Influence of the process speed in laser melt injection for reinforcing skin-pass rolls

Skin-pass rolling is the last process in a rolling line. In this process, the final sheet thickness and the surface properties are set. For improving the formability and paint adhesion of rolled sheets, specific textures are formed into sheets by textured skin-pass rolls.¹ Until now, there are at least six different technologies for texturing the skin-pass rolls that can be divided into stochastic and deterministic technologies according to the resulting texture. Stochastic textures can be generated by shot blast texturing, electrodischarge texturing, and precision texturing.² Shot blast texturing and electrodischarge texturing roughen the roll surface. In precision texturing, a chromium-based coating with spherical nubs is applied. Deterministic textures can be generated by laser texturing or electron beam texturing. With these technologies, usually, small craters are created by local melting of the roll surface. Another way of laser texturing is the generation of implants with titanium diboride.³ Each implant is generated by laser melt injection (LMI) using titanium diboride powder with an average powder diameter of 4 μm as the hard material. This two-step process consists of predepositing the titanium diboride powder with an organic binder and laser-induced implantation. However, the deterministic generation of implants with LMI is a time-consuming process.

Therefore, high-speed laser melt injection (HSLMI) for generating near-net shape metal matrix composite (MMC) layers on the rolls with high productivity has been developed. In this one-step process, wear-resistant MMC layers featuring a stochastic texture can be generated.

Until now, the applications of LMI are strongly limited due to low process speeds and low productivity. Usually, the process speeds in LMI are below 2 m/min.⁴ In contrast, process speeds between 5 (Ref. 5) and 200 m/min (Ref. 6) are possible in laser cladding. However, when adding hard particles to the cladding feedstock, the processed area per unit time is limited to 0.5 (Ref. 7) to 12 m/min when a laser spot with a diameter of 2 mm and an overlapping degree of 50% are used.⁸ The powder particles are transported by a feeding gas toward the weld pool. Usually, gas flow rates below 5 l/min are used.⁹ 

As hard materials for reinforcing metallic surfaces by LMI, metallic hard materials are used in particular. These are carbides, borides, and nitrides of transition metals of the IVa-, Va-, und VIa-group in the periodic table.¹⁰ Due to a poor wetting behavior, nitrides are less important for LMI.^11,12 Among the carbides, spherical fused tungsten carbide (SFTC) is a very suitable hard material for LMI due to a high hardness of up to 3100 HV0.1 and a good wetting behavior.¹³ However, it is well known that fused tungsten carbide particles partially degrade in welding-related processes. A degradation seam around the particles is formed in nickel-¹⁴ and cobalt-¹⁵based matrices. Furthermore, in LMI, the SFTC particles can partially melt and lose their spherical shape due to interactions with the laser beam. In lubricantfree deep drawing, a precise melting of SFTC particles can be used to generate fused tungsten carbide nuggets that reduce the contact pressure.¹⁶ 

For increasing the process speed, a model for the particle movement in LMI has been designed and analyzed, see Fig. 1. The distance travelled by the particle from the powder nozzle to the weld pool can be divided into s₁, s₂, and s₃. s₁+ s₂ is the distance from the powder nozzle to the surface of the weld pool, whereas s₃ is the incorporation distance that a particle travels from the first touch with the weld pool’s surface until full incorporation. The time a particle needs for travelling s₁+ s₂ is described by

v_1,2 describes the average particle velocity over s₁+ s₂. v_1,2 mainly depends on the volume flow rate of the feeding gas $V˙feedinggas$ and the weight force F_g,

The time a particle needs for being incorporated into the weld pool is described by

v₃ describes the average particle velocity over s₃. In contrast to v_1,2, v₃ depends on the surface tension force F_s, the drag force of the weld pool F_d, the weight force F_g, and the initial kinetic energy of the particle E_kin,

E_kin depends on v_1,2,

Consequently, v_1,2 has a major influence on both t_1,2 and t₃.

A particle interacts with the laser beam during s₂ and s₃. For HSLMI, experiments showed that high laser intensities of up to 353 kW/cm² are necessary to maintain a continuous weld pool at high process speeds. Consequently, strong interactions with the laser beam lead to partial melting and undesired deformations of particles. For avoiding this drawback, a new approach for LMI with a significantly higher particle velocity v_1,2 was taken.

The substrate was made of 1.2362 (X63CrMoV5-1), which is a cold work steel typically used for rolls. SFTC particles (Oerlikon MetcoClad 52001) featuring a grain fraction between 45 and 106 μm were injected into the 1.2362 substrate by LMI. SFTC is an eutectic material of 20%, …, 27% WC and 73%, …, 80% W₂C.¹⁷ The steel 1.3505 (100Cr6) was selected as a material for the counter bodies in the wear tests.

The laser beam of a disk laser (Trumpf TruDisk 12002) with a wavelength of 1030 nm was guided to a processing optic (Trumpf BEO D70) by an optical fiber with a diameter of 200 μm. The collimating lens of the processing optic provided a focal length of 200 mm, whereas the focusing lens provided a focal length of 300 mm. The processing optic was carried by a six-axis robot (Reis) that moved the processing optic parallel to the rotational axis of the workpiece. The workpiece was positioned 15 mm below the laser focus obtaining a laser spot with a diameter of 2 mm. A three-jet powder nozzle with a working distance of 16 mm was used. The SFTC particles were transported to the powder nozzle by a powder feeder (GTV PF 2/2). Argon was used as the feeding gas and as the shielding gas. The overlapping degree was set to 50%.

The MMC specimens were cut by an EDM machine (Mitsubishi MV 1200 V) and hot mounted into Struers LevoFast. After grinding the mounted specimens with granulation from P320 to P1000, they were polished by diamond suspension featuring a particle size of 3 μm and by a silica suspension featuring a particle size of 0.04 μm. For investigating the microstructure of the MMC layers, the samples were etched by Beraha I. Micrographs were taken with a light microscope (Zeiss AX10). The software ImageJ was used for measuring the layer thickness, the SFTC particle content, the cumulated crack length, the porosity, and the width of the degradation seam. For measuring the SFTC particle content and the porosity, an ROI with an area of 1.8 mm² was analyzed for MMC layers with a layer thickness up to 400 μm and two ROI with an overall area of 3.6 mm² were analyzed when the MMC layer thickness exceeded 400 μm. The cumulated crack length was defined as the sum of all crack lengths in an ROI of 2 mm² within the MMC layer. Only cracks within the 1.2362 matrix were considered. Furthermore, a hardness tester (Struers DuraScan 50) was used for measuring the Vickers hardness HV 0.1.

After LMI, the workpieces were machined by cylindrical grinding. Wear samples were cut out of the cylindrical workpieces by EDM. The counter bodies (pins) had a diameter of 10 mm and a spherical cap. Figure 2 shows the schematic setup for wear tests. For conducting wear tests, a tribometer (CETR UMT 3) was used. The tribometer was placed inside a box in which the temperature and the humidity could be controlled. The temperature was kept at 25 ± 0.1 °C and the humidity between 50% and 55%. For all tests, the frequency for the oscillating movement was 5 Hz and the distance between the turning points was 10 mm, see Fig. 2, leading to an average test speed of 100 mm/s. A test force of 30 N was applied over a test time of 4 h. For each LMI process speed, three tests were conducted. For determining the wear volume, the weight loss of the samples and the pins was measured with a high-precision scale (Shimadzu AUW120D, reading accuracy d = 0.1 mg). The calculation of wear volumes was based on combined densities considering the FTC particle content. For analyzing the wear tracks, an SEM (Zeiss EVO) and an EDX detector (Bruker Quantax) were used. For quantifying the material content of each area of interest, the average from 20 single measurements was used.

The particle velocity v_1,2 can be adjusted via the volume flow rate of the feeding gas. The correlation is shown in Fig. 3. It was found that the particle velocity v_1,2 needs to be increased to 9 m/s for avoiding undesired deformations of SFTC particles, see Fig. 4. For this, the volume flow rate needs to be increased from 5 to 15 l/min. Negative effects due to the increased volume flow rate of the feedings gas were not observed.

With a particle velocity v_1,2 of 9 m/s, the process speed of LMI could be increased up to 70 m/min with a laser power of 6.7 kW and up to 100 m/min with a laser power of 10 kW. A process speed of 100 m/min corresponds to a processed area per unit time of 1000 cm²/min. With HSLMI, near-net shape MMC layers featuring a small layer thickness can be generated, see Fig. 5. The interface between the SFTC particles and the 1.2362 matrix shows a good metallurgical bonding. Only a few particle deformations were detected. Furthermore, cracks within the 1.2362 matrix were detected that also run through SFTC particles. The cracks are mainly oriented perpendicular to the weld seam.

The layer thickness and the SFTC particle content depending on the process speed are shown in Fig. 6. The layer thickness decreases degressively with increasing process speed from 1 mm at a process speed of 10 m/min to 0.1 mm at a process speed of 70 m/min. However, the waviness of the MMC surface due to overlapping weld seams decreases significantly with the process speed. The SFTC particle content scatters between 14 and 30 vol. % with no clear correlation to the process speed.

Cracks and pores were detected at all investigated process speeds. Both the porosity and the cumulated crack length decrease degressively with increasing process speed, see Fig. 7. The porosity can be reduced from 6 to 1 vol. % and the cumulated crack length from 6 to 1 mm by increasing the process speed from 10 to 70 m/min. At a process speed of 10 m/min, the porosity and the cumulated crack length are significantly higher than at higher process speeds.

Furthermore, the process speed has a major influence on the microstructure of the MMC layer. A heterogenous microstructure consisting of martensite regions (etched blue to brown) and retained austenite regions (white) was found at all investigated process speeds between 10 and 70 m/min. The austenite regions feature a fine dendritic structure as shown in Fig. 8. Between a process speed of 10 and 50 m/min, the martensite grains feature a rounded shape, see Fig. 8. The SFTC particles are strongly degraded. When the process speed is further increased to 60 and 70 m/min, the microstructure of the martensite regions changes to a needle-shaped structure, see Fig. 9. The degradation seam of the SFTC particles is significantly smaller than at lower process speeds. A drop in hardness can be found both for the white regions and for the brown/blue regions of the 1.2362 matrix when increasing the process speed from 10 to 70 m/min, see Fig. 10. The average hardness of the white regions decreases from 907 to 612 HV 0.1, whereas the average hardness of the brown/blue regions decreases from 1078 to 795 HV 0.1. In contrast, the SFTC particles did show no significant drop in hardness. However, the hardness drops from 2853 HV 0.1 in the SFTC particles to 973 HV 0.1 in the SFTC degradation seam at a process speed of 10 m/min. No hardness could be measured for the SFTC degradation seam at a process speed of 70 m/min because the degradation seam width is too small.

The SFTC degradation at a process speed of 10 m/min was analyzed by EDX. Figure 11 (right) shows that the degradation seam is rich in tungsten and iron meaning that parts of the SFTC particles dissolve in the surrounding 1.2362 matrix. The average tungsten content within the SFTC degradation seam is 36.8 wt. %. The dependency of SFTC degradation from the process speed was investigated from 10 to 70 m/min. Figure 12 shows that the width of the degradation seam decreases degressively with increasing process speed. By increasing the process speed from 10 to 70 m/min, the width of the SFTC degradation seam can be reduced from 17 to only 3 μm.

For determining the wear resistance of the MMC layers, oscillating wear tests were carried out. Figure 13 shows the wear volume of nonreinforced and SFTC particle reinforced samples generated at different process speeds and the wear volume of the corresponding pins. The SFTC reinforcement of the samples leads to a reduction in wear volume from 3.6 to 0.1 to 0.3 mm³. The pins wore out stronger than the samples in all four configurations.

The wear tracks of the nonreinforced samples featured grooves and rugged areas, see Fig. 14 (left). In addition, small abrasive particles were attached to the surface of the wear track. The EDX mapping of the nonreinforced wear track indicates that parts of the attached abrasive particles derive from the pin since the particles are poorer in chromium than the main area of the wear track, see Fig. 14 (right). A quantitative analysis showed an average chromium content of 8.1 wt. % on the main area of the wear track and an average chromium content of 4.4 wt. % on the abrasive particles.

On the MMC wear track, only small signs of wear were detectable, see Fig. 15 (left). Both the SFTC particles and the 1.2362 matrix featured a slightly rugged topography. The EDX mapping of the MMC wear track indicates that parts of the SFTC particles are covered with the matrix material since small regions within the SFTC particles are rich in iron, see Fig. 15 (right).

It was shown that a high SFTC particle velocity v_1,2 of 9 m/s is necessary to enable HSLMI. By increasing the particle velocity v_1,2, both the travel time t_travel and the incorporation time t_{incorporation} can be reduced. Consequently, interactions with the laser beam are reduced as well and the initial geometry of the SFTC particles is preserved.

For explaining the degressive decrease in the layer thickness, two opposite influences need to be considered. On the one hand, the energy per unit length decreases with the process speed, which has a reducing effect on the layer thickness. On the other hand, the process efficiency increases with the process speed since the thermal losses are reduced, which has an increasing effect on the layer thickness.¹⁸ The MMC layers obtained at high process speeds feature a small layer thickness and a small waviness, which reduce the effort for final machining of SFTC reinforced parts significantly. Due to a high difference in hardness between the SFTC particles and the 1.2362 matrix, the final machining is challenging.

Another advantage of HSLMI is the significant reduction in the cumulated crack length and the porosity. At high process speeds, a low energy per unit length leads to a reduction in the heat input and accordingly to temperature-related defects. Furthermore, high process speeds result in a reduction in SFTC degradation. At high process speeds, the lifetime of the melt pool is significantly smaller, which reduces the time for SFTC degradation. The degradation process is based on diffusion that is time- and temperature-dependent.¹⁹ In order to preserve the desired material properties, the SFTC degradation should be minimized, especially since there is a significant drop in hardness in the degradation seam.

However, the influence of process speed on wear volume is negligible. Consequently, the differences in the microstructure in terms of the SFTC degradation seam and the matrix structure seem to have no decisive impact on the wear resistance. The rugged areas on the wear tracks are an indication for adhesion wear. However, the dominant wear mechanism was abrasion. Abrasive particles were attached to the surfaces of the wear tracks since the abrasive particles that are formed during the wear test were not removed during the tests.

HSLMI has been introduced. It was found that the particle velocity is a key factor for enabling LMI at high process speeds. With HSLMI, process speeds of up to 100 m/min can be reached.

The process speed has a major influence on the MMC microstructure. The cumulated crack length and the porosity within the 1.2362 matrix decrease when increasing the process speed. The SFTC particles show a degradation seam. The SFTC degradation can be reduced by increasing the process speed.

By reinforcing the steel 1.2362 with SFTC particles by HSLMI, the wear resistance is improved significantly. However, the influence of the process speed on the MMC microstructure has no decisive impact on the wear resistance.

The authors gratefully acknowledge the support of this work by Deutsche Forschungsgemeinschaft (DFG) within the project 424737129.

The authors have no conflicts to disclose.

Philipp Warneke: Conceptualization (equal); Data curation (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). Annika Bohlen: Supervision (equal). Thomas Seefeld: Supervision (equal).