Keywords

1 Introduction

In order to reduce environmental impact, it is important to further increase the service life of components. To this end, maintenance, repair and overhaul (MRO) is becoming increasingly important. Especially for components made of expensive materials such as titanium alloys, a cost advantage can arise in addition to the environmental aspect (Wits et al. 2016).

In order to extend the range of application of MRO, this subproject B6 of the Collaborative Research Center (CRC) 871 has researched possibilities to influence the joining zone during the repair of titanium components. In addition to the existing patch repair, which was extended by the use of fluxes and inoculants, a new repair method was developed: additive repair using wire-and-arc-additive-manufacturing (WAAM).

Within the CRC 871, the titanium blisk was selected as a component to be repaired. The titanium alloy Ti6Al4V was selected, as it is the most widely used titanium alloy. However, the methods investigated are not limited to the blisk and Ti6Al4V, but can also be applied to other alloys.

There are several repair methods in aerospace, these include patch repair, blend repair and tip repair. In patch repair, the defect is removed. In this process, the focus can be either to create a simple patch geometry or to transfer the seam to a lower-stressed area. A corresponding patch is manufactured for this application. The patch is welded to the component using one of various welding processes. This subproject investigated the repair possibilities of arc processes. After welding, the component is recontoured to restore the original geometry. In blend repair, a crack is not filled, but the crack is removed and a geometry is created that deviates from the initial geometry, but the notch effect of the crack is reduced. In tip repair, a seam is welded onto the tip to fill material and then the blade is recontoured to restore the initial geometry. An extension of the tip repair is additive repair. Here, instead of welding a single seam, the desired geometry is produced by a layer-by-layer technique. In order to be able to select a repair path, it must be clear in advance which properties the component attains for each repair path. With this knowledge the decision can be made if the component is still usable as is, if the part is scrap and needs to be replaced with a new part, or if it can be repaired, cf. Figure 1.

Fig. 1
A diagram of the usage of a highly stressed component. It is either usable or scrap. If it is usable, repair, followed by patching of another part, and a layer by layer deposition technique labeled W A A M are performed.

Possible decision paths for the repair of a highly-stressed component; three basic options are available for the component, i.e., the component can be used as is, it is scrap or it can be repaired

Full size image

Every repair weld involves an unavoidable heat treatment of the microstructure. In the fusion zone (FZ), a directionally solidified microstructure with isotropic properties is typically formed. In the heat-affected zone (HAZ), grain growth occurs and precipitates possible are dissolved and precipitated again. Welding is accompanied by extreme heating rates and very rapid cooling. For most titanium alloys, including Ti6Al4V, this means a hardening of the HAZ and FZ coupled with a significant decrease in ductility (Schulze 2010).

Usually the base material is homogenized after melting and then deformed. After forming, recrystallization usually takes place for α-alloys. For α-β-alloys, optional recrystallization is followed by an annealing. For β-alloys, either recrystallization annealing, annealing or both are employed. Heat treatment is then completed by (multiple) aging (Derby et al. 2003).

During welding, the homogenized base material is melted and cooled with cooling rates of up to several 100 °K/s (Schulze 2010). The welding of a conventional patch repair ends here. Due to the local heating and cooling, residual stresses are generated (Denkena et al. 2017); due to the lack of a subsequent annealing process, the residual stresses cannot be relieved and have a diminishing effect on the fatigue strength. How- ever, in order to achieve the grain refinement that the base material has, cooling rates of more than 1000 °K/s are necessary or the steps of forming with high degrees of deformation and recrystallization annealing are needed (Derby et al. 2003).

Various investigations have been carried out on the grain refinement of welds. On one hand, there are possibilities to design the process differently. On the other hand, one can add additives whose reactions influences the process.

In welding, heterogeneous nucleation is dominant and heat is mainly dissipated by the substrate. The grain growth progresses in the opposite direction to the heat flow, i.e. from the substrate into the molten pool. For grain refinement, Balasubramanian et al. (2008) pulsed a TIG arc to vibrate the molten pool. They used a frequency of 6 Hz, 40 A base current and 80 A pulse current. They were able to reduce the prior β-grain size to a third of the grain size achieved with the pulse-free arc. The grain size of the reference was 340 μm and the grain size of the pulsed sample was less than 100 μm.

Reisgen et al. and Henckell et al. have investigated active cooling. For this purpose, Reisgen et al. (2020) compared water bath cooling with aerosol cooling and reference welding when welding steel. Water bath cooling showed the greatest increase in hard- ness. Aerosol cooling with water and air also showed an increase in hardness compared to reference welding. Henckell et al. (2017) studied the influence of gas cooling in additive manufacturing of steel. They were able to both increase strength and reduce grain size through the use of a leading gas cooling system. Li et al. (2019) used hot wire in TIG welding of titanium. The use of hot wire allows a reduction in arc energy, which in turn reduces the weld pool temperature. Thus, the weld geometry becomes narrower. At the same time, grain refinement occurs and the mechanical properties in the vertical and horizontal directions equalize. Gou et al. (2020) stimulated the melt pool of Ti6Al4V with ultrasound. This reduced the average length of α-grains from 40 to 20 μm. In addition, the anisotropy of the component was reduced. Hönnige et al. (2017) introduced residual compressive stresses by ultrasonic impact treatment of the weld. By welding the next layer, recrystallization annealing occurs and grain refinement is achieved. Colegrove et al. (2017) rolled over the weld at 75 kN after welding Ti6Al4V, introducing residual compressive stresses. Grain refinement occurred in the HAZ of the next layer. Significant grain refinement was achieved and isotropic mechanical properties were obtained that exceed those of the reference. In the present study, the influence of mechanical deformation was also investigated. Mechanical working of the component significantly restricts the repair process. Forming usually takes place outside the inert gas atmosphere, so the component must be cooled down before treatment. This means that high forces are necessary to generate the required residual stresses in the component. Due to their geometry, components such as the blisk limit the force with which the residual stresses can be introduced without damaging the component. In order to take advantage of the mechanical processing, either a fixture must be manufactured to absorb the forces so that the component is not unintentionally loaded, or the energy input and the associated layer thickness must be reduced.

Another way to influence the process is by in-situ alloying. In the present subproject, the effect of fluxes was investigated. Among the fluxes used (AlCl3, NaCl, FeCl, VCl3, AlF3, CaF2), AlF3 had the greatest effect on the deep welding effect during joint welding. The results of AlF3 are presented below. Yin et al. (2017) also investigated the influence of adding CaF2 in additive manufacturing of Ti6Al4V and observed an effect on the weld geometry and mechanical properties.

The addition of inoculants was also studied in the subproject. With the idea of adding high melting components to the molten bath, which then act as heterogeneous nucleating agents, and thus lead to grain refinement of the FZ. Specifically, Si, SiC, TiC, C, B, TaC were used (Langen 2021). Bermingham et al. (2019) and Mereddy et al. (2017, 2018) have also investigated the grain refinement potential of adding Si, C, and La2O3 to titanium alloys.

2 Materials and Methods

2.1 Materials

For all tests, substrates made of the alloy Ti6Al4V (3.7164) were used. The chemical compositions for the different materials used in the studies are shown in Table 1. The samples were cut by plate shears or by water jet cutting.

Table 1 Chemical analysis of the substrate and welding wire in wt.%
Full size table

2.2 Methods

2.2.1 Shielding Gas Chamber

The shielding gas chamber was filled with argon with purity of 99.998%. Figure 2 shows the chamber used. The workspace is 2 m × 2 m × 1 m. Manual tests can be carried out via eight gloves. A six-axis robot, type KUKA Sixx R900 from KUKA AG, was employed for automated welding.

Fig. 2
A photograph of an argon welding chamber for processing refractory metals.

Argon welding chamber for processing refractory metals; workspace: 2 m3; handling system: KUKA Sixx R900

Full size image

A gas purification device, type E-Line Gasreinigung from GS GLOVEBOX Systemtechnik GmbH, automatically maintained a positive pressure of 1.5 mbar and allowed tests to be performed at a residual oxygen content of less than 10 ppm and a residual water content of less than 0.5 ppm. Various welding methods could be used in the chamber including: microplasma welding, tungsten inert gas welding (TIG), metal inert gas welding (MIG) and plasma nitriding with non-transferred arc.

2.2.2 Preparation and Welding of the Flux Samples

Specimens (180 mm × 210 mm × 2 mm) were cut from a plate. The specimens were cleaned with acetone. Using an airbrush gun, a solution of ethanol and AlF3 was applied to the specimens through a spray stencil. Thus, a strip of 20 mm × 150 mm, cf. Figure 3a and a weight of 0.13 ± 0.01 g was applied to the center of the sample. The area density was 0.04 mg/mm2. The welding parameters are listed in Table 2.

Fig. 3
Two diagrams at the top represent a block of 180.00 in length and 210.00 in width with 8 vertical line cuts in the rectangular strip in the block. A diagram of a dog bone specimen with a length of 200.00 and a width of 20.00 is represented at the bottom.

Process design of the flux samples: a representation of specimen geometry with flux application; b location of tensile specimens; c geometry of the tensile specimens

Full size image
Table 2 Welding parameters of additive repair and tests with flux or inoculant
Full size table

2.2.3 Preparation and Welding of the Inoculation Samples

The specimens for the tensile tests with inoculant were prepared in the same way as the flux specimens. However, the applied strip was smaller. The width was only 3.3 mm. The weight of the applied inoculant was 0.17 ± 0.02 g. The resulting area density was 0.34 mg/mm2.

The experiments for the micrographs were performed on samples with a size of 180 mm × 56 mm × 2 mm. The area on which inoculant was applied was 3.3 mm × 150 mm. The weight of the applied inoculant was 0.6 g; the area density was 1.21 mg/mm2.

As on the flux specimens, the inoculant specimens were welded without filler material in the welding chamber using the same welding system, cf. Figure 3. The welding parameters are listed in Table 2.

After welding, the weld buildup was removed by milling. The tensile specimens had the same design as the flux tensile specimens cut out using wire EDM.

2.2.4 Preparation and Welding of the Wire and Arc Additive Manufactured Samples

Substrate plates measuring 330 mm × 60 mm × 5 mm were cut from a titanium Ti6Al4V plate. The chemical composition of the substrate plates and the welding wire used are shown in Table 1. The substrate plates were cleaned with acetone. A single-wall 300 mm × 300 mm × 6 mm was welded in the shielding gas chamber. The process parameters are listed in Table 2 (WAAM TIG1). The welding power source used was a Tetrix 350 AC/DC from EWM AG. Waterjet cutting was used to cut out the ground and tensile specimens.

To increase the cooling rate, and thus decrease the prior β-grain size, a new wall was build. The dwell time was increased and the other process parameters had to be adjusted slightly, see Table 2 (WAAM TIG2). The specimen preparation was carried out in the same way as described before.

2.2.5 Determination of the Mechanical Properties

Zwick Z100 and Z250 universal testing machines from Zwick/Roell were used to perform the static tensile tests. The tensile specimens for flux and inoculation testing were designed to meet the standard DIN 50125 H 1.8 × 12 × 50, see Fig. 2c. The tensile specimens of the additive repair were designed to DIN 50125 E 2 × 6 × 20 with enlarged clamping surfaces. After cutting out of the welds, the tensile specimens were milled to final dimensions. Rotary bending tests were performed on round specimens to determine fatigue properties.

2.2.6 Microscopy

The samples were cut by water jet cutting, embedded, ground with SiC abrasive paper of the grit 500, 800, 1200 and 2500. Afterwards polished with an alcoholic diamond suspension (9 μm). Before etching with Kroll’s reagent, the samples were finished on a vibratory polisher VibroMet from Buehler. Images were taken using a BX61 incident light microscope with an Olympus XC30 camera.

2.2.7 Vickers Hardness Testing

The Vickers hardness tests were carried out according to DIN EN ISO 6507–1 with a loading time of 13 s. The micro hardness tester used was a Q10 A + from QNESS GmbH. A load of 9.81 N was applied to the test specimen for the flux and inoculation samples. The distance between the indentations was 0.22 mm. For the hardness measurements of the nitrated samples, a load of 0.245 N was applied. The tests were carried out with a Vickers Microhardness Tester FM-310 from Futer-tech. The distance between the indentations differed between 5 and 30 μm.

2.2.8 Local Atmospheric Nitriding with Non-transmitting Arc

Local atmospheric plasma nitriding was applied using various methods. These included TIG and plasma nitriding with non-transmitting arc. Nitrogen was used as the working gas. Figure 4 shows the nitriding process. Nitrogen flows to the arc as the working gas. There, the nitrogen molecules are dissociated in the arc. The nitrogen atoms are then ionized by the arc. The ions either diffuse into the titanium or they recombine at the cold surface to form molecules again. In TIG welding, the surface is melted and the nitrogen diffuses into the molten pool.

Fig. 4
A schematic represents the process of nitriding with non transferred arc. The labeled parts are nitrogen molecules are dissociated, nitrogen atoms are ionized, nitrogen recombines on the cold surface, atomic nitrogen is adsorbed on the surface, and atomic nitrogen diffuses into the titanium.

Schematic process of the nitriding with non-transferred arc with the steps dissociation, ionization and diffusion

Full size image

3 Results and Discussion

3.1 Reduction of Line Energy by Addition of Flux

Without flux, a line energy of at least 163 J/mm (current 60 A) was required to weld through the 2 mm thick specimens. By using 0.04 mg/mm2 aluminum fluoride, AlF3, as flux, the energy per unit length required for welding could be reduced from 163 to 83 J/mm (current 20 A). The flux influences the gradient of the surface tension dγ/dT, where T is the temperature. With a sufficiently large amount of flux, the gradient becomes positive (Schulze 2010). This means that the surface tension increases with increasing temperature. As a result, the direction of the Marangoni flow is reversed and the melt pool flow direction at the top of the melt pool is inverted. Figure 5a shows the usual melt pool flow for titanium. Figure 5b shows the course of the molten pool flow with a sufficiently large quantity of surface-activating substances (flux) to make the gradient positive.

Fig. 5
2 diagrams represent melt flow into the liquid weld. In A, 2 arrows on the left and right represent the melt flow. In B, 2 side by side downward facing arrows represent the melt flow.

Melt flow into the liquid weld: a shows the usual direction of the melt pool flow for titanium, b shows the direction of the flow when dγ/dT is >0 due to the influence of the flux (Schulze 2010)

Full size image

Figure 6 shows the reference weld and flux weld with 60 A, 40 A and 20 A. Only with 60 A, tests welded trough the 2 mm plate, cf. Figure 6a. With the addition of flux, all tests at 60 A, 40 A and 20 A welded trough, cf. Figure 6b. By reducing the energy required for joint welding, both the heat-affected zone and the FZ become significantly smaller compared to the reference. The smallest width of HAZ of the reference, which welded trough was about 12 mm and the smallest width of the HAZ of the weld with flux which welded trough was about 7 mm.

Fig. 6
6 micrographic image strips of cross sections of welded specimens. A, 3 micrographic strips represent 3 welded specimens welded through bead on plate in different current flows 60, 40, and 20 amperes. B, 3 micrographic strips represent 3 welded specimens welded through flux in different current flows.

Cross-section of bead on plate welds: a shows the reference welds with 60 A, 40 A and 20 A, b shows the weld with the use of flux with 60 A, 40 A and 20 A

Full size image

The coarse grain growth in the HAZ can be seen in both welds. The FZ of both specimens shows a martensitic structure. The flux weld shows smaller former β-grains due to the increase cooling rate compared to the reference. The hardness measurement shows a welding-induced hardening of the HAZ and FZ for the reference. The HAZ and FZ have approximately the same hardness at 360 HV1, which is 40 HV1 more than the base material. The hardness measurement of the flux sample shows the same hardness level of 360 HV1 for the HAZ, but the FZ is harder than for the reference. It reaches 400 HV1 and is thus 40 HV1 higher than in the reference. Due to the lower energy input, the meltpool has less overheating and cools down faster. Smaller β-grains are formed, but also more elements remain dissolved in the α'-martensite and harden it, cf. Figure 7.

Fig. 7
2 line graphs of hardness H V 1 versus distance from weld center labeled G T A and G T A A I F 3. Left, a fluctuating upward line is plotted. Right, a fluctuating downward line is plotted. In both graphs, there is a micrographic image at the bottom of a welded specimen, representing hardness.

Comparison of the hardness profile of the reference on the left, the current used is 60 A; the flux-welded specimen is on the right, welded with 20 A.; the dashed red line show the position of the hardness measurements; BM = base material; HAZ = heat affected zone; FZ = fusion zone

Full size image

The positive influence of the flux is visible in the mechanical properties of the overall joint, see Fig. 8. The energy reduction reduces the proportion of HAZ plus FZ of the repaired overall joint. Although the strength in the HAZ and FZ increases, the overall connection only reaches the level of the base material, since it breaks there. The worsened elongation due to the unwanted heat treatment can be seen. The weld with flux has visibly changed the microstructure in zone with a width of approx. 7 mm. For the reference the corresponding width was approx. 14 mm. The extensometer used to measure the strain had an initial length of 50 mm. Although less than 20% of the measured area of the flux had been optically changed, the sample only achieved 60% of the elongation. In the case of the reference, one third of the area was optically modified and only just under 35% of the strain of the base material was achieved. This is due to the fact that the HAZ and FZ hinder the movement of the dislocations.

Fig. 8
A grouped bar graph with error bars plots stress in megapascals and strain in percentage versus specimens. Ref has the highest tensile strength followed by flux and B M. B M has the highest elongation followed by flux and ref. Flux has the highest 0.2% offset yield strength followed by ref and B M.

Yield strength, tensile strength and elongation at break of base material, welded specimens and flux-welded specimens BM means the base material, Ref means the reference weld at 60 A and Flux means the weld at 20 A and the use of flux; error bars show the highest and lowest values obtained from the specimens

Full size image

3.2 Grain Refinement of the Prior β-grains with Inoculation

Inoculants are added to the melt during casting, the high-melting compounds act as heterogenic nuclei and stimulate grain formation, resulting in grain refinement (Bermingham et al. 2008). When inoculating welds with SiC, it has been shown that SiC melts and recombines to form TiC and Si (Langen et al. 2018). Thus, the inoculant does not act as an inoculant but as an in-situ alloy. With the SiC inoculant, a significant reduction in β-grain size can be achieved. In addition to the results presented here, grain refinement could also be achieved with TaC, VC, TiC, Si, and C (Langen 2021). In the FZ, the β-grains reached lengths above 1 mm. This is shown in Fig. 9a. Figure 9b shows a transverse section of the inoculated sample. First, the transition from FZ to HAZ is clearly visible, as grain refinement with the inoculant only affects the FZ. In the HAZ, coarse grain growth occurs as in the reference.

Fig. 9
A set of 4 micrographic images of a cross section welded specimen. In reference and with inoculation, there is a strip like micrographic image at the top labeled H A Z and F Z, and at the bottom, a smooth surface with multiple lines is observed.

Cross-section of reference and with SiC inoculated sample both welded at 90 A; on the left reference weld is shown, the prior β-grain size reach a length up to 2 mm; on the right an inoculated sample welded is shown; the prior β-grain size in the FZ is less than 100 µm; the precipitated carbides reach a length of <20 µm; however, in both samples, coarse grain formation occurs in the HAZ

Full size image

The hardness measurement on the cross-section shows that the inoculant only acts in the FZ. The coarse grain formation already seen in the microsection in the HAZ shows the same hardness as the HAZ of the reference. However, the addition of SiC results in further hardening of the FZ in the inoculated sample. Detailed EDX studies showed that the SiC dissolves in the melt and subsequently the carbon reacts with the titanium to TiC (Langen 2021). The precipitated titanium carbides further harden the FZ, see Fig. 10.

Fig. 10
2 line graphs of hardness H V 1 versus distance from weld center labeled G T A and G T A S i C. Left, a fluctuating upward line is plotted. Right, a fluctuating downward line is plotted. In both graphs, there is a micrographic image at the bottom of a welded specimen, representing hardness.

Comparison of the hardness profile of the reference on the left and the specimen welded with SiC inoculant on the right (Langen 2021)

Full size image

Tensile tests showed that incolation with an area density of 1.21 mg/mm2 SiC (1.7 wt.% Si in the FZ) reduced the elongation significantly to less than 0.5% (Langen 2021).

When inoculated with an area density of 0.34 mg/mm2 SiC (0.8 wt.% Si in the FZ), grain refinement is decreased but elongation at break is improved (Langen et al. 2018). Figure 11 shows the results of the tensile tests for the base material, the reference weld and inoculated specimens with SiC, B and TaC. From the values of the base material and the reference, it is clear that a single weld significantly degrades the mechanical properties of the component. Both strength and ductility are decreased. By using all inoculants, the strength of the weld zone is increased. The inoculated specimens broke outside the weld and showed a typically overmatching of the weld seam. The ductility of the specimens inoculated with B and TaC are similar to the reference weld. Only the specimens inoculated with SiC show better ductility of the overall joint.

Fig. 11
A grouped bar graph with error bars plots stress and strain versus specimens. S i C has the highest tensile strength followed by T a C, B, B M, and ref. B M has the highest and ref has the lowest 0.2% offset yield strength and elongation.

Yield strength, tensile strength and elongation at break from base material, welded specimens and specimens welded with inoculants; BM is the base material, Ref stands for the reference and SiC, B and TaC stand for the inoculants used; the current used for all welds was 90 A; error bars show the highest and lowest value obtained for the specimens (Langen et al. 2018)

Full size image

Mereddy et al. were also able to reduce the average length of the α-grains from 180 μm to below 40 μm with the addition of 0.41 wt.% carbon to Ti6Al4V. However, the TiC precipitates formed embrittled the material significantly, causing the elongation at break to drop from 10 to 1%. With the addition of 0.1 wt.% carbon, both strength and elongation at break are higher than the reference weld without carbon addition (Mereddy et al. 2018).

3.3 Additive Repair by Wire and Arc Additive Manufacturing

The first test series showed a structure stretching in the build-up direction. The β-grains partially grew over several layers and reached lengths of up to 15 mm. The microstructure appeared martensitic, cf. Figure 12. The tensile strength achieved ranged between 880 and 910 MPa, depending on the direction, cf. Figure 13. The elongation at fracture was also similar in all directions and, at least 6%, was above the values of the inoculated specimens and in the range of the specimens welded with flux. Although the grains grew predominantly in the build-up direction, the mechanical properties did not show any pronounced anisotropy. This is due to the fact that the cooling rate was reduced by the high interlayer temperature of 300 °C. As a result, the proportion of energy dissipated via the substrate decreases and the proportion dissipated via radiation and convection increases.

Fig. 12
2 micrographs of the 2 first test specimens, one transverse and one longitudinal. On the left, it represents a rough surface along with horizontal asymmetric lines. On the right, it represents horizontal asymmetric lines on a rough strip like surface.

Micrographs from the first test; build-up direction is upwards; a shows the longitudinal micrograph; b shows the transverse micrograph

Full size image
Fig. 13
A grouped bar graph with error bars plots stress and strain versus directions for the first survey. 45 degrees has the highest tensile strength and 0.2% offset yield strength followed by 0 and 90 degrees. 0 degrees has the highest elongation followed by 90 and 45 degrees.

Yield strength, tensile strength and elongation at break of the first survey; 0° is in the welding direction, 90° in the build-up direction and 45° is the diagonal between the welding direction and the build-up direction; error bars show the highest and lowest value obtained for the specimens

Full size image

In the second investigation, the cooling rate was increased to achieve grain refinement. Therefore the dwell time was increased to reduce the interlayer temperature. Figure 14 shows the longitudinal and transverse sections. The microstructure shows the α’ martensite. It can be clearly seen that the former β-grains have grown in the build-up direction. They reach lengths >20 mm. Due to the long cooling periods between the layers of 120 s, welding was performed with a significantly lower interpass temperature than in the first test. Cooling took place mainly via the substrate, as is usual in welding. The preferred growth direction of the grains can also be clearly seen in the mechanical properties, see Fig. 15. In the 45° direction, a tensile strength of 1050 MPa was achieved. In the other two directions, about 10% less was reached (940 MPa). In the 45° direction the ductility was less than in the other two directions, i.e., 7.5% versus 11% and 16%. The higher cooling rate led to smaller α-grains, which increased the strength. Compared to the first series of tests, the increase in cooling rate for the second series of tests increased the strength in all directions above the maximum of the first series of tests. The ductility was similar.

Fig. 14
2 micrographs of the 2 second test specimens, one transverse and one longitudinal. On the left, it represents a smooth surface along with thin horizontal asymmetric lines. On the right, it represents horizontal asymmetric lines on a smooth strip like surface.

Micrographs from the second test; a shows the longitudinal micrograph; b shows the transverse micrograph

Full size image
Fig. 15
A grouped bar graph for the second survey has error bars and plots stress and strain versus directions. 45 degrees has the highest tensile strength and 0.2% offset yield strength followed by 0 and 90 degrees. 90 degrees has the highest elongation followed by 0 and 45 degrees.

Yield strength, tensile strength and elongation at break of the second survey; 0° is in the welding direction, 90° in the build-up direction and 45° is the diagonal between the welding direction and the build-up direction; error bars show the highest and lowest value obtained for the specimens

Full size image

3.4 Local Nitriding

Nitrogen was introduced directly after solidification of the molten pool of a TIG weld. Figure 16 shows the hardness profile and a micrograph of a nitrided sample. A nitride layer of 10 to 20 μm is clearly visible as a white boundary layer. This is because nitrogen promotes the formation of the α-phase of titanium. Hardness values of over 1500 HV0.025 were achieved in the boundary layer. The hardness measurement shows that the diffusion depth was approx. 100 μm. Although the local treatment can achieve a significant hardening of the surface, stress is also induced. These stresses can lead to a reduction in strength or result in distortion. Fatigue tests on locally atmospherically plasma-nitrided specimens have shown that a hardness of over 1000 HV0.025 can be achieved. However, the heating required for this leads to distortion, so that the specimens break prematurely in fatigue tests. To reduce the induced stress that lead to distortion, the energy input can be reduced, resulting in lower surface temperatures. Lower temperatures reduce the nitrogen diffusion resulting in lower surface hardness or longer lead times. With similar fatigue and processing time a surface hardness of 800 HV could be reached.

Fig. 16
A line graph of hardness H V 0.025 versus distance from the surface in micrometers. It plots a downward line, which starts at (8, 1580), falls, and ends below the horizontal dashed line of limit hardness at (120, 380). Values are estimated.

Hardness profile of a locally atmospherically nitrided sample; limit hardness is 410 MPa (Gushchina et al. 2020); the nitriding hardness depth is approximately 110 µm

Full size image

4 Conclusions

Both inoculants and fluxes can promote the mechanical properties of the overall joint. They can also be combined, see Fig. 17 the narrow seam with small HAZ is clearly visible. The carbide precipitates also clearly distinguish the FZ from the HAZ.

Fig. 17
A micrograph of a joint represents combined inoculation and flux. At the center, a bright area for a narrow seam is observed between inoculation and flux.

Combined inoculation and flux effect

Full size image

The advantages of the inoculation and flux effect can also be used at least to some extent in additive repair. Grain refinement can also be achieved in additive manufacturing by adding carbon (Mereddy et al. 2018). On one hand, grain refinement helps to improve mechanical properties, and on the other hand, its use reduces the anisotropy of the microstructure (Mereddy et al. 2017). The effect of the flux causes the weld to become narrower. In addition, the average temperature of the molten pool can be reduced without causing bonding defects. The overheated material flows in- and downward rather than outward, preventing bonding defects. Arc stability is reduced by the use of flux. Especially when igniting above the flux layer, misfiring can occur. The arc needs enough energy to melt the flux as well, otherwise the arc jumps and welding defects can occur. The application with spray delivered reproducible results. Small inoculation amounts of 0.1 wt% C as Mereddy had used (Mereddy et al. 2018), are indeed also achievable with spray. But reproducibility is no longer guaranteed. As a result, some areas may reach higher carbon contents, resulting in degradation of mechanical properties. In addition, the spray process for applying inoculant or flux takes place outside the inert gas atmosphere. In order to be able to use the advantages of the inoculant and flux reproducibly with WAAM, the inoculant and flux must be applied differently.

The investigations have shown that the inoculant also can be dissolved during welding. During cooling, the precipitates are formed, which are available as heterogeneous nuclei. This means that no refractory compounds are needed for grain refinement, only the carbon (Langen 2021; Mereddy et al. 2018). Thus, inoculants and fluxes can also be fed directly into the process zone via gases.

Using gas the addition of the inoculant and flux can be carried out internally in the process via the shielding gas in the shielding gas atmosphere. Furthermore the amount of inoculant and flux can be adjusted more precisely and homogeneously.

Local atmospheric nitriding is one way to significantly increase the surface hardness of titanium components. However, one has to consider that the fatigue strengths might be affected by potential distortion. For cyclically loaded components, surface hardening to 800 HV can be achieved without degrading fatigue properties. For structural components not subjected to vibration, hardnesses in excess of 1500 HV can be exploited.

This subproject of CRC 871 is the link between the decision-making process of a possible repair and the repaired recontoured titanium component. The improved repair methods of titanium components developed in this subproject can further increase the mechanical properties of the repaired component. The determination of the achievable mechanical properties can be led back. Thus, they can be taken into account in a decision-making process preceding the repair and prevent unnecessary or inadequate repairs.