| Fe-Cr-Ni alloys are commonly used as structure material in nuclear power plants, for example 304 and 316 L stainless steel, Ni-based alloy 600 and 690, and so on. Longterm reliability of these metals is very important for the safety of nuclear power plant. However, intergranular damage was found in Fe-Cr-Ni alloy after long-term service in aggressive nuclear reactor environment, such as intergranular corrosion and stress corrosion cracking(IGC/IGSCC). Many research projects were carried out to strengthen intergranular-degradation-resistance of Fe-Cr-Ni alloys. Grain boundary engineering(GBE) is one of the most promising techniques to solve the problem, due to the formatin of high proportion of resistant GBs during GBE-processing. This technique has been successfully applied to many alloys. However, most of these researches of IGSCC and GBE were carried out using two-dimensional(2D) investigation, such as optical microscope(OM), scanning/transmission electron microscope(SEM/TEM) and electron backscatter diffraction(EBSD). Some information of 3D process and 3D microstructure lost in 2D observation, which induces uncertainty of research results. So, in the current work we investigated GBE and IGSCC in three-dimension using 3D-EBSD and 3D-OM. Thickness of the prepared 3Dmicrostructures is more than 250 μm, which is enough to analyze the real 3D GBnetwork of commercial Fe-Cr-Ni alloys.The main results were listed as below. 1) Grain geometry in 3D. The four geometric features of grain in 3D space are grain body, grain boundary(or grain face), triple-line(edge) and quadruple-point/node(vertex). So, a grain can be describe using these geometric characteristics: grain size, grain surface area, number of faces in grain, number of triple-lines in grain, number of quadruple-points in grain and face area. In polycrystalline materials, the distribution of all these characteristics is close to logarithmic-normal distribution: most grains have small value of these characters, but some grains have very high value of these characters; the non-uniformity is higher than that in 2D. The relationship of these characteristicswas was obtained by statistics: plot of mumber of grain face(number of neighboring grains) vs. grain diameter has exponential-function feature in fitting; grain surface area vs. grain diameter is power-function in fitting, similar with sphere surface area function; grain’s specific surface area is inversely corrlated with grain diameter, but direct correlation when grain diameter is larger than about 200 μm in GBE specimen; grain’s average face area is in direct proportion to grain diameter with line fitting; plots of number of grain triple-line and number of grain quadruple-points vs. grain diameter are power-function in fitting; number of sides of grain face is in direct proportion to grain face area. 2) 3D microstructure of GBE specimen. Results of investigations in 2D show that the twin-boundary fraction both in length and in number was increased significantly by GBE-processing. However, the 3D results of this work show that many large-area twinboundaries were formed during GBE, but no more number of twin-boundaries were formed. So, GBE-treatment increased the twin-boundary proportion significantly by area but not by number. In addition, from GB-network topology perspective, such as triple-junction, quadruple-junction, grain-topology and GB-topology, the twinboundary proportion in these structures increased with different extent. From graincluster perspective, some large grain-clusters, which have long twin-chain with complex tree-ring-shape topological structure, were formed during GBE; but the graincluster in conventional specimen is small, which has short twin-chain with simple treeshape topological structure. In conclusion, compared with conventional specimen, the microstructure of GBE-processed specimen has characteristics of large twins with complex morphology, large twin-boundaries with complex morphology, and large grain-clusters with complex topological structure of twin-chain. 3) Mechanism of GB-network evolution during GBE. Recrystallization during the annealing of GBE-processing is characterized by the ‘initiation’ and ‘growth’ of graincluster. The recrystallization front GBs moved into the deformed matrix and swept away the initial GBs, and twinning-events happen in the wake of GBs migrating. All twins formed by one multiple-twinning, a series of twinning-events from single nucleus, can be connected by a twin-chain. The assembly of all these twins is called grain-cluster. So all inner grains of grain-cluster have mutually ∑3n-misorientation, but its outer GBs are crystallography-randomly. Compared with conventional recrystallization, multistep multiple-twinning happened during GBE which result in long twin-chain and large grain-cluster, but few-step multiple-twinning happened in conventional recrystallization which result in short twin-chain and small grain-cluster. It is known that a grain could form many twins by twinning-events but only four twinningorientations. During the multistep multiple-twinning, twinning-events have propensity to access to some grain-orientations(maybe it is in low-order generation) of the orientation-twin-chain but not randomly form one of the four twinning-orientations. These orientations are preferential-orientations of twinning-events of one multipletwinning, and are dominant grain-orientations of the formed grain-cluster. So, multipletwinning has back-and-forth pattern which would result in ring-shape twin-chain. And a large number of twins with preferential-orientations are formed. Maybe these grains(twins) could encounter during growth, and a large twin and twin-boundary with complex morphology could form by merging. 4) Mechanism of IGSCC resistance strengthening by GBE. From local perspective, crack-resistant boundaries(special-GBs, ∑3 in most case) not only few cracked themselves but also constrained cracking of nearby crack-susceptible boundaries(random-GBs) by triple-junction, quadruple-junction and GB-topology. For example, IGSCC cannot propagate through a quadruple-junction with 3 twin-boundaries or more; a random-GB would never crack if it incorporating two 2T-TJs(triple-junction with 2 twin-boundaries). From the whole GB-network perspective, there are two model to analyze the special-GB effect on IGSCC. The first one is percolation theory. Percolation model studies the threshold of special-GB fraction to stop IGSCC propagation along GB-network. The twin-boundary fraction, even in GBE specimen, is much lower than the prediction threshold to form unconnected random-GB-network. So, GBE-treatment cannot avoid IGSCC completely, but large grain-cluster could stop IGSCC propagation in local. The second model is crack-bridging ligament. Random-GBs cracked in brittle model during IGSCC. But special-GBs show good ductility. They were uncracked zone and were left far behand the crack-front during IGSCC, maybe ductile crack finally. They bridge the cracked opposite surfaces like ligaments and hold back crack opening. So, the critical fracture stress and stress intensity factor of IGSCC would be increased by crack-bridging ligaments. |
PDF Full Text Request