Spatial distribution of pores in fly ash-based inorganic polymer gels visualised by Wood’s metal intrusion

https://doi.org/10.1016/j.micromeso.2009.05.016Get rights and content

Abstract

Inorganic polymer cements, or ‘geopolymers’, are now finding use as a replacement for Portland cement in concrete production, and have a complex pore structure which has proven difficult to measure accurately by gas or mercury porosimetry. These materials consist of an alkali aluminosilicate-based gel binder phase, within which are embedded unreacted precursor (usually coal fly ash and/or blast furnace slag) particles. Impregnation of the inorganic polymer samples with Wood’s metal, a low-melting-point alloy which solidifies at room temperature, and examination by scanning electron microscopy, allows both the size of pores and their physical distribution within the gel to be determined. Pore sizes as small as 10 nm are directly observable in high-resolution imaging. Much of the difficulty in applying standard porosimetry techniques to inorganic polymers may be identified as being related to the presence of numerous ‘ink-bottle’ pores, as well as the very wide distribution of pore diameters (spanning several orders of magnitude). The effect of gel chemistry on pore structure, and in particular the presence of calcium in the inorganic polymer formulation, is also considered.

Introduction

Inorganic polymer cement (IPC; including aluminosilicate ‘geopolymers’) is currently being developed as a less environmentally harmful material for construction applications, due mainly to the potential for Greenhouse emission reductions on the order of 80% compared to traditional Portland cement [1]. IPC is synthesised by reacting an aluminosilicate source – usually coal fly ash and/or blast furnace slag, produced as high-volume industrial wastes – with an alkali metal hydroxide or silicate solution. This initiates the partial dissolution of the solid particulate precursor and the subsequent formation of a nanostructured and microporous space-filling aluminosilicate gel, which has some significant similarities to the precursor gels used in many hydrothermal zeolite syntheses [2], [3], [4]. This phase is predominantly responsible for strength development in IPC, and can give mechanical performance comparable to the hydrated calcium silicate phases formed in Portland cement hydration.

The importance of the pore system in concrete cannot be overstated; the volume of pores and the distribution of their sizes control both the strength and durability of cement paste and concrete [5]. Attempts have been made using gas sorption porosimetry [6], [7], [8], and by combining gas sorption and mercury intrusion [9], to describe the pore structure of IPCs derived from both fly ash and metakaolin (calcined kaolinite clay). Vance et al. [10] also observed significant microporosity in metakaolin-derived IPC samples using positron annihilation lifetime spectroscopy. Despite these efforts, characterisation of the pore structure of IPC is still in its infancy and detailed understanding remains elusive. Thoroughly characterising the pore structure of IPC, and the factors which influence it, will greatly aid in advancing the unanswered question of the durability of IPC concretes.

Examination of the pore structure of Portland cement paste and concrete is more advanced than that of IPC, however, it remains an inexact science. In cement pastes, pores may be randomly shaped and their sizes distributed over 5–6 orders of magnitude, and this provides considerable experimental difficulty. A good example of these difficulties is provided by the challenges faced by the application of mercury intrusion porosimetry (MIP) to cements, as reviewed by León y León [11]. For many materials with a distribution of pore sizes, large-diameter pores are only accessible via narrow constrictions. This results in significant errors in MIP, as the volume of the large diameter region is attributed instead to the size of the constriction. This is known as the “ink-bottle effect” [11], and is significant in the analysis of complex materials such as hydrated Portland cement [12].

It is also possible to intrude a non-wetting low melting point liquid metal into the pore structure, then cool the sample under pressure so the metal solidifies and remains trapped in the pore system. The sample can then be examined microscopically, with the major advantage that the intruded metal is far more easily visible than the empty pore space of untreated samples. The technique has been used to examine Portland cement paste [13], [14], [15], [16] and concrete [17], [18], [19], natural rocks [20], [21], [22] and coal [23]. Clearly, metals or alloys with low melting points are required to minimise thermal damage to the sample. Wood’s metal, an alloy of Bi, Pb, Sn and Cd, has been used most frequently; gallium has also been used [15], [16]. The melting point of Wood’s metal varies from 65 to 80 °C depending on its composition [14], [21], while pure gallium melts at 29.8 °C [16].

The key benefit of Wood’s metal intrusion porosimetry (WMIP) is that it allows visual observation of the pore space within materials; this enables the physical distribution and connectivity of pores to be examined. For instance, Scrivener and Nemati [17] were able to observe the local porosity of the interfacial transition zone in Portland concrete using WMIP. Using image analysis with WMIP, Abell et al. [14] demonstrated the overestimation of small pores that occurs with conventional MIP due to the ‘ink-bottle’ effect: while MIP indicated the predominant pore diameter to be less than 0.1 μm, the pore diameters observed directly were in the region of 0.1–2 μm.

Previously [4], [24], the gel of IPC activated with alkali silicate solution was found via electron microscopy to consist of globular units of colloidal size, bonded where they intersect. The volume left unoccupied by the solid part of the gel comprises the pore space. Although difficult to observe directly, the pore space appeared to make up a significant part of the gel volume. In this paper, the pore space is examined and characterised by use of nitrogen sorption and Wood’s metal intrusion porosimetry. Both high-calcium and low-calcium IPC compositions are studied; it is well known that the reaction of alkaline solutions with high-calcium precursors such as blast furnace slag produces a gel that is predominantly calcium silicate hydrate in nature [25], in contrast to the aluminosilicate-based gel observed in low-calcium fly ash-derived IPC [4], [26], and this will have important implications for pore structure.

This study therefore has two key objectives: (a) to characterize the pore structure and hence the gel microstructure in a range of IPCs synthesized from waste materials, and (b) to determine the value of Wood’s metal intrusion in providing spatially resolved detail of the pore network of IPC.

Section snippets

Sample synthesis

The IPC samples studied here were synthesized using ASTM Class F fly ashes from Gladstone, Queensland, Australia (GFA; Pozzolanic Industries) and Port Augusta, South Australia (PAFA; Adelaide Brighton Cement), an ASTM Class C ash from Huntly, New Zealand (HFA; Golden Bay Cement), and ground granulated blast furnace slag (GGBS; Independent Cement and Lime). The oxide compositions of these materials are given in Table 1, X-ray diffractometry [32] showed the presence of some mullite, quartz and

Porosity in IPC

The effects of various compositional factors on the porosity of IPC pastes are shown in Table 2. Along with pore size distribution, porosity is critical in controlling the functional characteristics of porous media, such as convective and diffusive transport in cements [27]. It should be noted that the technique used here may underestimate porosity if the sample is not fully saturated at the beginning of the test. In addition, fully enclosed pores cannot be measured; however, this is not

Conclusion

The flow of ions into and out of cements and concretes controls most aspects of durability. Mass transfer is dependent on both the quantity and size of pores, and thus examination of the pore system is critical in assessing the durability of IPC. Unfortunately most conventional techniques for measuring pore sizes have significant shortcomings when applied to highly heterogeneous pore systems such as those found in cements, in particular with regard to the presence of ‘ink-bottle’ shaped pores,

Acknowledgements

This work was funded by the Australian Research Council (ARC), including partial funding via the Particulate Fluids Processing Centre, a Special Research Centre of the ARC.

References (33)

  • P. Duxson et al.

    Cement Concrete Res.

    (2007)
  • J.L. Provis et al.

    Chem. Eng. Sci.

    (2007)
  • R. Cioffi et al.

    Resour. Conserv. Recy.

    (2003)
  • P. Duxson et al.

    Colloid. Surf. A

    (2005)
  • C.A. León y León

    Adv. Colloid Interf. Sci

    (1998)
  • S. Diamond

    Cement Concrete Res.

    (2000)
  • K.L. Willis et al.

    Cement Concrete Res.

    (1998)
  • A.B. Abell et al.

    J. Colloid Interf. Sci.

    (1999)
  • K. Tanaka et al.

    Cement Concrete Res.

    (2002)
  • K. Kurumisawa et al.

    Cement Concrete Res.

    (2006)
  • K.L. Scrivener et al.

    Cement Concrete Res.

    (1996)
  • K.M. Nemati

    Int. J. Rock Mech. Min. Sci.

    (2000)
  • A. Hildenbrand et al.

    Mar. Petrol. Geol.

    (2003)
  • H. Yasuhara et al.

    Earth Planet. Sci. Lett.

    (2006)
  • S.D. Wang et al.

    Cement Concrete Res.

    (1995)
  • C. Gallé

    Cement Concrete Res.

    (2001)
  • Cited by (131)

    • Recent progress in understanding setting and hardening of alkali-activated slag (AAS) materials

      2022, Cement and Concrete Composites
      Citation Excerpt :

      For cement and concrete materials, the cutoff of pore size is somewhat arbitrary (lies in around 10 nm) between gel pores and capillary pores [104,105]. In generally, key pore parameters include porosity, pore size distribution and tortuosity of the pore network [106,107]. The diffusivity and permeability are largely (dependent on percolated pore size) associated with tortuosity of the pore network [108].

    View all citing articles on Scopus
    View full text