The MAI SN is very proud to present results coming from PhD works funded by the MAI. This gives an overview of material aging fundamental issues on material aging addressed by the MAI SN

Main contributor : PhD Géraldine RAPP

EDF Supervisor : Jonathan TIREAU

Collaboration : Sandrine THERIAS, Institut de Chimie de Clermont-Ferrand and Laurent Chazeau, MATEIS – INSA de Lyon

The objectives:

Predicting the lifetime of electrical cable in nuclear power stations is a major issue for EDF. Many researches are engaged to find new materials which will meet the ongoing regulations, especially fire retardant properties and halogen free composition. Blends of linear polyethylene (PE) with branched polyethylene (PEcB) cross-linked with peroxides seems to be a good candidate to replace the polymers currently used as insulate in the electric cables.

The aim of this work is

  • to determine the impact of thermo-oxidative aging on this new material, by using a multi-scale approach.
  • to understand the effect of the physical state of the branched polyethylene on the global aging of the blend, by using different aging temperatures (below and above melting temperature of PEcB).
  • to investigate the representativeness of accelerated aging versus natural aging in use conditions.

Main results

First, it has been shown that the modifications at molecular scale induced by thermal aging are the same between PE and PEcB. The kinetics of oxidation, followed by IR spectroscopy, are identical for all the samples except the 5050 blend, which oxidizes more than the other samples at all aging temperatures. Then, the approach based on the Arrhenius’ law has been applied on the IR results to study the validity of accelerated aging (T > 80°C) compared to aging in use conditions (T = 60°C) at the molecular scale. The results at molecular scale have shown that accelerated aging above 80°C are not representative of aging at lower temperatures. Thus, in the case of PE/PEcB blends, performing accelerated aging tests by increasing the temperature above 80°C does not allow to predict the lifetime of the material by extrapolation at lower temperatures with the Arrhenius’ law. The physical state of the two polyethylenes that can vary according to the temperature of thermal aging has no influence on the kinetics and on the non-representativeness of accelerated aging compared to aging in use conditions.

Then, the effects of thermal oxidation have been studied at other scales: microstructural, macromolecular and macroscopic. It has been shown that PE and PEcB have different morphologies before aging, according to their different branching content. The evolution of the morphology of PE during thermal aging is mainly the result of annealing, even if chemi-crystallization can occur at longer aging times. This leads to an increase of the crystallinity and lamellae thickness of PE, whereas the morphology of PEcB changes little during thermo-oxidation. The morphology of the blends is intermediate between that of PE and PEcB. Then, the gel fraction and tensiles tests above Tm have shown that the more the sample contains PEcB, the more the network is degraded during thermal aging. This is opposite to our previous infrared spectroscopy measurements, that have shown that the 5050 blend is more oxidized than the other samples. Moreover, the study of the mechanical properties before and after thermal aging demonstrates that the 5050 blend is the best compromise in terms of modulus and rupture behavior. It is therefore difficult to correlate the evolution of the molecular structure during thermal aging with the evolution of the macromolecular architecture and mechanical properties. This study highlights the importance of the multi-scale analysis, in order to take into account the impact of aging at every measurement scale.

In order to better understand the influence of the physical state of a polymer blend on the kinetics of thermo-oxidation, the two polymers of the blend should have different molecular structure and crystallinity degree. Indeed, in the present study, the very close molecular structure of PE and PEcB made them impossible to discriminate by IR spectroscopy, so no difference in their thermo-oxidation mechanism could be highlighted. Then, the small difference of crystallinity degree between PE and PEcB made the question of the influence of the physical state tough to answer.

Then, the characterization of the macromolecular architecture of the network, in particular the crosslink density and the nature of the crosslink bridges, is still to be improved. The development and the use of other analytical techniques, such as Nuclear Magnetic Resonance in the solid state applied to elastomers, could be considered to that end. Dynamic mechanical analysis or dielectric spectroscopy could also be used to study the impact of thermal aging on the molecular mobility in the polymers and blends.

Main contributor : PhD Leifeng Zhang

EDF supervisor : P.Todeschini et C.Domain

Collaboration : Bertrand Radiguet (Uni; Rouen)


Grain boundary (GB) segregation issue is important in materials science, and the quantitative research in multicomponent system is of great demand. The thesis had the following main objectives:

  • To clarify the influence of the GB type on solute segregation behavior;
  • To distinguish the influence of different factors(thermal aging, irradiation…)on the intergranular P segregation;
  • To make a comparison with the proposed segregation models.

Main results:

The main work of the present PhD thesis is to experimentally investigate the solute segregation behavior (especially P) in a French RPV steel, with a complex composition, under different conditions. Based on a large dataset of GBs for each with the joint chemical composition and crystallographic information being acquired on the same sample at the same location, the effects of GB type or carbide-ferrite interfaces and aging conditions (thermal aging and ion irradiation) on solute segregation behaviors (including both interstitial and substitutional segregations) were discussed. For a better understanding of the relationship between GB structure and its segregation behavior, 5 parameters were acquired for the studied GBs with a correlative EBSD/TKD/APT methodology. Afterwards, the segregation levels, measured by APT experiments, were compared to the predicted values from analytical models.

The adopted EBSD/TKD/APT methodology was described in details. In the current research, both the traditional EBSD and TKD techniques were adopted. The traditional EBSD technique was used to characterize the microstructure and classify the GB species, while the TKD technique was used to locate the GB position during annular milling and also helps to identify the GB plane with the combination of reconstructed APT atom maps. The principles, advantages and limitations of each technique were stated as well. The process to get a good GB is quite time-consuming.

The influence of boundary types (GBs and interfaces) on P segregation behavior was studied in the reference weld metal. After welding, the steel was stress-relief heat treated at 595~610°C for 16.5 h, followed by a controlled cooling at 15°C/h down to 350°C and then air cooling to ambient temperature.

The experimental results were compared with the predicted values from thermodynamic & kinetic considerations. Both the Langmuir-McLean binary model and the Guttmann models (ternary and multicomponent models) were considered for theoretical calculations. C, with fast migration ability, can diffuse during the cooling process after stress relief heat treatment, and its segregation level is probably deviated from the equilibrium condition at around 600°C. According to the comparison between experimental results and theoretical predictions, the Fe-P-C ternary model seems to fit the experimental results better than others.

After thermal aging, there is no obvious variation of the matrix composition from APT quantification, revealing no potential evolution of carbides. According to the analyses, there is neither any obvious evolution of the microstructure nor any evolution of GB distribution. However, the solute segregation behavior changes greatly. There is an increased P segregation for low angle GBs (LAGB), general high angle GBs (HAGB). On the contrary, the interstitial segregation of C is decreased after thermal aging.

A comparison with thermodynamic prediction in Fe-P-C ternary system reveals that the P segregation level was measured to be (0.06±0.02) atoms/nm2, lower than the predicted value of (0.11±0.04) atoms/nm2. This is contrary to thermodynamic considerations, and this could be explained by its slow diffusion kinetics at 350°Cwith 80000 h. In comparison, after thermal aging at 400°C, the measured P segregation level ( (0.11±0.04) atoms/nm2), though lower but is close to the predicted one, (0.18±0.03) atoms/nm2. This suggests that P has probably reached its equilibrium segregation level. C has faster diffusion ability and its transportation at low temperature (even below 250°C) is also efficient. In comparison to the reference weld metal, the lower C segregation levels for both thermally-aged steels could be due to the decreased sites at GBs because of the previously accumulated P (and other solutes). A lower aging temperature leads to a slightly higher C segregation level.

Besides GBs, solute segregation behavior at dislocations was also studied, and it seems that there is no obvious evolution of P segregation at dislocations compared with the reference weld metal.

The second part presents the results after ion irradiation. The irradiation experiment was performed under Fe3+ ion irradiation with energy of 5 MeV at 400°C. The studied GBs or interfaces are from a distance of 450±150 nm beneath the irradiation sample surface, ensuring a received dose of about 0.1dpa. According to the characterization, there is neither any obvious evolution of the microstructure nor any GB distribution in the ion-irradiated steel. Here again, two types of carbides, cementite and M2.0-3.2C carbides, were detected in the steel, consistent with the results reported in the reference weld metal. Mo shares over 50 at.% in M2.0-3.2Ccarbides, while Fe concentration is higher than 50 at.% in cementite. On average, LAGBs or general HAGBs share the highest P segregation level among all the detected GB or interface types in the ion-irradiated steel.

Besides, P segregation at dislocation was also detected. There is a decreased bulk P content due its segregation at PD sinks (GBs, interfaces, dislocations and others).

In ion-irradiated steel, there is a large variation of segregated C content among all the detected HAGBs (or among all the LAGBs). The substitutional segregation level remains the same level with the reference weld steel.

According to literature review and our present research, we could say that all the five parameters are important to the solute segregation behavior at GBs, and that carbide-ferrite hetero phase interfaces can also provide effective sites for solute segregation. In this research, by analysing numerous GBs and carbide-ferrite interfaces, the dependence of interstitial and substitutional segregation on the five-parameter GB crystallography was discussed for the French RPV weld metal.

This work could promote our understanding of GB segregation behavior. Besides, the adopted EBSD/TKD/APT methodology allows to perform relative research in other polycrystalline materials. Using this methodology, it could be interesting to study different model alloys to get material parameters with regard to GB types, which possibly helps to develop further simulations and design new materials with decorated GBs. Considering our experimental results, some efforts could also be devoted to directly observe the 3D GB structure with segregated chemical species. For irradiated materials, the detailed segregation behavior especially that with the presence of multiple segregants under various irradiation conditions, remains to be further clarified.

Main contributor : Post doc Aurélie JACOB

EDF supervisor : Patrick TODESCHINI

Collaboration : EDF Gilles ADJANOR/ Christophe DOMAIN

Supervisor : Erwin POVODEN-KARADENIZ, MatCalc Engineering, GmbH, Austria


There have been debates on the stability of MNS (MnNiSi intermetallic phase) precipitation in particular in RPV steels. Through the different studies and groups around the world, it is not clear if the MNS are thermodynamically stable or stabilised by radiation induced segregation (RIS). The present work aims at first to clarify the stability of the MNS through a comprehensive literature survey and to present a proper thermodynamic model for the MNS based on the assessment results.

There are a huge number of publications reporting MNS It is often referred to late blooming phase due to the fact that these precipitates are formed much later than “conventional” Cu precipitates in RPV steels. The designation of late blooming phase has been introduced by Odette and is now often used to describe the MNS in RPV.

Due to the controversy on the stability control of MNS, there have been recently a lot of studies using first principles calculations in order to help understanding the stability of these phases. According to Ngayam-Happy et al., MNS clusters are not thermodynamically stable since MnNi clusters interactions are energetically unfavourable by first-principles calculations. In contrast, Hosseinzadeh et al., by performing a systematic investigation on G-phase compounds as well as for related phases (B2, Ni2MnSi), found that the MNS G-phase would be thermodynamically stable. Moreover, they suggested favourable thermodynamics for precipitation. Accepting the results from Hosseinzadeh et al., the irradiation criterion for MNS G-phase stabilisation may not hold. In order to judge about the recent results , they are tested as input data for the thermodynamic model parametrisation of MNS G-phase in the present investigation.

Main results:

The G phase is a silicide phase with an overall formula Ni16M6Si7 (M=Mn,Nb,Ti). It is formed in high Ni containing steels, i.e ~10 wt. %, such as duplex stainless steels and reactor low pressure vessels steels. Among the different studies there is a lot of controversy on the chemical composition of the phase as well as the elemental distribution within the phase.

Previously the G-phase was modeled as (Ni)16(Cr,Fe,Mn,Nb,Ti)6(Si)7. This model does not allow any substitution of Ni nor Si. With such model the amount of Ni is fixed 55 % and Si to 24 % which is not consistent with experimental data. Thus, we develop a new thermodynamic model in order to be consistent with the experimentally determined site occupancies in the phase.

The present thermodynamic model proposed is (Cr,Fe,Ni)16(Cr,Fe,Mn,Mo)6(Si)7. Major constituents in respective sublattice are denoted in bold. With this element partitioning among the sublattices, we assume that Si is not substituted by transition metal elements, in accordance to the experimental crystallographic data and the nature of elements i.e. affinity of elements, similar chemical behavior of elements. With the present thermodynamic model Cr and Fe can substitute Ni on sublattice 1 and Mn on sublattice 2. We believe that due to the similarity of Fe and Cr the neglection of one or the other in respective replacements is not justified. Note that Cr is relevant in particular for G-phase stability questions in duplex stainless steel where G-phase formation is initiated by spinodal decomposition of the matrix into Cr-rich products.

The thermodynamic modeling of the G-phase has been elaborated based on high temperature phase stability and extrapolated to low temperature. In view of modeling materials for nuclear industry the working temperature are below 500°C, where it is difficult to obtain data at thermodynamic equilibrium in a reasonable experimental time (i.e several decades might be required). For the extrapolation at low temperature, we considered the G-phase to be stable in Mn-Ni-Si (in agreement with DFT and phase equilibria at intermediate temperature, i.e. T=800°C). Nevertheless, we do not expect the G-phase to be stable Fe-Mn-Si, Ni-Mo-Si, Cr-Mn-Si. So all the sub-systems were calculated at low temperature (i.e. below 500°C) to check the potential appearance of the G-phase in these systems. The G-phase was not stable in the different sub-systems except for Mn-Ni-Si.

DFT calculations reveal that all analysed compound energies of G-phase are negative (which can be seen analogue to G-phase stabilisation trend). This strongly suggests that dissolution of Fe and Cr in the crystal structure of NMS G-phase is relevant. Ni:Mn:Si and Ni:Mo:Si ternary “sub-phases” seem most preferred.

Using the newly developed thermodynamic modeling of G-phase, kinetic precipitation simulation of G-phase in austeno-ferritic was performed. The agreement between the experiments and the prediction is good.

Main contributor : Post Doc Binyan HE

EDF supervisor : Sebastien SAILLET


Supervisor : R.QIN, Open University and M.PEREZ, INSA – Lyon


This work aims to use electric current to regenerate the microstructure and properties of duplex stainless steel. It is well-known that electric-current enhances the mobility of atoms and dislocations and reduces kinetic barrier for microstructural transformation. The retardation of precipitation implies the electropulse-reduced thermodynamic driving force.

Aging is a typical microstructural transformation from non-equilibrium to equilibrium states. The cast duplex stainless steel (DSS) is implemented in many nuclear power plants across the world. This material undergoes solute decomposition and precipitates during aging. The mechanical properties, especially the charpy impact toughness, reduce drastically after aging. The precipitation in this steel has been studied intensively. The solute decomposition has been well characterized and modelled. Spinodal decomposition happens firstly in the aging processing, followed the formation and coarsening of G-phase and other precipitates. The aim of the present work was to regenerate the microstructure and mechanical properties of DSS which was caused by the early stage aging (with aging time up to 10,000 h at aging temperature up to 400 °C).

Main results

  1. A fast regeneration method has been developed to treat the aged duplex stainless steel charpy test specimens. According to the thermoelectric power measurement, one hour electropulsing treatment leads to the regeneration of>83% of aged microstructure in terms of TEP characterization. The charpy impact toughness has been recovered significantly. The Vickers micro-hardness was also recovered to that of near to the unaged materials.
  2. The regeneration has not changed the microstructure in grain scale. OM, SEM and EBSD characterization shows no change of the grain morphology, configuration of phases and crystallographic texture in the samples after electropulsing treatment. It was the spinodal decomposition and early stage G-phase to be reversed by the electropulsing treatment. TEM indicates the change of dislocations but no trance of precipitates.
  3. The regeneration is due to the electric-current-induced reverse of system free energy sequence. The electric current free energy causes a significant change of free energy difference between the decomposed and un-decomposed microstructures, which drives the segregated alloying elements to be homogenized in the ferrite phase. The mechanical properties are hence to be regenerated.
  4. The Ohm heat in the regeneration treatment is negligible according to the calculation. The electropulsing enhanced mobility enables the regeneration to be completed in very short time duration.

It would be interesting to examine the micro-chemical changes that would have been induced by treatment using atom probe technique in order to compare with the changes induced by high temperature-annealing for instance. Unfortunately, the application of this technique to real components would apply the elaboration of fairly big installation to produce the needed electromagnetic field which make it’s industrial potential rather limited.

Main contributor : Post Doc Kevin ARDON

EDF supervisor : Sophie DELAUNAY

Collaboration : Grégory LEFEVRE, Institut de Recherches Chimie Paris, Chimie Paris Tech

The objectives of this project are:

    • determine the characteristics of the deposit (porosity, thickness, chemical composition…) formed in representative SGs secondary circuit conditions to know the correlation between deposit characteristics and the physic-chemical conditions of deposit formation
    • determine correlation between deposits characteristics and thermal transfer
    • insert these data into a model to predict the impact of fouling on SGs performance

The first achievement realized is to have defined a sample preparation protocol for surface and transverse observations: cutting without lubricant, polishing with ethanol, nickel-plate and cold-coating.

For FORTRAND tubes, on surface, EDS analysis exhibited Fe and O presence and Raman spectroscopy identified magnetite and Fe2O3 compound in deposit.

SEM-characterization on surface permitted to compute a coverage rate on FORTRAND samples of about 82% in monophasic and 73% in biphasic with a mean particle size respectively of 0,47µm and 0,46µm.

SEM/FIB acquisition on sample from monophasic condition allowed to reconstruct deposit in 3D and compute porosity rate which is about 0,96%.

For IRIS tubes, characterization exhibit particle size was 0,40µm in monophasic and 0,15µm in biphasic conditions. Samples observed exhibited large area without any deposit which indicates a low coverage rate. SEM observations of magnetite powder presented presence of spherical particles with a diameter of 0,20µm which have been confirmed by granulometry.

Main results:

    • Implementation of appropriate methods to characterize deposits formed in SGs secondary side conditions
    • Comparison of characteristics of deposits formed in a same test at 275°C but with slight different thermohydraulic conditions
      • Presence of hematite in deposits formed in monophasic conditions
      • Both deposits with very low porosity
      • Thickness twice more important in diphasic conditions
    • Implementation and acquisition of thermal transfer measurement through SGs tube covered by different deposits
    • Insert these data in a predictive model to determine the impact of fouling on SGs performance