Effect of Long-Term High Temperature Oxidation on the Coking Behavior of Ni-Cr Superalloys

The service time of an industrial cracker is strongly dependent on the long-term coking behavior and microstructure stability of the reactor coil alloy. Super alloys are known to withstand temperatures up to even 1400 K. In this work, several commercially available alloys have been first exposed to a long term oxidation at 1423 K for 500 h, so-called metallurgic aging. Subsequently, their coking behavior was evaluated in situ in a thermogravimetric setup under ethane steam cracking conditions (Tgasphase = 1173 K, Ptot = 0.1 MPa, XC2H6 = 70%, continuous addition of 41 ppmw S/HC of DMDS, dilution δ = 0.33 kgH2O/kgHC) and compared with their unaged coking behavior. The tested samples were also examined using scanning electron microscopy and energy diffractive X-ray for surface and cross-section analysis. The alloys characterized by increased Cr-Ni content or the addition of Al showed improved stability against bulk oxidation and anti-coking behavior after application of metallurgic aging due to the formation of more stable oxides on the top surface.


Introduction
Surface technologies to reduce carbon deposition on the inner wall of industrial steam crackers are at the foreground of research because of their potential for improving the process economics. Due to coke deposition, the reactor cross-sectional area is reduced, leading to excessive pressure drop over its length. The latter results in a change in ethylene selectivity [1,2], while the coke layer decreases heat transfer from the furnace to the reactors. To keep the cracking severity stable over time and to compensate for the additional conductive heat resistance, the fuel flow rate to the furnace burners has to be increased, which makes that the reactor tube metal temperature rises over time. When the coke or the temperature of the coil is approaching the maximum allowable value, a decoking procedure with a steam/air mixture is necessary [3][4][5]. This cyclic cracking-decoking operation affects the annual production capacity if the plant is furnace limited, resulting in higher operating costs.
The focus of most of the technologies above mentioned is on the initial-fresh-coking behavior of the alloy, but not on its long-term behavior. Further scientific studies evaluated the coking performance of different alloys and coatings under industrially relevant conditions [23,44,[46][47][48][49][50]. However, the coking behavior after aging is not experimentally correlated with the microstructure of the samples in any of the studies. In the literature, typically only the effect on the microstructure and on the mechanical properties of the alloys is evaluated in long-term oxidation studies. Nevertheless, the coking behavior of an alloy when the temperature is close at its metallurgic limits is obviously of great importance and therefore it is studied in this work. The experimental data are compared both with non-aged and cyclically aged materials. SEM and EDX observations on both the surface and the cross sectional oxidation of the alloys as well as their microstructure allows for a better understanding of the studied phenomena.

Electro Balance Unit
The coking experiments were performed at the jet stirred reactor set-up of the Laboratory for Chemical Technology (LCT) of Ghent University. The Jet-stirred reactor (JSR) is designed in order to carry out thermogravimetric analyses on different reactor materials. Coke formed during the experiments is measured as a function of time under a broad range of conditions covering most of the industrial cases. The unit is capable of evaluating the effect of different pretreatment conditions (e.g., application of additives, temperature and duration of the pretreatment) as well as varying cracking conditions (e.g., the use of feed additives, steam dilution and cracking temperature) on coke formation, product distribution together with surface composition and morphology.
During an experiment the coke deposited on the metal sample, see Figure 1, and is accurately monitored online with resolution up to 0.1 µg per second. The weight signal is off-line processed afterwards in order to obtain coking rates. The reactor effluent is online analyzed via gas chromatography (GC) to evaluate the stability and reproducibility of the cracking conditions. The surface morphology and chemical composition of the coupons are typically analyzed off-line by means of SEM and EDX. The combination of the mentioned unit together with the experimental procedure gives valuable information on the coke formation phenomenon occurring during steam cracking of hydrocarbons and allows understanding the role of the studied material and conditions.  In the feed section, the hydrocarbon feedstock and water (or water with DMDS) are fed, evaporated at 523 K, mixed at 543 K and preheated at 903 K-which is the highest temperature that cracking does not occur-before sent to the reaction section, where steam cracking occurs leading to coke formation on the sample. The reactor effluent is quenched using an oil cooler set at 473 K to prevent liquefaction of the products while fraction of the effluent is withdrawn for online carbon oxides, C4 − and C5+ GC analysis. The rest is sent directly to the vent. The experimental unit is described in detail in earlier work [5,41,51]; however, for practical considerations a short description with the adaptation of some minor parts is included. Figure 2 gives a simplified diagram of the feed section of the unit.  In the feed section, the hydrocarbon feedstock and water (or water with DMDS) are fed, evaporated at 523 K, mixed at 543 K and preheated at 903 K-which is the highest temperature that cracking does not occur-before sent to the reaction section, where steam cracking occurs leading to coke formation on the sample. The reactor effluent is quenched using an oil cooler set at 473 K to prevent liquefaction of the products while fraction of the effluent is withdrawn for online carbon oxides, C4 − and C5+ GC analysis. The rest is sent directly to the vent. In the feed section, the hydrocarbon feedstock and water (or water with DMDS) are fed, evaporated at 523 K, mixed at 543 K and preheated at 903 K-which is the highest temperature that cracking does not occur-before sent to the reaction section, where steam cracking occurs leading to coke formation on the sample. The reactor effluent is quenched using an oil cooler set at 473 K to prevent liquefaction of the products while fraction of the effluent is withdrawn for online carbon oxides, C 4 − and C 5+ GC analysis. The rest is sent directly to the vent.

Metallurgic Aging
In this work, an additional oven was used to off-line pretreat the coupons that subsequently were tested in the JSR. In this oven, high temperature oxidation and/or nitration can be performed. The oven is capable of withstanding temperatures up to 1600 K for more than 500 h.

Cyclic Aging
Typically, the experiments conducted on the JSR unit consist of in total 8 cracking cycles-3 cycles of 6 h coking followed by decoking, followed by four short cycles of 1 h cracking and then decoking and a last cycle of 6 h-to evaluate the effect of cyclic cracking and decoking on each applied set of conditions. This procedure is called cyclic aging. An overview of the sequence kept in the aging experiments is given in Figure 3. In Tables 1 and 2 detailed process conditions of the cyclic and metallurgic aging procedures are summarized.

Metallurgic Aging
In this work, an additional oven was used to off-line pretreat the coupons that subsequently were tested in the JSR. In this oven, high temperature oxidation and/or nitration can be performed. The oven is capable of withstanding temperatures up to 1600 K for more than 500 h.

Cyclic Aging
Typically, the experiments conducted on the JSR unit consist of in total 8 cracking cycles-3 cycles of 6 h coking followed by decoking, followed by four short cycles of 1 h cracking and then decoking and a last cycle of 6 h-to evaluate the effect of cyclic cracking and decoking on each applied set of conditions. This procedure is called cyclic aging. An overview of the sequence kept in the aging experiments is given in Figure 3. In Table 1 and 2 detailed process conditions of the cyclic and metallurgic aging procedures are summarized.

Samples: Materials and Preparation
The metal coupons were made out 25/35 Cr/Ni, 35/45 Cr/Ni and 40/48 Cr/Ni alloys. Additionally, an Al-containing alloy is evaluated. These coupons are exposed to cyclic and metallurgic aging experiments. The coupons on which the coke will be deposited are cut as close as possible to the internal surface of industrial tubes using wire-cut electrical discharge machining (EDM) with an accuracy of 1 µ in dimensions 10 × 8 × 1 mm 3 (total surface area S = 1.96 × 10 −4 m 2 ). An illustration of the JSR coupon is depicted in Figure 4.

Samples: Materials and Preparation
The metal coupons were made out 25/35 Cr/Ni, 35/45 Cr/Ni and 40/48 Cr/Ni alloys. Additionally, an Al-containing alloy is evaluated. These coupons are exposed to cyclic and metallurgic aging experiments. The coupons on which the coke will be deposited are cut as close as possible to the internal surface of industrial tubes using wire-cut electrical discharge machining (EDM) with an accuracy of 1 μ in dimensions 10 × 8 × 1 mm³ (total surface area S = 1.96 × 10 −4 m 2 ). An illustration of the JSR coupon is depicted in Figure 4.

Coking Rate Determination
During the cracking experiment various data is logged, such as temperatures, mass flow rates and the evolution of the mass of the coupon and thus the deposited coke. Due to the movement of the coupon during cracking the raw balance output contains high frequency noise, which is removed by a low-pass filter in Matlab, resulting in a filtered weight curve. Since the raw balance output in the first 15 min of cracking is unstable due to the transient start-of-run effects the coking data is processed after these first 15 min. The filtering step also serves to withhold large disturbances from the raw coking data and to reduce the amount of data points. Next, the filtered weight curve is regressed by minimizing the total sum of squares to the following equation with parameters A, B, C and D: Equation (1) is called the fitted weight curve, with W the amount of coke (10 −6 kg) on the surface at time t (s). Via the fitted weight curve, data points for the first 15 min are calculated. The corresponding coking rate curve can be obtained by differentiating Equation (1): The reported coking rate is then easily obtained by converting to the desired units:

Coking Rate Determination
During the cracking experiment various data is logged, such as temperatures, mass flow rates and the evolution of the mass of the coupon and thus the deposited coke. Due to the movement of the coupon during cracking the raw balance output contains high frequency noise, which is removed by a low-pass filter in Matlab, resulting in a filtered weight curve. Since the raw balance output in the first 15 min of cracking is unstable due to the transient start-of-run effects the coking data is processed after these first 15 min. The filtering step also serves to withhold large disturbances from the raw coking data and to reduce the amount of data points. Next, the filtered weight curve is regressed by minimizing the total sum of squares to the following equation with parameters A, B, C and D: Materials 2018, 11, 1899 6 of 18 Equation (1) is called the fitted weight curve, with W the amount of coke (10 −6 kg) on the surface at time t (s). Via the fitted weight curve, data points for the first 15 min are calculated. The corresponding coking rate curve can be obtained by differentiating Equation (1): The reported coking rate is then easily obtained by converting to the desired units: With S coupon the surface area of the coupon in m 2 . The coking rate can also be calculated as a discrete derivative from the fitted weight curve as: From 15-60 min, the calculated coking rate is considered to be representative of the catalytic behavior of the material, while the one calculated from 5th to 6th h is referring to the long-term behavior of the material, corresponding to pyrolytic coking occurring when the complete surface of the coupon is covered by coke. In this work, the former is referred as initial coking rate while the latter as asymptotic.

Surface Characterization
SEM and EDX (JEOL JSM-7600F FEG SEM, Tokyo, Japan) are used to obtain information regarding the surface morphology and elemental composition of the samples. Both top surface and cross section analyses of coked samples were performed in this work. The top surface analysis gives a qualitative idea of the coke growth together with elemental quantitative data of the surface, performed in 10 kV and 20 kV. On the other hand, the cross section mappings explain the uniformity of the oxides generated due to exposure to an oxidative or carburizing environment. In this work, during the cross sectional analyses also the microstructure of the metallurgical aged alloys is evaluated. In the EDX analysis, the detected oxygen and carbon are omitted in the results both not to block viable results for other elements such as Cr, Mn and Fe, but also because their quantitative determination is not possible via EDX.

Experimental Results
In Table 3 an overview of the calculated initial and asymptotic coking rates for all the different experiments is given. The results can be interpreted in many different ways; initially the effect of the different aging procedures on the coking rates, keeping as reference the 1 cc and as fixed variable the elemental composition and subsequently, the effect of the different elemental composition of the coupon on the coking rates for all different applied conditions. Additionally, the effect of the different applied conditions on the elemental composition can be discussed using the SEM and EDX top and cross sectional analyses. Therefore, in the next sections, a more detailed discussion of the results is given.

Coke Formation
In Figure 5 the coking behavior for the different alloys under the complete range of the applied conditions is shown. The worst anti-coking performance is observed for the 25/35 Cr/Ni alloy, showing in all conditions at least 3 times higher coking rates than the other materials. The best coking behavior is observed for the 40/48 Cr/Ni alloy that is at least 10% better than the competition. For the catalytic coking rate, during the 1st cc, the differences are only significant for the 25/35 Cr/Ni alloy, coking almost a factor 3 more than the rest. For the remaining alloys the differences are negligible for the catalytic coking rates; however the ascending ranking in order of anti-coking performance is the following: 25/35 Cr/Ni < Al-containing < 35/45 Cr/Ni < 40/48 Cr/Ni Similar quantitative trends can be noted for the asymptotic coking rate, implying that the catalytic and pyrolytic coke formation are coupled phenomena. After application of cyclic aging even though the order of magnitude of the coking rates increases for all the alloys except for the 40/48 Cr/Ni one, the ranking of the materials remains the same. Here, the 25/35 Cr/Ni alloy cokes almost a factor 6 more than the best alloy, for both the catalytic and pyrolytic coking. Nevertheless, the differences remain similar among the other alloys. Application of cyclic aging leads to an increase of about double for the catalytic coking rates of the Al-containing and the 25/35 Cr/Ni alloy. For the asymptotic coking rate, a less pronounced increase of almost 30% is noted for these two alloys. For the remaining alloys-namely 35/45 and 40/48 Cr/Ni-the cyclic aging effect is negligible, implying their better anti-coking performance and stability during aging. The metallurgic aging, having as target to evaluate the materials after extreme oxidation conditions, similar to the end of run conditions of an industrial coil, is noted with orange in Figure 5. The anti-coking performance ranking of the alloys changes slightly when only considering the catalytic coking rate:

25/35 Cr/Ni < 35/45 Cr/Ni < Al-containing < 40/48 Cr/Ni
The Al-containing alloy cokes two times more than the high concentration of Cr/Ni alloys, while the 35/45 Cr/Ni alloy performs almost 15% worse than the former. The 25/35 Cr/Ni material remains the worst having almost a factor 10 higher coking rates. Unfortunately, the improved performance of the Al-containing alloy is not maintained during the pyrolytic coking stage. Here, the ranking is identical to the original one, with most of the alloys performing similarly to the cyclic aged ones, supporting the idea that the strongest factors influencing the radical pyrolytic coke formation are the cracking conditions, the applied pretreatment and the elemental composition, in agreement with previous observations [5].

Surface Analysis
The coke formation observations are in line with the top surface elemental compositions measured by means of EDX, see Table 4. In Figure 6 Ni and Fe are chosen as the representative catalytic coking elements, while in Figure 7 Cr is picked for all alloys and Al is picked for the Alcontaining, as representative for the passivating behavior [5,47]. The best anti-coking performance is measured for the 40/48 Cr/Ni alloy that has the lowest cumulative concentration of Fe and Ni on its surface, but also the highest amounts of Cr. The ranking of the tested alloys under the different conditions is fully aligned with the cumulative Fe and Ni content. It is clear that the existence of Cr in excess on the top surface of an alloy gives rise to the formation of Cr oxides having a good anticoking performance. However, for the Al-containing no clear conclusions can be extracted. Certainly

Surface Analysis
The coke formation observations are in line with the top surface elemental compositions measured by means of EDX, see Table 4. In Figure 6 Ni and Fe are chosen as the representative catalytic coking elements, while in Figure 7 Cr is picked for all alloys and Al is picked for the Al-containing, as representative for the passivating behavior [5,47]. The best anti-coking performance is measured for the 40/48 Cr/Ni alloy that has the lowest cumulative concentration of Fe and Ni on its surface, but also the highest amounts of Cr. The ranking of the tested alloys under the different conditions is fully aligned with the cumulative Fe and Ni content. It is clear that the existence of Cr in excess on the top surface of an alloy gives rise to the formation of Cr oxides having a good anti-coking performance. However, for the Al-containing no clear conclusions can be extracted. Certainly the absence of Fe and Ni in combination with the excess of elements responsible for the formation of oxides plays a significant role for the anti-coking performance of an alloy.    The ranking of the tested alloys under the different conditions is again fully aligned with the cumulative Fe and Ni content.

Cross Sectional Analysis
With the exception of the top surface elemental composition of the tested alloys, the cross sectional elemental composition is evaluated. It is found that the homogeneity of the oxides formed on a surface is also responsible for its subsequent anti-coking performance [5]. From Figure 8, it is clear that the Al-containing alloy has a very nicely oxidized surface after application of cyclic aging, forming mainly an Al oxide of roughly 2 to 3 μm thickness. After application of the metallurgic aging the oxide remains quite homogeneous although in some points it is replaced or overlaps with Cr oxides. The thickness is also decreased by almost 1 μm. This is in agreement with the coking observations. Probably the overlaying of Cr and Al improves the initial catalytic coking behavior of the alloy after metallurgic aging; however it does not affect significantly the pyrolytic coking rate to have an impact on the ranking of the alloys. The ranking of the tested alloys under the different conditions is again fully aligned with the cumulative Fe and Ni content.

Cross Sectional Analysis
With the exception of the top surface elemental composition of the tested alloys, the cross sectional elemental composition is evaluated. It is found that the homogeneity of the oxides formed on a surface is also responsible for its subsequent anti-coking performance [5]. From Figure 8, it is clear that the Al-containing alloy has a very nicely oxidized surface after application of cyclic aging, forming mainly an Al oxide of roughly 2 to 3 µm thickness. After application of the metallurgic aging the oxide remains quite homogeneous although in some points it is replaced or overlaps with Cr oxides. The thickness is also decreased by almost 1 µm. This is in agreement with the coking observations. Probably the overlaying of Cr and Al improves the initial catalytic coking behavior of the alloy after metallurgic aging; however it does not affect significantly the pyrolytic coking rate to have an impact on the ranking of the alloys. Materials 2018, 11, x FOR PEER REVIEW 11 of 18 In Figure 9-11, the superiority of the optimized pretreatment is vociferous. At the standard cyclic aging the alloys are forming a perfectly homogeneous Cr-Mn oxide layer of 2 to 3 μm thickness. Clearly the minor compositional differences imply that the Cr is less catalytic than Mn towards coke formation, while the increase in Ni is not enough to affect the coking performance of a material negatively if the latter is properly pretreated. A significant difference among the non Al and Alcontaining alloys is the thickness of the oxides formed after the metallurgic aging. Evidently, the addition of Al in the composition of an alloy halts its oxidation at higher temperatures [48,52]. Oxygen cannot move towards the bulk as easily as in the case of the non Al ones, probably because Al2O3-α is formed that is rather stable at high temperature. This phenomenon seems to be more pronounced in the 25/35 Cr/Ni alloy and less in the 35/45 Cr/Ni one, while in the 40/48 is almost absent. Additionally, for the 25/35 Cr/Ni alloy, Ni and Fe overlay with the passivating oxides for metallurgical aged sample, obviously catalyzing coke formation in line with the measured coking rates. In Figures 9-11, the superiority of the optimized pretreatment is vociferous. At the standard cyclic aging the alloys are forming a perfectly homogeneous Cr-Mn oxide layer of 2 to 3 µm thickness. Clearly the minor compositional differences imply that the Cr is less catalytic than Mn towards coke formation, while the increase in Ni is not enough to affect the coking performance of a material negatively if the latter is properly pretreated. A significant difference among the non Al and Al-containing alloys is the thickness of the oxides formed after the metallurgic aging. Evidently, the addition of Al in the composition of an alloy halts its oxidation at higher temperatures [48,52]. Oxygen cannot move towards the bulk as easily as in the case of the non Al ones, probably because Al 2 O 3 -α is formed that is rather stable at high temperature. This phenomenon seems to be more pronounced in the 25/35 Cr/Ni alloy and less in the 35/45 Cr/Ni one, while in the 40/48 is almost absent. Additionally, for the 25/35 Cr/Ni alloy, Ni and Fe overlay with the passivating oxides for metallurgical aged sample, obviously catalyzing coke formation in line with the measured coking rates.
The differences noted in the bulk of the alloys after exposure to the metallurgic aging process in the above mentioned cross sectional analyses led to the next section. This describes an attempt to identify possible microstructure patterns deeper in the bulk of the metallurgical aged alloys as a result of their exposure to temperatures close to their metallurgical limits. Materials 2018, 11, x FOR PEER REVIEW 12 of 18     The differences noted in the bulk of the alloys after exposure to the metallurgic aging process in the above mentioned cross sectional analyses led to the next section. This describes an attempt to identify possible microstructure patterns deeper in the bulk of the metallurgical aged alloys as a result of their exposure to temperatures close to their metallurgical limits.

Microstructure Analysis
In Figure 12 the microstructure of the four alloys after exposure to the long-term oxidation treatment is shown where in comparison to the ones from Figure 8 to Figure 11 even at higher resolution no elaborated microstructure is observed. Please note that the black spots are spots with increased concentration in Si, Al and C, implying the existence of SiC or Al4C3. Additionally, also Cr-C rich zones, on the one hand, and Nb-C rich zones, on the other hand, are observed in the microstructure. The dark grey areas indicate a precipitation of Cr and C while the white ones corresponds to an increased concentration of Nb and C. It is known that metallurgic crack initiation after high temperature oxidation happens mainly in Chromium carbides rich zones [53,54]. Also, when the microstructure of an alloy is not significantly elaborated, the alloy is not significantly aged [55,56]. The most pronounced microstructure is noted for the Fe-Ni-Cr alloys, justifying that the microstructure of an alloy is independent of its coking behavior. The 25/35 and 40/48 Cr/Ni alloys have almost identical microstructure pattern, but totally different coking behavior after metallurgic aging, while the 35/45 is the most pronounced, even though its coking behavior is better than the former and worse than the latter. The Al-containing alloy shows no elaborated microstructure, but its coking behavior is still not the best. Therefore no clear conclusions can be made towards coke formation from the current microstructural analysis.

Microstructure Analysis
In Figure 12 the microstructure of the four alloys after exposure to the long-term oxidation treatment is shown where in comparison to the ones from Figure 8 to Figure 11 even at higher resolution no elaborated microstructure is observed. Please note that the black spots are spots with increased concentration in Si, Al and C, implying the existence of SiC or Al 4 C 3 . Additionally, also Cr-C rich zones, on the one hand, and Nb-C rich zones, on the other hand, are observed in the microstructure. The dark grey areas indicate a precipitation of Cr and C while the white ones corresponds to an increased concentration of Nb and C. It is known that metallurgic crack initiation after high temperature oxidation happens mainly in Chromium carbides rich zones [53,54]. Also, when the microstructure of an alloy is not significantly elaborated, the alloy is not significantly aged [55,56]. The most pronounced microstructure is noted for the Fe-Ni-Cr alloys, justifying that the microstructure of an alloy is independent of its coking behavior. The 25/35 and 40/48 Cr/Ni alloys have almost identical microstructure pattern, but totally different coking behavior after metallurgic aging, while the 35/45 is the most pronounced, even though its coking behavior is better than the former and worse than the latter. The Al-containing alloy shows no elaborated microstructure, but its coking behavior is still not the best. Therefore no clear conclusions can be made towards coke formation from the current microstructural analysis.

Conclusions
The coking behavior of four different alloys was evaluated in a thermogravimetric unit after their exposure to a long-term high temperature oxidation for 500 h more representative of the end of run conditions. The results indicated that for the same conditions the coking resistance is improved by increasing the content of Cr and Ni of a high temperature alloy. Cyclic aging had only a minor effect on the tested alloys except for the 25/35 Cr/Ni alloy. The anti-coking performance of the materials decreases after the metallurgic aging. For the best tested material, a negligible effect was noticed after application of the long-term oxidation. The catalytic and pyrolytic coking behavior of the materials is quite aligned, implying that the two phenomena are interconnected. In line with previous work and literature it can be concluded that the presence of Fe and Ni is detrimental for the anti-coking performance of an alloy. Most of the alloys indicated a rather homogeneous coverage of their surface by oxides-mainly of Cr nature for the non Al-containing alloys. For the Al-containing alloy, the thickness of the oxides covering typically the surface after cyclic aging is reduced by almost 1 μm, in contrast with the other alloys for which the oxides' thickness was significantly increased. In line with the available literature, the addition of Al in the content of a high temperature alloy, forms Al oxides that protect the bulk from further oxidation. No microstructure was found in any alloy, which implies that the alloys all are quite resistant in terms of aging. However the non Al-containing alloys showed increased concentration of Carbides of Si and Nb in comparison with the Al-containing one. No correlation between the microstructure and the coking behavior of an alloy was found.

Conclusions
The coking behavior of four different alloys was evaluated in a thermogravimetric unit after their exposure to a long-term high temperature oxidation for 500 h more representative of the end of run conditions. The results indicated that for the same conditions the coking resistance is improved by increasing the content of Cr and Ni of a high temperature alloy. Cyclic aging had only a minor effect on the tested alloys except for the 25/35 Cr/Ni alloy. The anti-coking performance of the materials decreases after the metallurgic aging. For the best tested material, a negligible effect was noticed after application of the long-term oxidation. The catalytic and pyrolytic coking behavior of the materials is quite aligned, implying that the two phenomena are interconnected. In line with previous work and literature it can be concluded that the presence of Fe and Ni is detrimental for the anti-coking performance of an alloy. Most of the alloys indicated a rather homogeneous coverage of their surface by oxides-mainly of Cr nature for the non Al-containing alloys. For the Al-containing alloy, the thickness of the oxides covering typically the surface after cyclic aging is reduced by almost 1 µm, in contrast with the other alloys for which the oxides' thickness was significantly increased. In line with the available literature, the addition of Al in the content of a high temperature alloy, forms Al oxides that protect the bulk from further oxidation. No microstructure was found in any alloy, which implies that the alloys all are quite resistant in terms of aging. However the non Al-containing alloys showed increased concentration of Carbides of Si and Nb in comparison with the Al-containing one. No correlation between the microstructure and the coking behavior of an alloy was found. to revise and finalize the manuscript for submission and publication. K.V. provided important insight in the surface analysis presented in the manuscript. All authors contributed to a review of the manuscript before and after revision.
Funding: This research is funded by 'Long Term Structural Methusalem Funding by the Flemish Government'-grant number BOF09/01M00409.

Conflicts of Interest:
The authors declare no conflict of interest.