Alkali-Resistant Glass Fibres: Why AR Glass Makes or Breaks GRC

Walk up to almost any major architectural façade built with glass fibre reinforced concrete in the last 40 years and you’re looking at a material that, on the inside, is quietly engaged in a slow chemical battle. The cement matrix that gives GRC its compressive strength is simultaneously one of the most hostile environments a glass fibre can be asked to survive. Understanding that tension – and how the construction industry learned to manage it – is the key to understanding why AR glass matters so much, and why getting it wrong has such serious long-term consequences.

Why Ordinary Glass Fibre Fails in Cement

The problem starts with chemistry. When ordinary Portland cement hydrates, it releases calcium hydroxide as a byproduct, creating an intensely alkaline pore solution with a pH typically around 12.5. Standard borosilicate glass – the E-glass widely used in fibreglass composites – is no match for this environment. The –Si–O–Si– bonds that form the structural backbone of silicate glass are attacked and broken by hydroxyl ions, causing the fibre surfaces to corrode progressively. As the surface degrades, microscopic flaws develop, and because the extraordinary tensile strength of glass in fibre form (which can reach 3,000–3,500 MPa for a single filament) depends entirely on maintaining a pristine, defect-free surface, even limited surface damage causes a sharp and disproportionate drop in load-carrying capacity.

Early attempts to use E-glass fibres in OPC-based cement composites confirmed this in the most unambiguous way possible – the fibres simply degraded too quickly to provide durable reinforcement. This is why E-glass is not used in GRC production. The question then was whether a glass could be formulated that would hold up over decades in such a hostile environment.

The AR Glass Solution: Zirconium Dioxide

The answer came in the late 1960s from a UK research collaboration between A.J. Majumdar at the Building Research Station and A. Tallentire at Pilkington Glass Co. Their work produced the first commercially available alkali-resistant glass fibres – marketed as Cem-Fil – based on the incorporation of zirconium dioxide (ZrO₂) into the glass formulation. Japanese manufacturers, including Nippon Electric Glass and Asahi Glass, were developing similar high-zirconia fibres at the same time.

Zirconia works because it stabilises the silicate glass network, significantly slowing the rate at which hydroxyl ions can attack the –Si–O–Si– bonds. A minimum ZrO₂ content of 16% by weight is required to provide sufficient alkali resistance, and current commercial AR fibres contain between 16% and 19% ZrO₂. The typical chemical composition looks roughly like this:

Higher ZrO₂ contents improve resistance further, but also make the glass increasingly difficult and costly to manufacture, which sets a practical upper limit.

Physical Structure: From Filament to Roving

AR glass fibres don’t arrive at the production line as loose individual strands. They come in a carefully organised hierarchy that directly affects how they behave both during manufacture and in the hardened composite.

Individual filaments are drawn from molten glass with diameters typically between 12 µm and 18 µm. A size coating – applied at nanometre scale immediately after drawing – keeps filaments bundled together, protects their pristine surfaces from abrasion, and critically, controls how the fibre will bond chemically and mechanically to the cement matrix. This sizing is not a cosmetic detail: research on single-fibre pull-out behaviour has demonstrated that the nature of the size coating directly governs where interfacial cracks initiate and propagate – whether through the matrix material adjacent to the fibre, or through the weaker interface between the fibre and its coating. That distinction matters enormously for how energy is absorbed when the composite fractures.

Typically, 200 or so filaments are gathered into a strand, which in turn is wound into rovings – cylindrical packages of about 20 kg, each containing around 32 strands. For production use, rovings are typically chopped to lengths of 4–38 mm depending on the manufacturing process. In GRC, unlike many other fibre-reinforced composites, the fibres remain predominantly in bundled form throughout the composite rather than dispersing as individual filaments. This strand structure has a profound influence on how ageing and degradation occur.

What AR Glass Actually Delivers: Mechanical Properties

Fresh, undamaged AR glass fibre has genuinely impressive mechanical properties. A single pristine filament can achieve tensile strengths of 3,000–3,500 MPa, though practical strand strengths are typically in the range of 1,300–1,900 MPa due to the inevitable minor surface damage and non-uniform load distribution across filaments. The modulus of elasticity is approximately 70–74 GPa, with an ultimate tensile strain of 2.0–2.5%.

At the composite level, independent experimental work has shown that adding AR glass fibres to structural-grade concrete produces measurable improvements across multiple performance indicators. Compressive strength increases continuously with fibre content, with gains of up to around 28% at 4.5% fibre content by weight of cement at 28 days. Flexural strength showed the most dramatic improvements – increases of up to 50% were recorded at a fibre content of 4.0%, with an associated increase in measured deflection confirming that the fibres genuinely improve ductility rather than just peak strength. Split tensile and bond strengths also improve significantly, though bond strength peaks around 3.0% fibre content before declining as fibre balling becomes problematic at higher additions.

The key point is that these are the properties of fresh, well-produced GRC with undamaged fibres. The performance challenge – and the reason AR glass selection and quality matters so much – is retaining as much of this performance as possible over a service life measured in decades.

How the Fibres Degrade: Two Mechanisms, One Problem

Even with ZrO₂ protection, AR glass fibres are not completely inert in a cementitious matrix over the long term. Two distinct but interacting mechanisms drive degradation, and understanding both is essential for anyone making decisions about GRC design, mix composition, or fibre specification.

1. Chemical Attack at the Fibre Surface

At the fibre surface, cement pore fluids drive a series of chemical reactions. Hydroxyl ions attack the siloxane bonds on the glass surface, causing dissolution of silicon and sodium from the outermost glass layers, which migrate outward into the paste. Simultaneously, calcium and water diffuse inward from the matrix into the glass. The result, visible through electron microscopy, is a progressive transformation of the original solid glass fibre into a sequence of concentric zones: an inner residual glass core, a hydroxylated leached layer, and an outer reaction product zone consisting of cement and glass components.

At ambient temperatures, this process is slow and the fibres retain substantial reinforcing capability. Flexural strength measurements of composites cured at 20°C for one year showed that reinforcement was still meaningful, with strengths remaining above 12 MPa. At elevated temperatures – used in accelerated ageing tests – the process is much faster. Composites cured at 55°C showed dramatically accelerated dissolution; by 360 days, many fibres were reduced to isolated corroded fragments no longer in good mechanical contact with the surrounding matrix. CaOH₂ crystals impinging on fibre surfaces also cause localised accelerated attack known as notching, which effectively shortens fibres by creating stress-concentration points.

2. Growth of Hydration Products Within the Strand

The second mechanism is arguably more significant during the first years of service, and it operates inside the strand rather than at its outer surface. As cement continues to hydrate, calcium hydroxide (CH) crystals and calcium silicate hydrate (C-S-H) products can migrate into the inter-filament spaces within a strand and crystallise there.

The consequences were clearly demonstrated in classic comparative durability research at the Technion Institute, which studied GRC composites made with different commercial AR glass fibre types – referred to as Types 1, 2 and 3. The composite prepared with Type 1 fibres (first-generation Cem-Fil) showed the fastest degradation: after just half a year of water storage at 20°C, SEM examination showed filaments broken at fracture surfaces, surrounded by massive CH crystals, with hydration products having grown extensively between the filaments. The composite retained only around 40% of its initial modulus of rupture (MOR), and almost all of its toughness (work of fracture, WOF) was gone – the material had become essentially brittle.

The composite made with Type 2 fibres (second-generation Cem-Fil 2) told a different story. After the same period of exposure, SEM images showed fracture surfaces where filaments had pulled out rather than broken, with inter-filament spaces remaining largely empty and free of hydration products. This composite was still ductile at two years, retaining approximately 40% of its original WOF

The implications of this research are important. First, what ultimately causes embrittlement is the infilling of inter-filament spaces with dense hydration products – this drives up the internal bond between filaments, eliminating the relative movement between them that produces energy-absorbing pull-out behaviour. Once filaments can no longer slide relative to each other, a crack propagating through the composite encounters the full bundled strand rather than a system of individually extractable filaments, and the result is sudden, brittle fracture. Second – and this is often missed – the research found that even in the embrittled composites, the glass fibre surfaces showed no signs of severe chemical etching at the one-year mark. In other words, at shorter service durations in temperate conditions, it is physical densification of the strand interior, not direct chemical corrosion of the glass, that drives embrittlement. Chemical attack becomes more important over longer timescales.

The Interface Under Load: What Pull-Out Research Tells Us

Understanding degradation is only part of the picture. How fibres behave when the composite is actually loaded – including the nature of the bond at the fibre-matrix interface – governs how effectively they transfer stress and absorb energy.

Detailed single-fibre pull-out research using AR glass fibres with different size coatings revealed several things relevant to real-world GRC performance. The local interfacial shear strength (IFSS) measured under quasi-static loading was very similar across differently treated fibres – around 29–35 MPa regardless of whether a strong or weak sizing was used. However, under high-rate (impact) loading, the IFSS increased considerably for all fibre types, with apparent activation energies of 98–105 kJ/mol, consistent with strong interfacial bonds driven by hydrolytic decomposition of Si–O bridges – bonds that are strong but not as strong as direct covalent bonds.

The type of sizing was found to control where the interfacial crack propagates. For unsized fibres and those with a strong styrene-butadiene sizing, fracture occurred through the cement matrix layer immediately adjacent to the fibre – the matrix surface failed before the interface itself. For fibres with a weak polypropylene sizing, the crack ran through the weaker interface. This distinction matters in practice because controlling crack localisation and post-debonding friction determines whether a fibre pulls out gradually (absorbing energy) or breaks suddenly. The ability to tune this behaviour through size coating design is one of the key levers available to AR glass fibre manufacturers seeking to improve GRC toughness.

Testing Alkali Resistance: The SIC Test

How do specifiers and producers verify that a given AR glass fibre is genuinely alkali-resistant? The standard method is the Strand-in-Cement (SIC) test, standardised as EN 15422.

In this test, a glass strand is embedded along the centreline of a cylinder of cementitious matrix with a 20 mm contact length. After conditioning in hot water at 80°C for 96 hours, the strand is pulled in uniaxial tension. A tensile strength above 330 MPa is the minimum threshold for the fibre to be considered adequately alkali-resistant. The hot-water conditioning is an accelerated ageing regime intended to simulate years of ambient-temperature exposure in a compressed timeframe.

The SIC test is a useful comparative tool – it can reliably distinguish between better and worse-performing fibre types and assess the effect of different size coatings. However, it has real limitations that are worth being clear about:

• Decades of field evidence suggest that real-world alkaline attack at ambient temperatures is less severe than the hot alkaline solution used in the test, meaning accelerated SIC tests tend to underestimate the long-term performance of fibres in actual service
• The presence of polymer in the matrix changes the behaviour of the composite during the hot-water conditioning in ways that make predictions of real service performance difficult
• The test applies load directly to the strand ends, whereas in real GRC the load is transferred from the matrix to the fibre via the interfacial bond – a fundamentally different loading mechanism
• As AR fibre performance has improved through successive generations, strands increasingly fail outside the embedded gauge length, making the test harder to use for the most modern fibres

One particularly important insight from the durability research is that toughness is a more sensitive indicator of degradation than strength. In the comparative fibre study, the MOR of chopped-strand GRC composites made with different AR fibre types was nearly identical after two years – but the work of fracture showed dramatic differences, with one fibre type retaining substantially more ductility. Relying on strength testing alone to assess GRC durability can be seriously misleading; toughness or ductility measurements are essential.

Fibre Length and Content: Getting the Balance Right

Strand length in GRC is a compromise between reinforcing efficiency and production practicality. A longer embedded strand can, in principle, carry more load before pulling out – but longer fibres are harder to disperse, reduce fresh mix workability, and make compaction more difficult, especially in complex mould profiles.

In practice, strand lengths are constrained by production method:

• Premix process: up to 25 mm
• Spray-up process: up to 37 mm
• Concentric spray-heads: up to 42 mm

Fibre content is similarly process-dependent. The spray-up process produces predominantly two-dimensional fibre distributions, allowing an optimum content of around 5% by weight of the composite. The premix process generates a more three-dimensional distribution, with an optimum closer to 3% by weight. In experimental concrete work using AR glass fibres mixed directly into structural concrete, optimum fibre contents for different strength properties varied: maximum compressive and split tensile strength gains were found at 4.5%, while maximum flexural strength improvements occurred at 4.0%, and bond strength peaked at around 3.0% before declining as fibre balling increased. Exceeding optimum fibre content in any process reduces workability and can actually degrade performance.

Reducing the Threat: Matrix Modification Strategies

The most straightforward protection for AR fibres is to select the best available fibre type. But the matrix environment itself can also be modified to reduce the severity of attack.

Pozzolanic additions such as microsilica, fly ash, ground granulated blast-furnace slag (GGBS), and metakaolin consume calcium hydroxide during hydration, lowering matrix alkalinity and reducing the supply of CH available to crystallise in inter-filament spaces. Silica fume in particular has been shown to reduce notching damage on fibre surfaces by inhibiting direct CaOH₂ nucleation at the glass surface. Research using seed crystal additions (nucleation seeds) in combination with silica fume showed that while heat curing accelerated cement hydration, it also enlarged matrix defects and eventually reduced composite strength – underscoring that matrix modification needs to be carefully optimised rather than simply maximised.

Polymer modification – replacing 10% or more of cement with acrylic polymer – reduces matrix alkalinity and moisture transmission, offering some protection to fibres. However, the increased cost and significantly reduced fire resistance have prevented wide adoption in mainstream GRC construction.

Very low-alkali cements such as calcium aluminate cements can in principle reduce the chemical aggression dramatically, but risks including conversion reactions in warm humid conditions and incompatibility with standard admixtures have kept commercial use limited.

From Science to Specification: What This Means in Practice

The science described above translates into a set of practical decisions that anyone specifying or producing GRC needs to make consciously.

Only AR glass with a minimum ZrO₂ content of 16% should ever be used in an OPC-based matrix. This is non-negotiable – using E-glass or any non-AR formulation will result in rapid and progressive fibre degradation.

Second-generation AR fibres perform significantly better than first-generation ones over the long term. The Cem-Fil 2 generation – and its modern successors – demonstrated in real comparative testing that their surface treatments substantially reduce the rate of hydration product infilling between filaments, preserving ductility for far longer. This was not a marginal improvement. In the Technion research, the difference between fibre types was the difference between a composite that had essentially become unreinforced brittle paste at one year, and one that was still meaningfully ductile at two years

Toughness matters more than strength when evaluating long-term performance. A GRC panel can lose most of its energy-absorption capacity – and therefore its resistance to impact, seismic load, and sudden overload – while retaining a strength figure that looks reassuring on paper. Specifying and testing for work of fracture or ductility alongside MOR gives a far more honest picture of how a GRC composite will perform across its service life.

The sizing is not an afterthought. The chemistry of the size coating controls both initial bonding behaviour and how the interfacial crack propagates under load. Sizings that promote gradual fibre pull-out under quasi-static and impact loading give better energy dissipation than those that cause sudden, brittle fracture.

A Composite That Earns Its Place

GRC is, without question, one of the more technically demanding materials in the construction industry’s toolkit. Its internal structure – bundles of glass filaments dispersed throughout a cementitious matrix, each bundle in a continuous state of chemical and physical evolution – is more complex than almost anything else built at scale. The fact that it can be reliably produced by relatively small specialist manufacturers, with decades of documented performance in some of the world’s most demanding architectural environments, speaks to how well the fundamentals are understood when they’re properly applied.

The alkali resistance of the glass fibre sits at the heart of that reliability. Get it right – the right ZrO₂ content, the right generation of fibre, the right size coating, the right matrix design to moderate the chemical environment – and you have a composite with a 50-year track record of performance. Miss any part of that chain, and you risk the kind of rapid embrittlement that caused serious problems in GRC’s early years and temporarily set back an entire industry.

The lesson from five decades of research is straightforward: AR glass doesn’t just reinforce GRC. It is the condition on which GRC’s durability depends.

This article draws on research from multiple independent sources including: Bartos, P.J.M., Glassfibre Reinforced Concrete (Whittles Publishing, 2017); Bentur, Ben-Bassat and Schneider, Journal of the American Ceramic Society, 68(4), 1985; Yilmaz, Lachowski and Glasser, Journal of the American Ceramic Society, 74(12), 1991; Scheffler, Zhandarov and Mäder, Cement and Concrete Composites, 2017; Wu et al., Journal of Wuhan University of Technology, 28(4), 2013; and Ghugal and Deshmukh, Journal of Reinforced Plastics and Composites, 25(6), 2006.