Effect of grain boundary structure on the intercrystalline damaging of austenic steel during brazing

During brazing of austenitic stainless steel with copper based brazing material a common failure occurs, namely that the brazing material solutes along grain boundaries, which looks like cracks. This unfortunate effect occurred when AISI 304 and 310 steels are brazing. To avoid this unwanted effect – since the cracks propagate mainly on high angle grain boundaries – our goal was to enhance the number of special coincident site lattice type grain boundaries with thermomechanical treatment. Experiments were performed for 1, 48 and 72 hour heat treatments on materials cold rolled at different levels. After the thermomechanical treatment significant decrease in the crack size was found in depth and width, respectively.


Introduction
Corrosion resistant and stainless steels are widely used in the industry.Austenitic stainless steels, one group of such steels are often applied in the automotive industry [1].Many cases it is necessary to join austenitic parts air-proofly, for which welding and brazing are good solutions.Brazing is a fast and efficient way for joining such parts for low mechanical loads, for example in the case of exhaust systems of cars.For such applications, AISI 304 stainless steel is widely used due to the combination of good mechanical properties and excellent corrosion resistance over a large temperature range.To braze austenitic stainless steels mostly copper based braze materials are used.If AISI 304 steel is brazed with Boehler SG-CuSi3 brazing material, long intercrystalline cracks occur in the steel, full with copper (Fig. 1).This effect is similar to the intergranual corrosion of austenitic steels.Intergranual corrosion can be prevented or improved by increasing the fraction of so-called special grain boundaries [2][3][4].
Grain boundaries can be classified from different point of views.One popular division is based on the properties of the grain boundaries.The random or general boundaries are characterized by average values for specific parameters (energy, etc.).On the other hand there are so called special boundaries having values for specific parameter(s) which are very different from the values of the random boundaries.These differences are caused by the special geometry of the boundary.
The grain boundaries which have a given fraction of atoms in the grain boundary plane which are coincident to both lattices separated by the grain boundary are characterized by the Coincident Site Lattice (CSL) model [5].These boundaries are classified in terms of values.The value denotes the fraction of atoms in coincidence, e.g. in a 3 boundary every third atom is at coincident site.There are often multiple misorientations that can achieve a given value.For example 33 can be achieved by a 20.05 degree rotation about [110], or 33.56 degrees about [311], or 58.99 about [110].These three descriptions may be combined in the distribution.The frequency of 3 and 3 n type CSL-boundaries is much higher than that of other CSL-boundaries and they play an important role in the tun-ing of material properties [5].By increasing the number of the CSL-boundaries better corrosion and fatigue properties can be obtained [2], [5][6][7].To increase the fraction of CSL boundaries several thermomechanical treatments are known [2][3][4][5], [8][9][10].The parameters for the MIG brazing were: alternatingcurrent; 56 A, voltage 17 V, the wire feed was 4.4 m/min.As shielding gas 30 % He -70 % Ar mixture with 10 l/min flow was applied.
First the as received material was investigated.To increase the ratio of special CSL grain boundaries thermomechanical treatments were made.The austenitic samples were cold rolled with different plastic deformation rate φ, and then annealed at 950 • C in a furnace for different times.The plastic deformation rates and annealing times are listed in Tab. 4.  To quantify the effect of thermomechanical treatment on the crack propagation, the typical sizes of the cracks were determined as described below.Light microscope images were taken from the whole cross section of the brazed and metallographically prepared samples.The depth (normal to the brazed sur- face) and width (parallel to the brazed surface) of every crack on the samples were measured with image analysis software.
Than the frequency of crack length values was calculated and displayed in a histogram with a dedicated software.After this, different curves were fitted to the diagram.The Gauss curve was found to fit the best for the values (see for example Fig. 2) so the crack length was determined by this function.
To see the grain boundary characteristics electron backscattered diffraction (EBSD) measurements were made on the AISI 304 samples with a Philips XL 30 scanning electron microscope equipped with a TSL TexSEM EBSD-detector system.The scanned area at the border of the brazed region was approx.300 × 290 µm 2 with 1,7 µm step size.The EBSD measurements were made on the cross section of the samples prepared by traditional method [11].

Results and discussion
Crack sizes determined after brazing of the as received material and the thermomechanical treated samples are listed in Tab. 5.
In case of the ASI 304 steel it can be clearly seen that the average crack length decreased significantly after the thermomechanical treatments.In every case the crack depth reduced at least one third of that of the untreated sample (Fig. 3).The best result was achieved after 48 h of heat treatments, when the initial crack depth of the untreated sample (613 µm) was reduced by one sixth (100 µm).After 72 h treatments the crack depths increased slightly, but were significantly smaller than that of the untreated sample.The crack width values also decreased after the thermomechanical treatments, but with less rate than the crack depth values.
In case of the ASI 310 steel it it can be clearly seen that the average crack length decreased significantly after the thermomechanical treatments.In every case the crack depth reduced to half of that of the untreated sample (Fig. 4).The best result was achieved after 1 h heat treatment with 50 % plastic deformation, when the initial crack depth of the untreated sample (229 µm) was reduced by one tenth (23 µm).After 72 h treatments the crack depths increased slightly, but were significantly smaller than that of the untreated sample.The crack width values also decreased after the thermomechanical treatments, but with smaller rate than the crack depth values.
It is clear that the 301 material is much less affected by the intercrystalline cracking, it has significantly smaller cracks in both directions and thus it is more corrosion resistant also.
To see the effect of the heat treatment for decreasing the crack length in Fig. 5. the crack length / crack length of the base material ratio is shown for both steels and thermomechanical treatments.All the treatments decreased the crack length in the austenitic samples.The best result was achieved by the 1 h 50 % plastic deformation rate for the 310 steel. in the other cases generally the 48 h treatment decreased most the crack length ratio, and in the case of 72 h treatments a increase in the crack length occurred.For all the samples the crack length ratio was under 50 %.The SEM-EBSD measurements for the AISI 304 steel showed, that the intercrystalline cracks propagate mostly along the random grain boundaries (e.g.Fig.6).In Tab. 6 the length fraction of the different grain boundary types measured by EBSD can be seen.It is clear, that the fraction of the CSL grain boundaries increased significantly after every thermomechanical treatment.Thus the CSL/(random high angle grain boundary) fraction increased too.The fraction of twin ( 3 CSL) grain boundaries also increased after thermomechanical treatment.This increase of the fraction of special gain boundaries decreased the average crack depth after braze process significantly.In Fig. 7 the dimensions of cracks are plotted versus the frac-tion of CSL/random high angle grain boundaries.A clear decrease in crack size with the increased CSL/random high angle fraction can be observed.

Conclusion
Due to their higher Cr and N i content the more corrosion resistant AISI 310 steel is much less effected by the intercrystalline cracking than the AISI 304 steel.
In the case of the AISI 304 steel the relative fraction of CSL grain boundaries to the total amount of grain boundaries increased due to the thermomechanical treatments.The relative fraction of CSL grain boundaries to the amount of random high angle grain boundaries increased significantly due to the thermomechanical treatments.The average crack dimensions decreased significantly after brazing the thermomechanical treated samples, because the CSL grain boundaries, due to their lower surface energies, stopped and blocked the intercrystalline crack propagation.

Fig. 1 .
Fig. 1.Light microscope image of the cross section of a MIG brazed specimen (304_as_ rec).

Fig. 2 .
Fig. 2. Histogram of the crack length values of sample 304_5_1h in width directionand the fitted Gauss curve

Fig. 3 .Fig. 4 .
Fig. 3. Crack length into the surface (depth) according the plastic deformation rate andthe annealing time for AISI 304 steel

Fig. 5 .
Fig. 5. Crack length/crack length of the base material ratio (in depth) according the plastic deformation rate and the annealing time for AISI 304 and AISI 310 steels