Influence of Impact Conditions on Feedstock Deposition Behavior of Cold-Sprayed Fe-Based Metallic Glass
Cold spray is a promising method by which to deposit dense Fe-based metallic glass coatings on conventional metal substrates. Relatively low process temperatures offer the potential to prevent the crystallization of amorphous feedstock powders while still providing adequate particle softening for bonding and coating formation. In this study, Fe48Mo14Cr15Y2C15B6 powder was sprayed onto a mild steel substrate, using a variety of process conditions, to investigate the feasibility of forming well-bonded amorphous Fe-based coatings. Particle splat adhesion was examined relative to impact conditions, and the limiting values of temperature and velocity associated with successful softening and adhesion were empirically established. Variability of particle sizes, impact temperatures, and impact velocities resulted in splat morphologies ranging from well-adhered deformed particles to substrate craters formed by rebounded particles and a variety of particle/substrate interface conditions. Transmission electron microscopy studies revealed the presence of a thin oxide layer between well-adhered particles and the substrate, suggesting that bonding is feasible even with an increased oxygen content at the interface. Results indicate that the proper optimization of cold spray process parameters supports the formation of Fe-based metallic glass coatings that successfully retain their amorphous structure, as well as the superior corrosion and wear-resistant properties of the feedstock powder.
Keywordsbulk metallic glass (BMG) alloy cold spray deposition behavior splat morphology
Iron (Fe)-based amorphous alloys are recognized as important engineering materials due to their high strength and hardness, superior wear and corrosion resistance, and excellent soft magnetic properties, in addition to their relatively low cost (Ref 1, 2). In addition, researchers have been successfully establishing compositions of glassy iron alloys that will retain an amorphous microstructure under significantly lower cooling rates than early melt-spun alloys (Ref 3). At this time, iron-based bulk metallic glasses with critical cooling rates as low as on the order of 100 K/s have been found in Fe-based alloy systems containing metalloids (B, C, Si, and P) (Ref 4). However, their use as structural materials has been hindered by their generally low ductility and the difficulties associated with obtaining cooling rates high enough to produce large gauge glassy material. As a result, alternate material systems by which to exploit the superior properties of these amorphous alloys continue to be explored, including their use as thick corrosion and wear-resistant layers on top of crystalline metal substrates. This study is motivated by the goal of forming such material systems, including a fully amorphous Fe-based metallic glass coating on a mild steel substrate, in order to widen the assortment of their feasible industrial applications.
Much work has been done toward producing amorphous coatings using conventional deposition and thermal spray methods, but the majority of it has only achieved partial success due to the occurrence of oxidation during processing that provides nucleation sites (Ref 5-11). In many cases, the coatings are thin (~ 100 μm thick) and do not maintain a purely amorphous structure, having face-centered cubic Fe-based dendrites and fine crystalline precipitates within an amorphous matrix. The cold spray process is a novel alternative for producing dense coatings through an impaction process based on the acceleration of metallic powder particles toward a substrate. The feedstock powder particles travel at high velocities in an unreactive gas, such as nitrogen, and collide with the substrate such that their kinetic energy promotes coating formation by plastic deformation (Ref 12). As a solid-state process, cold spray can utilize temperatures that are consistently lower than the melting point of the powder material (Ref 13). As a result, successful cold-sprayed coatings typically do not demonstrate oxidation or residual stresses and are free from the phase transformations, microstructural changes, and grain growth that occurs in thermal spray processes that involve melting (Ref 12).
At this time, numerous conventional metals and alloys have been successfully synthesized as cold-sprayed coatings without oxidation or phase transformation, and characterized on the basis of various mechanical properties and corrosion resistance (Ref 14-18). An understanding of the effectiveness of cold spray for producing coatings from amorphous feedstock powders, however, is far less developed. The balance of particle impact temperatures and velocities is complicated by the brittleness of metallic glasses at ambient temperature (Ref 19), making it less clear if shear instabilities at the particle or substrate surfaces will cause the necessary local softening for the formation of strong and intimate interface bonding. The criteria for cold spray bonding require detailed study and the development of in-depth understanding of the plastic flow of these amorphous powders at very high strain rates.
Studies have shown that successful bonding of cold-sprayed powder can only be achieved when impacting particles reach a minimum critical velocity, which depends on powder and substrate properties (Ref 20, 21). The plastic deformation of feedstock particles is often insufficient to achieve bonding when their impact velocity is below the critical value. There is also experimental evidence that higher impact velocity does not always correlate with a larger fraction of adhered particles (Ref 22). This is due to a rebound phenomenon that typically occurs when particles either fail to reach the critical velocity or exceed the upper limit of impact velocity, known as the erosion velocity. In addition, when the velocity exceeds the erosion value, hydrodynamic penetration leads to strong erosion rather than bonding. These observations suggest that successful cold-sprayed particle adhesion is feasible when impact velocities fall within the two velocity limits (critical and erosion), and particle impact temperatures are favorable for deformation and bonding. This concept of a deposition window was introduced by Schmidt et al. (Ref 20) and previously developed for the cold spray of crystalline feedstock powder.
The current study investigates the critical velocity for cold-sprayed Fe-based amorphous powder Fe48Mo14Cr15Y2C15B6, (known as SAM1651) using individual particle impact (splat) tests. Splat tests are also used to study the onset of coating deposition as they allow for the observation of the behavior of individual feedstock particles (Ref 15). The developed temperature analysis accounts for the fact that amorphous metals do not behave in the same manner as crystalline materials with regard to thermal softening and impact mechanics. The critical conditions for particle adhesion are calculated on the basis of the experimentally determined hardness of amorphous SAM1651, the mass density (ρ) of the alloy, and an empirically defined softening temperature. The window of favorable impact conditions is validated using splat test results and detailed analysis of particle adhesion. Splat morphologies are characterized to study coating deposition onset and explore the cold spray feasibility of this Fe-based amorphous alloy, including consideration of optimal process parameters. The study contributes to the understanding of underlying mechanisms associated with the cold spray deposition of metallic glass powder. This is expected to advance efforts to form fully amorphous coatings on crystalline metal substrates and thereby expand the list of potential industrial applications that will benefit from the superior properties of metallic glasses.
Feedstock Powder Characterization
Amorphous Fe48Cr15Mo14C15B6Y2 (at.%), or SAM1651, was selected for this study due to its superior corrosion resistance and high hardness (Ref 23). The SAM1651 powder was commercially manufactured (Carpenter Powder Products, Pittsburgh, PA) by inert gas atomization and sieved to a target size range of 10-40 μm. Characterization of the feedstock powder was conducted using laser diffraction (Malvern Mastersizer, Malvern Instruments Inc., Westborough, MA) to verify the particle size distribution, x-ray diffraction (XRD) (PANalytical X’Pert Pro MPD, PANalytical, Almelo, Netherlands) to assess the crystallinity of the initial powder, and scanning electron microscopy (SEM) (JEOL JSM 500-type, JEOL, Inc., Peabody, MA) to inspect the initial powder morphology. Characteristic temperatures were determined by differential scanning calorimetry (DSC) (TA Instruments Q2000 MDSC, New Castle, DE) and differential thermal analysis (DTA) (TA Instruments Q600 STD, New Castle, DE).
Cold Spray Splat Tests
A commercial Plasma Giken cold spray system (PCS-304, at Mid-America Aviation, Webster, MA), with a maximum operating temperature of 1000 °C, was used to spray SAM1651 powder onto AISI 4140 steel coupons (25.4 × 25.4 mm). 4140 steel was selected due to its extensive use in industrial applications and the perceived benefit that a successful coating would improve its limited corrosion resistance. Prior to spray adhesion testing, the steel coupons were polished to a mirror-like finish (Ra < 1.5 μm).
Powder particles were preheated while passing through a 90-mm-long pre-chamber, together with the nitrogen carrier gas, prior to entering the nozzle. Splat tests were all conducted with gas pressure held constant at 4 MPa. Gas temperatures, in contrast, were systematically varied, so each run was completed at 850, 900, 950, or 1000 °C. The variation of gas temperature and the assortment of powder particle sizes resulted in a wide range of particle impact temperatures and velocities, as predicted by the developed computational model.
In order to optimize the cold spray process for the deposition of metallic glass powder particles, the effects of process parameters on particle impact conditions are needed. This information is then utilized to compute the kinetic and thermal energies of the particles just prior to substrate collision and to investigate cold spray adhesion, rebounding, and deposition efficiency. The accurate experimental measurement of particle temperatures and velocities at impact is difficult, however, due to the high speed and small size of the particles and the fact that their paths and locations in the gas stream affect their velocities. Experimental isolation of a specific particle at a defined location in the powder stream is also very difficult to achieve with accuracy. For these reasons, models have been developed to predict the impact conditions of cold-sprayed powder particles.
An analytical model based on one-dimensional, isentropic, gas-dynamic equations of particle flow in a fluid stream was previously developed by Champagne et al. (Ref 24, 25) and is presented in the literature together with experimental validation data. While the agreement between experimentally measured impact conditions and those predicted using the analytical model is reasonable, there are differences between them that are attributed to frictional effects and particle distribution within the stream. These effects, as well as the detailed particle behavior between the nozzle exit and the substrate surface in general, can be more accurately predicted with a computational fluid dynamic (CFD) model. For this study, a CFD model with the ability to parse the impact conditions of individual particles in specific locations within the non-uniform stream was developed to predict the gas and particle temperatures and velocities.
ANSYS® Fluent (Ref 26) was used to model the influences of nozzle geometry and significant process parameters, such as process gas, gas temperature and pressure, spray material type, and particle size, on particle impact conditions. The system fluid was modeled as nitrogen gas, and the powder particles were represented as a discrete phase using the characteristics of SAM1651 metallic glass. A two-dimensional axisymmetric model was implemented on the basis of the cylindrical nature of the de Laval nozzle and the associated flow through it. Dimensions were set to correspond to the previously described Plasma Giken nozzle, including consideration of a pre-chamber that allows the gas flow to be fully developed upon entering the converging section of the nozzle. The model includes flow exiting the nozzle and impacting a hard, solid wall at the defined standoff distance. The metallic glass particles are absorbed by the substrate; however, the gas is able to rebound from it. The flow is considered to be turbulent and is modeled using the standard k-omega equations (Ref 27). The gas is assumed to be compressible, and the solution was obtained using the density-based version of the Navier–Stokes equations to allow for changing density with Mach number.
To determine drag, particles were assumed to be spherical and monodisperse. The model defines gas pressure and temperature at the input of the pre-chamber. At the nozzle exit, air is assumed to be at ambient temperature and pressure. The model assumes that the particles have a small input velocity to enter the system; the particle velocity then equilibrates with the gas velocity upon entry to the nozzle. Due to the physical size and thermal properties of the particles, the model predicts that the temperature of the particles reaches the gas temperature by the time they exit the pre-chamber. The fluid velocity stays relatively constant throughout the pre-chamber and starts to increase slightly in the converging section of the nozzle. The velocity increases greatly at the throat as the fluid reaches sonic speed, and further increases to supersonic speeds as the nozzle diverges. After the initial high rate of increase, the velocity continues to increase at a decreasing rate until the nozzle exit is reached. The gas velocity fluctuates after exiting the nozzle due to pressure waves that are created from the fluid rebounding after impact with the substrate.
Specimen Inspection and Analysis
To study the effect of impact velocity and temperature on the adhesion of cold-sprayed metallic glass powder, the cold-sprayed specimens were examined using SEM. Each specimen includes a resulting band of sprayed particles that is approximately 2 mm wide and spans the full 25.4 mm length. This band includes multiple splats of different powder particle sizes, representing a variety of impact conditions. SEM inspection was used to count the number of adhered particles and rebound craters within four rectangular areas (each 500 μm × 500 μm) on each specimen; each area was randomly selected within one of the four quarters of the 25.4 mm length. These quantities were then categorized on the basis of particle or crater diameter, using 5-μm-size bins, and the associated impact velocities and temperatures were predicted by the described computational model. The ratio of bonds, defined as the number of adhered particles divided by the total number of impacted particles (craters + bonds) in a unit area of impact surface, was subsequently calculated for each spray condition based on the SEM results. Quantitative results were used to develop an empirical model of critical velocity and a temperature/velocity window of successful particle deposition.
Single-particle splats were also individually examined by SEM and categorized in one of three identified morphology types. Several splats were additionally sectioned using a focused ion beam (FIB) (FEI Quanta 200 3D Dual Beam FIB, Hillsboro, OR) and further studied by SEM in an effort to corroborate the level of perceived deformation and adhesion. Particle/substrate interfaces were also examined using transmission electron microscopy (TEM) (FEI Talos F200X STEM, Hillsboro, OR) in an attempt to characterize the quality of bond formed during cold spray. A layer of carbon was deposited on each particle in order to prevent milling artifacts and damage, and thin foils (~ 100 nm) were subsequently prepared using a FIB. Each of the foils was examined on the FEI Talos F200X scanning/transmission electron microscope (S/TEM) with a field emission source at 200 kV to investigate local interface phenomena.
Results and Discussion
Feedstock Powder Properties
Computational Model Validation
The two models predict similar velocity and temperature trends, but display distinct differences due to their respective underlying assumptions. The computational model shows fluctuations in the gas velocity and temperature upon exiting the nozzle due to pressure waves that result from the impact of gas flow against the substrate material. The analytical model does not capture this behavior. The analytical model predicts higher gas and particle velocities than the finite element model, in part due to the assumption that flow is inviscid in the analytical model. The computational model accounts for viscosity, which is small for nitrogen gas, but may be relevant to this discrepancy.
The computational model predictions of temperature also differ from those of the analytical model (Fig. 5b), with the 1D model predicting higher particle temperature and lower gas temperature than the computational model. In both models, the fluid temperature is high at the nozzle entrance, decreases at a rapid rate through the throat, and continues to decrease at a reduced rate as it moves through the diverging section of the nozzle. Upon exiting the nozzle, the pressure waves cause fluctuations in gas temperature. This fluid behavior also affects the metallic glass powder particles within the gas stream, whose velocity and temperature drop at a constant rate upon exiting the nozzle. The differences between the predicted temperatures of the two models are in part due to the way in which the heat transfer between the particles and the fluid is modeled. The analytical model uses a forced convection law, while the computational model uses natural convection and balances the energy equation.
The computational model is used to predict impact conditions in this study, as it better captures the behavior associated with pressure waves and it accounts for viscosity.
Effect of Impact Conditions and Particle Size on Adhesion
The size effect was further studied with respect to the adhesion results and the velocity and temperature predictions offered by the computational model. The data were consolidated on the basis of particle impact velocities and temperatures, which are significantly affected by particle size (Ref 28). Smaller particles achieve greater velocities within the nozzle due to their low mass, but this low mass also causes them to be more significantly affected by the shock waves produced as the gas flow stream rebounds from the substrate surface. In contrast, the larger particles achieve greater momenta, causing them to be less significantly affected by the shock wave, and thus continue to increase speed slightly from the nozzle exit until impact with the substrate (Ref 25).
Particle size also has an effect on impact temperature. As shown in Fig. 5(b), the gas temperature decreases significantly as it flows to the nozzle exit from the throat. The temperatures of smaller particles decrease more rapidly toward that of the gas, while the greater thermal mass of larger particles renders them less affected by the decrease in gas temperature. Similarly, the use of a pre-chamber is the most important for larger particles as they need more time than that spent in the converging length of the nozzle to reach the gas temperature. Upon exiting the nozzle, particle temperatures generally increase as they pass through the shock wave. This effect is more significant for smaller particles.
Temperature and Velocity Limits of Successful Deposition
The critical velocity for cold spray particle adhesion depends on a wide variety of factors, including the properties of feedstock powder and substrate materials, particle size and geometry (Ref 25), particle temperature (Ref 32), particle oxygen content (Ref 33), and substrate preparation (Ref 34, 35). Previous work suggests that the most significant of these factors are the temperature and thermomechanical properties of the sprayed material (Ref 36). In order to achieve successful deposition using cold spray, the powder particles must attain the critical velocity vc without exceeding the erosion velocity ve and achieve an impact temperature that permits the material softening.
The current study uses this same framework as the basis of an investigation of the limiting velocities associated with cold-sprayed amorphous SAM1651. For this feedstock powder, the temperature analysis was altered to account for the fact that amorphous metals do not work harden nor do they behave in the same manner as crystalline materials with regard to thermal softening and impact mechanics. Similarly, the melting temperature Tm is not a reasonable reference temperature for the softening of this alloy (Ref 38). The strain rate- and temperature-dependent yield strength expression was therefore adjusted to eliminate the effect of strain hardening and to include a reference softening temperature Ts that differs from Tm. The developed expression for temperature-dependent flow stress is described elsewhere (Ref 28).
The experimental adhesion results obtained for specimens with Tgas of 900, 950, and 1000 °C are presented in Fig. 9(b) and appear to suggest the same general trends as those predicted by the model. For example, for a gas temperature of 900 °C, the model predicts the successful deposition of SAM1651 particles with diameters between 14 and 22 μm and between 3 and 3.75 μm. The experimental results for the case of 900 °C gas indicate the highest proportion of adhered particles were those with diameters between 12 and 23 μm, with much smaller proportions of adhered particles of other sizes. This is in general agreement with the model, but some amount of error is also present. The empirical model is based on the CFD predictions for particle impact conditions, which include an undefined level of error, and the experimental adhesion results are manually gathered from four sampled regions on each specimen, which include human inspection error and are not based on the entire population of adhered particles of the specimen. As such, the agreement between the model and the experimental results is currently in the context of the limitations of both. More comprehensive CFD model validation and additional experimental results are expected as part of full coating studies.
Characterization of Splat Morphologies
The adhesion model and experimental validation suggest that the formation of bonds between cold-sprayed metallic glass particles and the steel substrate is influenced by the impacting temperature and velocity of the particle, which depend on the particle size and process parameter settings. More detailed information regarding the particle/substrate interface and the deformation of both particle and substrate has been gleaned from inspection of splat morphologies, which also assists in assessing the likelihood that particle adhesion will lead to successful coating formation. SEM was used to examine numerous splats associated with each of the tested spray conditions, and three broad impact morphology types were revealed. This examination process and impact categorization is detailed and previously presented elsewhere (Ref 30). A brief summary is provided again here, to categorize impact morphologies as (1) deformed/adhered particles, (2) undeformed/adhered particles, and (3) substrate craters formed by rebounded particles.
Transmission Electron Microscopy
The results of this study demonstrate the necessity of controlling cold spray process parameters in an effort to achieve desired impact conditions and promote the coating formation of Fe-based metallic glass powders. Adequate thermal softening of the Fe-Cr-Mo-C-B-Y metallic glass powder particles is required to support their deformation upon impact with the steel substrate, but temperature must be regulated to also ensure that phase transformations, microstructural changes, and grain growth do not occur. In order to achieve this softening, both computational and experimental results indicate that cold-sprayed Fe-based amorphous powders must be heated to a temperature that is within the supercooled liquid region. Results also suggest that effective impact velocities for these powders are attainable using a nitrogen carrier gas at a pressure of 4 MPa.
The results of the single-impact splat studies validate the developed empirical model of temperature and velocity limits associated with successful particle adhesion. Particle deformation is improved by the use of a pre-heating chamber and subsequently higher impact temperatures, which vary on the basis of particle size and process gas temperature. Splat examination suggests that shear instabilities and plastic flow are enhanced above the glass transition temperature, which is useful as the basis for defining the appropriate reference temperature for the softening of SAM1651 alloy, determined to be 1.4 Tg. TEM studies suggest that the deformed and adhered particles characterized as favorable cold-sprayed impacts retain their amorphous structure, and hence, their superior properties. TEM results also provide evidence that a thin oxide layer is often present between a well-adhered particle and the substrate, suggesting that bonding is still feasible with an increased oxygen content in regions of the splat/substrate interface. These results are favorable with regard to the expected success of forming thick cold-sprayed SAM1651 coatings that remain amorphous.
This study provides evidence that the deposition of cold-sprayed Fe-based amorphous coatings appears promising with the proper optimization of process parameters. Such coatings are expected to retain the exceptional properties of the feedstock powder, including exceptional hardness and wear and corrosion resistance. Full coating tests are currently planned to allow for the study of the coating/substrate bond and the detailed characterization of coating properties.
The authors thank the following colleagues for their collaboration in various aspects of this project: Ozan Ozdemir (South Dakota School of Mines and Technology) for performing DSC and DTA; Xiaojun Gu (Bucknell University) for casting bulk ingots and performing compression testing; Donna Ebenstein (Bucknell University) for nanoindentation; Dennis Helfritch (TKC Global at Army Research Laboratory) for access to spray facilities. This work involved shared facilities supported in part by the Penn State MRSEC, Center for Nanoscale Science, under the NSF award DMR-1420620. The authors acknowledge support from Bucknell University through a Dean’s Fellowship (Wright) and a Presidential Professorship (Ziemian).
- 3.R. Babilas and R. Nowosielski, Iron-Based Bulk Amorphous Alloys, Arch. Mater. Sci. Eng., 2010, 44(1), p 5-27Google Scholar
- 13.A. Papyrin, V. Kosarev, S. Klinkov, A. Alkhimov, V. Fomin, High Velocity Interaction of Particles with the Substrate. Experiment and Modeling, Cold Spray Technology, 1st ed., Amsterdam, Elsevier Science, 2007, 2, 33-108.Google Scholar
- 16.O. Nooirinah, A. Rohana, H.N. Nazrul, and M. Suhana, Supersonic Particle Deposition as Potential Corrosion Treatment Method for Helicopter Part in Malaysia, Int. J. Adv. Res. Eng. Technol., 2012, 3(2), p 275-279Google Scholar
- 22.J.W. Wu and H.Y. Fang, Critical and Maximum Velocities in Kinetic Spraying, Thermal Spray 2006: Building on 100 Years of Success, B.R. Marple, M.M. Hyland, Y.-C. Lau, R.S. Lima, and J. Voyer, Ed., ASM International, Seattle, 2006, p 396Google Scholar
- 25.D. Helfritch and V. Champagne, Optimal Particle Size for the Cold Spray Process, Thermal Spray 2006: Building on 100 Years of Success, B.R. Marple, M.M. Hyland, Y.-C. Lau, R.S. Lima, and J. Voyer, Ed., ASM International, Seattle, 2006, p 396Google Scholar
- 26.ANSYS Fluent User’s Guide, Release 15, Canonsburg, PA, 2013Google Scholar
- 27.B. Andersson, R. Andersson, L. Håkansson, M. Mortensen, R. Sudiyo, and B. van Wachem, Computational Fluid Dynamics for Engineers, 1st ed., Cambridge University Press, Cambridge, 2012, p 202Google Scholar
- 28.D.E. Cipoletti, C.W. Ziemian, W.J. Wright, D.J. Helfritch, K.V. Haile, M.N. Okwara, and K.A. Hetherington, Critical Velocity Window for the Deposition of Iron-Based Metallic Glass Particles using Cold Spray, Thermal Spray 2015: Proceedings from the International Thermal Spray Conference, A. McDonald, A. Agarwal, G. Bolelli, A. Concustell, Y.-C. Lau, F.-L. Toma, E. Turunen, and C. Widener, Ed., May 11-14, 2015 (Long Beach, California, USA), ASM International, 2015, p 1225Google Scholar
- 30.C.W. Ziemian and W.J. Wright, An Experimental Study of Cold Sprayed Fe-Based Metallic Glass Powder Particles, Proceedings of 30th International Conference on Surface Modification Technologies (SMT30), June 29-July 1, 2016 (Milan, Italy)Google Scholar
- 34.P. Richer, B. Jodoin, K. Taylor, E. Sansoucy, M. Johnson, and L. Ajdelsztajn, Effect of Particle Geometry and Substrate Preparation in Cold Spray, Thermal Spray 2005: Thermal Spray Connects: Explore Its Surfacing Potential!, E. Lugscheider, Ed., DVS-German Welding Society, Basel, 2005, p 193-198Google Scholar
- 37.G.R. Johnson and W.H. Cook, A Constitutive Model and Data for Metals Subjected to Large Strains, High Strain Rates and High Temperatures, Proceedings of the 7th International Symposium on Ballistics, April 19-21, 1983 (The Hague, Netherlands), 1983, pp 541-547.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.