Microstructural properties of the heterogeneous welded joints of LDX2101 and AISI 304 stainless steels T. Sándor1, J. Dobránszky2 1.
Budapest University of Technology and Economics, Department of Materials Science and Engineering, PhD student, Hungary
2.
Research Group for Metals Technology of the Hungarian Academy of Sciences, Senior Research Fellow, Hungary
Abstract. The recently developed LDX2101 type lean duplex stainless steels (LDSS) have similar corrosion and better mechanical properties than that of 304 austenitic grades, which ensure its extensive spreading in the industrial applications as a substitute of AISI 304 type austenitic grade. The only one existing problem concerning LDX2101 is that while a complete range of different fittings (flanges, elbows, tees, etc.) are available from 304, but not completely from LDX2101. This means the necessity of dissimilar welded joints between LDX2101 and 304 type steels. This paper examines the dissimilar welded joints carried out by TIG and ATIG welding process on 3 mm thick plates, using four types of cold wire (LDX2101, 316LSi, 2209 and 309L) and applying 3 sizes of joint gaps (0, 1.0 and 2.5 mm). The investigation focuses on the ferrite content of the dissimilar joints which is evaluated by Fischer Feritscope and metallographic examinations assisted by colour etching. The dependency of ferrite content of the weld metal is correlated to the consumable used as cold wire and the applied fitting gap.
1. Introduction The notoriety of the lean duplex stainless steel grades, intended for substitution of the standard austenitic types, is getting wider in the last years. The extreme price fluctuations of the alloying elements (in this case, mainly the nickel) of the recent decade empowered widely the spreading of the “new” stainless steel which could ensure stable price and similar or better corrosion properties [1], [2] compared to the 304 type austenitic grades. The superior mechanical properties (compared to the 304 type austenitic grades) provided by the duplex microstructure were additional positive effect welcomed by the market. However while lean duplex plate and tube / pipe can be quite easily purchased from the producers, the situation is not similarly good from fittings’ point of view. Lean duplex fittings like elbows, tees, flanges, etc. are seldom in the offer. On the other hand many system parts operate in the process industry segment with different corrosion or mechanical load levels, which are manufactured accordingly from different grades of stainless steels. This is another situation where welder companies may have to face with the welding of lean duplex to austenitic stainless steel. Based on the fact that lean duplex stainless steels are relatively young and their wide spreading is obvious in the future [3], the examination of the dissimilar joints of lean duplexes with conventional austenitic grades (for instance 304, 316, 347, etc.) and with duplex grades (for instance 2304 and 2205) is an obviously necessary task. This must be completed with the subject of similarly important dissimilar welds of lean duplex and mild steels. These considerations give importance to the subject of this paper regarding the dissimilar welds of 304 and lean duplex stainless steels. Based on the authors’ in-the-field earned industrial experience the role of fitting gap and different consumables were investigated as main influencing factors from the obtained ferrite/austenite ratio point of view.
Duplex World 2010
2. Experimental work For the simplicity of the execution of the experiments, commercially purchased AISI 304 (1.4301) and LDX2101 (1.4162) 3 mm thick plates were used (Table 1). Table 1. The chemical composition of the base materials applied in the experiments of the paper.
EN AISI 304
C
1.4301 0.041
Mn 1.2
Si
P
S
Cr
0.39 0.028 0.003 18.03
Ni
Mo
Cu
N
8.0
0.02 0.01 0.048
LDX 2101 1.4162 0.022 4.97 0.67 0.022 0.001 21.52 1.57 0.29 0.32 0.213 TIG (GTAW) and ATIG (HP-GTAW) welding processes were applied for welding. The ATIG welding is known about its superior penetration reported by several authors [4], [5], [6]. The SiO2 + acetone suspension, performed as active flux, was brushed manually to the surface of the prepared and tack welded plates. The welding of the dissimilar joints of 304 + LDX2101 was executed with 3 different fitting gap sizes: 0, 1.0 and 2.5 mm where the parameters were set to the gap size to provide complete penetration on all samples (0 mm – 200 A / 12.8 V; 1.0 mm – 185 A / 12.5 V; 2.5 mm – 125 A / 11.2 V in case of TIG welding; 0 mm – 130 A / 11.0 V; 1.0 mm – 115 A / 10.7 V; 2.5 mm – 110 A / 10.5 V in case of ATIG welding) (Annex 1). The arc length was constant 2 mm with 3.2 mm diameter ceriated tungsten electrode (sharpened to 90°) and only pure argon was used for shielding gas and for backing gas as well. The welding speed was kept on constant 12 cm/min (2 mm/s) value with a Yamaha linear driver, which ensured the precise welding speed at all, welds. The driver is assembled onto a table, which is provided with fixtures to stabilise the plates during welding. The welding equipment was an ESAB OrigoTig 200 welding machine that measured and displayed the welding voltage during the welding process. Besides the standard consumables of the dissimilar joints (309L and 2209) [7], the austenitic 316LSi and the 2101 type were used as cold wire addition (Table 2) with constant wire feed speed (52 cm/min in case of 0 mm gap; 75 cm/min in case of 1.0 mm gap; 182 cm/min in case of 2.5 mm gap with both welding process). The welds were square butt welds without any bevelling. The mechanically cut (with plate shears) edges were only brushed and acetone cleaned.
3. Results and discussion 3.1. The importance of the size of fitting gap The chosen 0, 1.0 and 2.5 mm gap size values were chosen for the examination because it’s a widely known problem that the required fitting gap (necessary for welding) is seldom can be kept precisely in real life conditions. So if the gap variation influences somehow the ferrite content of the welds, and it obviously does, the fabricators must be aware of this correlation. As the weld is built up from the base materials and the added cold wire, the dilution of the base materials were calculated from two cross-sectional samples of each welds. The cross sectional examinations showed the tendency that the increasing fitting gap results decreasing weld cross section in case of TIG welding, while no significant reduction can be observed with ATIG welding (Figure 1.).
Duplex World 2010 TIG welds
ATIG welds
Nr. 1
Nr. 2
Nr. 3
Nr. 1
Nr. 2
Nr. 3
Nr. 4
Nr. 5
Nr. 6
Nr. 4
Nr. 5
Nr. 6
Nr. 7
Nr. 8
Nr. 9
Nr. 7
Nr. 8
Nr. 9
Nr. 10
Nr. 11
Nr. 12
Nr. 10
Nr. 11
Nr. 12
Figure 1. The cross sections of the welded joints; plate thickness is 3 mm (parameters in Annex 1).
The weld cross sections are depending on the heat inputs. However it is clearly visible that the heat input of ATIG welding is significantly lower than that of TIG welding, while it still can provide complete penetration. This phenomenon was expected based on the known differences between TIG and ATIG welding, however no publication can be found in the literature reporting about ATIG welding with cold wire addition. Hereby two important observations can be done. Firstly, the applicability of ATIG welding with cold wire addition is out of question from now. All the welds, carried out with this process, had complete penetration. The profile of the weld cross section is beneficial from weld distortions point of view, but disadvantageous from risk of solidification crack sensitivity point of view [8], however no such a reported problem can be found in the publications in relation with ATIG welding. The second observation is that the increasing gap size reduces the heat input differences from ~40% to ~10% between the TIG and ATIG welding and together with it, the reduction of the difference of weld metal cross section is changing in similar ratio (Figure 2). 1,4
Weld bead cross section
Welding heat input
40
Heat input (kJ/mm)
TIG # LDX2101 TIG # 309L TIG # 316LSi TIG # 2209
1,0
30 ATIG # LDX2101 ATIG # 309L ATIG # 316LSi ATIG # 2209
25 20
0,8
15 0,6 0,0
0,5
1,0 1,5 Root gap (mm)
2,0
2,5
0,0
0,5
1,0 1,5 Root gap (mm)
2,0
2,5
10
Figure 2. Correlation between the weld cross section and the root gap and the heat inputs
Cross section area (mm2)
35 1,2
Duplex World 2010 The retrieved figures of the dilutions from macrographs show very well the correlation between the fitting gap and the obtained dilution (Figure 3). ATIG welding operates with lower base material dilution, which is a consequence of its inward weld metal flow instead of the outward that is the own property of TIG welding. The difference in dilution values is about 10-20% and may have importance in severe conditions.
Base metal dilution rate ( % )
90 80 70 60 TIG # LDX2101 ATIG # LDX2101 TIG # 309L ATIG # 309L TIG # 316LSi ATIG # 316LSi TIG # 2209 ATIG # 2209
50 40 30 20 10
0,0
0,5
1,0
1,5
2,0
2,5
Root gap (mm)
Figure 3. Dilution rates in correlation with the fitting gap for the different consumables.
3.2. The consumables It is obvious that when welds should be carried out with fitting gap bigger than zero, solid wire must be added to the weld pool to fill up the gap. It is also obvious that the applicable welding current must be reduced, with increasing fitting gap, so it also might be predicted that the ferrite content will tend to the welding wire by increasing gap. Theoretically the ferrite content might be varied from the low ferrite contents (2-6%) of 304 type austenitic stainless steels up to that of the typical duplex’s 45-55% value. However the choice of appropriate consumable is depending on several factors (surface quality, temperature, stress, etc.) [9], and includes non-scientific considerations (costs, availability, etc.) as well, so the four chosen welding wire is not a proposal of the authors, rather only a rational consideration of the possibilities. The ferrite contents of the chosen wires cover well the transition between 304 and LDX2101, from 3 to 41 FN (according to WRC’92) (Table 2). Table 2. Chemical composition of the used cold wires.
TYPE
CHEMICAL COMPOSITION C
Mn
Si
P
S
FN
Cr
Ni
Mo
2101
0.009 0.7 0.45 0.017 0.002 23.4
7.3
0.23 0.13 0.167
309L
0.01
1.7
0.4
0.015
0.01
23.5 13.4
0.1
0.08
0.05
316LSi
0.01
1.8
0.9
0.015
0.01
18.4 12.2
2.6
0.12
0.05
10
2209
0.01
1.6
0.5
0.015 0.002 23.0
3.2
0.1
0.16
40
8.6
Cu
N
Ti
WRC’92 41
0.001
20
Duplex World 2010 In view of the obtained dilution rates (Figure 3), the chemical compositions of the base materials (Table 1) and the consumables (Table 2) the expectable ferrite content prediction can be executed. For this purpose, the WRC’92 diagram was used (Figure 4).
Cold wire: LDX 2101
Cold wire: 309L
Cold wire: 316LSi
Cold wire: 2209 WRC-92 diagram
Figure 4. Ferrite content prediction of the dissimilar joints in WRC’92 diagram with the four chosen consumables and the measured dilution rates.
3.3. The ferrite measurements As in industrial environment the final product can be only checked by non-destructive tests, Fischer Feritscope FMP30 was chosen to carry out the necessary ferrite measurements. This decision is in harmony with IIW proposal [8]. The longitudinal centreline of the surface was measured on ten points while the cross sections were measured on three points. The obtained results were averaged and put into the diagram of Figure 6. Additionally image analysis was executed on all the samples with image analyser software (JMicroVision 1.2.7.). The N=20x images (Figure 5) from the root region were analysed after color etching (Beraha’s reagent).
Duplex World 2010 a)
c)
e)
b)
d)
f)
Figure 5. Weld metal of the ATIG-welded sample Nr.8. (a) and of the TIG-welded sample Nr.10. (b). (see the parameters in Annex 1). The other pictures (TIG-welded sample Nr.4) show the great difference in the width of the HAZ at the duplex (c, e) and austenitic (d, f) side, and at the face (c, d) and the root (e, f) side of the weld
The results of the ferrite content measurements show good accordance with each other. The average deviation is about 10% and the maximum is 20%. These results are acceptable based on the international experiences [11]. The results of image analyser show bigger deviation especially on the ATIG welds, which explanation might be the relatively small examined area of the weld and the different weld flow phenomenon providing higher cooling rates on the root side. However deeper analysis and more precise understanding of ATIG welding are necessary for the correct explanation. All diagrams display that the ferrite content of the weld tends to that of the consumable how it was predicted in the 3.2 subsection. The rather large deviation observed in case of 0 mm fitting gap might
Duplex World 2010 be explained by the asymmetrical penetration profiles that occurred in all cases of TIG welds with 0 mm gap (Figure 1). The asymmetric melting of the two base materials can be explained by the different heat conductivity of the two materials. All the diagrams show that the increasing gap assists the consumables to become more dominant. This dominancy is obvious when considering the ferrite values (Figure 6). However the ferrite measurements were carried in ferrite % and internationally rather the FN is approved, the measured % and calculated FN values are in very good accordance with each other. As a consequence it can be appointed that the prediction about the importance of the fitting gap is proved hereby. Its effect from ferrite content point of view is depending on the grade of the used consumable. R oot gap (m m ) 1 ,0 1 ,5
2 ,5
0 ,0
0 ,5
R oot gap (m m ) 1 ,0 1 ,5
2 ,0
2 ,5
3 0 9 L weldin g wire
L D X 2 1 0 1 weldin g wire
90
80
80
70
70
60
60
50
50
40
40
30
30
20
20
10 0 90
3 1 6 L S i weldin g wire
80 Ferrite content ( % )
2 ,0
T IG # F erits c ope - C ros s s ec tion T IG # F erits c ope - F ac e s ide T IG # Im age an aly s er A T IG # F erits c ope - C ros s s ec tion A T IG # F erits c ope - F ac e s ide A T IG # Im age an aly s er
10 0 2 2 0 9 weldin g wire
90 80
70
70
60
60
50
50
40
40
30
30
20
20
10
10
0
0 ,0
0 ,5
1 ,0 1 ,5 R oot gap (m m )
2 ,0
2 ,5
0 ,0
0 ,5
Ferrite content ( % )
Ferrite content ( % )
90
0 ,5
1 ,0 1 ,5 R oot gap (m m )
2 ,0
2 ,5
Ferrite content ( % )
0 ,0
0
Figure 6. Summarized results of the ferrite content measurements in the case of the four consumables.
3.4. Scanning electron microscopy analysis 3.4.1 SEM-EBSD analysis Based on SEM-EBSD analysis, which was executed to evaluate the distribution of the ferrite from the fusion line towards the centre of the weld, it could have been stated that the ferrite is quite evenly distributed in the weld in transverse direction (Fig. 7 and Fig. 8). However EBSD analysis is quite slow process this method is the most reliable and precise method for determination of the constituting phases. The stronger grain coarsening of LDX2101 can be observed in case of TIG welding compared to ATIG. This is due to the higher heat input values applied in case of TIG welding.
Duplex World 2010 304
Weld metal LDX2101 304
Weld metal
LDX2101
Figure 7. Grain orientation maps (on top) and phase maps (on bottom; austenite in red, ferrite in green) represent the crystallographic and phase constitutions at the two fusion lines of the ATIG-welded sample Nr.4
304
LDX2101
HAZ
Weld
Weld
Figure 8. Grain orientation maps (on top) and phase maps (on bottom; austenite in red, ferrite in green) represent the crystallographic and phase constitutions at the two fusion lines of the TIG-welded sample Nr. 4
Duplex World 2010 3.4.2 SEM-EDS analysis As final examination, SEM-EDS line analysis was executed to measure the line element distribution of Cr, Ni and Mn across the fusion line. As the weld pool convection of TIG and ATIG welding is significantly differs, the examination of the potential differences of the main alloying elements distribution is important. Both the ‘LDX2101 → weld metal’ and the ‘weld metal → 304’ transition (fusion line environment) were measured on the weld cross section near to the welding (top) and to the root (bottom) side of the joint. The deep analysis of line profiles, taking into account the composition of the wire addition, is not yet complete, but the strong equalization of the elements close to the fusion line can be observed (Figure 9. and 10.). The width of this transition zone (where the local homogenization takes place) can be characterised by the gradient of the element distribution line. In case of the ATIG welded sample (Nr. 4.) the width of the element transition zone is near two times large on the root side as that is on the welding (upper) side. This observation is probably caused by the non-perpendicularity of the measure line to the fusion line. Generally the width of the elements’ transition zone is about 60-120 µm in ATIG welded joints, but 200-300 µm in TIG welded joints. 0
Distance (micrometer) 400 600 800
200 Face side
Distance (micrometer) 1000
1000
1200
800
600
400
200
0
Face side
Chromium
20 Chromium
Austenitic base 12 metal
Measuring line
Measuring line Weld metal
Nickel
HAZ
Weld metal Nickel
LDX2101 base metal
15
10
6
Weight percent
Weight percent
18
5 Manganese
0
Manganese
Weight percent
Chromium
Austenitic 15 base metal
Weld metal HAZ
Chromium
LDX2101 base metal
Weld metal Measuring line
Measuring line
15
HAZ
10
10 Nickel
0
5
Nickel
5 Root side
0
500 Distance (micrometer)
Manganese
1000
Manganese
1200
1000
Root side
800 600 400 Distance (micrometer)
200
Figure 9. Line element distribution of Cr, Mn and Ni in the ATIG-welded sample Nr. 4.
0
0
Weight percent
20
20
Duplex World 2010 Distance (micrometer) 1500 1000 500
2000
0
Distance (micrometer) 2000
1500
Face side
1000
500
0
Face side
20
20 Chromium
15
15 HAZ
Measuring line
10
Measuring line Nickel
Weld metal
10
Nickel
5
5 Manganese
Manganese
Austenitic base metal
0
0
Weight percent
20
20 Weld metal
15
Chromium
Measuring line
10
Chromium
Weld metal
15 HAZ
Measuring line Nickel
HAZ
10 Nickel
Austenitic base metal 5
5 Manganese
0
Weight percent
Chromium
Root side
0
500
1000 1500 Distance (micrometer)
Manganese
Root side
2000
Weight percent
Weight percent
Weld metal
2000
1500 1000 500 Distance (micrometer)
0
0
Figure 10. Line element distribution of Cr, Mn and Ni in the TIG-welded sample Nr. 4.
4. Conclusions Based on the experiments carried out on 304 + LDX2101 dissimilar joints with 3 different fitting gap (0, 1.0 and 2.5 mm) and four kind of consumables (1.0 mm solid wire: 2101, 309L, 316LSi and 2209) the following observations were done: The ATIG welding can be applied with cold wire addition without any problem while the process keeps its own beneficial penetration properties. The base material dilution to the weld metal is significantly lower with ATIG welding, and it decreases with the increasing fitting gap. The differences of necessary heat inputs for complete penetration of TIG and ATIG welding are reducing (from the ~40% difference obtained at 0 mm gap) with increasing fitting gap, however a certain difference (approximately 10%) still remains in case of 2.5 mm gap. The asymmetric cross section of the LDX2101 + 304 dissimilar welds were observed, especially in case of 0 mm fitting gap, which phenomenon diminishes with increasing fitting gap, but a certain difference still remains at 2.5 mm. The phenomenon is very strong in case of TIG welding and weaker in case of ATIG welding and can be explained by the different heat conductivities of the two materials. The deeper investigation of this phenomenon and its effects on the quality of the joint are necessary for the correct evaluation. The dilution reduces with the increasing fitting gap, which consequently empowers the added consumable to dominate from ferrite content and chemical composition point of
Duplex World 2010
view. The prediction, based on WRC’92 diagram (and possessing the dilution rates), shows good matching to the measured values and follows the tendencies, which can be seen, on the diagram. SEM-EBSD analysis showed that the ferrite content is quite even in transverse direction of the weld from the base materials towards the weld metal. The SEM-EDS line analysis appointed that the alloying elements have a transition zone where homogenization takes place and its width strongly depends on the weld pool convection. Obviously the described experiments will continue with corrosion and mechanical tests to obtain a global picture about the effects of the fitting gap and the welding consumable.
References [1] M. Benson; ACOM Vol. 3, pp. 2-11. (2005). [2] C. Ericsson, V. Jägerström, J. Olsson; Proceedings of the International Symposium on Corrosion in the Pulp and Paper Industry 11. pp. 333-350, (2004). [3] J. Charles; Steel Research International, 79 pp. 455-465, (2008). [4] N. Ames, S. Babu; IIW doc., IX-2195-06. [5] J.J. Lowke, M. Tanaka, M. Ushio; J. Phys. D: Appl. Phys. 38. pp. 3438-3445, (2005). [6] T. Sándor, J. Dobránszky; Proc. of Stainless Steel World Conf., Maastricht, (2009). [7] L. Karlsson, J. Strömberg, S. Rigdal; Proc. of Duplex America 2000 Conference on Duplex Stainless Steels, pp. 273-280, (2000). [8] V. Shankar, T.P.S. Gill, S.L. Mannan, S. Sundaresan; Sadhana Vol. 28, Parts 3&4, June/August 2003, pp. 355-382. [9] T.L. Ladwein, M. Sorg, S. Schilling; Proc. of European Stainless Steel Conference, A02-14, (2008). [10] Welding consumables – Predicted and measured FN in specifications – Position statement of the experts of IIW Comission IX, Technical report ISO/TR 22824:2003. [11] J.C.M. Farrar; Welding in the World, Vol. 49. no 5/6, (2005).
Duplex World 2010
Annex 1 The welding parameters of the experiments. TIG-welded samples
1
2
3
4
5
Base material 1 Base material 2 Added cold wire Wire feed speed (cm/min) Shielding gas Root gap (mm) Welding current (A) Welding voltage (V) Welding speed (mm/s) Heat input (kJ / mm)
LDX 304 2101 52 Ar 0 200 12.7 2 1.270
LDX 304 2101 75 Ar 1 185 12.5 2 1.156
LDX 304 2101 182 Ar 2.5 130 11.2 2 0.728
LDX 304 309L 52 Ar 0 200 13 2 1.300
LDX 304 309L 75 Ar 1 185 12.5 2 1.156
ATIG-welded samples
1
2
3
4
5
LDX 304 309L 52 Ar 0 130 11.2 2 0.728
LDX 304 309L 75 Ar 1 115 10.7 2 0.615
Base material 1 LDX LDX LDX Base material 2 304 304 304 Added cold wire 2101 2101 2101 Wire feed speed (cm/min) 52 75 182 Shielding Ar Ar Ar Root gap (mm) 0 1 2.5 Current (A) 130 120 110 Voltage (V) 11 11.2 9.5 Welding speed (mm/s) 2 2 2 Heat input (kJ / mm) 0.715 0.672 0.523
6
7
8
9
LDX LDX LDX LDX 304 304 304 304 309L 316 LSi 316 LSi 316 LSi 185 52 75 185 Ar Ar Ar Ar 2.5 0 1 2.5 125 200 185 120 11.5 12.7 12.3 11.2 2 2 2 2 0.719 1.270 1.138 0.672
6
7
8
9
LDX LDX LDX LDX 304 304 304 304 309L 316 LSi 316 LSi 316 LSi 185 52 75 185 Ar Ar Ar Ar 2.5 0 1 2.5 110 130 115 110 11.2 11.7 11.0 11.5 2 2 2 2 0.616 0.761 0.633 0.633
10
11
12
LDX 304 2209 52 Ar 0 200 13 2 1.300
LDX 304 2209 75 Ar 1 185 12.7 2 1.175
LDX 304 2209 185 Ar 2.5 120 10.7 2 0.642
10
11
12
LDX 304 2209 52 Ar 0 130 10.5 2 0.683
LDX 304 2209 75 Ar 1 115 10.7 2 0.615
LDX 304 2209 185 Ar 2.5 110 11 2 0.605
