Introduction

Additive manufacturing (AM) is a rapidly evolving family of technologies capable of producing bulk, free-standing components. Their layer-by-layer buildup mechanisms allow fabrication of complex geometries from polymers, ceramics, or metals. Typical examples of AM technologies include laser- or electron beam-powder bed fusion (Ref 1, 2), wire arc additive manufacturing (Ref 3), or directed energy deposition (Ref 4). As such thermal energy processes are based on melting of the feedstock material, several detrimental phenomena could be triggered during the materials buildup, including oxidation or phase changes (Ref 5).

Cold spray additive manufacturing (CSAM) is a rapid, solid-state deposition process initially developed for surface applications (Ref 6, 7). In the process, material feedstock in the form of micron-sized particles is accelerated by compressed gas (typically nitrogen) and propelled toward a target substrate. Upon their impact, the particles experience severe, extreme-rate plastic deformation and conform to the underlying material (Ref 8). The bonding of the two partner materials is triggered at three different levels: mechanical anchoring exploiting the (frequently artificially introduced) surface asperities, vortex intermixing triggered by excessive plastic deformation, and, most of all, metallurgical bonding triggered by oxide shell rupture and removal and an intimate metal-to-metal contact upon the impact (Ref 9,10,11). In CS, the solid character of the process allows deposition of pure metals, including the oxygen-sensitive elements such as W, Cu, Al, Ti, and Zn (Ref 12,13,14).

In general, the CSAM deposits exhibit negligible porosity and particle cohesion superior to that of the high-temperature thermal spray processes (Ref 15). Also, they exhibit superior machinability and in fact can be tooled in their as-deposited state (Ref 15). These factors present—if critical—qualifications for the needs of the AM industry. Indeed, given the remarkable advantages of CSAM over the fusion-based additive manufacturing methods such as high throughput (typically in the order of ∼ 101 kg/h) and unique mode of operation below melting point of the materials, CSAM is being slowly introduced as a competing technology in this market (Ref 16).

Owing to its buildup principles, CSAM deposits exhibit an inimitable microstructure, where the individual particles are severely deformed at extremely high strain rates (Ref 17). The strain rate is the feature that differs CS from other severe plastic deformation methods such as high-pressure torsion or equal channel angular pressing where the deformation is significantly higher, but strain rates are several orders of magnitude smaller (Ref 18). The increase in the density of dislocations (as well as other lattice defects) associated with the plastic deformation of CSAM materials is tremendous, frequently leading to a significant rise in yield point and strength of the deposits (Ref 7), a phenomenon observed in cold-work processes. Naturally, in the as-deposited state, this leads to a trade-off and the ductility and further deformability of the CSAM materials is significantly reduced. In cases, where this might present an issue, post-treatment by various annealing methods is employed (Ref 7).

Positron annihilation spectroscopy (PAS) is a nondestructive technique for the characterization of open volume lattice defects in solids (Ref 19). PAS is sensitive not only to areal defects such as grain boundaries or line defects such as dislocations, but also to point defects, such as vacancies or vacancy clusters which are below the resolution limit of transmission electron microscopy. For a detailed description of the technique and its capabilities, the readers are kindly referred to a comprehensive review paper from 2018 (Ref 20).

Introducing the method using a more concise description in this paper, PAS offers several experimental techniques, each suitable for the characterization of different properties of materials. These include:

  • Positron lifetime (LT) spectroscopy, which enables identification of open volume defects and determination of their concentration (Ref 21).

  • Coincidence Doppler broadening (CDB) spectroscopy, which provides information about local chemical environment of the defects (Ref 22).

  • Angular correlation of annihilation radiation (ACAR), suitable for investigation of the electronic structure of materials (Ref 23).

  • Variable energy slow positron annihilation spectroscopy (VEPAS), which enables determination of the defect depth profile and measurement of the positron diffusion length (Ref 24).

  • Age momentum correlation (AMOC), which allows study of free volumes in polymers and porous materials and also chemical reactions involving positrons and positronium (Ref 25, 26).

Of these, the LT spectroscopy represents the most important method for defect characterization in metals. Here, positrons emitted from a radioactive source, most frequently 22Na, are implanted into the solids. After the implantation, the positrons are slowed down to thermal energies before they decay. Thermalized positrons are eventually annihilated by one of the surrounding electrons (Fig. 1). The positron annihilation rate is determined by the overlap of the positron density and the density of electrons. If the positron is trapped in an open volume defect such as a vacancy or dislocation, its annihilation rate is significantly decreased because of the reduced electron density in the defect. From the measured positron lifetime (inverse of the positron annihilation rate), it is possible to determine the type of lattice defect as well as the concentration of defects. For metals, the mean penetration depth of positrons emitted by 22Na is in the order of several tens of micrometers (Ref 20), and PAS with positrons from 22Na source, therefore, provides information about bulk properties of the studied samples.

Fig. 1
figure 1

Schematic illustration of positron lifetime (LT) spectroscopy and coincidence Doppler broadening (CDB) (Ref 20). Reprinted from Journal of Materials Science & Technology, Vol. 34, J. Čížek, Characterization of lattice defects in metallic materials by positron annihilation spectroscopy: A review, p. 577–598, Copyright 2018, with permission from Elsevier

Full size image

Considering its advantages, PAS represents an excellent tool for characterization of lattice defects. It does not require any special technique for sample preparation, exhibits a high sensitivity to open volume defects, and provides non-local information averaged from a macroscopic volume, and its non-destructive character allows repeated measurement of samples in order to monitor the development of defects during various treatments such as annealing, deformation, hydrogen loading, irradiation, etc. For cold spray and thermal spray, PAS can be used for both coatings as well as thick deposits (Ref 27).

In this first study, PAS was used for characterization of CSAM materials on the atomic scale. The character of vacancies and dislocations was studied, and their respective densities were quantified. Given its typical characterization dimensions, the results provided by PAS are complementary to the microstructural characterization that is typically obtained using SEM, EBSD, and even TEM methods. This said, the PAS results could aid in our understanding of the complex phenomena related to the extensive plastic deformation in the CS process.

Experimental

Materials

To assess the potential of using PAS for evaluation of cold sprayed deposits, four different materials have been tested: Al (20-50 µm, 99.99% purity, Toyal Europe, France), Cu (− 15 µm, Safina, Czech Republic), Ni (15-35 µm, Amperit 176.068), and Ti (− 45 µm, Amperit 155.090, both H.C. Starck, Germany). Given their different manufacturing methods, their morphologies ranged from spherical (gas atomized Cu, Ni), to semi-spherical (water atomized Al), to angular (fused and crushed Ti), as presented and discussed in more detail in our previous paper (Ref 28). The powders were polycrystalline, i.e., each individual particle consisted of multiple grains.

As a substrate material, AA6061 Al alloy was used. To prevent potential bending during the deposition, rather thick coupons of dimensions 50 × 50 × 10 mm3 were used. No grit blasting was performed prior to the cold spray deposition, but the surfaces were chemically cleaned by isopropanol.

Sample Preparation and Characterization

High-pressure Impact Innovations ISS 5/11 system with nitrogen as process gas was used for the deposition. Optimized conditions obtained in our previous studies were used to deposit samples of 50 × 50 mm2 and thickness exceeding 11 mm. More details on the deposition process can be found in our previous paper (Ref 28) as identical deposits were also used in this study.

From the deposits, cross-section samples were prepared using the following standard metallography procedure. The samples were ground with emery papers up to #4000 grit, followed by 1-µm diamond paste polishing. For the channeling contrast imaging, an additional polishing was performed in colloidal silica using the Vibromet vibratory polisher (Buehler, USA) for long processing times (16 h). The microstructure was then observed using SEM (Zeiss EVO MA 15, JEOL JSM-5000, both LaB6 cathode) in back-scattered electrons mode (BSE) as well as in a channeling contrast mode (ECCI) to emphasize the grain size orientation (Ref 28). The porosity of the deposits was determined using the SEM image analysis method (ImageJ software, NIH), with five SEM images used for each metal. This was done at various distances from the deposit–substrate interfaces to comprehend potential heterogeneity. The described procedures are identical to those used in our previous paper (Ref 28).

To further understand the development of microstructure of the CSAM deposits, crystallographic analysis was performed using electron back-scatter diffraction method (EBSD). FEG-SEM (Carl Zeiss ULTRA plus) equipped with HKL Nordlys (Oxford Instruments, UK) detector was used. The details on the analysis setup can be found in our previous paper (Ref 28). The EBSD analysis involved acquiring traditional inverse pole figures (IPF maps), as well as kernel average misorientation (KAM, also called local misorientation maps) and strain contouring maps (Ref 34), (114.5 ± 0.1) ps (Ref 35), (108 ± 1) ps (Ref 36), and (152 ± 1) ps (Ref 37). Both τ2 and τ3 lifetimes in our CSAM deposits are higher than the bulk positron lifetime of the corresponding materials. This testifies that the components with lifetimes τ2 and τ3 come from positrons trapped at defects. The free positron component was observed only in the Al deposit. In the other CSAM deposits, the concentration of defects is so high that virtually all positrons are trapped at defects (so called saturated positron trap**).

The lifetime τ2 is comparable with lifetimes reported for dislocations in the corresponding materials (Ref 34, 38,39,40,41). Hence, the component with lifetime τ2 represents a contribution of positrons trapped at dislocations introduced by plastic deformation during the CSAM deposition. Note that positron trap** at dislocations is a two-state process (Ref 42). The dislocation line itself is only a shallow positron trap with positron binding energy in the order of ∼ 0.1 eV (Ref 43). Positron is weakly localized in a dislocation line and rapidly diffuses along it until it is finally trapped in a vacancy bound to the dislocation. It is energetically favorable for a vacancy to be located in the compressive elastic field around dislocation, i.e., a vacancy located in the compressive elastic field in the vicinity of dislocation has lower energy than a vacancy located in the regular lattice (Ref 43). Hence, the positron is finally annihilated in a vacancy squeezed by the compressive elastic field of the dislocation. Consequently, lifetimes of positrons trapped at dislocations in metals are typically slightly shorter than lifetimes of positrons trapped in bare vacancies (Ref 44). Because of the saturated positron trap**, only the lower limit of the mean dislocation density in Cu, Ni, and Ti deposits can be estimated from the LT data using the three-state positron trap** model (Ref 45). With increasing concentration of defects, the intensity I1 of the free positron component decreases, and the lifetime τ1 shortens. The free positron component cannot be reliably resolved in the LT spectra when its intensity I1 falls below approximately 5%, and the lifetime τ1 becomes shorter than approximately 20 ps. It happens for the mean dislocation density exceeding 5 × 1014 m−2, which can be, therefore, considered as a threshold value for saturated positron trap** in dislocations. Hence, the mean dislocation density in Cu, Ni, and Ti deposits is at least 5 × 1014 m−2. In the case of the Al deposit, such saturated trap** does not occur. The free positron component with lifetime τ1 was detected in its LT spectra with a significant intensity of 44%. Hence, the density of defects in the Al deposit is significantly lower compared to the other three samples (Cu, Ni, and Ti). It indicates a significant dynamic recovery of deformation-induced defects in the Al deposit (Ref 46). Using the simple trap** model from (Ref 45) and the specific positron trap** rate for dislocations in Al of 0.5 × 10−4 m2 s−1 (Ref 47), the mean dislocation density can be calculated as (3.0 ± 0.4) × 1013 m−2. This value is more than one order of magnitude lower than the mean dislocation density reported by Liu Tian et al. (Ref 48) for Al-based coatings with added Cu. In that paper, the authors used line profile analysis of XRD reflections and obtained the dislocation density in the range of 4.3-7.5×1014 m−2 (correlated to the added Cu content of 2-5 wt.%). The difference from our dislocation density value is not surprising since in Ref 48, a low-pressure CS system was used at 1.21 MPa and 225 °C (cf. our 5.0 MPa and 500 °C used to deposit pure Al (Ref 28)). The dynamic recovery of defects was clearly more pronounced in our samples because of the higher process temperature that triggers an increased heat intake into the sprayed material and also because we used pure Al where recovery of dislocations occurs easier than in the Al-Cu alloys.

The longest component with lifetime τ3 can be attributed to positrons trapped at small vacancy clusters. The presence of vacancy clusters gives clear evidence that plastic deformation in the course of the CSAM deposition introduces not only dislocations but also vacancies created by non-conservative motion of dislocations, e.g., climbing and crossing of dislocations. Since vacancies in Al, Cu, and Ni are mobile at ambient temperature, they either disappear by diffusion to sinks at grain boundaries or bind together with other vacancies forming small vacancy clusters which have significantly lower mobility than monovacancies. The driving force for agglomeration of vacancies is reduction of the surface energy since the surface of vacancy clusters is smaller than sum of surfaces of the monovacancies the cluster consists of.

The mean size of vacancy clusters can be estimated from comparison of the lifetime τ3 with ab initio calculations of lifetimes of positrons trapped in vacancy clusters consisting of various numbers of vacancies (Ref 51). Figure 8 shows calculated lifetime of positrons trapped at vacancy clusters of various sizes plotted as a function of the cluster size. From comparison of the ab initio calculations with lifetimes τ3 measured in the experiment, one can conclude that Cu and Ni deposits contain vacancy clusters of similar size consisting of 2-4 vacancies (the last column in Table 1). The Al deposit contains larger vacancy clusters consisting of around 15 vacancies. The larger size of vacancy clusters in the Al deposit is due to the higher mobility of vacancies in Al compared to Cu and Ni. Interestingly, no vacancy clusters were detected in the CSAM Ti deposit. Instead, it only contains the dislocation component with lifetime τ2 = 175 ps. This is likely connected with spatial arrangement of dislocations. Metals with hcp structure such as Ti have only three active slip systems <001> {001} at ambient temperature. Hence, dislocations have limited degrees of freedom for slip and get stuck with each other easily. As a consequence, a forest of tangled, uniformly distributed dislocations is formed in Ti subjected to severe plastic deformation, as also observed in rolled or high-pressure torsion counterparts (Ref 52, 53). On the other hand, metals with fcc structure (Al, Cu, and Ni) have 12 slip systems \(<\!1 \bar{1} 0\!>\) {111}, and dislocations have more directions for slip. As a consequence, dislocations in plastically deformed fcc metals often form a cellular structure, where cells with low dislocation density are separated by dislocation walls with a high density of dislocations. The high density of uniformly distributed dislocations in hcp metals facilitates diffusion of vacancies into sinks due to pipe diffusion along the dislocation lines. In contrast, in fcc metals (Al, Cu, and Ni), diffusion of vacancies into sinks via pipe diffusion occurs in dislocation walls only, while inside the cells, the vacancies either disappear by diffusion to dislocation walls or agglomerate into clusters (Ref 38). For these reasons, the CSAM deposits with the fcc structure (Al, Cu, and Ni) contain vacancy clusters, while in the CSAM deposit with hcp structure (Ti), vacancy clusters are absent.

Fig. 8
figure 8

Calculated lifetimes of positrons trapped at vacancy clusters plotted as a function of the number of vacancies the cluster consists of. For comparison, the left panel also shows lifetimes of positrons trapped at various size vacancy clusters in Ni calculated by Puska and Nieminen (Ref 49) and Kuramoto (Ref 50)

Full size image

For comparison, Table 1 also shows results of PAS investigations of Al, Cu, Ni, and Ti deformed by conventional cold rolling up to the equivalent strain, e = 0.1. The LT spectra of cold rolled samples of all metals studied consist of two components. The first component with lifetime τ1 represents a contribution of free positrons, and the second component with lifetime τ2 comes from positrons trapped at dislocations. Hence, the cold rolled samples contain lower dislocation density than samples produced by CSAM, and saturated trap** at defects does not occur. The only exception is Al, for which the dislocation density in the cold rolled sample is higher than that in the CSAM deposit. This is clearly due to significant dynamic recovery of defects occurring in the CSAM Al deposit. Moreover, no vacancy clusters were observed in the cold rolled samples in contrast with the CSAM deposits. The absence of vacancy clusters in the cold rolled samples is clearly due to a significantly lower strain rate resulting in a low vacancy production rate. Hence, in contrast with CSAM deposits, vacancies created by non-conservative motion of dislocations during conventional cold rolling have enough time to disappear by diffusion to sinks.

Conclusions

PAS investigations of CSAM deposits of pure metals Al, Cu, Ni, and Ti revealed that plastic deformation during cold spray deposition introduced a high density of dislocations and also vacancies. The deformation-induced vacancies either disappeared by diffusion to sinks at grain boundaries or agglomerated into small vacancy clusters. The recovery of deformation-induced vacancies by diffusion to sinks is influenced by spatial distribution of dislocations. Hence, the CSAM deposits of fcc metals (Al, Cu, and Ni) contain dislocations and vacancy clusters. A significant dynamic recovery of the deformation-induced defects was observed in the Al deposit due to the relatively high process temperature for this metal. In the Ti deposit with hcp structure, the obtained results suggest a formation of a dense forest of uniformly distributed dislocations, observed also in other high deformation processes. Since dislocations facilitate diffusion of vacancies to sinks, most deformation-induced vacancies disappeared, and the vacancy clusters were not observed in the CSAM Ti deposit. In contrast with CSAM deposits, no vacancy clusters were detected in Al, Cu, Ni, and Ti metals deformed by conventional cold rolling. It testifies that the extremely high strain rate of particles occurring during CSAM deposition is crucial for high vacancy concentration in the CSAM deposits. The present study introduced PAS as a valuable technique for microstructure characterization of CSAM deposits. The PAS results confirmed a high production rate of vacancies during the CSAM deposition suggesting that vacancies play an important role in the CSAM deposits.