The properties of all frc are controlled to some degree by the interface between the fibres and the matrix, and the microstructural characteristics of this region have a significant impact on durability. Many investigators have studied this region using scanning electron and optical microscopy and their findings are summarised in the major reference sources (Majumdar and Laws, 1991, pp. 143±163, Bentur and Mindess, 1990, pp. 317±319, 412±413). This section will focus on key points relating to mechanisms of ageing. It is convenient to discuss frc with monofilament reinforcement in the form of single fibres (i.e., steel-frc) separately from frc with multifilament reinforcement in the form of strands or tows (i.e., glass, carbon, polypropylene and natural-frc) since there are a number of microstructural issues that only affect the latter category.
Multifilament frc
Figures 9.6 and 9.7 show the unaged interfacial microstructure for typical OPC matrix frc materials with multi-filament reinforcement, glass and coir respec- tively. The interface between the reinforcement unit and matrix is porous and spaces within the unit (interfilamental spaces in glass, and cell lumen in coir- frc) remain largely free from hydration product. Figures 9.8 and 9.9 show the microstructure of the same frcs after ageing such that significant degradation of mechanical properties has taken place. In both materials, a significant proportion of the interfacial and interfilamental/lumen space has been filled with hydration product, identified as Ca(OH)2 in the glass-frc and assumed to be the same in the coir-frc. This phenomenon is variously known as mineralisation, petrification and `bundle filling’. It is generally assumed to be associated with an increase in bond between the fibres and the matrix. In cases where the unit reinforcing element is a strand, e.g. glass- and carbon-frc, an increase in fibre- fibre bond is assumed as well. This can change the mode at failure from fibre pull-out to fibre fracture (i.e., loss of region IV behaviour, Fig. 9.2) if the critical value of bond is surpassed; matrix densification effects will also begin to occur (see Section 9.2.1). The transition from pullout to fracture has been observed in various types of glass-frc by Bartos and Zhu (1996), who also first identified an intermediate `telescopic’ pull-out mode (see Fig. 9.11 on page 345), and natural/cellulose-frc (Mohr et al., 2005, Bentur and Akers, 1989b). Enhanced fibre-fibre bond with ageing in glass-frc has been directly measured by Zhu and Bartos (1997) using novel micro-indentation apparatus. Purnell et al. (1999) showed that bond in glass-frc increased during the first few weeks of ageing and then stabilised about 2±3 times its initial value. Kim et al. (1993) suggest that the fracture toughness of the glass-frc interface increases by three orders of magnitude during curing.
Examination of fibre surfaces exposed in aged or weathered frc rarely shows any significant fibre corrosion, i.e. loss of section or gross pitting, regardless of the fibre type or the extent to which properties have been degraded (e.g., Purnell et al., 2000, ToleÃdo Filho et al., 2000, Katz and Bentur, 1995, Ball, 2003). In the short term at least, in frc where the fibres are known to be immune to alkali attack, i.e. carbon-frc, loss of strength and toughness may still be observed with ageing (Katz and Bentur, 1995, Katz 1996). For some time most investigators, therefore, have attributed loss of properties in all frc to matrix densification, bond enhancement and bundle filling/mineralisation effects (e.g. Bentur, 1985, Cohen and Constantiner, 1985, Diamond, 1985). For many frc composites this is true, but fibre corrosion should not be thus discounted. First, bundle filling and/ or bond increase is not always associated with degradation. In glass-frc compo- sites made with sulpho-aluminate based matrices, the interfilamental space is completely filled with hydration products even in unaged samples with no detriment to strength (Fig. 9.10); in OPC/2nd generation AR glass-frc, degrada- tion can occur without complete bundle filling being observed (see below) and bond can reach a maximum without detriment to mechanical properties (Purnell et al., 1999). Secondly, a transition in failure mode from pull-out to fracture does not necessarily require an increase in bond; a decrease in fibre strength will also trigger such a transition. In natural fibre composites, pull-out may be observed as the dominant failure mode even after mechanical properties have been significantly degraded (Mohr et al., 2005) and the mode may change from pull-out to fracture without mineralisation of the fibre (Bentur and Akers, 1989a). Thirdly, for glass and carbon, a brief examination of the fracture mechanics of fibres suggests that their strength is governed by the size and population distribution of surface flaws. The surface manifestation of critical flaws is likely to be so small (~10 nm) as to be very difficult to detect on in-situ fibres using an SEM; nucleation and growth of such flaws could thus cause strength loss without readily detectable fibre surface damage (Purnell and Beddows, 2005). Recent work using Weibull analysis and atomic force micro- scopy on AR-glass fibres subjected to various ageing treatments has established that the maximum surface defect size (as opposed to manifestation) is ~50 nm and confirmed that flaw size and population density of surface flaws control the strength of the fibres (Gao et al., 2003).
The microstructure of frc modified for enhanced durability is varied. The size applied to second generation AR-glass fibres reduces the precipitation of portlandite at the fibre-matrix interface and within the fibre bundle. The ubiquitous monolithic portlandite deposits completely surrounding fibres in first generation OPC matrix glass-frc after ageing (e.g,. Majumdar and Laws, 1991, p. 149) are not seen in modern glass-frc (e.g., Purnell et al., 2000). Reports on frc modified with csf are variable. Katz and Bentur (1995) did not observe a significant difference between the interfacial microstructure of OPC and OPC- csf matrix carbon-frc using SEM but mercury intrusion porosimetry (MIP) indicated that the csf induced a significant reduction in porosity after accelerated ageing. Bartos and Zhu (1996) reported that csf modification (10% cement replacement) to the matrix of glass-frc did not significantly change the develop- ment of interfacial and interfilamental microstrength with ageing (measured using a micro-indentation technique to `push out’ individual filaments). How- ever, matrix modification by metakaolin (25%) was effective in preventing development of microstrength and this was correlated with reduced degradation compared with OPC matrix glass-frc (Zhu and Bartos, 1997). Purnell et al. (2000) used thin section petrography to observe a change in the nature of inter- filamental deposits in metakaolin modified glass-frc; portlandite was not deposited between the filaments but an amorphous reaction product of meta- kaolin was. This was again correlated with reduced degradation during ageing. The microstructure of glass-frc made with sulpho-aluminate modified cements is quite different. In contrast to normal glass-frc, the matrix completely penetrates the fibre bundles and surrounds all the filaments even in the unaged condition, without detriment to mechanical properties, and the microstructure does not change with ageing; fibre pullout is still observed even after ageing (see Fig. 9.11). However, the matrix contains no portlandite. Thus it would appear that bundle filling is only detrimental if the precipitated material contains portland- ite. SEM examination of the interface in polymer-modified glass-frc shows a film of polymer partially covering the fibre surface at young ages (Bentur and Mindess, 1990, p. 262) but this film is not apparent in micrographs of naturally weathered samples (Ball, 2003) and the microstrength of the interfacial and interfilamental bond is not affected by polymer modification of the matrix (Zhu and Bartos, 1997). Given the fragility of the film coating the fibres, it is likely that polymer modification confers durability by impeding water and thus potential precipitate migration within the matrix in general with some temporary and/or minor enhancement at the interface. The microstructure of frc modified with pfa and ggbs has not received significant recent study and Majumdar and Laws (1991, p. 154±155) noted that under the SEM, the microstructure of glass- frc so modified was not markedly different from normal glass-frc.
Carbonation, either as a treatment or an ageing process, also induces micro- structural changes. In natural fibre reinforced concretes, carbonation changes the interface from being open and porous to very dense and homogenous (MacVicar et al., 1999) and also encourages the mineralisation of the porous fibre (Bentur and Akers, 1989b), correlating with increases in composite strength and decreases in toughness. Treatment of glass-frc with supercritical carbon dioxide also homogenised the matrix but did not promote interfilamental precipitation analogous to mineralisation, but in fact impeded subsequent precipitation during ageing. It did however fill void space at the interface with crypto-crystalline calcium carbonate and reduce the porosity of the matrix as measured using MIP (Purnell et al., 2003).
Monofilament frc The microstructure of monofilament frc ± i.e., steel-frc ± has not received the extensive attention afforded to multifilament frc, probably because it is rela- tively simple. The dominant feature of the steel-cement interface is a monolithic layer of calcium hydroxide around 10 -20um explicitly related observations of changes at the fibre-concrete interface to time- dependent behaviour, mainly because it is assumed to correlate with rusting of the steel fibres. Bentur (Bentur and Mindess, 1990) showed that there was a correlation between a reduction in fibre diameter and loss of both strength and toughness. A 30% reduction in fibre diameter correlated to a ~10% reduction in frc strength and a 50% reduction in frc toughness. The same degree of reduction was achieved in frc made with corroded fibres and accelerated aged frc samples, suggesting that the effect was almost solely due to fibre corrosion. Where significant toughness was lost, a change in failure mode from pull-out to fracture was observed. Some studies suggest that the steel fibre-cement bond decreases with time (Fu and Chung, 1997, Kim et al., 1993), in contrast to multifilament frc. In short, it seems that durability of steel-frc is controlled by simple fibre corrosion, which is only significant in extreme chloride-bearing environments and composites can be easily designed to account for this.
