Professor Hong S. Wong

Concrete is by far the most heavily used building material in the world. Current global production equates to over 3 tons for every person on Earth and demand will continue to rise. Because of the vast amounts used, the manufacture of cement currently accounts for ~8% of man-made global CO₂ emissions. However, concrete structures undergo ageing and gradually deteriorate. Degradation of concrete structures is increasingly becoming a worldwide problem with major economic, social and environmental consequences.

My research concerns the study of concrete at the microscopic level to understand the processes that govern its performance and deterioration mechanisms. This knowledge is important because it will help develop concretes with improved performance, lower environmental impact and greater resistance to degradation. It will also help develop more reliable methods to predict long-term performance. To achieve this, I employ a range of experimental and modelling techniques. The following summary presents some highlights of my research. 

The microstructure of concrete inevitably contains pores and microcracks that influence many of its properties such as strength and durability. However, the exact relationships are complex and not well understood. This is compounded by lack of accurate characterisation techniques, so my early work focussed on developing imaging techniques to quantify the microstructure. These have included optical and electron microscopy, X-ray microanalysis and laser scanning confocal microscopy. The methods were supported by developments in sample preparation to preserve the delicate microstructure [Wong et al., 2006], Monte Carlo simulations to help optimise imaging [Wong & Buenfeld, 2006] and image analysis to extract quantitative information. For example, we developed a new method (overflow) to segment pores [Wong et al., 2006b] that is more reliable, objective and overcomes the difficulty of defining the pore boundary in micrographs. We also developed a method based on Euclidean Distance Mapping to analyse microstructure gradients at interfaces that is more efficient and unconstrained by boundary conditions [Wong & Buenfeld, 2006b]. These techniques have enabled many of our subsequent work and provided opportunities to explore new ideas. For example, the techniques allowed us to make new observations regarding the characteristics of Hadley grains [Head et al., 2006], microstructural gradients at the ITZ [Wong & Buenfeld, 2006b; Head et al., 2008] and air voids [Wong et al., 2011], determine the relative contribution of pores and microcracks to transport, and study deterioration mechanisms [Wong et al., 2010]. These methods have also facilitated consultancy research activities such as examining the deterioration mechanism in various concrete structures and assessing the performance of cement grouts for radioactive waste containment [Wong et al., 2007]. Work is currently on-going to develop three-dimensional characterisation of pores and microcracks. 

All of the major degradation mechanisms affecting concrete are controlled by penetration of water and aggressive agents via pores and microcracks inherent in the microstructure. Understanding these processes is vital for developing durable concretes and reliable methods to predict performance. We carried out an EPSRC funded project [EP/F002955/1] in collaboration with Prof. Robert Zimmerman to address some of the technical challenges to achieve this. A focus of the study was to determine the contribution of pores, microcracks and other phases in the microstructure to transport. The aggregate-paste interfacial-transition zone (ITZ) is of particular interest because it statistically contains higher porosity and lower cement content compared to other regions. It is often thought that penetration of deleterious agents occurs mainly through the ITZ, but our study found that the net influence of the ITZ is small, even for concretes with a large fraction of overlapping ITZs [Wong et al., 2009 *MCR best paper award]. The influence of total porosity and microcracks, far outweigh any effects of the ITZ on transport. We also made new observations concerning how air voids can influence the microstructure and transport properties of concrete [Wong et al., 2011]. We found that entrained air voids can disrupt the packing cement grains, increases heterogeneity of the paste microstructure and produces gradients similar to those in the ITZ. Although the air voids appear isolated and do not form a continuous flow channel, they are in fact interconnected by capillary pores. Results show that air voids can increase or decrease transport, depending on the transport mechanism under consideration and the moisture content of the voids.

Experimental results are often influenced by many varying parameters that are difficult to isolate, masking important trends. Thus, numerical modelling was carried out to complement our experimental work and to support our understanding of underlying mechanisms. Images of microstructure can be used as starting point to predict transport properties [Wong et al., 2006c], and subsequently model deterioration. In Wong et al. [2012], we showed that the pore data extracted from micrographs can be used as input to a network model to predict permeability of cement-based materials with a wide range of mix proportion, age and microstructural characteristics. In collaboration with Prof. J.J. Zheng of ZJUT, a three-phase composite model was used to examine the relative influence of w/c ratio, curing, aggregate content, particle shape and gradation, and the aggregate-paste ITZ, on chloride diffusivity of concrete [Zheng et al., 2009]. In collaboration with Dr. Peter Grassl of Glasgow University, a nonlinear finite-element model was used to simulate shrinkage-induced microcracking and to investigate its effect on permeability of concrete [Grassl et al., 2010]. The study showed that the width of microcracks increases with increasing size and decreasing volume fraction of aggregates. These findings agreed well with experimental observations and highlighted the significance of microcracks to transport. The work has received follow-up funding from EU-FP7-ITN (264448) to support two Marie Curie ESRs to characterise microcracks and develop 3D meso-scale models to simulate diffusion [Dehghanpoor et al., 2013], capillary absorption [Dehghanpoor et al., 2014] and permeation in concrete.

Reinforcement corrosion is the most common deterioration affecting concrete structures. Corrosion initiation is primarily controlled by the ingress of Cl^-, CO₂ and H₂O. What is not well understood is the nature of the corrosion products and how they subsequently cause damage. In order to advance service-life modelling, we need a better understanding of how corrosion products form, disperse and accumulate at the steel-concrete interface, and the mechanical response of the surrounding concrete. In this study [Wong et al., 2010], we used a combination of imaging techniques to quantify the amount and distribution of corrosion products particularly at very early stages of corrosion, and more crucially, to follow its development to understand its damaging effects. The study showed initial corrosion produces soluble species that can migrate through the cement paste, and revealed several interesting aspects that have never been reported before. The corrosion products fill not only capillary pores and air voids, but also pores inherent in the hydrates, in the outer and inner products, in rims and relicts of reacted slag, and in pores created by dissolution of hydrates. When the paste region adjacent to corroding sites become filled or blocked, subsequent corrosion products are forced to accumulate at the steel-concrete interface, inducing expansive pressure that leads to bond failure and cracking. Once cracking initiates, corrosion products preferentially deposit in cracks rather than in pores in the cement paste.  This has important implications on modelling service-life because the ability of the microstructure to act as repositories to accommodate corrosion products, at least in the initial stages of corrosion, will extend the period between corrosion initiation, pressure development and cracking. In a subsequent paper [Zhao et al., 2011], finite-element analysis using the data from microscopy as input was carried out to compare the damage induced from non-uniform distribution of corrosion products to that of an idealised uniform corrosion. It showed that the form of corrosion has significant impact on the build-up of expansive pressure around the rebar and the duration of the propagation period. The work was carried out in collaboration with Prof. Yuxi Zhao and Prof. Wei-Liang Jin of Zhejiang University. The study will be extended by looking at a larger set of variables including concrete composition and degree of hydration, exposure environment, corrosion rate, and several forms of corrosion induced by cyclic salt spray, carbonation, admixed chloride and impressed current. Future work aims to exploit the findings to develop models to predict corrosion-induced cracking and spalling in concrete structures.

Superabsorbent polymers (SAP) are unique polymers that can absorb and retain vast amount of liquid, and swell to form a soft insoluble gel. The rate and amount of swelling is dependent on several factors including the composition and pH of the solution, chemistry and degree of cross-linking of the SAP. We believe that this property can be exploited to produce concrete that can self-seal cracks, thereby retaining its watertightness and durability [Lee at al., 2010]. When SAP is added during concrete batching, it swells only slightly because the mix water reaches very high pH and ionic concentration when in contact with cement. The initial swelling is also confined by the mixing and compaction processes. When the concrete hardens, the SAP gradually releases its absorbed water which promotes internal curing, then dries and collapses. The SAP then lies dormant until a crack releases the polymer. Subsequent ingress of fluids triggers the SAP to swell again, filling the crack and blocking further flow. Crack sealing is expected to be very effective because the SAP swells much more in the permeating fluid due to the reduced confining pressure, lower pH and ionic strength. It is also expected that SAP nearer to the external surface have larger swelling and better crack sealing ability, hence forming a manifold barrier system to fluid ingress. Preliminary results [Lee at al., 2010] show that SAP has a good repeated swelling capacity and that the crack sealing concept works for several fluids including synthetic groundwater. A subsequent paper [Lee at al., 2010b] demonstrates its effectiveness in sealing pastes and mortars with cracks up to 400 microns when exposed to a sodium chloride solution. This PhD project was supported by an EPSRC doctoral training grant. It has completed and publications are in preparation that will present findings from a wider range of experiments including concrete specimens, several SAP types, crack widths and ingress solutions. 

The water/cement (w/c) ratio is the most important parameter in designing concrete mixtures because it influences many properties of the hardened concrete. As such, the ability to measure the original w/c ratio is of great value for research, quality control and forensic analysis. However once concrete sets, it is very difficult to determine the exact amounts of water and cement that were added during batching. We have developed a new method to determine the original w/c of hardened concrete by measuring the amounts of capillary pores, hydration products and unreacted cement using backscattered electron microscopy. In contrast to existing methods, the new method is quantitative and does not require calibration to standards or comparison to reference samples. This is a major advantage because it allows concretes of unknown composition to be analysed. The method is also able to determine the original cement content, water content, and degree of hydration independently. Wong & Buenfeld [2009] reports the method development, assumptions/limitations and presents extensive experimental results showing its accuracy when tested on cement pastes with a range of w/c ratios and curing ages. A subsequent paper [Wong et al., 2013] presents its extension to mortars and concretes, where results from samples made with several cement types, w/c ratios, curing ages and aggregate content were presented. Laboratory and field-exposed samples were also tested. In a more recent paper [Yio et al., 2014], the method was extended to cement-based materials containing ground-granulated blastfurnace slag. Further research is ongoing to extend the method to samples containing other types of supplementary cementitious materials.