Misorientation Changes and Residual Stresses Redistribution after Welding. A Physical Simulation

Welding is widely used in the manufacture of steel products. Metallurgical processes during welding produce residual stresses in different zones of the welded parts. Welding residual stresses, particularly tensile stresses, can significantly impact the reliability and the integrity of the welded components. Our recent works report the effect of residual stress redistribution in a relatively short, up to two weeks, period after the welding procedure. The hypothesis that the observed stress relaxation phenomenon is associated to the microstructure evolution that occur in the material after the end of the welding procedure was proposed. To prove this hypothesis a physical simulation of the influence of the welding thermal cycles on the base metal with GLEEBLE equipment was performed. Using SEM EBSD, technique the variation of grain boundary misorientation accompanied by the residual stresses redistribution in the first days after welding thermal cycles simulation were observed.


Introduction
Welding is widely used in the manufacture of steel products. Metallurgical processes during welding, such as shrinkage, quenching, and phase transformations, produce residual stresses in different zones of the welded parts. These residual stresses, particularly tensile stresses, can significantly impact the reliability and the integrity of the welded components. Systematic studies have shown that residual stresses may result in failure mechanisms that are sensitive to localized stresses, such as the following: fatigue, brittle fracture, stress corrosion cracking, and creep cracking. Therefore, it is important to understand the distribution of the residual stresses on the surface of the welded components in and near the welding zone.
Our recent work reports the effect of residual stress redistribution in a relatively short, up to two weeks, period after the welding procedure 1,2,3 . The phenomenon was investigated by the daily monitoring of the welding residual stresses using X-ray diffraction and magnetic methods and the vertical displacements of the welded plate surface with a laser based technology equipment. The observed stress redistribution is characterized by a reduction and a uniformity of the values of the maximum shear stresses 2 weeks after welding. The microstructure analysis rules out the possibility of stress redistribution due to material failure.
The hypothesis that the observed stress relaxation phenomenon is associated to the microstructure evolution that occur in the material after the end of the welding procedure was proposed 2 . The dislocations originate from crystallisation and deformation processes. Both of these processes occurred during the welding procedure. Afterwards, the accumulation of dislocations can occur. Crystal orientations are changed due to the accumulation of dislocations and may exhibit fluctuations of several degrees, even within the same grain 4 . Thus, the accumulation of dislocations is expected to change the distributions of grain orientations and grain boundary misorientation angle. The formation of dislocations in solids requires energy and introduces residual stresses 5 . During plastic deformation, individual grains do not deform independently of each other and an excess of dislocations is observed at grain boundaries 6 . The accumulation of dislocations in large grains is bigger than in smaller grains, and stress concentration is higher in grains adjacent to larger grains. This means, for example, that in steels containing fine grain structure, greater stress has to be applied to cause slippage across the contour than in coarse grain steels 7 .
A direct relationship exists between the level of residual stresses presented in the material and grain boundary misorientation angle 8 . For low angle grain boundaries, the samples with higher stress levels have a higher percentage of misorientation. For high angle grain boundaries, the observed behavior is the opposite. This is corresponds to the fact that the energy of the low angle grain boundaries is the total energy of the dislocations within the contour area and depends on the dislocation spacing. For high angle grain boundaries, grain boundary energy is almost independent of misorientation 9 . The microstructure evolution is considered as a residual stress forming process in polycrystalline ceramics 10 . During formation of the microstructure, boundaries with higher mobilities and energies are likely to be eliminated, thereby changing the distribution of lattice orientations and grain misorientations. For multicrystalline silicon thin film, it was observed that stresses relax by generating recombination active extended defects such as dislocations and grain boundaries 11 . To prove this hypothesis a physical simulation of the influence of the thermo-mechanical processes on the base metal during the welding thermal cycles was performed with GLEEBLE equipment. The microstructure similar to intercritical coarse grained heat affected zone (IC CGHAZ) near to the fusion zone in a real multi pass welding joint was obtained. Using Electron Backscatter Difraction (EBSD) technique the variation of grain boundary misorientation in the first days after simulation were observed on IC CGHAZ area. These changes were accompanied by the residual stresses redistribution on the heat affected zone (HAZ) region. Residual stress measurements were performed with X-ray diffraction method.

Experimental Procedure and Techniques
The samples of dimension 10 x 10 x 71 mm were manufactured from ASTM DH36 steel. The material yield stress is 360 MPa. The chemical composition of the steel is given in Table 1.
A physical simulation of the influence of the welding thermal cycles on microstructure of base metal was performed with GLEEBLE 38000 machine. The specimen was maintained rigidly clamped in the machine jaws, as in Figure 1. The temperature was measured using a K-type thermocouples spot-welded in the middle of the specimen.
The thermal cycles simulation of double pass welding consists of following scheme. The middle region of the test specimen was heated to 1350°C with cooling rate of 450°C/s, and is held at this temperature for 0.35s, then cooled down to 800°C in 20s with cooling rate of 27,5°C/s and then to 200°C in 160s with cooling rate of 3,75°C/s for the first welding pass simulation. The second pass simulation involves heating from 200°C to peak temperature of 800°C with cooling rate of 266°C/s with holding at this temperature for 0.35s and next cooling to 200°C in 160s with cooling rate 3,75°C/s following by air free cooling, as showing in Figure 2.
This thermal scheme was chosen to obtain the microstructure similar to IC CGHAZ near to the fusion zone in the real multi pass welding joint, by simulating, two thermal passes. The   Microstructure of the central part of the specimen, as shown in Figure 3, is compatible with microstructure of IC CGHAZ region in real multi-pass welded samples, shown in Figure 4-a, followed by fine grain heat affected zone (FGHAZ) located at 5 mm from the center of the specimen, Figure 4-b. The FGHAZ was exposed at the temperature to 900°C. The microstructure of the unheated base metal is presented in Figure 5.   For data acquisition FEI Quanta 450 scanning electronic microscope (SEM) with tungsten filament and BRUKER E-FLASH EBSD automatic detector were used. BRUKER ESPRIT 2.0 package was applied for data processing. The standard polishing procedure was performed to prepare the specimens for microstructure analysis 13. The fist specimen was continuously maintained in the vacuum chamber of SEM in the same position for four days and the second one for seven days. The EBSD scanning of IC CGHAZ area was performed each day at the same time, without any equipment settings changes. In both cases, first scanning was made around two hours after the end of GLEEBLE simulation. For the first specimen, EBSD data were collected on 588.5x507.6 µm scan area, with 500x magnification, 0.74 µm step size, a 70° specimen tilt, using working distance of 15. The residual stress values were measured using a RAYSTRESS instrument, which is a portable X-ray equipment that uses the method of double exposure 1,2 . The principle of the stress measurements via the RAYSTRESS using double exposure is shown in Figure 6. Two cassette windows capture the diffraction lines in 2θ-angular intervals from 148° to 164°. The inclination of the specimen surface of 12° corresponds to measurements for steel specimens using Cr-K α radiation and the {211} reflection with θ 211 =78 o . The experimental accuracy of the stress measurements is 10 MPa. Specimen was maintained at the same measurement position throughout the all monitoring period of the three weeks. Stresses measurements were made in the center of the specimen on IC CGHAZ area, and on the area located 5 mm from the center of the specimen, whose microsturcture was identified as FGHAZ. The measurements on bouth the areas were performed simultaneously using two RAYSTRESS equipment. Measurements were started arround two hours after the end of the GLEEBLE simulation and were performed every 12 hours. Figure 7 present the results of residual stresses measurements performed on IC CGHAZ and FGHAZ areas of the specimen during the 21 monitoring days. Sign "-" denotes the compressive stress. The measurements were performed in the longitudinal direction. Although the stress state on the whole is defined as a tensor value, these measurements are sufficient for the aim of this paper to detect the change of the residual stress state on the analyzed areas. We observe the decreasing of the compressive residual stress on IC CGHAZ area accompanying by the increasing of the compressive residual stress on FGHAZ area. This process is not monotonic and has some oscillations. The initial difference of the values of the stresses on two analyzed areas is 20 MPa, and the final difference is 130 MPa. Stabilization in the process of stresses redistribution begins from the second week. Two weeks stabilization period is normally observed in the real welding samples 1,2,3 .

Misorientation changes
It is assumed that the residual strain manifests as local variation in lattice orientation 14 , and with increasing strain, low angle grain boundaries generate and their densities increase 15 . Hence local misorientaions provide an indication of the residual strain distribution in the crystalline materials and for its characterizing kernel based methods as Kernel Average Misorientation (KAM) is suitable 14 .
Analysis of the EBSD results based on KAM parameter is presented in the Figures 8, 9 and 10. The KAM results were obtained by calculating the difference in orientation from pixels three steps away in each direction (square) up to an orientation cut off of 5 degrees. A visible change of KAM was observed on the last monitoring day. This change consists on increasing of the fractions of angles grain boundaries that are less than 1.25 degrees and decreasing of the fractions of angles grain boundaries that are greater than 1.25 degrees. Thus, the redistribution of residual strain on the IC CGHAZ area of the specimen is detected.
Some authors seek for correlation between the level of residual stresses and the misorientation angles, relating that the specimen with highest absolute stress values has the highest fraction of misorientations present for low boundary angles (less that 15 degrees) and for high angle boundaries an opposite tendency is observed. The conclusion about  residual stresses level is based on the measurement of the stress in one direction only and the correlation results were observed for ASTM A106 Gr B steel 8 . It is worth noting that higher fractions of misorientation for high boundary angles is observed also in the annealed specimens 16 .
To observe the correlation between the variation of the residual stresses after welding and the changes in misorientation angles, for the second specimen, the seven day EBSD monitoring was performed. We note that on the first days of monitoring the compressive residual stresses measured in the longitudinal direction on IC CGHAZ area has value 195 MPa and seven day after this value becomes 112 MPa, i.e. absolute value of the residual stress decreases from first to seven day. Then, from first to seventh monitoring days we can have diminution of low boundary angles misorientation fractions and increasing of high boundary angles misorientation fractions 8 . We observe this result for over all phases misorientation angle distribution in the interval from 5 to 45 degrees, Figure 11, and also for misorentation of angle grain boundary distribution of ferrite phase, except the angels of degrees more that 55, Figure 12. Indeed, the influence of the ferrite phase, that is a greatest phase, is a principal on the over all phase misorientaion angle distribution result.  The retained austenite, Figure 13, and upper bainite, Figure 14, present well defined opposite tendencies of boundary angles misorientation fraction changes for the angles of 15, 20, 25, 40, 45, 55 degrees.
Comparing the 1-st and 7-th days, we can observe that the misorientation angle distribution for retained austenite and cementite, Figure 15, presents the same character of changes, except 35 degrees misorientation angle.   The summary of the misorientation grain boundary angle fraction changes from first to seventh days is given in Table  2. Here "+" denotes increasing and "-" denotes decreasing of the boundary angle misorientation fraction from 1-st to 7-th days. Signal "○" marks the absence of the fraction for comparison. Over all phases misorientation changes is compatible with the expected trend for the interval from 5 to 45 degrees. The expected trend means that for low angle grain boundaries, the samples with higher stress levels have a higher percentage of misorientation. For high angle grain boundaries, the observed behavior is the opposite 8 . Decreasing of misorientation for low angle grain boundary and its increasing for high angle grain boundaries correspond to decreasing of the absolute value of residual stresses measured at IC CGHAZ. The changes tendency for over all phases boundary angles misorientation fractions is following the ferrite one. Only for 2.5 and 50 degrees this tendency is slightly broken by increasing of the retained austenite misorientation fraction and decreasing of the upper bainite misorientation fraction, respectively. Over all phases changes is compatible with the expected trend for the interval from 5 to 45 degrees.

Conclusions
The results reported above demonstrate that the effect of residual stresses redistribution observed after welding is accompanied by a change in misorientation at least on IC CGHAZ area, simulated in this study. During the first two weeks after simulation, the compressive residual stress is decreasing on IC CGHAZ area and increasing on FGHAZ area, then the stabilization of this process occurs. At the same time we observe the changes on local misorientations and misorientation angle distribution for each phase and over all phases togehter. Over all phases misorientation changes is compatible with the expected trend for the interval from 5 to 45 degrees. We emphasize that the misorientation changes that occur after welding procedure has not been studied or reported previously in the literature.