- Email: [email protected]
Creep behavior of pack cementation aluminide coatings on Grade 91 ferritic–martensitic alloy
Surface & Coatings Technology 240 (2014) 32–39
Contents lists available at ScienceDirect
Surface & Coatings Technology journal homepage: www.elsevier.com/locate/surfcoat
Creep behavior of pack cementation aluminide coatings on Grade 91 ferritic–martensitic alloy B.L. Bates a, Y. Zhang a,⁎, S. Dryepondt b, B.A. Pint b a b
Department of Mechanical Engineering, Tennessee Technological University, Cookeville, TN 38505-0001, USA Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6156, USA
a r t i c l e
i n f o
Article history: Received 23 August 2013 Accepted in revised form 10 December 2013 Available online 19 December 2013 Keywords: Aluminide coatings Pack cementation Creep Ferritic–martensitic steel Hardness
a b s t r a c t The creep behavior of various pack cementation aluminide coatings on Grade 91 ferritic–martensitic steel was investigated at 650 °C in laboratory air. The coatings were fabricated in two temperature regimes, i.e., 650 or 700 °C (low temperature) and 1050 °C (high temperature), and consisted of a range of Al levels and thicknesses. For comparison, uncoated specimens heat-treated at 1050 °C to simulate the high temperature coating cycle also were included in the creep test. All coated specimens showed a reduction in creep resistance, with 16–51% decrease in rupture life compared to the as-received bare substrate alloy. However, the specimens heat-treated at 1050 °C exhibited the lowest creep resistance among all tested samples, with a surprisingly short rupture time of b 25 h, much shorter than the specimen coated at 1050 °C. Factors responsible for the reduction in creep resistance of both coated and heat-treated specimens were discussed. © 2013 Elsevier B.V. All rights reserved.
1. Introduction The demand for increased energy efficiencies and decreased emissions has been the driving force for development of coal-fired power plants with higher steam temperature and pressure [1]. Under these operating conditions, the class of 9–12% Cr ferritic–martensitic (FM) steels which, may be creep resistant to 650 °C, can suffer extensive steam-side oxidation [2–4] and thus protective coatings need to be considered. Al-rich coatings are of particular interest because of the slow growth of alumina and its stability in steam and exhaust gas environments, as compared to the coatings that form chromia or silica-rich scales [5–8]. One particular concern for the use of Al-rich coatings on FM steels is the effect of coatings on the mechanical integrity of coated alloys, especially their creep resistance. Studies on Ni-based superalloys [9,10] have suggested that several factors are responsible for the reduction of creep resistance by a coating application, including: (1) changing the microstructure of the substrate material during the thermal cycle of the coating process; (2) decreasing the load-bearing cross-section of the component owing to the weak mechanical strength of the coating; and (3) enabling premature crack initiation in the coating layer. For FM steels, although some of these factors have also been noticed to cause a decrease in creep resistance of coated alloys (e.g., the reduction of load-bearing cross-section) [11,12], in general, the creep performance of coated FM alloys has not been studied to the same degree as coated Ni-base alloys. ⁎ Corresponding author. Tel.: +1 931 372 3265; fax: +1 931 372 6340. E-mail address: [email protected] (Y. Zhang). 0257-8972/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.surfcoat.2013.12.008
This study focused on the effect of pack cementation coatings with various Al levels and thicknesses on the creep behavior of Grade 91 (Gr. 91) FM steel. The standard heat treatment procedure for commercial Gr. 91 alloys involves two steps [13]: (i) austenitizing at 1000–1150 °C for 10 min to 2 h, followed by rapid cooling to form martensite; and (ii) tempering at 730–780 °C for 1–2 h to promote carbide precipitation in the tempered martensite. Several European and US coating programs [7,8,14–16] have focused on synthesis of diffusion aluminide coatings at temperatures below the tempering temperature of the FM steel to preserve the tempered martensitic structure. However, in contrast to a typical aluminizing process at 900–1100 °C that produces phases such as FeAl, Fe3Al or ferritic Fe(Al) in the coating, the reduced coating temperature often leads to the formation of more brittle Al-rich intermetallic phases like Fe2Al5 or even FeAl3. Two temperature regimes were selected for coating fabrication in the present study. The low temperature (650 or 700 °C) was below the tempering temperature of Gr. 91, while the high temperature (1050 °C) was in the temperature range of its austenitizing treatment.
2. Experimental procedure The commercial Gr. 91 alloy (Fe–8.8Cr–1.0Mo–0.5Mn–0.3V– 0.2Ni–0.28Si–0.11C–0.06N–0.003S, in wt.%) was used as the substrate material. Small dog-bone specimens, as illustrated schematically in Fig. 1, were cut via electric-discharge machining (EDM). The gage length was ~7.6 mm, with a cross section of 2 × 2 mm2. The specimens were ground to a 600-grit finish and ultrasonically cleaned in acetone prior to pack cementation or heat treatment.
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
33
Fig. 1. Schematic showing the dimensions of the dog-bone creep specimen.
The pack mixture consisted of 1–2 wt.% NH4Cl activator, 10–20 wt.% masteralloy, and the balance inert Al2O3 filler. Masteralloys of pure Al, Cr–25Al, and Cr–15Al (in wt.%) were employed to vary the Al activity in the pack cementation process, and thus to achieve coatings with different Al levels and thicknesses (Table 1). The specimens coated at 650–700 °C were directly embedded in the pack powder in an alumina crucible, similar to the conventional pack process [5]. For the coatings synthesized at 1050 °C, the substrate was hung in a slotted alumina tube to be separated from the surrounding powder, Fig. 2. Approximately 30 vertical slits (~0.18 mm wide) were machined around the 19 mmOD alumina tube. The reagent gas species were able to interact with the specimen during pack aluminizing, whereas the powders were prevented from reaching the sample surface, leading to a cleaner coating. This non-contact assembly was not used at 650–700 °C, for the coating tended to be less uniform with the increased pack-to-specimen distance at lower aluminizing temperatures [16]. Additional Gr. 91 specimens were heat-treated at 1050 °C in the same arrangement as shown in Fig. 2, except that only inert Al2O3 powder (without masteralloy and activator) was placed around the slotted tube in the crucible. The crucible was then sealed with an alumina lid using an aluminabased cement [17]. After the cement was completely cured, the crucible was loaded into a horizontal resistance-heated tube furnace, and purged with high-purity argon. A vacuum pump was connected to the furnace to aid in the removal of air and moisture, and a vacuum level of 0.13–0.40 Pa was achieved. Pack aluminization was carried out at this vacuum level. The coating time, 6 or 12 h, was defined as the holding time at the aluminizing temperature. The furnace temperature was monitored with a K-type thermocouple positioned in the center of the heating zone, which was also connected to a NI-9211A data logger from National Instruments. In addition, in order to more accurately monitor the temperature of the specimen inside the crucible, small thermocouples were attached to the surface of several specimens. The
temperature–time profiles of both the furnace and the specimen were recorded with LabVIEW SignalExpress. After the aluminization was completed, the specimen was allowed to cool to room temperature in the furnace, and then was removed from the crucible and ultrasonically cleaned. Creep tests were carried out at 650 °C in laboratory air under constant uniaxial loading, with a nominal stress level in the range of 100–120 MPa. For the coated specimens, the stress was calculated on the basis of the sample cross-section before the coating was applied. The uncoated specimens were tested with a 600-grit surface finish, whereas the coated or heat-treated samples were tested in the as-processed condition. The creep testing procedure was based on ASTM standard E139-11 [18]. An average of 2–3 specimens was tested for each condition, and all specimens were crept to rupture. Specimens were examined by optical microscopy and scanning electron microscopy (SEM) equipped with energy dispersive X-ray analysis (EDXA). For cross-sectional observations, the coated specimens were copper-plated prior to metallographic sample preparation. Vickers microhardness test was conducted on the polished cross sections using a load of 500 gf. Each hardness value was taken as the average of five data points. Villela's reagent (100 ml methanol, 5 ml hydrochloric acid, and 1 g picric acid) was used to reveal the microstructure of the Gr. 91 alloy [19]. 3. Results and discussion 3.1. As-fabricated coatings Fig. 3 shows the cross sections of the as-fabricated aluminide coatings that were aluminized at different temperatures with various Al activities in the pack. When pure Al was used as the masteralloy, Fe2Al5 coatings of 80–125 μm thick were formed after 6 h at 650 °C (Fig. 3a), depending on
Table 1 Summary of creep specimens. Note that the pack cementation conditions are given as “temperature/time/masteralloy/amount of masteralloy in wt.%”. Treatment condition
Coating thickness (μm)
Creep rate Rate (h−1)
As-received 650 °C/6 h/Al/10 650 °C/6 h/Al/20 700 °C/12 h/Cr–15 Al/20 700 °C/12 h/Cr–25 Al/15 1050 °C/6 h/Cr–15 Al/20 1050 °C/6 h, heat-treated
0 105 125 18 35 290 0
3.4 × 6.0 × 8.1 × 5.2 × 4.8 × 8.7 × N/A
10−3 10−3 10−3 10−3 10−3 10−3
Time to rupture (h) Increase over “as-received”
Time (h)
Decrease over “as-received”
– 76% 138% 53% 41% 156% N/A
344 288 245 275 311 180 23
– 22% 34% 26% 16% 51% 94%
34
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
a
b
the amount of masteralloy (10 or 20 wt.%). Through-thickness cracks were observed in the Fe2Al5 coating, similar to what was observed by Agüero et al. in their Fe2Al5 slurry coatings [12]. The cracking was attributed to the thermal stress generated by the mismatch in coefficient of thermal expansion (CTE) between the coating and the substrate during cooling and the brittle nature of the Al-rich intermetallic phase Fe2Al5. By switching to the Cr–25Al masteralloy with a reduced Al activity, a thin layer of Fe2Al5 (~ 5 μm) on top of FeAl (~ 10 μm) was produced after 12 h aluminizing at 700 °C, as shown in Fig. 3b. Needlelike AlN precipitates (as pointed by the arrows) were observed in the inner layer [12,20]. Further lowering the Al activity by utilizing the Cr–15Al masteralloy resulted in a thin FeAl coating (~ 12 μm) without the Fe2Al5 outer layer, Fig. 3c. However, for the same Cr–15Al masteralloy, if the aluminizing temperature was increased to 1050 °C, much thicker coatings were obtained after 6 h, as indicated in Fig. 3d. The 1050 °C coating consisted of an outer layer of 15–25 μm and a very thick inner layer of ~275 μm, which were formed by outward diffusion of Fe and inward diffusion of Al, respectively. AlN precipitates were also observed in the 700 and 1050 °C coatings with the use of the Cr–15Al masteralloy, Fig. 3c–d. The Al concentration profiles of these pack coatings have been reported elsewhere [16,21], and a brief summary is given here. For the Fe2Al5 coating formed at 650 °C (Fig. 3a), the Al level remained nearly constant at ~ 72 at.% (55 wt.%) in the entire coating. Increasing the amount of Al masteralloy from 10 to 20 wt.% did not affect the Al content in the coating. For the coating synthesized at 700 °C using the Cr–25Al masteralloy (Fig. 3b), the Fe2Al5 outer layer also contained ~72 at.% Al. The Al content in the inner layer decreased gradually from ∼50 at.% (33 wt.%) to zero, suggesting that FeAl was formed immediately beneath the Fe2Al5 layer and eventually changed to ferritic Fe(Al). The Cr–15Al masteralloy, on the other hand, produced coatings
c
Fig. 2. The non-contact assembly using the slotted alumina tube: (a) schematic of the slotted tube, (b) sample hung on an alumina rod in the slotted tube, and (c) the specimen in the alumina crucible with pack powder.
a
b Cu Plating
Cu Plating
Coating Coating
Substrate Substrate
20µm
10µm
d
c Cu Plating Coating
Coating
Substrate Substrate
10µm
50µm
Fig. 3. SEM backscattered electron images of the cross-sections of various pack coatings on Gr. 91 alloy: (a) pure Al masteralloy (20 wt.% in the pack mixture), 6 h at 650 °C; (b) Cr–25 Al masteralloy (15 wt.% in the mixture), 12 h at 700 °C; (c) Cr–15 Al masteralloy (20 wt.% in the mixture), 12 h at 700 °C; and (d) Cr–15 Al masteralloy (20 wt.% in the mixture), 6 h at 1050 °C. Arrows in (b)–(d) point to the AlN precipitates.
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
3.2. Creep behavior of the as-received substrate alloy The as-received Gr. 91 specimens were creep tested first to validate the testing procedure and to obtain a baseline for comparison with the coated samples. Fig. 4 presents the creep curves of the bare alloy tested at stress levels of 100 and 120 MPa. When the applied stress was decreased from 120 to 100 MPa, the creep rupture time increased from ~65 to ~340 h. The results were consistent with the literature data for Gr. 91 with a tempered martensitic structure [13]. Subsequent creep testing on all coated and heat-treated specimens was conducted at 100 MPa.
a
16
Coated at 650-700°C
10
As-received
8 6 4
0 0
50
100
150
200
300
650°C/6h/Al/10 1050°C/6h/ 700°C/12h/ Cr-15Al/20 650°C/6h/Al/20 Cr-15Al/20
350
700°C/12h/ Cr-25Al/15
2.0 Heat-treated at 1050°C
1.6
1.2
As-received
0.8
0.4
0.0 0
50
100
150
200
250
300
Time (h) Fig. 5. Creep curves for bare and coated Gr. 91 specimens tested at 100 MPa: (a) the complete graph up to rupture and (b) up to 2% strain.
increase of coating thickness, demonstrating that the creep performance of the coated alloys was adversely affected by the reduction in load-bearing cross sections, due to the fact that these intermetallic coatings had essentially no creep strength relative to the substrate at 650 °C [11,12]. Such an effect was more profound for the small creep specimens used in this study. However, the load-bearing factor alone could not explain the extremely low creep resistance of the specimens heattreated at 1050 °C (open circles in Fig. 6). According to our recent oxidation studies [22,23], thin coatings (b50 μm) with relatively low Al contents were more durable and effective than thicker aluminide coatings (~250 μm) at 650–800 °C in humid air, in part because of the reduced CTE mismatch between coating and 400 As-received alloy
10
Coated specimens
120 MPa
100 MPa
TIme to Rupture (h)
Creep Strain (%)
250
Time (h)
Creep Strain (%)
Fig. 5 compares the creep curves of the as-received alloy with the specimens coated at three different temperatures (650, 700, and 1050 °C) and the ones heat-treated at 1050 °C. The creep rate and rupture time for each condition are summarized in Table 1. As compared to the as-received alloy, all coated specimens exhibited a reduction in creep resistance. For the specimens that were aluminized in the temperature range of 650–700 °C, the minimum creep rates increased ~ 50% for thinner coatings (b50 μm) and ~ 138% for thicker coatings (~125 μm). The decrease of rupture time was in the range of 16 to 34% for the 650–700 °C coatings. For the specimens coated at 1050 °C, an increase in creep rate of ~ 156% and a reduction in rupture time of ~51% were observed. However, a surprisingly different creep behavior was noticed for the specimens heat-treated in vacuum at 1050 °C, in spite of the same holding temperature/time and cooling rate as the samples aluminized at 1050 °C. The heat-treated specimens entered the tertiary creep regime once the test was started, with almost no steady state creep observed (Fig. 5a), and failed after only ~23 h (a decrease of ~94% in rupture time). The creep curves with up to 2% strain are replotted in Fig. 5b, to more clearly demonstrate the differences in creep rate for the specimens that were treated under different conditions. Again, the creep rates of the coatings fabricated at 1050 °C were much higher than the ones made at 650–700 °C. The decrease in creep resistance of the present coating specimens was similar to the FM alloy P92 with thick FeAl pack coatings [11] and Fe2Al5 slurry coatings [12] previously reported in the literature. Fig. 6 summarizes the relationship between the creep rupture lifetime and coating thickness. The specimens heat-treated at 1050 °C were also included for comparison. Since both the as-received alloy and the specimens heat-treated at 1050 °C could be considered as with zero coating thickness, different symbols were used to differentiate these two conditions. The rupture time decreased almost linearly with the
Coated at 1050°C
12
2
b
3.3. Creep behavior of the coated or heat-treated specimens
Heat-treated at 1050°C
14
Creep Strain (%)
(Fig. 3c and d) with lower Al levels of ~40 at.% Al at the surface, corresponding to the FeAl phase, which also gradually changed to zero at the coating/substrate interface.
35
8 6 4 2 0 0
100
200
300
400
Time (h)
Heat-treated specimens
300
200
100
0 -50
0
50
100
150
200
250
300
Coating Thickness (µm) Fig. 4. Creep curves of the as-received Gr. 91 alloy tested at stress levels of 100 and 120 MPa.
Fig. 6. Creep rupture time as a function of coating thickness.
350
36
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
substrate which mitigated cracking and coating failure. Therefore, the present creep testing results, in combination with the precious longterm oxidation studies, clearly indicate that thinner coatings with lower Al levels may have more potential to be adopted in power generation applications than thicker aluminide coatings due to the considerations of both oxidation and creep resistance. Fig. 7 shows the metallographic cross-sections of the coated specimens in the gauge region after rupture. Prior to creep testing, vertical cracks had already existed in the Fe2Al5 coating fabricated at 650 °C with the pure Al masteralloy (Fig. 3a), which were still present after the test (Fig. 7a). Although there was no cracking in the original coating with a thin Fe2Al5 top layer that was synthesized at 700 °C using the Cr–25Al masteralloy (Fig. 3b), cracks were observed after creep testing (Fig. 7b). Nevertheless, none of these cracks propagated into the substrate. Cracks were absent in the FeAl coatings aluminized either at 700 °C (Fig. 7c) or 1050 °C (Fig. 7d) in the packs containing the Cr–15Al masteralloy. The results further indicate that coatings with brittle Al-rich intermetallic phases, such as Fe2Al5, were more susceptible to cracking. In contrast to the coating thickness factor that contributed to the loss of load-bearing section, the cracks did not seem to significantly reduce the creep resistance of the current coated specimens. 3.4. Effect of creep exposure on hardness and microstructure of the alloy Vickers hardness was measured in the grip portion (away from the deformation region) of the crept specimens in the un-etched condition, and the values are given in Fig. 8, including the as-received alloy without being creep tested. The rupture lifetime of each specimen was also presented in the graph. It is clear that there was almost no change in hardness (205 vs. 207 HV) for the bare alloy before and after 328h of creep test at 650 °C. The specimens coated at 650–700 °C exhibited hardness of 205–211 HV after creep, close to that of the as-received alloy. A higher hardness value, ~ 251 HV, was detected in the alloy that was aluminized at 1050 °C and subsequently creep
tested at 650 °C for 192h. However, the specimen heat-treated at 1050 °C under vacuum only showed a hardness value of ~ 196 HV after the creep test. Even though no strong correlations can be found between microhardness and rupture time in the current study, changes in hardness have been used as a measure of microstructural stability to assess the residual creep life for some high chromium FM steels. Sawada et al. [24] reported a decrease in hardness of Mod.9Cr–1Mo steel after 2000–3000 h creep testing at 700 °C (under a nominal stress of 40 MPa), which was attributed to the increasing inter-particle spacing as a result of coarsening and coalescence of the precipitates. In contrast, because of the shorter exposure time (b350 h) and the lower exposure temperature (650°), the extent of precipitate coarsening was probably very limited, leading to negligible change in hardness of the alloy before and after creep testing. Fig. 9 compares the microstructures of the alloy before and after creep testing. Again, the optical photomicrographs were taken in the grip section of the dog-bone specimens. No distinct change in microstructure was found for the bare alloy before (Fig. 9a) and after the creep test (Fig. 9b), and both exhibited a tempered martensitic structure. For the specimens that were aluminized at 650–700 °C (i.e., below the alloy's tempering temperature), the substrate remained tempered martensite after creep, as shown in Fig. 9c–d. For the specimen coated at 1050 °C, a similar microstructure was also observed after creep, Fig. 9e. In contrast, the heat-treated specimen displayed a completely different microstructure after creep testing (Fig. 9f), where polygonal grains were noted, resembling the ferrite phase [25]. 3.5. Microstructural difference between specimens coated and heat-treated at 1050 °C In order to understand the different microstructures of the specimens coated and heat-treated at 1050 °C after creep testing (Fig. 9e and f), additional specimens that were not creep tested were examined.
b
a
Coating Coating Void Substrate Substrate
Nitride
50 µm
c
10 µm
d Coating
Coating
Nitride
Substrate
10 µm
Substrate
10 µm
Fig. 7. Optical micrographs of cross-sections of coated Gr. 91 specimens after creep testing at 650 °C: (a) pure Al masteralloy, 6 h at 650 °C; (b) Cr–25Al masteralloy, 12 h at 700 °C; (c) Cr–15Al masteralloy, 12 h at 700 °C; and (d) Cr–15Al masteralloy, 6 h at 1050 °C.
Vickers Hardness & Rupture Time
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39 328
The microstructures of the as-coated and as-treated alloys are shown in Fig. 10, and the corresponding microhardness values can be found in Table 2. The substrate in the as-coated specimen displayed a martensite-like microstructure (Fig. 10a) with a very high hardness value of ~390 HV. On the other hand, the specimen after heat treatment consisted of mainly polygonal ferrite grains with a small fraction of martensite (Fig. 10b), similar to what was observed for the heat-treated and creep-tested sample (Fig. 9f). The hardness of the heat-treated specimen was ~229 HV, higher than the same specimen after creep testing (~196 HV). Since both of the coated and heat-treated specimens were held at 1050 °C for 6 h, the initial tempered martensitic structure should have been converted to austenite before cooling started. The martensite start temperature (Ms) of Gr. 91 is around 370 °C [13]. According to the readings from the thermocouple that was attached to the specimen in the crucible, the cooling rates were very similar for the two specimens, with an average of 1.2–1.4 °C/min from 1050 °C to room temperature and slightly faster cooling (2.1–2.4 °C/min) from 1050 to 370 °C. As a result of such a slow cooling process, ferrite would be expected to form in both specimens. However, ferrite was only observed in the
HV 500gf 292
Rupture Time (h)
290 251
205
211
207
205
192
196
26
AsAsreceived, received, w/o creep creep test tested
Coated, Coated, Coated, Heat650°C/6h, 700°C/12h, 1050°C/6h, treated, 10Al Cr-15Al Cr-15Al 1050°C/6h
Fig. 8. Vickers hardness of the Gr. 91 alloy treated under different conditions after the creep test and the corresponding rupture life. The hardness was measured in the grip portion (away from the deformation region).
a
37
b
10µm
c
10µm
d
10µm
10µm
e
f
10µm
10µm
Fig. 9. Optical microstructures of the Gr. 91 alloy before (a) and after (b–f) the creep test: (b) bare alloy, (c) coated at 650 °C (6 h, pure Al masteralloy); (d) coated at 700 °C (12 h, Cr–15Al masteralloy), (e) coated at 1050 °C (6 h, Cr–15Al masteralloy), and (f) heat-treated at 1050 °C (6 h). The specimens were etched with Villela's reagent.
38
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
a
b
10 µm Fig. 10. Optical microstructure of the Gr. 91 alloy prior to the creep test: (a) coated at 1050 °C; and (b) heat-treated at 1050 °C.
heat-treated specimen (Fig. 10b), and both microstructure (Fig. 10a) and hardness data (Table 2) confirmed the formation of martensite in the coated specimen [13]. It is therefore reasonable to hypothesize that the development of martensite in the coated specimen was somehow associated with the pack cementation coating process. The as-received Gr. 91 steel consisted of tempered martensite with subgrains of high dislocation density and M23C6 (M = Cr) carbide precipitates along the subgrain boundaries, as well as MX (M = V, Nb and X = C, N) carbonitrides [26]. When the alloy was heated to 1050 °C during pack aluminizing, the transformation from tempered martensite to austenite occurred in the alloy, leading to the dissolution of most carbides [13]. Aluminum is known as a ferrite stabilizer and has very limited solubility in austenite [27]. Diffusion of Al into the alloy produced a ferrite layer at the specimen surface with very low carbon solubility, and carbon was therefore rejected into the austenite core [28], as illustrated in Fig. 11. Although it is difficult to determine the carbon content in the austenite core using techniques such as electron probe microanalysis (EPMA) or EDXA due to its low concentration, the amount of carbon rejected from the coating into the substrate can be estimated. As mentioned earlier in Section 3.1, for the coating synthesized at 1050 °C, the outer layer was formed via outward diffusion of Fe and the inner layer by inward diffusion of Al. The interface between the outer and inner layers (Figs. 3d and 11) corresponded to the original substrate surface, i.e., the inner coating layer had originally been part of the substrate alloy. The carbon content in the as-received Gr. 91 alloy was ~ 0.11 wt.%. Assuming that as the Al diffused into the coating inner layer at 1050 °C, carbon was completely rejected to the austenite core and was dissolved homogeneously in the austenite, the amount of carbon which was originally in the substrate region corresponding to the coating inner layer would be equivalent to what was received by the austenite core. For the 2-mm thick creep specimen, owing to the symmetry only half of the specimen was considered. If values of 25 and 275 μm were assumed as the thicknesses for the outer and inner layers, respectively, the austenite core would be 725 μm thick, Fig. 11. It was estimated that the carbon content in the core region would be increased to ~0.16 wt.% (i.e., an increase of 0.05 wt.% over the original carbon content) as a result of carbon rejection from the coating inner layer. Gr. 91 steel is a highly hardenable alloy, and it generally transforms completely to martensite after air cooling from the austenitizing Table 2 Comparison in hardness of selected specimens. Treatment time (h)
Treatment environment
As-received 6 6
N/A Aluminization Vacuum
Hardness (HV) Before creep
After creep
205 390 229
207 251 196
temperature to room temperature. Hardenability is defined as the relative ability of a steel to avoid forming the soft ferrite phase when cooled from the austenitizing temperature [25]. Increasing the hardenability has the same effect as increasing the cooling rate, which shifts the formation of ferrite to longer times so that the steel can be cooled more slowly while still forming a martensitic microstructure. Carbon is an important element that effectively increases hardenability. For the present thin austenite core (~725 μm), it is rational to assume that an increase of 0.05 wt.% in carbon content was sufficient to increase the hardenability such that martensitic transformation occurred when the specimen was furnace cooled from 1050 °C after pack cementation. Subsequent creep test of ~180 h at 650 °C exerted an effect of tempering, during which the conversion of martensite to tempered martensite occurred. For the commercial Gr. 91 alloy, tempering is normally carried out at 730–780 °C with a minimum of 1 h of holding time to allow carbides to precipitate out homogeneously [13]. The creep testing temperature of 650 °C (equivalent to a lower tempering temperature) thus produced higher hardness (251 HV) in the coated alloy than that of the as-received alloy (205 HV) which was originally tempered at 730–780 °C, as indicated in Table 2.
Specimen Centerline Pack Chemicals Al
Cr
AlCl
AlClx
Austenite Core (~725 µm)
C
Gas
Outer Coating Layer (~25 µm)
Inner Coating Layer (~275 µm)
Fig. 11. Schematic of the pack cementation coating process. The darkened area represents a zone of carbon enrichment in austenite. Adapted from Reference [28].
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
3.6. Effect of the coating/substrate interaction on the alloy creep behavior As compared to the ferritic microstructure formed in the specimens heat-treated at 1050 °C (Fig. 10b), the martensitic structure in the coated alloy (Fig. 10a) compensated for the weak creep strength of the FeAl coating. Hence, the specimens aluminized at 1050 °C exhibited higher creep resistance than the ones heat-treated at 1050 °C, as observed in Fig. 5. Similar effect of carbon enrichment on creep performance has been reported for Alloy 617 (Ni–22Cr–12Co–9Mo–1Al–0.08C, wt.%) after oxidation testing at 1000 °C [29]. As a result of selective oxidation of Cr, a zone depleted in chromium carbides was formed immediately beneath the Cr2O3 scale, which extended ~ 500 μm deep into the 2-mm thick specimens after 6000-h exposure [29]. Carbon released from the chromium carbides in the depletion zone diffused into the specimen core, producing a carburization effect. Extensive carbide precipitation was found in the central region of the specimen, which increased the creep strength locally, and counteracted the loss of strength due to carbide dissolution in the Cr-depleted zone near the specimen surface. The present study further corroborates that the effects of coatings on creep resistance of FM steels are sometimes more complicated than just the loss of load-bearing cross sections, even though it often plays an important role in the reduction of creep strength for coated alloys. Compositional/microstructural changes of the alloy can take place during the coating fabrication process, due to either the thermal treatment itself (temperature and time) or interactions between the coating and the substrate. For typical aluminizing and chromizing processes, the coating elements (Al and Cr) are ferrite stabilizers. Phase transformations triggered by deposition and diffusion of Al and/or Cr can change the solubility of other elements originally present in the substrate, leading to enrichment or depletion of certain elements in the alloy core, similar to the case of the 1050 °C coatings in this study. In addition, for alloys containing high nitrogen or carbon, diffusion of Al or Cr can lead to the formation of AlN precipitates [12,20] or chromium carbides [26]. Attention should also be paid to the interactions between coating and substrate, particularly the resultant changes in concentration and/or redistribution of the minor elements (such as carbon or nitrogen) in the substrate, for these elements have been meticulously balanced in alloy design for optimal creep performance [26]. For the FM alloys, several studies have concurred that the coating process (e.g., temperature, time, heating/cooling rate, etc.) needs to be compatible with the heat treatment cycle of the alloy [7,8,14–16], which is again confirmed by the present creep testing results. 4. Conclusions As compared to the bare Gr. 91 alloy with a tempered martensitic microstructure, decreases in creep resistance at 650 °C were observed for all specimens coated with pack cementation aluminide coatings. For the specimens aluminized at 650–700 °C with coating thickness in the range of 18–125 μm, a reduction of 16–34% in rupture time was noticed, whereas for the specimens aluminized at 1050 °C with a thicker coating
39
of ~ 290 μm, the decrease of rupture lifetime was ~ 51%. However, the greatest reduction in rupture time (~94%) was found for the specimens heat-treated in vacuum, following the same thermal cycle as the 1050 °C coatings. The drastic difference in creep performance between the specimens coated and heat-treated at 1050 °C was due to the different microstructures that were formed in the alloy. Martensite was developed in the coated substrate alloy, as carbon was rejected from the aluminide coating into the central region of the specimen, leading to increased hardenability. In contrast, ferrite was present in the alloy heat-treated at 1050 °C, resulting in a much lower creep resistance. Acknowledgments The authors would like to acknowledge funding by the Department of Energy (DOE) Advanced Coal Research at U.S. Colleges and Universities, under Grant No. DE-FG26-06NT42674. Additional support is from DOE Fossil Energy Advanced Materials Research Program, under contract DEAC05-00OR22725 with UT-Battelle LLC and subcontract 4000071336 with Tennessee Technological University. References [1] R. Viswanathan, W. Bakker, J. Mater. Eng. Perform. 10 (2001) 81. [2] H. Nickel, Y. Wouters, M. Thiele, W.J. Quadakkers, Fresenius J. Anal. Chem. 361 (1998) 540. [3] H. Asteman, J.-E. Svensson, L.-G. Johansson, M. Norell, Oxid. Met. 52 (1999) 95. [4] E.J. Opila, Mater. Sci. Forum 461–464 (2004) 765. [5] R. Bianco, M.A. Harper, R.A. Rapp, JOM 43 (1991) 20. [6] B.A. Pint, Y. Zhang, P.F. Tortorelli, J.A. Haynes, I.G. Wright, Mater. High Temp. 18 (2001) 185. [7] A. Agüero, R. Muelas, A. Paster, S. Osgerby, Surf. Coat. Technol. 200 (2005) 1219. [8] V. Rohr, M. Schütze, E. Fortuna, D.N. Tsipas, A. Milewska, F.J. Pérez, Mater. Corros. 56 (2005) 874. [9] H.J. Kolkman, Mater. Sci. Eng. 89 (1987) 81. [10] S. Osgerby, D.F. Dyson, Mater. Sci. Eng. A121 (1989) 645. [11] S. Dryepondt, Y. Zhang, B.A. Pint, Surf. Coat. Technol. 201 (2006) 3880. [12] A. Agüero, R. Muelas, M. Gutiérrez, R. Van Vulpen, S. Osgerby, J.P. Banks, Surf. Coat. Technol. 201 (2007) 6253. [13] G. Guntz, M. Julien, G. Kottmann, F. Pellicani, A. Pouilly, J.C. Vaillant, The T91 Book, Vallourec Industries, France, 1990. 30–34. [14] F.J. Pérez, M.P. Hierro, J.A. Trilleros, M.C. Carpintero, L. Sánchez, J.M. Brossard, F.J. Bolívar, Intermetallics 14 (2006) 811. [15] B.L. Bates, Y.Q. Wang, Y. Zhang, B.A. Pint, Surf. Coat. Technol. 204 (2009) 766. [16] Y.Q. Wang, Y. Zhang, D.A. Wilson, Surf. Coat. Technol. 204 (2010) 2737. [17] R. Bianco, R.A. Rapp, J. Electrochem. Soc. 140 (1993) 1181. [18] ASTM E139-11, Standard Test Methods for Conducting Creep, Creep-rupture, and Stress-rupture Tests of Metallic Materials, ASTM International, West Conshohocken, PA, 2006., http://dx.doi.org/10.1520/E0139-11 (2003, www.astm.org). [19] In: G.F. Vander Voort (Ed.), Metallography and Microstructures, ASM Handbook, vol. 9, ASM International, 2004, p. 211. [20] Y. Zhang, B.A. Pint, K.M. Cooley, J.A. Haynes, Surf. Coat. Technol. 200 (2005) 1231. [21] S. Velraj, Y. Zhang, E.W. Hawkins, B.A. Pint, Mater. Corros. 63 (2012) 909. [22] B.A. Pint, Y. Zhang, L.R. Walker, I.G. Wright, Surf. Coat. Technol. 202 (2007) 637. [23] B.A. Pint, Y. Zhang, Mater. Corros. 62 (2011) 549. [24] K. Sawada, K. Miyahara, H. Kushima, K. Kimura, S. Matsuoka, ISIJ Int. 45 (2005) 1934. [25] R.L. Klueh, D.J. Alexander, E.A. Kenik, J. Nucl. Mater. 227 (1995) 11. [26] K. Maruyama, K. Sawada, J.-I. Koike, ISIJ Int. 41 (2001) 641. [27] N.V. Bangaru, R.C. Krutenat, J. Vac. Sci. Technol. B 2 (1984) 806. [28] M. Zheng, R.A. Rapp, Oxid. Met. 49 (1998) 19. [29] P.J. Ennis, W.J. Quadakkers, H. Shuster, Mater. Sci. Technol. 8 (1992) 78.
Contents lists available at ScienceDirect
Surface & Coatings Technology journal homepage: www.elsevier.com/locate/surfcoat
Creep behavior of pack cementation aluminide coatings on Grade 91 ferritic–martensitic alloy B.L. Bates a, Y. Zhang a,⁎, S. Dryepondt b, B.A. Pint b a b
Department of Mechanical Engineering, Tennessee Technological University, Cookeville, TN 38505-0001, USA Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6156, USA
a r t i c l e
i n f o
Article history: Received 23 August 2013 Accepted in revised form 10 December 2013 Available online 19 December 2013 Keywords: Aluminide coatings Pack cementation Creep Ferritic–martensitic steel Hardness
a b s t r a c t The creep behavior of various pack cementation aluminide coatings on Grade 91 ferritic–martensitic steel was investigated at 650 °C in laboratory air. The coatings were fabricated in two temperature regimes, i.e., 650 or 700 °C (low temperature) and 1050 °C (high temperature), and consisted of a range of Al levels and thicknesses. For comparison, uncoated specimens heat-treated at 1050 °C to simulate the high temperature coating cycle also were included in the creep test. All coated specimens showed a reduction in creep resistance, with 16–51% decrease in rupture life compared to the as-received bare substrate alloy. However, the specimens heat-treated at 1050 °C exhibited the lowest creep resistance among all tested samples, with a surprisingly short rupture time of b 25 h, much shorter than the specimen coated at 1050 °C. Factors responsible for the reduction in creep resistance of both coated and heat-treated specimens were discussed. © 2013 Elsevier B.V. All rights reserved.
1. Introduction The demand for increased energy efficiencies and decreased emissions has been the driving force for development of coal-fired power plants with higher steam temperature and pressure [1]. Under these operating conditions, the class of 9–12% Cr ferritic–martensitic (FM) steels which, may be creep resistant to 650 °C, can suffer extensive steam-side oxidation [2–4] and thus protective coatings need to be considered. Al-rich coatings are of particular interest because of the slow growth of alumina and its stability in steam and exhaust gas environments, as compared to the coatings that form chromia or silica-rich scales [5–8]. One particular concern for the use of Al-rich coatings on FM steels is the effect of coatings on the mechanical integrity of coated alloys, especially their creep resistance. Studies on Ni-based superalloys [9,10] have suggested that several factors are responsible for the reduction of creep resistance by a coating application, including: (1) changing the microstructure of the substrate material during the thermal cycle of the coating process; (2) decreasing the load-bearing cross-section of the component owing to the weak mechanical strength of the coating; and (3) enabling premature crack initiation in the coating layer. For FM steels, although some of these factors have also been noticed to cause a decrease in creep resistance of coated alloys (e.g., the reduction of load-bearing cross-section) [11,12], in general, the creep performance of coated FM alloys has not been studied to the same degree as coated Ni-base alloys. ⁎ Corresponding author. Tel.: +1 931 372 3265; fax: +1 931 372 6340. E-mail address: [email protected] (Y. Zhang). 0257-8972/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.surfcoat.2013.12.008
This study focused on the effect of pack cementation coatings with various Al levels and thicknesses on the creep behavior of Grade 91 (Gr. 91) FM steel. The standard heat treatment procedure for commercial Gr. 91 alloys involves two steps [13]: (i) austenitizing at 1000–1150 °C for 10 min to 2 h, followed by rapid cooling to form martensite; and (ii) tempering at 730–780 °C for 1–2 h to promote carbide precipitation in the tempered martensite. Several European and US coating programs [7,8,14–16] have focused on synthesis of diffusion aluminide coatings at temperatures below the tempering temperature of the FM steel to preserve the tempered martensitic structure. However, in contrast to a typical aluminizing process at 900–1100 °C that produces phases such as FeAl, Fe3Al or ferritic Fe(Al) in the coating, the reduced coating temperature often leads to the formation of more brittle Al-rich intermetallic phases like Fe2Al5 or even FeAl3. Two temperature regimes were selected for coating fabrication in the present study. The low temperature (650 or 700 °C) was below the tempering temperature of Gr. 91, while the high temperature (1050 °C) was in the temperature range of its austenitizing treatment.
2. Experimental procedure The commercial Gr. 91 alloy (Fe–8.8Cr–1.0Mo–0.5Mn–0.3V– 0.2Ni–0.28Si–0.11C–0.06N–0.003S, in wt.%) was used as the substrate material. Small dog-bone specimens, as illustrated schematically in Fig. 1, were cut via electric-discharge machining (EDM). The gage length was ~7.6 mm, with a cross section of 2 × 2 mm2. The specimens were ground to a 600-grit finish and ultrasonically cleaned in acetone prior to pack cementation or heat treatment.
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
33
Fig. 1. Schematic showing the dimensions of the dog-bone creep specimen.
The pack mixture consisted of 1–2 wt.% NH4Cl activator, 10–20 wt.% masteralloy, and the balance inert Al2O3 filler. Masteralloys of pure Al, Cr–25Al, and Cr–15Al (in wt.%) were employed to vary the Al activity in the pack cementation process, and thus to achieve coatings with different Al levels and thicknesses (Table 1). The specimens coated at 650–700 °C were directly embedded in the pack powder in an alumina crucible, similar to the conventional pack process [5]. For the coatings synthesized at 1050 °C, the substrate was hung in a slotted alumina tube to be separated from the surrounding powder, Fig. 2. Approximately 30 vertical slits (~0.18 mm wide) were machined around the 19 mmOD alumina tube. The reagent gas species were able to interact with the specimen during pack aluminizing, whereas the powders were prevented from reaching the sample surface, leading to a cleaner coating. This non-contact assembly was not used at 650–700 °C, for the coating tended to be less uniform with the increased pack-to-specimen distance at lower aluminizing temperatures [16]. Additional Gr. 91 specimens were heat-treated at 1050 °C in the same arrangement as shown in Fig. 2, except that only inert Al2O3 powder (without masteralloy and activator) was placed around the slotted tube in the crucible. The crucible was then sealed with an alumina lid using an aluminabased cement [17]. After the cement was completely cured, the crucible was loaded into a horizontal resistance-heated tube furnace, and purged with high-purity argon. A vacuum pump was connected to the furnace to aid in the removal of air and moisture, and a vacuum level of 0.13–0.40 Pa was achieved. Pack aluminization was carried out at this vacuum level. The coating time, 6 or 12 h, was defined as the holding time at the aluminizing temperature. The furnace temperature was monitored with a K-type thermocouple positioned in the center of the heating zone, which was also connected to a NI-9211A data logger from National Instruments. In addition, in order to more accurately monitor the temperature of the specimen inside the crucible, small thermocouples were attached to the surface of several specimens. The
temperature–time profiles of both the furnace and the specimen were recorded with LabVIEW SignalExpress. After the aluminization was completed, the specimen was allowed to cool to room temperature in the furnace, and then was removed from the crucible and ultrasonically cleaned. Creep tests were carried out at 650 °C in laboratory air under constant uniaxial loading, with a nominal stress level in the range of 100–120 MPa. For the coated specimens, the stress was calculated on the basis of the sample cross-section before the coating was applied. The uncoated specimens were tested with a 600-grit surface finish, whereas the coated or heat-treated samples were tested in the as-processed condition. The creep testing procedure was based on ASTM standard E139-11 [18]. An average of 2–3 specimens was tested for each condition, and all specimens were crept to rupture. Specimens were examined by optical microscopy and scanning electron microscopy (SEM) equipped with energy dispersive X-ray analysis (EDXA). For cross-sectional observations, the coated specimens were copper-plated prior to metallographic sample preparation. Vickers microhardness test was conducted on the polished cross sections using a load of 500 gf. Each hardness value was taken as the average of five data points. Villela's reagent (100 ml methanol, 5 ml hydrochloric acid, and 1 g picric acid) was used to reveal the microstructure of the Gr. 91 alloy [19]. 3. Results and discussion 3.1. As-fabricated coatings Fig. 3 shows the cross sections of the as-fabricated aluminide coatings that were aluminized at different temperatures with various Al activities in the pack. When pure Al was used as the masteralloy, Fe2Al5 coatings of 80–125 μm thick were formed after 6 h at 650 °C (Fig. 3a), depending on
Table 1 Summary of creep specimens. Note that the pack cementation conditions are given as “temperature/time/masteralloy/amount of masteralloy in wt.%”. Treatment condition
Coating thickness (μm)
Creep rate Rate (h−1)
As-received 650 °C/6 h/Al/10 650 °C/6 h/Al/20 700 °C/12 h/Cr–15 Al/20 700 °C/12 h/Cr–25 Al/15 1050 °C/6 h/Cr–15 Al/20 1050 °C/6 h, heat-treated
0 105 125 18 35 290 0
3.4 × 6.0 × 8.1 × 5.2 × 4.8 × 8.7 × N/A
10−3 10−3 10−3 10−3 10−3 10−3
Time to rupture (h) Increase over “as-received”
Time (h)
Decrease over “as-received”
– 76% 138% 53% 41% 156% N/A
344 288 245 275 311 180 23
– 22% 34% 26% 16% 51% 94%
34
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
a
b
the amount of masteralloy (10 or 20 wt.%). Through-thickness cracks were observed in the Fe2Al5 coating, similar to what was observed by Agüero et al. in their Fe2Al5 slurry coatings [12]. The cracking was attributed to the thermal stress generated by the mismatch in coefficient of thermal expansion (CTE) between the coating and the substrate during cooling and the brittle nature of the Al-rich intermetallic phase Fe2Al5. By switching to the Cr–25Al masteralloy with a reduced Al activity, a thin layer of Fe2Al5 (~ 5 μm) on top of FeAl (~ 10 μm) was produced after 12 h aluminizing at 700 °C, as shown in Fig. 3b. Needlelike AlN precipitates (as pointed by the arrows) were observed in the inner layer [12,20]. Further lowering the Al activity by utilizing the Cr–15Al masteralloy resulted in a thin FeAl coating (~ 12 μm) without the Fe2Al5 outer layer, Fig. 3c. However, for the same Cr–15Al masteralloy, if the aluminizing temperature was increased to 1050 °C, much thicker coatings were obtained after 6 h, as indicated in Fig. 3d. The 1050 °C coating consisted of an outer layer of 15–25 μm and a very thick inner layer of ~275 μm, which were formed by outward diffusion of Fe and inward diffusion of Al, respectively. AlN precipitates were also observed in the 700 and 1050 °C coatings with the use of the Cr–15Al masteralloy, Fig. 3c–d. The Al concentration profiles of these pack coatings have been reported elsewhere [16,21], and a brief summary is given here. For the Fe2Al5 coating formed at 650 °C (Fig. 3a), the Al level remained nearly constant at ~ 72 at.% (55 wt.%) in the entire coating. Increasing the amount of Al masteralloy from 10 to 20 wt.% did not affect the Al content in the coating. For the coating synthesized at 700 °C using the Cr–25Al masteralloy (Fig. 3b), the Fe2Al5 outer layer also contained ~72 at.% Al. The Al content in the inner layer decreased gradually from ∼50 at.% (33 wt.%) to zero, suggesting that FeAl was formed immediately beneath the Fe2Al5 layer and eventually changed to ferritic Fe(Al). The Cr–15Al masteralloy, on the other hand, produced coatings
c
Fig. 2. The non-contact assembly using the slotted alumina tube: (a) schematic of the slotted tube, (b) sample hung on an alumina rod in the slotted tube, and (c) the specimen in the alumina crucible with pack powder.
a
b Cu Plating
Cu Plating
Coating Coating
Substrate Substrate
20µm
10µm
d
c Cu Plating Coating
Coating
Substrate Substrate
10µm
50µm
Fig. 3. SEM backscattered electron images of the cross-sections of various pack coatings on Gr. 91 alloy: (a) pure Al masteralloy (20 wt.% in the pack mixture), 6 h at 650 °C; (b) Cr–25 Al masteralloy (15 wt.% in the mixture), 12 h at 700 °C; (c) Cr–15 Al masteralloy (20 wt.% in the mixture), 12 h at 700 °C; and (d) Cr–15 Al masteralloy (20 wt.% in the mixture), 6 h at 1050 °C. Arrows in (b)–(d) point to the AlN precipitates.
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
3.2. Creep behavior of the as-received substrate alloy The as-received Gr. 91 specimens were creep tested first to validate the testing procedure and to obtain a baseline for comparison with the coated samples. Fig. 4 presents the creep curves of the bare alloy tested at stress levels of 100 and 120 MPa. When the applied stress was decreased from 120 to 100 MPa, the creep rupture time increased from ~65 to ~340 h. The results were consistent with the literature data for Gr. 91 with a tempered martensitic structure [13]. Subsequent creep testing on all coated and heat-treated specimens was conducted at 100 MPa.
a
16
Coated at 650-700°C
10
As-received
8 6 4
0 0
50
100
150
200
300
650°C/6h/Al/10 1050°C/6h/ 700°C/12h/ Cr-15Al/20 650°C/6h/Al/20 Cr-15Al/20
350
700°C/12h/ Cr-25Al/15
2.0 Heat-treated at 1050°C
1.6
1.2
As-received
0.8
0.4
0.0 0
50
100
150
200
250
300
Time (h) Fig. 5. Creep curves for bare and coated Gr. 91 specimens tested at 100 MPa: (a) the complete graph up to rupture and (b) up to 2% strain.
increase of coating thickness, demonstrating that the creep performance of the coated alloys was adversely affected by the reduction in load-bearing cross sections, due to the fact that these intermetallic coatings had essentially no creep strength relative to the substrate at 650 °C [11,12]. Such an effect was more profound for the small creep specimens used in this study. However, the load-bearing factor alone could not explain the extremely low creep resistance of the specimens heattreated at 1050 °C (open circles in Fig. 6). According to our recent oxidation studies [22,23], thin coatings (b50 μm) with relatively low Al contents were more durable and effective than thicker aluminide coatings (~250 μm) at 650–800 °C in humid air, in part because of the reduced CTE mismatch between coating and 400 As-received alloy
10
Coated specimens
120 MPa
100 MPa
TIme to Rupture (h)
Creep Strain (%)
250
Time (h)
Creep Strain (%)
Fig. 5 compares the creep curves of the as-received alloy with the specimens coated at three different temperatures (650, 700, and 1050 °C) and the ones heat-treated at 1050 °C. The creep rate and rupture time for each condition are summarized in Table 1. As compared to the as-received alloy, all coated specimens exhibited a reduction in creep resistance. For the specimens that were aluminized in the temperature range of 650–700 °C, the minimum creep rates increased ~ 50% for thinner coatings (b50 μm) and ~ 138% for thicker coatings (~125 μm). The decrease of rupture time was in the range of 16 to 34% for the 650–700 °C coatings. For the specimens coated at 1050 °C, an increase in creep rate of ~ 156% and a reduction in rupture time of ~51% were observed. However, a surprisingly different creep behavior was noticed for the specimens heat-treated in vacuum at 1050 °C, in spite of the same holding temperature/time and cooling rate as the samples aluminized at 1050 °C. The heat-treated specimens entered the tertiary creep regime once the test was started, with almost no steady state creep observed (Fig. 5a), and failed after only ~23 h (a decrease of ~94% in rupture time). The creep curves with up to 2% strain are replotted in Fig. 5b, to more clearly demonstrate the differences in creep rate for the specimens that were treated under different conditions. Again, the creep rates of the coatings fabricated at 1050 °C were much higher than the ones made at 650–700 °C. The decrease in creep resistance of the present coating specimens was similar to the FM alloy P92 with thick FeAl pack coatings [11] and Fe2Al5 slurry coatings [12] previously reported in the literature. Fig. 6 summarizes the relationship between the creep rupture lifetime and coating thickness. The specimens heat-treated at 1050 °C were also included for comparison. Since both the as-received alloy and the specimens heat-treated at 1050 °C could be considered as with zero coating thickness, different symbols were used to differentiate these two conditions. The rupture time decreased almost linearly with the
Coated at 1050°C
12
2
b
3.3. Creep behavior of the coated or heat-treated specimens
Heat-treated at 1050°C
14
Creep Strain (%)
(Fig. 3c and d) with lower Al levels of ~40 at.% Al at the surface, corresponding to the FeAl phase, which also gradually changed to zero at the coating/substrate interface.
35
8 6 4 2 0 0
100
200
300
400
Time (h)
Heat-treated specimens
300
200
100
0 -50
0
50
100
150
200
250
300
Coating Thickness (µm) Fig. 4. Creep curves of the as-received Gr. 91 alloy tested at stress levels of 100 and 120 MPa.
Fig. 6. Creep rupture time as a function of coating thickness.
350
36
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
substrate which mitigated cracking and coating failure. Therefore, the present creep testing results, in combination with the precious longterm oxidation studies, clearly indicate that thinner coatings with lower Al levels may have more potential to be adopted in power generation applications than thicker aluminide coatings due to the considerations of both oxidation and creep resistance. Fig. 7 shows the metallographic cross-sections of the coated specimens in the gauge region after rupture. Prior to creep testing, vertical cracks had already existed in the Fe2Al5 coating fabricated at 650 °C with the pure Al masteralloy (Fig. 3a), which were still present after the test (Fig. 7a). Although there was no cracking in the original coating with a thin Fe2Al5 top layer that was synthesized at 700 °C using the Cr–25Al masteralloy (Fig. 3b), cracks were observed after creep testing (Fig. 7b). Nevertheless, none of these cracks propagated into the substrate. Cracks were absent in the FeAl coatings aluminized either at 700 °C (Fig. 7c) or 1050 °C (Fig. 7d) in the packs containing the Cr–15Al masteralloy. The results further indicate that coatings with brittle Al-rich intermetallic phases, such as Fe2Al5, were more susceptible to cracking. In contrast to the coating thickness factor that contributed to the loss of load-bearing section, the cracks did not seem to significantly reduce the creep resistance of the current coated specimens. 3.4. Effect of creep exposure on hardness and microstructure of the alloy Vickers hardness was measured in the grip portion (away from the deformation region) of the crept specimens in the un-etched condition, and the values are given in Fig. 8, including the as-received alloy without being creep tested. The rupture lifetime of each specimen was also presented in the graph. It is clear that there was almost no change in hardness (205 vs. 207 HV) for the bare alloy before and after 328h of creep test at 650 °C. The specimens coated at 650–700 °C exhibited hardness of 205–211 HV after creep, close to that of the as-received alloy. A higher hardness value, ~ 251 HV, was detected in the alloy that was aluminized at 1050 °C and subsequently creep
tested at 650 °C for 192h. However, the specimen heat-treated at 1050 °C under vacuum only showed a hardness value of ~ 196 HV after the creep test. Even though no strong correlations can be found between microhardness and rupture time in the current study, changes in hardness have been used as a measure of microstructural stability to assess the residual creep life for some high chromium FM steels. Sawada et al. [24] reported a decrease in hardness of Mod.9Cr–1Mo steel after 2000–3000 h creep testing at 700 °C (under a nominal stress of 40 MPa), which was attributed to the increasing inter-particle spacing as a result of coarsening and coalescence of the precipitates. In contrast, because of the shorter exposure time (b350 h) and the lower exposure temperature (650°), the extent of precipitate coarsening was probably very limited, leading to negligible change in hardness of the alloy before and after creep testing. Fig. 9 compares the microstructures of the alloy before and after creep testing. Again, the optical photomicrographs were taken in the grip section of the dog-bone specimens. No distinct change in microstructure was found for the bare alloy before (Fig. 9a) and after the creep test (Fig. 9b), and both exhibited a tempered martensitic structure. For the specimens that were aluminized at 650–700 °C (i.e., below the alloy's tempering temperature), the substrate remained tempered martensite after creep, as shown in Fig. 9c–d. For the specimen coated at 1050 °C, a similar microstructure was also observed after creep, Fig. 9e. In contrast, the heat-treated specimen displayed a completely different microstructure after creep testing (Fig. 9f), where polygonal grains were noted, resembling the ferrite phase [25]. 3.5. Microstructural difference between specimens coated and heat-treated at 1050 °C In order to understand the different microstructures of the specimens coated and heat-treated at 1050 °C after creep testing (Fig. 9e and f), additional specimens that were not creep tested were examined.
b
a
Coating Coating Void Substrate Substrate
Nitride
50 µm
c
10 µm
d Coating
Coating
Nitride
Substrate
10 µm
Substrate
10 µm
Fig. 7. Optical micrographs of cross-sections of coated Gr. 91 specimens after creep testing at 650 °C: (a) pure Al masteralloy, 6 h at 650 °C; (b) Cr–25Al masteralloy, 12 h at 700 °C; (c) Cr–15Al masteralloy, 12 h at 700 °C; and (d) Cr–15Al masteralloy, 6 h at 1050 °C.
Vickers Hardness & Rupture Time
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39 328
The microstructures of the as-coated and as-treated alloys are shown in Fig. 10, and the corresponding microhardness values can be found in Table 2. The substrate in the as-coated specimen displayed a martensite-like microstructure (Fig. 10a) with a very high hardness value of ~390 HV. On the other hand, the specimen after heat treatment consisted of mainly polygonal ferrite grains with a small fraction of martensite (Fig. 10b), similar to what was observed for the heat-treated and creep-tested sample (Fig. 9f). The hardness of the heat-treated specimen was ~229 HV, higher than the same specimen after creep testing (~196 HV). Since both of the coated and heat-treated specimens were held at 1050 °C for 6 h, the initial tempered martensitic structure should have been converted to austenite before cooling started. The martensite start temperature (Ms) of Gr. 91 is around 370 °C [13]. According to the readings from the thermocouple that was attached to the specimen in the crucible, the cooling rates were very similar for the two specimens, with an average of 1.2–1.4 °C/min from 1050 °C to room temperature and slightly faster cooling (2.1–2.4 °C/min) from 1050 to 370 °C. As a result of such a slow cooling process, ferrite would be expected to form in both specimens. However, ferrite was only observed in the
HV 500gf 292
Rupture Time (h)
290 251
205
211
207
205
192
196
26
AsAsreceived, received, w/o creep creep test tested
Coated, Coated, Coated, Heat650°C/6h, 700°C/12h, 1050°C/6h, treated, 10Al Cr-15Al Cr-15Al 1050°C/6h
Fig. 8. Vickers hardness of the Gr. 91 alloy treated under different conditions after the creep test and the corresponding rupture life. The hardness was measured in the grip portion (away from the deformation region).
a
37
b
10µm
c
10µm
d
10µm
10µm
e
f
10µm
10µm
Fig. 9. Optical microstructures of the Gr. 91 alloy before (a) and after (b–f) the creep test: (b) bare alloy, (c) coated at 650 °C (6 h, pure Al masteralloy); (d) coated at 700 °C (12 h, Cr–15Al masteralloy), (e) coated at 1050 °C (6 h, Cr–15Al masteralloy), and (f) heat-treated at 1050 °C (6 h). The specimens were etched with Villela's reagent.
38
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
a
b
10 µm Fig. 10. Optical microstructure of the Gr. 91 alloy prior to the creep test: (a) coated at 1050 °C; and (b) heat-treated at 1050 °C.
heat-treated specimen (Fig. 10b), and both microstructure (Fig. 10a) and hardness data (Table 2) confirmed the formation of martensite in the coated specimen [13]. It is therefore reasonable to hypothesize that the development of martensite in the coated specimen was somehow associated with the pack cementation coating process. The as-received Gr. 91 steel consisted of tempered martensite with subgrains of high dislocation density and M23C6 (M = Cr) carbide precipitates along the subgrain boundaries, as well as MX (M = V, Nb and X = C, N) carbonitrides [26]. When the alloy was heated to 1050 °C during pack aluminizing, the transformation from tempered martensite to austenite occurred in the alloy, leading to the dissolution of most carbides [13]. Aluminum is known as a ferrite stabilizer and has very limited solubility in austenite [27]. Diffusion of Al into the alloy produced a ferrite layer at the specimen surface with very low carbon solubility, and carbon was therefore rejected into the austenite core [28], as illustrated in Fig. 11. Although it is difficult to determine the carbon content in the austenite core using techniques such as electron probe microanalysis (EPMA) or EDXA due to its low concentration, the amount of carbon rejected from the coating into the substrate can be estimated. As mentioned earlier in Section 3.1, for the coating synthesized at 1050 °C, the outer layer was formed via outward diffusion of Fe and the inner layer by inward diffusion of Al. The interface between the outer and inner layers (Figs. 3d and 11) corresponded to the original substrate surface, i.e., the inner coating layer had originally been part of the substrate alloy. The carbon content in the as-received Gr. 91 alloy was ~ 0.11 wt.%. Assuming that as the Al diffused into the coating inner layer at 1050 °C, carbon was completely rejected to the austenite core and was dissolved homogeneously in the austenite, the amount of carbon which was originally in the substrate region corresponding to the coating inner layer would be equivalent to what was received by the austenite core. For the 2-mm thick creep specimen, owing to the symmetry only half of the specimen was considered. If values of 25 and 275 μm were assumed as the thicknesses for the outer and inner layers, respectively, the austenite core would be 725 μm thick, Fig. 11. It was estimated that the carbon content in the core region would be increased to ~0.16 wt.% (i.e., an increase of 0.05 wt.% over the original carbon content) as a result of carbon rejection from the coating inner layer. Gr. 91 steel is a highly hardenable alloy, and it generally transforms completely to martensite after air cooling from the austenitizing Table 2 Comparison in hardness of selected specimens. Treatment time (h)
Treatment environment
As-received 6 6
N/A Aluminization Vacuum
Hardness (HV) Before creep
After creep
205 390 229
207 251 196
temperature to room temperature. Hardenability is defined as the relative ability of a steel to avoid forming the soft ferrite phase when cooled from the austenitizing temperature [25]. Increasing the hardenability has the same effect as increasing the cooling rate, which shifts the formation of ferrite to longer times so that the steel can be cooled more slowly while still forming a martensitic microstructure. Carbon is an important element that effectively increases hardenability. For the present thin austenite core (~725 μm), it is rational to assume that an increase of 0.05 wt.% in carbon content was sufficient to increase the hardenability such that martensitic transformation occurred when the specimen was furnace cooled from 1050 °C after pack cementation. Subsequent creep test of ~180 h at 650 °C exerted an effect of tempering, during which the conversion of martensite to tempered martensite occurred. For the commercial Gr. 91 alloy, tempering is normally carried out at 730–780 °C with a minimum of 1 h of holding time to allow carbides to precipitate out homogeneously [13]. The creep testing temperature of 650 °C (equivalent to a lower tempering temperature) thus produced higher hardness (251 HV) in the coated alloy than that of the as-received alloy (205 HV) which was originally tempered at 730–780 °C, as indicated in Table 2.
Specimen Centerline Pack Chemicals Al
Cr
AlCl
AlClx
Austenite Core (~725 µm)
C
Gas
Outer Coating Layer (~25 µm)
Inner Coating Layer (~275 µm)
Fig. 11. Schematic of the pack cementation coating process. The darkened area represents a zone of carbon enrichment in austenite. Adapted from Reference [28].
B.L. Bates et al. / Surface & Coatings Technology 240 (2014) 32–39
3.6. Effect of the coating/substrate interaction on the alloy creep behavior As compared to the ferritic microstructure formed in the specimens heat-treated at 1050 °C (Fig. 10b), the martensitic structure in the coated alloy (Fig. 10a) compensated for the weak creep strength of the FeAl coating. Hence, the specimens aluminized at 1050 °C exhibited higher creep resistance than the ones heat-treated at 1050 °C, as observed in Fig. 5. Similar effect of carbon enrichment on creep performance has been reported for Alloy 617 (Ni–22Cr–12Co–9Mo–1Al–0.08C, wt.%) after oxidation testing at 1000 °C [29]. As a result of selective oxidation of Cr, a zone depleted in chromium carbides was formed immediately beneath the Cr2O3 scale, which extended ~ 500 μm deep into the 2-mm thick specimens after 6000-h exposure [29]. Carbon released from the chromium carbides in the depletion zone diffused into the specimen core, producing a carburization effect. Extensive carbide precipitation was found in the central region of the specimen, which increased the creep strength locally, and counteracted the loss of strength due to carbide dissolution in the Cr-depleted zone near the specimen surface. The present study further corroborates that the effects of coatings on creep resistance of FM steels are sometimes more complicated than just the loss of load-bearing cross sections, even though it often plays an important role in the reduction of creep strength for coated alloys. Compositional/microstructural changes of the alloy can take place during the coating fabrication process, due to either the thermal treatment itself (temperature and time) or interactions between the coating and the substrate. For typical aluminizing and chromizing processes, the coating elements (Al and Cr) are ferrite stabilizers. Phase transformations triggered by deposition and diffusion of Al and/or Cr can change the solubility of other elements originally present in the substrate, leading to enrichment or depletion of certain elements in the alloy core, similar to the case of the 1050 °C coatings in this study. In addition, for alloys containing high nitrogen or carbon, diffusion of Al or Cr can lead to the formation of AlN precipitates [12,20] or chromium carbides [26]. Attention should also be paid to the interactions between coating and substrate, particularly the resultant changes in concentration and/or redistribution of the minor elements (such as carbon or nitrogen) in the substrate, for these elements have been meticulously balanced in alloy design for optimal creep performance [26]. For the FM alloys, several studies have concurred that the coating process (e.g., temperature, time, heating/cooling rate, etc.) needs to be compatible with the heat treatment cycle of the alloy [7,8,14–16], which is again confirmed by the present creep testing results. 4. Conclusions As compared to the bare Gr. 91 alloy with a tempered martensitic microstructure, decreases in creep resistance at 650 °C were observed for all specimens coated with pack cementation aluminide coatings. For the specimens aluminized at 650–700 °C with coating thickness in the range of 18–125 μm, a reduction of 16–34% in rupture time was noticed, whereas for the specimens aluminized at 1050 °C with a thicker coating
39
of ~ 290 μm, the decrease of rupture lifetime was ~ 51%. However, the greatest reduction in rupture time (~94%) was found for the specimens heat-treated in vacuum, following the same thermal cycle as the 1050 °C coatings. The drastic difference in creep performance between the specimens coated and heat-treated at 1050 °C was due to the different microstructures that were formed in the alloy. Martensite was developed in the coated substrate alloy, as carbon was rejected from the aluminide coating into the central region of the specimen, leading to increased hardenability. In contrast, ferrite was present in the alloy heat-treated at 1050 °C, resulting in a much lower creep resistance. Acknowledgments The authors would like to acknowledge funding by the Department of Energy (DOE) Advanced Coal Research at U.S. Colleges and Universities, under Grant No. DE-FG26-06NT42674. Additional support is from DOE Fossil Energy Advanced Materials Research Program, under contract DEAC05-00OR22725 with UT-Battelle LLC and subcontract 4000071336 with Tennessee Technological University. References [1] R. Viswanathan, W. Bakker, J. Mater. Eng. Perform. 10 (2001) 81. [2] H. Nickel, Y. Wouters, M. Thiele, W.J. Quadakkers, Fresenius J. Anal. Chem. 361 (1998) 540. [3] H. Asteman, J.-E. Svensson, L.-G. Johansson, M. Norell, Oxid. Met. 52 (1999) 95. [4] E.J. Opila, Mater. Sci. Forum 461–464 (2004) 765. [5] R. Bianco, M.A. Harper, R.A. Rapp, JOM 43 (1991) 20. [6] B.A. Pint, Y. Zhang, P.F. Tortorelli, J.A. Haynes, I.G. Wright, Mater. High Temp. 18 (2001) 185. [7] A. Agüero, R. Muelas, A. Paster, S. Osgerby, Surf. Coat. Technol. 200 (2005) 1219. [8] V. Rohr, M. Schütze, E. Fortuna, D.N. Tsipas, A. Milewska, F.J. Pérez, Mater. Corros. 56 (2005) 874. [9] H.J. Kolkman, Mater. Sci. Eng. 89 (1987) 81. [10] S. Osgerby, D.F. Dyson, Mater. Sci. Eng. A121 (1989) 645. [11] S. Dryepondt, Y. Zhang, B.A. Pint, Surf. Coat. Technol. 201 (2006) 3880. [12] A. Agüero, R. Muelas, M. Gutiérrez, R. Van Vulpen, S. Osgerby, J.P. Banks, Surf. Coat. Technol. 201 (2007) 6253. [13] G. Guntz, M. Julien, G. Kottmann, F. Pellicani, A. Pouilly, J.C. Vaillant, The T91 Book, Vallourec Industries, France, 1990. 30–34. [14] F.J. Pérez, M.P. Hierro, J.A. Trilleros, M.C. Carpintero, L. Sánchez, J.M. Brossard, F.J. Bolívar, Intermetallics 14 (2006) 811. [15] B.L. Bates, Y.Q. Wang, Y. Zhang, B.A. Pint, Surf. Coat. Technol. 204 (2009) 766. [16] Y.Q. Wang, Y. Zhang, D.A. Wilson, Surf. Coat. Technol. 204 (2010) 2737. [17] R. Bianco, R.A. Rapp, J. Electrochem. Soc. 140 (1993) 1181. [18] ASTM E139-11, Standard Test Methods for Conducting Creep, Creep-rupture, and Stress-rupture Tests of Metallic Materials, ASTM International, West Conshohocken, PA, 2006., http://dx.doi.org/10.1520/E0139-11 (2003, www.astm.org). [19] In: G.F. Vander Voort (Ed.), Metallography and Microstructures, ASM Handbook, vol. 9, ASM International, 2004, p. 211. [20] Y. Zhang, B.A. Pint, K.M. Cooley, J.A. Haynes, Surf. Coat. Technol. 200 (2005) 1231. [21] S. Velraj, Y. Zhang, E.W. Hawkins, B.A. Pint, Mater. Corros. 63 (2012) 909. [22] B.A. Pint, Y. Zhang, L.R. Walker, I.G. Wright, Surf. Coat. Technol. 202 (2007) 637. [23] B.A. Pint, Y. Zhang, Mater. Corros. 62 (2011) 549. [24] K. Sawada, K. Miyahara, H. Kushima, K. Kimura, S. Matsuoka, ISIJ Int. 45 (2005) 1934. [25] R.L. Klueh, D.J. Alexander, E.A. Kenik, J. Nucl. Mater. 227 (1995) 11. [26] K. Maruyama, K. Sawada, J.-I. Koike, ISIJ Int. 41 (2001) 641. [27] N.V. Bangaru, R.C. Krutenat, J. Vac. Sci. Technol. B 2 (1984) 806. [28] M. Zheng, R.A. Rapp, Oxid. Met. 49 (1998) 19. [29] P.J. Ennis, W.J. Quadakkers, H. Shuster, Mater. Sci. Technol. 8 (1992) 78.