Ionic Conductive Polyesters—Assessing the Risk of Corrosion in Steel-Reinforced Concrete

Resistivity sensors in concrete used to assess the moisture content of the cement matrix or the corrosion potential of the reinforcement usually consist of metal rods or rings. For retrofit applications, these need to be embedded in a grout mortar to establish the electrical coupling between concrete and the metallic sensor surface. Installations following this procedure, apart from being labour and, with it, cost intensive, suffer from two common problems: (i) the grout mortar has a different pore system than the existing concrete and the water introduced by the mortar changes the moisture content in the volume surrounding the sensor. Both change the resistivity and it is unclear, when reliable reading can be obtained; (ii) at the steel-mortar interface, the conductor type changes from electron transport in steel to ion transport in concrete. However, this interface is far from perfect, which increases the overall resistivity and the susceptibility to failure.

One possibility to circumvent these problems is to use conducting polymers as sensor material. However, electrical conductivity is a property rarely associated with polymers. Common polymers show resistivities in the range of 10⁵ to 10¹² Ω·m [1] and are, thus, used to make e. g. wire insulation and sockets. However, electrically conductive polymer composites have been known since the 1950s [2]. Initially, fine metal powders were incorporated into thermoplastic polymers at high loadings and the electrical conductivity was the result of a percolating network of particles inside an insulating matrix [3]. Due to the high filler content, these materials had poor mechanical properties and were hard to process. Bulk electrical conductivity in polymer materials was first demonstrated by Heeger, Shirakawa, and MacDiarmid in 1977 using doped poly(acetylene)s, where the conductivity originates in the transport of electrons along the delocalised π-electron system [4]. These materials are usually obtained as brittle films and require very controlled conditions during their synthesis. While these two are electron conductors, Wright started to follow a rather different approach in 1975. By dissolving lithium salts in poly(ethylene oxide)s, he discovered polymer materials with ionic conductivity [5], in which the ions move through the polymer in a hopping process.

Using polymers as sensor material would improve the concrete-sensor interface, as polymer resins easily adapt to rough surfaces and the transition from electron to ion transport occurs at the steel-polymer interface, which is much smoother and more adherent. In order to substitute conventional retrofit sensors in concrete structures with conducting polymers, the application needs to be simple, curing should be fast and little affected by environmental conditions such as temperature and moisture, and the material must be resistant to alkaline hydrolysis. These requirements and constraints rule out metal-filled thermoplastics and intrinsically conductive polymers as well as common epoxy, polyurethane, and silicone resins. The choice, therefore, fell on ion-conducting unsaturated polyester (UP) resins containing poly(ethylene oxide) segments and dissolved lithium ions which, which can be formulated as 2K resins and harden under a wide range of environmental conditions. The following article presents some of our results on the development of ion-conducting polymer resins and their initial lab-scale test as corrosion sensors in steel-reinforced concrete.

The ion-conducting polymer resins to be used as sensor materials are based on unsaturated polyesters, which are prepared by first reacting poly(ethylene oxide), PEO, or poly(propylene oxide), PPO, with maleic anhydride (Fig. 1). Both PEO and PPO form the conductive blocks along the chains, while maleic anhydride serves as connection to the cross-linker. In order to avoid high-molecular weight polymers, which would complicate subsequent handling, the theoretical degree of polymerisation was adjusted to 6 by using a calculated excess of PEO or PPO. Both polyesters were individually doped with lithium perchlorate to achieve ion conductivity and then cross-linked using styrene and the redox initiating system methyl ethyl ketone peroxide (MEKP)/cobalt(II) 2-ethylhexanoate. Cross-linking is necessary to obtain mechanical stability. Details concerning the synthesis of the materials can be found in [6].

Ionic conductivity in polymers is based on the solubility of small metal cations in the matrix and the movement of these cations along and between the chains [7]. In the well-documented system poly(ethylene oxide)/Li⁺, the solubility of the lithium ions is accomplished by their coordination to the oxygen atoms of the polyether. As a result, the atomic ratio of oxygen atoms in the chain to mobile lithium ions (O/Li⁺) plays an important role. In previous reports for similar systems, a V-shaped course of the resistivity with the O/Li⁺-ratio was reported [8]. The reason for the initial decrease in resistivity with increasing Li⁺ concentration is that more mobile charges facilitate the current flow. Beyond a certain concentration, the ions hinder each other, which results in an increase in the resistivity. For the present poly(ethylene oxide)-containing unsaturated polyesters, the minimum was found at O/Li⁺  = 50 (Fig. 2, circles) [9], which is in accordance with literature [8].

However, for the structurally related resins based on poly(propylene oxide), the minimum was found at O/Li⁺  = 10 (Fig. 2, squares). The reason for this might be the additional methyl group of poly(propylene oxide), which imposes a larger helix diameter. As a result, more lithium ions are needed to overcrowd the larger helix and hamper ion movement. Two further observations can be drawn from Fig. 2: (i) as expected, the resistivity decreases with increasing temperature (cf. Figure 2, PEO 0 ℃ and PEO 23 ℃), which is a direct consequence of the increased mobility and (ii) the minimum in the poly(propylene oxide) system (1941 Ω m) is lower than that of the poly(ethylene oxide) system at 0 ℃, but approx. 5 times higher than that at 23 ℃ (401·Ω m). Due to its lower polarity, the poly(propylene oxide) matrix is a poorer solvent for lithium ions than poly(ethylene oxide), resulting in higher resistivities.

Since the systems contain polymers as conductive blocks, it was interesting to investigate the influence of the block length on the resistivity of the cross-linked materials (Fig. 3). For the poly(ethylene oxide)-based systems at O/Li⁺  = 50 (Fig. 3, circles), the resistivity increases only moderately from approx. 200 Ω m for PEO blocks with 9 repeat units to 740 m for 22 repeat units. Up to this chain length, the PEO starting materials are liquid. The subsequent drastic increase is due to crystallisation of the polymer, which hampers the movement of the ions [10]. In contrast, the resistivity of lithium-doped poly(propylene oxide)-based unsaturated polyesters was found to decrease with increasing molecular weight from 15.4 kΩ m for PPO blocks with 7 repeat units to 194.3 Ω m for blocks with 69 repeat units (Fig. 3, squares). On the molecular level, the movement of ions along the chain (intrachain hopping) is faster than the change from one chain to another (interchain hopping) [11]. An increase in the molecular weight of the conductive block should, therefore, allow ions to travel longer distances along the same chain, rather than having to switch from one chain to the next (interchain hopping) with a loss of time. However, this effect is normally counteracted by the increased viscosity of high molecular-weights polymers, which hinders ion movement. The reason why this is not observed in the present system remains unclear.

Despite certain positive aspects of the poly(propylene oxide) system, application studies were run using the poly(ethylene oxide)-based unsaturated polymers at O/Li⁺  = 50. As the mixtures cross-link by radical polymerisation, it is not possible to formulate a 1K system with considerable shelf-life. Rather, the two components of the redox initiator (peroxide and catalyst) were individually mixed with solutions of the unsaturated polyester in styrene. Fabrication of the sensor from there is quite simple: the two solutions are simultaneously injected into a prepared drill hole be means of a commercial 2K applicator through a 3D-printed mixing section (Fig. 4A). The latter remains in the hole and serves a support for a mixed metal-oxide coated titanium-mesh, which provides the electrical contact with the polymer. The sensor is ready to use as soon as a sufficient degree of cross-lining is obtained.

For the initial lab trials, a 75 cm × 75 cm × 15 cm concrete slab with two layers of reinforcement is prepared from CEM I suitable for exposure class XD3 according to DIN EN 206–1. In one of the corners, an area of 25 cm × 25 cm of the upper reinforcement layer is left out during casting. Here, a reference electrode FORCE ERE20 is mounted and the section is then filled with a concrete mixture containing 3 wt% NaCl. Two polymer sensors are installed at distances of 18 and 24 cm to the macro element (Fig. 4B).

Before assessing the performance of the polymer sensors it is important to monitor the cross-linking reaction under application-relevant conditions. It has already been discussed in several places that ion mobility has an enormous influence on the resistivity of the final product. As cross-linking changes the macroscopic appearance of the mixture from viscous liquid to rubbery, the resistivity of the mixture increases as the cross-linking reaction proceeds. As can be seen from Fig. 5, the cross-linking reaction can be accomplished over a wide temperature range and accelerates with increasing temperature. The mixture even works at 0 ℃, but takes about a day to be ready to use. At ambient temperatures, reliable readings can be obtained after approx. 8 h, while at 60 ℃, cross-linking is essentially completed within 90 min. The presence of water, which is usually a challenge for thermoset resins and an absolute no-go for conventional ion-conducting systems, hardly affects the cross-linking reaction. Water lowers the recorded resistivity most likely as a result of an increased ion mobility at the concrete-polymer interface.

One of the samples from Fig. 4B is stored at 21 ℃ and 60 r. h. for 8 months (Fig. 6A), while a second sample is first stored at 21 ℃ and 85% r. h. for 96 days, after which the humidity is raised to 95% r. h (Fig. 6B). Shown are the measured macro-element currents between the depassivated and passive reinforcement as well as the short-circuit currents between the total reinforcement and the respective sensor (distance 18 and 24 cm). A plot with two reference axes was chosen to allow a better qualitative comparison of the current curves. As can be seen in Fig. 6A, the actual macro-element current between active and passive reinforcement decreases continuously over time. This is presumably due to the comparatively dry storage and corresponds to the desired desiccation according to principle 8.3 of DIN EN 1504–9.

In more humid climates (Fig. 6B), the corrosion current remains high from the beginning and increases significantly after the humidity has been increased. In both cases, the polymer sensors allow a good qualitative estimation of the actual good qualitative estimation of the actual macro-element fluxes. The exact distance of the sensors to the corrosion site obviously plays a subordinate role for qualitative considerations. Signs of a loss of function of the sensors, e. g. due to loss of adhesion, etc., could not be detected in the period under the test conditions.

The following conclusions can be drawn from the above results: