Through-Process Modeling of Aluminum Alloys for Cold Spray: EXPERIMENTAL CHARACTERIZATION AND VERIFICATION OF MODELS

by

Baillie McNally

A Dissertation

Submitted to the Faculty

of the

WORCESTER POLYTECHNIC INSTITUTE

In partial fulfillment of the requirements for the

Degree of Doctor of Philosophy

in

Materials Science and Engineering

January 2016

Approved:

Richard D. Sisson, Jr., Advisor

Director of the Material Science and Engineering Program

George F. Fuller Professor Mechanical Engineering

ABSTRACT The cold spray process is a cost-effective process for repairing damaged parts or creating thin coatings and structural bulk materials for military vehicles and aircraft that require high maneuverability, durability, and energy efficiency. This process can be made even more robust with a predictive tool that would tailor the material and processing parameters to a variety of applications. A through-process model that includes powder production, powder processing, cold spray particle impact, and post-processing would benefit the current trial and error efforts immensely and would aid in the search for optimal cold spray alloys for different applications. The powder production stage addresses the microstructure, phases and strength that result from the gas atomization process. The powder processing stage takes into account microstructural effects from heat treating or degassing the powder before it is cold sprayed. The particle impact stage includes a finite element model that simulates the temperature generation and strain that occurs during cold spray. An additive strength model, which is applied to the powder and used as an input into the impact model, determines the contributions of solid solution, microstructural, and precipitation strengthening and is a function of particle diameter, and time and temperature of powder processing. The important parameters that are experimentally characterized to verify and enhance the described models are grain size, phase morphology and fraction, and nanohardness. Characterization techniques such as optical, scanning and transmission electron microscopy, focused-ion beam, x-ray diffraction, and nanoindentation are used to verify the various stages of the through-process model for aluminum alloys 6061 and 5056.

1 ACKNOWLEDGEMENTS The author would like to acknowledge and thank the Army Research Laboratory, and Victor Champagne in particular, for the continued support throughout this research project. I would not have been able to pursue my Ph.D. if not for this support. I would also like to sincerely thank my advisor, Prof. Richard Sisson, for his guidance dating back to my undergraduate career when I first decided that materials science and engineering was the field I wanted to pursue. I would also like to thank the other members of my committee, Prof. Jianyu Liang, Prof. Satya Shivkumar, and Dr. Timothy Eden for their support and guidance. I have to thank Dr. Danielle Cote for not only being a member of my thesis committee, but for all of the support for the last 5 years. In my senior year at WPI, Danielle was my Major Qualifying Project advisor, and that experience helped cultivate my love for research. Thanks goes to the MTE department and MPI faculty and staff, especially Rita Shilansky. I also would like to thank the lab manager Dr. Li for all his help learning how to use the tools I have grown to love. I must thank members of my research group, both past and present: Aaron Birt, Luke Bassett, Rachel White, Caitlin Walde, Derek Tsaknopolous, Rob DelSignore, Kyle Fitzpatrick, and Matt Gleason. I also thank my fellow graduate students in the department. Lastly, I have to thank my family. My parents, Dan and Donna, first and foremost for their support throughout this adventure. My sister, Brittany, for blazing the trail for my career at WPI and my brother, Taylor, for always backing me up. Finally, to my boyfriend, Brandon, thank you for your constant support while I worked towards this ultimate goal.

2 TABLE OF CONTENTS Abstract ...... 1 Acknowledgements ...... 2 1.0 introduction & motivation ...... 5 1.1 Background ...... 5 1.2 Current Project Needs ...... 7 1.3 Approach ...... 8 1.3.1 Modeling ...... 8 1.3.2 Experimental Characterization ...... 8 1.3.3 Alloys Studied ...... 9 2.0 Literature review ...... 11 2.1 Powder Characteristics for Cold Spray/Additive Manufacturing ...... 11 2.2 Microstructures of Al 6061 and 5056 ...... 12 2.2.1 Aluminum 6061 ...... 12 2.2.2 Aluminum 5056 ...... 13 2.3 Thermal Processing of Aluminum Alloys ...... 14 2.3.1 Degassing of Aluminum Powders ...... 15 2.3.2 Microstructural Effects of Heat Treatment ...... 16 3.0 Through-Process Model for cold spray ...... 19 3.1 Powder Production Stage ...... 19 3.1.1 Al 6061 & 5056 Solidification Model ...... 19 3.1.2 Experimental Grain Size ...... 20 3.1.3 Al 6061 Thermodynamic and Kinetic Modeling ...... 23 3.1.4 Al 6061 Experimental Phase Identification ...... 26 3.1.5 Al 5056 Thermodynamic and Kinetic Modeling ...... 31 3.1.6 Al 5056 Experimental Phase Identification ...... 34 3.1.7 Experimental Verification of the Powder Production Stage Models ...... 37 3.2 Powder Processing Stage ...... 39 3.2.1 Al 6061 Heat Treatment Modeling ...... 39 3.2.2 Al 6061 Experimental Comparison ...... 40 3.2.3 Al 5056 Heat Treatment Modeling ...... 45 3.2.4 Al 5056 Experimental Comparison ...... 46 3.3 Additive Strength Model ...... 50 3.3.1 Model Description ...... 50

3 3.3.2 Experimental Comparison ...... 51 3.4 Particle Impact Model ...... 56 3.4.1 Model Description ...... 56 3.4.2 Experimental Comparison ...... 56 4.0 Recommendations for Future work ...... 64 4.1 Powder Processing Stage ...... 64 4.1.1 Model Development ...... 64 4.1.2 Powder Processing Experimental Verification ...... 64 4.2 Particle Impact Model ...... 64 4.2.1 Model Verification and Advancement ...... 64 5.0 Conclusions ...... 65 6.0 References ...... 67 INDEX ...... 69 List of Figures ...... 69 List of Tables ...... 71 Appendix ...... 73 List of Publications ...... 73 Conference Presentations ...... 73

4 1.0 INTRODUCTION & MOTIVATION 1.1 Background Aluminum alloys are widely used in many different industries, especially aerospace and automotive. One of the most important uses for aluminum alloys is for light-weighting in these industries. For example, steel components were replaced by a high-strength aluminum in the Ford Motor Co. F-150 truck. With over 6 years and $1 billion spent in development, the truck was debuted in 2015 at 700 pounds lighter than the previous year’s model. This light-weighting helps the automotive industry keep in compliance with emission and fuel-economy standards [1]. There are many other applications for aluminum alloys, and recently, using aluminum in additive manufacturing techniques has become very popular.

The work of this thesis focuses on an application for the U.S. military. Currently, the U.S. Army spends over $800,000 dollars to replace each corroded magnesium gearbox housing on their rotorcraft. With around 125 gearboxes failing each year, over $100 million annually is spent on these replacement parts. The U.S. Army Research Laboratory (ARL) is using the cold spray process, an additive manufacturing technique, to repair these parts. The corroded material is removed, cold spray aluminum is applied, and the part is machined to its original dimensions and put back into service [2].

A schematic of the cold spray process is shown in Figure 1. During this process, metallic powders are introduced into a heated gas, typically helium or nitrogen, and are carried through the de Laval nozzle. In the converging/diverging nozzle, the gas expands and the powders are brought to supersonic velocities. When the powder impacts the substrate, the metal plastically deforms and adheres. This process can create a thin coating or a near-net shape part.

Powder Feeder Cold sprayed Pre- material de Laval Substrate Chamber Nozzle

material Gas

Y sprayed Cold Heater

Gas Z X Figure 1: Schematic of the cold spray process [3]. The benefits of the cold spray process, and what makes it unique as an additive manufacturing technique is that it is a low temperature (100o – 500oC) and portable process. This means that parts can be repaired in-situ, which would reduce repair time and ultimately save money on inventory, lead time and labor costs. As for material properties, because cold spray is a completely solid-state deposition process, meaning the powder never reaches its melting temperature, thermal stresses and unfavorable tensile stresses from other additive techniques are avoided. In fact, the plastic deformation causes beneficial compressive residual stresses in the part.

Because of the great success using cold spray as an additive manufacturing technique, major modeling efforts are being conducted to help advance this fairly novel process. This is in alignment with the Materials Genome Initiative, issued by former U.S. President Obama in 2011 [4]. Computational modeling

5 is replacing trial-and-error operations and will shorten the new material development lifetime from 10- 20 years to just two to three years.

A research group at Worcester Polytechnic Institute is developing an all-encompassing through-process computational model that will use the initial properties of the powder to predict the final microstructure and mechanical properties of the cold sprayed deposit [5]. A schematic of the model is shown in Figure 2.

Figure 2: Schematic of through-process modeling for cold spray. The model includes all stages of the cold spray process, starting at the production of the powder. This stage involves gas atomization and utilizes thermodynamic and solidification models. The powder processing stage includes any preparation that is done to the powder before it is sprayed (degassing, milling, blending, etc.). This stage utilizes kinetic and diffusion-based modeling. The process parameters stage is being developed at ARL and includes parameters such as gas composition, velocity and temperature. The particle impact stage, developed at the Applied Research Laboratory at The Pennsylvania State University, is a finite element model of the plastic deformation during particle impact with the substrate. The post-processing stage includes any machining, heat treating, or forming that would be applied to the cold sprayed deposit. This final stage has yet to be developed.

In order to integrate the initial properties of the powder into the particle impact model, an additive strength model was developed that models the yield strength of the powder as a function of particle diameter, and time and temperature of any powder thermal processing. This equation is shown in Equation 1.

풎 풏 풑 흈풀푺(풅, 풕, 푻) = 흈풐 + ∆흈풔풔(풅, 풕, 푻) + ∆흈풎풊풄(풅, 풕, 푻) + ∆흈풑풑풕(풅, 풕, 푻) Equation 1

The strengthening contributions include the intrinsic strength of the material, 휎표, and terms for solid solution strengthening, ∆휎푠푠 , microstructural influence, ∆휎푚𝑖푐 , and precipitation and dispersion hardening, ∆휎푝푝푡. The superscripts, m, n and p, are included because the additive strength equation is a

6 non-linear model and are needed to calibrate the model to experimental results. The terms in the equation will be described in more detail in Section 3.3, the Additive Strength Model Section.

This predictive through-process model to be used by the U.S. Army and by industry, will help rapidly develop cold spray material design and selection processes, and ultimately make cold spray a more versatile process that can be used on numerous applications [6]. 1.2 Current Project Needs An important aspect of any model is the experimental data that verifies the predictive results of the model. Experimental characterization has been compared with model results of the powder production and particle impact stages of the through-process model and will be summarized in this dissertation. Current and future work focuses on the powder processing stage, thermal processing in particular. A current set-back in the development of these alloys is the low ductility. One way to improve ductility is to thermally process the powder before it is sprayed. This is done for two reasons: brittle hydroxides are removed from the surface of the powder, which is a typical degassing step for powder processing, and secondly, the microstructure is modified from the as-atomized condition to a more “sprayable” condition. These effects will be discussed in more detail in the Literature Review Section of this dissertation.

A bubble diagram created in GRANTA CES EduPack software shown in Figure 3, illustrates the range of tensile and elongation values for several aluminum alloys with different thermal treatments from the GRANTA database. The common T6 strengthening heat treatment is labeled for the heat treatable alloys. Also included in this diagram are initial tensile and elongation results of cold sprayed Al 6061 sprayed with heat treated powder, as indicated by the times and temperatures. After heat treating the powder at temperatures ranging from 200 to 260°C for twenty-four hours, the tensile strength decreases, but the ductility increases by a maximum of 60%, moving closer to the values of a T6 treatment of Al 6061.

Figure 3: Bubble diagram of tensile strength versus elongation for several aluminum alloys studied, including manual records of cold sprayed material. Diagram created in GRANTA CES EduPack.

7 The need to improve ductility in these Al 6061 samples has led to this research. The goals of this thesis work are to first fully characterize aluminum 6061 and 5056 powders as it pertains to the through-process model, and then investigate the effects of heat treatment on the microstructure and phases. 1.3 Approach 1.3.1 Modeling In the powder production and powder processing stages, solidification, thermodynamic and kinetic models were employed. The analytical solidification model uses properties from JMatPro® (Sente Software Ltd., United Kingdom). The thermodynamic models use the commercially available software Thermo-Calc (Thermo-Calc, Sweden). The kinetic models use JMatPro® and TC-PRISMA (Thermo-Calc, Sweden) to model the effects of the powder heat treatment. 1.3.2 Experimental Characterization Because modeling the properties of the powders is a function of particle diameter, the powder was sieved before characterization was performed. Firstly, fine particles less than approximately 20 microns were removed by sending nitrogen through a modified fluidized bed where the fine particles rose to the top and were ejected from the stream. This ensured the mesh of the sieves would not be clogged by fine particles. Secondly, the powder was sorted through stainless steel meshes in a mechanical sieve. The sizes of sieved particles ranged between 25-32, 32-38, 38-45, 45-53, and 53-63 m in diameter. An example of the particle size distribution of sieved Al 6061 powder was done by laser diffraction at United Technologies Research Center and is shown in Figure 4.

25

20

15 25-32 microns 32-38 microns

10 38-45 microns

45-53 microns Volume Volume Percent (%) 53-63 microns 5

0 0 10 20 30 40 50 60 70 80 90 100 Diameter (microns)

Figure 4: Particle size distribution of sieved Al 6061 powder, conducted by laser diffraction. For studying the heat-treated powders, small samples of powder (approximately 30 mg) were heat treated in a TA Instruments Q20 differential scanning calorimeter (DSC). Heat treatment times and temperatures, as well as heating rates, are stated for each sample. The DSC scan data was collected with a heating rate of 5oC/min and scanned from 20oC to 530oC. The atmosphere was nitrogen with a flowrate of

8 50mL/min. A Tzero Al pan and lid were used, with an empty Tzero Al pan and lid for the reference sample. The powder was removed from the DSC immediately after the allotted time and air-quenched.

Samples were cold mounted in epoxy, ground with 800-grit silicon carbide paper, and mechanically polished on a Struers Tegramin-20 auto-polisher to a final silica polish. For microstructural examination, Al 6061 powders were etched with a 0.5% hydrofluoric acid solution for 60 seconds, while the Al 5056 powders were etched with a variation of Keller’s reagent with the addition of nitric acid for 10 seconds.

Microstructural examination was done with a NIKON Epiphot 200 light optical microscope, and a JEOL JSM-7000F Field Emission scanning electron microscopy (FESEM) with an Oxford X-MAXN silicon drift detector for energy dispersive spectrometry (EDS). For further examination, a TEM lamellae was cross sectioned and prepared by a Helios NanoLab G3 UC DualBeam scanning electron microscope/focused ion beam (SEM/FIB) at the Institute of Materials Science (IMS) at the University of Connecticut. The lamellae were imaged in an FEI Talos F200X scanning/transmission electron microscope (STEM) also at IMS. Copper was not included in EDS analysis due to the sample being mounted on a copper grid. This would cause more copper to appear in the EDS map than what is actually present in the material. Transmission electron backscatter diffraction (t-EBSD) was conducted in a JEOL JSM-7000F field emission SEM at a voltage of 30V.

Nanohardness of the powders was determined by a Keysight Nano Indenter G200. Individual nanoindentations were prescribed to a depth of 250 nm. At least 30 indents were performed and the average and standard deviations are reported. The displacement resolution of this nanoindenter is <0.01 nm.

Phase fraction and phase measurements were performed in image analysis software Olympus Stream Essentials (Olympus Corporation, Japan). Automatic thresholds were employed. Area fraction results from image analysis was converted to volume fraction using a method described by Corti, and is further explained in Section 3.1.7, the Experimental Validation of the Powder Production Stage Models section [7]. 1.3.3 Alloys Studied The Al 6061 powder in this study was supplied by F.J. Bromann & Co. (Harvey, LA), and the Al 5056 powder was supplied by Valimet, Inc. (Stockton, CA). The compositions of the alloys used in this study are shown in Table 1, and are within specification limits [8, 9]. The composition was determined by direct current plasma emission spectroscopy (ASTM E 1097-12).

9 Table 1: Experimental compositions and specifications of Al 6061 and 5056 used in this study in weight percent [8, 9].

Al 6061 Al 6061 Al 5056 Al 5056 Spec.[8] Spec.[9] Cr 0.11 0.04-0.35 0.16 0.05-0.20 Cu 0.26 0.15-0.40 0.006 0.0-0.1 Fe 0.28 0.0-0.7 0.11 0.0-0.5 Mg 0.922 0.8-1.2 5.3 4.5-5.6 Mn 0.078 0.0-0.15 0.15 0.05-0.20 Ti 0.024 0.0-0.15 0 0 Zn 0.02 0.0-0.25 0.006 0.0-0.10 Si 0.591 0.40-0.80 0.05 0.0-0.3 Al 97.7 Rem. 94.2 Rem.

10 2.0 LITERATURE REVIEW 2.1 Powder Characteristics for Cold Spray/Additive Manufacturing Traditional additive manufacturing techniques, such as selective laser sintering (SLS), 3-D printing, electron beam melting (EBM) and laser engineered net shaping (LENS), highly depend on variables such as size distribution, morphology and flow of powders used in their processes. The sphericity of the powders will affect the flow of the powder through the powder feeder [10]. These variables matter in LENS in particular because a certain size range of powder, 36-150 microns in diameter, must be continuously injected onto the powder bed in a controlled fashion as the laser passes over it [11].

In contrast, the size distribution and morphology matter less for the cold spray process. Small particles, particularly less than 20 microns, do not reach the critical velocity at the substrate needed to plastically deform and adhere because of the bow shock effect on the substrate surface [12, 13]. Particles too large will not reach the critical velocity as well based on their mass. While some precautions on size distribution are made to prohibit clogging in the nozzle, it is not the main concern.

The most important feature of powder for cold spray is the microstructure. As cold spray is a completely solid-state process, the initial properties and phases in the microstructure of the aluminum powder greatly influence the final properties of the cold sprayed material. The phases that are present in the powder remain in the cold sprayed material. This is evident in Figure 5 and Figure 6. In the XRD pattern of Figure 5, there is no major phase transformation that occurs during the cold spray process. The same two phases indicated by the arrows in the powder sample of Figure 6a are also found in the corresponding cold spray sample in Figure 6b.

10000 Al Powder (111) Consolidated 8000

6000 Al (200) 4000 Al Al

Intensity (cps) Intensity (220) (311) 2000 Al Al (222) (400) 0 20 30 40 50 60 70 80 90 100 2θ Figure 5: XRD pattern for Al 6061 powder and consolidated material.

11 a) b)

100 nm 200 nm

Figure 6: STEM images of a) Al 6061 powder with a scale bar of 100 nm and b) Al 6061 cold spray with a scale bar of 200 nm. The white arrows indicate the dark phase in both images and the black arrow indicates the white phase. 2.2 Microstructures of Al 6061 and 5056 There is little to no literature data on the microstructural features and phases of Al 6061 and 5056 powder. This section will include information about the phases and precipitation sequences in these two alloys, mostly taken from data about their wrought or cast counterparts. It will provide a starting point for comparison with the experimental characterization of the powders. 2.2.1 Aluminum 6061 The 6xxx series aluminum alloys are primarily Al-Mg-Si systems and are further strengthened by additions of Cu, Mn, and Zn. They have good formability, machinability, and corrosion resistance but have lower strength than 2xxx and 7xxx series alloys. As an age-hardenable alloy, the primary strengthening phase is magnesium silicide, Mg2Si. This phase has a face-centered cubic structure, and its morphology is either fine Chinese script or dispersed particles. Other phases found in this alloy include AlFeSi and Al(FeSi), with the latter being a monoclinic phase that presents itself as needles [14]. Phases found in Al 6061 powder are most easily identified through TEM, as the most abundant phase, Mg2Si is easily eroded during prolonged polishing [15].

If there is sufficient Cu in the alloy, a quaternary phase Q (Al-Mg-Si-Cu) can form. It is not a well-recognized phase but is said to be found in large volume fractions by Chakrabarti [16]. This phase can be found with an intertwined structure, along with Mg2Si at the grain boundaries, as shown in Figure 7.

12

Figure 7: Optical micrograph of an ingot of Al 6061 showing the Q phase and Mg2Si [16]. The precipitation kinetics for Al 6061 are well studied and a DSC curve is shown in Figure 8 [17]. This curve shows what temperatures the metastable  phases will dissolve and precipitate. This information is helpful for the heat treatment studies and will be compared to an experimental DSC curve.

Figure 8: DSC curve for Al 6061 [17].

2.2.2 Aluminum 5056 There is a limited amount of literature data on the microstructure of Al 5056. The information presented here is for a direct chill semi-continuously cast ingot of Al 5056 and is thus a suggestion for what will be found in the powder Al 5056 [18]. It is an Al-Mg system that is a strong, work hardening and solid-solution strengthening alloy, with good corrosion resistance. The addition of Ti is as a grain refiner. Other alloying elements, like Fe and Si, are impurities and will form coarse intermetallics at the grain boundaries. Mn

13 and Cr are added for further strengthening and grain refinement. The intermetallics that might be present include Al7(Cr,Fe), Al6(Fe,Mn), Al10(Mg,Mn)3, or Al12Mg3Mn2. These intermetallics make up about 2% of the area fraction, have an irregular morphology, and vary in size [19].

It is difficult to precipitate Al3Mg2 in this alloy, but if present, would exist as rounded particles at the grain boundaries with an FCC structure [14]. These can form during natural or artificial aging. Authors do not agree on the nucleation of these precipitates; it has been said they nucleate on dislocation loops or on tetrahedron shaped voids. Zhu describes the artificial aging process and discusses the nucleation. ’ and  can precipitate directly from the matrix above 100oC. It will also precipitate at triple points, Al6Mn boundaries, and grain boundaries if there are enough dislocations to accommodate the semicoherency [18]

The addition of alloying elements will change the precipitation of Al3Mg2. The addition of Zn will accelerate the rate of precipitation within the grains, thus reducing the amount of this phase along the grain boundary area. Also, an AlMg(Zn,Cu) phase may form instead of Al3Mg2. With the addition of Cu, an AlMgCu phase may form instead, with a more homogeneous distribution in the matrix[18]. Al 5056 contains all of these alloying elements, so it is unlikely that the Al3Mg2 phase will exist.

The grain structure of the ingot contained equiaxed grains with serrated grain boundaries, which was typical of a cast structure. The average grain size for this sample was reported to be 144 microns. Spherical and long plate-like or needle-like intermetallics were found in the microstructure, as shown in Figure 9. There were also precipitate free zones around the grain boundaries and intermetallic particles [19].

Figure 9: Optical micrograph of Al 5056 alloy etched with Barker's reagent and viewed under polarized light [19].

2.3 Thermal Processing of Aluminum Alloys The reason that aluminum powder is thermally processed is two-fold: to degas the powder, and to change the microstructure. It is a fairly common industry practice to degas powder before it is used in various

14 powder metallurgy processes. As for changing the microstructure, most heat treatable aluminum alloys, including the 6000 series, undergo a three step strengthening heat treatment: solutionize, quench, and age harden. This process is outlined in the schematic in Figure 10. The important variables that will affect the results of the heat treatment include the heating rate, solutionizing temperature and time, the quench rate, and the aging temperature and time. These variables will be considered in this project. This section will focus on the effects of degassing and heat treating aluminum alloys 6061 and 5056.

Figure 10: Schematic of typical heat treatment steps for aluminum alloys, indicating important variables. 2.3.1 Degassing of Aluminum Powders A degassing procedure is a common practice in powder metallurgy processes. Aluminum powder is brought to elevated temperatures under a vacuum to remove physiosorbed and chemisorbed hydrogen and oxygen [20]. Harrell, et. al., describes degassing as an imperative step with hot isostatic pressing (HIP) aluminum powder. If this step is not included in the process, defects may be present in the compacted material.

Flumerfelt goes into detail about what occurs during a degassing step, describing the oxide film on aluminum powder as a duplex structure of alumina (Al2O3) and alumina trihydrate (Al2O33H2O) with physiosorbed water present on surface of the film [21]. The degassing step used in this experiment was to remove the gaseous hydrogen and water vapor from the film. The hydrogen is present because it is a product of the physiosorbed or chemisorbed water that reacts with the aluminum cations that can diffuse through the oxide film, which is illustrated by the reactions in Equations 2, 3 and 4 [21].

2퐴푙 + 퐻2푂 → 퐴푙2푂3 + 3퐻2 Equation 2

2퐴푙 + 6퐻2푂 → 퐴푙2푂3 ∙ 3퐻2푂 + 3퐻2 Equation 3

2퐴푙 + 4퐻2푂 → 퐴푙2푂3 ∙ 퐻2푂 + 3퐻2 Equation 4

The water vapor is removed by a dehydration sequence. The oxide film is hydrated as Al2O3·3H2O at o temperatures below 150 C. Between 150C and 175 C, the film is dehydrated to Al2O3·2.5H2O, Al2O3·H2O o o between 175-310 C and finally is dehydrated fully to Al2O3 at temperatures between 310-500 C. Some precautions for a degassing step include ensuring the temperature is below the solidus of the alloy, as well as keeping the time at that elevated temperature limited to minimize grain coarsening [21]. Other microstructural effects during a degassing heat treatment are discussed in the next section.

15 2.3.2 Microstructural Effects of Heat Treatment There are two main microstructural effects when applying a heat treatment: effect on grain size and phases present. Firstly, holding an alloy at elevated temperatures for considerable lengths of time will cause grain coarsening. Grain growth is an important consideration when heat treating any metal because of the Hall-Petch relationship, where the strength of the alloy is inversely proportional to the square root of the grain size, as given in Equation 5.

풌풚 흈풚 = 흈풊 + Equation 5 √풅

휎푦 is the yield strength of the material, 휎𝑖 is the intrinsic yield strength, ky is the Hall-Petch constant, and d is the grain size. Intuitively, a small grain size will yield a stronger material, thus grain coarsening should be avoided.

The second consideration when heat treating an aluminum alloy is the phase transformations that will occur at high temperatures. Two of the strengthening mechanisms in aluminum alloys are solid solution strengthening and precipitation hardening. During a solutionizing heat treatment step, alloying elements will completely dissolve into the aluminum matrix, causing a supersaturated solid solution (SSS) to form. This causes local distortion in the aluminum lattice and thus will increase the strength of the material by obstructing dislocation motion. An appropriate solutionizing temperature is 5oC below the solidus temperature. The alloy is then quenched to room temperature in order to preserve the high temperature composition of the SSS [22].

Some alloys are used in the solutionized condition with only solid solution strengthening, and others are furthered strengthened by precipitation hardening, or aging. A natural aging treatment is simply at room temperature, while an artificial aging step is at intermediate temperatures, below 200oC. During aging, fine particles are precipitated from the SSS. Guinier-Preston (GP) zones, or clusters, are the first to form and involve segregation of solute atoms on specific crystal planes. The zones nucleate homogeneously and are uniformly distributed in the alloy. Elastic stresses and strains are present around the zones and thus harden the alloy. After being held for longer times, coherent and semicoherent metastable precipitates will form in the alloy, which provide the largest strength contribution because of the large elastic distortions at the matrix and phase interfaces. These usually nucleate on grain boundaries or dislocations. For even longer aging times, these precipitates will coarsen, lose their coherency, and transform to the stable equilibrium precipitate. The main parameters of precipitation strengthening are the chemical composition of the SSS, composition of the aging products, kinetics of precipitation, temperature and time of heat treatment, and the properties of the precipitated phases such as shape, composition and time-temperature stability [22].

It is difficult to simply apply heat treatment schedules from literature to the powder for cold spray, as it is not certain that powders will respond to heat treatments in the same fashion as rods, sheets, plates, etc. Currently, there is not a specific heat treatment schedule for powders. This project will take into account the common heat treatments for the alloys, and use those as a starting point for further development.

Aluminum Alloy 6061 The heat treatment schedules, precipitation kinetics, and modeling for Al 6061 are very well studied [17, 23-29]. The most common heat treatment for precipitation strengthening Al 6061 is to solutionize at

16 530°C, quench, and artificially age between 160 and 175°C for 8-18 hours, depending on the fabrication method of the part (rolled, extruded, drawn, etc.) [30]. As for the aging step, the amount of time at the aging temperature determines the strength and ductility of the alloy, as shown by Figure 11 [31]. For each aging temperature, there is a peak hardness, where the metastable, coherent or semi-coherent strengthening precipitates have a large contribution to strengthening. At longer times, the strength decreases as the precipitates coarsen and become incoherent.

Figure 11: Al 6061 aging curves showing affect of heat treatment time and temperature on yield strength [31].

The main strengthening precipitate in Al 6061 is the  phase, Mg2Si. Because Mg and Si are fairly soluble in Al, they will completely dissolve into the supersaturated solid solution (SSS) after the solutionizing step. During aging, nucleation and growth begin and Mg2Si forms through the precipitation sequence shown in Figure 12. The sequence starts with alpha, the matrix phase, which is a supersaturated solid solution (SSS). During nucleation, GP zones will form, which are clusters of Mg and Si. ” forms, and is the peak strengthening phase. At longer hold times, another metastable phase ’ forms, followed by , the equilibrium incoherent phase.

 (SSS) GP Zones " '  (Mg2Si)

Figure 12: Precipitation sequence of Mg2Si in Al 6061 alloys. It has not yet been determined which condition of Al 6061 is the most suitable for cold spaying: the underaged condition with GP zones, the peak-aged condition with ”, or the overaged condition with . This dissertation will help present the experimental effects that will determine the solution. Aluminum Alloy 5056 As Al 5056 is not a heat treatable alloy, there are no prescribed heat treatments. This dissertation will focus on homogenizing the microstructure, as there has been initial success in cold spraying the powder in this condition. An as-atomized powder can be considered as a cast structure where the solidification rate has caused solutes to segregate to interdendritic regions. For many applications, a homogenous solid solution is a more desirable microstructure. Sheppard and Raghunathan conducted several different homogenization treatments on Al 5056 samples to study the effects on the microstructure [19]. The goal of the homogenization treatment is to put the segregated solute atoms into solution. The homogenizing

17 temperatures ranged from 450° to 570°C and their tests included single 24 soak times, and multi-step times of 24 hours followed by 8 hours at a lower temperature. The high temperature treatments resulted in course constituent particles in the microstructure. As for the hardness, it initially decreases rapidly at all temperatures, becomes constant for some time, and slightly increases for extended hold times. The structure was unchanged. Because of the dissolution of the eutectics in the interdendritic regions, the dendritic structure was not observed after homogenization. Another observation was high temperature treatments resulted in spheroidizing the grain boundary intermetallics [19].

These types of observations from the literature will be useful when conducting similar tests on Al 5056 powder. It is hypothesized that homogenization times for the powder will be less than what is seen in literature, based on the fact that the grain sizes are much smaller than the cast materials that were studied. Therefore, there is less distance for the atoms to diffuse.

18 3.0 THROUGH-PROCESS MODEL FOR COLD SPRAY This section will briefly describe the models for each of the stages of the through-process model. Previous work includes a detailed description of the modeling effort, which was not the focus of this dissertation. [3]. The results of the models are compared to experimental work. 3.1 Powder Production Stage This stage of the through-process model includes models that are used to predict both the grain size and the phases that are present upon gas-atomization of the powder particles. 3.1.1 Al 6061 & 5056 Solidification Model A solidification model that will be useful in the through-process model will predict the grain size as a function of particle size. During solidification of an aluminum particle, the matrix phase is the first to form and alloying elements are segregated to the grain boundaries. This is illustrated by the SEM image in Figure 13. A backscatter SEM images shows an Al 7075 powder particle with the segregation areas illuminated at individual grain boundaries. The size of these grains is an important determination in the strength of the powder, as illustrated by one of the contribution terms in Equation 1.

Figure 13: Backscatter SEM image of segregation and individual grains in an as-atomized Al 7075 powder particle. A cooling rate model and grain size relationship are used to create this model. A simplified droplet cooling rate model is used for gas-atomized powder [3]. A reduced heat balance, including several thermophysical properties of the atomizing gas and the particles, is given in Equation 6. The properties for modeling the cooling rate Al 6061 in argon gas is given in Table 2.

푑푇푑 12 푘푔 | | = (푇푑 − 푇푔) 2 Equation 6 푑푡 휌퐶푝 푑

3 The properties of the droplet include dTd/dt, the cooling rate [°C/s], , the density [kg/m ], Cp, the specific heat [J/(kgK)], Td, the droplet temperature [K], and d, the droplet diameter [m]. Tg is the gas atomizing temperature [K], and kg is the thermal conductivity of the gas [W/(mK)].

19 Table 2: Thermophysical properties of atomizing Al 6061 in argon gas (from JMatPro® and CES EduPack software).

Thermophysical Properties of Al 6061 Ar Atomizing Gas Molten Droplet

Tg Kg  Cp Td [K] [W/(mK)] [kg/m3] [J/(kgK)] [K] 300 1.79E-02 2380 1170 1473

A relationship commonly used in castings between secondary dendrite arm spacing and cooling rate is given in Equation 7. Because atomized powders are small castings, this relationship can be applied. −푛 푑푇푑 휆 = 휆 ( ) Equation 7 2 0 푑푡

2 is the secondary dendrite arm spacing [m], 0 [K/s][n] and n [dimensionless] are alloy system- dependent constants, T is the droplet temperature [C°], t is time [s], and dTd/dt is the cooling rate of the droplet [°C/s]. Using model data from JMatPro®, the constants 0 and n were found to be 100.99 and 0.33, respectively, for Al 6061.

If Equations 6 and 7 are combined, a useful relationship between grain size and particle size can be established and modeled for any alloy using different thermophysical properties. The relationship is given in Equation 8. −푛 12 푘푔 휆 = 휆0 ( (푇푑 − 푇푓) 2) Equation 8 휌퐶푝 푑 The results of this model will be compared with the experimental results in the next section. 3.1.2 Experimental Grain Size Powder particle cross sections were polished, etched and viewed under SEM and optical light microscopy. Al 6061 was etched with a with 0.5% hydrofluoric acid solution for 90 seconds and Al 5056 powder was etched with a variation of Keller’s reagent with the addition of nitric acid for 10 seconds. Images of the grains within each powder particle is shown in Figure 14.

Figure 14: SEM images of cross sections of Al 6061 powder etched with 0.5% hydrofluoric acid solution for 90 seconds (left) and Al 5056 powder etched with a variation of Keller’s reagent with the addition of nitric acid for 10 seconds (right).

20 Grains were measured from SEM and optical microscopy images using ASTM E112 – 13 intercept method. The grain size of different size particles was measured and are represented in Figure 15 by the diamonds. The dotted line represents a trendline from the experimental data while the solid line represents the model data from Equation 8. The hollow square points along the x-axis represent featureless grain morphology in small particles. This either means they solidified quickly enough to create an amorphous particle, or they are nanostructured and the grains were not perceived through SEM. 4.00 Al 6061 3.50 Al 6061 Experimental

m] 3.00  Al 5056 2.50 Al 5056 Experimental 2.00 1.50

1.00 Feature Size [ Size Feature 0.50 0.00 0 10 20 30 40 50 60 Particle Diameter [m]

Figure 15: Model and experimental comparison of relationship between grain size and particle diameter for AL 6061 and 5056. There is good agreement between experimental grain size and particle diameter. This comparison was also done for two other alloys systems not included in this thesis (Al 2024 and 7075) with similar results. It is revealed that smaller particles contain smaller grains, while larger particles contain larger grams. This is intuitive because the larger particles are cooling at a slower rate, while the small ones cool very quickly during gas atomization. This is also shown in Figure 16, where SEM images of Al 6061 powders are plotted against their calculated cooling rate showing the dependence on diameter.

21

Figure 16: SEM images of different sized Al 6061 powder particles plotted against calculated cooling rate [°C/s]. A preliminary electron backscatter (EBSD) study was conducted on as-received Al 6061 powder, to give more information about the grain structure in the powders. It is not clear in the literature if common powder microstructures contain dendrites, a cellular structure, or grains. As shown in the inverse pole figure of Figure 17, there are larger grains that seem to contain smaller sub-grains with the same orientation. There are other individual grains with different orientations that are not contained within a larger grain. For this thesis, the term “grain” will be used for the sub-grains. More EBSD on another and/or larger sample is necessary, as Figure 17 shows a small sample size of a single powder particle. Also, sample preparation is of utmost importance. The sample in Figure 17 was thinned by FIB, however a lower voltage for the final cleaning procedure needs to be used to reduce the beam damage.

22

Figure 17: EBSD analysis of Al 6061 as-received. (Left) Band contrast image of scanned area, and (Right) inverse pole figure color map. 3.1.3 Al 6061 Thermodynamic and Kinetic Modeling The thermodynamic model is important because it predicts what phases are going to be present in the alloy. This information is required as in input into the additive strength model, Equation 1. The precipitation term takes into account what phases are present, their size, and the amount, in order to calculate the amount of strengthening they contribute. This type of model is also useful because the composition of the alloy can be tailored to either minimize detrimental phases or to optimize the amount of the strengthening phases.

Using Thermo-Calc software, and the composition of Al 6061 given in Table 3, the equilibrium phase fractions as a function of temperature were predicted, as shown in Figure 18. Numerical values for phase fraction in weight and volume percent of the predicted phases at room temperature [25°] is given in Table 3. It was found that the most abundant secondary phase in Al 6061 is Mg2Si, with other iron-containing intermetallics present.

2.0

1.8 Liquid

1.5 Al 1.3

1.0 Alpha 0.8 Mg2Si Phase Fraction [wt %] [wt Fraction Phase 0.5

0.3 Al7Cu2Fe 0.0 -100 100 300 500 700 Temperature [°C] Figure 18: Equilibrium phase predictions for Al 6061 as a function of temperature.

23 Table 3: Weight percentages of equilibrium phases predicted for Al 6061 at room temperature [25°C].

Equilibrium Phases Phase Wt% Vol% FCC Al 96.55 96.52

Mg2Si 1.16 1.57 T (AlCuMgZn) 0.85 0.77

Al15Fe3Si2 0.73 0.58

Al45Cr7 0.38 0.32

Al13(Fe,Mn)4 0.27 0.19

Al3Ti 0.06 0.05

However, these equilibrium models must be used cautiously. The very fast cooling rates during gas atomization (approximately 104 – 105 °C/s) do not allow equilibrium to be reached. They are useful in that they are a guideline for the types of phases that could be present. A continuous cooling curve, a type of kinetic model, illustrates why the equilibrium phases may not form during gas atomization. The curve shown in Figure 19 includes cooling curves that are lower than that of gas atomization, i.e. the cooling curve for gas atomization is further to the left of the graph. The other curves represent different amounts of the formation of the stable Mg2Si phase. Because the cooling curve for gas atomization does not intersect any of these Mg2Si curves, it is hypothesized that the stable Mg2Si phase does not form during gas atomization. This will be investigated through TEM.

600 100 C/s

500 1000 C/s

10000 C/s C]

° 400 0.001% 300 0.01%

200 0.05% Temperature Temperature [ 0.10% 100 0.50% 0 1.00% 0.0000001 0.00001 0.001 0.1 10 1000 Time [hr] 1.25%

Figure 19: Continuous cooling curves for Mg2Si phase in Al 6061. The gray curves on the left represent cooling rates. As mentioned, thermodynamic equilibrium models can give a starting point for the types of precipitates that should be found in the aluminum alloys. Two types of non-equilibrium solidification models that are useful are a segregation model and a Scheil calculation. The inputs for the segregation model are the composition of the alloy, grain size and cooling rate. An experimental grain size of 2.15 microns was used with a calculated cooling rate of 1.16x104 oC/s, from Equation 7, for Al 6061 and the segregation results are shown in Figure 20. The composition at the grain boundary, which was rich in copper, silicon, and magnesium, was used as an input into Thermo-Calc to predict what phases would form at the grain

24 boundary based on what alloying elements were segregated to it. The weight and volume percent of the phases are shown in Table 4. This type of model may be more accurate since it considers the fast cooling rate during gas atomization, where the equilibrium calculator does not.

Figure 20: Segregation model output for Al 6061 (JMatPro®).

Table 4: Phases predicted to form at the grain boundary of Al 6061 based on the grain boundary composition predicted by the segregation model in Figure 20 (Thermo-Calc).

Phases at Grain Boundary Phase Wt% Vol % FCC Al 96.5 63.00

Al2Cu 1.41 15.55 Q (AlCuMgSi) 0.76 12.91 Si (Diamond) 0.65 7.97

Al15Si2(Fe,Mn)4 0.23 0.34

Al15Fe3Si2 0.17 0.24

In order to further understand the solidification phenomenon during atomization, a Scheil calculation is helpful. The input for this type of predictive model is the composition of the alloy. Figure 21 shows this plot for Al 6061, calculated in Thermo-Calc, using the composition in Table 1. It is shown that the FCC matrix phase is the first to form upon solidification, followed by the iron and silicon containing phases, Al13Fe4, Al8Fe2Si and Al15Fe4Si2. As the temperature decreases, Mg2Si, the main strengthening precipitate in Al 6061 forms. Following this, two more phases containing iron, magnesium and silicon form, Al9Fe2Si2 and Al18Fe2Mg7Si20. Towards the end of the solidification event, the remaining silicon precipitates in its diamond structure. These types of phases predicted by this plot will be identified experimentally in STEM.

25

Figure 21: Scheil solidification plot for Al 6061. 3.1.4 Al 6061 Experimental Phase Identification STEM has proven to be a beneficial tool in confirming identities and compositions of precipitates at the grain boundaries to compare with the equilibrium and non-equilibrium predictions. Figure 22a shows small precipitates at the grain boundaries, which are around 100-200 nm in size. The white arrow points to what is labeled a “white phase,” the dotted white arrow points out a “gray phase,” and the black arrow identifies a “dark phase.” Figure 22b shows a higher magnification of the phases at the grain boundary.

26 a) b)

Figure 22: STEM images of Al 6061 powder particle showing different precipitates and dispersoids at the grain boundaries. a) and b) scale bars read 500 nm. The average composition, in atomic percent, is given for each of these phases in Table 5. The dark phase is identified as Mg2Si, however, a diffraction pattern did not identify it as the stable Mg2Si. It is believed this is a metastable precipitate that has formed during gas atomization. The identity of the white and gray phases is difficult to pinpoint because of their similarity in composition, and the similarity of the phases predicted. They are identified as Al-Fe-Si containing precipitates predicted by the Scheil calculation as shown in Figure 21.

Table 5: Average composition in atom percentage of the phases depicted in Figure 22.

Average Atom % White Phase Gray Phase Dark Phase Mg 2.19 2.56 27.18 Al 79.4 89.3 61.13 Si 8.42 3.71 11.69 Fe 9.67 3.54 Cr 0.15 0.04 Zn 0.1 Mn 0.08 0.03

Further TEM on another particle, approximately 47 microns in diameter, is shown in Figure 23. This sample showed a much larger area of the particle than the previous sample. Fairly equiaxed grains were found inside the powder particle, with smaller phases speckled at the grain boundaries.

27

Figure 23: STEM image of Al 6061 as-received powder particle prepared by focus ion beam. EDS maps of various regions of interest are shown in Figures 24, 25, and 26. It is noted that copper was not included in the EDS maps because of influence from the copper grid that the lamellae is seated on. Figure 24 shows an interesting precipitate structure at a grain boundary. There appears to be a backbone type structure of an Mg-Si precipitate, possibly Chinese Script Mg2Si, with Al-Fe particles in the spaces between the “spines” of the Mg-Si precipitate.

Figure 24: HAADF STEM image of a grain and grain boundary area in as-received Al 6061; EDS maps of Al, Mg, Si, and Fe.

28 An inclined grain boundary is observed in Figure 25 and contains the interesting backbone structured Mg-

Si precipitates (Chinese Script Mg2Si). The Mg-Si precipitate structure at these grain boundaries seems dendritic, with the Al-Fe particles in between the secondary dendrite arms. This was not found in the first STEM sample shown in Figure 22, however, it is evident across the sample from Figure 23. It is a similar structure to the precipitates shown in Figure 7a of Chakrabarti et. al. [16].

Figure 25: HAADF STEM image of two large grain boundary areas in as-received Al 6061; EDS maps of Al, Mg, Si, and Fe. Though there are still similar dendritic Mg-Si precipitates in Figure 26, there are also discrete Al-Fe particles that do not seem to appear between the secondary dendrite arms, but still are contained within grain boundaries.

29

Figure 26: HAADF STEM image of discrete particles at grain boundaries of as-received Al 6061; EDS maps of Al, Mg, Si, and Fe. Image analysis was done on the EDS mapping areas that were taken at several locations in the TEM sample. A histogram of the size results of the Mg-Si and Fe particles is shown in Figure 27. It is shown that the majority of Mg-Si particles are around 40-80 nm in size. From the images analyzed, they also made up approximately 2.7 +/- 0.7 % area fraction. The Fe particles seem to greatly range in size from 20 to over 140 nm, and made up approximately 2.25 +/- 0.01 % area fraction.

If the interesting structure of the Mg2Si precipitates is considered to be dendritic, a secondary dendrite arm width and spacing can be reported. The SDA widths were measured be approximately 53 +/- 16 nm, while the SDAS measured 78 +/- 17 nm.

30 25

20

15 Mg 10

Frequency Fe Si 5

0 0 20 40 60 80 100 120 140 More Radius (nm)

Figure 27: Precipitate radius size distribution of Mg, Si, and Fe particles for as-received Al 6061. 3.1.5 Al 5056 Thermodynamic and Kinetic Modeling The same equilibrium predictions were conducted for Al 5056 in Thermo-Calc using the composition in Table 1. Figure 28 shows the mass percentage of equilibrium phases predicted by the software, while Table 10 gives the numerical values at room temperature [20°C]. The most abundant secondary phase is the beta, or Al3Mg2 phase at 0.133 mass percent.

0.010 0.009 0.008 Al45V7 0.007 Beta (Al3Mg2) 0.006 Al12Mn 0.005 T_Phase 0.004 Al6Mn 0.003 Mass Percent [%] Percent Mass FCC_L12 0.002 Al13Fe4 0.001 Liquid 0.000 Mg2Si 0 100 200 300 400 500 600

Temperature [°C] Figure 28: Mass percentage of predicted equilibrium phases from Thermo-Calc for Al 5056 as a function of temperature.

31 Table 6: Weight and volume percentages of equilibrium phases predicted for Al 5056 at room temperature [20°C].

Equilibrium Phases Phase Wt% Vol% FCC Al 85.05 87.89

 (Al3Mg2) 13.07 10.40

Al45Cr7 0.69 0.60

Al13Fe4 0.45 0.33

Al12Mn 0.45 0.40

Mg2Si 0.25 0.35 T Phase (AlMgCuZn) 0.02 0.02

Al3Ti 0.01 0.01

Two non-equilibrium solidification models were also calculated for Al 5056. The first is the segregation model, where the inputs are the composition of the alloy, grain size and cooling rate. An experimental grain size of 1.2 microns was used with a calculated cooling rate of 2.04x105 oC/s, from Equation 7, for Al 5056 and the segregation results are shown in Figure 29. The composition at the grain boundary, which was rich in magnesium, was used as an input into Thermo-Calc to predict what phases would form at the grain boundary. The weight and volume percent of the phases are shown in Table 7. As said previously, this type of model may be more accurate since it considers the fast cooling rate during gas atomization.

Figure 29: Segregation model output for Al 5056 (JMatPro®).

32 Table 7: Phases and phase fraction predicted to form at the grain boundary of Al 5056 based on the grain boundary composition predicted by the segregation model in Figure 29 (Thermo-Calc).

Phases at Grain Boundary Phase Wt% Vol% FCC Al 83.45 86.65

 (Al3Mg2) 15.58 12.46

Al12Mn 0.83 0.74 T Phase (AlMgCuZn) 0.06 0.07

Al13Fe4 0.04 0.03

Al45Cr7 0.03 0.02

Mg2Si 0.02 0.02

A Scheil non-equilibrium solidification plot was calculated for Al 5056 and shown in Figure 30. In contrast with Al 6061, the first phase to form is not the matrix phase, but an Al45Cr7 precipitate. Following this, the matrix forms. Later in the cooling process an Al-Fe phase, Al13Fe4, precipitates. Finally, towards the end of the solidification event, a familiar precipitate, Mg2Si, forms. Before the alloy completely solidified, the precipitate  (Al3Mg2) forms. It should be noted that although Al6Mn is shown in the legend, its phase fraction is negligible so it does not appear in the Scheil plot. These iron and chromium containing precipitates as well as Mg2Si will be identified experimentally in STEM.

33

Figure 30: Scheil solidification plot for Al 5056 calculated in Thermo-Calc. 3.1.6 Al 5056 Experimental Phase Identification It was determined to be very difficult to identify any phases from the as-polished surface of Al 5056 in

SEM. Even in backscatter mode, the magnesium segregation and Al3Mg2 phase does not have enough contrast against the matrix, and the iron containing phases are too small to image. STEM was the only tool where the phases could be imaged and identified.

A cross-sectioned and thinned Al 5056 as-received particle, approximately 45 microns in diameter, is shown in the STEM image in Figure 31. Particles and segregation outline the grain boundaries. Significant pores resulting from gas-atomization are also found in the center of the particle and are identified by arrows. More particles need to be cross-sectioned in order to make a conclusion about whether the presence of pores is a common occurrence in gas-atomized aluminum 5056 particles.

34

Figure 31: STEM image of Al 5056 as-received powder particle prepared by focus ion beam. EDS maps of a higher magnification bright field STEM image of the sample is shown in Figure 32. Copper was not included in any EDS maps because the sample was mounted on a copper grid. This image clearly shows the magnesium segregation that is found at the grain boundaries, as well as the dark precipitates speckled at the grain boundaries. The EDS map found the darker precipitates to be iron-containing, while the less abundant lighter precipitates are magnesium- and silicon-containing. Based on the thermodynamic and Scheil phase predictions, it is concluded that the magnesium silicon particles are Mg2Si. Further EDS was performed to identify the smaller particles and results are shown in Figure 33.

35

Figure 32: Bright field STEM image of discrete particles and segregation at grain boundaries of as-received Al 5056; EDS maps of Al, Mg, Fe, and Si. Figure 33 shows a higher magnification bright field STEM image of the smaller dark particles and corresponding EDS maps. Magnesium segregation was found at the grain boundary as well as small those Al-Fe-Cr-Mn particles. These particles could be any of the following phases, where iron, chromium and manganese are substituting for each other: Al45Cr7, Al13Fe4 or Al6Mn. It is difficult to identify these small particles because they contain all of the alloys in those phases. Mg2Si is found to form around those Al- Fe-Cr-Mn particles.

Figure 33: Bright field STEM image of discrete particles at grain boundaries of as-received Al 5056; EDS maps of Al, Mg, Fe, and Si.

36 Image analysis was done on the EDS mapping areas that were taken at several locations in the TEM sample. A histogram of the size results of the Al-Fe-Cr-Mn and Mg-Si particles is shown in Figure 34. It is shown that the majority of the Al-Fe-Cr-Mn particles are between 30 and 60 nanometers. There is a lower number of the magnesium silicon particles than the Al-Fe-Cr-Mn particles and the majority are between 30 and 40 nanometers in size. The area fraction of magnesium was calculated for 3 different regions of the sample to be an average of 43.8 +/- 3.4%. This result is concluded to be an overestimation based on the resolution of the EDS map.

60

50

40

30

Al/Fe/Cr/Mn Particles Frequency 20 Mg/Si Particles

10

0 0 10 20 30 40 50 60 70 80 90 More Particle Radius [nm]

Figure 34: Precipitate radius size distribution of Al-Fe-Cr-Mn and Mg-Si particles for as-received Al 5056.

3.1.7 Experimental Verification of the Powder Production Stage Models In order for the experimental results to be comparable with the output of the models, the area fractions of the phases found were converted to volume fractions using the method described by Corti [7]. The relationship in Equation 9 was used.

2 퐹 = 휋푟2푁 Equation 9 푣 3 퐴

Fv is the volume fraction of the particles, r is the average radius of spherical particles, and NA is the number of particles per unit area studied [7]. The variables used in the equation for each type of phase in the two alloys and the comparative results are found in Table 8. The experimental volume percent as well as the equilibrium and non-equilibrium output from the model are shown. The experimental volume percent was found to be larger than the model output for each phase in both alloys. This could be due to the sample size and the features of the STEM images. Only one particle from each alloy was studied through STEM, and the images focused on the grain boundaries. This means that each image contained a grain boundary, and images without a grain boundary were not included in the study. This would increase the total experimental volume percent of phases found. The other limitation of converting area percent to volume percent is that a spherical particle is assumed but may not be the case for all of the phases included in the table.

37 Table 8: Comparison of experimental results and outputs of the Powder Production Stage models for Al 6061 and 5056.

Al 6061 Al 5056 Phase Mg2Si Fe Al/Fe/Cr/Mn Mg/Si Area percent of phase (%) 2.70 2.25 3.30 2.10 Number of particles 45 110 268 165 Average phase radius (m) 0.07 0.08 0.05 0.05 Total Area Studied (m2) 38.59 58.40 84.55 100.78 Experimental Volume % 3.39 7.02 4.40 2.18 Equilibrium Volume % 1.57 0.77 1.33 0.35 Non-Equilibrium Volume% 0.00 0.58 0.797 0.023

This section of the dissertation showed model and experimental results for the Powder Production Stage of the Through-Process Model. The area percentage of the phases present in the experimental STEM images was converted to volume percent and compared to the model results. Though there is a discrepancy in these values, the method of experimental validation has been proven. In the future, more samples must be consulted with a larger total area examined that would be more representative of the sample.

38 3.2 Powder Processing Stage This stage of the model focuses on any preparation steps that are applied to the powder before it is sprayed. The following sections will focus on a heat treatment, or degassing, procedure. The prescribed heat treatments will remove brittle hydroxides from the surface and change the phases and grain size. This section will discuss the initial kinetic modeling and characterization of Al 6061. It is to be noted that natural aging that may be occurring to the powder at room temperature was not considered in this dissertation. 3.2.1 Al 6061 Heat Treatment Modeling Initial heat treatments of Al 6061 have been prescribed at 200°, 230°, and 260°C for twenty-four hours. These times and temperatures originate from initial studies on the aging effects on the microstructure, and will overage the alloy, as shown by the curves in Figure 11. A time temperature transformation (TTT) diagram will be critical in modeling the effect of heat treatment on the microstructure. Figure 35 shows a (TTT) diagram for the  phase in Al 6061. The star represents a heat treatment at 230°C for twenty-four hours on an Al 6061 powder sample. The diagram illustrates that almost all of the equilibrium Mg2Si will form during that heat treatment. This sample was investigated in STEM.

500 Mg2Si 450 Beta' 400 Beta'' 350

230°C, 24 hours C]

° 300

250

200 Temperature Temperature [

150

100

50

0 0.0001 0.001 0.01 0.1 1 10 100 1000 10000 100000 Time [hr] Figure 35: Time temperature transformation diagram for the  phase in Al 6061. For solutionizing tests, the thermodynamic phase prediction in Figure 18 and the Scheil model in Figure 21 were consulted for choosing the solutionizing temperature. The solubility chart in Figure 36 was also used to see the temperature where the elements in the alloy would reach their maximum solubility in aluminum. A temperature of 540oC was chosen based on the dissolution of the last precipitate before

39 melting occurs as well as the maximum solubility. The time of the heat treatment was set at one hour because it will simplify processes in industry.

Figure 36: Solubility of alloying elements in aluminum as a function of temperature for Al 6061, showing the maximum solubility at 540oC. 3.2.2 Al 6061 Experimental Comparison Microstructure and Phases Present One of the samples from the initial heat treatments was examined in TEM. The Al 6061 powder was heat treated at 230°C for twenty-four hours. Figure 37a shows precipitates and dispersoids at the grain boundaries of a single powder particle. A higher magnification image of several phases at a triple point in the grain boundary is shown in Figure 37b. The arrow points to the dark phase where a diffraction pattern was taken and is shown in Figure 37c. This phase was indexed as the stable Mg2Si phase. It is slightly darker in contrast, and a different shape than the metastable phase found in the as-atomized sample.

a) b)

c)

Figure 37: STEM images of an Al 6061 powder particle heat treated for twenty-four hours at 230°C. a) and b) scale bars read 500 nm. c) A diffraction pattern taken from the dark precipitate indicated by the white arrow in b).

40 The average composition of the two phases shown in Figure 37 is listed in Table 9. The darker phase confirms to be Mg2Si while the white phase remains unidentified, only known as an iron-containing intermetallic.

Table 9: Average composition in weight percent of phases found in Figure 37.

Average Weight % White Phase Dark Phase Mg 2.4 60.8 Al 80.8 3.6 Si 3.1 35.1 Cr 0.1 0.2 Mn 0.2 0.0 Fe 13.4 0.3

The solutionized sample was scanned in the DSC and held at 540oC for 1 hour. The DSC curve is plotted in Figure 38. The typical peaks for the precipitation and dissolution of the  phase are found, similar to the literature data in Figure 8. A powder particle approximately 40 microns in diameter, was examined in STEM and shown in Figure 39. Fairly equiaxed grains are found in the sample, with precipitates speckled at the grain boundaries. On closer look on the right side of Figure 39, there are bright, small precipitates, containing heavier elements, and larger dark precipitates at triple points of the grain boundaries. EDS will help to determine the identities of these phases.

Figure 38: DSC scan for Al 6061.

41

Figure 39: (Left) HAADF STEM image of Al 6061 particle heat treated at 540oC for 1 hour and prepared by focus ion beam. (Right) Higher magnification HAADF STEM image of heat treated Al 6061 particle showing precipitates at grain boundaries. Through EDS mapping in Figure 40 andFigure 41, it was found that the small bright white precipitates at the grain boundary contain almost all of the elements in the alloy; aluminum, silicon, iron, chromium, and manganese. This means these phases are difficult to identify and could be any one or a combination of the following phases from the predictive tools; Al13Fe4, Al8Fe2Si, Al15Si2Fe4, Al9Fe2Si2 and Al18Fe2Mg7Si20. The darker, larger precipitates mostly contained at triple points contain magnesium and silicon, and were hypothesized to be Mg2Si.

Figure 40: HAADF STEM image of discrete particles at grain boundaries of heat treated Al 6061 (540oC for 1 hour); EDS maps of Al, Mg, Si, Fe, Cr, Mn.

42

Figure 41: HAADF STEM image of discrete Fe-containing particles and larger Mg-Si particles at grain boundaries of heat treated Al 6061 (540oC for 1 hour); EDS maps of Al, Mg, Si, Fe, Cr, and Mn. The precipitate radius size distribution of Al-Fe-Si-Cr-Mn and Mg2Si particles is shown in Figure 42. It is found that the Al-Fe-Si-Cr-Mn precipitates are small, with the majority around 40-60 nm, and with a few over 200 nm in size. The average area fraction of these particles is approximately 4.1 +/-1.3 %. The Mg2Si phase has seemed to lower in number but coarser in size. Most of the precipitates are over 200 nm in size, with an average of 324 +/- 154 nm in size.

50

40

30

20 Fe/Si/Cr/Mn Frequency 10 Mg/Si 0

Radius (nm)

Figure 42: Precipitate radius size distribution of Al-Fe-Si-Cr-Mn and Mg-Si particles for heat treated Al 6061 (540oC for 1 hour). Grain Size For two of the initial heat treated samples (200° and 230°C), grain size was measured and compared to the as-atomized powder that did not receive a heat treatment. In Figure 43, the dots represent individual data points while the dotted line is a trendline for the corresponding data. In general, there was a slight increase in grain size during the heat treatments. More samples will have to be analyzed to make a conclusion about grain growth.

43 3.5

3.0

2.5

2.0 As-Received

1.5 Heat Treated (200C) Grain size Grainsize [um] Heat Treated 1.0 (230C) Amorphous

0.5

0.0 0 5 10 15 20 25 30 35 40 Particle size [um]

Figure 43: Plot of grain size as a function of particle size for as-atomized powder and heat treated powder (200° and 230°C). The grain size of the as-received Al 6061 sample and the solutionized sample (HP344-540G, 540oC for 1 hour) was determined by measuring 50 and 80 individual grains from the samples, respectively, and is shown in Table 10.

Table 10: Average grain size for as-received Al 6061 (P344) and heat treated Al 6061 (540oC for 1 hour, HP344-540G).

P344 HP344-540G Average 2.3 2.8 St. Dev. 0.9 0.7 Maximum 4.8 5.3 Minimum 0.9 1.4

Comparison The grain size slightly increases with increasing treatment temperature between the as-received and the initial heat treated samples (200° and 230°C). However, more data points are needed to prove that there is an increase. On the other hand, the grain size does not differ significantly between the as-received and solutionized Al 6061 samples. This shows that grain growth does not occur at high temperatures for short periods of time. More grain growth experiments at different solutionizing times and temperatures are needed.

There was a significant change in the morphology of the phases during the solutionizing step. Firstly, the Al-Fe-Si-Cr-Mn particles seemed to have refined in size. In the as-received sample, there was a large size range, from 20 to over 140 nm in size. In the solutionized sample, the majority of these particles were around 40-60 nm in size, and with only a few over 200 nm in size.

44 Secondly, the solutionizing step has dissolved the interesting dendritic structure of the Mg2Si phase. The non-dendritic precipitates found in the as-received sample were around 40-80 nm in size, with a 2.7 +/-

0.7 % area fraction. After solutionizing, the Mg2Si phase does not contain any precipitates in the dendritic structure, but it has seems to lesser in number and coarser in size. Most of these types of precipitates are between 200 and 500 nm in radius. There were approximately 15 Mg2Si precipitates in the entire sample, mostly at triple points. 3.2.3 Al 5056 Heat Treatment Modeling As a non-heat treatable alloy, kinetic modeling is difficult for Al 5056. To choose a homogenizing temperature, literature values were taken into account, as well as the information given by Figure 28,Figure 30, andFigure 44. Sheppard & Raghunathan reported successful homogenization temperatures ranging between 450 and 550oC [19]. Figure 28 shows at what temperatures different phases might dissolve; at temperatures reaching 560oC, most phases have dissolved and the liquid phase appears. The Scheil plot in Figure 30 shows that upon solidification of the alloy, an aluminum chromium phase forms even before the matrix phase. This means that this phase might not be dissolved during the homogenization treatment, as a temperature that high will not be chosen as it could involve melting. For initial tests and TEM characterization, 500oC was chosen as a suitable homogenization treatment based on all of the evidence described. To choose a time, the plot shown in Figure 44 was consulted. This plot o predicts the volume fraction of two phases in Al 5056, Al6Mn and Al13Fe4, as a function of time at 500 C. Inputs for this model are composition, temperature, and grain size. It is shown that after approximately 1 hour, the volume fraction of the phases starts to level out. It is also beneficial to choose a shorter heat treatment time so it is easier to implement in industry. STEM was conducted on only one sample (500oC for 1 hour). Nanohardness data for samples homogenized at various temperatures are shown in the next section.

45

Figure 44: Thermo-Calc prediction of volume fraction of two precipitates, Al6Mn and Al13Fe4, in Al 5056 as a function of time at 500oC.

3.2.4 Al 5056 Experimental Comparison Microstructure and Phases Present A STEM image of the cross-sectioned and thinned sample of the heat-treated Al 5056 (500oC for 1 hour) particle, approximately 40 microns in size, is shown in Figure 45. The bright white areas are pores resulting from gas-atomization. The larger dark spots that appear at grain boundaries are precipitates that did not dissolve, and may have coarsened during the heat treatment. The smaller dark spots seem to have remained at grain boundaries where the magnesium segregation has disappeared, i.e. the magnesium has dissolved into solution. There does not seem to be a large change in the grain size compared to the as- received powder.

46

Figure 45: STEM image of Al 5056 particle heat treated at 500oC for 1 hour and prepared by focus ion beam. Figure 46 shows an EDS map taken from a representative area from the sample shown in Figure 45. It is to be noted that even though copper is present in the alloy, it was not included in the EDS maps because the sample was mounted to a copper grid. In the area analyzed, a few magnesium silicon particles are found. Medium-sized particles include Al, Fe, and Mn are found at grain boundaries. Even smaller particles containing Al and Cr are scattered throughout the sample, mostly located at grain boundaries. The largest visible difference between the homogenized and the as-received samples is the lack of magnesium segregation at the grain boundaries.

47

Figure 46: HAADF STEM image of discrete particles at grain boundaries of heat treated Al 5056 (500oC for 1 hour); EDS maps of Al, Mg, Si, Fe, Cr, and Mn. A histogram of the precipitate radius sizes found in the heat treated Al 5056 is shown in Figure 47. Separate Al-Cr and Al-Fe-Mn phases were found at the gain boundaries and prior grain boundaries where magnesium segregation has disappeared. There is a larger number of Al-Cr particles in the sample and the majority are between 30-60 nm in size. There are less Al-Fe-Mn particles and they range mostly between 30-50 nm with a few larger ones around 150 nm. Eight Mg2Si particles were found in this sample and they ranged from 230 to 580 nm in size. No magnesium segregation was found, and thus no phase fraction is available.

35 30 25 20

15 Al/Cr Particles Frequency 10 Al/Fe/Mn Particles 5

0

0

30 10 20 40 50 60 70 80 90

100 110 120 130 140 150 160 More Particle Radius [nm]

Figure 47: Precipitate radius size distribution of Al-Cr and Al-Fe-Mn particles for heat treated Al 5056 (500oC for 1 hour). Grain Size A comparison of the grain size for as-received and the solutionized sample is shown in Table 11. Approximately 90 grains in each sample were measured and the averages, standard deviations, maximums and minimums are reported. There is a level of uncertainty based on the sample size.

48 Table 11: Average grain size for as-received Al 5056 (P743) and heat treated Al 5056 (500oC for 1 hour, HP743-500G).

P743 HP743-500G Average 1.4 1.6 St. Dev. 0.3 0.4 Maximum 2.2 2.9 Minimum 0.8 0.7

Comparison When comparing the solutionized Al 5056 sample to the as-received sample, the most defining difference is the loss of the magnesium segregation at the grain boundaries. All of the Mg has either gone into solution, or is contained within the eight Mg2Si particles found throughout the sample, mostly at triple points of the grain boundaries. These particles ranged between 230 to 580 nm in radius. These particles have also coarsened from the as-received sample, mostly ranging between 30-40 nm in size.

As for the other precipitates in the sample, an interesting phase transformation has occurred. In the as- received sample, small particles between 30-60 nm in size were found to contain Al/Fe/Cr/Mn. After the heat treatment, two distinct particles were found that were similar in size: an Al/Cr precipitate (30-60 nm) and another Al/Fe/Mn (30-50 nm). One of these may be the Al/Cr precipitate, Al45Cr7, since it was hypothesized this would not dissolve based on the thermodynamic and Scheil plots. The Al/Fe/Mn phase could be Al13Fe4 or Al6Mn with iron and manganese substituting for each other.

Though it is not completely clear what types of phase transformations are occurring during this solutionizing treatment, it is sure that the magnesium that was previously segregated to the grain boundary has dissolved into the matrix and thus will increase the solid solution strengthening term.

49 3.3 Additive Strength Model As mentioned in the Background Section, the additive strength model, given in Equation 1, is used to integrate the powder production and preparation steps with the impact model stage. The strength model takes into account several modes of strengthening that occur within the powder particle. This additive strength can be used as an input into a plasticity model of the particle impact finite element model. 3.3.1 Model Description The strengthening contributions included in the additive strength model are solid solution strengthening, microstructural influence, and precipitation hardening. The equations included in the model will be briefly described [32]. The equation used for solid solution strengthening is given in Equation 10. This is applicable to substitutional solutes.

ퟑ/ퟐ ퟏ/ퟐ ∆흈풔풔 = ∑풊(푮휺풔 풄풊 )/ퟕퟎퟎ Equation 10 G is the shear modulus of the matrix,  is a strain hardening constant, and c the concentration of the solute. The limitation of this equation is that a spherical distortion is assumed, where the hardening is much less than for tetragonal distortions.

The microstructural influence is governed by the Hall-Petch behavior, in Equation 11:

풌품풔 ∆흈 = Equation 11 풎풊풄 √풅

푘푔푠 is the Hall-Petch constant and d is the microstructural feature size, both of the matrix phase. The precipitation hardening term can be broken into two types of strengthening: precipitation and dispersion. For coherent precipitates, Equation 12 can be used. The three terms that make up the coherent precipitate strengthening equation are strengthening from coherency, modulus and chemical. Those are given by Equations 13, 14 and 15.

∆흈풑풑풕, 푪풐풉 = ∑풊[흈Coherency, i+ 흈Modulus, i+ 흈Chemical, i] Equation 12

ퟑ/ퟐ 풓풇 ퟏ/ퟐ ∆흈 = ퟕ휺 푮 ( ) Equation 13 푪풐풉풆풓풆풏풄풚,풊 풄풐풉,풊 풃

ퟑ/ퟐ 풓풇 ퟏ/ퟐ ∆흈 = ퟐ휺 푮 ( ) Equation 14 푪풉풆풎풊풄풂풍,풊 푪풉,풊 풃

ퟑ/ퟐ 풓풇 ퟏ/ퟐ ∆흈 = ퟎ. ퟎퟏ휺 푮 ( ) Equation 15 푴풐풅풖풍풖풔,풊 푮풑,풊 풃 The variables in Equations 13, 14, and 15 include hardening constants for strain, interfacial, and modulus, given by 휀푐표ℎ, 휀퐶ℎ, 휀퐺푝, respectively. The other variables include r for the precipitate radius, f is the volume fraction, and b is the burgers vector. The coherency, chemical, and modulus contributions of Equations 13, 14, and 15 are for early stage precipitation and are analogous to the size effect in solid solution hardening. The limitation of these equations is that they assume a spherical precipitate. The author points out that these equations overestimated the strength when compared with experimental results. It is also noted that chemical strengthening does not play an important role in precipitation hardening [32].

For an incoherent particle, either a precipitate or dispersoid, Equation 16 can be utilized. There is one term and it is given by Equation 17.

50 ∆흈풑풑풕,푰풏풄풐풉 = ∑풊[흈Incoherency, i] Equation 16

푮풃 ∆흈 = Equation 17 푰풏풄풐풉풆풓풆풏풄풚,풊 푳−ퟐ풓 G is the shear modulus, b is the burgers vector, L is the interparticle spacing, and r is the particle radius. This strength contribution is due to dislocations bowing around non-deforming incoherent particles [32].

To illustrate the different contributions from the strengthening mechanisms as a function of particle diameter, Figure 48 was created. The black line represents the total strength contributions for Al 6061. The matrix strengthening, or intrinsic strength, is the leading contribution. The microstructural influence has the next highest contribution, followed by solid solution and precipitation strengthening.

Figure 48: Graph of yield strength versus particle diameter for Al 6061 showing individual contributions from strengthening mechanisms of the additive strength model. 3.3.2 Experimental Comparison As an initial comparison to the general idea that smaller particles will have a higher strength than larger particles, mostly due to the microstructural term, nanohardness indents were performed on the different size ranges of the sieved, as-received Al 6061 powder. The results are shown in Figure 49, where the columns represent the average hardness for each size range indicated below it. There was not a significant difference in the nanohardness between the size ranges.

51 1.60 1.40 1.24 1.16 1.12 1.20 1.05 1.00 0.80 0.60

0.40 Nanohardness(GPa) 0.20 0.00 25-32 32-38 38-45 45-53

Range of Particle Sizes [m] Figure 49: Average nanohardness values for different ranges of sieved particles. Nanoindentation was also done on conventional Al 6061 and 5056 powder, without heat treatment, as well as the initial heat treated/degassed samples for those alloys. The average nanohardness, of all size particles, is shown in Figure 50. These results show that degassing/heat treating the powder will reduce its hardness, thus reducing the flow stress of the material and making it more “sprayable.” As discussed earlier, the Al 6061 samples heat treated at 200oC and 230oC were not solutionized, but simply overaged, thus a decrease in hardness is expected. Again, a significant difference in nanohardness was not found.

1.60 1.26 1.40 1.06 1.09 1.20 1.00 0.80 0.60 0.40

Nanohardness(GPa) 0.20 0.00 As-Received 6061 Degassed 200 Degassed 230

Figure 50: Average nanohardness values for as-received and “degassed” samples of Al 6061. The variation in nanohardness values for difference solutionizing times at 530oC is shown in Figure 51. It is shown that there is an initial increase in nanohardness at 0 hours when the sample is brought to 530 oC and immediately quenched. As you increase the time held at the solutionizing temperature, the hardness tends to decrease, with the lowest value after 4 hours. This could be due to the coarsening and loss of coherency with the main strengthening phase, Mg2Si.

52 1.6 1.29 1.23 1.4 1.12 1.19 1.14 1.2

1.0 0.88

0.8

0.6

Nanohardness(GPa) 0.4

0.2

0.0 As-Received 0 hours 1/3 hours 1 hour 2 hours 4 hours

Figure 51: Variation of average nanohardness values for Al 6061 solutionized at 530oC for various times for particle sizes between 38-45 microns. The variation in nanohardness between Al 5056 samples that were solutionized at different temperatures ranging from 400oC to 550oC is shown in Figure 52. The initial drop in hardness even at a low homogenization temperature could be due to a loss of dislocations and thermal stresses that could have resulted during the rapid solidification, similar to the study by Zhu [18]. A further increase and stabilization in strength at higher temperatures could mean that the sample has fully solutionized. Though there are less Mg2Si particles, all of the magnesium that was segregated to the grain boundaries has diffused to the matrix, thus contributing highly to the solid solution strengthening term. There are still a great number of dispersoids at the grain boundaries that will also contribute to the strength.

2.0

1.8

1.6

1.4 1.16 1.13 1.2 0.99 1.07

1.0 0.84

0.8

0.6 Nanohardness(GPa) 0.4

0.2

0.0 As-Received 400C 450C 500C 550C

Figure 52: Variation of average nanohardness values for Al 5056 solutionized for 1 hour at various temperatures for particle sizes between 38-45 microns.

53 Further nanoindentation work was done to Al 6061 to compare to the additive strength model. Figure 53 shows an example of a nanohardness grid that was conducted on an individual Al 6061 powder particle.

Figure 53: SEM image of a 5x5 nanohardness grid on a powder particle. Depth of indents is approximately 250 nm. Nanohardness values were averaged and converted to Vickers hardness for various sized particles and plotted in Figure 54. The purple dots represent the experimental values, the dotted line represents the experimental trendline, and the red solid line represents the additive strength model, converted to Vickers hardness. It is intuitive that particle size is inversely proportional to hardness, knowing that one of the major contributors to the additive strength equation, the microstructural influence, is also inversely proportional to particle size. Therefore, the high cooling rate of the small powder particles will lead to a higher hardness. Particles less than approximately 20 microns in diameter were not included in this study. At this point, those particles were observed not to have any microstructural features, or seemingly amorphous. Overall, there is good agreement between the experimental data and the model for Al 6061. This model should be applied to other aluminum alloys.

54

Figure 54: Model and experimental comparison of the relationship between particle diameter and hardness for Al 6061.

55 3.4 Particle Impact Model 3.4.1 Model Description The final stage of the through-process model that has been developed is the particle impact stage, which involves using a finite element model with either the Johnson-Cook or Preston-Tonks-Wallace plasticity model. One of the inputs to this model will be property data from the powder production and processing stages, i.e. the additive strength model. The impact of a single particle, and eventually multiple particles, will then be modeled and completely understood so that mechanical property data can be predicted for the final cold sprayed material. Initial model results from the Applied Research Laboratory at The Pennsylvania State University are shown here.

The model utilizes Abaqus/Explicit software. The outputs will provide key data needed for the prediction of material properties and microstructure, and include plastic strain, temperature and residual stress. Figure 55 shows the plastic equivalent strain output from a single particle of Al 6061-O impacting a substrate of the same material. It is a 20-micron diameter particle with an impact velocity of 700 m/s. The particle and substrate initial temperature is set to 150°C. It is evident that the most plastic strain occurs at the boundary of the particle, while the interior of the particle is less deformed. Qualitative and quantitative experimental characterization will be used to both verify the deformed shape of the particle and the quantitative output of the model.

Figure 55: Plastic equivalent strain output from single particle impact model. 3.4.2 Experimental Comparison Qualitative Verification A sample was created that included many single particle splats by increasing the raster speed of the gun. This meant that individual particles could be singled out and characterized, instead of building up a continuous and dense coating. A single particle splat, approximately 40 microns in diameter, was sprayed with similar conditions to the model in Figure 55. The splat was cross sectioned by FIB at the Center for Nanoscale Systems (CNS) at Harvard University. The initial cross section cut is shown in Figure 56.

56

Figure 56: Initial cross section cut of Al 6061 single splat. After cleaning cross section cuts, the microstructure of the powder was revealed, in Figure 57. The particle and substrate interface is highlighted by a lighter contrast in the image. The shape of the deformed particle agrees with what was predicted by the model in Figure 55. Additionally, strain contrast in the image shows very high plastic deformation in the substrate and in the bottom of the particle, while the top/middle of the particle is deformed to a lesser degree.

Figure 57: Cross section of Al 6061 single particle splat. A higher magnification image of the material jetting at the top right of the deformed particle is shown in Figure 58. This also agrees with the material jetting predicted by the model. It is believed that this material jetting is from the substrate and not the particle itself.

57

Figure 58: Top right edge of cross section of Al 6061 single particle splat illustrating the jetting that occurred at the particle/substrate interface. A laser scanning confocal microscope was employed to better investigate the topography of the single particle splat. The left image in Figure 59 is a top down view of two single particle splats, where the material jetting is evident. The right image is the topography output, or height profile from the confocal microscope, highlighting the material jetting on either side of the profile and the rounded particle surface in the center. The characteristics of the deformed particle agree with the model output.

Figure 59: (Left) Top down view on Al 6061 single particle splat and (Right) height profile of Al 6061 single particle splat measured in a confocal laser scanning microscope. Quantitative Verification Comparative nanohardness was used to quantitatively show the amount of plastic deformation that occurred during the cold spray process. Figure 60 shows grids of nanohardness indents on both a cross section of an Al 6061 powder particle and on the spray direction surface of an Al 6061 consolidated coating. The coating was slightly etched with a 0.5% hydrofluoric acid solution for 5 seconds to reveal powder particle boundaries. A particle boundary is highlighted in blue. The average nanohardness for all sizes of powder for Al 6061, Al 2024, Al 7075, and Al 5056 is shown by the solid bar in Figure 61, while the average nanohardness across a consolidated sample is shown by the dashed bar. In almost all alloys, there is a slight increase in nanohardness after the cold spray process. Al 5056 experiences the largest increase in nanohardness as it is a strain hardening alloy. Further nanohardness was done on single particle impacts to compare to the single particle impact model.

58

Figure 60: (Left) SEM image of a grid of nanohardness indents on a single Al 6061 powder particle. (Right) SEM image of a grid of nanohardness indents on the spray direction face of a consolidated Al 6061 coating. The coating was etched with 0.5% hydrofluoric acid solution for 5 seconds to reveal particle boundaries. A blue line highlights one particle boundary.

2.50

2.00

1.50

1.00

Nanohardness (GPa) Nanohardness 0.50

0.00 Al 6061 Al 2024 Al 7075 Al 5056 Figure 61: Graph of average nanohardness for Al 6061, Al 2024, Al 7075 and Al 5056. Average nanohardness for all sizes of powder is shown by the solid bar, while the dashed bar represents the consolidated sample. Nanohardness on single particle impacts was used to quantitatively show the amount of plastic deformation. Firstly, Figure 62 shows a polished cross section of a single particle impact from the fast raster test. The particle/substrate interface is highlighted by a blue line. An array of nanoindentations starts at the top of the deformed particle and goes into the substrate. The values of these nanoindentations were plotted in Figure 63.

59

Figure 62: SEM image of single particle impact showing location of profile of nanoindentations across the particle and substrate. It is shown in Figure 63 that the negative position values are in the interior of the particle, x=0 is the interface, and positive position values are contained within the substrate. The results show an increase as you move to the right in the particle, towards the substrate. An indent just near the interface has a low value of hardness, suggesting an impact with a poorly bonded region. The hardness and modulus then greatly increase at after this interface, into the substrate. Using the model output in Figure 55, this result was expected based on the amount of plastic deformation that occurred.

4.0 120 3.5 100 3.0 80 2.5 2.0 60 Hardness 1.5 40

Modulus(GPa) Modulus 1.0 Nanohardness(GPa) 20 0.5 0.0 0 -30 -20 -10 0 10 20 30 40 Distance from Interface (microns)

Figure 63: Plotted nanohardness and modulus values for line profile conducted in Figure 62. Negative position values are within the particle, x=0 is the particle/substrate interface, and positive position values are within the substrate. For a true quantitative comparison, an Al 6061 single particle impact was created at UMASS for characterization and comparison to the finite element model. The goal of this quantitative verification was to quantitatively match nanohardness values taken from a cross-section of a single particle impact to plastic equivalent strain values taken from the same locations in the finite element model. The particle diameter is 18.9 microns and the velocity was calculated to be 927 m/s. These are important inputs into

60 the particle impact model. SEM images of the top down view of the specific particle characterized, and a cross-section that was prepared by plasma FIB at UCONN, is shown in Figure 64. Normal material jetting is found in the top down view, similar to the jetting found in the other single particle splats in Figure 59. The right side of Figure 64 shows the cross-section of the particle, with platinum deposit on the surface to protect the sample during milling. A large gap between the particle and substrate is found, showing an area of de-bonding during the cold spray process. This de-bonding could have resulted from poor adhesion of the particle and substrate, or from polymer contamination that may have been introduced during the process UMASS used to spray a single particle.

Figure 64: (Left) SEM image of top down view of the Al 6061 single particle splat, and (Right) cross-sectioned Al 6061 single particle impact prepared by plasma FIB. Two nanohardness line profiles were conducted on the cross section of the single particle impact and are shown in Figure 65. The lines labeled as 1 and 2 start at the top of the particle and travel across the particle/substrate interface and continue into the substrate.

1

2

Figure 65: SEM image of cross-sectioned Al 6061 single particle impact showing locations of nanoindentations. The nanohardness points from the line profiles shown in Figure 65 are plotted in Figure 66 as a function of position. The negative position values represent the particle while the positive position values are in the substrate. The y-axis represents the particle-substrate interface. It is shown that the nanohardness for both profiles slightly increases in the substrate just after the interface, where the most plastic deformation would have occurred.

61 1.8 1.6 1.4 1.2 1.0 Line Profile 1 0.8 Line Profile 2 0.6 Nanohardness(GPa) 0.4 0.2 0.0 -15 -10 -5 0 5 10 15 Position (mm) Figure 66: Plotted nanohardness values for line profiles conducted in Figure 65. Negative position values are within the particle, x=0 is the particle/substrate interface, and positive position values are within the substrate. To compare these nanohardness profile values to the particle impact model, the model was run by The Pennsylvania State University using the Preston-Tonks-Wallace plasticity model as well as the same inputs as the experimental powder (18.9-micron diameter and a velocity of 927 m/s). The plastic equivalent strain output is shown in Figure 67.

Figure 67: Output of the finite element single particle impact model, showing plastic equivalent strain, calculated using the Preston-Tonks-Wallace plasticity model. The SEM image with the nanohardness profile locations was overlaid onto the single particle impact model in Figure 68. The black dots represent the locations and a numerical value for the plastic equivalent strain was extracted from these locations. The plastic equivalent strain was plotted against the corresponding

62 nanohardness values for both line profiles and this plot is shown in Figure 69. It was hypothesized there would be a relationship between the predicted strain and the experimental nanohardness, however, there was some variation in the slopes the two plotted lines. This type of quantitative verification method needs to be optimized.

1

2

Figure 68: SEM image of nanoindentations overlaid on the output of the single particle impact model.

2.5

2.0

PEEQ y = -1.825x + 3.8698 - 1.5 R² = 0.5873 Line Profile 1 Line Profile 2 1.0 y = -3.9236x + 6.0426 R² = 0.8083 ModelOutput 0.5

0.0 0.5 1.0 1.5 2.0 2.5 Nanohardness (GPa) Figure 69: Model output (plastic equivalent strain) plotted against nanohardness from the line profiles illustrated in Figure 68. Linear trendlines, equations, and R2 values are reported.

63 4.0 RECOMMENDATIONS FOR FUTURE WORK 4.1 Powder Processing Stage 4.1.1 Model Development Previous success in cold spraying powder in the solutionized condition has led to more research in the area of heat treating or degassing the powder before it is sprayed. An optimal heat treatment schedule is needed, as it will differ from the traditional heat treatment schedules reported in literature for each alloy. If a homogenized structure is preferred for cold spray, a dissolution model will be helpful in predicting when all of the soluble solute atoms have gone into solution and the dissolvable precipitates are eliminated. This type of model could be an analytical model or can utilize commercially available diffusion software such as DICTRA. 4.1.2 Powder Processing Experimental Verification More experiments on the effect of heat treatment time and temperature need to be conducted. This thesis provided a preliminary exploration of the microstructural effects when exposed to high temperatures for a short period of time for two alloys. Experiments that are designed specifically to compare to the newly developed kinetic models will be essential. Characterization techniques that have proved successful for these types of experiments are S/TEM with image analysis of solutionized samples to determine phase fractions, and nanoindentation to compare to the additive hardness model. Another interesting technique worth looking into is atom probe tomography. This would be beneficial as the EDS in STEM has shown us what elements are in each precipitate, however, it is difficult to identify the precipitates when they contain several elements that are in different phases. Atom probe tomography will give you more information on the composition of the micron and nano-sized precipitates as well as the coherency. 4.2 Particle Impact Model 4.2.1 Model Verification and Advancement The thermodynamic and kinetic models in conjunction with STEM and phase identification will be helpful when attempting to overlay a microstructure onto the finite element model to study the effects of precipitates at grain boundaries during impact. The impact model will be more accurate if microstructural features are implemented.

An initial attempt at quantitatively verifying the impact model was shown in this thesis. That comparison technique should be used as a stepping stone to designing an experiment that can quickly, accurately and quantitatively verify different outputs of the impact model. The impact model is a difficult model to calibrate and experimental verification is going to help propel those results.

64 5.0 CONCLUSIONS A through-process model, with experimental verification, does not yet exist for the cold spray process. This model, to be used by the U.S. Army and by industry, will greatly advance the cold spray process by making it a more versatile and robust process. Fewer critical parts on military vehicles and rotorcraft will be replaced and can be repaired at a fraction of the cost when compared to the cost of raw components. In addition, time and money spent on trial-and-error operations will be reduced by the predictive nature of the through-process model.

The first goal of this thesis work was to fully characterize aluminum 6061 and 5056 powders and quantitatively compare the microstructures to thermodynamic and solidification model predictions of the powder production stage of the through-process model. The grain size model agreed with the experimental results. The experimental phase identification and quantification results varied from the model predictions. These models need to be calibrated to more accurately describe the condition of the gas-atomized powders. Overall, these results have contributed to the development of the through- process model for cold spray aluminum alloys.

In addition, the basic microstructural research of aluminum particles will be of great contribution to the powder metallurgy and additive manufacturing field. Powder microstructures of aluminum differ greatly from their wrought and cast counterparts because of the unique atmosphere and solidification conditions during atomization. There is very little to no literature data on the microstructures of these alloys in the field of powder metallurgy, particularly of the powders of interest to applications in this research.

The second goal of this thesis work was to investigate the effects of heat treatment on the microstructure and phases of aluminum 6061 and 5056 powders used for cold spray. Initial solutionizing heat treatments were performed on both alloys and the samples were fully characterized and compared to the as- atomized powder. For both Al 6061 and 5056, grain growth was not found after the solutionizing heat treatment. The microstructure, however, changed significantly for both alloys. For Al 6061, the Al-Fe-Si-

Cr-Mn intermetallic phase refined in size after solutionizing. The Chinese script Mg2Si phase was dissolved, but a few larger Mg2Si particles, between 200 and 500 nm in radius, formed and remained. For Al 5056, the most defining difference between the as-atomized powder and the solutionized sample is the loss of the magnesium segregation at the grain boundaries. Similarly to the solutionized Al 6061 sample, large

Mg2Si particles were found throughout the sample, ranging between 130 and 580 nm in radius.

This work demonstrated the effects of heat treatments on aluminum powders, and with more research, a complete understanding of the effect of thermal processing of Al 6061 and 5056 powders will be beneficial to powder metallurgy community. The microstructure of the powders for additive manufacturing has yet to be the focus of research, as in most additive manufacturing techniques particle melting usually occurs. Understanding the effects of the heat treatments will lead to optimizing cold spray and other solid state process properties with heat treated powders. Much success has already been accomplished using only trial-and-error methods at this point, and optimized heat treatments will be unleashed with the power of the through-process model.

Another accomplishment of this thesis work was quantitatively characterizing the strength of the powder and validating the additive strength model. Nanohardness for different size particles and different heat treatments for Al 6061 and 5056 powder was shown. Model and experimental nanohardness as a function of particle size was also presented for Al 6061 and there was good agreement between the results. This model can be applied to other alloys as well.

65 The final achievement of this thesis was the qualitative and quantitative validation of a single particle impact model with experimental characterization of Al 6061 single particle impacts. The morphology of an impacted particle agreed with the impact model, including the shape of the deformed particle and material jetting at the edges. A quantitative method for validating the impact model was also introduced using nanohardness profiles across a single particle impact. Optimization of this method is currently underway.

The types of modeling and characterization techniques in this thesis could easily be applied to different material systems. Aluminum, being the current material of interest, was to develop this model and characterization techniques, but many materials are being used in the cold spray process that could benefit from this work, including titanium, copper, stainless steel, etc.

Not only will this through-process model be utilized for cold spray process advancement, it can be modified and applied to other powder processing and additive manufacturing techniques where the properties and microstructure of the powder play a significant role in the consolidated material microstructure. This model and experimental characterization is a foundation for a type of modeling and characterization techniques that could be employed by many different fields and manufacturing techniques.

66 6.0 REFERENCES 1. Lippert, J., Ford's Crown Jewel, the F-150, Has a Big Problem. Bloomberg Pursuits, 2016. 2. Belsito, D.C. and R.D. Sisson, Cold Spray Manufacturing for Repair, in Department of Defense Rapid Innovation Fund. 2012. 3. Cote, D.B., Application of Computational Thermodynamic and Solidification Kinetics to Cold Sprayable Powder Alloy Design, in Materials Science and Engineering. 2014, Worcester Polytechnic Institute. p. 68. 4. Council, N.S.a.T., Materials Genome Initiative for Global Competitiveness. 2011. 5. Belsito, D., et al. Through-Process Modeling for Cold Spray Alloy Optimization. in MS&T. 2013. Montreal, QC, Canada. 6. Champagne, V.K., The cold spray materials deposition process: Fundamentals and applications. 2007. 7. Corti, C.W., P. Cotterill, and G.A. Fitzpatrick, Evaluation of the Interparticle Spacing in Dispersion Alloys. International Metallurgical Reivews, 1974. 19: p. 77-88. 8. Army, U.S., Military Specification MIL-DTL-32262 (MR), D.o. Defense, Editor. 2007. 9. Aircraft Materials. Aluminum Alloy 5056 12/01/2016]; Available from: https://www.aircraftmaterials.com/data/aluminium/5056.html. 10. Strondl, A., et al., Characterization and Control of Powder Properties for Additive Manufacturing. JOM, 2015. 67(3): p. 549-554. 11. Zheng, B., et al., Powder Additive Processing with Laser Engineered Net Shaping (LENS), in Powder Metallurgy Research Trends, L.J. Smit and J.H.V. Dijk, Editors. 2009, Nova Science Publishers, Inc. p. 125-190. 12. Pattison, J., et al., Standoff distance and bow shock phenomena in the Cold Spray process. Surface & Coatings Technology, 2008. 202(8): p. 1443-1454. 13. Fukumoto, M., et al., Deposition of Copper Fine Particle by Cold Spray Process. MATERIALS TRANSACTIONS, 2009. 50(6): p. 1482-1488. 14. Borchers, C., et al., Microstructural bonding features of cold sprayed face centered cubic metals. Journal of Applied Physics, 2004. 96(8): p. 4288-4292. 15. Mondolfo, L.F., Metallography of Aluminum Alloys. 1943, New York: John Wiley and Sons, Inc. 16. Chakrabarti, D.J., B.-k. Cheong, and D.E. Laughlin, Precipitation in Al-Mg-Si-Cu Alloys and the Role of the Q Phase and its Precursors. Automotive Alloys II, 1998: p. 27-44. 17. Gómez De Salazar, J.M. and M.I. Barrena, The influence of Si and Mg rich phases on the mechanical properties of 6061 Al-matrix composites reinforced with Al2O3. Journal of Materials Science, 2002. 37(8): p. 1497-1502. 18. Zhu, Y., Characterization of Beta Phase Growth and Experimental Validation of Long Term Thermal Exposure Sensitization of AA5XXX Alloys, in Metallurgical Engineering. 2013, The University Of Utah. p. 133. 19. Sheppard, T. and N. Raghunathan, Modification of cast structures in Al-Mg alloys by thermal treatments. Materials Science and Technology, 1989. 5(3): p. 268-280. 20. Harrell, T.J., et al., Microstructure and Strengthening Mechanisms in an Ultrafine Grained Al-Mg- Sc Alloy Produced by Powder Metallurgy. Metallurgical and Materials Transactions A, 2014. 45(13): p. 6329-6343. 21. Flumerfelt, J.F., Aluminum powder metallurgy processing. 1998: Retrospective Theses and Dissertations. 22. Handbook of Aluminum. Physical Metallurgy and Processes, ed. G.E. Totten and D.S. Mackenzie. Vol. 1. 2003, Boca Raton: CRC Press.

67 23. Couper, M.J., M. Cooksey, and B. Rinderer, Effect of Homogenisation Temperature and Time on Billet Microstructure and Extruded Properties of Alloy 6061. 24. Aouabdia, Y., A. Boubertakh, and S. Hamamda, Precipitation kinetics of the hardening phase in two 6061 aluminium alloys. Materials Letters, 2010. 64(3): p. 353-356. 25. Du, Q. and J. Friis. Modelling Precipitation Kinetics During Aging of Al-Mg-Si Alloys. in 2nd World Congress on Integrated Computational Materials Engineering. 2013. Salt Lake City, Utah: Wiley. 26. Marioara, C.D., et al., The influence of alloy composition on precipitates of the Al-Mg-Si system. Metallurgical and Materials Transactions A, 2005. 36(3): p. 691-702. 27. Clinch, M.R., et al. The annealing response of cold worked 6000 and 7000 series aluminium alloys. 28. Maisonnette, D., et al., Effects of heat treatments on the microstructure and mechanical properties of a 6061 aluminium alloy. Materials Science & Engineering A, 2011. 528(6): p. 2718- 2724. 29. Mrówka-Nowotnik, G., Influence of chemical composition variation and heat treatment on microstructure and mechanical properties of 6xxx alloys. Archives of Materials Science and Engineering, 2010. 46(2): p. 98-107. 30. Chandler, H., M. Knovel, and C. Metallurgy Library - Academic, Heat treater's guide: practices and procedures for nonferrous alloys. 1996, Materials Park, OH: ASM International. 31. ASM Handbook: Volume 4: Heat Treating, ed. A.H. Committee. 1991. 32. Courtney, T.H., Mechanical behavior of materials. Vol. 2nd. 2005, Long Grove, Ill: Waveland Press.

68 INDEX List of Figures Figure 1: Schematic of the cold spray process [3]...... 5 Figure 2: Schematic of through-process modeling for cold spray...... 6 Figure 3: Bubble diagram of tensile strength versus elongation for several aluminum alloys studied, including manual records of cold sprayed material. Diagram created in GRANTA CES EduPack...... 7 Figure 4: Particle size distribution of sieved Al 6061 powder, conducted by laser diffraction...... 8 Figure 5: XRD pattern for Al 6061 powder and consolidated material...... 11 Figure 6: STEM images of a) Al 6061 powder with a scale bar of 100 nm and b) Al 6061 cold spray with a scale bar of 200 nm. The white arrows indicate the dark phase in both images and the black arrow indicates the white phase...... 12 Figure 7: Optical micrograph of an ingot of Al 6061 showing the Q phase and Mg2Si [16]...... 13 Figure 8: DSC curve for Al 6061 [17]...... 13 Figure 9: Optical micrograph of Al 5056 alloy etched with Barker's reagent and viewed under polarized light [19]...... 14 Figure 10: Schematic of typical heat treatment steps for aluminum alloys, indicating important variables...... 15 Figure 11: Al 6061 aging curves showing affect of heat treatment time and temperature on yield strength [31]...... 17 Figure 12: Precipitation sequence of Mg2Si in Al 6061 alloys...... 17 Figure 13: Backscatter SEM image of segregation and individual grains in an as-atomized Al 7075 powder particle...... 19 Figure 14: SEM images of cross sections of Al 6061 powder etched with 0.5% hydrofluoric acid solution for 90 seconds (left) and Al 5056 powder etched with a variation of Keller’s reagent with the addition of nitric acid for 10 seconds (right)...... 20 Figure 15: Model and experimental comparison of relationship between grain size and particle diameter for AL 6061 and 5056...... 21 Figure 16: SEM images of different sized Al 6061 powder particles plotted against calculated cooling rate [°C/s]...... 22 Figure 17: EBSD analysis of Al 6061 as-received. (Left) Band contrast image of scanned area, and (Right) inverse pole figure color map...... 23 Figure 18: Equilibrium phase predictions for Al 6061 as a function of temperature...... 23

Figure 19: Continuous cooling curves for Mg2Si phase in Al 6061. The gray curves on the left represent cooling rates...... 24 Figure 20: Segregation model output for Al 6061 (JMatPro®)...... 25 Figure 21: Scheil solidification plot for Al 6061...... 26 Figure 22: STEM images of Al 6061 powder particle showing different precipitates and dispersoids at the grain boundaries. a) and b) scale bars read 500 nm...... 27 Figure 23: STEM image of Al 6061 as-received powder particle prepared by focus ion beam...... 28 Figure 24: HAADF STEM image of a grain and grain boundary area in as-received Al 6061; EDS maps of Al, Mg, Si, and Fe...... 28 Figure 25: HAADF STEM image of two large grain boundary areas in as-received Al 6061; EDS maps of Al, Mg, Si, and Fe...... 29 Figure 26: HAADF STEM image of discrete particles at grain boundaries of as-received Al 6061; EDS maps of Al, Mg, Si, and Fe...... 30

69 Figure 27: Precipitate radius size distribution of Mg, Si, and Fe particles for as-received Al 6061...... 31 Figure 28: Mass percentage of predicted equilibrium phases from Thermo-Calc for Al 5056 as a function of temperature...... 31 Figure 29: Segregation model output for Al 5056 (JMatPro®)...... 32 Figure 30: Scheil solidification plot for Al 5056 calculated in Thermo-Calc...... 34 Figure 31: STEM image of Al 5056 as-received powder particle prepared by focus ion beam...... 35 Figure 32: Bright field STEM image of discrete particles and segregation at grain boundaries of as-received Al 5056; EDS maps of Al, Mg, Fe, and Si...... 36 Figure 33: Bright field STEM image of discrete particles at grain boundaries of as-received Al 5056; EDS maps of Al, Mg, Fe, and Si...... 36 Figure 34: Precipitate radius size distribution of Al-Fe-Cr-Mn and Mg-Si particles for as-received Al 5056...... 37 Figure 35: Time temperature transformation diagram for the  phase in Al 6061...... 39 Figure 36: Solubility of alloying elements in aluminum as a function of temperature for Al 6061, showing the maximum solubility at 540oC...... 40 Figure 37: STEM images of an Al 6061 powder particle heat treated for twenty-four hours at 230°C. a) and b) scale bars read 500 nm. c) A diffraction pattern taken from the dark precipitate indicated by the white arrow in b)...... 40 Figure 38: DSC scan for Al 6061...... 41 Figure 39: (Left) HAADF STEM image of Al 6061 particle heat treated at 540oC for 1 hour and prepared by focus ion beam. (Right) Higher magnification HAADF STEM image of heat treated Al 6061 particle showing precipitates at grain boundaries...... 42 Figure 40: HAADF STEM image of discrete particles at grain boundaries of heat treated Al 6061 (540oC for 1 hour); EDS maps of Al, Mg, Si, Fe, Cr, Mn...... 42 Figure 41: HAADF STEM image of discrete Fe-containing particles and larger Mg-Si particles at grain boundaries of heat treated Al 6061 (540oC for 1 hour); EDS maps of Al, Mg, Si, Fe, Cr, and Mn...... 43 Figure 42: Precipitate radius size distribution of Al-Fe-Si-Cr-Mn and Mg-Si particles for heat treated Al 6061 (540oC for 1 hour)...... 43 Figure 43: Plot of grain size as a function of particle size for as-atomized powder and heat treated powder (200° and 230°C)...... 44

Figure 44: Thermo-Calc prediction of volume fraction of two precipitates, Al6Mn and Al13Fe4, in Al 5056 as a function of time at 500oC...... 46 Figure 45: STEM image of Al 5056 particle heat treated at 500oC for 1 hour and prepared by focus ion beam...... 47 Figure 46: HAADF STEM image of discrete particles at grain boundaries of heat treated Al 5056 (500oC for 1 hour); EDS maps of Al, Mg, Si, Fe, Cr, and Mn...... 48 Figure 47: Precipitate radius size distribution of Al-Cr and Al-Fe-Mn particles for heat treated Al 5056 (500oC for 1 hour)...... 48 Figure 48: Graph of yield strength versus particle diameter for Al 6061 showing individual contributions from strengthening mechanisms of the additive strength model...... 51 Figure 49: Average nanohardness values for different ranges of sieved particles...... 52 Figure 50: Average nanohardness values for as-received and “degassed” samples of Al 6061...... 52 Figure 51: Variation of average nanohardness values for Al 6061 solutionized at 530oC for various times for particle sizes between 38-45 microns...... 53

70 Figure 52: Variation of average nanohardness values for Al 5056 solutionized for 1 hour at various temperatures for particle sizes between 38-45 microns...... 53 Figure 53: SEM image of a 5x5 nanohardness grid on a powder particle. Depth of indents is approximately 250 nm...... 54 Figure 54: Model and experimental comparison of the relationship between particle diameter and hardness for Al 6061...... 55 Figure 55: Plastic equivalent strain output from single particle impact model...... 56 Figure 56: Initial cross section cut of Al 6061 single splat...... 57 Figure 57: Cross section of Al 6061 single particle splat...... 57 Figure 58: Top right edge of cross section of Al 6061 single particle splat illustrating the jetting that occurred at the particle/substrate interface...... 58 Figure 59: (Left) Top down view on Al 6061 single particle splat and (Right) height profile of Al 6061 single particle splat measured in a confocal laser scanning microscope...... 58 Figure 60: (Left) SEM image of a grid of nanohardness indents on a single Al 6061 powder particle. (Right) SEM image of a grid of nanohardness indents on the spray direction face of a consolidated Al 6061 coating. The coating was etched with 0.5% hydrofluoric acid solution for 5 seconds to reveal particle boundaries. A blue line highlights one particle boundary...... 59 Figure 61: Graph of average nanohardness for Al 6061, Al 2024, Al 7075 and Al 5056. Average nanohardness for all sizes of powder is shown by the solid bar, while the dashed bar represents the consolidated sample...... 59 Figure 62: SEM image of single particle impact showing location of profile of nanoindentations across the particle and substrate...... 60 Figure 63: Plotted nanohardness and modulus values for line profile conducted in Figure 62. Negative position values are within the particle, x=0 is the particle/substrate interface, and positive position values are within the substrate...... 60 Figure 64: (Left) SEM image of top down view of the Al 6061 single particle splat, and (Right) cross- sectioned Al 6061 single particle impact prepared by plasma FIB...... 61 Figure 65: SEM image of cross-sectioned Al 6061 single particle impact showing locations of nanoindentations...... 61 Figure 66: Plotted nanohardness values for line profiles conducted in Figure 65. Negative position values are within the particle, x=0 is the particle/substrate interface, and positive position values are within the substrate...... 62 Figure 67: Output of the finite element single particle impact model, showing plastic equivalent strain, calculated using the Preston-Tonks-Wallace plasticity model...... 62 Figure 68: SEM image of nanoindentations overlaid on the output of the single particle impact model. 63 Figure 69: Model output (plastic equivalent strain) plotted against nanohardness from the line profiles illustrated in Figure 68. Linear trendlines, equations, and R2 values are reported...... 63

List of Tables Table 1: Experimental compositions and specifications of Al 6061 and 5056 used in this study in weight percent [8, 9]...... 10 Table 2: Thermophysical properties of atomizing Al 6061 in argon gas (from JMatPro® and CES EduPack software)...... 20 Table 3: Weight percentages of equilibrium phases predicted for Al 6061 at room temperature [25°C]. 24

71 Table 4: Phases predicted to form at the grain boundary of Al 6061 based on the grain boundary composition predicted by the segregation model in Figure 20 (Thermo-Calc)...... 25 Table 5: Average composition in atom percentage of the phases depicted in Figure 22...... 27 Table 6: Weight and volume percentages of equilibrium phases predicted for Al 5056 at room temperature [20°C]...... 32 Table 7: Phases and phase fraction predicted to form at the grain boundary of Al 5056 based on the grain boundary composition predicted by the segregation model in Figure 29 (Thermo-Calc)...... 33 Table 8: Comparison of experimental results and outputs of the Powder Production Stage models for Al 6061 and 5056...... 38 Table 9: Average composition in weight percent of phases found in Figure 37...... 41 Table 10: Average grain size for as-received Al 6061 (P344) and heat treated Al 6061 (540oC for 1 hour, HP344-540G)...... 44 Table 11: Average grain size for as-received Al 5056 (P743) and heat treated Al 5056 (500oC for 1 hour, HP743-500G)...... 49

72 APPENDIX List of Publications B. McNally, D.B. Cote, R. Handel, V. K. Champagne, Jr., R.D. Sisson, Jr., “Experimental Verification of Through-Process Modeling of Cold Spray Al Alloys,” Proceedings 2015 Microstructural Characterization of Aerospace Materials and Coatings Conference, Long Beach, CA, May 11-14, 2015.

B. McNally, D.B. Cote, R. Handel, D. Apelian, R.D. Sisson, Jr., “Powder Characterization for Through- Process Modeling of Cold Spray Al Alloys,” Proceedings POWDERMET 2015, San Diego, CA, May 18-20, 2015.

D. Belsito, B. McNally, L. Bassett, V. Champagne, Jr., R.D. Sisson, Jr., “'A through Process Model for Cold Sprayed Aluminum Alloys,” Proceedings 5th International Conference on Thermal Process Modeling and Computer Simulation Conference 2014, Orlando, FL.

D. Belsito, B. McNally, L. Bassett, V. Champagne, Jr., R.D. Sisson, Jr., “Through-process Modeling for Cold Spray Alloy Optimization,” Proceedings MS&T 2013, Montreal, Quebec, Canada. Conference Presentations D. Cote, B. McNally, J. Schreiber, V. K. Champagne, Jr., R. D. Sisson, Jr., “Experimental Verification and Computational Modeling for High Strain Rate Additive Manufacturing,” Materials Science and Technology 2016, Salt Lake City, UT, October 24-27, 2016.

D.B. Cote, B. McNally, V. K. Champagne, Jr., R. D. Sisson, Jr., “Advanced Powder Characterization and Modeling,” Cold Spray Action Team Conference, Worcester, MA, June 21-June 22, 2016.

B. McNally, D.B. Cote, V. Champagne, Jr., R D. Sisson, Jr., “A Predictive Strength Model with Experimental Characterization for Gas-Atomized Aluminum Powder,” 27th Advanced Aerospace Materials and Processes (AeroMat) Conference 2016, Bellevue, WA, May 23-26, 2016. Accepted.

B. McNally, D.B. Cote, V. Champagne, Jr., R D. Sisson, Jr., “Verification of a Predictive Strength Model for Gas-Atomized Aluminum Powder,” The Minerals Metals and Materials Society 2016 Annual Meeting & Exhibition, Nashville, TN, February 14-18, 2016. Accepted.

B. McNally, D.B. Cote, V. Champagne, Jr., R D. Sisson, Jr., “Experimental Characterization for Through- Process Modeling of Cold Spray Aluminum Alloys,” Materials Science & Technology Conference 2015, Columbus, OH, October 3-8, 2015.

B. McNally, D.B. Cote, R. Handel, D. Apelian, R D. Sisson, Jr., “Powder Characterization for Through- Process Modeling of Cold Spray Al Alloys,” PowderMet 2015 Conference, San Diego, CA, May 17-20, 2015.

B. McNally, D.B. Cote, L. Bassett, V. Champagne, Jr., R D. Sisson, Jr., “Experimental Verification of Through-Process Modeling of Cold Spray Al Alloys,” Microstructural Characterization of Aerospace Materials and Coatings Conference, Long Beach, CA, May 11-14, 2015.

B. McNally, D. Belsito, R D. Sisson, Jr., V. Champagne, Jr., “Experimental Characterization for Through- Process Modeling of Cold Spray Al Alloys,” The Minerals Metals and Materials Society 2015 Annual Meeting & Exhibition, Orlando, FL, March 16-19, 2015.

73 B. McNally, D. Belsito, V. Champagne, Jr., R D. Sisson, Jr., “Experimental Verification for Through-Process Modeling of Cold Spray Al Alloys,” The Minerals Metals and Materials Society 2015 Annual Meeting & Exhibition, Orlando, FL, March 16-19, 2015.

B. McNally, D. Belsito, L. Bassett, V. Champagne, Jr., R D. Sisson, Jr., “Experimental Verification of Through-Process Modeling of Cold Spray Al Alloys,” Materials Science & Technology Conference 2014, Pittsburgh, PA, October 12-16, 2014.

B. McNally, D. Belsito, V. Champagne, Jr., R D. Sisson, Jr., “Microstructural Evolution in Cold Spray Al Alloys 6061 and 5083,” AeroMat Conference, Orlando, FL, June 16-19, 2014.

B. McNally, D. Belsito, R D. Sisson, Jr., V. Champagne, Jr., “Microstructural Characterization and Analysis of Cold Spray Al Alloys,” The Minerals Metals and Materials Society 2014 Annual Meeting & Exhibition, San Diego, CA, February 16-20, 2014.

B. McNally, D. Belsito, V. Champagne, R.D. Sisson, Jr., “Through-Process Modeling of Cold Spray Alloy Optimization,” Nanotech for Defense Conference 2013, Tucson, Arizona, November 4-7, 2013.

B. McNally, D. Belsito, R.D. Sisson, Jr., V. Champagne Jr., “Analysis of Thermal Powder Pre-Processing for Cold Spray Consolidation,” Materials Science & Technology Conference 2013, Montreal, Canada, October 27-31, 2013.

B. McNally, D. Belsito, V. Champagne Jr. “Microstructural Analysis of Cold Sprayed Al Alloys,” Cold Spray Action Team Annual Meeting, Worcester, MA, June 18-19, 2013.

74