SX-717

Two Cr-based SX-717 laser coatings on mild steel with different magnitudes of dilution were tested. Coating, which Fe content was 6–7 wt.% survived at least 32 cycles similar to those shown in Figure 158. After 32 cycles the test was stopped. Coating with lower amount of Fe (1–2 wt.%) survived at least 50 cycles. After 50 cycles the test was stopped. Both of these coatings were initially defect-free except several micropores in interdendritic regions as was shown in Figure 78b in section 3.3.1.2. Nominal CTE for this alloy was 11.7 x 10^-6 K^-1 (at 100°C), which does not deviate much from CTE of mild steel. It was also important to note that micropores in interdendritic regions did not act as stress raisers and initiate cracks as reported in Ref. [311].

0 100 200 300 400 500 600 700 800 900 1000

3000 8000 13000

Time (s)

Temperature (ºC) ^{T/C-K 1}

Figure 158. Thermal cycles encountered by Inconel 625 laser coating on mild steel. Average peak temperature was 685ºC. Heating period from 18ºC to peak temperature lasted approximately 70 s. Cooling period from the peak temperature back to 18ºC lasted approximately 60 s.

3.7.3 Tribaloy T-800

Initially defect-free T-800 laser coatings were tested on mild steel and austenitic SS. Coating on mild steel was more heavily diluted (28–29 Fe wt.%) than coating on austenitic SS (3–4 wt.% Fe). Despite high dilution and low mismatch in CTEs (T-800: 12.6 (20-500°C) vs. Fe52:

12.7 (20-500°C)), T-800 on mild steel survived only 1 cycle but cracked severely during the next 5 cycles. Similarly, low diluted T-800 on austenitic SS survived only 1 cycle but cracked severely during the next 5 cycles. In the latter case, cracks perpendicular to coating/base material interface observed on transverse cross-section continued to base material, which was initially in tension as was shown in Figure 156.

4 DISCUSSION

In this chapter, the obtained results are discussed separately for back reflection measurements, wet corrosion, hot corrosion, abrasion wear, sliding wear, residual stress and thermal fatigue studies.

4.1 Back reflection characteristics

As the traditional blown powder laser cladding process suffers from low productivity and cost efficiency in large area applications, one way to overcome this is to use high power levels of short-wavelength laser beam produced by relatively cheap and efficient laser sources. Direct high power diode lasers (HPDL) fulfill these criteria well. Despite short-wavelengths, energy couplings in liquid state surface treatments including cladding are typically rather low as discussed in section 1.3.2. For instance, in Nd:YAG cladding 50% of the delivered laser power reflects off the melt pool and powder cloud [20]. For this reason, using high levels of power, back reflection targeted to laser itself and cladding tool should be considered. Direct HPDLs, particularly, are vulnerable to back reflection due to their open structure. There is clear path for the reflected beam to enter, for instance, inside the laser head in various surface treatment processes.

The studies conducted here were carried out to define the direction of back reflection in 1-step blown powder laser cladding process while using direct HPDL equipped with off-axis or coaxial type powder feeding nozzles. In-house built temperature monitoring system was utilized in measurements. It measured temperatures on several locations including prisms inside the laser head. In cladding experiments, the coating alloy was Stellite 12, which does not usually oxidize but remains rather shiny and smooth during processing. As the oxidation improves absorption, cladding conditions in this study can be considered harmful for laser head and nozzles because high reflection was expected. According to temperature measurements, it was noted that despite rather long working distance of ~0.5 m, back reflection indeed increased the temperatures inside the laser head while using laser power of 4.8 kW. The amount of back reflection inside the laser head depended on the incidence angle of laser beam with respect to the surface of the melt pool. As the leading edge of the melt pool, which shape depends on factors like bead height, melt pool length and perhaps the melt viscosity, reflects the beam forward towards the direction where cladding proceeds, laser head should be tilted towards “pushing” direction in order to avoid additional heating inside the laser head. This was particularly true for single bead cladding. It was proved that tilting the laser head towards the “pushing” direction as much as 5.0°, temperature increase inside the laser head due to back reflection was fully eliminated. This was obtained while using coaxial type cladding nozzle. Off-axis cladding configuration was not tested in “pushing” position.

During large area coating application, where the beads were deposited side by side with 50%

overlapping, heating inside the laser head diminished because the shape of the leading edge of the melt pool changed. Due to 50% overlapping bead height increased and melt pool inclined directing the reflected beam more towards the direction where the inter-track advance proceeded. This result was obtained while using off-axis nozzle. Coating thickness was 1.7 mm in these large area experiments.

If two different nozzles, off-axis and coaxial type, are compared in single bead cladding, with used parameters and bead thickness, the coaxial type nozzle generated conditions which caused less back reflection inside the laser head. This was obviously due to larger amount of

material between the melt pool and laser head to block the back reflection. As the coaxial type nozzle provided better shielding for the melt pool than off-axis nozzle, the total amount of back reflection was probably higher while cladding with coaxial type nozzle. This also speaks for the more efficient blocking of back reflection in case of coaxial type nozzle.

4.2 Wet corrosion properties

Wet corrosion studies consisted of open circuit potential (OCP) and potentiodynamic cyclic polarization measurements both in aqueous 3.5 wt.% NaCl solution at room temperature (RT). OCP measurements proved that defect-free Inconel 625 laser coatings were true corrosion barrier coatings, which restricted the less noble mild steel base material perfectly from the surrounding environment. Plasma transferred arc (PTA) overlay welding produced also impervious coating layers but high-velocity oxy-fuel (HVOF) sprayed coating allowed the solution to penetrate into the base material and corrode it quickly. Coating itself was corroded selectively since areas adjacent to oxidized splat boundaries were degraded. Dent et al. [401] noticed similar degradation in HVOF sprayed NiCrMoB coating exposed to 0.5 M H2SO4, especially when significant oxidation occurred during spraying. According to Edris et al. [402], these areas suffered from Cr depletion since Cr was oxidized predominantly to Cr2O3 but also to spinel NiCr2O4. Consequently, local galvanic pairs were generated inside the coating structure. Another galvanic pair was generated between the coating and base material due to established electrolyte connection. Since the NiCrMo alloy occupies much higher position than Fe in the galvanic series in seawater [348] (as well as in 3.5 wt.% NaCl as shown by OCP measurements in Figure 49), the former became cathode and the latter anode, i.e. coating was cathodically protected, which accelerated the base material corrosion further. Corrosion performance of HVOF coating was, however, improved significantly by subsequent laser remelting. With optimal parameters, perfect corrosion barrier coatings with low dilution and fusion bond were produced. In addition, undesired oxide layers in the as-sprayed structure disappeared probably by floating and accumulating at the surface during solidification. One of the benefits of this 2-step process was that relatively thin coatings (0.3-0.4 mm) with low dilution and fusion bond could be obtained. Corrosion barrier coating in aqueous environments does not necessarily have to be very thick and the production of thin and low diluted coating by 1-step method is difficult. With the inter-track advance of 8-9 mm, i.e. with overlapping of approximately 10-20%, and traverse speed of 1900 mm/min, coverage rates of 0.9-1.0 m²/h were reached with a 4 kW Nd:YAG laser. This rather high coverage rate originates exclusively from the low thickness since melting efficiencies, 17-22%, for this process was not any spectacular compared for instance with 26-33% calculated for 1-step HPDL cladding. In these calculations, the melt pool temperatures were assumed to be between 1350-2000°C. Compared with thicker coatings produced by 1-step method, thinner laser remelted coatings were, however, more susceptible to coating defects. Interconnected paths were formed in overlapped areas due to too cold parameters and/or lack of overlapping.

With too hot parameters single interconnected pores formed even on the centre of remelted tracks. Such single interconnected paths were never encountered in thicker Inconel 625 coatings produced by 1-step cladding on mild steel. Despite their low frequency of occurrence and small size in diameter in plane parallel to coating/base material interface, transverse cross-sections prepared from the exposed coatings revealed that these singe interconnected pores lead to severe and rapid deterioration in base material mainly via mechanism familiar from crevice corrosion.

In general, these exposure tests were not very harsh for Inconel 625 alloy. For this reason, there could not be detected any differences in behaviour between wrought, PTA overlay

welded and laser coatings in spite of remarkable differences, for instance, in dilution and microsegregation. PTA coating was exception since its OCP curve showed clearly lower potentials compared with other impervious coatings. This was due to significantly higher Fe content in coating in consequence of dilution.

In cyclic polarization measurements, far higher amount of alloys produced by different methods were tested. These tests revealed clearer differences in their corrosion performance than in OCP measurements. The results are discussed here starting from the Ni-based alloys.

Ni-based alloys tested comprised Inconel 625 and Alloy 59. Inconel 625 coatings were produced with 1-step HPDL cladding, PTA overlay welding, HVOF spraying and laser remelting of HVOF sprayed coating. Wrought Inconel 625 was used as reference material.

Cyclic polarization curves and microstructural studies indicated that impervious coatings outperformed clearly the HVOF sprayed coating for the same reasons as explained already earlier. HPDL clad coatings, which exhibited higher amount of Fe in consequence of dilution, macrosegregation of Fe and microsegregation of Mo and Nb than wrought alloy, was inferior to wrought alloy. This was evidenced by the microstructural studies and characteristics extracted from the polarization curves. For instance, HPDL clad coating, which exhibited the lowest amount of Fe (6.0 wt.%) among HPDL coatings, showed lower repassivation (Erp) potential than wrought alloy, which contained 3.6 wt.% Fe. In addition to this, characterization of tested surfaces revealed that HPDL coating suffered from slight crevice corrosion under the gasket, preferential dissolution of macrosegregated areas and dendrite cores, which were depleted in Mo and Nb. All these characterized aspects were absent in wrought alloy. The difference in performance between HPDL coatings and wrought alloy increased when the dilution of the coating increased. More heavily diluted laser coatings (9.4 and 19.4 wt.% Fe) exhibited lower Eb and Erp as well as more severe crevice corrosion under the gasket. Evidently, Fe was very detrimental to crevice corrosion resistance of Inconel 625 alloy. As the initial Fe content of the powder was 1.2 wt.%, the compositional dilution of the coating, which included 6.0 wt.% Fe, was approximately 5%. In 1-step HPDL cladding even lower compositional dilutions were obtained without loosing fusion bond as evidenced in hot corrosion studies where compositional dilutions were approximately 2%. This kind of extremely mildly diluted HPDL coating would have managed better in polarization tests than coatings studied here. On the other hand, considering real applications, it would be safer to use little bit too hot parameters rather than too cold ones to secure fusion bond. As these HPDL coatings were manufactured with rather low power density and high interaction time rather high microsegregation of Mo and particularly Nb to the interdendritic regions took place. This microsegregation behaviour was in good agreement with the results presented by Tinoco [384], who reported on segregation of Mo and Nb to interdendritic regions and Cr to dendrite cores in Inconel 625. This microsegregation was, however, not so severe that it would have caused initiation of corrosion pits in areas depleted in Mo, which is often reported in the context of arc welded alloys exposed to chloride bearing environments.

Microsegregation can be, however, diminished by using higher traverse speeds and shorter interaction times. Laser remelted Inconel 625 characterized with low dilution and microsegregation showed behaviour equivalent to wrought alloy. These two samples could be still tested in harsher environments to verify their equivalent resistance. Whitney et al. [403]

noticed recently that laser clad Ni-based superalloys (Inconel 625, Alloy 59 and C-276) exhibited lower pitting resistance at elevated temperature in chloride bearing environments as compared to the same alloy in wrought condition. This was attributed to the microsegregation, which was absent in wrought alloys.

Co-based alloys subjected to cyclic polarization measurements included Stellite grades 21 and 6. Stellite 21 was manufactured by Nd:YAG laser cladding and PTA overlay welding. Stellite 6 was prepared by Nd:YAG cladding and HIPping. According to data extracted from the polarization curves, low diluted Stellite 21 laser coating (1.3 wt.% Fe) exhibited behaviour similar to wrought Inconel 625 and low diluted Inconel 625 and Alloy 59 laser coatings. Erp

was even better than that for wrought Inconel 625. It was also immune to crevice corrosion under the gasket and pitting corrosion in used conditions. SEM studies, however, revealed preferentially dissolved areas in micro-level, where Cr- and Mo-rich interdendritic regions in consequence of microsegregation dissolved considerably less than dendrite cores. As opposed to Inconel 625, chromium tended to segregate to the interdendritic regions instead of dendrite cores. Similar Mo segregation in laser clad Stellite 21 was reported in Refs. [186, 361] but segregation of Cr was not mentioned. When the both key elements Cr and Mo segregate to the interdendritic regions, this makes dendrite cores exceptionally vulnerable to pitting. In fact, PTA coatings which suffered from higher dilution and microsegregation than laser coating suffered from pitting corrosion.

Compared with almost equally diluted Stellite 6 laser coating (1.6 wt.%), the used cyclic polarization conditions did not reveal any major differences. According to Rogne et al. [183], Stellite 21 has higher critical crevice corrosion temperature in chloride bearing solutions than equally diluted Stellite 6, which does not contain Mo. It could be anticipated that dendrite cores in Stellite 6 become susceptible to pitting not only due to lack of Mo but also due to decreased Cr content because part of the Cr is bound to interdendritic carbides. HIPped Stellite 6 (1.0 wt.% Fe) exhibited even lower Cr content in the matrix than laser coating in dendrite cores (~20.8 vs. 23.9 wt.%). Nevertheless, data extracted from the polarization curves did not show any major differences. Erp was, however, higher for laser coating (+483 vs. +361 mV). As the HIPped Stellite 6, closely reminding the one studied here, showed in Ref. [404] significantly lower current densities and higher Eb than cast Stellite 6 in 3.5 wt.%

NaCl polarization tests at RT, it can be assumed that Stellite 6 laser coating would be far better than corresponding cast alloy.

4.3 Hot corrosion properties

Hot corrosion studies revealed that Ni and Cr in the tested alloys built up rather dense corrosion products, nickel ortho- (Ni3V2O8) and chromium (CrVO4) vanadates, on the exposed surfaces in comparison with Fe in Fe-based wrought alloy. These continuous and compact oxide layers, particularly CrVO4, clearly reduced the diffusion of oxygen and other agents from the Na2SO4-V2O5 salt and environment inwards to the metal resulting in the reduction of the kinetics of the anodic reactions (Ni -> Ni²⁺ + 2e^-, Cr -> Cr³⁺ + 3e^-) at the alloy/salt and cathodic reactions in molten salt (for instance S2O72- and VO3- are reduced [327]) and/or at the salt/atmosphere interface (½O2 + 2e^- -> O^2-). As the melting temperatures of Ni3V2O8 and CrVO4 are 1220 and 810°C, respectively [386], they might have remained solid underneath the regularly added salt layer since each salt addition involved cooling down the sample. These reactions were not, however, fully prevented since even with one of the best alloys studied here, SX-707 HVOF exposed to salt (Na:V = 0.22), deterioration was more like accelerating rather than constant or decelerating as a function of time as was shown in Figure 75b. It could have been expected that increase in the corrosion product thickness would have decreased the corrosion rates despite regular salt addition.

If trying to describe what actually may have happened on the surfaces of the alloys in the test, initially, Ni- and Cr-based alloys were covered with thin but dense protective oxide layers, which were formed after grinding procedure since the oxygen activity (= partial pressure, PO2) in air at RT (PO2 = 0.2 atm) readily exceeded the equilibrium values to oxidize Ni to NiO (PO2

= 10^-74 atm, 2Ni + O2 = 2NiO, ΔG = -423 kJ) and Cr to Cr2O3 (PO2 = 10^-124 atm, 4/3Cr + O2 = 2/3Cr2O3, ΔG = -705 kJ). In addition to these, protective spinel oxides like NiCr2O4 may have formed (PO2 = 10^-112 atm, 1/2Ni + Cr + O2 = 1/2NiCr2O4, ΔG = -636 kJ). It is also possible that protective oxide layers developed further on the top of test alloys in the beginning of the hot corrosion test. At the test temperature of 650˚C in reaction chamber, oxygen activity should have been more than 10^-18 atm for Ni to form NiO, 10^-30 atm for NiCr to NiCr2O4 and 10^-34 atm for Cr to form Cr2O3. Equilibrium oxygen activities at 650˚C were obtained from the phase stability diagrams of ternary Ni-O-S and Cr-O-S systems calculated with

“Outokumpu HSC Chemistry® 4.0 for Windows” –software (Figure 159). In the reaction chamber atmosphere at 650˚C, the oxygen partial pressure was presumably higher than the pressure needed to form protective oxide layer at least in the case of Cr, particularly when the air was constantly blown to the reaction chamber and elevated temperature enhanced the reaction rates.

Another question is, was the oxygen activity high enough beneath the molten salt. As Lai explained [405], the reaction kinetics of hot corrosion is often characterized by an initial incubation state with a relatively low rate of reaction and a later stage with rapid materials degradation by oxidation and/or sulfidation. In other words, the initial stage involves frequently a formation of a protective oxide layer of reaction products on the metal. On the other hand, oxygen activity may have been locally too low at the metal/salt interface to form the protective oxide layers. In fact, molecular oxygen solubility in Na2SO4–V2O5 salt is low according to Rapp [406] (oxidizing solute is predominantly dissolved SO3) and environment is often reducing at the metal/salt interface promoting sulfidation attack. Signs of this were observed in wrought Nimonic 80A, where some sulphur was detected below alloy surface as was mentioned in section 3.3.2.2. However, irrespective of whether the protective oxide layer formed in the beginning of test or not, at later stage the salt mixture finally fluxed or dissolved the protective oxide layers away and prevented their reformation, which enabled the corrosion, i.e. reactions between alloy, atmosphere and salt, to proceed.

According to studies in Refs. [407, 408], sodium sulphate salt’s ability to flux and thus prevent the formation of protective oxide layers depends strongly on acid/base character of the molten salt. Acid/base character is in turn determined by the salt composition and the surrounding gas atmosphere. Salt is acidic when its oxygen ion (O^2-) activity is low and dissolved SO3 from the atmosphere is high. Salt is basic when its oxygen ion activity is high and dissolved SO3 is low. V2O5 additions tend to decrease oxygen ion activity and increase the melt acidity and corrosion rates [409]. Rapp [407] has shown that solubility of certain

According to studies in Refs. [407, 408], sodium sulphate salt’s ability to flux and thus prevent the formation of protective oxide layers depends strongly on acid/base character of the molten salt. Acid/base character is in turn determined by the salt composition and the surrounding gas atmosphere. Salt is acidic when its oxygen ion (O^2-) activity is low and dissolved SO3 from the atmosphere is high. Salt is basic when its oxygen ion activity is high and dissolved SO3 is low. V2O5 additions tend to decrease oxygen ion activity and increase the melt acidity and corrosion rates [409]. Rapp [407] has shown that solubility of certain