Critical assessment of the evaluation of thermal desorption spectroscopy data for duplex stainless steels: A combined experimental and numerical approach

Abstract

The present study evaluates thermal desorption spectroscopy (TDS) data measured for UNS S32205 duplex stainless steel. Variations in the TDS spectra are obtained by electrochemical hydrogen charging for different times and by applying different heating rates for desorption to evaluate the desorption activation energy. Good agreement is found when comparing the experimental TDS curves with the desorption flux based on a numerical diffusion model using a homogeneous average hydrogen diffusion coefficient for the two-phase (ferrite-austenite) duplex microstructure. Trapping cannot be distinguished from the experimental TDS data since hydrogen diffusion in austenite is the rate-determining process during desorption. An average diffusion activation energy of 43.4 kJ/mol is determined from the experiments. Moreover, similar findings are obtained with a finite-element model that includes the heterogeneous hydrogen-related properties of the two phases of this duplex stainless steel.

Introduction

Thermal desorption spectroscopy (TDS) is a technique used to assess the distribution of hydrogen within a metallic crystal lattice and its discontinuities [1]. It has the ability to investigate hydrogen trapping at different microstructural features and defects such as grain boundaries, dislocations and carbides. Every type of trap is characterised by a specific activation energy. This energy needs to be supplied, in the case of TDS thermally, for hydrogen to be released from the trap [2,3]. Activation energies are typically determined by the Kissinger equation [4] by performing TDS measurements at different heating rates as initiated by Lee and co-workers [5,6]. Recent literature, however, focussed on the limitations of the Kissinger equation, especially in the case of diffusion-controlled processes, as the possible delay by diffusion is neglected [2,7].

Lattice diffusion in body centered cubic (BCC) iron occurs with an activation energy of approximately 8 kJ/mol [8]. Lattice defects such as dislocations, grain boundaries etc. have binding energies ranging from 20 kJ/mol to 60 kJ/mol [9], [10], [11], [12], [13]. Carefully designed carbides can reach higher binding energies [14,15]. These features can thus be regarded as deep traps as compared to lattice diffusion in BCC steels and the detrapping activation energy can be determined reliably through the Kissinger equation. Lattice diffusion in face centered cubic (FCC) iron, however, has an activation energy of 51–55 kJ/mol. The binding energy between hydrogen and microstructural defects is similar to BCC iron, leading to the inability to characterize or evaluate them with TDS [16,17]. This is also clear from experimental TDS data which showed that the application of cold work could be distinguished from TDS signals in the case of BCC steel while this was not the case for FCC steel since an identical spectrum was obtained after cold work [18]. Duplex stainless steels combine both phases, i.e. ferrite (BCC) and austenite (FCC), leading to difficulties in interpreting the TDS data with respect to hydrogen trapping. The derived approximate solutions of Fick's second law to simulate TDS data which are controlled by both diffusion and (different types of) trapping published by Kirchheim [7], could provide important insight in hydrogen-microstructure interactions for this steel type.

Few authors have reported TDS measurements on duplex stainless steels so far. Yagodzinskyy et al. [19] e.g. performed TDS on low alloyed 2101 duplex stainless steel and observed two peaks in their TDS spectrum. The authors stated that the first peak corresponded to lattice hydrogen and the second peak arose from hydrogen detrapped from dislocations. Dabah et al. [20] tested SAF 2507 DSS and found three peaks which they attributed to grain boundaries with an activation energy of 22.5–28.5 kJ/mol, dislocation cores with an activation energy of 34.8–40.3 kJ/mol and vacancies and/or austenite/ferrite interfaces with an energy of 50.2–57.4 kJ/mol. Silverstein et al. [21] performed TDS measurements on SAF 2205 DSS and observed four main peaks. The activation energies of the peaks were 24, 37, 44 and 62 kJ/mol, respectively. The authors mentioned various possible microstructural trapping features: elastic strain fields of dislocations (0–20 kJ/mol), screw dislocation cores or grain boundaries (20–30 kJ/mol), high angle grain boundaries, vacancies and austenite/ferrite interfaces (40–50 kJ/mol) and martensite (60 kJ/mol). The authors had extensive surface cracking due to the charging process which could have influenced the TDS spectra. It should be noted that all previous studies were performed on inhomogeneously electrochemically hydrogen charged specimens where hydrogen did not reach the centre. This clearly affects the resulting desorption profile since hydrogen effusion is controlled by a chemical potential difference throughout the material. Consequently, hydrogen can diffuse both towards the edge and the centre of the specimen.

In another study, Silverstein et al. [22] reported on TDS measurements performed on gaseous charged lean DSS, which contained a homogeneous hydrogen distribution after charging. The authors deconvoluted the signal into three peaks. While the first two peaks were attributed to hydrogen trapped at microstructural features, the third peak was ascribed to the sigma phase formed during gaseous hydrogen charging (67–72 kJ/mol). Park et al. [23] studied a DSS by gaseous hydrogen charging as well. The material was left in air for one day before testing. The resulting spectrum showed one broad peak. The authors concluded that hydrogen originated from austenite because of the similar appearance compared to TDS spectra of austenitic stainless steel. Further in-depth data analysis was however not provided.

Consequently, no consensus has been reached so far on the shape of TDS spectra performed on duplex stainless steels nor on the analysis of possible hydrogen detrapping activation energies. The present study therefore aims to perform TDS on electrochemically hydrogen charged duplex stainless steel where no damage was initiated and no phase transformations occurred [24]. The experimental study will be combined with numerical diffusion modelling based on the work of Kirchheim [7] to increase the understanding on the way hydrogen desorbed from the specimens during the TDS tests. The influence of treating a duplex stainless steel as a homogeneous material on the TDS spectra is therefore evaluated.