay and metakaolin blends can refine pores comparatively early, but their fineness and reactivity may increase water demand and autogenous shrinkage if proportions are not adjusted [6,36]. Silica fume is mainly relevant in dense ternary or high-performance blends; it densifies the matrix, but in low-w/b systems the associated self-desiccation and cracking risk must be managed [10,20]. Natural pozzolans and emerging SCMs are source-dependent: reactivity, fineness, water demand, and curing sensitivity should be assessed, not inferred from the material name [5,35].

2.2. Limestone-Containing and LC3-Type Binders

Limestone-containing cements reduce clinker content mainly through dilution, particle packing, nucleation effects, and formation of carbonate-bearing AFm phases [8,39]. These effects can support early hydration and packing, but carbonation behavior depends on limestone content, w/b, curing, and the remaining alkalinity buffer. Limestone-containing binders have shown greater carbonation susceptibility than corresponding Portland systems where the portlandite reserve is reduced and porosity is not compensated [11,45]. LC3-type systems combine clinker, calcined clay, limestone, and adjusted sulfate balance, exploiting calcined clay reactivity and limestone–alumina interactions to form carboaluminates at high clinker reduction [6,40]. Well-proportioned LC3 concretes can achieve mechanical performance comparable to Portland-cement references [12,36]. However, carbonation resistance remains sensitive to curing, exposure conditions, and the balance between transport resistance and buffering capacity [11,41]. Greater carbonation susceptibility has been reported for calcined clay systems than for plain Portland cement, but the comparison depends on the mixtures and test conditions [41]. Shrinkage depends on mixture proportions, paste volume, and aggregate restraint [22,23]. Restrained cracking additionally depends on tensile-strength development and stress relaxation [24,25]. These constraints matter most in elements with early drying, a high surface-to-volume ratio, or limited curing access.

2.3. Binder-Efficient Mixture Design

Binder-efficient design is a low-carbon route in its own right: instead of only replacing cement, the mixture reduces the total binder and paste volume required for a given workability, strength class, and cover-zone performance [7,8,9]. Practical levers include optimized aggregate grading, controlled w/b, high-range water-reducing admixtures (HRWR), and limits on unnecessary paste volume. The route matters here because paste volume and w/b influence both transport and shrinkage demand. Lower paste volume can reduce drying-shrinkage demand, while the aggregate skeleton restrains paste deformation. Differential shrinkage between paste and aggregate can also generate local microcracking [23,46]. The benefit is conditional: low w/b can increase self-desiccation and autogenous shrinkage even under sealed conditions, particularly in refined-pore systems [20,21,22]. Binder-efficient design is therefore most effective when paired with curing discipline, adequate compaction, and crack-width control.

2.4. Boundary Routes and Exclusions

Two low-carbon routes are retained only as boundaries of scope. CO₂ curing and mineralization are manufacturing-stage strategies, most relevant to precast elements, masonry units, and other controlled-production products [42]; intentional early-age mineralization binds CO₂ into solid carbonates and must not be confused with progressive service-life carbonation of reinforced cover. Alkali-activated binders are low-clinker or clinker-free, but their precursor and activator chemistry, pore-solution composition, carbonation response, shrinkage behavior, and standardization pathway differ substantially from blended Portland systems [43]. Their atmospheric carbonation acts primarily on the alkaline pore solution and, in calcium-bearing systems, on C-(A)-S-H, rather than on a portlandite reserve, and both laboratory and eight-year field data show carbonation and steel-passivity behavior that cannot be extrapolated from Portland-system experience [47,48,49]; they are therefore acknowledged but not synthesized here. Recycled concrete aggregate (RCA) is not a binder route; where used, its influence on absorption, interfacial transition zone (ITZ) quality, shrinkage, cracking, and transport should be evaluated as a mixture-level variable within the same framework [50].

2.5. Link to Transport, Carbonation, and Cracking

The classification matters because each route modifies the same controls in different proportions: alkalinity buffering, pore refinement versus connectivity, moisture retention, paste volume, w/b, and crack susceptibility. A slag, fly ash, or LC3 blend may perform well as a mature, well-cured matrix yet become carbonation-sensitive when early drying leaves the cover connected or cracked; a binder-efficient mixture may reduce shrinkage demand, but only if workability, compaction, and near-surface quality are maintained. Binder route is thus not a durability outcome by itself; it preconditions the transport and cracking state that later controls carbonation. Section 3 therefore treats transport and microstructure as the link between binder route, moisture state, carbonation rate, and cracking susceptibility.

3. Transport and Cover-Zone Microstructure

The binder routes discussed in Section 2 must be assessed in terms of the concrete cover separating the exposed surface from the reinforcement. Atmospheric CO₂ enters through this cover, where transport resistance and alkalinity buffering govern carbonation spread towards the steel [11,15]. Direct moisture exchange with the environment can produce a different moisture state from the more protected core, while inadequate curing can limit near-surface hydration and pore refinement [11,51]. Consequently, bulk compressive strength alone does not establish the protective quality of the cover [13,14]. This emphasis concerns carbonation-related reinforcement protection, rather than implying that core properties are unimportant. The following subsections examine pore connectivity, moisture state, and transport indicators as the link between binder route (Section 2) and carbonation behavior (Section 4).

3.1. Pore Refinement Versus Pore Connectivity

Clinker-reduced binders commonly show microstructural refinement at later ages as SCM reaction continues, reducing characteristic pore sizes and increasing tortuosity [4,52]. A matrix can exhibit reduced average pore size yet retain a connected capillary network when curing is insufficient or early drying limits further hydration and pozzolanic reaction [11,51]. Self-desiccation can additionally increase the gas-filled fraction of an existing connected network [20,21]. Clinker-reduced systems can therefore display apparent refinement while remaining transport-active under unsaturated conditions. This effect is amplified in building elements with high surface-to-volume ratio and variable curing, where the near-surface microstructure is less mature than the bulk. Connectivity is further influenced by ITZ and by microcracking: a porous ITZ, bleed channels, or early shrinkage microcracks can provide a connected network that dominates transport even when the paste matrix is refined [26,46,53]. This is one reason mixtures that appear dense in uncracked laboratory specimens can carbonate comparatively quickly in service.

3.2. Moisture State and RH-Controlled Transport

Transport coefficients in cementitious materials are not material constants; they depend on degree of saturation. CO₂ diffuses much faster through gas-filled than water-filled porosity, while dissolution and carbonate precipitation require an aqueous phase, so carbonation peaks at intermediate RH and slows at very low or very high saturation [29,54]. Low-carbon binders can be more moisture-sensitive than Portland references because continued hydration and SCM reaction depend on internal water availability: premature drying can lock in a more connected pore structure and raise sorptivity and gas diffusivity simultaneously [11,51]. Durability assessment must therefore consider the moisture preconditioning history and service RH regime explicitly, not treat transport indicators as intrinsic properties.

3.3. Key Transport Indicators: Sorptivity, Diffusivity, and Permeability

Three widely reported indicators are most informative for screening cover quality. Sorptivity (capillary absorption coefficient) is sensitive to near-surface pore connectivity and microcracking and is directly relevant to wetting–drying exposure; it typically decreases with age under sustained curing but remains elevated when curing is short [13,55]. Effective gas diffusivity controls CO₂ ingress under concentration gradients; its trend in clinker-reduced systems reflects the balance between refinement and saturation, so it should be interpreted together with curing duration and moisture state [15,29]. Permeability (to air or water) reflects the presence of continuous pathways under a pressure gradient and is especially defect-sensitive: a refined paste can still show high permeability when microcracks or connected ITZ regions are present, which makes site air-permeability methods useful screening tools for cover quality [13,26,56]. None of these indicators is a stand-alone durability rating; each describes an effective cover-zone quality under a specific conditioning path.

3.4. Why Intact-Matrix Transport Does Not Fully Describe Cracked Cover

Comparisons of intact-matrix transport indicators require comparable conditioning and a documented moisture state at testing, as discussed in Section 3.2 and Section 3.3 [11,13,51]. Carbonation depth in sound concrete is primarily diffusion-controlled, but real building cover deviates from idealized diffusion because of cracking and variable saturation. Surface-connected cracks can markedly increase transport through the cover. Water-permeability studies demonstrate strong crack-width dependence [26,57]. The influence of continuous cracks on steady-state diffusion has also been examined in cement-based materials [58]. CO₂-specific studies show preferential penetration and localized carbonation along crack paths [28,31]. Intact-specimen transport measurements remain relevant to the matrix between cracks but do not, by themselves, characterize preferential ingress through cracked cover. The chain developed in this review follows from that limitation: binder route and curing set pore refinement and saturation state; these set intact-matrix transport; shrinkage restraint sets the cracking state; and the combination sets the effective transport and carbonation response of the cover zone. Figure 1 summarizes this framework, which Section 4, Section 5 and Section 6 develop mechanism by mechanism.

Figure 1. Conceptual framework linking low-carbon binder route, curing/RH history, member-site conditions, and cover-zone controls to the effective carbonation front and reinforcement depassivation risk.

4. Carbonation in Low-Carbon Building Concrete

Carbonation is a coupled transport-reaction process: atmospheric CO₂ penetrates the partially saturated pore system, dissolves in pore water, and reacts with alkaline constituents, precipitating carbonates and reducing pore-solution alkalinity [15,16]. In reinforced concrete, the practical consequence is the progressive loss of the high-pH environment that passivates steel, with corrosion becoming possible once the carbonation front approaches bar depth [17,18]. In clinker-reduced systems, the response is governed by the balance between CO₂ transport capacity in the cover zone and the alkalinity buffering capacity of the hydrate assemblage, both of which are modified by the binder route, curing, and moisture history [11].

4.1. Alkalinity Buffering and Portlandite Availability

In Portland systems, portlandite is the primary carbonation buffer: its dissolution consumes CO₂ while sustaining high pH. Clinker reduction lowers this buffer in two ways. Dilution reduces the absolute portlandite generated per unit binder, and pozzolanic reaction consumes portlandite to form additional C–S–H and related hydrates [4,37]. Many low-carbon systems consequently carry a reduced reactive-CaO reserve, which increases the sensitivity of carbonation depth to CO₂ ingress even when later-age refinement improves transport resistance; analysis of 1044 concrete and mortar mixes examined relationships between composition, curing, exposure, and carbonation performance, with the statistical and machine-learning models using carbonation depth as the response variable [34]. Here, reactive CaO denotes the compositional descriptor calculated in [34], excluding calcium assigned to initial carbonate and sulfate carriers, rather than measured portlandite content. Carbonation in SCM-rich and LC3 binders proceeds not only through portlandite but also through decalcification of C–(A)–S–H and transformation of AFm/AFt phases, which alters microstructure in the carbonated zone and can coarsen porosity in low-portlandite systems [16,37,59]. The practical implication is that transport resistance and buffering are not necessarily aligned: a matrix can be less permeable yet lose alkalinity faster once CO₂ penetrates.

4.2. RH Dependence and Moisture Preconditioning

Carbonation requires both gas-phase CO₂ transport and an aqueous phase for reaction, so the rate peaks at intermediate RH, commonly reported around 50–70%, and falls at near-saturation or strong dryness [29,54]. For low-carbon concretes, this dependence is sharpened by microstructure: refined pores can remain water-filled and slow gaseous diffusion, while near-surface drying under poor curing shifts the cover into the moisture range favorable for CO₂ ingress [11,51]. Preconditioning before accelerated testing acts on the same physics. Severe pre-drying alters saturation, gas diffusivity, and moisture profiles, potentially biasing measured carbonation depths in short-duration accelerated tests [32,33]. Drying-induced microcracking is an additional possible effect [46]. Moisture histories in sheltered or intermittently wetted building elements depend on local climate and exposure. Carbonation comparisons should therefore consider the relevant RH and wetting conditions rather than infer them from the element type [19,38,54].

4.3. Curing Sensitivity of Clinker-Reduced Binders

Curing is a dominant variable for carbonation in clinker-reduced concrete because hydration and SCM reaction are moisture-dependent and often slower than Portland clinker hydration. Short curing leaves a less mature near-surface microstructure with higher connected porosity and elevated effective diffusivity precisely in the layer that governs ingress [11,12]. The effect is not an early-strength issue alone: curing duration determines capillary-network discontinuity, the degree of reaction achieved, and the extent of early drying microcracking, all of which persist into service. In building construction, early demolding, limited moist curing, wind exposure, and thermal gradients accelerate surface drying; for clinker-reduced systems, these conditions can simultaneously reduce transport resistance, increase shrinkage microcracking, and reduce buffering through incomplete hydrate formation [38,44]. For slag concrete, parameter analysis by Zhang et al. [44] predicts lower carbonation depth with longer curing before exposure. The broader database analysis by Vollpracht et al. [34] also identifies curing and pre-storage conditions as relevant to carbonation performance. Curing should therefore be specified as part of the material system.

4.4. Sources of Divergence in Reported Carbonation Performance

Published carbonation results for low-carbon binders often disagree because genuine material variables and test boundary conditions act together. Material variables include clinker factor, reactive-CaO/portlandite reserve, SCM reactivity and fineness, w/b, paste volume, air content, admixture system, and pore connectivity; boundary conditions include curing, preconditioning, CO₂ concentration, exposure RH, specimen size, exposed versus sealed faces, crack state, and the method used to define carbonation depth. These two groups of variables are not independent. For example, the same pre-drying regime can affect a slag-rich, fly ash-rich, or LC3-type concrete differently because hydration maturity, pore-size distribution, buffering capacity, and moisture retention are not the same. At least 17 standards and recommendations are used for carbonation testing, and they differ in curing, preconditioning, CO₂ concentration, RH, exposure geometry, and measurement procedure [32]. These differences can alter, and in some cases invert, reported rankings, especially where the tested concretes differ in curing sensitivity or buffering capacity. Table 2 summarizes the main divergence drivers, and Figure 2 illustrates for a single binder pair how test choices alone can reverse the apparent ranking.

Figure 2. Conceptual schematic showing how test conditions can reverse the carbonation-depth ranking of two hypothetical binder systems, A and B. Bar heights are illustrative, not measured. Greater depth indicates lower carbonation resistance within each test set. RH = relative humidity.

A central distinction is needed between relative ranking under matched accelerated protocols and direct conversion of accelerated carbonation coefficients into natural-exposure rates. Elevated CO₂ accelerates testing and is useful for controlled comparison, but it may alter carbonate phases, reaction-front morphology, moisture gradients, water-retention behavior, and microstructural evolution relative to natural exposure [16,34,60]. Direct comparison between natural exposure and 3% CO₂ exposure has shown mineralogical and microstructural differences, including effects on water retention and carbonation-induced cracking [60]. Similarly, five-year natural exposure data for concretes containing fly ash, slag, and limestone–calcined clay showed that accelerated and natural carbonation coefficients could not be directly correlated without a mixture- and concentration-dependent conversion approach [61]. These findings do not invalidate accelerated testing; they define its proper use. The 22-laboratory RILEM comparison reported analogous cement-type rankings under natural and accelerated exposure, while differences between indoor and sheltered outdoor mean rates were not statistically distinguishable within the interlaboratory scatter [62]. Accelerated carbonation may therefore support relative ranking under matched conditions, but it should not be used for direct service-life rate prediction unless the conversion is calibrated against natural-exposure data for the same mixture and exposure class.

Mixture variables beyond binder label, including w/b, paste volume, air content, admixtures, and reactive-CaO/portlandite reserve, can confound replacement-level comparisons [34,38]. Specimen geometry, sealing, surface preparation, and exposure directionality affect transport length and moisture gradients, while crack state and crack-generation method determine whether carbonation remains matrix-controlled or becomes locally crack-governed [27,28,31].

Endpoint definition adds another layer. Phenolphthalein identifies a pH threshold rather than complete carbonation, and two binders can reach the color boundary at different degrees of chemical carbonation because pore-solution chemistry, alkali buffering, moisture state, and carbonated hydrate assemblages differ [16,32]. The revised RILEM CPC-18R1 recommendation provides guidance on pH-indicator testing and reporting mean and maximum carbonation depths for nonuniform fronts [63]. Finally, the model used to extract a carbonation coefficient from depth–time data, including whether non-zero initial carbonation depth is considered, can change the reported coefficient [64].

4.5. Building-Relevant Interpretation

For building applications, carbonation resistance should be interpreted as a condition-dependent performance indicator, not as an intrinsic material property. Carbonation depth or carbonation coefficient is most meaningful when reported together with curing method and duration, age at first exposure, preconditioning route, exposure RH, CO₂ concentration, exposed-face configuration, specimen geometry, and crack state. At least one defect-sensitive transport proxy, either sorptivity or air permeability, should preferably be measured under documented, test-specific conditioning, with moisture state reported [13,14,32]. Gas diffusivity may provide additional transport information, while electrical resistivity should be treated as a complementary indicator, with its dependence on saturation and pore-solution chemistry acknowledged, as stated in Section 8.2. This reporting is particularly important for clinker-reduced systems, where strength-class equivalence does not necessarily imply equal carbonation resistance: concretes of similar compressive strength may differ in buffering capacity, pore-solution chemistry, near-surface maturity, and crack susceptibility.

Taken together, these findings support a conditional interpretation rather than a simple “better” or “worse” ranking of low-carbon binders. Clinker-reduced concretes tend to show higher measured carbonation when reduced portlandite/reactive-CaO buffering coincides with an immature, dried, or connected near-surface layer [11,34,38]. They may approach Portland-cement reference performance for a defined indicator, such as carbonation depth after a specified exposure or a calibrated carbonation coefficient, when curing and moisture state allow pore refinement to dominate ingress [41,61]. Natural-exposure studies provide mixture- and exposure-specific carbonation data for durability assessment [61,65,66]. Meeting a design service-life target additionally requires specified cover, a defined limit state, and an accepted prediction or qualification procedure (Section 7.3). Not all disagreement is therefore a testing artifact; real material differences remain central. The less-resolved building issue is the cracked cover state, where intact-matrix carbonation data may no longer represent the effective carbonation front. This motivates the shrinkage and coupled carbonation–cracking assessment developed in Section 5 and Section 6.

Table 2. Test and boundary-condition variables that control reported carbonation performance in low-carbon concrete.

Variable	Why It Matters	Typical Effect on Measured Carbonation	Minimum Reporting Requirement	Refs.
CO2 concentration: natural vs. accelerated	Alters reaction front, carbonate phases, moisture gradients, and microstructural evolution	Elevated CO2 accelerates testing; rankings may shift; direct coefficient conversion to natural exposure is not generally valid	CO2 level, temperature, RH, exposure duration, conversion/calibration approach if used	[16,34,60,61,67]
Preconditioning: drying severity and duration	Controls saturation, moisture gradients, and gas diffusivity; sufficiently severe drying may also induce microcracking	Severe drying can inflate depth and penalize moisture-sensitive or curing-sensitive binders	Target RH, temperature, duration, mass-stabilization criterion, sealing during preconditioning	[32,33,46]
Exposure RH	Sets the balance between gas diffusion and dissolution/reaction	Carbonation commonly peaks at intermediate RH; suppressed near saturation or strong dryness	Chamber RH, tolerance, monitoring frequency, exposure duration	[29,54]
Curing duration and age at first exposure	Controls SCM reaction maturity, near-surface pore continuity, and early drying damage	Short curing or early exposure can increase depth in clinker-reduced systems; extended moist curing may improve matrix resistance	Curing method, start time, duration, temperature/RH, age at carbonation exposure	[11,34,38,44]
Binder chemistry/buffering capacity	Controls CO2 demand, pH buffering, carbonation-front chemistry, and pH-indicator response	Lower portlandite or reactive-CaO reserve can increase measured depth even when matrix transport is refined	Clinker factor, SCM type/level, binder composition, CaO or reactive-CaO descriptor if available	[11,16,34,37,59]
Mixture variables beyond binder label	w/b, paste volume, air content, compaction, and admixtures affect transport and moisture retention	Higher w/b or inadequate compaction can increase depth. Effects of binder and paste content depend on binder chemistry, curing, and moisture conditions.	Full mix proportions, w/b, binder content, paste volume estimate, air content, admixture type/dosage	[34,38]
Specimen geometry and exposed faces	Sets transport length, edge effects, and moisture gradients	Multi-face exposure or small specimens can accelerate ingress relative to one-sided building exposure	Geometry, exposed/sealed faces, surface preparation, exposure directionality	[32,67]
Crack state and crack-generation method	Cracks shift control from matrix transport to crack-domain transport	Surface-connected cracks can raise local crack-wall depth and influence average depth; method changes connectivity	Crack width distribution, depth, density/spacing, connectivity, generation method, age at cracking, RH during measurement	[27,28,31]
Endpoint definition and model fitting	pH-indicator threshold and coefficient-extraction model differ between studies	Indicator moisture state and fitting assumptions can change reported depth/coefficient	Indicator/pH procedure, splitting condition, measurement locations, fitting model, treatment of initial depth	[16,32,63,64]

RH = relative humidity; SCM = supplementary cementitious material.

Table 2. Test and boundary-condition variables that control reported carbonation performance in low-carbon concrete.

Variable	Why It Matters	Typical Effect on Measured Carbonation	Minimum Reporting Requirement	Refs.
CO2 concentration: natural vs. accelerated	Alters reaction front, carbonate phases, moisture gradients, and microstructural evolution	Elevated CO2 accelerates testing; rankings may shift; direct coefficient conversion to natural exposure is not generally valid	CO2 level, temperature, RH, exposure duration, conversion/calibration approach if used	[16,34,60,61,67]
Preconditioning: drying severity and duration	Controls saturation, moisture gradients, and gas diffusivity; sufficiently severe drying may also induce microcracking	Severe drying can inflate depth and penalize moisture-sensitive or curing-sensitive binders	Target RH, temperature, duration, mass-stabilization criterion, sealing during preconditioning	[32,33,46]
Exposure RH	Sets the balance between gas diffusion and dissolution/reaction	Carbonation commonly peaks at intermediate RH; suppressed near saturation or strong dryness	Chamber RH, tolerance, monitoring frequency, exposure duration	[29,54]
Curing duration and age at first exposure	Controls SCM reaction maturity, near-surface pore continuity, and early drying damage	Short curing or early exposure can increase depth in clinker-reduced systems; extended moist curing may improve matrix resistance	Curing method, start time, duration, temperature/RH, age at carbonation exposure	[11,34,38,44]
Binder chemistry/buffering capacity	Controls CO2 demand, pH buffering, carbonation-front chemistry, and pH-indicator response	Lower portlandite or reactive-CaO reserve can increase measured depth even when matrix transport is refined	Clinker factor, SCM type/level, binder composition, CaO or reactive-CaO descriptor if available	[11,16,34,37,59]
Mixture variables beyond binder label	w/b, paste volume, air content, compaction, and admixtures affect transport and moisture retention	Higher w/b or inadequate compaction can increase depth. Effects of binder and paste content depend on binder chemistry, curing, and moisture conditions.	Full mix proportions, w/b, binder content, paste volume estimate, air content, admixture type/dosage	[34,38]
Specimen geometry and exposed faces	Sets transport length, edge effects, and moisture gradients	Multi-face exposure or small specimens can accelerate ingress relative to one-sided building exposure	Geometry, exposed/sealed faces, surface preparation, exposure directionality	[32,67]
Crack state and crack-generation method	Cracks shift control from matrix transport to crack-domain transport	Surface-connected cracks can raise local crack-wall depth and influence average depth; method changes connectivity	Crack width distribution, depth, density/spacing, connectivity, generation method, age at cracking, RH during measurement	[27,28,31]
Endpoint definition and model fitting	pH-indicator threshold and coefficient-extraction model differ between studies	Indicator moisture state and fitting assumptions can change reported depth/coefficient	Indicator/pH procedure, splitting condition, measurement locations, fitting model, treatment of initial depth	[16,32,63,64]

RH = relative humidity; SCM = supplementary cementitious material.

For resource-limited comparisons, prioritize matched curing method and duration, age at first exposure, and moisture preconditioning, with controlled exposure RH, CO₂ concentration, temperature, and common measurement times [32,54]. Maintain consistent specimen geometry, exposed/sealed faces, and depth-measurement and coefficient-fitting procedures [32,64]. State the mixture-comparison basis and, for cracked specimens, use a common crack-generation protocol and document crack width, depth, and surface connectivity [31]. Initial screening can use one curing/preconditioning regime and one carbonation environment, with additional regimes and geometries deferred to qualification. Advanced imaging may be omitted when conventional methods adequately characterize cracks (Section 6.3). These reductions concern testing breadth, not environmental tolerances, exposure duration, or replication required by the adopted standard. Rankings apply only to the tested conditions.

5. Shrinkage and Restrained Cracking

This section distinguishes autogenous, drying, and plastic shrinkage and examines their contribution to restrained cracking in building elements. Table 3 summarizes the mechanisms and their timing, while Section 5.1, Section 5.2, Section 5.3 and Section 5.4 address strain demand, tensile resistance, transport effects, and building-specific risk factors. Coupled carbonation behavior and mitigation are developed in Section 6 and Section 7, respectively.

Table 3. Shrinkage and restrained-cracking mechanisms relevant to carbonation performance of low-carbon building concrete.

Mechanism	Main Driver	Typical Timing	Low-Carbon Sensitivity	Durability Relevance	Refs.
Autogenous shrinkage	Self-desiccation driven by chemical shrinkage, internal RH reduction, and capillary tension in refined pores	After setting; early age to weeks/months, mainly at low w/b	More relevant in low-w/b and refined-pore systems; may increase in reactive calcined clay/LC3, silica-fume, and other dense blends; sensitive to clay reactivity, fineness, and sulfate balance	Can generate early tensile stress and microcracking under restraint before the cover zone develops sufficient tensile resistance	[20,21,22,68,69]
Drying shrinkage	Moisture loss to the environment; governed by paste volume, RH gradient, pore structure, and aggregate restraint	After exposure to drying; long-term under sustained or cyclic drying	Sensitive to curing duration, paste volume, pore structure, aggregate skeleton, and one-sided drying; not intrinsic to binder type alone	Can generate tensile stress and near-surface microcracking when differential or external restraint opposes drying shrinkage	[23,46]
Plastic shrinkage	Evaporation exceeding bleeding before final set	Fresh stage; before final set, often within the first hours after placement	Risk increases in low-bleeding, fine-powder, low-w/b systems under wind, heat, or low RH	Can create surface-connected cracks from day one, especially in slabs, toppings, and exposed horizontal members	[25,70]
Restrained cracking	Free shrinkage converted into tensile stress by internal or external restraint; cracking occurs when stress exceeds tensile capacity after relaxation/creep	Early age through drying exposure; may continue whenever shrinkage develops under restraint	Sensitive to curing, tensile-strength development, creep/relaxation, w/b, paste volume, SCM type, LC3 reactivity, and member restraint	Creates crack-connected transport paths; surface-connected cracks can accelerate crack-wall carbonation and shift the carbonation front from matrix-controlled to crack-influenced	[24,25,26,27,28,31,71,72]

5.1. Autogenous and Drying Shrinkage in Low-Clinker Binders

Autogenous shrinkage arises from self-desiccation: chemical shrinkage empties fine capillaries, internal RH drops, and the resulting capillary tension contracts the solid skeleton. It becomes significant at low w/b and in refined pore systems, where menisci form in smaller pores [20,21]. Several low-clinker systems sit in exactly this regime, because fine, reactive calcined clays refine porosity early. At concrete scale, the investigated LC3 mixtures showed higher later-age autogenous shrinkage than the Portland references, while total shrinkage developed similarly [22]. Paste-scale studies are useful for identifying the mechanism-level drivers, i.e., early fine-pore volume, clay reactivity, fineness, and sulfate balance, but they should not be transferred directly to building concrete because aggregate volume, aggregate stiffness, paste fraction, curing path, and member restraint all alter the concrete-scale response. With that caution stated, paste evidence confirms the mechanism and its conditionality: autogenous shrinkage of LC3 pastes rises with early fine-pore volume, clay reactivity, and fineness [68], while adjusting the sulfate level of the same LC3 paste measurably reduces shrinkage at equal clinker factor [69], and the trend is tunable, not intrinsic. Replacement level, clay reactivity, sulfate balance, and w/b can therefore move autogenous shrinkage in either direction relative to the reference [36,69]. Drying shrinkage results from moisture loss to the environment and is governed by paste volume, pore-size distribution, degree of hydration, ambient RH, and the aggregate skeleton that restrains the paste [23,46]. Not all low-clinker systems increase it: blends that slow moisture loss can reduce the shrinkage rate, while mixtures with immature, dried surfaces show higher early shrinkage and steeper gradients. Paste volume and aggregate skeleton often matter as much as the identity of the SCM [23,36]. In building elements under variable RH, drying shrinkage is frequently the dominant cracking driver, because the surface layer carries both the largest moisture gradient and the restraint of the wetter core. Boundary routes such as alkali-activated binders and CO₂-cured products should not be inferred from Portland-system shrinkage correlations: their gel chemistry, pore solution, and curing paths differ, so shrinkage under representative restraint is a qualification requirement for them rather than something assumed [42,43]. They are retained here only as boundary cases.

5.2. Early-Age Curing, Drying Gradients, and Tensile Resistance

Many low-clinker binders depend more strongly on sustained moisture curing to develop both microstructure and tensile capacity. When curing is shortened, the near-surface layer stays more porous and mechanically weaker while losing moisture faster, a combination of higher strain demand and lower cracking resistance [11,25]. Yuan et al. [73] reported that extending curing from 7 to 14 days reduced free shrinkage after 30 days of drying from 287 to 190 µε (33.8%) in concrete with 60% Grade 100 slag replacement by volume, compared with 277 to 260 µε (6.1%) in the Portland-cement control. The ranking between the two mixtures therefore reversed with curing duration: after 7 days of curing the slag mixture shrank slightly more than the control, whereas after 14 days it shrank appreciably less. In the same slag mixture, the curing effect narrowed at longer drying periods, to 383 versus 347 µε (9.4%) at 365 days, indicating that the curing benefit is largest at early drying ages. Both mixtures used granite aggregate, matched w/b and paste volumes, and drying conditions of 23 ± 2 °C and 50 ± 4% RH. Strains were referenced to demolding at approximately 24 h. These results quantify a mixture-specific curing effect on free shrinkage, not a corresponding reduction in restrained-cracking risk.

Early drying can also damage that layer through distributed microcracking that is hard to see but raises connectivity and transport [26,46]. Restrained-ring tests capture the competition directly: residual stress builds as shrinkage develops against restraint and is only partly relieved by creep and relaxation, so time-to-cracking depends on the race between stress development and tensile-strength growth [24,71]. Basic compressive-creep measurements on limestone–calcined clay pastes showed lower compliance than Portland reference pastes [74]. However, concrete-scale tests by Canda et al. [75] found higher early-age tensile creep coefficients in the investigated LC3 concretes than in the Portland reference, with the response varying with clay reactivity, replacement level, and age. Paste compressive creep should therefore not be used alone to infer reduced tensile-stress relaxation in LC3 concrete. The cracking outcome is also grade-specific: in a 21-mix ring program, slag blends and high-strength grades cracked earliest while fly ash delayed cracking, with stress-development rate the best predictor [72]. High surface-to-volume ratios can accelerate moisture loss, while one-sided drying can produce differential shrinkage. The resulting stress and cracking depend on moisture gradients, internal and external restraint, and tensile-property development [23,24,46,51].

These mechanisms all act after setting. A further risk precedes them: before the final set, evaporation exceeding the bleeding rate generates capillary suction at the fresh surface and can open plastic shrinkage cracks, which in slabs and toppings are wide and surface-connected from the first day [70]. Mixtures with reduced bleeding, including low-w/b systems, fine SCMs, and limestone–calcined clay combinations, are more exposed under wind, low RH, or elevated temperature, so evaporation control at placement is part of cracking-risk management rather than only a hot-weather precaution. The corresponding measures are addressed in Section 7.

5.3. Restrained Cracking as a Transport Accelerator

Cracking changes transport through the cover in two ways. First, discrete surface-connected cracks form preferential transport paths. Water-permeability studies demonstrate strong crack-width dependence [26,57], while the influence of continuous cracks on steady-state diffusion has been examined separately [58]. Second, drying-induced microcracks can alter the continuity of the pore and interfacial network [46]. Their potential transport consequences should be assessed through transport measurements together with crack characterization. Matrix refinement and cracking can coexist, so intact-specimen measurements alone may not represent cracked cover performance. Reported crack-width thresholds should not be treated as universal, because gas diffusion, water permeability, and crack-wall carbonation each respond to different aspects of crack geometry and are measured by different methods. Shrinkage and cracking therefore cannot be decoupled from carbonation: they act as a multiplier on transport-controlled deterioration rather than as a parallel serviceability topic.

5.4. Cracking Risk Drivers in Building Elements

Restrained-cracking risk is the imbalance between tensile-stress development and tensile resistance. For shrinkage-driven cracking, the dominant drivers in practice are: (a) high strain demand: low w/b with self-desiccation, high paste volume, refined pore structure, low ambient RH, wind, and heat [20,23]; (b) low early tensile resistance: short curing, slow strength development, and early demolding [11,25]; (c) insufficient stress relaxation relative to tensile-stress development [24,25]; (d) high restraint: reinforcement, continuity, subgrade, and layered systems [24]; (e) differential drying: nonuniform moisture loss and shrinkage, particularly under one-sided exposure [23,46,51]; and (f) crack connectivity: surface-connected cracks are particularly relevant because they provide preferential paths for CO₂ ingress. Most of these are construction- and detailing-dependent, so cracking performance in buildings is governed by curing practice and restraint conditions as much as by binder selection. The microcrack–macrocrack distinction should be kept explicit. Distributed microcracks may escape routine visual inspection and can be characterized by microscopy or image analysis [46]. Transport testing provides an indirect indication of their effects but also reflects pore structure and moisture state [13,51]. For discrete macrocracks, durability relevance is set less by surface width than by depth relative to cover, spacing, and surface connectivity, criteria developed in Section 6.

6. Coupled Carbonation–Cracking Risk

Existing models describe matrix carbonation [15], while cracked-concrete models and experiments address crack-assisted penetration [27,30,31]. The present framework integrates these mechanisms with binder-dependent buffering, curing, restrained shrinkage, and crack geometry to assess low-carbon building cover. Its contribution is to interpret mean matrix depth alongside localized penetration and surface connectivity using the effective carbonation front and descriptive crack-amplification index (Section 6.2). These measures inform the coupled test matrix and reporting checklist (Table 4 and Section 8.2). The framework complements predictive models through integrated assessment rather than by introducing a new transport law or claiming improved predictive accuracy.

Table 4. Recommended coupled test matrix for carbonation–shrinkage-cracking assessment of low-carbon building concrete. RH = relative humidity; w/b = water-to-binder ratio.

Test	Specimen Condition	Key Output	Relevance to Carbonation–Cracking Coupling	Minimum Reporting Requirement	Refs.
Plastic-shrinkage cracking (restrained panel)	Fresh concrete in a restrained panel with stress riser under controlled evaporation (wind, temperature, RH), with exposure duration and crack-reading ages specified by the adopted method (Tier 1 screening; applied to slabs, toppings, and members with high surface-to-volume ratio).	Crack-width statistics and supplementary crack area/density at stated observation ages, including post-setting readings where prescribed; time of first cracking; measured evaporation rate.	Captures the only window in which surface-connected cracks form before the cover has developed tensile capacity; such cracks enter the crack domain from day one.	Test method and deviations; exposure duration; setting-time determination; crack-reading ages; evaporation rate; wind speed, air and concrete temperature, RH; bleeding rate; time from placement to exposure; panel and restraint geometry; finishing practice; fiber type and dosage where used.	[70]
Autogenous shrinkage	Sealed; early-age temperature control (Tier 1; Tier 2 for low-w/b binders).	Autogenous strain vs. time.	Captures self-desiccation strain demand that may promote microcracking.	Method; material scale and specimen dimensions; binder and w/b; sealing and moisture-loss check; strain-zero age relative to setting; temperature history and thermal-strain treatment.	[20,21,76]
Drying/free shrinkage	Specimens prepared and conditioned under the adopted shrinkage method; supplementary sealed companions where component separation is investigated (Tier 1–2).	Length change and mass loss under stated conditioning; estimated autogenous and drying contributions where supplementary measurements permit.	Quantifies the strain demand feeding restrained cracking.	Standard used; curing; sealing; drying RH/temperature; start of measurement.	[23,77,78]
Restrained cracking (ring or comparable)	Representative restraint; curing matched to intended practice (Tier 2 qualification; Tier 3 verification).	Time to cracking and strain/stress-development indicators; supplementary crack-width and spacing measurements at stated RH.	Quantifies crack-domain formation under representative restraint.	Method and restraint geometry; curing/drying history; exposed/sealed faces; RH/temperature; crack-detection criterion; test duration; cracking age or no cracking by test end.	[24,71]
Crack connectivity and depth assessment (core)	After restrained-cracking testing; non-destructive mapping before carbonation exposure and destructive or moisture-altering assessment on matched companion specimens; RH/temperature recorded (core for Tier 2–3; simplified Tier 1 screening where cracks occur).	Surface-connectivity class; crack depth relative to cover; width distribution; spacing/density; detection method.	Distinguishes surface-connected pathways that can dominate carbonation acceleration from non-connected internal cracks.	Connectivity criterion; method and detection limit; RH/temperature at measurement; crack-generation method; age at cracking.	[27,28,31]
Carbonation (accelerated and/or natural)	Controlled preconditioning RH/duration; defined CO2/RH/temperature; one-face exposure preferred for building cover.	Matrix depth versus time; carbonation coefficient; maximum local depth normal to the original exposed surface; and penetration normal to crack walls, reported separately where cracks are present.	Measures matrix-domain front and crack-wall acceleration (two-domain behavior).	Curing and age at exposure; preconditioning; CO2/RH/temperature and duration; exposed faces; indicator formulation and reading time; measurement locations and crack-exclusion rule; fitting model; crack-connectivity class.	[32,63,64,67]
Transport proxy (sorptivity or air permeability)	Matched initial curing where compatible with the adopted methods; test-specific conditioning; companion specimens where testing would alter specimens reserved for carbonation.	Sorptivity or air-permeability coefficient with moisture state.	Defect-sensitive connectivity indicator bridging matrix and crack domains.	Moisture state at testing; test-specific conditioning; specimen identity; measurement location relative to cracks.	[13,55,56]
Advanced connectivity imaging (optional)	Cracked specimens at defined moisture state (optional Tier 3/research).	Crack-path evolution and length; spatial carbonation distribution.	Mechanistic validation where crack topology is central.	Resolution; segmentation criterion; field of view; moisture state.	[79]

Note: RH = relative humidity; w/b = water-to-binder ratio. Crack connectivity assessment is a core component for Tier 2 and Tier 3 coupled testing; for Tier 1 screening, simplified visual or optical classification may be sufficient. Standard ring testing provides a relative material comparison, not direct prediction of member cracking. Project-specific adaptations and departures from the standard must be identified [71]. Differences between sealed and unsealed strains provide an approximate component separation and should be interpreted with the effects of drying on hydration in mind [76].

6.1. A Two-Domain Framework: Matrix Domain and Crack Domain

The coupling can be described without additional chemistry through two transport domains. The matrix domain is the uncracked cover, where CO₂ ingress is diffusion-controlled through partially saturated porosity and moderated by buffering capacity (Section 4). The crack domain comprises surface-connected cracks and their adjacent affected zones, which shorten CO₂ transport paths and expose additional crack-wall material to carbonation [27,28,30]. Local moisture conditions remain an exposure-dependent control. Crack type determines which domain controls. Distributed microcracks mainly amplify matrix-domain connectivity; discrete macrocracks act as conduits and can dominate ingress when they are surface-connected and penetrate a substantial fraction of the cover [26,58]. Internal, non-connected cracks have limited influence unless they later link to the surface. Crack kinematics matter as well: openings vary with RH and temperature, so transient opening modulates CO₂ access during cycling, whereas sustained opening under restraint maintains a persistent shortcut [27,80]. Experimental evidence supports the framework. In cracked mortar with openings from 9 to 400 μm, diffusion along the crack path persisted down to about 9 μm, whereas carbonation perpendicular to the crack wall diminished markedly below roughly 41 μm [28]. Work on functionally layered concrete shows that flexural cracks penetrating a higher-cement durability layer allow the carbonation front to advance locally into the leaner interior, producing local extremes in carbonation depth around crack locations well ahead of the planar front [31]. Sustained mechanical loading interacts with the same physics: interlaboratory testing on SCM concretes found carbonation depths reduced under moderate compression but increased under high compressive or tensile load, consistent with load-induced changes in crack and pore connectivity; a review of load–carbonation interaction reports the same overall trend, while noting variability between test methods [81,82].

6.2. The Effective Carbonation Front in Cracked Cover

For cracked cover, it is useful to define an effective carbonation front as the envelope connecting the matrix-controlled carbonation depth between cracks with the locally deeper carbonated zones along surface-connected cracks. Carbonation depth is therefore treated as a spatial field rather than a single plane. Assessment of carbonation-related depassivation risk should consider localized penetration towards reinforcement, particularly where surface-connected cracks intersect the bar, rather than the average matrix front alone [27,30,31]. Tomographic observations of carbonation through surface-opening cracks in paste specimens provide additional support for this spatial interpretation [79]. For reporting within this framework, the dimensionless crack-amplification index at exposure time (t) is defined as:

$$I_{\mathrm{c}\mathrm{r}}\left(t\right)={\displaystyle \frac{x_{\mathrm{e}\mathrm{f}\mathrm{f},\mathrm{m}\mathrm{a}\mathrm{x}}\left(t\right)}{{\overline{x}}_{\mathrm{m}}\left(t\right)}}$$

where $x_{\mathrm{e}\mathrm{f}\mathrm{f},\mathrm{m}\mathrm{a}\mathrm{x}}$ is the maximum observed carbonation depth, including localized penetration along surface-connected cracks, and ${\overline{x}}_{\mathrm{m}}$ is the mean matrix depth outside crack-affected zones, determined using a stated exclusion distance. Both depths are measured within the same sampling window, perpendicular to the original exposed surface, using the same front-identification criterion. Crack length alone is not counted as carbonation depth. The crack-excluded matrix mean used in this index should be distinguished from the mean calculated under the adopted carbonation-depth measurement procedure. Values near unity indicate similar maximum and mean matrix depths, whereas larger values indicate greater localization. Because front irregularity also affects the ratio, cracking effects should be interpreted using crack maps and, where needed, matched uncracked controls. The index is descriptive, not a carbonation-rate multiplier or service-life predictor. Comparisons require matched conditioning, exposure, and measurement procedures. Both depths and the sampling details specified in Section 8.2 should be reported. The index should not be calculated when a matrix region outside the crack influence cannot be identified, or its mean depth cannot be reliably distinguished from zero.

For illustration, consider concrete with a 25 mm cover, measured to the nearest reinforcement surface, and a surface-connected shrinkage crack directly above the bar. At a given exposure time (t), assume a mean matrix carbonation depth of ${\overline{x}}_{\mathrm{m}}$ = 8 mm away from the crack and a maximum local depth of $x_{\mathrm{e}\mathrm{f}\mathrm{f},\mathrm{m}\mathrm{a}\mathrm{x}}$ = 20 mm beside it. Both depths are measured perpendicular to the exposed surface, giving $I_{\mathrm{c}\mathrm{r}}=20/8=2.5$. The matrix depth alone suggests a remaining distance of 17 mm to reinforcement, whereas the local crack-associated front is only 5 mm away. The two-domain interpretation thus captures localized penetration that the mean matrix depth would overlook. These assumed values illustrate the framework at one exposure time and do not represent experimental data or a service-life prediction.

Two practical implications follow from this spatial interpretation. First, maximum crack width alone is an incomplete predictor: while wider cracks generally produce deeper local penetration [31], modeling and experiments indicate that crack depth relative to cover and surface connectivity determine whether that local penetration reaches the reinforcement zone [28,30]. Second, autogenous self-healing can partially close fine cracks under intermittent wetting, particularly at early ages, so crack-domain transport is time-dependent; this mitigation is real but should not be assumed in design without verification under representative RH cycling [80]. Figure 3 illustrates the two-domain geometry.

Figure 3. Schematic of two-domain carbonation in cracked concrete cover, contrasting a planar matrix-controlled front in intact cover with localized crack-wall carbonation and enhanced connectivity in cracked cover.

6.3. What to Measure in Coupled Assessment

A coupled assessment treats crack state and carbonation state as co-evolving variables. Within the two-domain framework, whether carbonation remains matrix-controlled or shifts into crack-domain transport depends on connectivity to the exposed surface. Crack connectivity is therefore a core measurement rather than an optional observation: a narrow surface-connected crack can be more relevant to carbonation than a wider internal one because it provides direct CO₂ access to crack walls and can promote localized advance of the carbonation front [27,28,31]. Crack metrics should be reported as distributions rather than single values, comprising width distribution with the RH and temperature at measurement, density or spacing, depth relative to cover, an explicit surface-connectivity class, and the generation method and age at cracking [27,32]. A connectivity class should record whether a crack is open to the exposure face and continuous over a measurable depth, and whether that depth is shallow relative to cover or sufficient to influence the reinforcement zone; classes are project- or study-defined rather than universal. Established methods are adequate for most purposes, including optical crack mapping, image analysis or digital image correlation (DIC), dye penetration, sectioning, microscopy, and crack-depth gauges; X-ray computed tomography with suitable image analysis can provide additional information on crack-path evolution and spatial carbonation distribution, but is not required for routine qualification in the proposed workflow [31,79]. Carbonation should be mapped for localization, reporting matrix depth over time and local penetration normal to crack walls under fully specified exposure conditions [28,31]. The indicator procedure, depth-calculation method, and timing of readings after indicator application should also be reported [63]. At least one defect-sensitive transport proxy, either sorptivity or air permeability measured under documented, test-specific conditioning with moisture state reported, should accompany the carbonation data as a connectivity indicator [13,56]. Acceleration factors from different studies remain incomparable unless curing, preconditioning, RH, CO₂ concentration, exposure faces, crack-generation method, age at cracking, and connectivity class are all reported [32,33].

7. Practical Mitigation and Performance-Based Testing

This section translates the preceding mechanistic discussion into mix-design measures, curing and construction actions, and a tiered performance-based assessment workflow.

7.1. Mix-Design Levers

Reducing w/b with high-range water reducers rather than added paste lowers capillary connectivity and slows CO₂ transport in the intact matrix; because self-desiccation and autogenous shrinkage can increase at low w/b even without external drying, w/b reduction should be coupled to curing discipline or internal curing [20,21]. Binder selection and replacement level should balance carbonation buffering, transport resistance, and curing sensitivity [11,41]. The suitability of moderate replacement levels or ternary blends should be verified under the intended curing and restraint conditions rather than inferred from binder category. Total binder content and paste volume should be controlled to limit unnecessary material use and drying-shrinkage demand [7,8,23]. Where clinker replacement requires additional binder to meet performance requirements, the net carbon benefit should be evaluated from constituent quantities and their emission factors rather than replacement level alone [2]. Aggregate grading, volume fraction, and stiffness influence paste demand and internal restraint of shrinkage [23,46]. Internal curing with pre-wetted lightweight fines or superabsorbent polymers is most valuable in low-w/b, SCM-rich systems prone to self-desiccation, where it maintains internal RH and reduces early microcrack connectivity in the cover; mechanisms, dosing strategies, and side effects are consolidated in dedicated reviews [83,84,85]. Shrinkage-reducing admixtures lower capillary tension and shrinkage magnitude but require compatibility checks with the binder system and curing regime and remain crack-control tools rather than substitutes for curing [76,86]. Their benefit is binder-dependent: early-age restrained-stress tests under mixture-specific temperature histories showed a more pronounced SRA effect in the investigated slag concrete than in the fly ash concrete [76]. Dosage should therefore be qualified for the binder and test conditions. Polypropylene fibers can mitigate plastic-shrinkage cracking, with effectiveness depending on fiber geometry and dosage [70]. SCM fineness and limestone filler improve packing and later refinement but can raise early water demand and shrinkage, so adjustments should be verified with defect-sensitive proxies rather than compressive strength alone [8,39].

7.2. Curing and Construction Actions

The first priority is to prevent early drying of the surface layer through an appropriate curing method. Membrane-forming curing compounds are liquid products, commonly based on waxes or resins, that form a surface film to reduce evaporation [87]. They retain existing moisture rather than supply additional curing water [87,88]. Their effectiveness depends on water-retention performance, application timing and rate, and uniform film coverage [87]. Compatibility with subsequent coatings or finishes should also be checked, with removal where the film would impair adhesion [87]. Their suitability should therefore be assessed for the intended mixture and site conditions rather than assumed equivalent to wet curing [87]. For clinker-reduced binders, curing duration should be specified as part of the material system because inadequate curing can leave a less mature, more connected surface layer [11]. Model-based analysis of slag concrete also predicts a lower carbonation depth with longer curing before exposure [44]. At the construction level, restraint should be managed as a durability variable: joint layout, saw-cut timing, and casting sequence determine where cracks form and whether they remain narrow and confined to locations that can be sealed or detailed. Finishing timing and protection against wind and low RH limit plastic and early drying shrinkage in elements with high surface-to-volume ratios. Crack inspection should be scheduled after formwork removal and initial drying, and again after early RH cycling where relevant; recorded crack width should be accompanied by the RH and temperature at measurement and a surface-connectivity class, because openings are transient. Thresholds should be project- and exposure-specific, used as triggers for further assessment or mitigation rather than as universal service-life limits [27].

7.3. Performance-Based Assessment Workflow

A performance-based approach specifies indicators relevant to carbonation and crack-domain formation rather than binder recipes alone, consistent with the broader shift toward performance-based durability specification [14,89]. A three-tier workflow is proposed. Tier 1 (screening) compares candidate mixes using defect-sensitive transport proxies and basic shrinkage indicators using a consistent protocol across candidate mixes within each test, with test-specific conditioning, together with plastic-shrinkage cracking where the intended members have a high surface-to-volume ratio, eliminating mixes with high near-surface connectivity or high shrinkage demand. Tier 2 (qualification) couples the mechanisms: restrained-cracking testing representative of the intended restraint, followed by carbonation exposure under controlled RH, with both matrix and crack-wall carbonation measured to capture two-domain behavior. At this stage, formal carbonation-resistance qualification is established using companion specimens that meet the preparation, curing, conditioning, and exposure requirements of the adopted procedure, with a qualified reference used where permitted [32,67]. Intentionally cracked specimens provide supplementary evidence of localized ingress and should not be used to assign a standard resistance class unless the accepted procedure explicitly permits their use. Tier 3 (verification) repeats the qualified mix under project-representative curing, with method and duration reproducing site practice and one-sided exposure where relevant, to confirm that transport proxies, crack metrics, and carbonation progression remain within the qualified envelope. Acceptance limits should be taken from applicable standards or from project service-life design rather than from fixed generic values. Accelerated carbonation results remain comparable only when curing, preconditioning, exposure RH and CO₂ concentration, and exposure faces are controlled and reported, since standardized methods differ substantially on precisely these parameters [32,33,67]. Table 4 consolidates the corresponding test matrix across the three tiers.

Acceptance should reflect exposure class, design service life, cover requirements and tolerances, and applicable national provisions. Where adopted, EN 1992-1-1:2023 links carbonation exposure-resistance classes (XRC) to exposure, minimum durability cover, and design life [90]. Within the proposed workflow, Tier 1 permits progression when predefined transport and shrinkage criteria are met, without establishing service-life compliance. Tier 2 requires an accepted carbonation-resistance classification or equivalence to a reference already qualified for the project requirements, alongside project-defined cracking criteria. Extrapolation from accelerated tests to natural-exposure service life requires appropriate calibration [32]. Tier 3 verifies cover, curing execution, and project-representative performance against qualified limits. Exceedances require reassessment or corrective measures before acceptance. Crack measurements supplement code requirements, while the crack-amplification index remains descriptive rather than a universal pass/fail criterion.

8. Discussion: Synthesis, Reporting Needs, and Research Gaps

The preceding sections can now be read together as a single conditional statement about when clinker-reduced cover performs and when it does not.

8.1. Conditional Performance: When the Governing Balance Changes

Across Section 2, Section 3, Section 4, Section 5, Section 6 and Section 7, performance under building carbonation exposure is controlled by the balance of three states: the alkalinity buffer, which falls as clinker factor decreases; the effective CO₂ transport of the partially saturated cover; and the presence or absence of a crack domain that bypasses matrix tortuosity. Favorable outcomes arise when curing duration and method allow the near-surface pore network to mature before exposure, when the RH history avoids aggressive early drying, when surface-connected cracking is limited in width and continuity under restraint, when w/b is reduced through admixture strategy rather than paste addition, and when paste volume is balanced against an optimized aggregate skeleton so that shrinkage demand remains moderate [11,23,34]. Performance deteriorates when short or ineffective curing and early drying leave an immature, connected, microcracked cover; when restraint in thin, one-sided-drying members produces sustained surface-connected cracks; when binder-heavy proportions raise shrinkage demand; and when aggressive pre-drying or unrepresentative accelerated conditions stand in for service exposure [32,33,38]. Table 5 condenses the representative evidence behind this reading, and Figure 4 summarizes its microstructural basis: the same clinker reduction that dilutes the portlandite buffer can produce either a refined, poorly connected pore system under sustained curing or a connected, microcracked one after early drying, and cracking then redirects carbonation along the paths of least resistance.

Table 5. Evidence synthesis for the coupled carbonation–cracking assessment of low-carbon building concrete. Representative studies are illustrative, not exhaustive.

Evidence Theme	Representative Studies	Condition/Range Extracted	Semi-Quantitative Observation	Critical Interpretation	Transfer Limitation	Building Implication
Carbonation of clinker-reduced binders in intact specimens	[11,34]	RILEM database-level evidence: 1044 concrete/mortar mixes with carbonation depth–time data; variables include binder family, w/b, curing/pre-storage, exposure, and reactive-CaO descriptors [34]	The w/(reactive CaO) ratio combines transport- and buffering-related descriptors, but its association with carbonation must be interpreted alongside curing, exposure, and other mixture variables.	Binder family alone is insufficient; buffering capacity, w/b, curing, and exposure define the measured carbonation response	Database combines heterogeneous protocols and mixture classes	Specify clinker factor, SCM type/level, w/b, curing/preconditioning, and reactive-CaO or binder-chemistry descriptors where available
Accelerated vs. natural carbonation	[60,61,62,65,66]	Natural vs. 3% CO2 comparison [60]; 34-concrete program: five-year natural open/sheltered exposure and 112-day accelerated exposure at 1% CO2, with seven mixes also tested at 3% CO2 [61]; 12-year natural-exposure data for blended cements [65]	Accelerated and natural coefficients are not directly interchangeable; harmonized accelerated protocols may still support relative ranking [61,62]	Separate ranking from rate conversion; accelerated tests are useful for screening, not for uncalibrated service-life prediction	Climate, exposure mode, binder system, and conversion model remain dataset-specific	Use accelerated carbonation for comparative qualification only when conditioning, RH, CO2 level, exposed faces, and fitting model are reported
Conditioning and standardization sensitivity	[32,33,46,67]	RILEM review documents numerous standards and recommendations with differing curing, preconditioning, CO2/RH exposure, and measurement procedures [32]; severe preconditioning analyzed in accelerated tests [33]	Drying/preconditioning changes moisture profiles and gas transport; drying-induced microcracking is an additional possible effect [46]	Test boundary conditions can shift reported rankings, especially for curing- and moisture-sensitive binders	Few protocols explicitly couple crack state, RH history, and carbonation measurement	Report the complete curing and preconditioning path; use matched companion specimens for transport, carbonation, and crack assessment
LC3/SCM shrinkage and early-age cracking tendency	[22,68,69,72]	LC3 concrete evidence across 25, 32, and 45 MPa grades [22]; LC3 paste experiments using six calcined clays from Australia, France, and India, linking early pore refinement to autogenous shrinkage [68]; sulfation-level effects [69]; restrained-cracking data for fly ash/slag concretes [72]	LC3/SCM autogenous shrinkage can increase with fine-pore development and internal-RH drop, but total concrete shrinkage and cracking depend on aggregate skeleton, grade, sulfation, and curing	Shrinkage is a mixture- and curing-dependent response, not an intrinsic property of “low-carbon” binders	Paste-scale evidence cannot be transferred directly to building concrete; aggregate restraint and drying gradients alter response	Distinguish paste, mortar, and concrete evidence; verify restrained cracking under representative curing and member restraint
Crack width, depth, and connectivity effects on carbonation	[26,27,28,31,57,58,79,81,82]	Crack openings of 9–400 μm tested; measurable carbonation perpendicular to the crack wall falls off below ~41 μm, with diffusion along the crack still occurring at 9–41 μm [28]; low-pressure water-permeability tests on specimens with crack widths of 50–350 μm showed strongly increasing permeability with increasing crack width [57]; loaded SCM concretes investigated in an interlaboratory comparison [81], with the broader loading-carbonation literature reviewed in [82]	Gas diffusion, crack-wall carbonation, and water permeability do not share one universal crack-width threshold	Crack effect depends on width, depth, density/spacing, surface connectivity, loading/restraint state, and RH-dependent opening	Saw-cut and mechanically loaded cracks may not reproduce restrained-shrinkage microcrack halos	Report width distribution, depth, density/spacing, surface connectivity, crack-generation method, and RH at measurement; avoid relying on maximum width alone
Mitigation effectiveness: internal curing, SRA, fibers	[70,76,83,84,85,86]	Internal-curing/SAP evidence from RILEM round-robin and review data [83,84,85]; SRA at 1% of total binder mass was evaluated in concretes based on Portland, fly ash, and slag cements, all also containing silica fume, with deformation monitored for approximately 2.5 years [76]; polypropylene-fiber geometry tested for plastic-shrinkage cracking [70]	Internal curing, SRA, and fibers can reduce shrinkage strain or crack formation, but the magnitude depends on binder, dosage, curing, and restraint	Mitigation effectiveness is not directly transferable across binder systems	Dosage, workability, strength, and curing interactions differ between mixtures	Qualify mitigation per binder system using matched curing and restrained-cracking tests; treat fibers and SRA as crack-control tools, not carbonation cures
Performance-based testing and reporting comparability	[14,32,67,71,77,78]	Current standards cover separate indicators: accelerated carbonation [67], restrained-ring cracking [71], and free shrinkage [77,78]; RILEM review documents variability in carbonation procedures [32]	Comparable coupled assessment requires documented shared histories, test-specific conditioning, and companion-specimen allocation	A carbonation coefficient without crack-state and conditioning alignment describes only the tested specimen state	The standards reviewed here do not provide a single integrated protocol for the proposed carbonation–restrained-cracking assessment	Use a tiered matrix: screening by transport/shrinkage, qualification by restrained cracking + carbonation, verification under project-representative curing

Note: RH = relative humidity; SCM = supplementary cementitious material; LC3 = limestone–calcined clay cement; SRA = shrinkage-reducing admixture; SAP = superabsorbent polymer. Accelerated tests can support relative ranking under matched protocols, but carbonation coefficients should not be converted to natural-exposure rates without mixture- and exposure-specific calibration.

Figure 4. Schematic microstructural basis of carbonation response in Portland and SCM/LC3 cover concrete. Well-cured SCM/LC3 cover combines lower CH reserve with refined, poorly connected pores, whereas early drying and cracking increase capillary/microcrack connectivity and localize carbonation along transport-favorable paths.

Much of the apparent contradiction in the literature is reconciled by this conditional interpretation without recourse to any new mechanism. SCM-rich and LC3-type systems may approach Portland-reference carbonation performance under favorable mixture, curing, and exposure conditions, but equivalence is not guaranteed. Reduced buffering can remain consequential even in well-cured cover, while early drying and cracking can introduce additional transport pathways [11,41]. Because the test choices identified in Section 4.4 act on precisely this balance, studies differing in preconditioning, conditioning path, or crack state may report reversed rankings within a single binder family [32,33,34]. Accelerated carbonation consequently retains value as a comparative tool within explicitly defined conditioning while becoming misleading across unmatched conditions, and intact-specimen carbonation depth should not be treated as a stand-alone material property for curing- and cracking-sensitive systems [16,34].

The framework carries further implications for specification. Prescriptive mixture and cover provisions establish a compliance route, but do not directly quantify every project-specific interaction between curing, restrained cracking, and localized carbonation. The proposed coupled assessment supplements the applicable durability provisions with additional performance evidence [14,89]. Performance-based specification is more defensible, though only where the tests are themselves defect-sensitive, since a specification founded on intact-core transport indices may certify a mixture whose in-service cover is microcracked and freely carbonating. Such frameworks should therefore pair a transport indicator with a restrained-cracking criterion and verify both under project-representative curing (Section 7.3). It should be noted that much of the coupled evidence derives from laboratory crack-generation methods of uncertain field representativeness; the quantitative thresholds cited in this review are accordingly indicative ranges rather than design values [28,81].

Embodied-carbon savings should be assessed against the structural function and target service life of the building element, rather than production-stage emissions per cubic meter alone [2,44]. Comparisons should use a common assessment period and consistent system boundaries, considering initial emissions alongside any protection, repair, or replacement needed to maintain performance. Such interventions may reduce the initial carbon advantage, but this must be evaluated rather than assumed [2]. For carbonation exposure, the model-based optimization by Zhang et al. [44] illustrates how carbon-efficient mixture design changes when carbonation-durability constraints are added to strength requirements. Within the present framework, both matrix resistance and crack-associated penetration should therefore inform the durability qualification of a low-carbon mixture. Carbonation reaching reinforcement, however, is not equivalent to damaging corrosion or the end of structural service life [18]. The framework identifies cover-performance evidence relevant to carbon-service-life assessment without providing a quantitative prediction of lifetime emissions.

8.2. Reporting Checklist for Comparable Studies

Comparability across coupled studies requires, as a minimum:

• Binder composition: clinker factor, SCM types and replacement levels, limestone and calcined clay contents, fineness descriptors.
• Mixture proportions: w/b, total binder content, estimated paste volume, aggregate volume fraction and grading.
• Curing regime: method, start time after finishing, duration, temperature and RH, demolding time, and evaporation conditions during the pre-set period.
• Preconditioning: target RH or moisture state, temperature, duration, mass-stabilization criterion, sealing.
• Carbonation exposure: CO₂ concentration, RH, temperature, duration, exposed versus sealed faces, specimen geometry, surface preparation.
• Crack generation: method (restrained shrinkage, plastic shrinkage under controlled evaporation, mechanical, saw-cut), restraint description, age at cracking, stabilization before exposure.
• Crack metrics under stated RH: width distribution, density or spacing, depth, surface-connectivity classification.
• Carbonation measurement: front-identification criterion, indicator procedure and reading time, specimen moisture condition, and sampling locations in the matrix and beside cracks. For the crack-amplification index, report both component depths, sampling-window dimensions, crack-exclusion distance, measurement resolution, and numbers of sections and specimens examined. State the model and time units used when fitting carbonation coefficients.
• Transport proxy: at least one of sorptivity or air permeability, with test-specific conditioning, moisture state at testing, and specimen identity reported. Electrical resistivity may be included as a complementary indicator, with its dependence on saturation and pore-solution chemistry acknowledged.
• Conditioning alignment statement: shared initial histories where appropriate, test-specific conditioning, and specimen allocation across transport, shrinkage, crack, and carbonation measurements. Identify matched companion specimens used for destructive or moisture-altering measurements and report relevant differences in age, prior history, and crack state.

Where carbonation depth is determined with a pH indicator, item 8 and the exposure-duration element of item 5 should follow the revised RILEM recommendation CPC-18R1, which specifies the indicator procedure, the definitions of carbonation depth and carbonation front, and the associated reporting requirements [63].

8.3. Targeted Research Gaps

The gaps below are framed as minimum study designs rather than open themes, because the principal deficit in the field is not a shortage of data but a shortage of mutually comparable data; uncoordinated testing under heterogeneous conditioning would add volume without resolving the disagreements traced in Section 4.4 [32,34]. Five are most consequential.

• Standardized RH preconditioning for coupled tests remains under-defined, although drying level influences moisture profiles and CO₂ transport. The minimum design is fixed preconditioning at two RH states, with documented duration and mass stabilization, before identical carbonation exposure [32,33].
• Crack-wall carbonation profiling is rarely quantified, obscuring the effective carbonation front. The minimum design is simultaneous mapping of matrix and crack-wall carbonation depth at defined spatial intervals under a controlled RH history [28,31].
• The representativeness of crack-generation methods is unresolved, since saw-cuts and mechanically induced cracks may not reproduce the microcrack halos and connectivity of restrained shrinkage cracks. The minimum design is a restrained-shrinkage cracking protocol with quantified width distribution, depth, and connectivity established prior to carbonation exposure [71,81].
• Field-relevant curing variability is under-sampled for clinker-reduced binders despite their curing sensitivity. The minimum design compares short with extended curing, and membrane with wet or sealed curing, followed by identical preconditioning and exposure [34,44].
• Autogenous self-healing is not consistently separated from moisture effects under RH cycling, limiting interpretation of crack-domain persistence. The minimum design is controlled wet–dry conditioning with repeated crack-width and connectivity mapping and localized carbonation profiling at fixed exposure RH [80].

9. Conclusions

This review examined carbonation and restrained shrinkage cracking as coupled controls on the atmospheric service life of low-carbon concrete in buildings. The following conclusions are drawn.

• Low-carbon concrete is best treated as a measurable mix-design and cover problem rather than a binder label. Clinker factor, SCM type and level, limestone or calcined clay content, w/b, paste volume, and binder efficiency are the descriptors that govern carbonation and cracking and should be reported as such.
• Carbonation in clinker-reduced binders is conditional, set by the balance between cover transport capacity and alkalinity buffering. Because pore refinement and buffering are not necessarily aligned, strength equivalence does not imply carbonation equivalence; clinker-reduced concrete can nonetheless approach Portland-cement reference performance for specified carbonation indicators under defined curing, RH, and crack-state conditions.
• Curing and RH history must form part of the performance definition, particularly for curing-sensitive clinker-reduced binders, since short curing and early drying leave an immature, connected, microcracked cover. Reported carbonation rankings depend on curing, preconditioning, CO₂ concentration, RH, exposed faces, and endpoint and model choices, so many apparent contradictions reflect the interaction between real material differences and unharmonized boundary conditions. Accelerated tests can support relative rankings within a defined protocol, while natural-exposure predictions require appropriate calibration.
• Shrinkage and restrained cracking are durability controls rather than serviceability issues alone. Autogenous, drying, and plastic shrinkage act through different mechanisms and timings, and the resulting behavior is mixture- and curing-specific rather than intrinsic to clinker reduction or to LC3; under restraint, shrinkage can generate surface-connected cracks when tensile-stress development exceeds the available resistance.
• Cracked cover should be interpreted through the effective carbonation front, which combines matrix carbonation with locally deeper crack-wall carbonation. Maximum crack width alone is insufficient: crack depth relative to cover, density, surface connectivity, RH-dependent opening, and the crack-generation method together determine whether carbonation remains matrix-controlled or shifts to crack-connected transport.
• Durability in practice follows from coordinating binder selection, paste-volume and w/b control, curing discipline, restraint-aware detailing, and crack-connectivity management. Performance-based assessment should combine defect-sensitive transport proxies, restrained-cracking tests, carbonation testing, and crack-wall and connectivity mapping under project-representative curing.

Durable low-carbon building concrete therefore requires carbon reduction, cover-zone maturation, and crack-connectivity control to be specified together rather than treated as separate objectives.