CC BY
109
Interfacial diffusion and local mechanical properties in a bimetallic specimen
L. Bedel, V. Rougier1, L. Briottet, G. Delette, and M. Ignat2
CEA/G, DRT/LITEN/DTEN/S3ME, Grenoble, 38054, France 1 EDF/UTO/SIS, Noisy-le-Grand, 93192, France 2 CNRS/LTPCM, ENSEEG, Saint Martin d’Heres, 38402, France
In this paper, we present an analysis of the transmission of displacement and generated stress field across an interface of a bimaterial specimen, when subjected to an external tensile loading. The bimaterial is composed by stainless steel (316LN) and low alloyed ferritic steel (16MND5). Applying an image correlation technique on SEM in-situ tensile tests, it was possible to evaluate locally the displacement field. The displacement field indicated a strong interdiffusive zone as observed by SEM and microhardness measurements. Using iterative FEM solutions, the behaviour of the materials affected by diffusion was finally deduced.
1. Introduction
The high temperature isostatic pressure bonding process (HIP) allows us to bond together a wide domain of couples of different materials, including metal and ceramics. Putting materials in contact, then applying a high isostatic pressure at high temperatures will activate solid state interdiffusion processes at the level of their common interface. The aim of this work was to analyse materials at the vicinity of the bonded interface between ferritic steel (16MND5) and auste-nitic one (316LN) to obtain the local mechanical parameters which will control their behaviour. When bonding these two steels, the interdiffusion processes will produce microstructural changes likely to induce strong modifications on the local mechanical behaviours [4-6]. In order to derive the local mechanical parameters the strain fields on a bimetallic specimen have been measured during an in-situ tensile test through an image analysis formalism, the analysed zone being localized at the interface area. Finally, the local mechanical parameters are extracted, iterating the solutions between local image analysis and finite element calculations.
2. Experimental methods (bonding process and reaustenization cycle)
The main alloying elements in the bonded materials are reported in Table 1. The experimental parameters for the HIP bonding process of the materials under isostatic pressure of 100 MPa during 30 minutes at 1000 °C.
Following HIP bonding, to microstructurally restore the ferritic steel, a thermal treatment was performed [7]. The specimen was then resubjected to two heating cycles; the first one corresponding to a solutionning at at 900 °C during
30 minutes. Then, after quenching the specimen in oil, the specimen was tempered at 640 °C during 5 hours.
The obtained bimaterials were then sliced by spark erosion longitudinally in beam type flat specimens, presenting the boundary between steels in the middle. From these rectangular specimens, flat “dog bone” type tensile specimens were obtained. Total length of the tensile specimens did not exceed 35 mm, and their thickness varied from 1 to 1.5 mm. The bonded zone was localized in the centre of the specimens; their longitudinal axis (pulling direction) perpendicular to the material shared interface. The specimens were pulled with a micromechanical device, installed in a scanning electron microscope. This experimental procedure allowed us to follow step by step the microstrutural evolution on the specimen surface and in the bonded area. Detailed micrographs were extracted, corresponding to progressive steps of specimen deformation. The experimental setup including the specimen geometry and micromechanical device have been described elsewhere [1, 8].
The collected step by step micrographs were used to determine from image analysis the displacement fields and their evolution. Each image was centered on the bonded interface, then discretized in 2000 zones. Each zone represented a mean displacement for a given global strain. We may recall briefly that the analytical procedure to establish
Table 1
Chemical composition of the base materials
(wt. %) C Mn Cr Ni Mo N Si
316LN 0.022 1.73 17.5 12.16 2.4 0.068 0.41
16MND5 0.17 1.31 0.18 0.74 — — 0.015
© L. Bedel, V Rougier, L. Briottet, G. Delette, and M. Ignat, 2004
1000 500 0 500 1000
Fig. 1. Microstructure of the joint Distance from the interface, jim
the image correlation is based on the Newton-Raphson method. Detailed description of the above mentioned methods have been given in [1, 9].
3. Experimental results (microstructure, stress/strain curves and displacement fields)
Particular attention has been given to the microstructure which developed at the vicinity of the bonded interface. As a matter of fact predictive calculations of the diffusion path of carbon, from carbon rich ferritic steel towards low carbon austenitic steel, showed that diffusion affected zones could reach 0.3 mm on the austenitic side and around 0.5 mm on the ferritic steel side [10]. These results agree with microstructural observation (Fig. 1). On the austenitic side, an extended zone with intergranular precipitation of carbides is observed following preferential orientations (Fig. 2). By
Fig. 2. TEM micrograph showing strong intragranular carbide precipitation in the austenitic side (zone axis <211>)
Fig. 3. Microhardness profile across the interface showing the influence of carbon diffusion
contrast, on the opposite side (ferritic steel), there is depletion of carbon inducing a layer enriched in ferrite near the interface. Then the classical tempered martensite microstructure is observed far from the joint.
This heterogeneous redistribution of carbon exerts local permanent effect on mechanical response of the material. As a matter of fact, Vickers indentations performed with a 100g load and regularly spaced at 20 ^m were performed across the interface. The obtained hardness profile (Fig. 3) confirmed microstructural changes: the precipitation of carbides near the boundary (austenitic side) increases strongly hardness, while on the ferritic side close to the boundary hardness decreases. It is expected then that concerning other mechanical parameters as for instance the Young modulus and/or the yield strength, they will also be affected by the observed microstructural evolution.
The global response of the bimaterial specimen was analysed through tensile experiments performed also on the monomaterial specimens to obtain the mechanical behaviour of the materials unaffected by diffusion. The maximum load of the bimaterial specimen is similar to that of the 316LN one (Fig. 4). Due to heterogeneity as well as geometry of the bimetallic specimen, its overall strain cannot be compared to the monomaterials one. Close to the boundary between both materials the microstructural effects are strong, diminishing when diverging from the boundary. Consequently, we focused our attention on the evolution of displacement fields generated on each side of the boundary when progressively
/'\ ...
1 \
1 3-^^ \
$ — 16MND5 — 316LN — Bimaterial
0 10 20 30 40 50 60
Strain, %
Fig. 4. Tensile curves of the monomaterials and bimaterial specimens (1 and 2 marked the two steps where the mechanical behaviour optimisation has been performed, see below)
Mí®
“isüidB
i ï ; ; ; ; MWMt ; ï | j i 11\ ¡i J::;.pÆ
k^üíí*i,.•-• .<*-•"*■*'*■’■' i i : í ¡ í í -i'v'í'íí íí í j
V*. *. * . I 1 I .. - !*U i ni/-, t »-T^Ti Ir *** * * * *■" lililí* íS^.Víííjí»;
lf Í.<(< <-t-i * tJjjji' ; 1 î r I ï Ï *:
111 T ! "n ~7 T/f /: .',/, .
0.06-
E
E
c\T
=>
0.04-'
0.02
0.00
Black: ^ Grey: Full ci Dash Solid • I : step 1 step 2 rcles: experimental results line: initial FEM results line: optimised FEM results
0 16MND5 side 316LN side
-0.3 -0.2 -0.1 0.0 0.1 0.2
Distance from interface, mm
Fig. 5. Experimental displacement field (a), longitudinal displacements along the interface at steps 1 and 2 (b)
0.3
pulling the specimen. Establishing the displacement fields has been recalled previously: it was performed from successive image analysis [1].
4. Discussion
To derive local mechanical behaviour in the boundary zone affected by diffusion process, the results obtained from image analysis have been confronted to those obtained by finite element method. This has been performed by the following steps.
1. First, a mesh of tensile specimens is performed including several layers of 50 jum thick on each side of the joint to allow a quasi-continuous evolution of mechanical behaviour with respect to the interface distance. Since the tests are performed at room temperature and under monotonic loading, elasto-plastic behaviour based on the von Mises criterion associated to an exponential type of isotropic hardening is used for each layer.
2. For several given steps during pulling test, the displacement fields through the interface are determined.
3. The parameters of mechanical models in the diffusion affected zone are optimised by comparison of the displacements measured by image analysis and FEM solutions.
The experimental displacement field at step 2 (see Fig. 4) is displayed in Figure 5(a). Since displacements are measured on the free surface of the specimen, a 2D plane stress FEM simulation is performed using the Zebulon software. The displacements obtained using only the behaviour of the two base materials are compared to those obtained using the optimised multilayered behaviours and to the image analysis ones in Figure 5(b). The displacement fields are compared along a line in the middle of the specimen at two different loads of the tensile test history corresponding step 1 and 2 (see Fig. 4). As can be seen in this figure, to fit the experimental data, it has been necessary to modify the local mechanical behaviour of the layers on a width of 100 ^m in the 316LN side and on more than 500 ^m on the 16MND5 side. This is in good agreement with microhardness measurements as well as with metallographical observations. The very different mechanical behaviour observed between the
16MND5
316LN
16MND5
316LN
100 jam min-40 jam, max 20 [im
100 (im min-20 jam, max 0 jLim
Fig. 6. Comparison of the displacement fields at step 2 between FEM and image correlation analysis: U2 (a), U1 (b)
hardened layer in the 316LN side and the softened layer in the 16MND5 side leads to strong strain localisation inducing numerical convergence problems.
The in-plane displacements obtained at step 2 by the FEM simulation and image correlation analysis are compared in Figure 6. A good correlation is observed between these displacements.
5. Conclusion
The mismatch between two metallic materials bonded by diffusion bonding has been carefully analysed by means of metallurgical characterization, mechanical tests and FEM simulations. The diffusion induces on both side of the interface microstructural modifications. These changes are likely to strongly modify the local mechanical behaviour of the materials. The comparison between the local displacement fields measured by the in-situ image correlation method and FEM analysis allows us to identify local mechanical behaviour in the diffusion affected zone. The results obtained in this study show that this methodology based on local field measurements, simulation and metallurgical analysis is very useful in order to properly understand the mechanical behaviour of diffusion bonded junctions.
References
[1] L. Bedel, M. Ignat, Y. Dextre, and B. Riccetti, Determination of displacement fields near an interface obtained by diffusion bonding, Phys. Mesomech., 6, No. 5-6 (2003) 83.
[2] M. Schwartz, Modern Metal Joining Techniques, John Wiley & Sons, 1969.
[3] Y. Bienvenu, T. Massart, L. Van Wouw, M. Jeandin, and J. Morrison, The Metallurgy of Diffusion Bonding, Proceed. of the Int. Conf. on Diffusion Bonding, Cranfield, England, 7-8 July, 1987.
[4] R. Prader, B. Buchmayr, H. Cerjak, J. Peterselm, and I. Fleg, Microstructures and Mechanical Properties of Graded Composition Joints between Different Heat Resistant Steels, Conf. PGM94. Presses Polytechniques et Universitaires Romandes (1995) 479.
[5] T. Enjo, K. Ikeuchi, and S. Yoshisaki, J. High Temp. Soc. Jpn., 14, (2) (1988) 55.
[6] J. Agren and T. Helander, Metall. Mater. Trans. A, 28A (1997) 303.
[7] L. Bedel, H. Burlet, I. Chu, and B. Ricetti, Etude d’une jonction 16MND5/ INC0NEL690/316LNLN elaboree par compaction isostatique a chaud, Note technique CEA/DTA/DEM №98/02, 1998.
[8] M. Ignat, C. Josserond, and L. Debove, Bull. SFM. Elec., 14 (1995) 10.
[9] L. Bedel, I. Chu, Y. Dextre, and B. Ricetti, Optimisation of the spatial resolution of a digital image correlation. Application on Eurofer-316LNLN and CuCrZr-316LN junctions. Technical Note DTEN #2001:133.
[10] V. Rougier, Microstructure et endommagement d’une liaison bimetal-lique elaboree par soudage-diffusion. These E.N.S.P., Novembre 2000.
