1. Introduction
In 1995, Xiang et al. [1] published research on a “combinatorial materials chip” as a cover article in Science. On one substrate, simultaneous integrated growth and characterization were achieved with 128 new materials with different components, and also high-throughput material screening was achieved, along with a systematic description of the “phase diagram of the material.” Since then, similar high-throughput material syntheses and characterization technologies have been developed and applied in various fields. Examples include a “diffusion multiple approach” [2], used to accelerate the design of structural materials; highthroughput film-growth technology [3], used to screen semiconductor materials with discrete and continuous components; and an ink-jet delivery system, used to screen composite powders [4]. Other high-throughput experimental techniques include the rapid characterization of composition [5], structure [6], electrochemical properties [7], catalytic properties [8], electromagnetic properties [9], magnetic properties [10], optical properties [11], thermal properties [12], and mechanical properties [13]. The screening of unknown materials has been accelerated significantly, and work that took years with conventional methods can be achieved in one week with high-throughput characterization technology, resulting in revolutionary breakthroughs. Nevertheless, the “combinatorial materials chip” technology has shortcomings; chip preparation often differs significantly from the actual material production, which suggests that significant effort is still required to scale up production to enable the transfer from laboratory to application. Wang et al.[14,15] proposed a high-throughput statistical spatial-mapping characterization technology based on originalposition statistical-distribution analysis (OPA). Based on a recognition of the intrinsic material heterogeneity, the proposed technique carried out high-throughput and across-scale span statistical-distribution characterization of samples. Massive datasets were acquired promptly with an exact location correspondence between macroscopic, mesoscopic, and microscopic composition, structure, and properties. A spatial-mapping relationship between datasets was established so that a correlation was obtained between the composition, structure, and properties. Genetic units with superior performance were screened rapidly and selected to guide the modification of existing materials and for production optimization. Through machine learning of massive spatial-mapping datasets, these screened genetic units could be reconstructed reversely “bottom up” from the microscale to the macroscale, so that an optimum combination of composition, structure, and properties could be achieved at the macroscopic level. Thus, a reverse design of new materials could be accomplished and accelerated. Therefore, an intrinsically heterogeneous actual material is also considered a “natural chip” with various combinations. Because this kind of “natural chip” originates from an actual production process, it has special significance in guiding the modification of existing materials and the optimization of the production process.
2. High-throughput statistical spatial-mapping characterization technology
2.1. Intrinsic material heterogeneity
Macroscopic materials are heterogeneous, multi-elementary, and complex. No material is homogeneous or isotropic at a certain small scale. Parts of the material that differ from one another can be termed “natural chips.” At different spots of the material, the composition, structure, and properties vary slightly, and the combination of these slight differences establishes the overall material performance. Like human genes, a material has a smallest unit of matter, which is a genetic unit and reflects the material properties. The genetic units vary for different materials, and could be a natural atom, molecule, or ion that makes up a substance, phases, clusters, functional groups, units, or grains formed by a combination of these particles. Certain processes or technologies can be used to combine the same or different genetic units into “genomes” or even into final materials with certain properties. This combinatorial process or technology can be referred to as an “assembly.” Because of the complexity and diversity of the different materials and processes, the “assembly” of genetic units is diverse. Research into materials genome engineering includes high-throughput synthesis, characterization, screening, assembly, and repeated verification of target genetic units that are designed on demand. Thus, a correlation can be established between the microscale, mesoscale, macroscale, and across-scale span, as well as the composition, structure, and properties, to guide the research and development of new materials and the modification of existing materials with less time and cost. Wang et al. [16] used OPA characterization technology to study the distribution of Nb in the disk forging of a superalloy compressor. Nb was distributed unevenly in the colddie-affected area on the lower die and in the center of the forging, which led to a failure of the disk forging. Through results analysis, uneven parts were removed in the process, which raised the final product quality to standard. The physical properties of amorphous alloys in different directions are the same, because their atomic structures are amorphous and isotropic at the macroscopic scale. Many studies have shown that the static structure and dynamics of amorphous alloys at the nanoscale or microscale are heterogeneous[17–24]. Therefore, the material heterogeneity should be acknowledged. However, if genetic units of material could be identified more clearly based on their intrinsically heterogeneous nature, faster and better designs and the development of new materials with improved properties could be achieved.
2.2. The high-throughput statistical spatial-mapping characterization technology
High-throughput statistical spatial-mapping characterization is based on the intrinsic material heterogeneity. Through the acrossscale span characterization of a material, different compositions, structures, and properties are acquired from tens of thousands of material microarrays. A statistical spatial-mapping model between the sets of parameters is established based on the original material position. By using high-throughput computations, databases can be formed by screening genetic units, which determines the screened material properties and forms a database of material with the help of high-throughput computations. Materials-design optimization is carried out to guide material modification, process optimization, and the discovery of new materials, as shown in Fig. 1.
Fig. 1. Combined technology of microarray statistical spatial-mapping based on material heterogeneity.
The high-throughput statistical spatial-mapping characterization process is shown in Fig. 2, and uses a series of rapid characterization techniques to analyze the across-scale span from the macroscale to the microscale. Datasets of composition, structure, and properties at each location are acquired rapidly for practical sample sizes. Through accurate positional coordinate information, and with reference to a point-to-point correspondence, the database is set up to contain spatial-mapping lattices. The design requirements of the research and development of materials are used to select spatial-mapping datasets within the required target intervals from the database. Statistical analyses, including statistical frequency within the range of parameters, statistical correlation between parameters, statistical elimination of the outliers, reasonable criteria, and models, can be used to select the design that best meets the targeted requirements. After repeated verification, genetic units that reflect the material properties are determined. Established pointing parameters in the process optimization are used to verify the assembly of these genetic units on the mesoscale, and quantitative correlations are established between the microscale, mesoscale, macroscale, practical sample, across-scale span composition, structure, and properties.
Fig. 2. Process of characterization of high-throughput statistical mapping.
High-throughput statistical spatial-mapping characterization is like walking through a maze. By using the multi-path parallel trial method, the efficiency of discovering new materials can be improved significantly. The research and development single trial-and-error mode that was applied to existing materials can be discarded, and a new approach for the rapid research and development of new materials is presented. An innovation system for material research and development is constructed herein, which involves a “high-throughput trial-and-error method” that reduces the cost and shortens the research and development cycle for new materials.
2.3. Across-scale span original-position statistical-distribution analysis
OPA is a technology that performs a quantitative statisticaldistribution analysis of chemical composition and morphology at the macroscopic scale (cm2 ). This analysis uses the location distribution information of chemical composition, content distribution statistical information, and state-distribution statistical information. The information can be divided into one-, two-, and threedimensional OPA based on a regional division[25–32]. OPA technology has been applied to characterize composition, structure, and mechanical properties. Advanced technologies have been developed, such as Spark-OPA, laser-induced breakdown spectroscopy (LIBS)-OPA, laser-ablation (LA) inductively coupled plasma mass spectrometry (LA-OPA), micro-X-ray fluorescence (μXRF) OPA, full-view-metallography (FVM)-OPA, macroscopic scanning-electron microscopy (SEM)-OPA, and fluid microprobeOPA. A series of instruments and devices with independent intellectual property rights has also been developed.
2.3.1. Spark-OPA
As the main means to characterize the macroscopic statistical distribution of materials, Spark-OPA technology was developed from the optical emission spectroscopy (OES). Traditional OES pre-sparks the sample and integrates the signals in order to acquire reliable results, but this method cannot analyze the original inclusions information, since the inclusions have been remelted by the pre-spark. Spark-OPA derives a series of technologies including single-spark discharge, signal-resolution extraction, non-pre-spark continuous excitation, and synchronous-scanning positioning, by which millions of pieces of data on the original content and state information of each element are obtained to correspond to their original positions within the material. Statistical analysis is used to quantitatively characterize parameters such as segregation, looseness, and the inclusion distribution of materials. A single-spark discharge can be regarded as a kind of unconstrained probe with a size of 1–10 μm, and a spark point contains millions of these probes. Thus, Spark-OPA can provide accurate information on the position distribution, state distribution, and quantitative distribution of each component on a macroscopic scale (hundreds of square centimetres) [33]. Spark-OPA is an across-scale span-characterization technology that could reflect the microstate at the macroscale. The original-position metal analyzer OPA-100 was developed in 2002, with seven associated patents[34–40], and was awarded the second-place prize at the Chinese National Technological Invention in 2008 [41]. It has been upgraded to a fourth-generation product (as shown in Fig. S1 in Appendix A).
Spark-OPA technology has been developed into a mature commercial application for composition segregation and the quantitative characterization of inclusions in macroscopic metal samples. This technology has played an important role in guiding production optimization. Compositional segregation characterization has been applied to the characterization of carbon steel and stainless steel materials, such as continuously cast bloom, cord steel, and ship plate steel[42–81]. For example, Li et al. [82] studied the round billet of No. 35 carbon steel and found a bright white band at the edge of the billet that was produced by electromagnetic stirring; an obvious negative segregation of elemental C, Si, Mn, and P was the main reason for the uneven distribution of grain structure and Vickers hardness (HV) (Fig. 3). The composition segregation and quality defects both have been explored in detail for nonferrous metals such as aluminum alloys and brass, etc.[83–85].
Fig. 3. Application of Spark-OPA technology in the composition segregation of an electromagnetically stirred round billet [82]. (a) Analyzed region (106 mm × 52 mm); (b) two-dimensional elemental distribution in the analyzed region; (c) relationship between composition distribution and HV of C, Mn, and Si.
For the quantitative characterization of inclusions, Wang’s research team[86–89] established an analytical method for the content and size distribution of inclusions, such as Mn, Al, Ti, and Si in steel, through the statistical analysis of several abnormal sparks in a single discharge. Spark-OPA technology has been applied to the statistical distribution analysis of inclusions in various metal materials, such as carbon steel, stainless steel, heavy rail steel, beam steel, gear steel, and high-pressure boiler-tube steel [90–107]. For example, Luo et al. [108] used Spark-OPA technology to conduct a full-scale characterization of the cross-section of a continuously cast plate of stainless steel (Samples A–N) and found that, in the macroscopic test, the proportion of Al–Ca inclusions and that of small-particle inclusions were slightly higher at the edge of the white band. The proportion of Al–O inclusions and that of large-particle inclusions were slightly higher in the central region, which was consistent with the SEM analysis. An edge-tocenter distribution of aluminum inclusions was obtained in the sample (as shown in Fig. 4).
Fig. 4. Application of Spark-OPA in the characterization of inclusions in continuously cast steel plate [108]. (a) Analysis of low-magnification component distribution and statistical segregation of sample M; (b) diagram of abnormal spark channels for Al and Ca; (c) bar graph of Al particle-size distribution inclusions in regions 2 and 4; (d) bar graphs of particle-size distribution of Al inclusions in regions 2 and 4 by SEM and OPA; (e) distribution of Al inclusions in the plate; (f) distribution of elemental composition of Al in the plate.
2.3.2. LIBS-OPA
LIBS-OPA technology is based on a quantitative analysis of the spectrum signal of the atomic emission that is generated by the action of a high-energy laser beam on the material surface. The beam spot is micron-to-millimeter sized. This technology provides the advantages of non-contact analysis, micro-area analysis, and in-depth analysis, with point and line scanning. One-dimensional depth analysis and two-dimensional surface analysis enable a fine positioning of the sample surface and are an effective means of mesoscopic-to-macroscopic across-scale span materials characterization [109–111]. The first commercial LIBS-OPA100 was developed in 2010, with several associated patents [112–115]. It has since been upgraded to a second-generation product (Fig. S2).
LIBS-OPA technology has advantages in the characterization of composition segregation of small samples. Therefore, LIBS-OPA100 have been widely used to analyze the distribution of various components in small samples, such as medium- and low-alloy steel plate, cord wire rod, the surfacing fusion zone of X80 pipeline steel, and electromigration gadolinium rods [116–122]. Their work helped to pinpoint existing problems in production through the characterization of composition segregation. Fig. 5 shows a LIBS-OPA characterization of the composition distribution and a study of the relevance between the microstructure and the microscopic distribution of HV in the surfacing zone of X80 pipeline steel.
Fig. 5. LIBS-OPA characterization, microstructure, and microscopic distribution of HV in the surfacing zone of X80 pipeline steel [116]. (a) Scanning area of the surfacing zone of X80 pipeline steel (3.3 mm × 3.9 mm); (b) full-view microstructure and microscopic distribution of HV; (c) microscopic distribution of HV and composition in the surfacing zone.
Many studies have been carried out with LIBS-OPA technology on the analysis of sample defects. Various shape defects on the surface of auto and cold-rolled hot-galvanized sheets were tested by LIBS-OPA100 with modes of line scanning, surface scanning, and depth analysis, respectively [123–126]. The results showed that defects were accompanied by elemental segregation, most of which were caused by the introduction of mold powder into the production. Their research was significant in guiding improvements in coated-sheet production. Fig. 6 shows an in-depth and line-scanning study of the defects of an auto sheet.
Fig. 6. In-depth and line-scanning study of the defects of an auto sheet with LIBS-OPA technology [125]. (a) Line defect of a galvanized sheet; (b) depth analysis of the defect; (c) depth analysis of non-defect spots; (d) band defects of an auto sheet; (e) line scanning of surface defects; (f) line scanning of surface non-defect spots; (g) a scratch defect and artificial defects on an auto sheet; (h) line scanning of scratch defects; (i) line scanning of artificial defects.
In recent years, LIBS-OPA technology has progressed to inclusion analysis. Yang et al. [127–130] found that the number of abnormal signals in the laser spectrum reflected the number of inclusions, and that the intensity of the abnormal signals was related to the size of the inclusions. They used this information to analyze the contents of acid-insoluble aluminum, MnS inclusions, and Si–Al– Ca–Mg composite inclusions in steel. The results of their analysis agreed with those of traditional wet-chemical analysis. Fig. 7 depicts a study of the LIBS-OPA quantification of MnS inclusions.
Fig. 7. LIBS-OPA study on MnS inclusions [128]. (a) MnS inclusions; (b) partially burnt inclusions; (c) completely burnt inclusions; (d) intensity distribution of Mn; (e) intensity distribution of S; (f) intensity of S and Mn; (g) linear fitting relation of the intensity of S and the area of inclusions; (h) linear fitting relation of the intensity of Mn and the area of inclusions.
2.3.3. LA-OPA
The principle of LA-OPA technology is that the sample can be stripped and gasified layer by layer by a microbeam-focused laser, transferred to an inductively coupled plasma (ICP) source, be ionized or atomized in an inert gas environment as an aerosol, and be analyzed quantitatively by mass spectrometry. The laser-beam spot of this technology is on the order of microns, with a low detection limit and a high sensitivity, and is suitable for the statistical analysis of the surface distribution of low-content and trace components in hetero-morphed or small samples. This technique provides another effective means of mesoscopic-to-macroscopic across-scale span material characterization. In 2008, the LA-OPA technology has been developed and applied to the statisticaldistribution characterization of various hetero-morphed surfaces of small samples, such as spherical flat steel, galvanized steel pipes, welded pipes, superalloy turbine blades, dysprosium bars, pipeline steel cracks, and impact fractures [131–147]. They focused on the position distribution, statistical segregation, and maximum segregation of components; the relevance between these properties; and the material quality. In 2015, commercialization of the proprietary intellectual property rights was achieved for the LA-OPA characterization equipment (including a laser-ablation sampling device and an inductively coupled plasma mass spectrometer), as shown in Fig. S3. Two patents were applied for Refs. [148,149], and the Beijing Conference and Exhibition on Instrumental Analysis (BCEIA) Gold Prize established by the China Association for Instrumental Analysis was won in 2015. Fig. S4 shows the statistical-distribution characterization of directionally solidified superalloy turbine blades by LA-OPA technology; the precipitation of elements with low melting points in the polycrystalline zone resulted in blade defects [150].
2.3.4. μXRF-OPA
μXRF-OPA technology focuses X-rays into a small beam with a diameter of about 20 lm by using capillary lenses. The chemical compositions on the material surface can be detected by nondestructive surface scanning by μXRF-OPA, and statisticaldistribution analysis is conducted on the large acquired dataset. μXRF-OPA technology is a non-destructive testing method with improved resolution but a low fluorescence-intensity loss, which enables a scanning range up to centimeters and provides an efficient method for characterizing materials from the mesoscale to the macroscale. In 2017, a prototype with proprietary intellectual property rights was developed, as shown in Fig. S5. Yang et al. [151] conducted statistical-distribution characterization of the composition segregation in microregions of weatherproof steel sheets with μXRF-OPA technology. The results showed that the segregation of Ti, Mn, P, and S in the fractured zone may be the main cause of cracking, as shown in Fig. 8. Li et al. [152] conducted statistical-distribution characterization of superalloy compositions that were produced by different processes with μXRF-OPA technology. Heat treatment improved the distribution uniformity of elemental Nb, Ti, Mo, and W, and the maximum segregation decreased significantly, as shown in Fig. 9.
Fig. 8. Statistical-distribution characterization of composition segregation in microregions of weatherproof steel sheet [151].
Fig. 9. Statistical-distribution characterization of the elements Nb, Ti, Mo, and W in superalloys produced by different processes [152]. (a) Cast; (b) heat treatment; (c) comparison of statistical components distribution. RSD: relative standard deviation.
2.3.5. FVM-OPA
FVM-OPA technology, which is based on full-automatic scanning metalloscopy, rapidly acquires an atlas of the metallographic structure and position information of the entire sample surface, and then pieces data into a complete image with precise location information. Through a statistical analysis of the original digital signal (gray value) at each pixel in the image, automatic identification and quantitative statistical-distribution characterization are achieved for various structural property types (e.g., looseness, cracks, shrinkage cavities, defects, crystalline grains, precipitated phases, and inclusions). Through statistical analysis of the structural orientation in the entire sample range, FVM-OPA technology resolves the issues of subjectivity, randomness, and contingency when artificially selecting the field of view, which results in a more comprehensive characterization of the metallographic structure. Wang et al. [153] conducted statisticaldistribution analyses of martensite and ferrite in a ferrosilicon alloy by FVM-OPA technology. The gray value of the structure was quantitatively correlated with the C content, Si content, C/Si ratio, and HV (Fig. S6).
2.3.6. Macroscopic SEM-OPA
SEM-OPA technology uses an electron source with a high brightness and field-emission, high-resolution electromagnetic compound objective lens, and direct electron detector to achieve the high-throughput acquisition of images in macroscopic samples. The scanning time of an image of the same quality is 1/50th that of traditional SEM. Intelligent software integrates professional image libraries of a wide variety of specific materials, and automatically acquires and calibrates the category and characteristics of the observed structures. Graphic processor unit multi-threaded parallel computing and large data mining are used to conduct a more comprehensive statistical analysis of the overall distribution of the structural parameters of the macroscopic samples and to establish an improved statistical mapping correlation with the distribution of the composition and performance. Wang et al. [154] characterized a 12 mm diameter sample of nickel (Ni)-based single-crystal superalloy by SEM-OPA technology and obtained full-view surface-distribution information of the 
phase, as shown in Fig. 10. The results showed that small -phase grains were mainly distributed in dendritic stems while larger -phase grains occurred between dendrites, as shown in Fig. 11.
phase, as shown in Fig. 10. The results showed that small -phase grains were mainly distributed in dendritic stems while larger -phase grains occurred between dendrites, as shown in Fig. 11.
Fig. 10. Full-view and magnified -phase crystal superalloy [154].
-phase crystal superalloy [154].
Fig. 11. Distribution -phase grains of different sizes [154].
-phase grains of different sizes [154].2.3.7. Fluid-microprobe-OPA
Fluid-microprobe-OPA technology is based on the principle of isostatic pressure. Under the action of a high-pressure fluid (gas or liquid), because of the heterogeneous nature of the sample, the positions of different structures yield different deformations. By establishing a correlation between the micro deformations at each position and structure, a statistical distribution of the stress and strain at an original position is achieved. Fluid can be regarded as a continuously distributed and uniformly pressed microprobe. Therefore, fluid-microprobe-OPA technology performs, in a real sense, a continuous across-scale span, from nanometer to centimeter, and provides a high-throughput characterization of the mechanical properties. Feng et al. [155] studied the surface deformation, structural distribution, and HV of samples of high-chromium (Cr) white cast iron using fluid-microprobe-OPA technology. The deformation was related closely to the elastic modulus, equivalent modulus, and hardness, as shown in Fig. 12.
Fig. 12. Study of high-Cr white cast iron by fluid-microprobe-OPA technology [155]. (a) Metallographic structure; (b) microhardness distribution; (c) distribution of isostatic pressure-induced deformation; (d) correlation between elastic modulus and deformation; (e) correlation between equivalent modulus and deformation; (f) correlation between HV and deformation.
3. Demonstrative application of high-throughput statistical spatial-mapping characterization technology
3.1. Statistical spatial-mapping characterization of copper in the aging of G115 heat-resistant steel for ultra-supercritical power-generation units
Ultra-supercritical coal-fired power generation is an important measure to achieve energy conservation and emissions reduction. Research and development of heat-resistant materials is the greatest bottleneck restricting the development of advanced ultrasupercritical thermal-power units. G115 martensitic heatresistant steel is developed based on existing 9 wt%–12 wt% Cr heat-resistant steel. Its ultimate service temperature exceeds 650 °C, which is of great engineering significance. G115 steel strengthens precipitation by 1.0 wt% copper (Cu) addition in the alloy design. Because of difficulties in the characterization of the precipitated Cu phase in martensitic steel, the form of existence, distribution, and strengthening mechanism of Cu in G115 steel is still unclear. Yang [156] used lXRF-OPA technology to conduct mesoscopic-to-macroscopic across-scale span characterization of a full sample surface (8.1 mm × 8.1 mm). A two-dimensional distribution of the intensity shows that elements were distributed uniformly in a mesoscopic state without obvious segregation, which indicates that the microbeam-fluorescence resolution was insufficient to characterize the discrepancy between the Cu in the sample (Fig. 13). SEM-energy dispersive spectroscopy (EDS) was used to locate the region with the highest intensity in microbeam-fluorescence analysis. At a 1000× magnification, because of the relatively low spatial resolution, elements in the region of characterization (300 μm × 300 μm) were in a relatively evenly dispersed state (Fig. 14). A white boxed area with slightly enriched Cu in the image was selected for microscopic characterization at a 20 000× magnification. Genetic units that contained Cu could be screened rapidly in this region (15 μm × 15 μm). The surface distribution of the energy spectrum showed that Cu-enriched areas were distributed at the interface or grain boundary, whereas other elements were negatively segregated, which indicates that Cu existed separately at the interface or grain boundary in the Cu-rich particles and did not form phases with other elements (Fig. 15). To determine the form of the Cu-rich particles, scanning transmission electron microscopy (STEM) was used to characterize the interfacial region (3 μm × 300 μm) in the thin area of the sample. Cu-rich particles in G115 steel formed a face-centered cubicstructured Cu-rich phase, which contained about 90.28 wt% Cu (Table 1). The particles were elliptical or spherical with an equivalent diameter of 50–242 nm and an average diameter of 114 nm. They often coexisted with M23C6 and Laves phases along the lath boundary, and could exist independently at the lath boundary with many dislocations (Fig. 16). A three-dimensional atomic probe (3DAP) was used to characterize the different ages of the Cu in the G115 steel matrix. The results showed that a prolonged aging time facilitated Cu precipitation (Table 2). High-throughput statistical spatial-mapping characterization was used to locate and screen from the macroscopic to the microscopic areas, to achieve a high-resolution characterization of the existing form and a distribution of genetic units that contain Cu characteristics, and to reveal the evolution of Cu through a systematic characterization of Cu during the aging of G115 steel (Fig. 17).
Fig. 13. Characterization of the mesoscopic-to-macroscopic across-scale span distribution of μXRF-OPA technology [156]. (a) μXRF distribution of Cu; (b) μXRF distribution of Cr; (c) μXRF distribution of W.
Fig. 14. Characterization of microscopic-to-mesoscopic across-scale span distribution under 1000× magnification of EDS-SEM [156]. (a) SEM distribution of Cu; (b) SEM distribution of Cr; (c) SEM distribution of W.
Fig. 15. Characterization of microscopic-to-mesoscopic across-scale span distribution with 20 000× magnification of EDS-SEM [156]. (a) SEM distribution of Cu; (b) SEM distribution of Cr; (c) SEM distribution of W.
Fig. 16. Characterization of microscopic distribution by STEM [156]. (a) STEM distribution of Cu; (b) STEM distribution of Cr; (c) STEM distribution of W; (d) transmission electron microscopy (TEM) image of Cu; (e) electron diffraction pattern of Cu in selected region.
Fig. 17. Characterization of Cu-rich phase in G115 steel by across-scale span high-throughput statistical spatial-mapping [156].
Table 1 STEM-EDS analysis of the characteristic genetic units of Cu in G115 steel [156].
Table 2 Analysis results of Cu content in the matrix of G115 steel during aging by 3DAP [156].
3.2. Statistical spatial-mapping characterization of the composition, structure, and properties of macroscopic deforming FGH96 turbine disks
Ni-based superalloy is a key material in aeroengines and gas turbine disks, but its modification and optimization in research and development cycles are long. Because its chemical composition is complex, service environments are hostile and performance requirements are strict. Macroscopic deforming FGH96 turbine disks use electroslag remelting continuous directional solidification (ESR-CDS) to prepare ingots and multidirectional and isothermal forging for molding. The mechanical properties of FGH96 turbine disks made by ESR-CDS are similar to those of the powder FGH96 alloy and this deforming alloy still exists in the engineering-development stage. Lu et al. [157,158] characterized slices of macroscopic deforming FGH96 turbine disks using multiple statistical-distribution characterization technologies, including Spark-OPA, FVM-OPA, and SEM-OPA. Data on the distribution of various disk parameters on the disk were obtained, and included various compositions; the total amount of the phase, primary phase, secondary phase, tertiary phase, and particle size of the phase; grain size; carbide phase; microhardness; roomtemperature stress and strain; and creep at high temperature (Fig. 18). Statistical mapping with point-to-point correspondence was established for these data. For 0–100 nm, the relative mass fraction of the phase and the atomic fraction of cobalt (Co) and Mo that entered the phase had a significant impact on the creep properties at high temperature. A mathematical model of relevance to the regional statistical mapping was established between the genetic unit of the superalloy phase and the properties of creep at high temperature (Fig. 19), which plays an important role in guiding the modification of superalloy turbine disks.
phase, primary phase, secondary phase, tertiary phase, and particle size of the phase; grain size; carbide phase; microhardness; roomtemperature stress and strain; and creep at high temperature (Fig. 18). Statistical mapping with point-to-point correspondence was established for these data. For 0–100 nm, the relative mass fraction of the phase and the atomic fraction of cobalt (Co) and Mo that entered the phase had a significant impact on the creep properties at high temperature. A mathematical model of relevance to the regional statistical mapping was established between the genetic unit of the superalloy phase and the properties of creep at high temperature (Fig. 19), which plays an important role in guiding the modification of superalloy turbine disks.
Fig. 18. Characterization of multi-parameter high-throughput statistical distribution. MC, M23C6, and M3B6 are type carbides [154].
Fig. 19. Mathematical model of the relevance of statistical spatial-mapping between the phase and high-temperature creep.
phase and high-temperature creep.4. Prospects
High-throughput statistical spatial-mapping characterization is a new technology based on OPA theory, which has led to the development of a series of new methods and new apparatuses with independent intellectual property rights such as Spark-OPA, LIBS-OPA, LA-OPA, μXRF-OPA, FVM-OPA, SEM-OPA, and fluid microprobe-OPA. Many application results have been achieved using high-throughput statistical spatial-mapping characterization technology. Across-scale span high-throughput characterization of composition, structure, and mechanical properties has been realized for all kinds of carbon steel, stainless steel, nonferrous metal, continuous casting slabs, coated plates, superalloy, and heat-resistant steel. In practice, materials, parts, and components in use at the macroscale are inherently non-uniform or heterogeneous in regard to their composition/structure/properties at different scales. The applications of high-throughput experimental tools are extremely useful to fully characterize materials’ composition/ structure/properties at different scales, and to establish various valuable composition–structure–property relationships. Hence, the seven high-throughput experimental characterization tools presented here, and the composition–structure–property data and relationships that can be obtained through their use, powerfully enable the materials genome engineering methodology.
Through top-down analysis from the macroscale to the microscale, genetic units that affect material properties are screened out, and a model of across-scale span statistical spatial-mapping of the material composition, structure, and properties can be established. The most significant advantage of this method is its similarity to the production process that helps to provide insight into material modification and process optimization. Its disadvantage is that the design-regulated degree of freedom is limited by the process. Future research should focus on the integration of these tools, establish a data analysis process, establish further tools such as machine learning tools, and improve the calibration or coordination of positions while applying two or more characterization tools. Such research would make it possible to reconstruct the highresolution composition, structure, and properties of the overall macroscopic material at each position and in each microregion, in order to significantly accelerate the discovery and reverse design of new materials.
Acknowledgements
This research was supported by the National Key Research and Development Program of China (2016YFB0700300). The authors acknowledge helpful discussions with Profs. Hong Wang, Xiaodong Xiang, and Liang Jiang. We thank Laura Kuhar, Ph.D. from Liwen Bianji, Edanz Group China (www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript.
Compliance with ethics guidelines
Haizhou Wang, Lei Zhao, Yunhai Jia, Dongling Li, Lixia Yang, Yuhua Lu, Guang Feng, and Weihao Wan declare that they have no conflicts of interest or financial conflicts to disclose.
Appendix A. Supplementary data
Supplementary data associated with this article can be found in the online at https://doi.org/10.1016/j.eng.2020.05.005.