Dimensional Effects of Hybrid Bond Layers on CFRP Metallized by Ti Cold Spray

Hybrid bond layers (BLs) were designed, fabricated, and evaluated for cold spray metallization of CFRP. The bond layers consisted of metal mesh embedded in a polymer ﬁlm adhesive co-cured to the CFRP. Efforts were devoted to identifying the critical opening ratio—i.e., the ratio of mesh opening size to powder diameter, for deposition of an adherent coating. Analysis of powder deposited at mesh openings show a transition from erosion (at a mesh opening ratio of 6.4) to mechanical interlocking and formation of a continuous coating with decreasing opening ratio. Selection of opening ratio yielded either (a) a grid of consolidated thin-walled deposits atop mesh wires separated by microchannel openings, or (b) densiﬁed coatings of cold-sprayed Ti. The effective opening ratio increased with increasing diameter ratio—i.e., the ratio of wire diameter to powder size, a consequence of eroded wire peripheries at shallow impact angles. These ﬁndings inform the design of future hybrid BLs, in concert with the selection of powder size, for cold spray metallization of CFRP.


Introduction
Adhesion of cold-sprayed (CS) deposits relies primarily on interactions between the impinging powder and the substrate surface. For dissimilar classes of materials, particularly brittle substrates, bond layers are often required to render substrates compatible to the deposition process and to promote adhesion. Incorporation of a bond layer is widespread and has enabled cold spray (Ref 1) and thermal spray metallization (Ref 2, 3) of thermoset composites. However, the adhesion of CS deposits to thermoset composites generally does not meet service requirements.
In the present study, hybrid bond layers comprised of woven wire meshes with select opening sizes were embedded in an epoxy film adhesive to address this shortfall. The objective of the work was to determine the effects of feature dimensions within the bond layer with respect to powder size on the metallization of carbon fiber reinforced polymer (CFRP) by cold spray. The relative dimensions of powder, wire, and mesh opening governed the CS metal deposition and affected adhesive strength. Judicious selection of dimensional parameters led to fabrication of strongly bonded deposits on thermoset composites.
Polymer metallization generally imparts erosion resistance, electrical and thermal conductivity, and biocompatibility, and can be accomplished through vapor phase deposition, electroless plating, and recently -cold spray (Ref 4). This solid-state material consolidation technique affords high deposition rates at low cost without the risk of thermal damage to the substrate or the formation of undesirable phases. The in situ formation of mechanical interlocks between CS powder and the substrate can create adherent interfaces without pre-treatment. Additional surfaces for interlocking and metallic bonding can be introduced by incorporating a wire mesh within the hybrid bond layer. The approach of hybrid bond layers aligns with a general progression toward scalable, cost-effective, ecofriendly surface preparation to form mechanical interlocks essential to thick (mm) composite metallization where chemical bonding yields insufficient adhesion (Ref 4 In the present work, we demonstrate an alternative process route to join Ti to thermoset composites in a single step. The process involves direct CS metallization of a composite featuring a hybrid bond layer comprised of metal mesh and epoxy, and the approach is widely applicable. The purpose of the bond layer is to ''compatibilize'' both process and material, and to promote adhesion. The effect of the reinforcing mesh within the bond layer was investigated, focusing on geometric features. The opening ratio, defined as the mesh opening size to powder diameter, strongly affected the depth and content of embedded powder at the mesh openings after cold spray deposition. The diameter ratio, defined as the wire diameter to powder diameter, led to select erosion of wire peripheries. The dependence of deposit continuity and adhesion on these parameters was identified and can guide the selection of BL and powder dimensions. The findings demonstrate the ability to tailor hybrid bond layers to the desired form of the deposit to the intended application.

Bond Layer Preparation and Cold Spray Deposition
Commercial purity Ti powder (CP Ti, D90 = 40 lm, 325 9 325 mesh) was cold sprayed onto a hybrid metalpolymer bond layer (BL) on the surface of a CFRP laminate (1.5 mm thick), comprised of ten plies (0/90) of prepreg (Toray 2510, Toray Industries, Inc., Tokyo, Japan). Figure 1 illustrates the two types of bond layer substratesembedded and supported mesh. The wire mesh was fully embedded in the film adhesive ( Fig. 1a) or positioned atop and supported by the film adhesive (Fig. 1b). The hybrid bond layer (Fig. 1a) was reinforced with Al 5056 wire mesh (McMaster-Carr, Elmhurst, IL, USA) embedded in an epoxy film adhesive with Al filler (LOCTITEÒ EA 9658 AERO, Henkel, Düsseldorf, Germany). The bond layer and CFRP were co-cured in a vacuum-bag only (VBO) process following manufacturer's guidelines and detailed in a previous study (Ref 1). To decouple the effects of the film adhesive and the metal mesh on deposit adhesion, supported mesh ( Fig. 1b) was also cold sprayed to investigate interactions between CP Ti powder and Al wire. Three mesh opening sizes -BL50, BL120, and BL200 -were investigated as described in Table 1. The substrates were cold sprayed with He at 200°C, 2.76 MPa, raster speed of 400 m/s, and a stand-off distance of 25.4 mm using a VRC Gen III CS system (VRC Metal Systems, Rapid City, SD, USA). The mean powder velocity of 594 m/s was measured at the nozzle exit using a laser velocimetry system (HiWatch HR1 CS, Oseir, Tampere, Finland).

Microstructural Analysis and Lug Shear Test
The hybrid metal-polymer bond layer and the deposits were sectioned and ion polished (JEOL SM-09010, Tokyo, Japan) for SEM (Helios G4 PFIB UXe, Thermo Fisher Scientific, Waltham, MA, USA) imaging to analyze both plan and cross sections. The supported mesh was cold mounted before sectioning and polishing for cross-sectional imaging. EDS (Ultim Max, Oxford Instruments, Abingdon, UK) maps and line scans were used to detect the presence of Ti, Al, and O. The largest powder diameter -44 lm, was used to calculate the opening ratio (mesh opening size to powder diameter) and the diameter ratio (wire diameter to powder diameter). Adhesive shear strength of the deposit (6-pass CP Ti followed by 30-pass CP Al) was measured according to MIL-J-24445A. The shear strength was calculated by taking the average of 5 measurements. The resulting fracture surfaces were examined by SEM to identify the adhesion mechanisms.

Finite Element Analysis
Two types of impact simulation were conducted, both with 2D thermo-mechanical coupling, using ABAQUS/Explicit with Lagrangian formulae. The first model simulated the impact of a single Ti particle onto supported mesh, as shown in Fig. 2(a). The wire mesh was positioned atop an epoxy substrate, and normal contact was defined between the surface of the wire mesh and the top surface of the epoxy. This simulation demonstrated wire erosion upon impact, as the position of impinging powder particle varied along the wire cross section. The second model simulated the impact of multiple Ti particles impinging onto BL120 and BL200, as shown in Fig. 2(b) and (c). This simulation depicted the impact load transferred to the wire mesh and the accumulation of CS powder at the mesh openings. In this simulation, all the particles traveled at the same impact velocity as the measured in-flight velocity. As shown in Fig. 2, the particles were distributed in the same alternating grid arrangement for both BL120 and BL200 to reflect the different deposition behaviors on the bond layers. 38 particles in total accumulated to two particle layers for comparison with experimental results.
In both simulations, the Johnson-Cook (JC) model was used to simulate plasticity of the material. The erosion of epoxy in multiple particle impact was simulated using the Johnson-Cook damage criterion. The material parameters and properties used are listed in Table 2. The simulated impact velocity was 600 m/s, which was chosen based on the measured powder velocity (593.8 m/s), and the initial temperature was set to 298 K. The simulation grid size was 1=50d p for single particle impact, and 1=20d p for multiple particle impact. A coarser grid size was selected for the multiple particle simulations because of the larger computational resources compared to the single particle model. The element type was CPE4RT with reduced integration and hourglass control.

Results and Discussion
Microstructure Figure 3 shows images of CS deposits produced with different bond layer designs. The images show that the continuity and structure of the deposit depended on the selected opening ratio (mesh opening size to powder diameter) and the pattern of the underlying mesh. An opening ratio of 6.4 (BL50) produced a discontinuous   Fig. 3a and b), while a ratio of 1.7 (BL200) yielded a continuous coating ( Fig. 3c and d).    Fig. 3(d). BL120, which is not shown here, spontaneously detached after cold spray because of weak deposit adhesion. Images in Fig. 4 show deposition onto mesh wires and erosion of wire edges after 6-pass CS onto a supported 50 9 50 mesh (opening ratio: 6.4, diameter ratio: 5.2). The grid pattern of the wire was preserved, resulting in rectangular openings (Fig. 4a). The wire edges yielded no deposit and formed erosion zones at impact angles \ 45°( calculated using Fig. 4(a) and (b). Figure 4(c) and (d) support this description and present a similar observation when the sample is viewed after tilting to 40°. The images display vertically stacked CP Ti powder at the center of the wires with steeply inclined sides. Powder consolidated only in the vertical direction, while no powder fused to the sides of the wire or deposit. Compared to the 6-pass deposit in Fig. 4(a), the wall spacing of the 36-pass deposit in Fig. 3(b) widened by * 10% because of increased erosion of off-center deposits. The formation of a discontinuous structure was a function of both the opening ratio and the impact angle. The eroded regions of wires increased the effective opening ratio by widening the mesh opening required for powder to bridge and form a continuous deposit. Figure 5 shows the deposition and degree of erosion of the wire at different impact angles. The total simulation time was 500 ns, and each image was captured immediately after the equivalent plastic strain (PEEQ) became constant over time. Deposition and erosion of the wire at a 90°spray angle depended on the local impact angle of impingement, as depicted in Fig. 5. Local impact angles greater than 45°led to cratered mesh wire around the impact site (Fig. 5a, b and c). However, impact angles less than or equal to 45°led to severe plastic deformation of the wire periphery, causing erosion (Fig. 5d, e, f). The velocity of the impinging particle can be decomposed into the normal component (v p sinh) and the tangential component ðv p coshÞ, (shown previously in Fig. 2a). For impact sites at increasing distances from the wire center, the impact angle decreased. Correspondingly, the normal impact velocity reduced by a multiple of sinh, reducing the crater depth and resulting in insufficient deformation for bonding. In concurrence, the tangential momentum elongated the crater and facilitated particle rebound. The progressively shorter crater rim failed to capture and retain subsequently impinging particles. Similar conclusions can be drawn from analysis of the kinetic energy of the impinging particle, which transformed into friction dissipation, plastic dissipation, and recoverable strain energies ( Ref 25,26). A greater plastic dissipation energy (ALLPD variable in Abaqus) from the induced kinetic energy (ALLKE) correlated with the adhesion energy (Ref 27-29), i.e., the probability of particle   Fig. 6. Plastic dissipation energy decreased in conjunction with the impact angle. At impact angles less than 45°, the plastic dissipation energy was less.
These findings were consistent with studies where impact angles less than 45-60°for various material combinations yielded negligible deposition (Ref [30][31][32][33][34]. Figure 7 shows bridging of wire openings and deposition onto film adhesive with 6-pass CS onto supported 120 9 120 mesh (opening ratio: 2.5). In the images in Fig. 7(a) and (b), over half of the 120 9 120 mesh openings are filled or partially covered by CP Ti powder (in contrast to those in Fig. 4, which showed consolidated powder only on the 50 9 50 mesh wires). The bridging of consolidated powder over wires at an elevation * 150 lm above the mesh is shown in cross-section in Fig. 7(c). Within the deposit, porosity was present primarily above the wire openings. Figure 7(d) shows CP Ti powder stacked on the film adhesive after traveling through 120 9 120 mesh openings. Tamping by subsequently impinging powder did not effectively fuse powder above the mesh to powder directly on the film adhesive. The observation indicates that for embedded BLs, deposit adhesion depended primarily on wire attachment to the BL matrix.   Figure 8 shows the effects of opening ratio and diameter ratio on mechanical interlocking and wire separation in the hybrid bond layer. Figure 8(a) shows * 85 lm powder embedment in the area surrounded by three wires, framed in white. The sample (BL120, opening ratio 2.5), shows two parallel wires present top center, and one serpentine wire crossing beneath. In contrast, Figure 8(b) shows * 30 lm powder embedment in a similar region for BL200 (opening ratio 1.7). At the mesh opening, loosely bonded powder was observed (white frame), a result of detachment during sample preparation. The contribution of interlocking to adhesive strength would increase with embedment depth if the deposit extending into wire openings were fully consolidated, as demonstrated with CP Al in a previous study (Ref 1). However, the chosen CS parameters in this work yielded loosely bonded powder that was inadequately interlocked. For wires not in mutual contact, the separation distance from the epoxy matrix depended on the diameter ratio. Figures 8(c) and (d) show separation distances of 11 lm and 19 lm for diameter ratios of 2.3 and 1.2 (BL120 vs BL200).
The effect of opening ratio on mechanical interlocking was also demonstrated in multiple particle simulations shown in Fig. 9. The depth and content of powder embedment resembled the observations shown in Fig. 8. Under the same impact conditions, Ti powder penetrated more deeply into the opening of BL120 than BL200. In simulations, 1500 ns after impact, the residual stress generated in the Al wire exceeded the stress generated in the epoxy film adhesive. The stress distribution of both BL120 and BL200 indicated that the impact loads were transmitted primarily to the Al wire, mitigating the extent of epoxy erosion.
The consequence of load transfer between Al wire and epoxy resembles the direct impact of CS powder onto substrates. Figure 10(a) shows that only the wire nearer the surface interacted with CS-Ti and separated from the film adhesive. The separation between Al wires and epoxy was attributed to CTE mismatch and rebound kinetic energy after impact. Rebound of CS powder resulted in partial detachment and gap formation between the CS powder and the substrate ( Ref 35,36). Beneath the detached wire was a thin metal layer composed of oxidized Al and Ti as shown in Figures 10(b) Shear strength measurements of BL200 (36.0 ± 3.0 MPa) revealed insights into the role of the wire mesh in adhesion, and provided context for assessing the suitability of the bond layer approach in practice. Figure 11 shows mixed mode failure between the CP Ti  Figure 11(c) shows sparsely dispersed Ti powder (white, indicated by cyan arrows) that remained on the substrate fracture surface. Figure 11(d) shows detachment of Al wires (light grey, orange arrows) and Ti deposit (white, cyan arrows). Regions of the Ti deposit were interlocked with the film adhesive (dark grey), which was mixed with Al filler particles (light grey, orange arrows). The mixed failure mode indicated gradual crack propagation, and the failure mode undoubtedly contributed to fracture toughness. Both phenomena were dictated by the design of the hybrid bond layer.
In aerospace applications, the effectiveness of leadingedge protection by a metallic surface layer relies on the adhesion of the surface treatment as well as erosion mitigation in service. Poor adhesion even in a relatively small area can lead to entire debonding of leading-edge tapes ( Ref 38). Erosion of leading edges can create surface asperities and trigger propagation of shear stress waves from the damage initiation site ( . In this work, the measured shear strength exceeded the overlap shear strength of a structural adhesive commonly used for aircraft maintenance and attaching leading-edge protection  deposit-substrate interface. With further refinement of BL designs, the proposed approach can be fully automated for both consistent application and rapid restoration of adherent leading-edge protection.

Conclusions
The effectiveness of the hybrid bond layer design for CS deposition onto CFRP was demonstrated. The mesh opening ratio (mesh opening size-to-powder diameter) determined the extent to which cold-sprayed powder mechanically interlocked and/or eroded epoxy exposed at mesh openings. Larger openings led to erosion of exposed adhesive, while smaller openings led to insufficient interlocking. The findings further indicated that as a general rule, mesh openings [ 10 9 the mean powder diameter or \ 1 9 the mean powder diameter are unlikely to result in acceptable mechanical interlocks for a continuous coating.
The design of the hybrid bond layer in conjunction with selection of powder diameter was major factors that controlled the microstructure of Ti deposits on CFRP. By incorporating an appropriately designed hybrid bond layer, cold-sprayed titanium, and possibly Ti-6Al-4 V, can be bonded to CFRP without adhesives. The process may be suitable for metallization of leading edges in air vehicles and of bearing surfaces in medical implants. The findings demonstrate an approach to expand applications of cold spray Ti alloys in protective coatings that can be restored, as well as in direct additive manufacturing onto composites.
Acknowledgments This work was sponsored by Pratt & Whitney Institute for Collaborative Engineering (PWICE).
Funding Open access funding provided by SCELC, Statewide California Electronic Library Consortium.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.