Alkali–silica reaction

This deleterious chemical reaction causes the expansion of the altered aggregate by the formation of a soluble and viscous gel of sodium silicate (Na₂SiO₃ · n H₂O, also noted Na₂H₂SiO₄ · n H₂O, or N-S-H (sodium silicate hydrate), depending on the adopted convention). This hygroscopic gel swells and increases in volume when absorbing water: it exerts an expansive pressure inside the siliceous aggregate, causing spalling and loss of strength of the concrete, finally leading to its failure.

ASR can lead to serious cracking in concrete, resulting in critical structural problems that can even force the demolition of a particular structure. The expansion of concrete through reaction between cement and aggregates was first studied by Thomas E. Stanton in California during the 1930s with his founding publication in 1940.

Chemistry

To attempt to simplify and to stylize a very complex set of various reactions, the whole ASR reaction, after its complete evolution (ageing process) in the presence of sufficient Ca²⁺ cations available in solution, could be compared to the pozzolanic reaction which would be catalysed by the undesirable presence of excessive concentrations of alkali hydroxides (NaOH and KOH) in the concrete. It is a mineral acid-base reaction between NaOH or KOH, calcium hydroxide, also known as portlandite, or (Ca(OH)₂), and silicic acid (H₄SiO₄, or Si(OH)₄). For simplification, after a complete exchange of the alkali cations with the calcium ions released by portlandite, the alkali-silica reaction in its ultimate stage leading to calcium silicate hydrate (C-S-H) could be schematically represented as follows:

Ca(OH)₂ + H₄SiO₄ → Ca²⁺ + H₂SiO₄²⁻ + 2 H₂O → CaH₂SiO₄ · 2 H₂O

Here, the silicic acid H₄SiO₄, or Si(OH)₄, which is equivalent to SiO₂ · 2 H₂O represents hydrous or amorphous silica for the sake of simplicity in aqueous chemistry.

Indeed, the term silicic acid has traditionally been used as a synonym for silica, SiO₂. Strictly speaking, silica is the anhydride of orthosilicic acid, Si(OH)₄.

SiO₂↓ + 2 H₂O ⇌ Si(OH)₄

An ancient industrial notation referring to H
₂SiO
₃, metasilicic acid, is also often used to depict the alkali-silica reaction. However, the metasilicic acid, H
₂SiO
₃, or SiO(OH)
₂, is a hypothetic molecule which has never been observed, even in extreme diluted solutions because H
₂SiO
₃ is unstable and continues to hydrate.

Indeed, contrary to the hydration of CO₂ which consumes only one water molecule and stops at H₂CO₃, the hydration of SiO₂ consumes two water molecules and continues one step further to form H₄SiO₄. The difference in hydration behaviour between SiO₂ and CO₂ is explained by thermodynamic reasons (Gibbs free energy) and by bond energy or steric hindrance around the central atom of the molecule.

This is why the more correct geochemical notation referring to the orthosilicic acid Si(OH)
₄ really existing in dilute solution is preferred here. However, the main advantage of the now deprecated, but still often used, industrial notation referring to the metasilicate anion (SiO^2–
₃), which also does not exist in aqueous solution, is its greater simplicity and its direct similitude in notation with the carbonate (CO^2–
₃) system.

One will also note that the NaOH and KOH species (alkali hydroxides, also often simply called alkali to refer to their strongly basic character) which catalyze and accelerate the silica dissolution in the alkali-silica reaction do not explicitly appear in this simplified representation of the ultimate reaction with portlandite, because they are continuously regenerated from the cation exchange reaction with portlandite. As a consequence, they disappear from the global mass balance equation of the catalyzed reaction.

Silica dissolution mechanism

The surface of solid silica in contact with water is covered by siloxane bonds (≡Si–O–Si≡) and silanol groups (≡Si–OH) sensitive to an alkaline attack by OH⁻
ions.

The presence of these oxygen-bearing groups very prone to form hydrogen bonds with water molecules explains the affinity of silica for water and makes colloidal silica very hydrophilic.

Siloxane bonds may undergo hydrolysis and condensation reactions as schematically represented hereafter:

≡Si–O–Si≡ + H
₂O ↔ ≡Si–OH + HO–Si≡

=Si=O + H
₂O ↔ =Si(–OH)
₂

On the other hand, silanol groups can also undergo protonation/deprotonation:

≡Si–OH ↔ ≡Si–O⁻
+ H⁺
.

These equilibria can be shifted towards the right side of the reaction leading to silica dissolution by increasing the concentration of the hydroxide anion (OH^–), i.e., by increasing the pH of the solution.

Alkaline hydrolysis of siloxane bonds occurs by nucleophilic substitution of OH^– onto a silicon atom, while another ^–O–Si group is leaving to preserve the tetravalent character of Si atom:

≡Si–O–Si≡ + OH⁻
→ ≡Si–OH + ^–O–Si≡

=Si=O + OH⁻
→ =SiO(OH)⁻

Deprotonation of silanol groups:

≡Si–OH + OH⁻
→ ≡Si–O⁻
+ H
₂O.

In the pH range 0 – 7, the solubility of silica is constant, but above pH=8, the hydrolysis of siloxane bonds and deprotonation of silanol groups exponentially increase with the pH value. This is why glass easily dissolves at high pH values and does not withstand extremely basic NaOH/KOH solutions. Therefore, NaOH/KOH is released during cement hydration attacks and dissolves the tridimensional network of silica present in the aggregates. Amorphous or poorly crystallized silica, such as cryptocrystalline chalcedony or chert present in flints (in chalk) or rolled river gravels, is much more soluble and sensitive to alkaline attack by OH^– anions than well crystallized silica such as quartz. Strained (deformed) quartz or chert exposed to freeze-thaw cycles in Canada and Nordic countries are also more sensitive to alkaline (high pH) solutions.

The species responsible for silica dissolution is the hydroxide anion (OH^–). The high pH conditions are said to be alkaline and one also speaks of the alkalinity of the basic solutions. For the sake of electroneutrality, (OH^–) anions need to be accompanied by positively charged cations, Na⁺ or K⁺ in NaOH or KOH solutions, respectively. Na and K both belong to the alkali metals column in the Periodic Table. When speaking of alkalis, one systematically refers to NaOH and KOH basic hydroxides, or their corresponding oxides Na₂O and K₂O in cement. Therefore, it is the hydroxide, or the oxide, component of the salt which is the only relevant chemical species for silica dissolution, not the alkali metal in itself. However, to determine the alkali equivalent content (Na₂O_eq) in cement, because of the need to maintain electroneutrality in solids or in solution, one directly measures the contents of cement in Na and K elements and one conservatively considers that their counter ions are the hydroxide ions. As Na⁺ and K⁺ cations are hydrated species, they also contribute to retain water in alkali-silica reaction products.

Osmotic processes (Chatterji et al., 1986, 1987, 1989) and the electrical double layer (EDL) play also a fundamental role in the transport of water towards the concentrated liquid alkali gel, explaining their swelling behavior and the deleterious expansion of aggregates responsible of ASR damages in concrete.

Catalysis of ASR by dissolved NaOH or KOH

The ASR reaction significantly differs from the pozzolanic reaction by the fact that it is catalysed by soluble alkali hydroxides (NaOH / KOH) at very high pH. It can be represented as follows using the classical geochemical notation for representing silica by the fully hydrated dissolved silica (Si(OH)₄ or silicic acid: H₄SiO₄), while in an older industrial notation the non-existing (H₂SiO₃, hemihydrated silica, is considered in analogy to carbonic acid):

2 Na(OH) + H₄SiO₄ → Na₂H₂SiO₄ · 2 H₂O

The so-produced soluble alkali silicagel can then react with calcium hydroxide (portlandite) to precipitate insoluble calcium silicate hydrates (C-S-H phases) and regenerate NaOH, thus continuing the initial silica dissolution reaction:

Na₂H₂SiO₄ · 2 H₂O + Ca(OH)₂ → CaH₂SiO₄ · 2 H₂O + 2 NaOH

The combination of the two above mentioned reactions gives a general reaction resembling the pozzolanic reaction, but it is important to keep in mind that this reaction is catalysed by the undesirable presence in cement, or other concrete components, of soluble alkaline hydroxides (NaOH / KOH) responsible for the dissolution of the silica (silicic acid) at high pH:

Ca(OH)₂ + H₄SiO₄ → CaH₂SiO₄ · 2 H₂O

Without the presence of dissolved NaOH or KOH, responsible for the high pH (~13.5) of the concrete pore water, the amorphous silica of the reactive aggregates would not be dissolved and the reaction would not evolve. Moreover, the soluble sodium or potassium silicate is very hygroscopic and swells when it absorbs water. When the sodium silicate gel forms and swells inside a porous siliceous aggregate, it first expands and occupies the free porosity. When this latter is completely filled, and if the soluble but very viscous gel cannot be easily expelled from the silica network, the hydraulic pressure rises inside the attacked aggregate and leads to its fracture. The hydro-mechanical expansion of the damaged siliceous aggregate surrounded by calcium-rich hardened cement paste is responsible for the development of a network of cracks in concrete. When the sodium silicate expelled from the aggregate encounters grains of portlandite present in the hardened cement paste, an exchange between sodium and calcium cations occurs and hydrated calcium silicate (C-S-H) precipitates with a concomitant release of NaOH. In its turn, the regenerated NaOH can react with the amorphous silica aggregate, leading to an increased production of soluble sodium silicate. When a continuous rim of C-S-H completely envelops the external surface of the attacked siliceous aggregate, it behaves as a semi-permeable barrier and hinders the expulsion of the viscous sodium silicate while allowing the NaOH / KOH to diffuse from the hardened cement paste inside the aggregate. This selective barrier of C-S-H contributes to increase the hydraulic pressure inside the aggregate and aggravates the cracking process. It is the expansion of the aggregates which damages concrete in the alkali-silica reaction.

Portlandite (Ca(OH)₂) represents the main reserve of OH^– anions in the solid phase as suggested by Davies and Oberholster (1988) and emphasized by Wang and Gillott (1991). As long as portlandite, or the siliceous aggregates, has not become completely exhausted, the ASR reaction will continue. The alkali hydroxides are continuously regenerated by the reaction of the sodium silicate with portlandite and thus represent the transmission belt of the ASR reaction driving it to completeness. It is thus impossible to interrupt the ASR reaction. The only way to avoid ASR in the presence of siliceous aggregates and water is to maintain the concentration of soluble alkali (NaOH and KOH) at the lowest possible level in concrete, so that the catalysis mechanism becomes negligible.

Analogy with the soda lime and concrete carbonatation

The alkali-silica reaction mechanism catalysed by a soluble strong base as NaOH or KOH in the presence of Ca(OH)₂ (alkalinity buffer present in the solid phase) can be compared with the carbonatation process of soda lime. The silicic acid (H₂SiO₃ or SiO₂) is simply replaced in the reaction by the carbonic acid (H₂CO₃ or CO₂).

In the presence of water or simply ambient moisture, the strong bases, NaOH or KOH, readily dissolve in their hydration water (hygroscopic substances, deliquescence phenomenon), and this greatly facilitates the catalysis process because the reaction in aqueous solution occurs much faster than in the dry solid phase. The moist NaOH impregnates the surface and the porosity of calcium hydroxide grains with a high specific surface area. Soda lime is commonly used in closed-circuit diving rebreathers and in anesthesia systems.

The same catalytic effect of the alkali hydroxides (function of the Na₂O_eq content of cement) also contributes to the carbonatation of portlandite by atmospheric CO₂ in concrete although the rate of propagation of the reaction front is there essentially limited by the CO₂ diffusion within the concrete matrix less porous.

The soda lime carbonatation reaction can be directly translated into the ancient industrial notation of silicate (referring to the never observed metasilicic acid) simply by substituting a C atom by a Si atom in the mass balance equations (i.e., by replacing a carbonate by a metasilicate anion). This gives the following set of reactions also commonly encountered in the literature to schematically depict the continuous regeneration of NaOH in ASR:

If NaOH is clearly deficient in the system under consideration (soda lime or alkali-silica reaction), it is formally possible to write the same reactions sets by simply replacing the CO₃^2- anions by HCO₃⁻ and the SiO₃^2- anions by HSiO₃⁻, the principle of catalysis remaining the same, even if the number of intermediate species differs.

Main sources of OH− in hardened cement paste

One can distinguish several sources of hydroxide anions (OH⁻
) in hardened cement paste (HCP) from the family of Portland cement (pure OPC, with BFS, or with cementitious additions, FA or SF).

Direct sources

OH⁻
anions can be directly present in the HCP pore water or be slowly released from the solid phase (main buffer, or solid stock) by the dissolution of Ca(OH)
₂ (portlandite) when its solubility increases when high pH value starts to drop. Beside these two main sources, ions exchange reactions and precipitation of poorly soluble calcium salts can also contribute to release OH⁻
into solution.

Alkali hydroxides, NaOH and KOH, arise from the direct dissolution of Na
₂O and K
₂O oxides produced by the pyrolysis of the raw materials at high temperature (1450 °C) in the cement kiln. The presence of minerals with high Na and K contents in the raw materials can thus be problematic. The ancient wet manufacturing process of cement, consuming more energy (water evaporation) that the modern dry process, had the advantage to eliminate much of the soluble Na and K salts present in the raw material.

As previously described in the two sections dealing respectively with ASR catalysis by alkali hydroxides and soda lime carbonatation, soluble NaOH and KOH are continuously regenerated and released into solution when the soluble alkali silicate reacts with Ca(OH)
₂ to precipitate insoluble calcium silicate. As suggested by Davies and Oberholster (1988), the alkali-silica reaction is self-perpetuating as the alkali hydroxides are continuously regenerated in the system. Therefore, portlandite is the main buffer of OH⁻
in the solid phase. As long as the stock of hydroxides in the solid phase is not exhausted, the alkali-silica reaction can continue to proceed until the complete disparition of one of the reagents (Ca(OH)
₂ or SiO
₂) involved in the pozzolanic reaction.

Indirect sources

There exist also other indirect sources of OH⁻
, all related to the presence of soluble Na and K salts in the pore water of hardened cement paste (HCP). The first category contains soluble Na and K salts whose corresponding anions can precipitate an insoluble calcium salts, e.g., Na
₂SO
₄, Na
₂CO
₃, Na
₃PO
₄, NaB(OH)
₄, Na
₂B
₄O
₇, ... .

Hereafter, an example for calcium sulfate (gypsum, anhydrite) precipitation releasing sodium hydroxide:

Na
₂SO
₄ + Ca(OH)
₂ → CaSO
₄ + 2 NaOH

or, the reaction of sodium carbonate with portlandite, also important for the catalysis of the alkali–carbonate reaction as emphasized by Fournier and Bérubé (2000) and Bérubé et al. (2005):

Na
₂CO
₃ + Ca(OH)
₂ → CaCO
₃ + 2 NaOH

However, not all Na or K soluble salts can precipitate insoluble calcium salts, such as, e.g., NaCl-based deicing salts:

2 NaCl + Ca(OH)
₂ ← CaCl
₂ + 2 NaOH

As calcium chloride is a soluble salt, the reaction cannot occur and the chemical equilibrium regresses to the left side of the reaction.

So, a question arises: can NaCl or KCl from deicing salts still possibly play a role in the alkali-silica reaction? Na⁺
and K⁺
cations in themselves cannot attack silica (the culprit is their counter ion OH⁻
), and soluble alkali chlorides cannot produce soluble alkali hydroxide by interacting with calcium hydroxide. So, does it exist another route to still produce hydroxide anions in the hardened cement paste (HCP)?

Beside portlandite, other hydrated solid phases are present in HCP. The main phases are the calcium silicate hydrates (C-S-H) (the "glue" in cement paste), calcium sulfo-aluminate phases (AFm and AFt, ettringite) and hydrogarnet. C-S-H phases are less soluble (~ 10⁻⁵ M) than portlandite (CH) (~ 2.2 10⁻² M at 25 °C) and therefore are expected to play a negligible role for the calcium ions release.

An anion-exchange reaction between chloride ions and the hydroxide anions contained in the lattice of some calcium aluminate hydrates (C-A-H), or related phases (C-A-S-H, AFm, AFt), is suspected to also contribute to the release of hydroxide anions into solution. The principle mechanism is schematically illustrated hereafter for C-A-H phases:

Cl⁻
+ (C-A-H)–OH → (C-A-H)–Cl + OH⁻

As a simple, but robust, conclusion, the presence of soluble Na and K salts can also cause, by precipitation of poorly soluble calcium salt (with portlandite, CH) or anion exchange reactions (with phases related to C-A-H), the release of OH⁻
anions into solution. Therefore, the presence of any salts of Na and K in cement pore water is undesirable and the measurements of Na and K elements is a good proxy (indicator) for the maximal concentration of OH⁻
in pore solution. This is why the total alkali equivalent content (Na
₂O
_eq) of cement can simply rely on the measurements of Na and K (e.g., by ICP-AES, AAS, XRF measurements techniques).

Alkali gel evolution and ageing

The maturation process of the fluid alkali silicagel found in exudations into less soluble solid products found in gel pastes or in efflorescences is described hereafter. Four distinct steps are considered in this progressive transformation.

1. SiO
₂ dissolution and Na
₂SiO
₃ formation (here, explicitly written in the ancient industrial metasilicate notation (based on the non-existing metasilicic acid, H
₂SiO
₃) to also illustrate the frequent use of this later in the literature):

2 NaOH + SiO
₂ → Na
₂SiO
₃ · H
₂O (young N-S-H gel)

This reaction is accompanied by hydration and swelling of the alkali gel leading to the expansion of the affected aggregates. The pH of the fresh alkali gel is very high and it has often a characteristic amber color. The high pH of young alkali gel exudations often precludes the growth of mosses at the surface of concrete crack infilling.

2. Maturation of the alkali gel: polymerisation and gelation by the sol–gel process. Condensation of silicate monomers or oligomers dispersed in a colloidal solution (sol) into a biphasic aqueous polymeric network of silicagel. Ca²⁺
divalent cations released by calcium hydroxide (portlandite) when the pH starts to slightly drop may influence the gelation process.

3. Cation exchange with calcium hydroxide (portlandite) and precipitation of amorphous calcium silicate hydrates (C-S-H) accompanied by NaOH regeneration:

Na
₂SiO
₃ + Ca(OH)
₂ → CaSiO
₃ + 2 NaOH

Amorphous non-stoechiometric calcium silicate hydrates (C-S-H, the non-stoechiometry being denoted here by the use of dashes) can recrystallize into rosettes similar to these of gyrolite. The C-S-H formed at this stage can be considered an evolved calcium silicate hydrate.

4. Carbonation of the C-S-H leading to precipitation of calcium carbonate and amorphous SiO₂ stylized as follows:

CaSiO
₃ + CO
₂ → CaCO
₃ + SiO
₂

As long as the alkali gel (Na
₂SiO
₃) has not yet reacted with Ca²⁺
ions released from portlandite dissolution, it remains fluid and can easily exude from broken aggregates or through open cracks in the damage concrete structure. This can lead to visible yellow viscous liquid exudations (amber liquid droplets) at the surface of affected concrete.

When pH slowly drops due to the progress of the silica dissolution reaction, the solubility of calcium hydroxide increases, and the alkali gel reacts with Ca²⁺
ions. Its viscosity increases due to gelation process and its mobility (fluidity) strongly decreases when C-S-H phases start to precipitate after reaction with calcium hydroxide (portlandite). At this moment, the calcified gel hardens, hindering therefore the alkali gel transport in concrete.

When the C-S-H gel is exposed to atmospheric carbon dioxide, it undergoes a rapid carbonation, and white or yellow efflorescences appear at the surface of concrete. When the relatively fluid alkali gel continues to exude below the hardened superficial gel layer, it pushes the efflorescences out of the crack surface making them to appear in relief. Because the rates of the gel drying and of the carbonation reactions are faster than the gel exudation velocity (liquid gel expulsion rate through open cracks), in most of the cases, fresh liquid alkali exudates are not frequently encountered at the surface of civil engineering concrete structures. Decompressed concrete cores can sometimes let observe fresh yellow liquid alkali exudations (viscous amber droplets) just after their drilling.