Determining microstructures of additively manufactured metallic materials using in-situ process monitoring data

Abstract

Methods, devices and systems are described for improving the manufacturing of components, devices, and materials. The described technique can predict microstructures including defects, such as cracks, porosity, undesirable phases, or non-optimal grain structures, segregations, and dendrite structures of alloys and superalloys manufactured by additive manufacturing, such as three-dimensional (3D) printing including laser powder bed fusion (LPBF) and laser melting deposition (LMD). Example alloys and superalloys include Rene 65, IN625, Mar-M-247, CM247LC, Rene 108, IN738, Ti-6Al-4V, AlSi10Mg, or IN718.

Description

DETERMINING MICROSTRUCTURES OF ADDITIVELY MANUFACTUREDMETALLIC MATERIALS USING IN-SITU PROCESS MONITORING DATACROSS-REFERENCE TO RELATED APPLICATION(S)

[0001] This application claims priority to the provisional application with serial number 63 / 603,840 titled “Determining Microstructures of Additively Manufactured Metallic Materials Using In-Situ Process Monitoring Data,” filed November 29, 2023. The entire contents of the above noted provisional application are incorporated by reference as part of the disclosure of this document.STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH

[0002] This invention was made with government support under Grant No. 80NSSC21M0100 awarded by NASA. The government has certain rights in the invention.TECHNICAL FIELD

[0003] This patent document is generally related to additive manufacturing methods and device, and in particular to improved additive manufacturing with reduced defects or microstructures.BACKGROUND

[0004] Additive manufacturing processes, such as laser powder bed fusion, produce material by localized melting of a powder feedstock layer by layer. The small melt pools and high energy density generate very different microstructures in nickel superalloys when compared to more traditional cast or wrought processing, including features such as cellular structures and epitaxial grain growth. The features of these microstructures vary depending on local thermal history, alloy chemistry, and processing parameters. There is a need to better understand the nature of such microstructures and how their formations are influenced by various factors in order to improve manufacturing of the components.SUMMARY

[0005] The disclosed embodiments relate to methods, devices and systems for improving the manufacturing of components, devices, and materials that, among other features and benefits,predict microstructures including defects, segregations, and dendrite structures of alloys manufactured by additive manufacturing, such as three-dimensional (3D) printing including laser powder bed fusion (LPBF) and laser melting deposition (LMD). Example alloys include Rene 65, IN625, Mar-M-247, CM247LC, Rene 108, IN738, Ti-6A1-4V, AlSilOMg, or IN718. The microstructures to be predicted typically include solidification microstructures, such as epitaxial grains, cellular dendrite spacing, or micro-segregation. The methods can also predict defects, such as cracks, porosity, undesirable phases, or non-optimal grain structures.BRIEF DESCRIPTION OF THE DRAWINGS

[0006] FIG. 1 illustrates an example layout of a sample material used for preparing additively manufactured components in an experiment conducted in accordance with the disclosed embodiments.

[0007] FIG. 2 illustrates example optical microscope images of multiple samples, and their porosities, obtained based on an experiment conducted in accordance with the disclosed embodiments.

[0008] FIG. 3 illustrates an example color-scaled longwave infrared image of samples obtained based on an experiment conducted in accordance with the disclosed embodiments.

[0009] FIG. 4 illustrates example layer-based local thermal history and cooling profile for a sample obtained based on an experiment conducted in accordance with the disclosed embodiments.

[0010] FIG. 5 illustrates fitted curves for the cooling profile based on example embodiments.

[0011] FIG. 6 illustrates cooling profiles at the centroid of five samples obtained in accordance with example embodiments.

[0012] FIG. 7 illustrates images of cellular dendrites and their average spacing for select samples obtained in accordance with example embodiments.

[0013] FIG. 8 illustrates Scheil model elemental segregation results for IN718 solidification in accordance with example embodiments.

[0014] FIG. 9 illustrates Energy Dispersive Spectroscopy (EDS) maps of a IN718 sample in accordance with example embodiments.

[0015] FIG. 10 illustrates EDS line scan data corresponding to a sample in accordance with example embodiments.

[0016] FIG. 11 illustrates a set of operations that can be carried out to determine amicrostructure of an additively manufactured part in accordance with an example embodiment.

[0017] FIG. 12 illustrates a set of operations that can be carried out to improve additive manufacturing of a component in accordance with an example embodiment.

[0018] FIG. 13 illustrates a set of operations that can be carried out for an improved additive manufacturing process in accordance with an example embodiment.

[0019] FIG. 14 illustrates a set of operations that can be carried out for an improved additive manufacturing process in accordance with another example embodiment.DETAILED DESCRIPTION

[0020] The disclosed embodiments use process monitoring data to provide real-time and localized initial and intermediate conditions to computerized models to compute the microstructures. One of the advantages of the disclosed embodiments is that the process monitoring data allows the thermodynamics models to yield accurate predictions of the solidification microstructures. Model predictions can be compared to microscopic images to improve the algorithms such as through fine tuning thermodynamics coefficients used by the computerized models, making future predictions even more accurate. Many types of process monitoring parameters can be used, such as plasma emission, plasma absorption, reflected light, scattered light, radiated light, frequency response of acoustic signals, thermal data comprising a temperature or a cooling rate, thermal tomography, longwave infrared imaging, spatter imaging, recoat imaging, or coaxial melt pool thermal imaging.

[0021] The disclosed embodiments can be used to improve microstructures of a region or of a part being manufactured by analyzing the relationships between process monitoring data and microstructures of previous regions or previous parts. The disclosed methods and systems can also be used to provide alerts if microstructures or defects of a part being manufactured are expected to exceed a threshold.

[0022] As noted earlier, the features of microstructures formed as part of additive manufacturing processes vary depending on local thermal history, alloy chemistry, and processing parameters. There is a need to develop a systematic understanding of the influence the local thermal conditions during solidification have on the resulting microstructure. Such understanding is useful in predicting and ultimately avoiding microstructural defects, such as undesirable phases or non-optimal grain structures. Among the technical solutions described in this patent document,in-situ Longwave Infrared imaging of a laser powder bed fusion process is used to characterize the local thermal conditions throughout additively manufactured builds for an example alloy IN718 processed using systematically varied process parameters. This information is then correlated to observations of the microstructural features of these alloys in the as-built condition. This correlation analysis shows clear influence of the local thermal conditions during solidification on the dimensions of the dendritic microstructures formed during the build process for IN718. These dendritic structures arise due to segregation of elements, such as niobium, during solidification, an observation which can be predicted in some embodiments using a Scheil modeling approach.

[0023] Nickel-based superalloys are known for maintaining strength, toughness, fatigue performance and resistance to corrosion and oxidation at elevated temperature and pressure. These properties make nickel-based superalloys key materials in turbine engines for aircraft and power generation. Some components of gas turbines require complex geometries for which nickel-based superalloys can be difficult to machine and fabricate. Additive manufacturing (AM), a method that builds parts in a layer-wise process, enables the production of parts with complex geometries where traditional manufacturing methods have limited ability. Thus, AM, especially Laser Powder Bed Fusion (LPBF), has found growing applications in fabricating intricate part geometries with hard-to-machine metals used in, for example, aviation and aerospace. In a LPBF system, as the focused laser beam scans each layer of powder, it melts the powder into the shape of the cross section of the part to form the designed 3D model. This building process generates microstructures with features including cellular structures and epitaxial grain growth, which is very different from those formed in traditional processes, such as forging and casting. The local thermal history, alloy chemistry, and processing parameters all contribute to the features of the microstructures.

[0024] In the description that follows, the local thermal conditions during the LPBF building process of alloy IN718 are used as an example to facilitate the understanding of the disclosed technology. Specifically, the local thermal conditions during the LPBF building process of alloy IN718 are recorded using an in-situ process monitoring system for different parts built using varying process parameters. The monitoring data is correlated to the microstructural features of these alloys in the as-built condition characterized by electron microscopy and optical microscopy. This information is then compared to computational modeling of the segregation and dendrite structures formed during the solidification. These provide a deeper systematic understanding ofthe interplay between local thermal condition variation during solidification and the resulting microstructure, and allow for better process designs to mitigate not only macroscopic defects, such as porosity and cracks, but also microstructural defects, such as undesirable phases or non-optimal grain structures.General background on LPBF of IN718

[0025] IN718 is a common nickel -based superalloy. IN718 is a precipitation hardening alloy, its excellent mechanical properties benefit from the precipitation phase y" (Ni3Nb). For the last few decades, IN718 has been used in components such as casings, shafts, disks and compressor blades and vanes of jet engines.

[0026] The deposition of layers in LPBF involves highly localized laser energy input, high laser scanning velocity and short interaction time with the melt pool. These conditions result in large thermal gradients and high cooling rates during the solidification and thus cause directional grain growth accompanied by micro-segregation and precipitation of metastable phases. Studies on IN718 fabricated by LPBF have shown epitaxial columnar dendrites approximately along the build direction. In the as-built material, carbides and Laves precipitates were found in the interdendritic spaces with segregation of niobium, titanium, and molybdenum, and the presence of y' can be indicated by X-Ray Diffraction (XRD) analysis. After heat treatment, partial recrystallization and homogenization may occur depending on the temperature and y', y" and 8 precipitates were found at the grain boundaries.Observations of solidification structures of AM superalloys- Rene 65, IN625, Mar-M-247, CM247LC, Rene 108, IN738

[0027] Similar solidification structures have been found in some nickel-based superalloys whose chemistries are similar to IN718. For example, some prior work has studied the microstructure of Rene 65, a y' strengthened alloy, processed by LPBF. Cellular dendritic structures rich in y' forming elements were found in the as-built material. The microstructures developed in heat treatment at temperatures below and above y' solvus were examined. The subsolvus grains still showed some elongation (aspect ratio about 1.5) in the build direction. The supersolvus grains were close to equiaxed. Both heat treated microstructures showed a multimodal distribution of y' precipitates, and the supersolvus material had a generally finer distribution than subsolvus material. IN625 is a solid solution strengthened alloy. The as-built LPBF IN625consists of cellular dendrites and elongated grains in the build direction. Others have observed columnar arrays of fine y" nanoparticles along the boundaries of the melt pool. C247LC is a nickel-based superalloy considered hard to weld due to the high Ti and Al content and is susceptible to strain-age cracking. Another study has observed y' cellular dendrites and microcracking in LPBF fabricated CM247LC. After heat treatment above 750°C up to 975°C, precipitation of intra-cellular y', y' films at cell and grain boundaries and M23C6 carbides at grain boundaries were observed. Alloy IN738LC has poor weldability and is susceptible to hot cracking as well. In one study of IN738LC, dendrite structures with no distinct y' precipitates were found in the as-built condition. A unimondal y ' distribution was developed after solution and aging heat treatment. When the hot isostatic pressing (HIP) was applied before solution and aging heat treatments a bi-modal y' distribution with fine secondary y' between coarser y' precipitates was observed. Microcracks formed during LPBF process can be reduced by HIP.Modeling of solidification structures in AM, emphasis on superalloys

[0028] Although the essentials of AM process make the grains prone to grow into columnar dendrites, the transition of columnar to fine equiaxed grains is found at the top of each layer in some experiments. The mechanisms of nucleation and grain growth in AM process have been studied by modeling the microstructures formed during solidification. This research concluded that the growth of equiaxed grains requires: a low temperature gradient G, a high cooling rate and an alloy containing grain refining solute and potent nucleant particles. These conclusions are in accordance with what is demonstrated in the experiments. For instance, one study has reported the columnar to equiaxed transition (CET) at the top of each layer where the temperature gradient is low and the cooling rate is high during laser melting deposition (LMD) of Ti-2AL7Mo alloy. Another study reported that the CET in cylindrical samples of AlSilOMg alloy produced by direct metal laser sintering (another name for LPBF) is dependent on the build direction. Another study reported a mix of columnar and equiaxed grains in titanium alloys containing La2Ch particles fabricated by a wire arc AM process. While these models describe the mechanisms that drive grain structure formation, so far there is limited work available on modeling the formation of solidification structures, such as cellular dendrites in LPBF manufactured nickel superalloys.Process Monitoring Examples in LPBF

[0029] The development of in-situ process monitoring and control remarkably improves therobustness of the AM process and the quality of additively manufactured parts. The basic idea of in-situ monitoring in the LPBF process is to collect the information of “process signatures” associated with the melt pool and surrounding heat affected zone (HAZ). Process signatures can include electromagnetic signatures, such as plasma emission / absorption, reflected / scattered light, and radiated light. Prior research has also examined acoustic signatures, and the frequency response of acoustic signals has shown correlation with several weld quality metrics and process phenomena, such as keyhole formation, plasma formation, and crack propagation. Due to practical system integration limitations, there has been limited research on the adoption of acoustic monitoring in LPBF system. The electromagnetic signals can be used to determine the melt pool geometry, temperature, cooling rate and thermal history in the HAZ. Many in-process monitoring systems based on electromagnetic sensing have been developed for LPBF. The electromagnetic sensors may include photodiodes, pyrometry, and visible light or infrared cameras. In one prior experiment, a high-speed camera and a photodiode have been utilized to measure the dimension of the melt pool and mean radiation emitted, respectively. With this method, only the active area in LPBF is monitored. Single channel detectors such as photodiodes and pyrometers have small field of view but high sensitivity, fast data collection speed and low cost. The cameras using CCD or COMS detectors, either visible or IR, enable the spatial resolution of imaging and thus allow the mapping of the build area in LPBF. However, the cameras have slower response time than single-channel detectors and post processing of data is a great challenge.

[0030] In these examples described herein, we utilize in-situ process monitoring data streams to describe the local thermal history and its effect on microstructure formation in PBF-LB builds of IN718. In particular, we use Longwave Infrared data to construct the time-temperature history at locations throughout a designed experiment and correlate these histories with features of solidifications structures such as dendrite spacing that are determined by cooling rates during the solidification process. We used a designed experiment with intentional variation of energy density, but in contrast to most prior studies that have taken this approach, we allowed for relatively modest variation in process parameters such that the majority of the material studied would be free from defects such as cracks and porosity and our analysis focused on the variation in microstructure seen in as-built materials that would be generally deemed of high quality by end users.Example Experimental Methods

[0031] All parts in this example study were built with gas atomized powder. The chemical composition of the IN718 powder provided by the manufacturer is given in Table 1. The IN718 samples were built using a custom designed and fabricated PBF-LB test bed by Open Additive. This machine has a notional build volume of 6” by 6” by 6” delivering the 1070 nm fiber laser light via a SCANLAB varioSCAN and intelliSCAN, which enables coaxial meltpool thermal imaging. Each IN718 sample is 10 mm tall - the bottom 7 mm is inverted pyramid shaped, and the top 3 mm is square cuboid shaped with 10 mm x 10 mm square cross-section. The samples were built with varied laser power and laser scanning velocity. All samples were built with 30 pm build layers using a 70 pm hatch spacing within a 5 mm tile size in a hatch strip approach. The 7 mm tall, inverted pyramid structured base of each sample was built with a common parameter using a 175 W laser power and 1100 mm / s scan speed to reduce the effects of proximity to the build plate heat sink in the samples. The top square cuboid portion of each sample was built with varying processing condition in the experiment by varying laser power and scan speed across the experimental layout. The layout of the samples is illustrated in FIG. 1. Figure 1 shows layout and building parameters for IN718. The vertical axis in FIG. 1 denotes the laser scan speed, the horizontal axis shows the laser power (W), and the samples are identified using row-column numbers (e g., 11 represents the sample in row 1, column 1).Table 1: Composition of IN718 powderExample Process Monitoring Methods

[0032] The machine used to create the metallurgical samples was equipped with both commercially available and R&D sensors. Longwave IR, thermal tomography, spatter, recoat imaging, galvo position data, and coaxial melt pool thermal imaging were recorded for each build. Some features of these devices are described in further detail below.

[0033] Recoat imaging takes an image before and after recoat using a standard silicon-based CMOS sensor. White light LEDs are used for illumination and the optical filtration is just to prevent the passage of laser light. These images reveal insights into the quality of the recoat process, including short feeds, damaged recoater blades, part peel up, or other process anomaliesthat can disturb the layer of powder. Thermal tomography is also based on a silicon-based CMOS camera that images the entire powder bed while being filtered to being sensitive only in the Near Infrared (NIR) with a long integration time (e g., 250 ms) and a frame rate of, for example, 4 fps. Individual images are stitched together to create a composite image on a layer-wise basis. Thermal tomography is sensitive to many process anomalies such as spatter events, uneven gas flow, and part warping among many others. The spatter camera is also based on a silicon-based CMOS sensor filtered for the same band in the NIR but has a shorter integration time and operates at a higher frame rate, such as 150 fps. Typical exposure times are on the order of 500 ps but for extremely bright materials, such as tungsten, can be as low as 25 ps. A GPU accelerated analytic works to detect the presence of slow-moving hot objects, referred to as ‘welded spatter’ that tend to get incorporated into the resultant microstructure and, given that their mass is much larger than individual powder particles, result in Lack of Fusion (LOF). Other metadata from the images is also measured and logged in an HDF5 file for post process analysis as desired.

[0034] Longwave IR imaging is a valuable tool for measuring the temporal dynamics of the solid state cooling in the LPBF process. By its nature, LWIR imaging is done off axis, typically viewing the process through a germanium window. The Optris PI640 is a VGA resolution microbolometer that can run at 32 Hz, is sensitive from 7 - 13 pm, and has four temperature ranges that will measure as low as -20 °C to 1500 °C. The temperatures returned from this camera are useful for relative temperature changes but absolute temperatures can vary by 50 °C or more due to changes or ambiguity in the material’s emissivity. For the builds, the camera was run at full resolution at 32 Hz in the 150° - 900°C range for its full duration. The field of view captured the whole build area allowing different parts to be segmented from the data set.

[0035] Coaxial melt pool thermal imaging is accomplished using a NIR filtered CMOS camera that has been blackbody calibrated to return temperature measurements. It is windowed down to run between 1000 and 1500 Hz depending on application and for these builds was set to 1000 Hz. The camera was situated so that it images down the beamline and before the scanner such that the field of view of the camera is always centered on the melt pool. To enable registration of the images to their location on the build, an FPGA was developed to record the commanded scanner position that was written to disk in a CSV file. Using the time stamps of the galvo positions and the melt pool images, enable registration of one to the other facilitating more detailed analysis.Example Metallurgical Analysis

[0036] The samples were removed from the build plate and cut along center x-z cross section by a wire Electrical Discharge Machining (EDM) device. The sectioned samples were mounted in phenolic resin by compression mounting process. Samples for microstructure analysis were prepared using metallographic methods including successive grinding using 240, 320, 400, 600, 800 and 1200 grit silicon carbide papers. Final polishing was performed progressively with 9- micron diamond slurry, 1 -micron diamond slurry and 0.05-micron Alumina abrasive paste. The porosity in the as-polished x-z cross section was examined by KEYENCE VHX-700 optical microscope. IN718 samples were electrolytically etched in 10% phosphoric acid solution at 2.5 V. A TESCAN MIRA3 Scanning Electron Microscope (SEM) with Schottky field emission gun (FEG) was used to characterize the microstructures including epitaxial grains, cellular dendrite spacing, and micro-segregation.Example Image Analysis Methods

[0037] The porosity was measured by the optical microscope based on the contrast difference between the holes and surrounding the surface. The measurement of the dendrite spacing was performed on the SEM micrographs of the x-z cross sections at lOkx magnification using the image analysis software ImageJ. On the lOkx SEM images, clusters of cellular dendrites growing over a couple of melt pools along z axis (build direction) were targeted. The width of each cluster was measured and the number of dendrites within each cluster was counted. The dendrite spacing of a sample was calculated by averaging the total width of the targeted clusters over the total number of dendrites.Example Results

[0038] Porosity: The porosity of IN718 samples 11, 22, 44, 66, 77 and 88 were measured to evaluate the defect density of samples across the broadest range of energy densities in the experimental matrix. The examined area of each sample in the center x-z plane was 10 mm x 3 mm. The optical microscope images of samples 11, 44 and 88 are listed in FIG. 2 (panels (a), (b) and (c), respectively). The porosities of IN718 diagonal samples are shown in panel (d) of FIG. 2. Sample 11 was built with the following parameters: 125 W laser power and 1300 mm / s scan speed (which has the lowest energy density input). The formation of holes in sample 11 could be attributed to the lack of fusion while printing. The irregular shape of holes shown in panel (a) alsoindicates the lack of fusion. Sample 88 was built with 300 W and 600 mm / s. It has the highest energy density input among samples. The formation of holes in this sample are more spherical in nature and likely due to keyhole formation at higher energy density.Examples of Local Thermal History Data Collection and Processing

[0039] FIG. 3 shows a sample color-scaled LWIR image. Longwave IR imaging data was collected throughout the build and analyzed first at layers of interest and then locations of interest for extracting local thermal history data. FIG. 3 shows an adjusted and color-scaled LWIR image. It is important to note that the temperatures captured by the LWIR camera undergo spatial and temporal averaging. Therefore, the local thermal history recorded by the camera is not absolute but rather relative. Hence, though the complete melting of IN718 takes place at a temperature range of 1370-1430 °C in the LPBF process, the maximum temperature visible in the color-scaled image of FIG. 3 is 430°C.

[0040] FIG. 4 shows sample layer-based local thermal history - (a) thermal history at a pixel point for the time of printing one layer; (b) corresponding cooling profile starting from the highest peak temperature. Note that the asterisk (*) symbol indicates the time when the temperature reaches steady ambient temperature starting from the highest peak temperature. Layer-wise local thermal history is obtained by collecting the calibrated temperature values from all the LWIR images captured during the printing of a single layer. Panel (a) shows a sample local thermal history of a pixel point for the time of printing one layer of the build. In this example, our primary focus is on the cooling profile starting from the highest peak temperature and until the temperature reaches a steady ambient temperature, which was around 225°C for the experiments conducted in this study. A pixel-level cooling profile is illustrated in panel (b) of FIG. 4. Assuming that the cooling rate itself is not significantly affected by the aforementioned spatial and temporal averaging, the cooling profile can be backward extrapolated to obtain a profile starting from the melting temperature of the build material in the LPBF process. In doing so, the measured pixellevel cooling profile is used to estimate the temperature decay model for a specific pixel.

[0041] FIG. 5, in panel (a), shows second-order exponential (Exp2) fitted curve to the cooling profile and its two components; and in panel (b) shows extrapolation of the 2nd order exponential (Exp2) fitted curve up to 1500°C. The cooling data is best approximated by the second-order exponential (Exp2) decay function of the form= a x ebxt+ c X edxtsince it canclosely model the physics of cooling in the LPBF process with the superposition of its two components. In the LPBF process, after the laser passes by a point and melting happens, the material temperature starts cooling exponentially towards the ambient temperature at a rapid decay rate - this effect is captured by the first component of the / EXP2( - However, this decay in temperature changes its rate after a short period of time, thanks to heat conduction from the neighboring laser-traversed points - this effect is captured by the second component of the EXPZC - Panel (a) in FIG. 5 shows the Exp2 curve fitting to the cooling profile shown in FIG. 4’s panel (b) along with the two components of the fitted curve. The equation of the Exp2 fitted curve in panel (a) is fExp2. = 221.5 x e-2,85t+ 225.1 xe-3-56xl°6tand the / ?2-value of the fit is 0.9962. Using the fitted second-order exponential model, the local cooling profile can be backward extrapolated to the melting point of the build material in the LPBF process. Panel (b) of FIG. 5 illustrates the extrapolation of the Exp2 fitted curve shown in panel (a) up to 1500°C.

[0042] FIG. 6 shows cooling profiles at a selected pixel point of the IN718 samples 11, 22, 44, 77 and 88, where the differences in cooling rate and cooling time are clearly visible. The selected points are located at the centroid of the top square cuboid portion of the samples. Recall that the samples are 10 mm tall - the bottom 7 mm is inverted pyramid shaped, and the top 3 mm is square cuboid shaped with 10 mm x 10 mm square cross-section. Therefore, the selected points are located at a height of 8.5 mm (7 mm + 3 / 2 mm) and at the center of the 10 mm x 10 mm square cross-section of the square cuboid. Note that the thermal history and cooling profile presented in FIGS. 4 and 5 correspond to the selected point in Sample 88. The differences in the local thermal history can be correlated to the microstructural features of the samples under investigation - which will be discussed in the next section.Example Solidification Structure

[0043] FIG. 7, in panel (a) shows cellular dendrites along z direction found in IN718 samples 11, 22, 44, 77 and 88. Images were taken by SEM at 10k magnification. Panel (b) in FIG. 7 shows a long range of dendrites over a couple of melt pools in sample 88. The image was taken by SEM at 2k magnification. The epitaxial grain structure is found in all of the IN718 diagonal samples. Panels (a) and (b) of FIG. 7 show the clusters of cellular dendrites along build direction found in the samples. The dendrite cell spacing of 11, 22, 44, 66, 77 and 88 are measured and provided in panel (c) of FIG. 7. As the volumetric energy density input increases from sample 11 to sample88, the dendrite spacing increases. The higher energy input leads to a higher thermal gradient and longer cooling time, which allows more time for dendrite growth and micro- segregation. The trend of increasing dendrite cell spacing increasing with corresponding increases in powder density is consistent across the entire range of the experiment, however there is an observed deviation from the trend of increasing cooling time with increasing power density for the highest power density sample 88. In FIG. 6, the cooling time is less for sample 88 than for sample 77, despite a higher power density parameter used for sample 88. While the exact cause of this deviation is not yet clear, analysis of alternate locations in samples 77 and 88 shows significant variability in cooling time for sample 88 relative to other samples. This suggests that at very high power densities that result in the generation of significant keyhole porosity defects, which are shown in panel (c) of FIG. 2 for sample 88, there may be phenomena such as vaporization and condensation in turbulent melt pools that cause additional complexities in the thermal history of the material.

[0044] FIG. 8 shows Scheil model elemental segregation results for IN718 solidification. These dendritic structures are formed during solidification due to the preferential segregation of elements such as niobium to the liquid phase. This segregation can be predicted using CALPHAD models such as Pandat. Utilizing a Scheil model in the Pandat PanPhaseDiagram module, the solidification segregation predictions are shown in FIG. 8. As IN718 solidifies, elements such as chromium and iron are expected to preferentially incorporate in the solid dendrite cores, while elements such as niobium, molybdenum, titanium and aluminum are pushed into the interdendritic regions where they form the dendrite cell boundaries upon complete solidification.

[0045] FIG. 9 shows Energy Dispersive Spectroscopy (EDS) maps of IN718 sample 88. Panel (a) shows secondary electron image at 50kx. Panels 9(b)-(e) show EDS signal of Nb, C, Ti and Ni. Niobium and titanium are found rich in the interdendritic regions as the CALPHAD model predicts. FIG. 10 shows EDS line scan data of IN718 sample 88. Notably, FIG. 10 shows the EDS line scan data on another site of the same sample. The titanium peak indicates a Ti-rich particle on the cell wall, presumably a titanium containing carbide or carbonitride particle due to the presence of carbon also noted in the cell walls.

[0046] The results show that the dendrite spacing in as-built IN718 vary as a function of processing parameters and energy density, and that this variation can be related to timetemperature information that can be obtained using Longwave IR in-situ process monitoring. Thisvariation is observed even within a relatively narrow range of processing parameters that can be used to build material free of significant numbers of defects such as cracks and porosity. Material evaluated in this work would be generally considered acceptable if analyzed via nondestructive evaluation or metallographic analysis of witness coupons.

[0047] Significant variation can be present in material built with a single fixed process parameters but having complex variations in thermal history due to geometric complexity in the component design. This variation in thermal history and dendritic structure leads to variations in the spatial distribution of elements that are important in the formation of phases in these alloys, for example the segregation of niobium in IN718 can lead to regions of Laves phase formation. This size and distribution of these phases can have a significant impact on mechanical properties of the material. Control of the microstructure of these alloys via closed loop process control enabled by in-situ process monitoring could produce materials with more consistent mechanical behavior for the end user.

[0048] As evident from the foregoing, longwave IR data provides a useful measure of temporal variation of solid state cooling throughout the build, by nature of the wide viewing area and relatively low data collection rate this method captures heat flow in a lower temperature regime across the build plate. This analysis has shown that this can be correlated to trends in microstructure formations and the data can be extrapolated to higher temperatures to construct location specific cooling curves. However, addition of additional process monitoring data streams such as thermal tomography and melt pool thermal imaging can capture complementary higher temperature data which can be used to improve the accuracy of thermal history reconstruction at specific locations in the build. These reconstructed cooling curves can be used in more sophisticated solidification models such as Computherm’s PanSolidification module that utilize mobility information to make predictions of solidification structures that include spatial information such as dendrite arm spacing. The coupling of in-situ monitoring of local thermal conditions and prediction of microstructure can be used to control LPBF processes and validate component quality for AM parts that can be difficult to inspect adequately using available NDE methods.

[0049] As illustrated herein, in-situ process monitoring provides useful information towards understanding the variation observed in solidification structures of alloys and superalloys,including nickel superalloys such as IN718. The following are some of the features, benefits, and observations:

[0050] Dendrite spacing varies in IN718 as a function of energy density in LPBF builds.

[0051] This dendrite spacing variation is driven by variations in local cooling rates in the material.

[0052] The time-temperature history of the material throughout the build can be characterized and understood by utilizing in-situ process monitoring methods including Longwave IR data collection and processing.

[0053] Dendritic solidification leads to micro- segregation of elements such as niobium in IN718 that could have important implications for phase formation during subsequent processing.

[0054] FIG. 11 illustrates a set of operations that can be carried out to determine a microstructure of an additively manufactured part in accordance with an example embodiment. At 1102, one or more sets of process monitoring data is received by a processing device, where each set of the process monitoring data corresponds to a portion of the additively manufactured part, and each set of the process monitoring data further comprises timing information corresponding to when the portion is manufactured. At 1104, a computerized model representing the additively manufactured part is obtained by the computing device; the computerized model comprises a plurality of elements.. At 1106, by the computing device and based on applying each set of process monitoring data to the plurality of elements, a microstructure of the additively manufactured part is determined.

[0055] In one example embodiment, the additively manufactured part is manufactured by a three-dimensional (3D) printing process of laser powder bed fusion (LPBF) or laser melting deposition (LMD). In another example embodiment, the additively manufactured part comprises a Ti-based alloy or superalloy, an Al-based alloy or superalloy, an Fe-based alloy, a high entropy alloy or a Ni-Cr-based alloy or superalloy. In yet another example embodiment, each set of process monitoring data comprises process monitoring parameters for at least one of the following: plasma emission, plasma absorption, reflected light, scattered light, radiated light, frequency response of acoustic signals, thermal data comprising a temperature or a cooling rate, thermal tomography, longwave infrared imaging, spatter imaging, recoat imaging, or coaxial melt pool thermal imaging.

[0056] According to one example embodiment, the above noted method further includesextrapolating the one or more sets of process monitoring data to counter an averaging effect caused by a physical size of each portion of the additively manufactured part or an averaging effect caused by a time duration of process monitoring data measurement. In another example embodiment, each portion of the additively manufactured part is determined by at least one of the following: three coordinates; an identifier of a layer and two coordinates within the layer; or position data provided by a galvanometer laser. In still another example embodiment, the computerized model is a finite element analysis (FEA) model, a regression model, or a phase field model. In yet another example embodiment, determining the microstructure comprises performing thermodynamics computations on each element using the one or more sets of process monitoring data as initial or intermediate conditions.

[0057] In another example embodiment, the above noted method includes comparing the microstructure of the additively manufactured part with a microscopic image to optimize thermodynamics coefficients in the computerized model. In yet another example embodiment, determining the microstructure comprises determining a solidification microstructure of the additively manufactured part comprising at least one of the following: epitaxial grains, cellular dendrite spacing, or micro-segregation. In still another example embodiment, determining the microstructure comprises determining a formation of defects comprising at least one of the following: cracks, porosity, undesirable phases, or non-optimal grain structures. In another example embodiment, the one or more sets of process monitoring data includes a plurality of sets of process monitoring data, and a total number of the plurality of elements is determined by a total number of the plurality sets of process monitoring data.

[0058] FIG. 12 illustrates a set of operations that can be carried out to improve additive manufacturing of a component in accordance with an example embodiment. At 1202, a first set of process monitoring data corresponding to a first layer of an additively manufactured part is received by a computing device, where the first set of process monitoring data comprises first timing information corresponding to when the first layer is manufactured. At 1204, by the computing device and based on applying the first set of process monitoring data to a first element of a computerized model representing the additively manufactured part, a first microstructure of the first layer is determined. At 1206, a second set of process monitoring data corresponding to a second layer of the additively manufactured part is received by the computing device, wherein thesecond set of process monitoring data comprises second timing information corresponding to when the second layer is manufactured, and wherein the second set of process monitoring data is adjusted based on the first microstructure. At 1208, by the computing device and based on applying the second set of process monitoring data to a second element of the computerized model, a second microstructure of the second layer is determined, wherein the second microstructure is improved in an epitaxial grain structure, a cellular dendrite spacing, or a micro-segregation spacing compared with the first microstructure.

[0059] In one example embodiment, the second set of process monitoring data is caused to be adjusted by adjusting input process parameters on process monitoring devices. In another example embodiment, the second microstructure is further improved in terms of a formation of defects comprising at least one of the following: cracks, porosity, undesirable phases, or non-optimal grain structures. In yet another example embodiment, the first and second sets of process monitoring data each comprises process monitoring parameters corresponding to at least one of the following: plasma emission, plasma absorption, reflected light, scattered light, radiated light, frequency response of acoustic signals, thermal data comprising a temperature or a cooling rate, thermal tomography, longwave infrared imaging, spatter imaging, recoat imaging, or coaxial melt pool thermal imaging.

[0060] In one example embodiment, the computerized model is a finite element analysis (FEA) model, a regression model, or a phase field model. In another example embodiment, determining the first and second microstructures comprises performing thermodynamics computations on the first and second elements using the first and second sets of process monitoring data as initial or intermediate conditions. In still another example embodiment the method described in FIG. 12 further includes comparing the first and second microstructures with a microscopic image to optimize thermodynamics coefficients in the computerized model.

[0061] FIG. 13 illustrates a set of operations that can be carried out for an improved additive manufacturing process in accordance with an example embodiment. At 1302, a first set of process monitoring data corresponding to a first additively manufactured part is received by a computing device, wherein the first set of process monitoring data comprises first timing information corresponding to when the first additively manufactured part is manufactured. At 1304, it is determined, by the computing device and based on applying the first set of process monitoringdata to a first computerized model representing the first additively manufactured part, a first microstructure of the first additively manufactured part. At 1306, a second set of process monitoring data corresponding to a second additively manufactured part is received by the computing device, wherein the second set of process monitoring data comprises second timing information corresponding to when the second additively manufactured part is manufactured, and wherein the second set of process monitoring data is adjusted based on the first microstructure. At 1308, it is determined, by the computing device and based on applying the second set of process monitoring data to a second computerized model representing the second additively manufactured part, a second microstructure of the second additively manufactured part, wherein the second microstructure is improved in an epitaxial grain structure, a cellular dendrite spacing, or a microsegregation spacing compared with the first microstructure.

[0062] FIG. 14 illustrates a set of operations that can be carried out for an improved additive manufacturing process in accordance with another example embodiment. At 1402, a set of process monitoring data corresponding to a portion of an additively manufactured part is received by a computing device, where the set of process monitoring data comprises time information corresponding to when the portion is manufactured. At 1404, by the computing device and based on applying the set of process monitoring data to an element of a computerized model representing the additively manufactured part, a microstructure of the portion is determined. At 1406, the microstructure of the portion is compared with a reference microstructure by the computing device. At 1308, by the computing device and based on a difference between the microstructure of the portion and the reference microstructure exceeding a threshold, an instruction is transmitted to pause or stop a manufacturing process of the additively manufactured part.

[0063] In one example embodiment, the difference between the microstructure of the portion and the reference microstructure comprises at least one of the following: an epitaxial grain size difference, a cellular dendrite spacing difference, or a micro-segregation spacing difference. In another example embodiment, transmitting the instruction to pause or stop the manufacturing process is further based on a total number of defects exceeding a threshold; and the defects comprise at least one of the following: cracks, porosity, undesirable phases, or non-optimal grain structures. In still another example embodiment, the operations of FIG. 14 further include applying the difference between the microstructure of the portion and the reference microstructure tooptimize thermodynamics coefficients in the computerized model.

[0064] Another aspect of the disclosed embodiment relates to a device that includes a processor and a memory with instructions stored thereon, wherein the instructions upon execution by the processor cause the processor to perform the operations recited any one of FIGS 11-14, and methods described herein.

[0065] Various operations disclosed herein can be implemented using a processor / controller configured to include, or be coupled to, a memory that stores processor executable code that causes the processor / controller carry out various computations and processing of information. The processor / controller can further generate and transmit / receive suitable information to / from the various system components, as well as suitable input / output (IO) capabilities (e.g., wired or wireless) to transmit and receive commands and / or data. The processor / controller may, for example, provide signals to control the operation of various components such as light sources and detectors that are disclosed herein. The processor / controller may be further configured to perform various method steps and computations that are disclosed in this patent document.

[0066] Various information and data processing operations described herein may be implemented in one embodiment by a computer program product, embodied in a computer- readable medium, including computer-executable instructions, such as program code, executed by computers in networked environments. A computer-readable medium may include removable and non-removable storage devices including, but not limited to, Read Only Memory (ROM), Random Access Memory (RAM), compact discs (CDs), digital versatile discs (DVD), etc. Therefore, the computer-readable media that is described in the present application comprises non-transitory storage media. Generally, program modules may include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Computer-executable instructions, associated data structures, and program modules represent examples of program code for executing steps of the methods disclosed herein. The particular sequence of such executable instructions or associated data structures represents examples of corresponding acts for implementing the functions described in such steps or processes.

[0067] Only a few implementations and examples are described and other implementations, enhancements and variations can be made based on what is described and illustrated in this patent document.

Claims

CLAIMS1. A method for determining a microstructure of an additively manufactured part, comprising: receiving, by a computing device, one or more of sets of process monitoring data, wherein the one or more process monitoring data corresponds to a portion of the additively manufactured part, and wherein the one or more process monitoring data further comprises timing information corresponding to when the portion is manufactured; obtaining, by the computing device, a computerized model representing the additively manufactured part, wherein the computerized model comprises a plurality of elements; and determining, by the computing device and based on applying the one or more sets of process monitoring data to the plurality of elements, a microstructure of the additively manufactured part.

2. The method of claim 1, wherein the additively manufactured part is manufactured by a three-dimensional (3D) printing process of laser powder bed fusion (LPBF) or laser melting deposition (LMD).

3. The method of claim 1, wherein the additively manufactured part comprises a Ti-based alloy, an Al-based alloy, an Fe-based alloy, a high entropy alloy or a Ni-Cr-based alloy or superalloy.

4. The method of claim 1, wherein each set of process monitoring data comprises process monitoring parameters corresponding to at least one the following: plasma emission, plasma absorption, reflected light, scattered light, radiated light, frequency response of acoustic signals, thermal data comprising a temperature or a cooling rate, thermal tomography, longwave infrared imaging, spatter imaging, recoat imaging, or coaxial melt pool thermal imaging.

5. The method of claim 1, further comprising extrapolating the one or more sets of process monitoring data to counter an averaging effect caused by a physical size of each portion of the additively manufactured part or an averaging effect caused by a time duration of process monitoring data measurement.

6. The method of claim 1, wherein each portion of the additively manufactured part is determined by at least one of the following: three coordinates; an identifier of a layer and two coordinates within the layer; or position data provided by a galvanometer laser.

7. The method of claim 1, wherein the computerized model is a finite element analysis (FEA) model, a regression model, or a phase field model.

8. The method of claim 1, wherein determining the microstructure comprises performing thermodynamics computations on each element using the one or more sets of process monitoring data as initial or intermediate conditions.

9. The method of claim 1, further comprising comparing the microstructure of the additively manufactured part with a microscopic image to optimize thermodynamics coefficients in the computerized model.

10. The method of claim 1, wherein determining the microstructure comprises determining a solidification microstructure of the additively manufactured part comprising at least one of the following: epitaxial grains, cellular dendrite spacing, or micro-segregation.

11. The method of claim 1, wherein determining the microstructure comprises determining a formation of defects comprising at least one of the following: cracks, porosity, undesirable phases, or non-optimal grain structures.

12. The method of claim 1, wherein the one or more sets of process monitoring data includes a plurality of sets of process monitoring data, and wherein a total number of the plurality of elements is determined by a total number of the plurality sets of process monitoring data.

13. A method for improving additive manufacturing of a component, comprising: receiving, by a computing device, a first set of process monitoring data corresponding to a first region of an additively manufactured part, wherein the first set of process monitoring data comprises first timing information corresponding to when the first region is manufactured; determining, by the computing device and based on applying the first set of process monitoring data to a first element of a computerized model representing the additively manufactured part, a first microstructure of the first region; receiving, by the computing device, a second set of process monitoring data corresponding to a second region of the additively manufactured part, wherein the second set of process monitoring data comprises second timing information corresponding to when the second region is manufactured, and wherein the second set of process monitoring data is adjusted based on the first microstructure; and determining, by the computing device and based on applying the second set of process monitoring data to a second element of the computerized model, a second microstructure of the second region, wherein the second microstructure is improved in an epitaxial grain structure, a cellular dendrite spacing, or a micro-segregation spacing compared with the first microstructure.

14. The method of claim 13, wherein the first region is part of a first layer of the additively manufactured part, and the second region is part of a second layer of the additively manufactured part.

15. The method of claim 13, wherein the second set of process monitoring data is caused to be adjusted by adjusting input process parameters on process monitoring devices.

16. The method of claim 13, wherein the second microstructure is further improved in term of a formation of defects comprising at least one of the following: cracks, porosity, undesirable phases, or non-optimal grain structures.

17. The method of claim 13, wherein the first and second sets of process monitoring data each comprises process monitoring parameters associated with at least one of the following: plasmaemission, plasma absorption, reflected light, scattered light, radiated light, frequency response of acoustic signals, thermal data comprising a temperature or a cooling rate, thermal tomography, longwave infrared imaging, spatter imaging, recoat imaging, or coaxial melt pool thermal imaging.

18. The method of claim 13, wherein the computerized model is a finite element analysis (FEA) model, a regression model, or a phase field model.

19. The method of claim 13, wherein determining the first and second microstructures comprises performing thermodynamics computations on the first and second elements using the first and second sets of process monitoring data as initial or intermediate conditions.

20. The method of claim 13, further comprising comparing the first and second microstructures with a microscopic image to optimize thermodynamics coefficients in the computerized model.

21. A method, comprising: receiving, by a computing device, a first set of process monitoring data corresponding to a first additively manufactured part, wherein the first set of process monitoring data comprises first timing information corresponding to when the first additively manufactured part is manufactured; determining, by the computing device and based on applying the first set of process monitoring data to a first computerized model representing the first additively manufactured part, a first microstructure of the first additively manufactured part; receiving, by the computing device, a second set of process monitoring data corresponding to a second additively manufactured part, wherein the second set of process monitoring data comprises second timing information corresponding to when the second additively manufactured part is manufactured, and wherein the second set of process monitoring data is adjusted based on the first microstructure; and determining, by the computing device and based on applying the second set of process monitoring data to a second computerized model representing the second additively manufactured part, a second microstructure of the second additively manufactured part, wherein the second microstructure is improved in an epitaxial grain structure, a cellular dendrite spacing, or a microsegregation spacing compared with the first microstructure.

22. A method, comprising: receiving, by a computing device, a set of process monitoring data corresponding to a portion of an additively manufactured part, wherein the set of process monitoring data comprises time information corresponding to when the portion is manufactured; determining, by the computing device and based on applying the set of process monitoring data to an element of a computerized model representing the additively manufactured part, a microstructure of the portion; comparing, by the computing device, the microstructure of the portion with a reference microstructure; and transmitting, by the computing device and based on a difference between the microstructure of the portion and the reference microstructure exceeding a threshold, an instruction to pause or stop a manufacturing process of the additively manufactured part.

23. The method of claim 22, wherein the difference between the microstructure of the portion and the reference microstructure comprises at least one of the following: an epitaxial grain size difference, a cellular dendrite spacing difference, or a micro-segregation spacing difference.

24. The method of claim 22, wherein transmitting the instruction to pause or stop the manufacturing process is further based on a total number of defects exceeding a threshold, and wherein the defects comprise at least one of the following: cracks, porosity, undesirable phases, or non-optimal grain structures.

25. The method of claim 22, further comprising applying the difference between the microstructure of the portion and the reference microstructure to optimize thermodynamics coefficients in the computerized model.

26. A device, comprising: a processor and a memory with instructions stored thereon, wherein the instructions uponexecution by the processor cause the processor to perform the operations recited in any one of claims 1-25.

Images