Design of Oxidation Resistant Alloys using Combinatorial Approaches with Chemically Graded Materials

Design of Oxidation Resistant Alloys using Combinatorial Approaches with Chemically Graded Materials | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Design of Oxidation Resistant Alloys using Combinatorial Approaches with Chemically Graded Materials Sabrina GHANES, Mikael PERRUT, Enrica EPIFANO, Matthieu DEGEITER, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4742772/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 28 Aug, 2024 Read the published version in High Temperature Corrosion of Materials → Version 1 posted 7 You are reading this latest preprint version Abstract This work introduces a new high-throughput method to characterize the oxidation behavior of chemically graded Ni-based alloys in order to feed databases destined to numerical metallurgy approaches. A Ni-wCr-3Al (w ∈ [0, 30]) chemically graded material was obtained from two homogeneous samples by a diffusion couple method at 1300°C for 100h. The composition range was selected in order to observe the three types of oxidation behavior identified in the reference work of Giggins and Pettit [ 10 ]. The excellent agreement between simulated and experimental diffusion profiles validated the experimental method used to manufacture the chemically graded material (CGM). The CGM was then oxidized at 1200°C in air. Surface and cross-section characterization were conducted by SEM/EDS and Raman spectroscopy to identify the oxides formed on the CGM. To accelerate the Raman characterization treatment, a method linking Principal Component Analysis (PCA) and K-means unsupervised clustering algorithm was developed [ 11 – 12 ]. It allowed for the identification of the oxide type without peak indexation issues and is well-suited for CGM. These results show that results similar to well-recognized reference experiments [ 10 ] can be achieved using only one CGM. alloy by design high-throughput method diffusion gradient combinatorial metallurgy Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Introduction In materials science and engineering, the development of new materials with improved properties remains the key to technological progress for industry. Integrated Computational Materials Engineering (ICME) has emerged as the new approach to accelerate the development and optimization of materials. ICME integrates computer modelling tools and experimentations into all stages of the materials development process, from design to elaboration, processing and application. This multi-scale approach enables the understanding and prediction of the behavior of materials under real conditions by considering the interactions among structure, processing and performance [ 1 – 2 ]. Initiatives such as the Materials Genome aim to develop the use of artificial intelligence and modelling techniques to create databases and algorithms that are able to predict the performance of materials based on their atomic structure. Computational methods such as CALPHAD (CALculation of PHAse Diagrams) have been widely used to predict domains of interest. However, the effectiveness of these methods is highly dependent on the quality of the databases available. Current databases are often incomplete as the acquisition of data can be difficult and time-consuming. For this reason, it is necessary to develop high-throughput manufacturing and characterization methods to feed the databases. Xiang [ 3 ] and others have implemented strategies such as combinatorial metallurgy, which involves rapidly investigating and optimizing metallic material properties through the generation and evaluation of alloy libraries [ 3 – 5 ]. The first works used vapor deposition techniques such as co-evaporation and co-deposition to obtain thin films with composition gradients [ 3 – 4 ]. This approach enabled the utilization of masks to selectively target designated regions within the composition space or to enhance the uniformity of the applied gradient [ 3 – 5 ]. However, extrapolating properties of solid materials from thin films is difficult due to size effects, residual stresses and potential compositional differences. For this reason, methods such as diffusion couples/multiples have been developed to study bulk material microstructures and properties directly [ 6 – 9 ]. In the field of aeronautics, there is a need for improved creep and oxidation resistant nickel-based alloys for turbine blades applications. However, the trial-error method used until now to develop alloys is expensive and time-consuming thus making it incompatible with the acceleration of alloy design. The aim of this study is to develop a high-throughput manufacturing and characterization method adapted to nickel-based chemically graded material (CGM) in order to feed databases for the application of oxidation resistant alloys. For this purpose, the Ni-Cr-Al ternary system was used as it is the basis for the development of nickel-based superalloys. According to Giggins and Pettit [ 10 ], three types of oxidation behaviors can occur for this model system. The type I consists in the formation of an outer layer of nickel oxide over internal alumina and chromia precipitates. The type II exhibits the formation of an outer layer of chromia and internal alumina precipitates. Finally, the type III consists in the formation of a continuous external alumina layer. These three types of oxidation behaviors are represented on Fig. 1 as an oxidation map at 1200°C [ 10 ]. The aim of this study is to characterize the three types of oxidation behavior of the Ni-Cr-Al system, with only one sample. Materials and Methods To observe the composition limits for the formation of oxides predicted by the oxide map [ 10 ] in Fig. 1 while working in the single-phase γ domain, the composition was set at 3 wt% Al with a chromium gradient. First, five Ni-wCr-3Al model alloys (w = 0, 5, 10, 20 and 30 wt%) were produced through arc melting as reference for comparison with the CGM. The five compositions were selected to study the limits of the occurrence domains of oxides expected to form in the Ni-Cr-Al system (cf Fig. 1 ). The five ingots were cut into little plates of 6 mm x 6 mm x 2.5 mm. Then, a chemically graded material was developed for the Ni-Cr-Al model system to study the oxide frontiers at 3 wt% Al with a chromium gradient from 0 to 30 wt% Cr. To create this CGM, two alloys were assembled, i.e. Ni-0Cr-3Al and Ni-30Cr-3Al, and annealed for 100 h at 1300°C under argon atmosphere. The alloys used for the diffusion couple had the same dimension as the reference samples. To prevent the chromium volatilization during the interdiffusion, a piece of Ni-30Cr-3Al was added in the alumina cylinder to saturate the atmosphere in Cr-rich gazeous species. The interdiffusion duration and temperature were selected on the basis of thermo-kinetic calculations, performed using Thermo-Calc and DICTRA software with TCNI8/MOBNI5 databases, in order to obtain a composition gradient spreading across 1 mm. This length was chosen to facilitate the CGM characterization using techniques with spatial resolutions ranging between 1 and 50 micrometers. For the diffusion simulations, closed system boundary conditions were used. The initial aluminum profile was set at a constant value of 3 wt%, the chromium profile was a step function from 30 wt% to 0 wt% and the nickel was the dependent component. A double-geometric grid of 100 points was used with a lower ratio of 0.9 and an upper ratio of 1.1. Finally, after the interdiffusion treatment at 1300°C, a HIP post-treatment was performed at 1150°C under 1300 bar during 4h in argon atmosphere to close any pores. Two CGM, C01 (underwent a HIP treatment) and C03 (no HIP), were produced. C01 was used for surface characterization and C03 for cross-sectional characterization. The reference samples and the CGM were oxidized in laboratory air at 1000°C and 1200°C, for 20 hours, to compare with Giggins and Pettit results. All oxidation treatments were conducted in a horizontal furnace calibrated before each treatment. Cross sections of samples were prepared using conventional metallographic methods including copper-nickel plating, grinding with SiC papers and polishing with diamond suspension (1/4 µm). Surface and cross section analyses were performed using optical microscopy, SEM/EDS on Zeiss MERLIN with SamX detector (the concentration profiles after assembling were quantitatively measured in comparison to pure Al, Cr, Ni standards, using IDFIX software [ 9 ]) and Raman spectroscopy with the Renishaw inVia Reflex confocal micro-Raman spectrometer with a laser wavelength of 532 nm. Results 1. Oxidation behavior of reference samples The cross sections of the reference samples oxidized at 1200°C for 20 h are illustrated in Fig. 2 . Several oxides were identified on the cross section of the oxidized Ni-5Cr-3Al alloy (Fig. 2 . a ) by means of EDS analyses. The reaction between NiO and Al 2 O 3 results in the formation of mixed oxide rich in aluminum and nickel with a small amount of chromium likely to be a spinel. XRD analysis confirmed the presence of spinel (not shown here). This oxidation behavior corresponds to the type I described by Giggins and Pettit. On the Ni-20Cr-3Al in Fig. 2 . b , a chromia layer at the surface and internal alumina precipitates have been detected. In some parts of the sample (as seen on Fig. 2 . b ) the internal precipitates have coalesced to form a continuous alumina layer beneath the chromia layer. This oxidation behavior is characteristic of a type II. In Fig. 2 . c , Ni-10Cr-3Al formed a thin and continuous protective alumina layer with a limited amount of mixed oxide rich in Ni, Cr and Al (a spinel as confirmed by XRD) on the top in some areas. Since the alumina is continuous, the alloy is considered as type III. The same characterizations were performed on the reference samples oxidized at 1000°C and the same three oxidation behaviors were identified. For practical reasons, only the results obtained at 1200°C are presented in this article, but the conclusions remain the same. 2. Characterization of chemically graded samples The thickness of the interdiffusion layer in the diffusion couple is a critical factor for the high-throughput method. In order to ease the characterization of the CGM and of its oxidation, a length of several millimeters would be the best choice. However, the diffusion length given by equations (1) and (2) is limited to restrictions on time and temperature. A compromise was necessary and the diffusion length of the chromium gradient was set at 1 mm. $$\:{L}_{d}=4\sqrt{D\cdot\:t}\:\:\:\:\:\:\:\left(1\right)$$ $$\:D={D}_{0}{exp}\left(\frac{-Q}{RT}\right)\:\:\left(2\right)$$ L d : Diffusion length (m) D: interdiffusion coefficient (m²/s) Once the length of the gradient is set, the second requirement is to limit the processing time, so increasing the temperature is a key. Nevertheless, temperature is limited by two conditions. The first one is the alloy lowest solidus temperature, which is estimated to be around 1360°C. The second condition is imposed by the oxidation experiment, the aim is to prevent an evolution of CGM in the metal during the oxidation experiment. Considering these conditions, the interdiffusion treatment was set at 1300°C. Finally, interdiffusion simulations were performed to assess the duration required to obtain a gradient of about 1 mm at 1300°C. For that, 100 h were necessary. To evaluate experimentally the diffusion length obtained in these conditions, the cross section of the couple (Fig. 3 . a) was analyzed by EDS. The results are compared to the simulation predictions in Fig. 3 . b , using DICTRA. An excellent agreement between the experimental and the simulated profiles for both Cr and Al can be observed. 3. Characterization of oxidized chemically graded sample 3.1. Surface characterization Secondary electron mapping revealing the sample topology was performed on the C01 oxidized sample (Fig. 4 . a ). It appears that the thickest oxide formed on the part of the sample with the lowest chromium content. The oxide in the chromium-rich zone is thinner than the oxide in the chromium-poor zone, but thicker than the oxide in the transition zone. It should be specified that the thickness of the thickest oxide was even larger than shown in Fig. 4 because a part of the oxide spalled off after the oxidation treatment. The correlation between thickness and the type of oxide can be established using Fig. 4 . b . The thickest oxide corresponds to the nickel rich oxide. Thanks to the reference samples, it may be assumed that it corresponds to NiO. The thinnest oxide is an aluminum-rich oxide, which should be α-alumina. The oxide formed in the chromium-rich part of the CGM is rich in chromium, thinner than nickel oxide but thicker than alumina, and should correspond to chromia. To verify theses hypothesis, a Raman spectroscopy profile was realized on the surface of the oxidized CGM and presented in the Fig. 5 . Raman spectroscopy allowed the identification of phases through their characteristic vibrational spectra, unlike EDS which only identified chemical elements. Each spectrum was compared to reference spectra of expected oxides among nickel oxide, α-alumina and chromia. According to Fig. 5 . b , nickel oxide was detected in the chromium-poor area, α-alumina in the transition area and chromia in the chromium-rich part of the CGM. Given their different thicknesses, focusing all the three oxides simultaneously was not possible, resulting in the one detected in the chromium-poor area being blurred. Several tests were carried out to evaluate the importance of focusing. For nickel oxide and chromia the signal was so intense that it was not necessary to set the focus to detect the peaks. In contrast, the characteristic peaks of alumina were weak, so setting the focus was necessary to optimize the signal. To overcome the focus problem, which requires changing the setting for each point, Raman spectroscopy mapping was performed by focusing on the transition zone (Fig. 6 ) where 781 spectra were acquired. Each spectrum was compared automatically with reference spectra, which were carried out on the reference samples, to determine the oxide type. However, peak shifts in a significant number of spectra made the oxide identification challenging, as the direct comparison with the reference spectra often failed. Given the independent processing of each spectrum is time-consuming, a high-throughput post-treatment method was set up. For this purpose, a data-driven analysis suited to the classification of Raman spectra by oxide type was developed. First, data was pre-treated to scale and smooth the spectra after removal of their baseline. A principal component analysis (PCA) was carried out, which amounts to expressing the full spectra values in the basis where the data covariance matrix is diagonal, with its eigenvalues sorted in descending order [ 11 ]. We chose this linear dimension reduction technique for its simplicity, its ease of implementation and for its strong advantage of being fully explainable. Figure 6 . b. highlights the cumulative variance of our data, explained with an increasing number of principal component (PC). 94% of the data variance is carried by the first two PCs, while keeping five PCs guarantees a 99% reconstruction of the data variance. Therefore, Raman spectra were projected from the 781-dimensional space they are observed in, to a 2-dimensional linear subspace spanned by the first two PCs, while preserving 94% of the explained variance (cf Fig. 6 . c ). Each point on this figure is the projection of a spectrum, and they tend to cluster into either narrow or widespread distributions. Underlying our analysis is the idea of measuring distances between Raman spectra: spectra similar to one another will have their projections clustered in neighboring regions of the linear subspace. Matching each cluster with the corresponding oxide-type – chromia, α-alumina, or nickel oxide – requires a clustering analysis for which we present a first approach using the K-means algorithm among Machine Learning techniques. The K-means unsupervised clustering algorithm was used to divide the dataset into multiple clusters [ 12 ]. The cluster number K is fixed during the initialization step of the algorithm, and we chose K = 3 because, according to the reference samples, we expect three types of oxides. The data is then sorted according to the nearest clustering center. Once the clusters were identified, each point was assigned a label and a color. Mapping back the labeled spectra to our oxidized chemically graded sample, we obtained the oxide map reconstruction presented at Fig. 6 . d , where an oxide type is assigned to each pixel. The Raman mapping results validated the hypothesis established by the EDS characterizations. The three oxide types which were expected to form from the literature and reference samples were found i.e. nickel oxide in the chromium-poor zone, alumina in the transition zone and chromia in the chromium-rich part. 3.2. Cross section characterization of the oxidized chemically graded alloy A cross section of the C03 oxidized sample is represented in Fig. 7 . a . During polishing the copper-nickel plating layer was detached from the oxide layer, which explains the black hole observed in the optical micrograph. EDS maps of Al, Cr, Ni and O were carried out to identify the oxide and inner precipitates composition. According to Fig. 7 . b the thinner oxide layer and the inner precipitates are aluminum and oxygen rich. The oxide layer in the chromium-rich part of the CGM is chromium and oxygen rich, and the oxide island is rich in nickel and oxygen. Note that a part of the thicker oxide has been removed right after the oxidation treatment because of low adhesion. In certain areas, the nickel oxide layer has been preserved, enabling the measurement of its thickness. As the surface observation has already highlighted, the nickel oxide, which has grown in the chromium poor part of the CGM, is much thicker than the other oxides. As a comparison, the thickness of the nickel oxide layer is about 113 ± 3 µm, while the chromia and alumina layers are about 13 ± 3 µm and 3 ± 1 µm, respectively. Moreover, there is a large area with internal oxide precipitates of 180 ± 8 µm under NiO. The pores present in the chromium rich part of the CGM in Fig. 7 . a. were formed during the interdiffusion treatment, prior to oxidation, and they are due to Kirkendall effect [ 13 ]. To complete the EDS analysis, a Raman map has been performed and illustrated in Fig. 7 . c. The comparison of our reference spectra with spectra acquired by mapping confirms that the thin layer and inner precipitates are alumina, the thicker layer is chromia and the oxide island is nickel oxide. Based on EDS and Raman map, the compositions required to form a type I is situated in [0Cr-3Al; 5Cr-3Al] ± 1%w composition range, a type III in [5Cr-3Al; 12Cr-3Al] ± 1%w and a type II in [12Cr-3Al, 30Cr-3Al] ± 1%w. Discussion Thanks to the excellent agreement between the diffusion simulations and experiments, the CGM was successfully obtained with the desired chromium gradient, spreading over of 1 mm. The same three types of oxide observed on the references were identified on the oxidized chemically graded alloy. It means that three types of behavior with very different oxidation kinetics can be observed on the same sample. From the measured oxide thicknesses, the parabolic rate constants values were calculated for the three type of oxides and presented in Table 1 . Table 1 Parabolic rate constants estimated from local oxide scale thicknesses for the oxidation of CGM at 1200°C in air during 20h and parabolic rate constants for the oxidation of a typical Group (I, II and III) alloy at 1200°C in 0.1 atm of oxygen during 20h [ 10 ] k p (mg².cm -4 .s -1 ) k p (Giggins & Pettit [10]) (mg².cm -4 .s -1 ) Type I 4.10 -3 10 − 2 Type II 8.10 -5 10 − 4 Type III 4.10 -6 10 − 5 Comparing k p of the three types of oxide with those reported by Giggins & Pettit [ 10 ], it is observed that the parabolic rate constants for alumina and chromia are slightly lower than the reference value. As part of the nickel oxide layer has peeled off, the thickness would be expected to be higher resulting in a higher k p . Nonetheless, all three k p values remain relatively close to the reference. Thus, thickness difference helps to identify the transition areas between Type I, II and III. The compositions required to form a type I is situated in [0Cr-3Al ; 5Cr-3Al] ± 1%w composition range, a type III in [5Cr-3Al ; 12Cr-3Al] ± 1%w and a type II in [12Cr-3Al, 30Cr-Al] ± 1%w; which is very close to the limits announced in the literature [ 10 ]: Type I ∈ [0Cr-3Al; 5Cr-3Al] Type III ∈ [5Cr-3Al ; 15Cr-3Al ] and Type II ∈ [15Cr-3Al, 30Cr-Al and > 30Cr-Al]. These results allowed to measure the critical contents of chromium needed to promote external growth of a protective alumina layer rather than aluminum internal oxidation. For turbine blades, the aim is always to form an outer layer of the most protective oxide, i.e. α-alumina. These results suggest that the critical content of chromium required for the transition from internal to external oxidation of aluminum in Ni-wCr-3Al is ≈ 5 ± 1 wt% Cr. It demonstrates in a new and direct way the third element effect [ 10 , 14 ], i.e. adding chromium to Ni-Al alloys promotes the formation of external layer of alumina at lower aluminum content. In this work, EDS and Raman spectroscopy were used as characterization methods to identify the oxide type. However, the current post treatment was not always adapted to the CGM characterization. Thus, this article proposes a high-throughput characterization linking Raman spectroscopy and unsupervised machine learning to accelerate post processing. The first results presented here are encouraging, identifying the oxide type without peak indexing problems. The clustering part of the analysis is a work still in progress, for which we have already identified several improvement directions. We note that the K-means algorithm used as a first clustering method has its limitations, and is known to perform poorly when the clusters have differing variances and densities, as well as when they display anisotropy. In a high-throughput characterization approach, this remains acceptable, given that it is possible to identify the oxide formation regions. Raman enables the direct identification of the oxide phases, contrary to EDS which only yields their chemical composition. Moreover, the chemically graded material can be precisely characterized by Raman spectroscopy with a resolution of 0.6 µm, better than with EDS (few µm). However, in the cross-sectional study, characterizing the surface oxide was difficult because of a weak signal, specially of alumina. Additionally, it was not possible to characterize the substrate because metals generally have no Raman active modes or present modes with very weak intensity. For these reasons, both characterization techniques are complementary. Conclusions and Perspectives This work shows that it is possible to obtain similar results as Giggins and Pettit, using far fewer samples thanks to CGM. There were doubts about forming a continuous external alumina layer with only 3 wt% Al. However, these experimental results confirm, with only one sample, how this is possible, thanks to the ‘third element effect’ [ 10 , 14 ]. Indeed, the addition of Cr allows alumina to be formed at lower Al contents than required for a binary Ni-Al alloy. The excellent agreement between simulated and experimental diffusion profiles validated the experimental method used to design CGM. The next step is to extend this method to more complex systems, in order to increase the data collected in only one experiment. For this purpose, diffusion multiplet will be realized to obtain a 2D gradient in one sample. In addition, this study introduced a high-throughput characterization method combining EDS and Raman maps. The CGM was characterized precisely combining surface and cross-sectional analysis. EDS provided a preliminary analysis that was confirmed by Raman spectroscopy, which directly identifies the characteristic spectrum of each oxide. However, the current method of indexing Raman peaks required a certain amount of time to work on a case-by-case basis, whenever there were shifts or broadening of the peaks. The post-treatment characterization proposed in this article overcomes this problem, offering an identification method suitable for any oxide. By using the PCA method, the high number of dimensions of the initial problem was reduced to two dimensions, thereby reducing calculation time. The PCA projection of the data led to a new 2-dimensional representation of the data in which clusters were formed. Each cluster represented an oxide type in the subspace. The K-means unsupervised clustering algorithm allowed the identification of the three clusters formed by nickel oxide, chromia, and alumina spectra. The Raman map was then reconstructed and compared with the EDS results. Both EDS and Raman results were in line with reference samples and literature, proving the validity of this method. This method can be reliably correlated to the oxide type, but sometimes struggles to identify certain points that are far from the cluster centers. For this reason, a forthcoming work will be dedicated to improving the clustering algorithm and compare this method with a deep learning approach. Declarations Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Author Contribution S.G did the experiment and write the first draft of the paper including figs and tables.M.P coordinates the project in which this work takes place.S.G, M.P, E.E, and D.M wrote the main manuscript text. S.G and M.D wrote the PCA and K-means part of this article.T.V and Y.C helped during the conception and the writing of the article. All authors reviewed the manuscript. Acknowledgement Financial supports by the French Armed forces Ministry and the French Aerospace Lab are gratefully acknowledged. The authors are grateful to C. Rio, N. Horezan, and Q. Barrès for their valuable help with SEM/EDS mapping, J.S. Mérot for technical support and his involvement and A. Andrieux for providing access to the Raman spectrometer. References Alberi K et al. The 2019 materials by design roadmap. J. Phys. D: Appl. Phys. 52, 013001, 2019. McDowell DL et al. Chapter 3 - Overview of the Framework for Integrated Design of Materials, Products, and Design Processes in Integrated Design of Multiscale, Multifunctional Materials and Products, 39–64 , 2010. Xiang XD et al. A Combinatorial Approach to Materials Discovery. Science 268, 1738–1740, 1995. Takahashi R et al. Design of Combinatorial Shadow Masks for Complete Ternary-Phase Diagramming of Solid State Materials. J. Comb. Chem. 6, 50–53, 2004. Potyrailo R et al. Combinatorial and High-Throughput Screening of Materials Libraries: Review of State of the Art. ACS Comb. Sci. 13, 579–633, 2011. Zhao JC. Phase Diagram Determination Using Diffusion Multiples. in Methods for Phase Diagram Determination 246–272, 2007). Zhao JC. Combinatorial approaches as effective tools in the study of phase diagrams and composition–structure–property relationships. Progress in Materials Science 51, 557–631, 2006. Perrut M. Life Prediction Methodologies for Materials and Structures, Journal Aerospace Lab, 2015. Lacour-Gogny-Goubert A et al. Effect of Mo, Ta, V and Zr on a duplex bcc + orthorhombic refractory complex concentrated alloy using diffusion couples. Intermetallics 124, 106836, 2020. Giggins CS & Pettit FS. Oxidation of Ni-Cr-AI Alloys Between 1000°C and 1200°C. J.Electrochem.Soc 118, 9, 1971. Greenacre M et al. Principal component analysis. Nat Rev Methods Primers 2, 1–21, 2022. Pedregosa F et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research 12, 2825–2830, 2011. Nesbitt JA & Heckel RW. Interdiffusion in Ni-Rich, Ni-Cr-Al alloys at 1100 and 1200°C: Part I. Diffusion paths and microstructures. Metall Trans A 18, 2061–2073, 1987. Niu Y, Zhang XJ, Wu Y & Gesmundo F. The third-element effect in the oxidation of Ni–xCr–7Al (x = 0, 5, 10, 15 at%) alloys in 1 atm O2 at 900–1000°C. Corrosion Science 48, 4020–4036, 2006. Additional Declarations No competing interests reported. Supplementary Files Supplementarymaterial.png Cite Share Download PDF Status: Published Journal Publication published 28 Aug, 2024 Read the published version in High Temperature Corrosion of Materials → Version 1 posted Editorial decision: Accepted 21 Jul, 2024 Reviews received at journal 21 Jul, 2024 Reviewers agreed at journal 21 Jul, 2024 Reviewers invited by journal 20 Jul, 2024 Editor assigned by journal 20 Jul, 2024 Submission checks completed at journal 18 Jul, 2024 First submitted to journal 15 Jul, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4742772","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":329798209,"identity":"76052f67-84fb-4767-bf56-fa496c52847f","order_by":0,"name":"Sabrina GHANES","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABAElEQVRIiWNgGAWjYFACHgaGBBAtAebZQEVtcKjGoiUNKpqGQzVMCwNCy2HCWuTbzx778HCHHYN8dPOzBx/+nJczZ+99+IEh4R5OLQZn8pJnJJ5JZjC8c8zccGbbbWPLnuPGEgwJxbi1MOQYMyS2MTMYzkgwk+ZtuJ244UYagwTjjwTcDut/A9JSD9SS/k36z59ziRvuP2P+wZCAWwvDDbAthxnkJXLMpBnYDgBtYWOTwKfF4Ma7ZKCW4zwGEjllkr1tycYGZ9LYLBLwaJHvzz3M+LOtWk5+Rvo2iR9/7OQMjh9jvvEBn8OggMfgADKXsAaQdQ3EqBoFo2AUjIIRCQAYI1IaFKd5VgAAAABJRU5ErkJggg==","orcid":"","institution":"Office National d'Études et de Recherches Aérospatiales","correspondingAuthor":true,"prefix":"","firstName":"Sabrina","middleName":"","lastName":"GHANES","suffix":""},{"id":329798212,"identity":"8b12938a-1793-4fd8-95e0-66f9ee282ff6","order_by":1,"name":"Mikael PERRUT","email":"","orcid":"","institution":"Office National d'Études et de Recherches Aérospatiales","correspondingAuthor":false,"prefix":"","firstName":"Mikael","middleName":"","lastName":"PERRUT","suffix":""},{"id":329798213,"identity":"4b158825-eaa6-4c67-86fc-4429703ac684","order_by":2,"name":"Enrica EPIFANO","email":"","orcid":"","institution":"CNRS, CIRIMAT, Université de Toulouse","correspondingAuthor":false,"prefix":"","firstName":"Enrica","middleName":"","lastName":"EPIFANO","suffix":""},{"id":329798214,"identity":"b9a49c3e-48f9-43d6-bcfb-8f2d7a4ad0a3","order_by":3,"name":"Matthieu DEGEITER","email":"","orcid":"","institution":"Office National d'Études et de Recherches Aérospatiales","correspondingAuthor":false,"prefix":"","firstName":"Matthieu","middleName":"","lastName":"DEGEITER","suffix":""},{"id":329798216,"identity":"e2c78770-28be-4ea5-b1cc-b4c86e11f3d6","order_by":4,"name":"Thomas VAUBOIS","email":"","orcid":"","institution":"SAFRAN Tech","correspondingAuthor":false,"prefix":"","firstName":"Thomas","middleName":"","lastName":"VAUBOIS","suffix":""},{"id":329798218,"identity":"3b1eac44-1aa1-47a9-907a-5ff84d28ff17","order_by":5,"name":"Yohan COSQUER","email":"","orcid":"","institution":"DGA Techniques aérospatiales","correspondingAuthor":false,"prefix":"","firstName":"Yohan","middleName":"","lastName":"COSQUER","suffix":""},{"id":329798219,"identity":"f8041e3e-2fc8-4e5e-b9f6-3a35d3a7c248","order_by":6,"name":"Daniel MONCEAU","email":"","orcid":"","institution":"CNRS, CIRIMAT, Université de Toulouse","correspondingAuthor":false,"prefix":"","firstName":"Daniel","middleName":"","lastName":"MONCEAU","suffix":""}],"badges":[],"createdAt":"2024-07-15 12:17:03","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4742772/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4742772/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s11085-024-10284-5","type":"published","date":"2024-08-28T15:58:02+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":62395331,"identity":"1eb3fbde-97f9-4bc5-87dd-f9a1ba3dfb3f","added_by":"auto","created_at":"2024-08-13 16:56:43","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":122736,"visible":true,"origin":"","legend":"\u003cp\u003eOxide map of the Ni-Cr-Al model alloy in 0.1 atm of oxygen at 1200°C for 20 h or more [10]. The red line corresponds to the concentration gradient observed in the diffusion couple of the present work.\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-4742772/v1/0f1755dbc52648e5611bc834.png"},{"id":62395335,"identity":"b30f44e0-bb1b-4828-8e3c-8d808a5fc380","added_by":"auto","created_at":"2024-08-13 16:56:43","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":320065,"visible":true,"origin":"","legend":"\u003cp\u003eMicrograph (SEM/BSE) of a) Ni-5Cr-3Al b) Ni-20Cr-3Al c) Ni-10Cr-3Al oxidized at 1200 °C during 20h in air.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-4742772/v1/ef90943c9cacdf92b9c633b7.png"},{"id":62395333,"identity":"f358b229-563e-4ebd-a2ac-3b41eb138f21","added_by":"auto","created_at":"2024-08-13 16:56:43","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":195178,"visible":true,"origin":"","legend":"\u003cp\u003ea) SEM secondary electron (SE) image of C01 post HIP treatment and interdiffusion b) Comparison of simulated (DICTRA with TCNI8/MOBNI5 databases) and experimental diffusion profiles for Cr and Al, after 100h interdiffusion at 1300°C.\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-4742772/v1/f6e9b150ac4898f7b3191fd4.png"},{"id":62395336,"identity":"5b77b63d-6515-4f24-b3e8-691d0d15928d","added_by":"auto","created_at":"2024-08-13 16:56:43","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":344722,"visible":true,"origin":"","legend":"\u003cp\u003ea) SEM-SE image b) EDS-based Ni, Al and Cr weight fractions map of C01 surface, oxidized 20h at 1200°C in air\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-4742772/v1/f1a00e41891a54ede38983c1.png"},{"id":62395338,"identity":"5cb2a009-249d-4223-ba96-f77737b1526d","added_by":"auto","created_at":"2024-08-13 16:56:43","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":456787,"visible":true,"origin":"","legend":"\u003cp\u003ea) Location of the Raman spectra on the surface of C01 oxidized 20 h at 1200°C in air b) Raman spectra\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-4742772/v1/9e9b0b9e1c22ac408627367e.png"},{"id":62395334,"identity":"b8268cc3-e65f-4c03-bae0-0ae31f96a1c6","added_by":"auto","created_at":"2024-08-13 16:56:43","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":396309,"visible":true,"origin":"","legend":"\u003cp\u003ea) Normalized Raman spectroscopy map spectra b) Cumulated variance of data as a function of the number of Principal Components (PC) c) Data projection on the two-dimensional linear subspace spanned by the first two principal components PC1 and PC2 d) Oxide map reconstruction from PCA + Kmeans post-treatment.\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-4742772/v1/be0ae2dfe970749233577d75.png"},{"id":62395339,"identity":"c5b740e6-5a01-46f4-a82b-7ffe19c744e9","added_by":"auto","created_at":"2024-08-13 16:56:43","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":366673,"visible":true,"origin":"","legend":"\u003cp\u003ea) Optical micrograph of the C03 oxidized CGM at 1200°C for 20 h in air after copper-nickel coating b) EDS maps of Al, Cr, Ni and O c) Raman phase map of oxidized CGM\u003c/p\u003e","description":"","filename":"image7.png","url":"https://assets-eu.researchsquare.com/files/rs-4742772/v1/de39a1e1b1ffec554ffa1537.png"},{"id":63821067,"identity":"27d09ec9-f7f9-4af0-9b33-5796ed5b0b39","added_by":"auto","created_at":"2024-09-02 16:11:26","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2998199,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4742772/v1/c34ff249-c4e8-414c-a43a-bfc1778d7fbc.pdf"},{"id":62395685,"identity":"50f4f1ef-466c-465d-89ae-fe4050c98111","added_by":"auto","created_at":"2024-08-13 17:04:43","extension":"png","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":129816,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterial.png","url":"https://assets-eu.researchsquare.com/files/rs-4742772/v1/041bb2e62b8bb91cdce16db2.png"}],"financialInterests":"No competing interests reported.","formattedTitle":"Design of Oxidation Resistant Alloys using Combinatorial Approaches with Chemically Graded Materials","fulltext":[{"header":"Introduction ","content":"\u003cp\u003eIn materials science and engineering, the development of new materials with improved properties remains the key to technological progress for industry. Integrated Computational Materials Engineering (ICME) has emerged as the new approach to accelerate the development and optimization of materials. ICME integrates computer modelling tools and experimentations into all stages of the materials development process, from design to elaboration, processing and application. This multi-scale approach enables the understanding and prediction of the behavior of materials under real conditions by considering the interactions among structure, processing and performance [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Initiatives such as the Materials Genome aim to develop the use of artificial intelligence and modelling techniques to create databases and algorithms that are able to predict the performance of materials based on their atomic structure. Computational methods such as CALPHAD (CALculation of PHAse Diagrams) have been widely used to predict domains of interest. However, the effectiveness of these methods is highly dependent on the quality of the databases available. Current databases are often incomplete as the acquisition of data can be difficult and time-consuming. For this reason, it is necessary to develop high-throughput manufacturing and characterization methods to feed the databases. Xiang [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e] and others have implemented strategies such as combinatorial metallurgy, which involves rapidly investigating and optimizing metallic material properties through the generation and evaluation of alloy libraries [\u003cspan additionalcitationids=\"CR4\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. The first works used vapor deposition techniques such as co-evaporation and co-deposition to obtain thin films with composition gradients [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. This approach enabled the utilization of masks to selectively target designated regions within the composition space or to enhance the uniformity of the applied gradient [\u003cspan additionalcitationids=\"CR4\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. However, extrapolating properties of solid materials from thin films is difficult due to size effects, residual stresses and potential compositional differences. For this reason, methods such as diffusion couples/multiples have been developed to study bulk material microstructures and properties directly [\u003cspan additionalcitationids=\"CR7 CR8\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIn the field of aeronautics, there is a need for improved creep and oxidation resistant nickel-based alloys for turbine blades applications. However, the trial-error method used until now to develop alloys is expensive and time-consuming thus making it incompatible with the acceleration of alloy design. The aim of this study is to develop a high-throughput manufacturing and characterization method adapted to nickel-based chemically graded material (CGM) in order to feed databases for the application of oxidation resistant alloys. For this purpose, the Ni-Cr-Al ternary system was used as it is the basis for the development of nickel-based superalloys. According to Giggins and Pettit [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], three types of oxidation behaviors can occur for this model system. The type I consists in the formation of an outer layer of nickel oxide over internal alumina and chromia precipitates. The type II exhibits the formation of an outer layer of chromia and internal alumina precipitates. Finally, the type III consists in the formation of a continuous external alumina layer. These three types of oxidation behaviors are represented on Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e as an oxidation map at 1200\u0026deg;C [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. The aim of this study is to characterize the three types of oxidation behavior of the Ni-Cr-Al system, with only one sample.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003eTo observe the composition limits for the formation of oxides predicted by the oxide map [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e while working in the single-phase γ domain, the composition was set at 3 wt% Al with a chromium gradient. First, five Ni-wCr-3Al model alloys (w\u0026thinsp;=\u0026thinsp;0, 5, 10, 20 and 30 wt%) were produced through arc melting as reference for comparison with the CGM. The five compositions were selected to study the limits of the occurrence domains of oxides expected to form in the Ni-Cr-Al system (cf Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The five ingots were cut into little plates of 6 mm x 6 mm x 2.5 mm.\u003c/p\u003e \u003cp\u003eThen, a chemically graded material was developed for the Ni-Cr-Al model system to study the oxide frontiers at 3 wt% Al with a chromium gradient from 0 to 30 wt% Cr. To create this CGM, two alloys were assembled, i.e. Ni-0Cr-3Al and Ni-30Cr-3Al, and annealed for 100 h at 1300\u0026deg;C under argon atmosphere. The alloys used for the diffusion couple had the same dimension as the reference samples. To prevent the chromium volatilization during the interdiffusion, a piece of Ni-30Cr-3Al was added in the alumina cylinder to saturate the atmosphere in Cr-rich gazeous species. The interdiffusion duration and temperature were selected on the basis of thermo-kinetic calculations, performed using Thermo-Calc and DICTRA software with TCNI8/MOBNI5 databases, in order to obtain a composition gradient spreading across 1 mm. This length was chosen to facilitate the CGM characterization using techniques with spatial resolutions ranging between 1 and 50 micrometers. For the diffusion simulations, closed system boundary conditions were used. The initial aluminum profile was set at a constant value of 3 wt%, the chromium profile was a step function from 30 wt% to 0 wt% and the nickel was the dependent component. A double-geometric grid of 100 points was used with a lower ratio of 0.9 and an upper ratio of 1.1.\u003c/p\u003e \u003cp\u003eFinally, after the interdiffusion treatment at 1300\u0026deg;C, a HIP post-treatment was performed at 1150\u0026deg;C under 1300 bar during 4h in argon atmosphere to close any pores. Two CGM, C01 (underwent a HIP treatment) and C03 (no HIP), were produced. C01 was used for surface characterization and C03 for cross-sectional characterization.\u003c/p\u003e \u003cp\u003eThe reference samples and the CGM were oxidized in laboratory air at 1000\u0026deg;C and 1200\u0026deg;C, for 20 hours, to compare with Giggins and Pettit results. All oxidation treatments were conducted in a horizontal furnace calibrated before each treatment. Cross sections of samples were prepared using conventional metallographic methods including copper-nickel plating, grinding with SiC papers and polishing with diamond suspension (1/4 \u0026micro;m). Surface and cross section analyses were performed using optical microscopy, SEM/EDS on Zeiss MERLIN with SamX detector (the concentration profiles after assembling were quantitatively measured in comparison to pure Al, Cr, Ni standards, using IDFIX software [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]) and Raman spectroscopy with the Renishaw inVia Reflex confocal micro-Raman spectrometer with a laser wavelength of 532 nm.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e1. Oxidation behavior of reference samples\u003c/h2\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe cross sections of the reference samples oxidized at 1200\u0026deg;C for 20 h are illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Several oxides were identified on the cross section of the oxidized Ni-5Cr-3Al alloy (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003cb\u003ea\u003c/b\u003e) by means of EDS analyses. The reaction between NiO and Al\u003csub\u003e2\u003c/sub\u003eO\u003csub\u003e3\u003c/sub\u003e results in the formation of mixed oxide rich in aluminum and nickel with a small amount of chromium likely to be a spinel. XRD analysis confirmed the presence of spinel (not shown here). This oxidation behavior corresponds to the type I described by Giggins and Pettit. On the Ni-20Cr-3Al in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003cb\u003eb\u003c/b\u003e, a chromia layer at the surface and internal alumina precipitates have been detected. In some parts of the sample (as seen on Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003cb\u003eb\u003c/b\u003e) the internal precipitates have coalesced to form a continuous alumina layer beneath the chromia layer. This oxidation behavior is characteristic of a type II. In Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003cb\u003ec\u003c/b\u003e, Ni-10Cr-3Al formed a thin and continuous protective alumina layer with a limited amount of mixed oxide rich in Ni, Cr and Al (a spinel as confirmed by XRD) on the top in some areas. Since the alumina is continuous, the alloy is considered as type III. The same characterizations were performed on the reference samples oxidized at 1000\u0026deg;C and the same three oxidation behaviors were identified. For practical reasons, only the results obtained at 1200\u0026deg;C are presented in this article, but the conclusions remain the same.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2. Characterization of chemically graded samples\u003c/h2\u003e \u003cp\u003eThe thickness of the interdiffusion layer in the diffusion couple is a critical factor for the high-throughput method. In order to ease the characterization of the CGM and of its oxidation, a length of several millimeters would be the best choice. However, the diffusion length given by equations (1) and (2) is limited to restrictions on time and temperature. A compromise was necessary and the diffusion length of the chromium gradient was set at 1 mm.\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:{L}_{d}=4\\sqrt{D\\cdot\\:t}\\:\\:\\:\\:\\:\\:\\:\\left(1\\right)$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:D={D}_{0}{exp}\\left(\\frac{-Q}{RT}\\right)\\:\\:\\left(2\\right)$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003eL\u003csub\u003ed\u003c/sub\u003e: Diffusion length (m)\u003c/h2\u003e \u003cdiv id=\"Sec7\" class=\"Section4\"\u003e \u003ch2\u003eD: interdiffusion coefficient (m\u0026sup2;/s)\u003c/h2\u003e \u003cp\u003eOnce the length of the gradient is set, the second requirement is to limit the processing time, so increasing the temperature is a key. Nevertheless, temperature is limited by two conditions. The first one is the alloy lowest solidus temperature, which is estimated to be around 1360\u0026deg;C. The second condition is imposed by the oxidation experiment, the aim is to prevent an evolution of CGM in the metal during the oxidation experiment. Considering these conditions, the interdiffusion treatment was set at 1300\u0026deg;C. Finally, interdiffusion simulations were performed to assess the duration required to obtain a gradient of about 1 mm at 1300\u0026deg;C. For that, 100 h were necessary.\u003c/p\u003e \u003cp\u003eTo evaluate experimentally the diffusion length obtained in these conditions, the cross section of the couple (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003cb\u003ea)\u003c/b\u003e was analyzed by EDS. The results are compared to the simulation predictions in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003cb\u003eb\u003c/b\u003e, using DICTRA. An excellent agreement between the experimental and the simulated profiles for both Cr and Al can be observed.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e3. Characterization of oxidized chemically graded sample\u003c/h2\u003e \u003cp\u003e3.1. Surface characterization\u003c/p\u003e \u003cp\u003eSecondary electron mapping revealing the sample topology was performed on the C01 oxidized sample (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e.\u003cb\u003ea\u003c/b\u003e). It appears that the thickest oxide formed on the part of the sample with the lowest chromium content. The oxide in the chromium-rich zone is thinner than the oxide in the chromium-poor zone, but thicker than the oxide in the transition zone. It should be specified that the thickness of the thickest oxide was even larger than shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e because a part of the oxide spalled off after the oxidation treatment. The correlation between thickness and the type of oxide can be established using Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e.\u003cb\u003eb\u003c/b\u003e. The thickest oxide corresponds to the nickel rich oxide. Thanks to the reference samples, it may be assumed that it corresponds to NiO. The thinnest oxide is an aluminum-rich oxide, which should be α-alumina. The oxide formed in the chromium-rich part of the CGM is rich in chromium, thinner than nickel oxide but thicker than alumina, and should correspond to chromia. To verify theses hypothesis, a Raman spectroscopy profile was realized on the surface of the oxidized CGM and presented in the Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eRaman spectroscopy allowed the identification of phases through their characteristic vibrational spectra, unlike EDS which only identified chemical elements. Each spectrum was compared to reference spectra of expected oxides among nickel oxide, α-alumina and chromia. According to Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003cb\u003eb\u003c/b\u003e, nickel oxide was detected in the chromium-poor area, α-alumina in the transition area and chromia in the chromium-rich part of the CGM.\u003c/p\u003e \u003cp\u003eGiven their different thicknesses, focusing all the three oxides simultaneously was not possible, resulting in the one detected in the chromium-poor area being blurred. Several tests were carried out to evaluate the importance of focusing. For nickel oxide and chromia the signal was so intense that it was not necessary to set the focus to detect the peaks. In contrast, the characteristic peaks of alumina were weak, so setting the focus was necessary to optimize the signal. To overcome the focus problem, which requires changing the setting for each point, Raman spectroscopy mapping was performed by focusing on the transition zone (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e) where 781 spectra were acquired.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eEach spectrum was compared automatically with reference spectra, which were carried out on the reference samples, to determine the oxide type. However, peak shifts in a significant number of spectra made the oxide identification challenging, as the direct comparison with the reference spectra often failed. Given the independent processing of each spectrum is time-consuming, a high-throughput post-treatment method was set up.\u003c/p\u003e \u003cp\u003eFor this purpose, a data-driven analysis suited to the classification of Raman spectra by oxide type was developed. First, data was pre-treated to scale and smooth the spectra after removal of their baseline. A principal component analysis (PCA) was carried out, which amounts to expressing the full spectra values in the basis where the data covariance matrix is diagonal, with its eigenvalues sorted in descending order [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. We chose this linear dimension reduction technique for its simplicity, its ease of implementation and for its strong advantage of being fully explainable. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e.\u003cb\u003eb.\u003c/b\u003e highlights the cumulative variance of our data, explained with an increasing number of principal component (PC). 94% of the data variance is carried by the first two PCs, while keeping five PCs guarantees a 99% reconstruction of the data variance. Therefore, Raman spectra were projected from the 781-dimensional space they are observed in, to a 2-dimensional linear subspace spanned by the first two PCs, while preserving 94% of the explained variance (cf Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e.\u003cb\u003ec\u003c/b\u003e). Each point on this figure is the projection of a spectrum, and they tend to cluster into either narrow or widespread distributions. Underlying our analysis is the idea of measuring distances between Raman spectra: spectra similar to one another will have their projections clustered in neighboring regions of the linear subspace. Matching each cluster with the corresponding oxide-type \u0026ndash; chromia, α-alumina, or nickel oxide \u0026ndash; requires a clustering analysis for which we present a first approach using the K-means algorithm among Machine Learning techniques.\u003c/p\u003e \u003cp\u003eThe K-means unsupervised clustering algorithm was used to divide the dataset into multiple clusters [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. The cluster number K is fixed during the initialization step of the algorithm, and we chose K\u0026thinsp;=\u0026thinsp;3 because, according to the reference samples, we expect three types of oxides. The data is then sorted according to the nearest clustering center. Once the clusters were identified, each point was assigned a label and a color. Mapping back the labeled spectra to our oxidized chemically graded sample, we obtained the oxide map reconstruction presented at Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e.\u003cb\u003ed\u003c/b\u003e, where an oxide type is assigned to each pixel. The Raman mapping results validated the hypothesis established by the EDS characterizations. The three oxide types which were expected to form from the literature and reference samples were found i.e. nickel oxide in the chromium-poor zone, alumina in the transition zone and chromia in the chromium-rich part.\u003c/p\u003e \u003cp\u003e3.2. Cross section characterization of the oxidized chemically graded alloy\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eA cross section of the C03 oxidized sample is represented in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e.\u003cb\u003ea\u003c/b\u003e. During polishing the copper-nickel plating layer was detached from the oxide layer, which explains the black hole observed in the optical micrograph. EDS maps of Al, Cr, Ni and O were carried out to identify the oxide and inner precipitates composition. According to Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e.\u003cb\u003eb\u003c/b\u003e the thinner oxide layer and the inner precipitates are aluminum and oxygen rich. The oxide layer in the chromium-rich part of the CGM is chromium and oxygen rich, and the oxide island is rich in nickel and oxygen. Note that a part of the thicker oxide has been removed right after the oxidation treatment because of low adhesion. In certain areas, the nickel oxide layer has been preserved, enabling the measurement of its thickness. As the surface observation has already highlighted, the nickel oxide, which has grown in the chromium poor part of the CGM, is much thicker than the other oxides. As a comparison, the thickness of the nickel oxide layer is about 113\u0026thinsp;\u0026plusmn;\u0026thinsp;3 \u0026micro;m, while the chromia and alumina layers are about 13\u0026thinsp;\u0026plusmn;\u0026thinsp;3 \u0026micro;m and 3\u0026thinsp;\u0026plusmn;\u0026thinsp;1 \u0026micro;m, respectively. Moreover, there is a large area with internal oxide precipitates of 180\u0026thinsp;\u0026plusmn;\u0026thinsp;8 \u0026micro;m under NiO. The pores present in the chromium rich part of the CGM in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e.\u003cb\u003ea.\u003c/b\u003e were formed during the interdiffusion treatment, prior to oxidation, and they are due to Kirkendall effect [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. To complete the EDS analysis, a Raman map has been performed and illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e.\u003cb\u003ec.\u003c/b\u003e The comparison of our reference spectra with spectra acquired by mapping confirms that the thin layer and inner precipitates are alumina, the thicker layer is chromia and the oxide island is nickel oxide. Based on EDS and Raman map, the compositions required to form a type I is situated in [0Cr-3Al; 5Cr-3Al]\u0026thinsp;\u0026plusmn;\u0026thinsp;1%w composition range, a type III in [5Cr-3Al; 12Cr-3Al]\u0026thinsp;\u0026plusmn;\u0026thinsp;1%w and a type II in [12Cr-3Al, 30Cr-3Al]\u0026thinsp;\u0026plusmn;\u0026thinsp;1%w.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThanks to the excellent agreement between the diffusion simulations and experiments, the CGM was successfully obtained with the desired chromium gradient, spreading over of 1 mm.\u003c/p\u003e \u003cp\u003eThe same three types of oxide observed on the references were identified on the oxidized chemically graded alloy. It means that three types of behavior with very different oxidation kinetics can be observed on the same sample. From the measured oxide thicknesses, the parabolic rate constants values were calculated for the three type of oxides and presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eParabolic rate constants estimated from local oxide scale thicknesses for the oxidation of CGM at 1200\u0026deg;C in air during 20h and parabolic rate constants for the oxidation of a typical Group (I, II and III) alloy at 1200\u0026deg;C in 0.1 atm of oxygen during 20h [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\"\u0026minus;\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ek\u003csub\u003ep\u003c/sub\u003e (mg\u0026sup2;.cm\u003csup\u003e-4\u003c/sup\u003e.s\u003csup\u003e-1\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ek\u003csub\u003ep (Giggins \u0026amp; Pettit [10])\u003c/sub\u003e (mg\u0026sup2;.cm\u003csup\u003e-4\u003c/sup\u003e.s\u003csup\u003e-1\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eType I\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026minus;\" colname=\"c2\"\u003e \u003cp\u003e4.10\u003csup\u003e-3\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eType II\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026minus;\" colname=\"c2\"\u003e \u003cp\u003e8.10\u003csup\u003e-5\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10\u003csup\u003e\u0026minus;\u0026thinsp;4\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eType III\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026minus;\" colname=\"c2\"\u003e \u003cp\u003e4.10\u003csup\u003e-6\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10\u003csup\u003e\u0026minus;\u0026thinsp;5\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eComparing k\u003csub\u003ep\u003c/sub\u003e of the three types of oxide with those reported by Giggins \u0026amp; Pettit [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], it is observed that the parabolic rate constants for alumina and chromia are slightly lower than the reference value. As part of the nickel oxide layer has peeled off, the thickness would be expected to be higher resulting in a higher k\u003csub\u003ep\u003c/sub\u003e. Nonetheless, all three k\u003csub\u003ep\u003c/sub\u003e values remain relatively close to the reference. Thus, thickness difference helps to identify the transition areas between Type I, II and III. The compositions required to form a type I is situated in [0Cr-3Al ; 5Cr-3Al]\u0026thinsp;\u0026plusmn;\u0026thinsp;1%w composition range, a type III in [5Cr-3Al ; 12Cr-3Al]\u0026thinsp;\u0026plusmn;\u0026thinsp;1%w and a type II in [12Cr-3Al, 30Cr-Al]\u0026thinsp;\u0026plusmn;\u0026thinsp;1%w; which is very close to the limits announced in the literature [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]: Type I \u0026isin; [0Cr-3Al; 5Cr-3Al] Type III \u0026isin; [5Cr-3Al ; 15Cr-3Al ] and Type II \u0026isin; [15Cr-3Al, 30Cr-Al and \u0026gt;\u0026thinsp;30Cr-Al]. These results allowed to measure the critical contents of chromium needed to promote external growth of a protective alumina layer rather than aluminum internal oxidation. For turbine blades, the aim is always to form an outer layer of the most protective oxide, i.e. α-alumina. These results suggest that the critical content of chromium required for the transition from internal to external oxidation of aluminum in Ni-wCr-3Al is \u0026asymp;\u0026thinsp;5\u0026thinsp;\u0026plusmn;\u0026thinsp;1 wt% Cr. It demonstrates in a new and direct way the third element effect [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e], i.e. adding chromium to Ni-Al alloys promotes the formation of external layer of alumina at lower aluminum content.\u003c/p\u003e \u003cp\u003eIn this work, EDS and Raman spectroscopy were used as characterization methods to identify the oxide type. However, the current post treatment was not always adapted to the CGM characterization. Thus, this article proposes a high-throughput characterization linking Raman spectroscopy and unsupervised machine learning to accelerate post processing. The first results presented here are encouraging, identifying the oxide type without peak indexing problems. The clustering part of the analysis is a work still in progress, for which we have already identified several improvement directions. We note that the K-means algorithm used as a first clustering method has its limitations, and is known to perform poorly when the clusters have differing variances and densities, as well as when they display anisotropy. In a high-throughput characterization approach, this remains acceptable, given that it is possible to identify the oxide formation regions.\u003c/p\u003e \u003cp\u003eRaman enables the direct identification of the oxide phases, contrary to EDS which only yields their chemical composition. Moreover, the chemically graded material can be precisely characterized by Raman spectroscopy with a resolution of 0.6 \u0026micro;m, better than with EDS (few \u0026micro;m). However, in the cross-sectional study, characterizing the surface oxide was difficult because of a weak signal, specially of alumina. Additionally, it was not possible to characterize the substrate because metals generally have no Raman active modes or present modes with very weak intensity. For these reasons, both characterization techniques are complementary.\u003c/p\u003e"},{"header":"Conclusions and Perspectives","content":"\u003cp\u003eThis work shows that it is possible to obtain similar results as Giggins and Pettit, using far fewer samples thanks to CGM. There were doubts about forming a continuous external alumina layer with only 3 wt% Al. However, these experimental results confirm, with only one sample, how this is possible, thanks to the \u0026lsquo;third element effect\u0026rsquo; [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Indeed, the addition of Cr allows alumina to be formed at lower Al contents than required for a binary Ni-Al alloy. The excellent agreement between simulated and experimental diffusion profiles validated the experimental method used to design CGM. The next step is to extend this method to more complex systems, in order to increase the data collected in only one experiment. For this purpose, diffusion multiplet will be realized to obtain a 2D gradient in one sample.\u003c/p\u003e \u003cp\u003eIn addition, this study introduced a high-throughput characterization method combining EDS and Raman maps. The CGM was characterized precisely combining surface and cross-sectional analysis. EDS provided a preliminary analysis that was confirmed by Raman spectroscopy, which directly identifies the characteristic spectrum of each oxide. However, the current method of indexing Raman peaks required a certain amount of time to work on a case-by-case basis, whenever there were shifts or broadening of the peaks. The post-treatment characterization proposed in this article overcomes this problem, offering an identification method suitable for any oxide. By using the PCA method, the high number of dimensions of the initial problem was reduced to two dimensions, thereby reducing calculation time. The PCA projection of the data led to a new 2-dimensional representation of the data in which clusters were formed. Each cluster represented an oxide type in the subspace. The K-means unsupervised clustering algorithm allowed the identification of the three clusters formed by nickel oxide, chromia, and alumina spectra. The Raman map was then reconstructed and compared with the EDS results. Both EDS and Raman results were in line with reference samples and literature, proving the validity of this method. This method can be reliably correlated to the oxide type, but sometimes struggles to identify certain points that are far from the cluster centers. For this reason, a forthcoming work will be dedicated to improving the clustering algorithm and compare this method with a deep learning approach.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eDeclaration of Competing Interest\u003c/h2\u003e \u003cp\u003eThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eS.G did the experiment and write the first draft of the paper including figs and tables.M.P coordinates the project in which this work takes place.S.G, M.P, E.E, and D.M wrote the main manuscript text. S.G and M.D wrote the PCA and K-means part of this article.T.V and Y.C helped during the conception and the writing of the article. All authors reviewed the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eFinancial supports by the French Armed forces Ministry and the French Aerospace Lab are gratefully acknowledged. The authors are grateful to C. Rio, N. Horezan, and Q. Barr\u0026egrave;s for their valuable help with SEM/EDS mapping, J.S. M\u0026eacute;rot for technical support and his involvement and A. Andrieux for providing access to the Raman spectrometer.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAlberi K et al. The 2019 materials by design roadmap. J. Phys. D: Appl. Phys. 52, 013001, 2019.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcDowell DL et al. Chapter 3 - Overview of the Framework for Integrated Design of Materials, Products, and Design Processes in Integrated Design of Multiscale, Multifunctional Materials and Products, \u003cem\u003e39\u0026ndash;64\u003c/em\u003e, 2010.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXiang XD et al. A Combinatorial Approach to Materials Discovery. Science 268, 1738\u0026ndash;1740, 1995.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTakahashi R et al. Design of Combinatorial Shadow Masks for Complete Ternary-Phase Diagramming of Solid State Materials. J. Comb. Chem. 6, 50\u0026ndash;53, 2004.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePotyrailo R et al. Combinatorial and High-Throughput Screening of Materials Libraries: Review of State of the Art. ACS Comb. Sci. 13, 579\u0026ndash;633, 2011.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao JC. Phase Diagram Determination Using Diffusion Multiples. in \u003cem\u003eMethods for Phase Diagram Determination\u003c/em\u003e 246\u0026ndash;272, 2007).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao JC. Combinatorial approaches as effective tools in the study of phase diagrams and composition\u0026ndash;structure\u0026ndash;property relationships. Progress in Materials Science 51, 557\u0026ndash;631, 2006.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePerrut M. Life Prediction Methodologies for Materials and Structures, Journal Aerospace Lab, 2015.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLacour-Gogny-Goubert A et al. Effect of Mo, Ta, V and Zr on a duplex bcc\u0026thinsp;+\u0026thinsp;orthorhombic refractory complex concentrated alloy using diffusion couples. Intermetallics 124, 106836, 2020.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGiggins CS \u0026amp; Pettit FS. Oxidation of Ni-Cr-AI Alloys Between 1000\u0026deg;C and 1200\u0026deg;C. J.Electrochem.Soc 118, 9, 1971.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGreenacre M et al. Principal component analysis. Nat Rev Methods Primers 2, 1\u0026ndash;21, 2022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePedregosa F et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research 12, 2825\u0026ndash;2830, 2011.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNesbitt JA \u0026amp; Heckel RW. Interdiffusion in Ni-Rich, Ni-Cr-Al alloys at 1100 and 1200\u0026deg;C: Part I. Diffusion paths and microstructures. Metall Trans A 18, 2061\u0026ndash;2073, 1987.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNiu Y, Zhang XJ, Wu Y \u0026amp; Gesmundo F. The third-element effect in the oxidation of Ni\u0026ndash;xCr\u0026ndash;7Al (x\u0026thinsp;=\u0026thinsp;0, 5, 10, 15 at%) alloys in 1 atm O2 at 900\u0026ndash;1000\u0026deg;C. Corrosion Science 48, 4020\u0026ndash;4036, 2006.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"high-temperature-corrosion-of-materials","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [High Temperature Corrosion of Materials](https://www.springer.com/journal/11085)","snPcode":"11085","submissionUrl":"https://submission.nature.com/new-submission/11085/3","title":"High Temperature Corrosion of Materials","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"alloy by design, high-throughput method, diffusion, gradient, combinatorial metallurgy","lastPublishedDoi":"10.21203/rs.3.rs-4742772/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4742772/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis work introduces a new high-throughput method to characterize the oxidation behavior of chemically graded Ni-based alloys in order to feed databases destined to numerical metallurgy approaches. A Ni-wCr-3Al (w \u0026isin; [0, 30]) chemically graded material was obtained from two homogeneous samples by a diffusion couple method at 1300\u0026deg;C for 100h. The composition range was selected in order to observe the three types of oxidation behavior identified in the reference work of Giggins and Pettit [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. The excellent agreement between simulated and experimental diffusion profiles validated the experimental method used to manufacture the chemically graded material (CGM). The CGM was then oxidized at 1200\u0026deg;C in air. Surface and cross-section characterization were conducted by SEM/EDS and Raman spectroscopy to identify the oxides formed on the CGM. To accelerate the Raman characterization treatment, a method linking Principal Component Analysis (PCA) and K-means unsupervised clustering algorithm was developed [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. It allowed for the identification of the oxide type without peak indexation issues and is well-suited for CGM. These results show that results similar to well-recognized reference experiments [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] can be achieved using only one CGM.\u003c/p\u003e","manuscriptTitle":"Design of Oxidation Resistant Alloys using Combinatorial Approaches with Chemically Graded Materials","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-08-13 16:56:38","doi":"10.21203/rs.3.rs-4742772/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Accepted","date":"2024-07-21T17:50:06+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-07-21T09:29:32+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"22023397579724329343500945674751299733","date":"2024-07-21T09:28:27+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-07-20T21:44:36+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-07-20T21:42:54+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-07-18T14:04:23+00:00","index":"","fulltext":""},{"type":"submitted","content":"High Temperature Corrosion of Materials","date":"2024-07-15T12:15:44+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"high-temperature-corrosion-of-materials","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [High Temperature Corrosion of Materials](https://www.springer.com/journal/11085)","snPcode":"11085","submissionUrl":"https://submission.nature.com/new-submission/11085/3","title":"High Temperature Corrosion of Materials","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"0bc7721f-c88c-4fab-8e65-04ad6c132a98","owner":[],"postedDate":"August 13th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2024-09-02T16:04:10+00:00","versionOfRecord":{"articleIdentity":"rs-4742772","link":"https://doi.org/10.1007/s11085-024-10284-5","journal":{"identity":"high-temperature-corrosion-of-materials","isVorOnly":false,"title":"High Temperature Corrosion of Materials"},"publishedOn":"2024-08-28 15:58:02","publishedOnDateReadable":"August 28th, 2024"},"versionCreatedAt":"2024-08-13 16:56:38","video":"","vorDoi":"10.1007/s11085-024-10284-5","vorDoiUrl":"https://doi.org/10.1007/s11085-024-10284-5","workflowStages":[]},"version":"v1","identity":"rs-4742772","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4742772","identity":"rs-4742772","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}