Exjobb_LDX2101_svetsoxid_080627

Investigation and ranking of Localized Corrosion Resistance of Welded austenitic and lean duplex stainless steel

ABSTRACT

ii

SAMMANFATTNING

iii

Content1 Introduction_____________________________________________________________1

2 Theory_________________________________________________________________2

2.1 Stainless steel_______________________________________________________22.1.1 Ferritic stainless steels_____________________________________________22.1.2 Austenitic stainless steels___________________________________________42.1.3 Ferritic-austenitic duplex stainless steels_______________________________52.1.4 Martensitic stainless steels__________________________________________62.1.5 Alloying elements________________________________________________7

2.2 Corrosion__________________________________________________________92.2.1 Uniform corrosion_______________________________________________102.2.2 Galvanic corrosion_______________________________________________112.2.3 Localised corrosion______________________________________________11

2.3 Corrosion testing___________________________________________________152.3.1 Polarization curves_______________________________________________162.3.2 The Avesta Cell_________________________________________________182.3.3 The Multicell___________________________________________________20

2.4 TIG Welding_______________________________________________________21

2.5 Weld oxide formation_______________________________________________22

2.6 Post-weld cleaning__________________________________________________24

2.6 GDOES___________________________________________________________25

3 Experimental___________________________________________________________26

3.1 Material___________________________________________________________27

3.2 Welding___________________________________________________________27

3.3 Post weld cleaning__________________________________________________28

3.4 Corrosion testing___________________________________________________293.4.1 Avesta Cell_____________________________________________________293.4.2 Multicell_______________________________________________________29

3.5 GDOES___________________________________________________________30

4 Results________________________________________________________________31

4.1 Base metal_________________________________________________________31

4.2 Topside___________________________________________________________324.2.1 Avesta Cell_____________________________________________________324.2.2 Multicell_______________________________________________________364.2.3 GDOES_______________________________________________________38

4.3 Root side__________________________________________________________39

4.4 LDX 2101 weld metal microstructure__________________________________40

5 Discussion_____________________________________________________________41

5.1 Base metal_________________________________________________________41

iv

5.2 Topside___________________________________________________________425.2.1 Avesta Cell_____________________________________________________435.2.2 Multicell_______________________________________________________455.2.3 GDOES_______________________________________________________45

5.3 Root side__________________________________________________________465.3.1 Multicell_______________________________________________________465.3.2 GDOES_______________________________________________________47

5.4 Ferrite content_____________________________________________________47

6 Conclusions____________________________________________________________48

7 Suggestion to further work________________________________________________49

8 Acknowledgements______________________________________________________49

9 References_____________________________________________________________50

Appendix A________________________________________________________________52

Appendix B________________________________________________________________53

v

1 IntroductionThe passive film naturally formed in oxidizing environments protects the stainless steel and makes it corrosion resistant. When welding these materials, the high temperature will cause thermal growth of the passive film forming weld oxide, often referred to as heat tint. The resulting reduction in corrosion resistance is often explained by a chromium-depleted layer under the chromium-rich weld oxide, but the pitting attacks normally occur where the oxide is most iron-rich [1, 2, 3]. As the low-alloyed grades appear to be most sensitive, most work has been performed on austenitic steels of 304 and 316 type. However, recent work has shown that the lean duplex stainless steel LDX 2101® is more sensitive to residual weld oxide in the standardized pitting resistance test ASTM G150 than 316. An initial current peak in the test prevents actual measurement and this needs to be confirmed by other means [SSW 2008]. An important difference between the weld oxides formed on the lean duplex grades LDX 2101 and 2304 is that these are considerably more manganese-rich than the heat tint on 304 and 316 that is more chromium-rich [4]. It is not known to which extent this manganese fraction affects the corrosion resistance. Normally it is assumed that all weld oxide will be removed with a suitable post-weld cleaning procedure before being used in a real application. The most efficient way is to remove it by chemical pickling. Environmental aspects can sometimes limit the cleaning to mechanical treatments; often known to cause reduced corrosion resistance compared to pickling. For instance polishing of the heat tint can visually give it a shiny surface concealing residual oxide. The effect of insufficient post-weld cleaning of lean duplex grades of LDX 2101 type on the pitting resistance is unknown and the manganese-rich oxide could play an important role for the corrosion performance. Recent work has also shown that most manganese in the heat tint is deposited from the weld metal, which is further enhanced by nitrogen additions to the shielding gas [4].

For this purpose, GTA welding is performed on the lean duplex grade LDX 2101 with and without nitrogen additions to the shielding gas. The austenitic grades 304 and 316L are welded with the same procedure using only argon as comparison. The welds are cleaned by either laboratory pickling, polishing or left in as-welded condition. Each combination is corrosion tested in the Avesta Multicell based on the ASTM G61 standard (pitting potential), with six specimens for improved statistics. GDOES is used for measuring oxide composition and thickness, and to explain phenomena revealed during corrosion testing. The weld microstructure is shown and a literature review of the previous work on weld oxides is presented.

The aim of this diploma work was to investigate the pitting resistance of welded LDX 2101. Previous examinations have shown that residual weld oxides formed on this material prevents reliable measurements using the standard ASTM G150 pitting test. An initial current peak exceeding the threshold value according to the ASTM G150 standard was suspected to give large underestimations of the pitting resistance. This current peak was investigated by recording polarization curves based on the ASTM G61 standard. Pitting potentials was measured in as-welded condition for LDX 2101 using 304 and 316L as reference materials. The second aim was to rank the three grades (304, LDX 2101, 316L) based on the pitting potential of the base metal, in as-welded condition and after post-weld cleaning by either polishing or pickling.

1

2 TheoryThis first section gives an orientation of the materials and the experimental work that has been carried out. A general description of the most common types stainless steel and their properties is given together with a presentation of the different corrosion types that commonly occur in stainless steels. Theory behind corrosion testing is given and the used welding method is presented. Finally different types of post-weld cleaning methods are described.

2.1 Stainless steelStainless steel contains a minimum of 10.5% chromium. The outer surface of such an alloy will react spontaneously with atmospheric oxygen or in aerated aqueous solutions and form a 2-3 nm thin protective chromium oxide layer [5, 6, 7]. This oxide makes the alloy passive, hence the name “passive film”, and improves the general corrosion resistance of the alloy. The oxidation and formation of the thin passive layer on the steel surface are dependent on how clean the surface is. Anything that blocks the steel surface from contact with oxygen such as dirt, oil, grease, coatings and paint could prevent the formation of the passive layer and result in decreased corrosion resistance. The passive film will be destroyed locally if the metal is scratched, but if sufficient amounts of oxidants are available, the metal will react spontaneously with oxygen and repassivate the scratch.

Stainless steels are generally weldable, but special procedures might be needed. Welding normally results in a local change of the microstructure in the heat-affected zone (HAZ) and the weld metal, compared with that of the base metal. This can shift the phase fractions in the material and alter the desired phase balance, which can result in segregation, grain growth and formation of intermetallic phases that could change the corrosion performance and mechanical properties of the steel [8].

Stainless steels can be divided into different application areas such as stainless steel, acid-proof steel and high-temperature steel. Another way to arrange the steels is by their alloying elements; chromium steel, chromium-nickel steel and chromium-nickel-molybdenum steel. However, the most common arrangement is dividing them according to their microstructure into the different phases:

Ferritic steels Austenitic steels Ferritic-austenitic (duplex) steels Martensitic steels

2.1.1 Ferritic stainless steelsFerritic stainless steels are in principle ferritic at all temperatures. These grades are alloyed with 13 – 18% chromium, and in some cases up to 30% [9]. Fully ferritic steels (1.4016) normally contain 17% chromium. Titanium or niobium is often added since these elements bind carbon easier than chromium and thus prevent formation of chromium carbides. Figure 1 shows the microstructure of a ferritic stainless steel. Low alloyed (12 – 13% Cr) ferritic stainless steels are primarily used in the automotive industry for applications such as exhaust systems. Ferritic stainless steels with less than 20% chromium are often used in applications

2

such as household utensils and vehicle components. The high alloyed grades containing, 20 – 30% chromium are most common in high temperature applications such as flues and furnaces [9].

Figure 1. Microstructure of a ferritic stainless steel.

Ferritic stainless steels have higher yield strength but a lower tensile strength than the common types of austenitic stainless steels. Another difference is that their elongation at fracture is only half of that of austenitic stainless steels [10]. Small increases in strength can be achieved by performing solution hardening with carbon and nitrogen.

The corrosion resistance of ferritic stainless steels can vary depending on application area. The risk of stress corrosion cracking (SCC) is decreased when a ferritic microstructure is present in the steel. Low alloyed (12 – 13% Cr) ferritic stainless steels are becoming more and more common, where the modern types e.g. type 430 have low carbon content and good resistance to atmospheric corrosion and SCC [10].

Some of the disadvantages with ferritic stainless steels are embrittlement due to the precipitation of the brittle intermetallic sigma phase in the 500 – 800C interval during long-time exposure. Another is the 475C embrittlement. Higher chromium contents increase the risk of sigma phase precipitation and 475C embrittlement [11].

The weldability of ferritic stainless steels is highly dependent on the chemical composition of the steel. For older types of stainless steels the problem with precipitation of chromium carbides, due to higher carbon content, sometimes resulted in intergranular corrosion. This is not a large problem with modern ferritic stainless steels. The weldability has been improved by lower carbon and nitrogen contents, which results in a fully ferritic structure at all temperatures. Ferritic stainless steels with higher carbon content can, however, be rather sensitive to intergranular corrosion. All ferritic stainless steel are sensitive to grain growth in the heat-affected zone (HAZ) [10]. This grain growth can reduce the ductility of the steel, requiring that the heat input during welding should be kept as low as possible. The HAZ can be brittle, mainly caused by partial toughening of the area adjacent to the weld and by grain coarsening. The embrittlement caused by the formation of martensite can be reduced by post weld heat treatment at 750C, but embrittlement caused by grain coarsening will remain [9]. The stabilized ferritic grades have improved weldability by preventing grain growth.

Hydrogen embrittlement can also be a problem when welding ferritic stainless steels, but it can be avoided if all sources containing hydrogen such as in electrodes and shielding gases is limited to a low level. When welding ferritic stainless steels, austenitic filler materials are

3

recommended for optimal ductility, while ferritic fillers are used when the application requires a fully ferritic microstructure [10].

2.1.2 Austenitic stainless steelsThe most common type of stainless steel is the austenitic grade 1.4301, 304 containing 18%-Cr and 8%-Ni. This steel has good corrosion properties in many environments, and comes in different types with varying carbon and nitrogen contents. Some of them are stabilized with titanium or niobium. Figure 2 shows the microstructure of a fully austenitic stainless steel. Austenitic stainless steels can be used in a wide range of applications such as processing and high corrosive environments, e.g. offshore installations, high temperature equipment, paper and chemical industries.

Figure 2. Typical microstructure of an austenitic stainless steel.

Austenitic stainless steels are a group of steels with varying properties; the most important being corrosion resistance. The various grades have different corrosion performance and can be used in a wide range of applications. Austenitic stainless steels can be divided according to their properties into the following groups:

Austenitic without molybdenum (e.g. 304 and 304L) Austenitic with molybdenum (e.g. 316, 316L, 317L and 904L) Stabilized austenitic (e.g. 321, 321H and 316Ti) Fully austenitic with high molybdenum (e.g. 254 SMO®) Heat and creep-resistant (e.g. 321H, 253 MA and 310S)

Austenitic stainless steels with or without molybdenum have a microstructure consisting of austenite and in some cases a small fraction of delta ferrite. The compositions for these types of alloys are typically 17 – 20% chromium and 8 – 13% nickel. The most common stainless steel type is 1.43XX, 304 with 18% chromium and 8 – 9% nickel. This is used in mild corrosive environments, for instance within pulp and paper. When alloyed with molybdenum these austenitic grades commonly contain 2 – 3% molybdenum, which increases the pitting resistance of the alloy [10]. 1.44XX, 316 is an example and this material can be used in environments with somewhat higher chloride contents.

Stabilized austenitic stainless steels have an addition of titanium or niobium to prevent precipitation of chromium carbides. These grades are preferably used in high temperature applications.

4

Fully austenitic stainless steels are high alloyed with 20 – 25% chromium, 18 – 35% nickel and up to 0.4% nitrogen. These are used in highly demanding corrosive environments such as in offshore and chemical industries.

The most characteristic mechanical property for austenitic stainless steels is their excellent ductility at both high and low temperatures. Fully austenitic stainless steels have good impact toughness and a generally high yield and tensile strength. Austenitic stainless steels have excellent corrosion resistance. The carbon content is normally low (<0.03%), which increases the resistance to intergranular corrosion. There are many different grades of austenitic stainless steels available to meet the different types of corrosive environments. Chromium and nickel alloys have good general corrosion, but to acquire good resistance to pitting and crevice corrosion, molybdenum contents up to approximately 5% might be necessary. For improved hot cracking resistance and better weldability austenitic stainless steel grades such as type 316 are often manufactured to result in some ferrite fraction. Resistance to SCC increases with increased nickel and molybdenum content, but the same elements also promote formation of intermetallic phases [10].

The weldability of standard austenitic stainless steels is generally very good, while the fully austenitic grades might need some more attention due to increased risk of hot cracking, segregation and precipitation of intermetallic phases. If ferrite is present in the microstructure during solidification, impurities such as sulphur and phosphorus will dissolve into the ferrite. When a fully austenitic structure is present the impurities tend to segregate in the austenitic grain boundaries, which increase the risk of hot cracking. To avoid hot cracking and precipitations of intermetallic phases, specially developed filler metals are used to obtain the desirable microstructure and phase fractions. The heat input during welding of fully austenitic stainless steels might be restricted and need to be controlled. In most cases no preheating or post-weld heat treatment is necessary.

2.1.3 Ferritic-austenitic duplex stainless steelsDuplex means twofold, which in this case designate phases i.e. duplex stainless steels contain a two-phase microstructure of ferrite and austenite. The purpose of duplex stainless steels is to combine the benefits from each phase for overall improved properties. Figure 3 shows the microstructure of a duplex stainless steel.

Figure 3. Typical microstructure of a duplex stainless steel.

The duplex grades have an optimised composition typically containing 18 – 26% chromium, 4 – 8% nickel, 0 – 4.5% molybdenum, 0.1 – 0.50% nitrogen and 0.02 – 0.03% carbon [12].

5

These are heat treated during manufacturing to result in equal propositions of ferrite and austenite. Duplex stainless steels are often divided into three different groups with increasing alloy content:

Lean alloyed duplex stainless steels (LDX 2101® and 2304) High alloyed duplex stainless steels (2205) Superduplex stainless steels (2507)

The most common duplex stainless steel is 1.4462 (2205), but the use of lean duplex grades has grown the last years.

Duplex stainless steels have very high mechanical strength, which can sometimes be up to two times higher than that of the austenitic grades. The ductility is good, especially at room temperature, but not as good as for austenitic stainless steels [9].

In duplex stainless steels the high chromium content, and often additions of molybdenum and nitrogen, give the steel a high resistance to pitting and crevice corrosion. The duplex grade 2205 has e.g. a higher pitting resistance than the austenitic type 316. Furthermore, the duplex microstructure has superior resistance to SCC, and the low carbon content makes them resistant to intergranular corrosion [10].

The weldability of duplex stainless steel is generally good. However, the fluidity of the melt is somewhat lower compared with austenitic stainless steels. The welding parameters should therefore be adapted for welding in duplex stainless steel and may not be the same as for austenitic stainless steel. Duplex stainless steel solidifies with a fully ferritic structure, with subsequent austenite precipitation and growth during cooling. If the cooling rate is very high, e.g. if the heat input is too low, then there is increased risk of excessive ferrite contents due to the fact that there is insufficient time for nitrogen diffusion and associated austenite transformation. Ferrite contents exceeding 65% – 75% or more can result in reduced corrosion resistance and ductility. A high content of nitrogen is added in modern duplex stainless steels in order to stabilize austenite at higher temperatures when welding. When welding superduplex grades, slow cooling in the temperature range of 700 – 980C should if possible be avoided to minimize the risk of formation of intermetallic phases, which can have a negative effect on corrosion resistance and toughness. Heat input should be controlled and is typically 0.5 to 3.0 kJ/mm. The filler metals used when welding duplex stainless steels should be matched with the parent metal and are in most cases over-alloyed. The nickel content in the filler metal is higher than in the parent metal in order to stabilize the precipitation of austenite during the rapid cooling after welding [10].

Duplex stainless steels are mainly used in applications where it is possible to make use of their high mechanical strength combined with a good corrosion resistance. Typical applications are; heat exchangers, pressure vessels, pulp digesters, chemical industry equipment, rotors, fans and huge tanks used for transporting or storing chemicals.

2.1.4 Martensitic stainless steelsMartensitic stainless steels have a chromium content of typically 12 – 17%, a carbon content exceeding 0.1% and if added, a nickel content of a few percent [9]. The formation and maintenance of a martensitic microstructure is possible because of the high carbon content, compared with e.g. austenitic and duplex stainless steels. A martensitic microstructure can be

6

seen in Figure 4. Martensitic stainless steels are used for knifes and razor blades where a sharp edge is important. The strength in martensitic stainless steels can be enhanced by precipitation hardening and formation of intermetallic phases when adding certain alloying elements. The heat treatment procedure of these types of stainless steels should be carefully controlled. The corrosion resistance of martensitic stainless steels are modest, but can be improved by addition of molybdenum. Molybdenum alloyed martensitic stainless steels have high strength, good erosion resistance and are used in applications such as process vessels, within the petroleum industry and for propellers.

Figure 4. Typical microstructure of a martensitic stainless steel.

The mechanical properties for martensitic stainless steels are characterised by very high yield strength, tensile strength and a high hardness. The elongation and impact toughness of martensitic stainless steels are normally poor [10].

Fully martensitic stainless steels have poor weldability due to the high hardness and low ductility. This makes these types of stainless steels very sensitive to hydrogen cracking, unless a careful preparation is carried out before welding. To minimize the risk of hydrogen cracking preheating and post weld annealing should be performed. To obtain optimal mechanical properties matching filler material should be used, otherwise austenitic and duplex fillers are most common, giving highest weld metal ductility.

2.1.5 Alloying elements

Chromium – CrChromium is the most important element in stainless steels. It stabilizes ferrite and minimum 10.5% is needed to create and maintain the passive film. Chromium is known to increase the pitting potential, the critical pitting temperature (CPT) and the critical crevice temperature (CCT). An increased chromium content helps to increases the passive film stability in acidic environments and decreases the pitting propagation. Coordinated increases of nickel, or nitrogen are required to maintain the phase balance and addition of chromium is presently limited to about 27% due to risk of precipitation of intermetallic phases [13]. For austenitic steels the chromium content is typically 17 – 18%, while the duplex stainless steels have 20 – 29%.

Nickel – Ni

7

Nickel is a strong austenite stabilizer. To be able to attain a fully austenitic microstructure in steels with 18% chromium, additions of approximately 8 – 9% nickel is required to balance the ferrite stabilizing element chromium. Nickel is an important element for improving the pitting and crevice corrosion resistance of welded duplex components and consumables are consequently somewhat over-alloyed in nickel [14]. Low nickel levels can result in the formation of a high proportion of ferrite in the duplex microstructure with subsequent loss in toughness. It should be noted that the principle role of nickel in duplex stainless steels is not to modify corrosion resistance, but rather to control ferrite/austenite balance in the microstructure [12].

Molybdenum – MoMolybdenum is a ferrite stabilizer that increases the corrosion resistance to localised and uniform corrosion. In steels where molybdenum is added, additions of austenite stabilizing elements such as nickel or manganese are needed to maintain the austenitic microstructure. A molybdenum content in excess of 3 or 3.2% is required to have a good resistance to localized corrosion in hot sea water and to ferric chloride tests [15]. For lower molybdenum contents, complete resistance to pitting cannot be achieved in acidic environments even by increasing the chromium content of the alloy [16,13].

Nitrogen – NNitrogen is an austenite stabilizing element and is increasingly used for improving the corrosion resistance of duplex steels. Apart from increased localised corrosion resistance such as pitting and crevice corrosion, nitrogen increases the strength of the alloy when it is in solution. The duplex grades normally contain 0.1 – 0.3% nitrogen to have an improved austenite formation when welding. The addition of nitrogen in duplex stainless steel is associated with an increase in the A4 temperature, leading not only to an increase in austenite content at peak temperatures, but the transformation also starts at higher temperatures during the weld cooling cycle. Sakai et al. [17] demonstrated the beneficial effects of nitrogen and molybdenum on pitting corrosion resistance and Truman et al. [18] showed that the beneficial effects of nitrogen seem to have been further enhanced in the presence of molybdenum. Nitrogen also improves the ductility and toughness of duplex stainless steels due to its austenite stabilizing effect. Ogawa and Koseki [19] indicated that an increase in the nitrogen content of duplex stainless steels improves the pitting corrosion resistance of the weld. [12].

Manganese – MnManganese stabilises austenite and is used in the lean duplex stainless steels together with nitrogen to compensate the lower nickel content. This makes the price of the material more stable since the nickel price has fluctuated significantly over the last years.

Carbon - CCarbon is a strong austenite stabilizing element. Carbon atoms are dissolved interstitially in the steel, which results in build up of internal stresses that results in increased strength. The new and modern process metallurgy techniques used today allow production of steels with very low carbon content. Carbon in duplex stainless steels is limited to levels of 0.03% to minimize the formation of carbides and/or carbonitrides that could make the material susceptible to intergranular corrosion. The negative effect is that by decreasing the carbon content, which is a strong austenite stabilizer, another austenite stabilizing element must be added to maintain an austenitic microstructure. Thus, the composition of the older standard austenitic stainless steels has changed from 18% chromium and 8% nickel to 18% chromium and 9 – 10% nickel [12].

8

Silicon – SiSilicon is a ferrite stabilizing element. Silicon is common in high-temperature and special grade steels, improving the hardenability and tensile strength.

Titanium – Ti and Niobium – NbTitanium and niobium are stabilizing ferrite and are used when it is difficult to maintain the carbon content at low levels. These two alloying elements have higher affinity to bind carbon than chromium, hence prevent the formation of chromium carbides that could cause intergranular corrosion. As stainless steels normally have low carbon content, titanium and niobium additions are primarily used in high-temperature steels with higher carbon contents.

2.2 CorrosionCorrosion is defined as dissolution of a metal or metal alloy due to the reaction with the surrounding environment. Most metals are in a thermodynamically unstable form, and corrosion means that the same amount of energy needed to extract these metals from their minerals in nature are emitted during chemical reactions. When the material has corroded it has transformed to the more thermodynamically stable form, and corrosion is thus the result of the driving force when nature struggles towards thermodynamic equilibrium [20]. During the corrosion process two different types of reactions will take place at the same time; the anodic and the cathodic [20].

The anodic reaction (1) occurs when the metal oxidise and electrons are going into the electrolyte (conductive medium):

(1)

A cathodic reaction can either occur as oxygen reduction in acidic solutions (2) or hydrogen evolution (3):

(2)

(3)

For neutral or basic solutions oxygen reduction in form of Reaction (4) is the most common:

(4)

Corrosion of stainless steels can occur in many different forms depending on the environment, the metallurgical properties of the steel and the local stress. This result in different types of corrosion attacks, the most common forms being:

Uniform corrosion Galvanic corrosion Localised corrosion Pitting corrosion Crevice corrosion

9

Stress corrosion cracking Corrosion fatigue Intergranular corrosion

To avoid corrosion attack of the steel, designers and engineers have to be aware of the service environment to be able to choose the right type of stainless steel for use in corrosive conditions. The pitting resistance equivalent number (PRE) [21, 22] in Equation (5) is generally accepted as a rough estimation of the corrosion resistance of duplex grades.

PRE = Cr + 3.3 × Mo + 16 × N (%) (5)

For austenitic stainless steels, the effect of nitrogen is even attributed to 30 × N [23].

2.2.1 Uniform corrosionUniform corrosion is characterised by an even corrosion of a surface that is exposed to a corrosive environment, Figure 5. Therefore the corrosion rate is often measured in terms of reduction of thickness per unit time, often expressed in millimetres per year (mm/y). Uniform corrosion of stainless steels mainly occurs in acid or hot alkaline solutions. The corrosion resistance for stainless steels in oxidising acids such as nitric acid is generally good. However, in non-oxidising acids the maintenance of the passive film is not always possible. Hydrochloric and hydrofluoric acids are very aggressive environments for most types of stainless steels, thereby limiting the service environments to low temperatures and concentrations [24].

Figure 5. Uniform corrosion on the outside of a steam tube that has been exposed to sulphuric acid [24].

The resistance to uniform corrosion can be improved by addition of chromium, nickel and molybdenum. One difference is that molybdenum containing stainless steels show less resistance to uniform corrosion in strongly oxidising environments such as warm concentrated nitric acid, compared with steels without molybdenum.

Uniform corrosion results in decomposition of the steel at an even rate compared to other types of corrosion, which is why uniform corrosion is relatively easy to control. Failure caused by uniform corrosion is often due to improper selection of the type of steel grade for the service environment [24].

10

2.2.2 Galvanic corrosionGalvanic corrosion takes place when two dissimilar materials are connected electrically to each other e.g. in a conductive liquid, electrolyte, Figure 6. This is often called a galvanic cell. In a galvanic corrosion process, the more noble metal, the cathode, will be protected against corrosion while the less noble, the anode, is attacked. To classify different types of conducting materials a series called the galvanic series can be utilised. The galvanic series are valid for a special environment, often seawater at certain temperatures. Some factors that will affect the galvanic corrosion are e.g. the difference in nobility between the two metals, the surface area ratio of the two metals and the conductivity of the electrolyte [24].

2.2.3 Localised corrosionIn contrast to uniform corrosion, where all or large sections of the passive layer breaks down, localised corrosion involves a local breakdown of the passive layer. Many different forms of localised corrosion can take place in stainless steels. The most common forms are pitting corrosion, crevice corrosion, stress corrosion cracking, corrosion fatigue and intergranular corrosion.

Pitting corrosion

Pitting corrosion often appears in acidic or neutral chloride solutions. A local pitting attack is initiated when chloride ions penetrates the protecting passive film. When the passive film is locally destroyed, the base metal under it will no longer be protected. This can be considered a galvanic cell, where the small area of base metal can be considered as an anode and the large area of the remaining passive film acts as a cathode. At the anode an oxidation reaction takes place, while a reduction reaction occurs at the cathode. The effect of having a small anode area compared to a much bigger cathode area is rapid corrosion of the base metal, hence creating a pit [24]. Pits often appear to be small at the surface, but it is common that they have a rather large cross-section area deep inside the metal, Figure 7.

Figure 6. Galvanic corrosion on carbon steel (right) that has been welded to stainless steel (left) and exposed to seawater [24].

11

Figure 7. Pitting corrosion in a tube for cooling water [25].

The reactions that take place between the dissolved metal ions at the anode and the electrolyte results in decreased pH in the pit, Figure 8. In order to still have an electroneutral balance, the negative charged chloride ions migrate into the pit to balance the positive charge of the metal ions. This will further increase the already aggressive environment, and this process is accelerating with increasing size of the pit. The passivation of the metal is thus obstructed and the result is often small and deep pits growing at a high rate. The high propagation rate makes this form of corrosion dangerous, and the risk of pitting corrosion increases with increased chloride concentration, increased temperature and decreased pH [24].

Figure 8. Reaction mechanisms for pitting corrosion [26].

Crevice corrosion

Crevice corrosion is as pitting corrosion a form of localised corrosion that occurs under same conditions, in acidic or neutral chloride solutions. The main difference is that this form of corrosion occurs in crevices e.g. in flange joints, instead of an open surface. An example of crevice corrosion is shown in Figure 9.

12

Figure 9. Crevice corrosion under a valve flange in a pipe that carried heated seawater [24].

The reactions inside and outside the crevice are initially similar, but the transportation of oxygen into the crevice is limited, Figure 10. The reduction of oxygen will thus cease after some time while the oxidation of the metal continues both outside and inside the crevice. An excess of metal ions with positive charge in the crevice attracts the negative charged chloride ions outside the crevice, and in order to maintain electroneutral balance the chloride ions will migrate into the crevice. As for pitting corrosion this process increases the chloride concentration and decreases the pH in the crevice, which gives an even more aggressive environment and the passive film breaks down. Initiation of crevice corrosion is easier than pitting corrosion due to the more rapid process of changing the environment in the crevice [24].

O2OH-O2

OH-

Cl-

Me2+

Me2+OH-O2 Cl-

Cl-Cl-

H+

H+

Me2+O2OH-O2

OH-

Cl-

Me2+

Me2+OH-O2 Cl-

Cl-Cl-

H+

H+

Me2+

Figure 10. Reaction mechanisms for crevice corrosion [25].

Stress corrosion cracking

The combination of tensile stress and a corrosive environment can result in stress corrosion cracking (SCC) causing brittle failure of the material. SCC mainly occurs in chloride containing solutions at temperatures exceeding 50°C. The attacks have a crack like

13

appearance propagating as branches through the material, Figure 11. In contrast to pitting and crevice corrosion, SCC often requires lower concentrations of chlorides and oxidising substances. If the chloride solution contains hydrogen sulphide (H2S), which exists in e.g. oil and gas wells, the risk of cracking increases. Another type of environment that is known to cause SCC is strong alkaline solutions at high temperatures, pH over 14 and a temperature above 120°C [24].

Figure 11. Micrograph showing typical stress corrosion cracking [24].

Corrosion fatigue

Corrosion fatigue causes brittle failure of the material, Figure 12. It occurs when exposed to a cyclic load in a corrosive environment. Cyclic loads can cause fatigue below the ultimate tensile strength, but when a corrosive environment is present failure may occur at even lower loads. This form of corrosion mainly takes place at ambient temperature and in solutions that can be considered relatively harmless compared to other forms of corrosion. Residual stresses from production can have a negative affect on the resistance to corrosion fatigue, but can be reduced by heat treatment. To decrease the risk of corrosion fatigue for susceptible applications, a material with high mechanical strength should be used such as duplex instead of austenitic stainless steels [24].

Figure 12. Corrosion fatigue cracks in a paper machine suction roll shell, made of austenitic 1.4432 (ASTM 316L) [24].

14

Intergranular corrosion

Intergranular corrosion might occur as a result of chromium carbide (Cr23C6) or intermetallic phase precipitation in stainless steels. Due to the high chromium content, chromium carbide precipitation makes the areas adjacent to the grain boundary depleted in chromium. The chromium-depleted areas along the grain boundaries can cause corrosion if the material is exposed to a corrosive environment, Figure 13. Intergranular corrosion occurs mainly in non-reducing acids and strong oxidising conditions are often needed to initiate an intergranular attack. The resistance to intergranular corrosion can be reduced by for example solution annealing, low carbon content and alloying (stabilising) with titanium or niobium. Modern stainless steels have low carbon contents making intergranular corrosion comparably rare [24].

Figure 13. Intergranular corrosion caused by carburisation inside of a tube that carried dilute sulphuric acid [24].

2.3 Corrosion testingCorrosion testing can be used to predict a typical service lifetime for a steel grade, which depends on many factors such as environment and temperature. Laboratory experiments do not perfectly correspond to the corrosion performance in real-life, but provide good prediction of the corrosion behaviour for many applications. Corrosion testing estimates the reduction of corrosion resistance when the material properties have been changed compared with the base material. This can often be the case when the material is joined together with welding techniques.

In this work, the purpose of the corrosion tests is to rank the three materials in certain specified conditions. Ranking using CPT is often used, but in this case the ASTM G61– 86 method was needed. 304 returns invalid (too low) results [unpublished work, E. M. Westin] when using the ASTM G150 method, so the ranking was instead based on the pitting potential to be able to compare 304 with LDX 2101 and 316L. This also makes it possible to study the materials in as-welded condition. All tests were carried out at 35°C to ensure that the measurements were above the CPT for all grades and especially 316L, to facilitate comparison of 316L with LDX 2101 and 304. Both pitting potentials and repassivation potentials are obtained from the ASTM G61 – 86 measurement (Figure 14). All corrosion measurements in this work are based on the ASTM G61 – 86 standard, but the scanning direction is not reversed as in Figure 14. The measurement is instead stopped when reaching a

15

current of approximately 5 mA. This type of procedure saves time, and only the pitting potentials are of interest in this work. Another reason is that Turner et al. [27] performed corrosion tests on 316L and 304L and showed that when performing cathodic polarization experiments the weld oxide was removed by reduction, thus resulting in lower reproducibility. Therefore, the polarization experiments were limited to anodic polarization to obtain better reproducibility of the experiments. For that reason the measurements in this work is also limited to anodic polarization. The main purpose was to use the Multicell for all corrosion tests, which have shown to provide reliable and rapid results for estimating the pitting potentials from flat surfaces without forming crevice corrosion [28]. The Multicell measurements are also based on ASTM G61 – 86.

When the ASTM G61 – 86 standard is followed the result of a measurement will be a polarization curve with a different appearance than in Figure 15. This can clearly be seen in Figure 14, measurements carried out by R. Qvarfort [29], where the ASTM G61 – 86 has been followed. After reaching the pitting potential, EP, the current will increase to around 1 to 50 mA and then the scanning direction will be reversed, resulting in a loop that gives the repassivation potential (ERP). The polarization curve will have different appearance depending on the temperature where the measurement has been carried out, above or under the critical pitting temperature (CPT), Figure 14 (a) and (b), respectively. The observed appearance in curve (b) is due to transpassive corrosion [29]. CPT is defined as the lowest test temperature at which pitting corrosion occurs in a specific environment [24].

EPERP

EPERP

Figure 14. Polarization curves of a passive metal obtained in 1 M NaCl, at temperature (a) above the CPT (b) below the CPT [29].

2.3.1 Polarization curvesWhen connecting a metal surface to an external power source it will be polarized and there are many different types of correlations between current and potential. When measuring a polarized metal it is possible to plot a current-potential curve, more known as a polarization curve [30]. When performing a polarization measurement the tested metal is used as working electrode, which is further connected to a counter electrode and a reference electrode that are in connection with an electrolyte. A potentiostat is used to maintain a constant potential

16

between the work and the reference electrode, resulting in a measurable current flow. These measurements are often carried out in an electrochemical cell, in this case the Avesta Cell, which will be described later. There are many different types of measuring techniques available, but in this work, only potentiodynamic and potentiostatic methods were used.

Potentiodynamic polarization curves are characterized by measurements with a linear change of potential with time. A value of 10 mV/s is often used according to ASTM G61 – 86 and the scanning direction is reversed when a current of 5 mA is reached [31]. In potentiodynamic measurements, parameters such as scan rate and potential step can be set. The drawback is that this type of testing is rather time-consuming. Potentiostatic polarization curves on the other hand are obtained by maintaining the applied potential at specified discrete steps until the current has become constant, or reaches the defined maximum time value. When a constant current or the maximum time is reached, the potential will change to a higher potential defined by the user from the start. For the potentiostatic measurement both, potential step and the maximum time at each potential can be set, but also the sensitivity that is defining the change in current to reveal whether the current is constant or not.

A schematic polarization curve for a passive metal is illustrated in Figure 15, showing an active passive behaviour in an acidic environment. The corrosion potential, ECorr is the potential that a metal generates between itself and the counter electrode in the cell, which is caused by the difference in nobility between the two metals. For example platinum can be considered noble and less susceptible to corrosion compared to zinc. The corrosion potential is also known as open circuit potential (OCP), often used in software for polarization measurement e.g. general purpose electrochemical system (GPES). Platinum has a potential of approximately 0.2 mVSCE and zinc –1.0 mVSCE in seawater [20]. Where SCE stands for saturated calomel electrode. The current at ECorr is the same for both the anode and the cathode, but with different signs. With increasing potential, the active area is reached where the anodic process will dominate and the metal will be actively dissolved, Reaction (1). The passivation process takes place at the passivation potential EPP. At this point the corrosion rate of the metal is reduced 103 to 106 times below the rate in the active area [20]. This passivity is caused by the formation of a thin protective corrosion product surface film consisting mainly of hydrated oxide. This film acts as a barrier to the anodic dissolution, Reaction (1). At the transpassive potential, ETR, the current will increase caused by either oxygen evolution (Reaction (2) or (4), depending on the pH) and/or oxidation of the metal to higher valence states, without passive film formation. When the electrolyte is acidic, often in H2SO4 of different molarities, the measured curve will follow the schematic in Figure 15 from the cathodic to the transpassive area. However, if the electrolyte is containing aggressive halides such as chloride and bromide ions, the current can rapidly increase before the transpassive area, still in the passive area, resulting in pitting attack. This potential is called the pitting potential EP.

17

Figure 15. Schematic polarization curve of a passive metal [32].

Figure 16 shows a potentiostatic polarization curve for as-received 316L, which has a typical appearance for a passive metal in sodium chloride.

316L As received 1 M NaCl 35°C

Potentiostatic

-0.050 0 0.050 0.100 0.150 0.200 0.250 0.300 0.350-91.000x10

-81.000x10

-71.000x10

-61.000x10

-51.000x10

-41.000x10

-31.000x10

-21.000x10

-11.000x10

01.000x10

Figure 16. Passive metal in 1 M NaCl at 35°C.

IR drop has not been taken into consideration when performing measurements in the Avesta Cell, as the resistance of the used electrolyte was approximately 20 ohm and at 100 A/cm2 the resulting IR drop will be approximately 2 mV [32]. IR drop was not compensated in the Avesta Multicell for the same reason.

18

Potential, VSCE

Cu

rren

t d

ensi

ty [

A/c

m2 ]

316L As-received 1 M NaCl at 35°C

Potentiostatic

2.3.2 The Avesta CellThe Avesta Cell provides the possibility to study pitting corrosion without forming crevice corrosion. This is achieved by pumping distilled water, usually 2 – 6 ml/h, to replace the chloride solution in the microcrevice between the sample and specimen holder, Figure 17 [33].

Figure 17. Schematic picture of the Avesta Cell design [32].

The cell used in this work is relatively small, containing approximately one decilitre of electrolyte solution. A larger version is available containing approximately one litre of electrolyte used for larger samples. The temperature of the electrolyte is maintained at the set temperature, in this case 35°C, by using the Haake Phoenix II thermostat that circulates thermostated liquid in the cell wall. As can bee seen in Figure 17, the gas inlet, the counter electrode, the thermometer and the luggin capillary are inserted from the top of the cell. No extra mixing bar is needed when using a small Avesta Cell at a fix temperature. The thermometer is of type PT 100 and the reference electrode is a saturated calomel electrode (SCE). The reference electrode is in connection with the sample via a salt bridge. During the measurement N2 gas is flushed into the electrolyte to provide stirring and to maintain a low oxygen level in the electrolyte. The sample is mounted to the bottom of the cell using an o-ring and two filter papers to obtain tight connection between the sample and the cell to prevent that electrolyte can percolate. The filter paper is used to distribute the distilled water evenly on the sample and thus prevent crevice corrosion.

19

All samples that were corrosion tested in the Avesta Cell or the Multicell were of the size 30×30 mm (Figure 18) to have a test area of 1 cm2, according to the ASTM G61 – 86 standard.

Figure 18. Schematic illustration of sample dimensions used in the Avesta Cell and the Multicell.

2.3.3 The MulticellThe Avesta Multicell used in this work was developed by C.-O.A. Olsson [34]. The cell design is similar to the Avesta Cell, but instead of one sample, six samples can be measured simultaneously. In the Multicell approximately one litre of electrolyte is used for one measurement. In similarity with the Avesta Cell, a pump is used to provide the samples with distilled water flow of approximately 6 ml/h per channel, in this case a six-channel Alitea 403U/C6 pump. A schematic representation of the electrical connections of the Multicell is illustrated in Figure 19. The connections are similar to the ones in the Avesta Cell, but with the difference that all samples in the Multicell are connected to a box before the computer and the potentiostat.

20

Figure 19. Schematic illustration of electrical connections of the Multicell [32].

The box contains a shunt resistor and a switch, where the shunt resistor is a low resistance precision resistor used to measure electrical currents caused by voltage drops. For all channels a maximum current is set and when that value is reached the switch disconnects the connection to the sample. The value of the shunt resistor is adapted to provide a correct estimation of the current just below 100 A, which is in the region where the pitting potential is evaluated in the Multicell. When performing a measurement with the Multicell the resulting data is not representing a polarization curve as for the Avesta Cell. Currents measured under the range of the shunt resistor are only detected as noise [32]. As for the Avesta Cell nitrogen gas is flushed into the electrolyte and the samples are connected to the cell using o-rings and filter paper. The same type of thermostat as for the Avesta Cell is used, Haake Phoenix II, maintaining a temperature of 35°C in the electrolyte (1 M NaCl) using a PT 100 thermometer. The reference electrode used in this cell was a saturated Ag/AgCl type.

2.4 TIG WeldingAll welding methods used for construction steel can also be used for stainless steel. Tungsten inert gas (TIG) or gas tungsten arc welding (GTAW) is characterised by high quality weld metal deposits, great precision, superior surface homogeneity and high strength. In TIG welding an electric arc is created between the tungsten electrode and the workpiece, where the arc is melting the parent metal. Filler material can be added from the side if needed, Figure 20. TIG welding can be performed either manually or automatically with or without filler material, Figure 21.

Figure 20. The principle of TIG welding [10].

21

Figure 21. Schematic overview of the TIG welding process [35].

The tungsten electrode can vary between 1.0 to 4.8 mm in diameter and be made of pure tungsten or a tungsten alloy (1 – 2% thorium, zirconium or cerium oxides), where the tungsten-thorium alloy is the most common type of electrode. The angle of the electrode is an important parameter, which has a major effect on the penetration of the material. As can be seen from Figure 22, small angles (15 – 30) result in a wide arc with low penetration and wider angles (60 – 75) result in a narrower arc, thus giving deeper penetration. The effect of different polarities can be seen in Figure 23. In this work, the direct-current electrode negative (DC-) type is used [10].

Figure 22 The electrode angle and its effect on penetration [10].

Figure 23. Effect of different polarities on weld penetration [35].

The productivity of the TIG welding method is relatively low both if the amount of melted weld metal is determined per unit time or in terms of welding speed. Preheating of the TIG wire (hot wire TIG) can increase the productivity. TIG welding is favourable in tube joints and for root beads and is often used when a high requirement on the surface appearance is needed. TIG welding is common in the nuclear and process industry where good weld quality is required, but also for thin material in general and one-side welding of tubes. Another possibility with this method is the ability to re-melt eventual geometrical deviations in the

22

junction between weld and base metal and therefore improve the fatigue properties of the weldment. It is very important for optimum corrosion resistance and mechanical properties to use a suitable filler material when welding high alloy stainless steels such as 2205 and 254 SMO®. It is also vital to use shielding and purging gas to achieve optimal corrosion properties after welding.

2.5 Weld oxide formation The oxide formation that occurs during welding is a complex process to predict and hard to describe in detail. The basic concept of oxide formation is that alloying elements with the highest affinity to oxygen (most negative Gibbs free energy of formation) preferentially reacts and forms an oxide containing the alloy element and oxygen. According to Pettersson and Flyg [1] weld oxides forms extremely fast and will not reach equilibrium conditions, which means that kinetics will dominate the process. A chromium depletion is frequently reported to be present at the surface of weld oxides and if the diffusion rate of chromium to the outer surface is too low then iron or other elements with affinity to oxygen will tend to oxidise, often resulting in lower corrosion resistance. Gibbs free energy of formation for the most common reactions can be obtained using the Ellingham diagram [36], but the reactions are based on equilibrium reactions, which do not correspond to the reactions formed in weld oxides. It can be determined what element has the highest affinity to oxygen from the Ellingham diagram, but this is for the most common reactions only. The weld oxide can have different stoichiometric formulas depending on what is the most thermodynamic stable oxide composition. Pettersson and Flyg [1] found iron oxide on the topside of TIG welded 316L stainless steel at approximately 3 mm from the FL. When studying TIG welded 316L stainless steel an obvious difference in oxide thickness between the top and root side could be seen, 200 nm compared to 30 nm for the root side. This was explained by the controlled level of oxygen at the root surface, which will slow down the oxidation kinetics and thus result in a thinner oxide. Furthermore, no iron oxide layer could be identified on the root side, meaning that the diffusion of chromium was sufficiently high for maintaining and causing growth of the chromium oxide.

Von Moltke et al. [2] reported that a chromium-rich oxide layer is present under the iron-rich layer and means that the oxidation temperature influences the diffusion rate of iron and chromium. Chromium reduction in the weld oxide at the highest temperatures was related to evaporation of chromium oxide, CrO3 [2]. Furthermore, von Moltke et al. [2] concluded that the iron-rich oxides are most harmful and that chromium-rich oxides are more resistant to corrosion. Von Moltke et al. [2] also suggest that the iron-rich oxide acts as an ion-selective membrane that absorbs chlorides, but also prevents them from leaving by diffusion. This results in increased chloride concentration followed by pit initiation.

A schematic description of mechanisms occurring during welding has been suggested [4] for heat tint formation on LDX 2101 and is illustrated in Figure 24. Weld oxide formation is often described by oxidation of the base metal and diffusion from the weld metal [4]. Westin et al. [4] showed that manganese is evaporated from the weld metal and then forms weld oxide by subsequent redeposition from the gas phase. Furthermore, nitrogen addition to the shielding gas showed enhanced evaporation from the weld metal caused by the higher temperatures that were attained, which resulted in a thicker weld oxide that also contained oxynitrides. Another suggestion from Westin et al. [4] was that the weld oxide might also absorb nitrogen from the shielding gas. They also found an oxide at 14 mm from the fusion line where the maximum

23

temperature was 200C. This oxide was suggested to have formed by redeposition and not due to oxidation, because of the relatively low temperature.

Figure 24. Schematic description of mechanism occurring during welding of LDX 2101 [4].

Figure 25 shows an illustration of oxides that forms when welding LDX 2101 depending on which shielding gas that has been used, suggested by Westin et al. [4]. LDX 2101 welded with pure argon has a chromium and nitrogen-rich manganese oxide. A double layer oxide is formed on LDX 2101 when nitrogen is added to the shielding gas, which most probably contains manganese oxynitrides. The manganese fraction and the thickness of the weld oxide became thicker when Ar + 2% N2 shielding gas was used, which was suggested to be related to the absorption-evaporation-deposition process.

Figure 25. Schematic illustration of oxides formed on LDX 2101 with and without nitrogen addition to the shielding gas [4].

2.6 Post-weld cleaningPost-weld cleaning is important to attain the desired aesthetic appearance, hygienic and corrosion requirements of stainless steel after the welding process. A clean, smooth and faultless surface is normally required to achieve high corrosion resistance. Many different types of cleaning methods, both mechanical and/or chemical, can be utilized depending on the requirements of the corrosion resistance for the application in the specific environment. The most common cleaning processes are grinding, blasting, brushing, electropolishing, pickling and combinations of mechanical and chemical methods. Before using any of these methods a

24

careful cleaning with a suitable degreasing agent should be used to remove any organic contaminants from the steel surface.

Grinding is suitable for removing weld defects and deep scratches on the surface by using a grinding disc. For minor surface defects a flapper wheel is often sufficient. If coarse grinding is needed (40 – 60 mesh), then a subsequent finer grinding (180 mesh) should be performed [10]. Polishing can be performed (220 mesh or higher) after the grinding process when even higher demands are set on the material. The chromium-depleted zone beneath the weld oxide can be removed if proper grinding is performed [10].

Polishing can sometimes be sufficient for cleaning thin oxides e.g. TIG weld oxides. The chromium-depleted zone is not removed when this method is used, but oxides and dirt that cause localized corrosion are removed. Scotchbrite or some type of polishing cloth is often used.

Sand and grit blasting, is mostly used for cleaning of large surfaces. The blasting material must be clean to obtain good results when using this method. Common blasting materials are olivine sand and smooth glass beads. Also blasting should be carried out with caution to minimize the coarseness of the surface. Performing sand blasting without a followed cleaning procedure with other methods, results in the lowest corrosion resistance compared with most other post-weld cleaning methods.

Brushing can often be sufficient for removing weld oxide and maintaining a relatively good corrosion resistance. Stainless steel or nylon brushes are often used. This method does not result in excessively coarse surfaces, but with no guarantee for complete removal of the chromium-depleted zone. However, the risk of contamination is relatively high and therefore use of clean tools is important.

Pickling is the best method to remove oxides and iron-contamination, and thus to restore the corrosion resistance to a level close to that of the base metal. When pickling, a mixture of approximately 8 – 20% nitric acid and 0.5 – 5% hydrofluoric acid is often used. Rinsing with clean water is important after the pickling process. Chloride containing agents such as hydrochloric acid should be avoided, since these can increase the risk of pitting corrosion. Different pickling methods can be utilized depending on the requirements and size/geometry of the material. Some of the most common types are; pickling in a bath (20 – 65 C), pickling with pickling paste (10 – 40C), pickling with pickling solution (gels) and electropolishing. Pickling in bath is used when such equipment is available and when the material match the size of the bath. Pickling paste is used when pickling a material that cannot be moved, when no pickling bath or when no other equipment is available. When pickling solution is utilized a gel like solution is sprayed on the material, as for blasting this method is favourable when pickling large surfaces. Electropolishing can give optimal corrosion resistance due to controlled corrosion of the material surface and may also improve the surface finish [10]. Shot blasting before using pickling paste or pickling solutions sprayed on the material can be beneficial since it reduces the needed pickling time.

2.6 GDOESThe glow discharge optical emission spectroscopy (GDOES) instrument can be used to for bulk, surface and depth profiling of both conducting and non-conducting materials e.g. weld oxides [37]. The instrument is based on argon glow discharge, which is a low-pressure plasma

25

that is electrically neutral, containing negatively charged electrons and positively charged argon ions, Figure 26. Low pressure of argon is filled in the cavity between the anode and cathode to initiate the plasma. At room temperature and with no applied potential between the anode and cathode, the gas acts as an isolating medium. If applying a sufficiently high potential between the anode and cathode, a current can flow creating a low-pressure plasma. Electrons flow from the electrode with a negative potential, in this case the cathode, to the electrode with a positive potential. The electrons are colliding with argon atoms to form positive argon ions that are attracted to the negative electrode (sample). Positive argon ions with sufficiently high energy will knock off atoms, but also secondary electrons, when they hit the sample surface. The secondary electrons are attracted to the positive electrode (anode) and some of them will collide with argon atoms on their way to the anode, thus maintaining the discharge. However, the sputtered cathode atoms will collide with high-energy electrons or metastable argon ions to form exited atoms. Those excited atoms are in a metastable state so they will de-excite. When the atoms de-excite they will emit photons with a characteristic wavelength that can be detected with the optical spectrometer to measure the composition of the sample. All elements are analysed simultaneously during the measurement. Radio frequency (RF) is used, which alternates the potential at the anode and cathode, in order to avoid charge accumulation in the cavity and thus allowing analysis of both conductive and non-conductive materials [38]. In this work, GDOES results can give information on oxide composition, thus provide better understanding of the corrosion test results.

Ar

Vacuum

Window

Cooling

Copper anode

Cathode (sample)RF

Photons

Cathode atoms in fundamental state

Positive ionised argon atoms

Cathode atoms in excited state

Electrons

Ar

Vacuum

Window

Cooling

Copper anode

Cathode (sample)RF

Photons

Cathode atoms in fundamental state

Positive ionised argon atoms

Cathode atoms in excited state

Electrons

Photons

Cathode atoms in fundamental state

Positive ionised argon atoms

Cathode atoms in excited state

Electrons

Figure 26. Schematic illustration of the glow discharge process.

It is important that the sample surface is flat to be able to obtain a vacuum before the measurement starts. If measurements are carried out on a non-flat surface, a special sample holder can be used to obtain valid results, Figure 27.

26

Figure 27. Picture of the special sample holder for non-flat surfaces.

27

3 ExperimentalAll polished, ground and/or pickled samples were left in laboratory atmosphere for at least 24 hours before measurements.

3.1 MaterialThree different stainless steel grades were used in this work; two austenitic (304 and 316L) and one lean duplex (LDX 2101®). Their chemical compositions and properties are summarised in Table 1 and 2, respectively. The 304 and 316L grade were chosen as comparison to LDX 2101.

Table 1. Chemical composition and PRE number for 304, 316L and LDX 2101.

Name EN C Si Mn P S Cr Ni Mo Cu N PRE*

LDX 2101 1.4162 0.03 0.62 5.0 0.02 0.001 21.5 1.58 0.2 0.25 0.20 25.4

304 1.4301 0.05 0.45 1.5 0.03 0.005 18.2 8.02 0.2 0.46 0.05 19.7

316L 1.4404 0.02 0.36 1.8 0.03 0.002 17.5 10.06 2.1 0.41 0.05 25.2

*PRE=pitting resistance equivalent, calculated with Equation 5.

Table 2. Mechanical properties of grades 304, 316L and LDX 2101.

Name EN Rp0.2

[Mpa]Rm

[Mpa]A5

[%]Mill finish Thickness

[mm]

LDX 2101 1.4162 591 787 40 2E (brushed) 2.2

304 1.4301 309 640 56 2B* 3.0

316L 1.4404 376 636 53 2B* 3.0

*2B=Cold rolled, heat-treated, pickled, skin passed.

All material was delivered by Nyby. All samples that were corrosion tested in the Avesta Cell and/or the Multicell were cut in 30×30 mm. The material used for characterisation of the heat tint in GDOES were initially ground and polished to 3 m mirror finish on the side that will be analysed.

All samples were carefully cleaned with acetone and methanol before welding, corrosion testing and GDOES measurement.

3.2 WeldingAll stainless steel grades were mechanically TIG welded bed-on-plate (BOP) with a modified EWM TIG 450 DC-P using a tungsten-thorium electrode with an angle of 60° and a diameter of Ø1.6 mm. The arc and electrode stick-out lengths were 3.0 and 7.0 mm, respectively, and the gas cup inner diameter was Ø12 mm. In this work no filler material was used during the welding process. Three different types of shielding gases were used with a gas flow of 15 l/min; Ar or 90% N2 + 10% H2 was used as purging gas and Ar or Ar + 2% N2 as shielding gas, Table 3.

Table 3. Welding parameters, HI=heat input=U×I/v.

28

Name Analyzed side

Shielding gas

Purging gas U [V]

I [A] V [mm/s]

HI [kJ/mm]

Thickness[mm]

304 Cap Ar Ar 14.3 163 6.7 0.3 3.0304 Root Ar Ar 14.3 163 6.7 0.3 3.0304 Root Ar none 14.3 163 6.7 0.3 3.0304* Cap Ar Ar 16.1 163 10.0 0.3 3.0304* Root Ar Ar 15.5 163 10.0 0.3 3.0304* Root Ar none 15.7 163 10.1 0.3 3.0316L Cap Ar Ar 14.6 163 7.0 0.3 3.0316L Root Ar Ar 14.6 163 7.0 0.3 3.0316L Root Ar none 14.6 163 7.0 0.3 3.0316L* Cap Ar Ar 15.7 163 10.0 0.3 3.0316L* Root Ar Ar 15.6 163 10.1 0.3 3.0316L* Root Ar none 15.5 163 10.1 0.2 3.0

LDX 2101 Cap Ar Ar 14.8 163 9.2 0.3 2.2LDX 2101 Root Ar Ar 14.8 163 9.2 0.3 2.2LDX 2101 Root Ar none 14.8 163 9.2 0.3 2.2LDX 2101 Cap Ar+2% N2 90% N2+10% H2 14.9 163 10.3 0.2 2.2LDX 2101 Root Ar+2% N2 90% N2+10% H2 14.9 163 10.3 0.2 2.2LDX 2101* Cap Ar Ar 15.3 163 10.1 0.2 2.2LDX 2101* Root Ar Ar 15.8 163 10.1 0.3 2.2LDX 2101* Root Ar none 15.4 163 10.1 0.2 2.2LDX 2101* Cap Ar+2% N2 Ar 15.6 163 10.1 0.3 2.2LDX 2101* Root Ar 90% N2+10% H2 15.4 163 10.1 0.2 2.2LDX 2101* Root Ar+2% N2 none 15.9 163 10.1 0.3 2.2* GDOES specimens polished to 3 m mirror finish before welding.

Cross-sections of the LDX 2101 welds were ground and polished with SiO2 as a final step. The welds were then etched and the ferrite content was measured with light optical image analysis.

3.3 Post weld cleaningBoth welded samples and the base metal were pickled. Different pickling times were chosen, depending on the cleanness of the welds. The samples were pickled until they looked visually clean. LDX 2101 required the longest pickling time, Table 4.

Table 4. Pickling time for grades 304, 316L and LDX 2101 in 3.4M HNO3 + 3M HF at 60°C.

Name EN Pickling time, base metal [minutes]

Pickling time, welded material [minutes]

LDX 2101 1.4162 0.5 19.0 – 21.0

304 1.4301 0.5 2.0

316L 1.4404 0.5 2.5

Polishing on the welded samples and the base metal was performed using a Flex LP 1503 VR electrical machine with a rotation speed of 2700 rpm and an Schleif-Vlies polishing wheel with K80 coarseness, Table 5. The welded samples were polished until they looked visually clean.

Table 5. Polishing with K80 coarseness on 304, 316L and LDX 2101 with a rotation speed of 2700 rpm.

Name Analyzed Shielding gas Purging gas Times* Thickness

29

side [mm]304 --- --- --- 40 3.0304 Cap Ar Ar 50 3.0304 Root Ar Ar 30 3.0304 Root Ar none 30 3.0316L --- --- --- 40 3.0316L Cap Ar Ar 60 3.0316L Root Ar Ar 30 3.0316L Root Ar none 30 3.0LDX 2101

--- --- --- 40 2.2

LDX 2101

Cap Ar Ar 100 2.2

LDX 2101

Root Ar Ar 30 2.2

LDX 2101

Root Ar none 40 2.2

LDX 2101

Cap Ar+2% N2 90% N2+10% H2 80 2.2

LDX 2101

Root Ar+2% N2 90% N2+10% H2 30 2.2

* Back and forth, without applied external force (until no coloured heat tint could be seen).

3.4 Corrosion testingIn this work base metal and welded samples were corrosion tested in five different conditions; as-welded, as-received, polished, wet ground to 320 mesh and pickled. All corrosion tests were performed in 1 M NaCl at 35C. The corrosion tests were mainly performed using the Multicell, but complementary measurements was also carried out using the Avesta Cell.

3.4.1 Avesta CellThe cell is connected to the Autolab PGSTAT30 potentiostat and a computer using the GPES interface. Standard values of parameters used for potentiodynamic measurements in the GPES are; step potential to 2.44 mV, scan rate to 0.1667 mV/s, start scan at OCP, wait 10 minutes for OCP and current boundaries disabled (no change in scan direction after pitting corrosion occurs, Figure 15). The temperature was set to 35°C and the measurement is not started until that temperature is reached in the electrolyte. On the other hand, when performing potentiostatic measurements the following parameters were used in the software; step potential of 25 mV, maximum time interval of 4 minutes, minimum abs(di/i) per second to zero, minimum abs(di) (A) to 1e-7. Pitting potentials are evaluated where the current density exceeds 100 A/cm2 for at least 60 seconds.

3.4.2 MulticellSolatron 1287A potentiostat was used for the Multicell experiments together with the Labview software interface. Standard set values of parameters used for Multicell measurements in the Labview software were; scan rate to 0.1667 mV/s, Measure interval to 1000 ms, Current limit to 1 mA, wait 15 minutes for OCP. In this case the OCP value is the mean value for all six samples. The temperature was set to 35°C and the measurement was not started until that temperature was reached. Pitting potentials were evaluated where the current density exceeded 100 A/cm2 for at least 60 seconds.

30

3.5 GDOESThe weld oxide composition was characterised at different depths using an RF GDOES, the GD-Profiler HR, from Horiba Jobin Yvon. Measurements were carried out using a 2 mm copper anode. The sample is attached using vacuum, and then an RF applicator together with a cooling block is positioned on the backside of the sample. Finally a mechanical arm is placed on the back of the cooling block to fasten the sample, Figure 28. The topside of the welded samples was mounted directly in the instrument, due to the relatively flat surface on the topside. The root side was too uneven to be tested with the available sample holders.

Mechanical arm Applied Radio Frequency

Cooling block

SampleMechanical arm Applied Radio Frequency

Cooling block

Sample

Figure 28. Mounting of sample, before (left) and after (right) the mechanical arm is in right position.

The flush time was 60 s, the background measurement lasted 5s, the sputter time was 70 s at an applied pressure of 650 Pa. Figure 29 illustrates how the GDOES measurements were performed on the topside of the welded samples.

31

BM

LT

HT

WM

BM

LT

HT

WM

BM

LT

HT

WM

Figure 29. Schematic illustration of how the measurements were performed on the topside where BM = base metal, LT = low temperature oxide, HT = high temperature oxide and WM = weld metal.

4 ResultsThe visual impression when comparing the cleanness of the polished and pickled samples to the as-welded condition was that the pickled appeared to be cleanest. The polished surface was shinier than the pickled one, but some darker areas could be identified 1-2 mm from the fusion line (FL), especially for LDX 2101 welded with nitrogen addition to the shielding gas.

4.1 Base metalThe base metal condition was primarily corrosion tested with the Multicell and only a few measurements were performed using the Avesta Cell for comparison. In as-received condition, LDX 2101 had the lowest pitting potential, Figure 30. The pitting potential for both LDX 2101 and 304 increased when polished with K80 coarseness, compared with the as-received condition. After polishing to 320 mesh, LDX 2101 was in parity with that of 304 but not 316L. 316L even showed somewhat lower values after polishing than in the as-received condition. An increase in pitting potential could be seen when as-delivered LDX 2101 samples were pickled. However, this was not the case for 304 and 316L where 304 returned the lowest value. 316L returned a somewhat higher value in the pickled condition compared with 320 mesh. As can be seen from Figure 26, all cleaning methods increased the pitting potential for LDX 2101, and the best was wet grinding with 320 mesh. 304 in polished condition only resulted in small increase in the pitting potential compared to the as-received condition. For 304, the pickled condition resulted in the lowest pitting potentials, and the highest was obtained in the polished condition. Detailed information on the of pitting potential results from Figure 30 can be found in appendix A, Table 1.

32

Base metal 1M NaCl at 35°C

0

50

100

150

200

250

300

350

400

450

500

304 LDX 2101 316L

Grade

Pit

tin

g P

ote

nti

al [

mV

SC

E]

As-receivedPolished (K80)320 meshPickled

Figure 30. Pitting potential of base metal in 1 M NaCl at 35°C.

4.2 TopsidePitting potential measurements using the Multicell for as-welded condition resulted in low pitting potential values and no pitting attack occurred on LDX 2101 and 304 samples. The Avesta Cell was for that reason used on LDX 2101, 304 and 316L in as-welded condition.

4.2.1 Avesta CellFigure 31 shows the potentiodynamic and potentiostatic curves of the as-welded samples of LDX 2101 using Ar + 2% N2 as shielding gas. These were rather similar to each other with an active passive behaviour. For the potentiodynamic curve (a) an initial top was present around –450 mVSCE, a smaller at approximately –90 mVSCE and finally a relatively sharp increase at the end of the curve. The potentiostatic curve (b) is smoother with an initial top around –480 mVSCE, a smaller at approximately –100 mVSCE and finally a sharp increase at the end. After the corrosion measurement, pitting attacks could be identified 1 – 2 mm from the FL where a dark oxide was present. Only a few attacks could be identified in the weld metal.

33

LDX 2101 As welded Ar + 2N2 Top 1 M NaCl 35°C

Potentiodynamic

-0.600 -0.500 -0.400 -0.300 -0.200 -0.100 0 0.100-91.000x10

-81.000x10

-71.000x10

-61.000x10

-51.000x10

-41.000x10

-31.000x10

-21.000x10

-11.000x10

01.000x10

LDX 2101 As welded Ar + 2N2 Top 1 M NaCl 35°C

Potentiostatic

-0.600 -0.500 -0.400 -0.300 -0.200 -0.100 0 0.100-91.000x10

-81.000x10

-71.000x10

-61.000x10

-51.000x10

-41.000x10

-31.000x10

-21.000x10

-11.000x10

01.000x10

Figure 31. (a) Potentiodynamic polarization curve and (b) Potentiostatic polarization curve of topside of as-welded LDX 2101 using Ar + 2% N2 as shielding gas.

LDX 2101 welded with pure argon as shielding gas showed a somewhat similar potentiodynamic curve as LDX 2101 welded with Ar + 2% N2, Figure 32a. An initial peak is present at approximately –385 mVSCE a smaller at 63 mVSCE and a slightly sharper increase after 48 mVSCE. However, the potentiostatic curve had a different appearance, showing a decreased current density in the middle of the curve at approximately –324 mVSCE. The decrease is followed by a sharp increase at –300 mVSCE, and then further decreasing to –124 mVSCE where the cureve finally increases to the end (Figure 32b). The current density was lower for samples welded with pure argon most of the time compared with samples welded with nitrogen additions. Pitting attacks mainly occurred 1 – 2 mm from the FL and almost no attacks were found in the weld metal. This was not the case for LDX 2101 welded with nitrogen additions to the shielding gas where attacks occurred both 1 – 2 mm from the FL and in the weld metal.

34

Cu

rren

t d

ensi

ty [

A/c

m2 ]

Potential, VSCE

Potential, VSCE

LDX 2101 As-welded Ar + 2N2 Top 1 M NaCl at 35°C

Potentiostatic

(a)

(b)

LDX 2101 As welded Ar Top 1 M NaCl 35°C

Potentiodynamic

-0.450-0.400-0.350-0.300-0.250-0.200-0.150-0.100-0.050 0-91.000x10

-81.000x10

-71.000x10

-61.000x10

-51.000x10

-41.000x10

-31.000x10

-21.000x10

-11.000x10

01.000x10

LDX 2101 As welded Ar Top 1 M NaCl 35°C

Potentiostatic

-0.450 -0.350 -0.250 -0.150 -0.050 0.050-91.000x10

-81.000x10

-71.000x10

-61.000x10

-51.000x10

-41.000x10

-31.000x10

-21.000x10

-11.000x10

01.000x10

Figure 32. (a) Potentiodynamic polarization curve and (b) Potentiostatic polarization curve of topside as-welded LDX 2101 performed using pure argon as shielding gas.

The results for as-welded 304 did not show the same appearance as for LDX 2101 (Figure 33). It can be noticed that no initial peak is present, but only a smooth top at approximately –20 mVSCE for the potentiodynamic curve. The potentiostatic curve also showed a smooth top, but at approximately –15 mVSCE, and no initial peak could be identified in the beginning of the curve. Pitting primarily occurred 1 – 5 mm from the FL and in the weld metal.

35

Cu

rren

t d

ensi

ty [

A/c

m2 ]

Potential, VSCE

Potential, VSCE

LDX 2101 As-welded Ar Top 1 M NaCl at 35°C

Potentiostatic

(a)

(b)

304 As welded Ar Top 1 M NaCl 35°C

Potentiodynamic

-0.250 -0.200 -0.150 -0.100 -0.050 0 0.050-91.000x10

-81.000x10

-71.000x10

-61.000x10

-51.000x10

-41.000x10

-31.000x10

-21.000x10

-11.000x10

01.000x10

304 As welded Ar Top 1 M NaCl 35°C

Potentiostatic

-0.175-0.150-0.125-0.100-0.075-0.050-0.025 0 0.025 0.050-91.000x10

-81.000x10

-71.000x10

-61.000x10

-51.000x10

-41.000x10

-31.000x10

-21.000x10

-11.000x10

01.000x10

Figure 33. (a) Potentiodynamic polarization curve and (b) Potentiostatic polarization curve of the topside of as-welded 304.

The results for the as-welded 316L samples showed a similar polarization curve as for LDX 2101, but at much lower current densities, Figure 34a. A smooth initial top is present in the beginning at approximately –99 mVSCE. The current density is then decreased until a sharp peak at 100 mVSCE and then another sharp peak at 120 mVSCE decreasing to 127 mVSCE where the current density finally increases at the end of the curve. The potentiostatic curve is similar to the curve of 304 with no initial top at the beginning but a sharp increase at the end at approximately 41 mVSCE. For 316L the pitting attacks occurred in the weld metal.

36

Cu

rren

t d

ensi

ty [

A/c

m2 ]

Potential, VSCE

Potential, VSCE

304 As-welded Ar Top 1 M NaCl at 35°C

Potentiostatic

(a)

(b)

316L As welded Ar Top 1 M NaCl 35°C

Potentiodynamic

-0.200 -0.150 -0.100 -0.050 0 0.050 0.100 0.150 0.200-91.000x10

-81.000x10

-71.000x10

-61.000x10

-51.000x10

-41.000x10

-31.000x10

-21.000x10

-11.000x10

01.000x10

316L As welded Ar Top 1 M NaCl 35°C

Potentiostatic

-0.200 -0.150 -0.100 -0.050 0 0.050 0.100 0.150-91.000x10

-81.000x10

-71.000x10

-61.000x10

-51.000x10

-41.000x10

-31.000x10

-21.000x10

-11.000x10

01.000x10

Figure 34. (a) Potentiodynamic polarization curve and (b) Potentiostatic polarization curve of the topside of as-welded 316L.

4.2.2 MulticellFigure 35 shows the pitting potential for the various surface conditions examined in the Multicell. Detailed information on the pitting potential values can be found in appendix B, Table 2. 316L reached pitting potential values over 300 mVsce, LDX 2101 between 200-300 mVSCE and 304 reach values around 200 mVSCE. LDX 2101 welded with Ar + 2% N2 shielding gas reached a somewhat higher value than samples welded with pure argon. Polishing improved the pitting potential for 316L to a level close to the pickled condition, where it was insufficient for 304 and LDX 2101. Polishing was particularly inefficient for the LDX 2101 welds performed with Ar + 2% N2 as shielding gas. Pitting potential values in as-welded condition could only be obtained for 316L.

37

Cu

rren

t d

ensi

ty [

A/c

m2 ]

Potential, VSCE

Potential, VSCE

316L As-welded Ar Top 1 M NaCl at 35°C

Potentiostatic

(a)

(b)

Topside 1M NaCl at 35°C

0

50

100

150

200

250

300

350

400

450

500

Ar Ar Ar+2% N2 Ar

Shielding gas

Pit

tin

g P

ote

nti

al [

mV

SC

E]

As-weldedPolished (K80)Pickled

Ar + 2% N2

LDX 2101 316L

304

Figure 35. Pitting potentials of topside welds in 1M NaCl at 35°C.

As can be seen from Figure 36, no pitting attacks occurred in the weld metal for the two austenitic grades in pickled condition. The opposite was observed for pickled LDX 2101 welded with pure argon as shielding gas, Figure 36c. However, when nitrogen was added to the shielding gas pitting also occurred 1 – 2 mm from the FL, 36d.

Figure 36. Pitting attacks on topside pickled samples of (a) 304, (b) 316L, (c) LDX 2101 pure argon as shielding gas and (d) LDX 2101 using Ar + 2N2 as shielding gas.

38

4.2.3 GDOESThe result from the GDOES measurements on the topside in the as-welded condition can be seen in Figure 37. A high fraction of manganese is present at the surface in weld metal, high temperature and low temperature oxide on LDX 2101 welded with pure argon and with nitrogen additions. This was not the case for 304 and 316L. The thickest oxide formed on the weld metal can be observed for LDX 2101 welded with Ar + 2% N2 shielding gas and the thinnest was formed on 316L.

The results from the high temperature oxide showed that a high fraction of manganese was present in the outer oxide layer of LDX 2101. This is especially true for LDX 2101 welded with nitrogen additions. The oxygen curve follows the manganese curve for all grades but with different fractions, where LDX 2101 welded with nitrogen additions is the most distinct example. Nickel depletion can be observed for almost in almost all cases, except in the low temperature oxide of 304 and 316L. The silicon content in the high temperature oxide at the surface of LDX 2101 welded with pure argon is almost ten times higher compared with LDX 2101 base metal, approximately 6% compared to 0.62% for the base metal. A somewhat higher chromium and nitrogen content can be observed for LDX 2101 welded with pure argon than welded with nitrogen addition to the shielding gas in the weld metal and the high temperature oxide. Chromium enrichment can be observed in the high temperature oxide for LDX 2101 and a small fraction for 316L, but not for 304. Iron depletion occurs at the surface for 316L and LDX 2101, almost none in 304. The manganese content in the high temperature oxide is extremely high for LDX 2101 welded with nitrogen additions to the shielding gas, where approximately 55% manganese can be identified at the surface. The nitrogen content is also highest for LDX 2101 welded with nitrogen additions, but 304 have almost three times higher nitrogen content at the surface. The silicon content is not very high except for LDX 2101 welded with pure argon, where a content of over 6% can be observed at the surface. The oxygen curve follows the manganese curve relatively well for LDX 2101.

An interesting observation is that a chromium depletion was present in the weld metal for 304 and 316L, but not for LDX 2101. It can also be seen that the silicon content in the weld metal is almost four times higher at the surface of 304 than in 316L.

Almost no chromium depletion occurs for the austenitic grades in the low temperature oxide. LDX 2101 welded with nitrogen additions showed the highest depletion at the surface. Manganese enrichments are observed for LDX 2101 at the low temperature oxide surface.

39

Figure 37. GDOES results in as-welded condition on the topside of the weld metal, the high temperature (HT) oxide and the low temperature (LT) oxide of 304, 316L and LDX 2101.

4.3 Root sideRegarding the root side the majority of the measured samples, both as-welded, polished and the pickled welds returned valid results. Only a few measurements on the root side were performed in the Avesta Cell. No GDOES measurements were performed due to problems with high oxygen and nitrogen signals.

The results of the Multicell root side measurements in as-welded condition clearly show the difference when welding with or without purging gas. 316L appeared to be least sensitive to residual oxide. The highest pitting potential for LDX 2101 in as-welded condition could be observed when the 90% N2 + 10% H2 purging gas was used, Figure 38. A more detailed information on the pitting potential results can be found in appendix B, Table 3.

40

Figure 38. Pitting potentials for the root side of as-welded samples.

Figure 39 shows that the pitting potentials became much higher when the welds were polished or pickled compared to the as-welded samples. One interesting observation is that LDX 2101 welded with pure argon reached higher pitting potential than 316L in the pickled condition. Furthermore, samples welded with 90% N2 + 10% H2 as purging gas reached the lowest values of all pickled root side samples. Detailed information on pitting potential results can be found in appendix B, Table 3.

Root side 1M NaCl at 35°C

0

100

200

300

400

500

600

none Ar none Ar 90% N2 +10% H2

none Ar

Purging gas

Pit

ting

Po

ten

tia

l [m

VS

CE]

As-weldedPolished (K80)Pickled

LDX 2101

304

316L

90% N2 +

10% H2

Figure 39. Pitting potential of the root side welds in 1 M NaCl at 35°C.

4.4 LDX 2101 weld metal microstructure

Figure 40 shows the weld metal microstructure at the cap surface for LDX 2101 welded with pure argon and 2% nitrogen addition to the shielding gas. LDX 2101 welded with pure argon showed somewhat less Widmanstätten austenite and more chromium nitride precipitates, seen as dark areas at the surface. These are of Cr2N type [39]. The ferrite content in weld metal was 74 3% when using pure argon as shielding gas. An average ferrite content of 71 3% was obtained when Ar + 2% N2 shielding gas was used, and less chromium nitrides precipitated at the surface.

41

(a) Welded with pure argon as shielding gas (b) Welded with Ar + 2%N2 as shielding gasFigure 40. Cross-section of LDX 2101 weld metal microstructure using different types of shielding gas.

5 DiscussionThe used corrosion testing method in this work did not classify the materials fairly in all conditions. LDX 2101 returns lower pitting potential result than expected when using the ASTM G61 – 86 test method compared to the ranking obtained using the CPT method for the same material. This is probably due to the higher surface sensitivity in the pitting potential corrosion test ASTM G61 – 86 compared with the CPT test ASTM G150. The main reason why the ASTM G61 – 86 method was used was to be able to perform corrosion measurements in the as-welded condition and also to compare the results with 304. The CPT method works well with pickled welds, as can be seen in Figure 42.

5.1 Base metalFigure 41 shows typical CPT measurements for various materials. The ranking of the base metal in this work in as-received condition according to ASTM G61 (Figure 30), did not follow the ranking obtained using the more established ASTM G150 method [24]. For the ASTM G61 method 304 showed higher pitting potentials than LDX 2101, but in the ASTM G150 LDX 2101 reaches much higher values being rather more in parity with 316 than 304. The calculated PRE numbers in Table 1 correspond relatively well with the CPT values in Figure 41 regarding the ranking of the base metal. One explanation for the lower pitting potential of LDX 2101 in ASTM G61 method can probably be the coarser surface of LDX 2101 (2E) compared to the smoother on 304 and 316L (2B).

42

Figure 41. Typical values of critical pitting temperatures (CPT) for base metal in 1 M NaCl according to ASTM G150 using the Avesta Cell [24].

5.2 TopsideThe results from the as-welded samples were difficult to interpret and that is why LDX 2101 welded with Ar + 2% N2 as shielding gas, was studied in more detail using the Avesta Cell. This was carried out to see whether the initial top (Figure 31b) was caused by low corrosion resistance or by oxide dissolution. However, the curve for 304 is harder to interpret compared with LDX 2101 and that is why 304 were not further investigated. For the pickled welds, only small difference of the curves could be identified, and that is why the Multicell was used, to get improved statistics and faster measurements.

Figure 42 shows some critical pitting potential values for 316L, LDX 2101 and 2304 [SSW 2008, unpublished + en till, prata med west]. All samples were at least polished and some also pickled. It can be noticed that LDX 2101 base metal have similar results as 316L, but even better when welded both with and without filler metal and nitrogen additions to the shielding gas. What is interesting in Figure 42 is that no measurable results were obtained for polished but not pickled LDX 2101 welds. This was concluded to have been caused by oxide dissolution. 304 were also included in Figure 42, but no measurable values could be obtained since 304 have too low pitting resistance. This ranking could not be observed in this work when ASTM G61 – 86 standard was used, Figure 30 and 35, respectively.

43

0

5

10

15

20

25

30

BM None 316L-Si

BM None LDX2101

2205 BM None LDX2101

Filler metal type

CP

T [

ºC] Base metal

Ar

Ar + 2.5% N2

Polished

316L

LDX 2101

2304

Figure 42. Critical pitting temperatures for pickled topside on 316L, LDX 2101 and 2304 [ref Elin, unpublished work].

5.2.1 Avesta CellThe corrosion test in the Avesta Cell was performed on the as-welded samples that strongly deviated from a typical pitting potential curve of a passive metal (Figure 16) by having an initial peak at the beginning before reaching a lower stable current, Figure 31b. This was especially the case for LDX 2101 when using Ar + 2% N2 as shielding gas. 316L followed the standard curve (Figure 16) relatively well and only a smooth top could be observed. When following the standard potentiodynamic pitting potential corrosion test ASTM G61 – 86 a scan rate of 10 mV/min is used. However, this rate was probably too high for measuring as-welded samples because of weld oxide dissolution. The scan rate needs to be tested and adapted for the as-welded samples. In order to study the as-welded samples a potentiostatic measurement was utilized, providing faster measurements compared to using a potentiodynamic method with very low scan rate. The potentiostatic polarisation curve of LDX 2101 with Ar + 2% N2 shielding gas appeared similar to a curve for a passive metal in an acidic solution (Figure 15), by showing one top at the beginning of the curve (Figure 31) followed by a sharp increase at the end. The potentiostatic curve of as-welded 304 (Figure 33b) and 316L (Figure 34b) had a similar appearance. The major difference was that 304 had higher current densities, probably due to oxide dissolution as for LDX 2101. Two peaks could be observed for potentiodynamic measurement of 316L in Figure 34a, which can be pittinitiations that repassivates.

A difference on where the pitting attacks occurred could be seen for the three grades in as-welded and pickled condition. The difference for LDX 2101 between as-welded and pickled was; there was no attacks 1 – 2 mm from the FL when pickling LDX 2101 welded with pure argon, but no difference could be seen when nitrogen additions was used. This was probably due to residual weld oxides caused by a too short pickling time. However, for 304 the attacks did not appear in the weld metal after pickling, which was also the case for 316L. The as-welded 316L samples had pitting attacks in the weld metal, but after pickling all attacks occurred 1 – 2 mm from the FL, Figure 36.

The most interesting welds, LDX 2101 welded with Ar + 2% N2, were further investigated. A number of different measurements on LDX 2101 with Ar + 2% N2 were carried out and some of the measurements were terminated between the first peak and the sharp increase at the end,

44

numbered 1, 2 and 3 in Figure 43. This was performed to verify that no pitting occurred before the sharp increase at the end of the curve, thus identify an approximately potential were the pitting actually occurs in as-welded condition on LDX 2101. As the potentiostatic measurement does not follow the ASTM G61 – 86 standard, the results are not fully verified, but should rather be interpreted as a rough estimation of the pitting potential it confirms that LDX 2101 does not show pitting attacks before number 2 in Figure 43. However, IR drop compensation has not been taken into consideration, which can affect the result somewhat. According to the obtained results the initial top is most probably due to dissolution of oxides. This is supported by the fact that no pitting attack could be observed on the samples terminated at number 1 (Figure 43). Pitting attacks mainly occurred in the weld metal or 1 – 2 mm from the FL. These probably initiate under the dark oxide located 1 – 2 mm from the FL followed by pit initiation in the weld metal.

LDX 2101 As Welded Ar+2N2 / Formier 10 1M NaCl 35C

080407

-0.600 -0.500 -0.400 -0.300 -0.200 -0.100 0 0.100-51.000x10

-41.000x10

-31.000x10

-21.000x10

-11.000x10

01.000x10

E / V

i / A

1 2

3

LDX 2101 As Welded Ar+2N2 / Formier 10 1M NaCl 35C

080407

-0.600 -0.500 -0.400 -0.300 -0.200 -0.100 0 0.100-51.000x10

-41.000x10

-31.000x10

-21.000x10

-11.000x10

01.000x10

E / V

i / A

1 2

3

Start Nr 1 Nr 2 Nr 3

Figure 43. Polarization curve of the topside of LDX 2101 welded with Ar + 2% N2 with three different stop potentials marked with corresponding pictures.

One interesting observation is that LDX 2101 samples welded with Ar + 2% N2 showed a dark oxide adjacent to the FL (Figure 44b) after corrosion measurements. As can be seen in Figure 44a, this was not the case for LDX 2101 welded with pure argon as shielding gas. This dark oxide that forms or is revealed during the corrosion measurement is probably one important factor that affects the pitting potential for LDX 2101 in as-welded condition, but also the difference in polarization curves of samples welded with pure argon and those with Ar + 2% N2. The samples in Figure 44 have, however, not been corroded to the same current level, which can probably be the reason for the higher amount of pits in Figure 44b. The oxide formed when using Ar + 2% N2 is thicker and contains more manganese. It is possible that this reduces the pitting resistance, but this is difficult to measure due to the oxide dissolution. One reason for this was most likely due to formation of an oxide that contains higher fraction of manganese when nitrogen addition is used in the shielding gas, but also due to redeposition

45

Cu

rren

t d

ensi

ty [

A/c

m2 ]

Volt, VSCE

LDX 2101 As-welded Ar + 2N2 Top 1 M NaCl 35°C

Potentiostatic

of manganese from the gas phase and formation of oxynitrides, see weld metal and high temperature oxide in Figure 37 [4].

The dark oxide 1 – 2 mm from the FL on LDX 2101 with nitrogen addition to the shielding gas represents the “high temperature” (HT) oxide measured with GDOES, Figure 37. If comparing GDOES results of LDX 2101 welded with pure argon and welded with nitrogen addition it can clearly be seen that the latter contains a much higher fraction of manganese. The chromium fraction is also lower for LDX 2101 welded with nitrogen addition according to the results in Figure 37. This can maybe explain the somewhat lower pitting potential for samples welded with nitrogen addition in as-welded condition. No such reduction of the pitting resistance when using Ar + 2% N2 was seen for the pickled condition in the ASTM G150 tests in Figure 42, but it has been stated that welds performed on LDX 2101 with nitrogen additions requires pickling for avoiding discoloration and to have measurable corrosion resistance [SSW 2008, Weldox paper].

(a) Welded with pure argon as shielding gas (b) Welded with Ar + 2%N2 as shielding gasFigure 44. Difference between Ar and Ar + 2%N2 as shielding gas on the oxide adjacent to the fusion line.

5.2.2 Multicell304 reached the lowest pitting potential, 316L the highest and LDX 2101 values between 304 and 316L. One interesting observation is that pickled LDX 2101 welded with nitrogen addition to the shielding gas reached a somewhat higher pitting potential value than samples welded with pure argon. This is in accordance with CPT results showing that improved austenite formation improves the pitting resistance [SSW 2008]. The reason for that is most probably the reduction of the amount of chromium nitride precipitates or due to the removal of the manganese-rich weld oxide formed when nitrogen addition was used in the shielding gas [4]. Pickling was the only cleaning method that removed the oxides present on the LDX 2101 samples and restored the corrosion resistance to an acceptable level, Figure 35. Also the austenitic grade 304 restored the corrosion resistance to an acceptable level when pickling was performed. Polishing was not sufficiently effective for restoring the corrosion resistance for either LDX 2101 or 304. As no valid result could be obtained on the as-welded samples, most likely caused by weld oxide dissolution, they were instead measured in the Avesta Cell.

46

5.2.3 GDOESHigh chromium-enrichment can be observed under the surface for 316L and LDX 2101 in the weld metal. Furthermore an iron-enrichment can be seen close to the surface for the two austenitic grades, but not for LDX 2101. Another interesting observation is that all grades have higher silicon content than 316L at the surface, especially 304. LDX also shows high manganese content at the surface of the weld metal compared with the two austenitic grades.

The GDOES measurements from the topside (Figure 37) clearly show significantly higher manganese content on the LDX 2101 specimens. High manganese fractions were seen in the weld metal, and in both the high temperature and low temperature oxide formed on LDX 2101. This was further enhanced by using nitrogen additions to the shielding gas. This can probably be explained by the suggested absorption-evaporisation-deposition process [4]. It states that the majority of the manganese in the weld oxide originates from deposited weld metal species and only to some extent from base metal diffusion.

5.3 Root sideThe majority of the measurements carried out on the root side followed the standard curve (Figure 16) for a passive metal relatively good. The reason why the Multicell measurements turned out well can probably be explained by the fact only oxidation could occur at the root side since the root was not molten. Previous work have shown that on the topside both oxidation and deposition from the weld metal can occur, which is probably why the results of the weld oxide on the topside is different from the root side [4].

5.3.1 MulticellThe Multicell results for the root side in as-welded condition were in agreement with earlier work regarding the ranking between the grades LDX 2101, 304 and 316L. One interesting observation is that LDX 2101 increased the pitting potential from approximately 100 mVSCE

in as-welded condition to almost 300 mVSCE in the pickled condition. The low value for the as-welded condition can probably be explained by manganese oxides formed on LDX 2101, due to the higher manganese content in the alloy compared to 304 and 316L [4]. One reason for the high pitting potential values obtained in pickled condition compared to the pickled base metal can maybe be explained by different pickling times, see Figure 34 and 26. Pickling time for the welded samples was most likely more adapted than for the base metal. Only one pickling time was tested for the base metal and that time was probably not the most optimal. Another observation is that pickled LDX 2101 welded with no purging gas resulted in higher pitting potential than for samples welded with 90% N2 +10 H2, which is known to be a superior purging gas. This can also be a result of the pickling time. As all LDX 2101 was pickled for equally long time, this time might be relatively optimal for some of the oxides, but not for all of them. However, the same pickling time can be to long for the oxide formed when 90% N2 +10 H2 purging gas was used, but perfect for the oxide formed when pure argon was used. Detta är föreslag från Elin istället för markerad text ovan: “What has previously been known is that 90% N2 + 10% H2 gives better corrosion resistance than pure argon in as-welded condition. In pickled condition the results deviates significantly from what is previously known from ASTM G150 measurements and ?…? experience.” Vad hände med pickling time?

Ericsson et al. [40] showed how the pitting resistance of 316 stainless steel changed with different post-weld cleaning methods, Figure 45. The pitting potential increased with

47

increased degree of surface finish, which means that the susceptibility to pitting is decreased. In this work, this pattern could only be seen for the topside, Figure 30 and 39. The root side results from measurements of 316L with pure argon as purging gas, LDX 2101 with 90% N 2

+10 H2 purging gas and 304 without any purging gas did not follow the pattern. One important factor is that only two measurements were carried out on the polished welds, which makes is hard to interpret the results when they have similar pitting potential values. The initial ranking between the materials also deviated from expected values.

Figure 45. Pitting resistance for different cleaning methods after annealing at 1050°C for five minutes in 0.1 M NaCl at 25°C [40].

5.3.2 GDOESTo perform measurements on the root side a special sample holder was used, but the results were not fully validated. The high oxygen and nitrogen signals obtained during test measurements with the special sample holder were most probably caused by air leakage into the cavity where the glow discharge process occurs. Measurements on the root side can be performed when the special sample holder is fully tested and verified. This will give valuable information that could help the interpretation of the corrosion results.

5.4 Ferrite contentThe effect of nitrogen additions to the shielding gas on the average weld metal austenite content was small 74 3% to 71 3% when Ar + 2% N2 was used. The austenite formation at the cap side that is normally exposed to the corrosive media was, however, significantly improved, and the amount of chromium in the nitrides at the surface was reduced. This can be explain why the majority of the pitting attacks occurred in the weld metal for LDX 2101 welded with pre argon, while nitrogen additions improved the corrosion performance. Earlier work [39] has shown that nitrogen addition to the shielding gas increases the austenite formation.

48

6 ConclusionsThree different types of stainless steels (304, LDX 2101®, 316L) were TIG welded bed on plate without filler material. Two post-weld cleaning methods were performed on the welded samples, polishing with K80 coarseness and pickling in pickling bath. Two types of ASTM G61 – 86 based measurements were performed, one was recording polarization curves in the Avesta Cell and the other was measuring pitting potentials in the Multicell. Recording of polarization curves in the Avesta Cell was further investigated on the topside of LDX 2101 welded with nitrogen additions to the shielding gas. Surface analysis with GDOES was performed on the welded samples to provide better understanding for the corrosion results.

Residual weld oxides reduced the corrosion resistance according to ASTM G61 – 86, especially for 304 and LDX 2101. Polishing to K80 coarseness somewhat improved the corrosion resistance on both top and root side. Pickling restored the corrosion resistance to the highest level in most cases, especially on the topside for 304 and LDX 2101. 316L generally showed the highest pitting potential values both on top and root side.

The ranking of the three investigated grades were not the same if comparing the ASTM G61 – 86 used in this work with the more common ASTM G150 corrosion test method. It seems like the ASTM G61 – 86 method was more dependent on the surface roughness of the sample than ASTM G150.

LDX 2101 welded with Ar + 2% N2 developed an oxide that was hard to remove with polishing and pickling, only pickling was sufficiently good for improving the corrosion resistance. However, when the oxide was removed a somewhat higher pitting potential was obtained compared with samples welded with pure argon. This is probably due to the improved austenite formation in the weld metal when using nitrogen additions in the shielding gas, which decreases the chromium nitride precipitates.

The high temperature oxide formed on the cap side of LDX 2101 welded with nitrogen addition to the shielding gas has higher manganese content in the outer oxide layer compared with LDX 2101 welded with pure argon, which will probably have a negative effect on corrosion resistance if not removed.

The initial current peak that occurs for LDX 2101 in the beginning of the ASTM G150 measurement can also be identified when using ASTM G61 – 86 method in the Multicell, which in both cases exceeds the set threshold value of 1 mA. The ASTM G61 – 86 Avesta Cell measurements of LDX 2101 in as-welded condition showed that a sharp increase in current density occurs at the beginning of the measurement, exceeding the threshold value 1 mA. This increase in current density is most probably due to oxide dissolution and results showed that no visible pitting attacks could be identified after the first current peak, which indicates that the corrosion resistance is higher than what the standard measuring techniques show.

The relatively high pitting potential values on the root side compared with the topside of LDX 2101 in pickled condition can probably be explained by the redeposition of manganese from the weld metal, which most probably occurs on the topside.

49

The weld oxide formed on LDX 2101 when adding nitrogen to the shielding gas becomes thicker and the composition deviates somewhat from that of the weld oxide formed when using pure argon. LDX 2101 with nitrogen addition showed slightly lower pitting potential values in as-welded condition, but somewhat better than pure argon in pickled condition. The former is addressed to oxynitrides formed from deposition from the weld metal and the latter due to improved weld metal pitting resistance.

7 Suggestion to further workIt would be beneficial to validate and adjust the special sample holder, for non-flat specimens, to obtain valid GDOES measurements on the root side of the welded samples, which in turn could give some explanation to the pitting potential results on the root side.

As different materials and different shielding gas types needed different pickling times to become visually clean, the effect of different pickling times should be investigated.

GDOES measurements on samples that have been corrosion tested to see how the surface composition, and especially on LDX 2101, is changed after aborted measurements showing oxide dissolution, but no pitting attack

The GDOES instrument should be calibrated for using the 1 mm anode instead of the 2 mm used in this work. This would give better precision when analysing the different oxide zones on the welded specimen.

Performing ASTM G150 tests with increased threshold values to determine the CPT more accurately.

CPT measurements of all conditions should be performed for better comparison with earlier work. Investigations of other post-weld cleaning methods such as pickling paste, shot blasting, polishing, wet grinding with different coarseness and combining mechanical and chemical methods would show the necessity of post-weld cleaning.

Performing polarization curves on welded samples of LDX 2101, 304 and 316L using ASTM G61 – 86 standard with lower scan rate.

8 Acknowledgements

50

9 References[] R. F. A. Pettersson, J. Flyg, Influence of weld oxides on the corrosion of 316L stainless steel, Eurocorr

2005, Corrosion Control for Sustainable Development, European Corrosion Congress, EFC Event 273, Lissabon, Portugal, 4-8 Sept. 2005. Paper 644. 9 pp. Publ: Rolo & Filhos Lda, Portugal; Producao Grafica; 2005. ISBN 972-95921-1-X.

[2] T. von Moltke, P.C. Pistorius, R.F. Sandenbergh, The influence of heat-tinted surface layers on the corrosion resistance of stainless steels, Proc. First Internatrional Chromium Steel & Alloys Congress, INFACOM 6. Vol. 2. Chromium Steel and Alloys, Cape Town, South Africa, Johannesburg, SAIMM, 1992, pp. 185-195.

[3] S. Azuma, H. Miyuki, J. Murayama, T. Kudo, Corros. J. 29 (1994) 78-80.[4] E. M. Westin, Weld Oxide Formation on Lean Duplex Stainless Steel.[5] S. Turner, F.P.A. Robinson, Corrosion 49 (1989) 710-716.[6] J.R. Cahoon, R. Bandy, Corrosion 36 (1982) 299-305.[7] K. Asami, K. Hashimoto, Corros. Sci. 19 (1979) 1007-1017.[8] John C. Lippold, Damian J. Kotecki, Welding Metallurgy and Weldability of Stainless Steels, 2005. p. 2.

John Wiley & Sons, Inc., Hoboken, New Jersey, ISBN 0-471-47379-0.[9] Outokumpus egen materialutbildning, ”KiR – Konstruera i Rostfritt”, section ”Materialval”, 11pp. [10] The Avesta Welding Manual: Practice and products for stainless steel welding, 2nd Edition (2005), Edita

Västra Aros Sweden, ISBN 91-631-5713-6.[11] John C. Lippold, Damian J. Kotecki, Welding Metallurgy and Weldability of Stainless Steels, 2005. p. 97.

John Wiley & Sons, Inc., Hoboken, New Jersey, ISBN 0-471-47379-0.[12] I. Varol, W.A. Baeslack III, J.C. Lippold, Welding of duplex stainless steels, Key Engineering materials

(Switzerland), Vol. 69-70, (1992) pp. 217-251.[13] P. Combrade, J.P. Audouard, Duplex stainless steelsls and localized corrosion resistance, Conference

Proceedings of Duplex Stainless Steels ’91, Beaune, France, Vol. 1, 28-30 Oct. 1991, pp. 257-281. 1991.[14] M. Miura, M. Koso, T. Kudo, H. Tsuge, Welding International, 1990, 4 (3), p. 200.[15] Sakai et al. DSS (1983) p. 211., Charles et al. Corrosion 1989., Miyuki et al, DSS 1983, p. 95., Nagano et al.

Métaux-Corrosion-Industrie, 56 (1981) p. 61.[16] Srindhar and Kolts, Corrosion 43 (1987) p. 646., Miyuki et al, DSS 1983, p. 95., Nagano et al. Métaux-

Corrosion-Industrie, 56 (1981) p. 61.[17] J. Sakai, I. Matsushima, Y. Kamemura, M. Tanimura, T. Osuka, Duplex stainless steels, 1983, p.211, Edited

by R.A. Lula, American Society for Metals, Metals Park, Ohio.[18] J.E. Truman, M.J. Coleman, K.R. Pirt, British Corrosion Journal, 1977, 12, 4, 236.[19] T. Ogawa, T. Koseki, Welding Journal, 1989, 68, 5, p181s.[20] Denny A. Jones, David Johnstone (editor), Principles and Prevention of Corrosion, 2nd Edition (1992). pp 4-

5, 5-9, 7-9, 168-169, Macmillan Publishing Company, 866 Third Avenue, New York, New York 10022, ISBN 0-02-946439-0.

[21] Tsuge, H., Tarutani, Y. & Kudo, T. (1988) The effect of nitrogen on localised corrosion resistance of duplex stainless simulated weldments, Corrosion 44, 305-314.

[22] Matsunaga, H., Sato, Y.S., Kokawa, H. & Kuwana, T. (1998) Effect of nitrogen on corrosion of duplex stainless steel weld metal, Sci. Technol. Weld. Joi. 3 225-232.

[23] G. Herbsleb, Werkt. Korros. 33 (1982) pp. 334-340.[24] Outokumpu Stainless, Corrosion Handbook, 9th Edition (2004), Edita Västra Aros, Västerås, Sweden.[25] Outokumpus egen materialutbildning, ”KiR – Konstruera i Rostfritt”, section ”Korrosion”, 12pp.[26] Sandvik Steel, Corrosion Handbook Stainless Steels, (1999), printed in Sandviken, Sweden, ISBN 91-630-

2124-2.[27] S. Turner, F. B. A. Robinson, The effect of The Surface Oxides Produced during Welding on The Corrosion

Resistance of Stainless Steels, Corrosion, Sept.-89, p 710-716.[28] M. Berner, Estimating the Localized Corrosion Resistance of Low-Alloy Stainless Steels: A Comparison of

Pitting Potentials and Critical Pitting Temperatures Measures on the Lean Duplex Stainless Steel LDX 2101 after Sentitisation.

[29] R. Qvarfort, Crittical Pitting Temperature Measurements Of Stainless Steels With An Improved Electrochemical Method, Corrosion Science, 29 No 8, 987-993 (1989).

[30] Philippe Marcus, Florian Mansfeld, Analytical methods in corrosion science and technology, 2006. p 445. CRC Press, Taylor & Francis Group, LLC, Boca Raton, FL 33487-2742, ISBN 0-8247-5952-4.

51

[31] Annual Book of ASTM Standards, ASTM G 61-86 (2007): Standard test method for conducting cyclic potentiodynamic polarization measurements for localized corrosion susceptibility of iron-, nickel-, cobalt-based alloys, ASTM international, West Conshohocken, PA 19428-2959, United States.

[32] M. Berner, M. Sc. Thesis, EPFL 2006 and ARC Internal report SE20060301.[33] R. Qvarfort, New Electrochemical Cell For Pitting Corrosion Testing, Corrosion Science, 28 No 2, 135-140

(1998).[34] C.-O. A. Olsson, The Avesta Cell, Simultaneous Six Samples Version, Internal report 2006.02XX.[35] Sindo Kou, Welding metallurgy, 2nd Edition (2003), John Wiley & Sons, Inc., Hoboken, New Jersey, ISBN

0-471-43491-4.[36] Ellingham H. J. T., J Soc Chem Ind (London) 63 125 (1944).[37] Horiba Jobin Yvon information PDF,

http://www.jobinyvon.com/SiteResources/Data/MediaArchive/files/Emission/GDPROFILER_Series.pdf, 2008-05-19.

[38] Thomas Nelis, Richard Payling, Glow Discharge Optical Emission Spectroscopy: A practical Guide, The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge CB4 0WF, UK ISBN 0-85404-521-X.

[39] E. M. Westin, E. Keehan, M. Ström, B. von Brömssen, Laser welding of a lean duplex stainless steel, Proc. ICALEO 2008, Paper 609, 10 pp.

[40] R. Ericsson, L. Schön, B. Wallén, Effect of some oxide removal treatements on the corrosion resistance of 316 stainless steel, Proc. Of the 8th Scand. Corr. Congress, NKM 8, Helsinki, 1978, pp. 321-329.

52

Appendix A

MULTICELL MEASUREMENTS

TABLE 1. Pitting potentials of base metal. BASE METAL

Material Epit (Average) [mVAg/AgCl] Epit (Average) [mVSCE] STDAV* Number of samples

As-received304 284 239 30 2

LDX 2101 160 115 4 2316L 437 392 11 2

320 mesh304 294 249 16 9

LDX 2101 303 258 45 6316L 397 352 25 6

Pickled base metal304 176 131 none 1

LDX 2101 224 179 14 2316L 400 355 10 2

Polished base metal (K80)304 302 257 6 2

LDX 2101 278 233 1 2316L 416 371 none 1

1

*

2

N

xxSTDAV

53

Appendix B

MULTICELL MEASUREMENTS

TABLE 2. Pitting potentials of welded topside. WELDED TOPSIDE

Material Shielding gas Epit (Average) [mVAg/AgCl] Epit (Average) [mVSCE] STDAV* Number of samples

As-Welded 304 Ar --- --- none 0

LDX 2101 Ar --- --- none 5LDX 2101 Ar+2% N2 --- --- none 0

316L Ar 138 93 4 6

Polished (K80)304 Ar 87 42 9 2

LDX 2101 Ar 73 28 13 2LDX 2101 Ar+2% N2 --- --- none 6

316L Ar 333 288 8 2

Pickled304 Ar 247 202 28 6

LDX 2101 Ar 287 242 62 4LDX 2101 Ar+2% N2 294 249 68 3

316L Ar 378 333 13 6

TABLE 3. Pitting potentials of welded root side. WELDED ROOT SIDE

Material Purging gas Epit (Average) [mVAg/AgCl] Epit (Average) [mVSCE] STDAV* Number of samples

As-Welded304 none 93 48 5 5304 Ar 131 86 17 6

LDX 2101 none 52 7 1 3LDX 2101 Ar 167 122 17 6LDX 2101 90% N2 + 10% H2 187 142 50 4

316L none 159 114 2 4316L Ar 203 158 6 3

Polished (K80)304 none 257 212 3 2304 Ar 249 204 11 2

LDX 2101 none 253 208 20 2LDX 2101 Ar 280 235 25 2LDX 2101 90% N2 + 10% H2 263 218 55 6

316L none 321 276 78 2316L Ar 434 389 1

Pickled304 none 253 208 56 6304 Ar 300 255 20 6

LDX 2101 none 299 254 72 8LDX 2101 Ar 426 381 102 10LDX 2101 90% N2 + 10% H2 227 182 64 10

316L none 375 330 17 6316L Ar 389 344 11 6

1

*

2

N

xxSTDAV

54