Degradation of epoxy coatings under gamma irradiation

Abstract

Epoxy networks based on Diglycidyl ether of bisphenol A (DGEBA) and cured with Jeffamine® (POPA) or polyamidoamine (PAA) were gamma irradiated at 25 °C in air. Dose rates of 50, 200 or 2000 Gy h⁻¹ for doses up 100 kGy were used. Structural changes were monitored by IR spectrophotometry, DSC and sol–gel analysis.

Both networks display some common features: for I≥200 Gy h⁻¹, reaction products grow proportionally to time and the rate is a decreasing function of dose rate. The simplest explanation is that peroxy radicals are the main precursors of these products (in the dose rate domain under study), through a unimolecular rearrangement of which an hypothetical mechanism is proposed. DGEBA–POPA are more reactive then DGEBA–PAA networks (according to IR criteria), that can be attributed to the high reactivity of tertiary CH bands in polyoxypropylene segments. The oxidation of these sites leads to methyl ketones. A simple kinetic model in which methyl ketones result from rearrangements of tertiary peroxyls and from tertiary alkoxyls was proposed. It leads to an expression of the radiochemical yield of methyl ketones (G(MK)) of the form

$$G(\text{MK})=a+b{I}^{-1/2}$$

where a and b are parameters depending of elementary rate constants. Experimental G(MK) values are reasonably well fitted by this equation. In DGEBA–PAA networks, a wide variety of oxidation products, among which amides predominate, can be observed. In these networks, chain scissions predominate over crosslinking, whereas a slight predominance of crosslinking was observed, at least for the lowest dose rate, in DGEBA–POPA.

Introduction

Epoxy networks of relatively low glass transition temperatures (typically 50 °C≤T_g≤80 °C) are often used as coatings in nuclear industry applications. Recently considerable attention has been paid to the improvement of certain properties such as: scratch and abrasion resistance, tensile strength, impact resistance or heat stability. One approach involved the incorporation of nanoscopic fillers (Chazeau et al., 2004). These fillers are expected to play a positive chemical role by trapping radicals or degradation by-products and hence decreasing the ageing rate. However, these fillers can also undergo degradation and generate undesirable by-products.

There is a wide range of chemical processes that can be induced by radiochemical or thermal oxidation. Chain scission and cross-linking are particularly important because, at moderate conversions, only the changes in crosslink density are capable of influencing the thermomechanical behavior. The quantitative determination of the number of chain scission and cross-linking events and their relationship with the elementary radical oxidation processes is thus of primary importance in understanding the overall mechanism.

Even taking into account the wide range of relevant structures, the recent literature on the radiochemical ageing of epoxy networks remains relatively scarce (Ngono-Ravache et al., 2001, Devanne et al., 2005, Ngono and Marechal, 2001, Vignoud et al., 2001, Longieras et al., 2006, Davenas et al., 2002, Crăciun et al., 2011). It seems reasonable to distinguish fully aromatic systems, essentially those used as composite matrices, from system-cured by aliphatic amines such as triethylene tetramine (Garcia et al., 2011, Benard et al., 2006, Benard et al., 2007), Jeffamine (Zaman et al., 2011), polyamidoamine (Astruc et al., 2009) or cycloaliphatic amines (Zhou et al., 2006). Among other possible applications, the latter can be used as coatings.

It is useful to summarize the published data as follows:

• Nature of the monomers

  The constitutive repeat unit (CRU) of an amine cross-linked epoxy polymer can be represented as follows:

  Here E is the diepoxide nucleus and A is the diamine nucleus. For example, in the case of PAA, the bisphenol A is the nucleus E and the aliphatic polyamide is the nucleus A.

  All these networks have in common the isopropanol segment (ip). This segment is, no doubt, a “weak point”, as well in purely radiochemical processes as in radical oxidation ones.

  The three carbons (a, b and c) are relatively susceptible to radical attack because they are linked to electronegative atoms (O or N) whose destabilizing effects are well known (Bellenger and Verdu, 1985). Oxidation of either carbon (a) or (b) is expected to lead to a wide range of products. Of these, the carbonyl-containing ones such as aldehydes, ketones, acids, peracids and others have their IR stretching bands in the 1710–1770 cm⁻¹ range. Carbon (c) can give rise to several products among which is a tertiary amide. Although this is also a carbonyl compound, its IR stretching band can be easily distinguished from the preceding ones since it is found around 1660 cm⁻¹. The presence of amides has been observed by practically all the authors who used IR spectrophotometry. They sometimes appear as the major oxidation products thus confirming that an important part of the oxidation events occur in the immediate vicinity of the tertiary amines which are in fact the nodes of networks.

  In the case of coating systems, the diamine moiety A contains also carbons which have a relatively high reactivity to oxidation, for example carbon (d) in PAA and carbon (e) in POPA.

  From a practical point of view, the important questions are: for a given epoxide such as DGEBA, what is the influence of the diamine structure on stability? Is it possible to predict this influence?

• Diffusion controlled oxidation

  Since oxidation is diffusion controlled, it is clear that the totality of available oxygen is consumed in a superficial layer. In thick samples, the deep layers undergo only anaerobic radiochemical ageing whereas the superficial layers undergo oxidation. However, anaerobic reactions can have a positive effect on cross-link density changes (as discussed below). According to Audouin et al. (Girois et al., 1997, Richaud et al., 2010), the thickness of the oxidized layer (TOL) can be estimated using a scaling law:

  $$\mathrm{TOL}\approx {\left(\frac{D}{K}\right)}^{1/2}$$

  Here D is the diffusion coefficient of oxygen in the polymer and K is the pseudo first order rate constant of oxygen consumption and is defined by

  $$K=\left(\frac{r_{ox}}{C}\right)$$

  where r_ox is the rate of oxygen consumption and C is the oxygen concentration in the polymer.

  The thickness of the oxidized layer is thus expected to depend on all the exposure parameters influencing D and K. In particular, these are temperature, dose rate and oxygen partial pressure.

• Changes in cross-link density

  At reasonably low conversions, the radiochemical or oxidation reactions are expected to influence the thermomechanical properties only if they induce changes in the cross-link density. Such changes can result only from chain scissions or from cross-linking events.

  If ν, s and x are the numbers of elastically active chains, chain scissions and cross-linking events per mass unit, respectively, then at low conversions for a fully cured sample, the following relationship can be written (Galant et al., 2010):

  $$\nu ={\nu }_0-3s+2x$$

  Indeed, since chain scission and cross-linking have opposite effects, then a knowledge of s and x or at least ν, is crucial for the understanding of the mechanical property changes. On the other hand, in view of kinetic modeling and lifetime prediction, it is also necessary to associate chain scission and cross-linking events to the elementary steps of the radiochemical ageing process.

• Radiochemical vs thermochemical initiation

  Let us consider the simplest case of oxidation in oxygen excess at constant initiation rate r_i. According to a very classical kinetic model, the whole oxidation rate r_ox will be given by

  $$r_{ox}=k_3\left[PH\right]{\left(\frac{r_i}{2k_6}\right)}^{1/2}$$

Here k₃ and k₆ are the rate constants of propagation (III) and termination (VI) reactions, respectively. These reactions are the following:

• Initiation: Polymer+O₂→radicals POO (r_i)

• Propagation: POO+PH→POOH+P (k₃)

• Termination: POO+POO→Inact. prod.+O₂ (k₆)

Initiation can be due to polymer radiolysis or hydroperoxide decomposition. This latter can be simply written

δPOOH→αP+βPOO

Here α=2 and β=0 for δ=1 (unimolecular decomposition) and α=1 and β=1 for δ=2 (bimolecular decomposition).

One sees that initiation by hydroperoxyde decomposition becomes favored when hydroperoxides can accumulate in the polymer. Schematically r_i∝[POOH]^δ.

In the absence of irradiation at moderate temperatures, initiation can result only from POOH decomposition. Oxidation is thus auto-accelerated and can display an induction period.

Propagation is relatively slow. Structure-reactivity relationships are well known and k₃∼10^−3±1 l mol⁻¹ s⁻¹ for most common hydrocarbon substrates at 30 °C (Korcek et al., 1972). Carbons having electronegative atoms in α position and tertiary carbons are among the most reactive ones.

Termination is also slow because it involves a bimolecular combination of rare species (POO° radicals). Such reactions are inevitably diffusion controlled in polymer matrices where molecular motions are relatively slow. In addition, for tertiary radicals, terminations are about one thousand times slower than for secondary or primary ones.

The combination of relatively fast propagation with a relatively slow termination can explain a high reactivity towards oxidation. Whereas radiochemical initiation is almost independent of temperature and its rate is proportional to dose rate, thermal initiation by POOH decomposition is dose rate independent but highly activated by a temperature increase. It is also sharply linked to POOH concentration. So that the domain of predominance of POOH decomposition in radical chain initiation, in a time-temperature space, is expected to have the shape of Fig. 1.

A kinetic model corresponding to this situation has been already published (Khelidj et al., 2006). Finally, the current research work in this domain is aimed to at clarifying the position of the system (material-exposure conditions) within a set of alternatives.

• Stability of isopropanol segment vs stability of diamine moiety.

• Importance of pure radiochemical reactions vs oxidation reactions.

• Crosslinking vs chain scission.

• Thermal initiation vs radiochemical initiation.

With this aim, a study involving two types of epoxy networks will be presented here namely DGEBA–PAA and DGEBA–POPA.