Miniaturization and performance improvements of electronic devices in recent decades have significantly increased heat dissipation rates. To overcome this, researchers have developed heat sinks with miniature fluid channels to maintain small device footprints with increased heat transfer performance. These channels are often fabricated using either subtractive fabrication methods, such as etching or micro-milling, or additive methods such as direct metal laser sintering (DMLS). These methods are limited by their long processing times, low geometric accuracy, or high cost. To overcome these limitations, a novel additive manufacturing method is developed using twin wire-arc spray. Wire-arc spray was used to build complex aluminum structures with length scales varying from 0.5 mm to 74 mm. Surface structures were built on a metal plate by spraying aluminum through a 3D printed polymer mask. Internal flow passages were made by filling surface channels with a water-soluble polyvinyl alcohol (PVA) paste that was allowed to harden, spraying metal over it, and then dissolving the PVA. The influence of wire-arc spray process parameters, such as standoff distance and scanning speed, on coating solid PVA with aluminum, were also investigated.

The piezoresistivity of flame-sprayed NiCoCrAlTaY on an electrically insulated surface of a steel substrate was investigated through cyclic extension and compression cycles between 0 and 0.4 mm for 1000 cycles and uniaxial tensile test. The sprayed NiCoCrAlTaY was in grid form with grid thickness of 3 mm and grid length of 30 mm while the electrical insulation was fabricated by flame spraying alumina on the surface of the steel. During mechanical loading, instantaneous electrical resistance measurements were conducted to evaluate the corresponding relative resistance change. Images of the loaded samples were captured for strain calculations through Digital Image Correlation (DIC) technique. After consolidation of the pores within the coating, the behavior of the flame-sprayed NiCoCrAlTaY was consistent and linear within the cyclic compression and extension limits, with strain values of approximately -1000 με and +1700 με, respectively. The coating had a consistent and steady maximum relative resistance change of approximately 5% within both limits. The tensile test revealed that the coating has two gauge factors due to the bi-linearity of the plot of relative resistance change against strain. The progression of damage within the coating layers was analyzed from its piezoresistive response and through back-scattered scanning electron microscopy images. Based on the results, the nickel alloy showed high piezoresistive sensitivity for the duration of the loading cycles, with little or no damage to the coating layers. These results suggest that the flame-sprayed nickel alloy coating has great potential as a surface damage detection sensor.

Metals were deposited on components made by 3-D printing with polyvinyl alcohol (PVA), a water-soluble polymer. The polymer was then dissolved, leaving a metal layer whose surface topography was the negative of that of the polymer. This is a rapid and low-cost alternative to 3D printing directly using metal, but to succeed it is essential for the sprayed metal to adhere to the polymer substrate. Tests were done in which aluminum and copper were sprayed using a twin-wire arc spray system onto 3D printed coupons, 50 mm x 50 mm in size, made from polylactic acid (PLA), PLA mixed with metal (aluminum, copper) or carbon fiber, and PVA. Adhesion depended on substrate roughness (minimum 1-2 μm) and substrate temperature (above the glass transition temperature but below the melting temperature of the polymer). It was shown that surface features could be made with high resolution on metal components using this technique.

Porous copper coatings, which act as wicks for liquid transport, were fabricated using a flame spraying process. Copper and aluminum powders were fed independently into the spray torch and deposited on copper substrates to form a composite coating. The aluminum was subsequently removed using chemical leaching leaving a porous copper coating behind. Varying the feed rate of aluminum powder allowed the coating porosity to be controlled. Channels to enhance liquid flow were made in some of the porous copper coatings by placing pieces of aluminum wire mesh on the copper substrate before spraying. During spraying the sprayed powders passed through the mesh opening and created pyramid shaped arrays on the substrates. The groove width was controlled by using different wire mesh sizes. Coatings were made with porosity varying from 2 to 44 %, and groove width ranging from 0.16 to 0.53 mm. The capillary performance of the coatings was evaluated experimentally by measuring the rate of rise of ethanol in the coatings. The rate of rise increased with coating porosity, and decreased with groove width.

In this study, twin wire arc spraying is used to bond wire mesh to the outside surfaces of stainless steel pipes in order to increase heat transfer surface area. At the optimum spray distance, the oxide content, porosity, and adhesion strength of the coatings are shown to be 6.6%, 2.1%, and 24 MPa, respectively. Pipes with different wire mesh configurations were placed in an oven and heated to temperatures from 300 °C to 900 °C. Water temperatures were measured at the inlet and outlet of the pipe for flow rates between 0.2 and 0.5 gpm. A maximum water temperature rise of 13 °C was achieved, corresponding to a total heat flux of 57 kW/m2. Heat transfer efficiency is shown to depend strongly on the quality of the bonds between the wire mesh and pipe and the spacing of wires in the mesh.

A numerical investigation of fluid flow and heat transfer through thermal spray formed metal foam heat exchangers is presented. Experimentally obtained fluid flow and heat transfer parameters are used in the simulations. Analytically obtained values of effective thermal conductivity are used to model heat transfer. A 3D CFD model was created for a metal foam heat exchanger with a square cross-section. The external walls were deposited on the foam using a wire-arc process. The channel walls of the foam were exposed to a constant temperature of 400 K and an air flow with an inlet velocity of 2 m/s. The model was verified by comparing sample results to experiments. The effect of the foam on heat transfer was then studied by varying thermal conductivity values.

Thermal spray coating processes have been employed in the current study to deposit well-adhered, dense skins on the surfaces of open-cell nickel foams. Using foam with 10 and 40 PPI (pores per inch) pore sizes, square channels were made with a height of 20mm and having a length of 250mm. In a unique process that prevents the deposited skin from penetrating the foam substrate via a paste comprised of a thermoset resin and powder particles, a dense stainless steel skin with an average thickness of 400 μm is applied to the exterior of the foam sample. The result is a channel that consists of a Ni foam core and a stainless steel skin wall that can be used as a compact heat-exchanger by directing the coolant flow through the foam. To study the feasibility of the metallic foam heat-exchangers, hydraulic and heat-transfer characteristics were investigated experimentally. The local wall and fluid temperature distribution and the pressure drop along the length of the heat exchanger were measured for heat-flux of 1540.35 – 9627.38 W/m 2 . Experiments were conducted using air as the coolant and varying flow velocity from 10 – 80 L/min. For non-Darcy flow with inertia effects in the porous media, the Dupuit and Forchheimer modification is employed with the experimental results to determine foam characteristics such as permeability (K), Ergun coefficient (CE) and the friction factor (f). To measure the heat-transfer performance of the metal foam filled channels, a length average Nusselt number is derived based on the local wall and fluid temperatures. Heat transfer was shown to have nearly doubled compared to that of a channel without a foam core.

This paper presents the development of a new thermal spray gun for the so-called warm spraying process in which powder particles are not melted but heated to temperatures much higher than those typically found in a cold spray process. The increased heating leads to a reduction in the particle impact velocity required to deposit the coating and hence reduces operating cost. The new gun utilizes methane-oxygen combustion for particle heating and features a swirl-type combustion chamber to create a turbulent mixture of the fuel and oxidizer for efficient combustion. Powder can be fed axially or radially into the gun. To control particle temperature independently, combustion gases are diluted by adding nitrogen gas through axial or radial ports provided in the gun. A converging-diverging nozzle with a downstream cylindrical barrel accelerates the burnt gases to supersonic velocities. The design of the nozzle and barrel was optimized using numerical simulations. Mass flow rates of methane, oxygen, and nitrogen were calculated using energy balance, stoichiometric combustion, and nozzle flow rate equations. The gun is designed to operate up to 200 kW and is water-cooled. Experiments were conducted to test the performance of the new gun in which tungsten carbide coatings were deposited on aluminum substrates. Coatings were analyzed using standard methods and showed promising results.

Open pore foams can be used as gas filters, catalyst supports, and heat exchangers due to their high gas permeability and heat conductivity. In this study, Ni-Cr skins were deposited on each side of a foam sheet by HVOF spraying to form a sandwich structure for use as a heat exchanger. The microstructure of the skins and the interface with the nickel foam struts were examined and the hydraulic characteristics and heat transfer properties of the sandwich structure were experimentally determined. Pressure drops across the heat exchanger were measured and found to be proportional to the square of the velocity of the coolant and a least square fit was used to solve for the permeability, K, and form coefficient, C, of the foam.

Nickel-based superalloys can be used at temperatures up to 1050 °C in air. Superalloy open cell foam sheets with skin layers plasma sprayed on both sides can be used as high temperature heat exchangers provided that the two deposited skins are dense and well adhered to the open cell foam. In this study alloy 625 skins were deposited on each side of a sheet of metal foam by APS and HVOF to form a sandwich structure. Two densities of open cell foams, 20 and 10 pores per linear inch (ppi), were used in this study as the core. The initial Ni foam was converted to an alloy composition by plasma spraying aluminum and chromium on the foam’s struts with subsequent diffusion/solutionizing heat treatments before the alloy 625 skins were deposited. The microstructure of the coatings and the interface between the struts and skins was investigated. A layer of Ni-Al alloy was formed near the surface of the struts as a result of the heat treatment. The foam struts were imbedded more deeply into the coatings deposited by HVOF than the coatings deposited by APS.

Plasma-sprayed, molten molybdenum particles (~55 µm diameter) were photographed during impact on grit-blasted glass surfaces that were maintained at either room temperature or at 350°C. Droplets approaching the surface were sensed using a photodetector and after a known delay, a fast charge-coupled device (CCD) camera was triggered to capture time-integrated images of the spreading splat from behind the glass. A rapid two-color pyrometer was used to collect the thermal radiation from the spreading droplets to follow the evolution of their temperature and calculate the splat cooling rates. It was found that as the surface roughness increased, the maximum spread diameters of the molten molybdenum droplets decreased, while the splat cooling rates increased. Impact on non-heated and heated roughened glass with similar roughness values produced splats with approximately the same maximum spread diameters, skewed morphologies, and cooling rates. On smooth glass, the splat morphologies were circular, with larger maximum spread diameters and smaller cooling rates on non-heated smooth glass. An established model was used to estimate the splat-substrate thermal contact resistances. On highly roughened glass, the thermal contact resistance decreased as the glass roughness increased, suggesting that splat-substrate contact was improved as the molten metal penetrated the spaces between the large asperities.

Plasma-sprayed, molten nickel particles (60 µm diameter) were photographed during impact on oxidized 304L stainless steel surfaces that were maintained at room temperature or at 350oC. The steel samples were oxidized at different temperatures. Droplets approaching the surface were sensed using a photo detector and after a known delay, a fast charge-coupled device (CCD) camera was triggered to capture time-integrated images of the spreading splat from the substrate front surface. A two-color pyrometer was used to collect the thermal radiation from the particles to follow the evolution of their temperature after impact. Molten nickel particles impacting on oxidized steel at room temperature fragmented significantly, while heating the surfaces produced splats with disk-like morphologies. Impact on steel that was highly oxidized induced the formation of finger-like splash projections at the splat periphery. The splat cooling rate and thermal contact resistance between the splat and non-heated oxidized steel varied significantly as the degree of oxidation increased; heating the oxidized steel greatly reduced the variations. It was suggested that the large variations in splat cooling rates and thermal contact resistances on the non-heated oxidized steel was due primarily to the presence of adsorbates on the steel surface.

Splats formed during a thermal spray process may be either highly fragmented or intact and disk-like. To predict this change in splat morphology, a dimensionless solidification parameter (Θ), which takes into account factors such as the substrate temperature, splat and substrate thermophysical properties, and thermal contact resistance between the two, has been defined. Θ is the ratio of the thickness of the solid layer formed in the splat while it is spreading, to the splat thickness. The value of Θ can be calculated from simple analytical models of splat solidification and spreading. If the solid layer growth is very slow (Θ << 1), the droplet spreads out to a large extent. Once it reaches maximum spread it becomes so thin that it ruptures, producing fragmented splats. If, however, the solid layer thickness is significant (Θ ~ 0.1 – 0.4), the droplet is restricted from spreading too far and does not become thin enough to rupture. Under such circumstances, disk-type splats are expected. When the solid layer growth is rapid (Θ~1), spreading of the droplet is significantly obstructed by the solid layer, producing splats with fingers around their periphery. Predictions from the model are compared with experimental data.

In this work we present the numerical simulation results for the molten nickel and zirconia (YZS) droplets impact on different micro-scale patterned surfaces of silicon. The numerical simulation clearly showed the effect of surface roughness and the solidification on the shape of the final splat, as well as the pore creation beneath the material. The simulations were performed using a computational fluid dynamic software, Simulent Drop, The code uses a three-dimensional finite difference algorithm solving full Navier Stokes Equation with heat transfer and phase change. Volume of fluid (VOF) tracking algorithm is used to track the droplet free surface. Thermal contact resistance at the droplet– substrate interface is also included in the model. Specific attention is paid to the simulation of droplet impact under plasma spraying conditions. The droplet sizes ranged from 15 to 60 microns with the initial velocities of 70-250 m/s. The substrate surface was patterned by a regular array of cubes spaced at 1 µm and 5 µm from each other. The peak to valley height of each cube was between 1 to 3 µm. Different splat morphologies will be compared with those obtained from the experimental results under the same impact and surface conditions.

The curling up of splats of molten metal deposited on a cold substrate was investigated both experimentally and numerically. An analytical model based on mismatch of thermal expansion between the splat and substrate was developed to calculate the deformation of the splat after curling up. The curling up angle at the edges of splats was predicted using the analytical model and compared with the experimental measurements. The prediction shows good agreement with the experiments.

Plasma-sprayed, molten molybdenum particles (~40 µm diameter) were photographed during impact (with velocity ~110 m/s) on Inconel surfaces that were preheated or maintained at room temperature or 400oC. A droplet approaching the surface was sensed using a photodetector and after a known delay, a fast CCD camera was triggered to capture images of the spreading splat from the substrate front surface. A rapid two-color pyrometer was used to collect the thermal radiation from the impacting particles to follow the evolution of their temperature and size after impact. Molten molybdenum particles impacting on surfaces at room temperature disintegrated and splashed, after achieving a maximum diameter larger than 400 µm. Impact on preheated and heated Inconel produced splats with maximum diameters between 200 µm and 300 µm and with less splashing. The cooling rate of splats on the preheated Inconel was larger than that of splats on non-heated Inconel, suggesting that the splat-substrate contact was improved.

Plasma-sprayed yttria-stabilized zirconia particles (~40 µm diameter) were photographed during impact (velocity ~200 m/s) on a glass surface that was maintained at either room temperature or 400°C. A droplet that approached the surface was sensed using a photodetector and after a known delay, a light source was triggered to illuminate the particle in order to photograph it with a charge-coupled device (CCD) camera. A rapid two-color pyrometer was used to collect the thermal radiation from the particles to follow the evolution of their temperature and size, in-flight and after impact. The fully molten particles spread into a thin liquid splat after impacting the surfaces. The partially molten particles disintegrated into small satellite fragments immediately upon impact. The surface area, as indicated by the pyrometric signals, of the partially molten particles during spreading were almost an order of magnitude smaller than that of the fully molten particles. The pyrometric signals, characteristic of the impact of partially molten zirconia, provide a novel method of identifying partially molten ceramic particles after impact on a flat surface.

The impact of plasma-sprayed molybdenum particles on glass surfaces held at 25 and 400°C was photographed. A two-color pyrometer was used to collect thermal radiation from the particles to follow their temperature evolution and to calculate the splat cooling rate. Significant fragmentation of the splat on the surface at 25°C was observed. A 3D model of droplet impact and solidification was used to estimate the thermal contact resistances between the splat and glass. It was found that the thermal contact resistance was approximately two orders of magnitude smaller on the surface at 400°C, indicating faster solidification, which reduced splashing. The larger thermal contact resistance between the non-heated glass and splat was attributed to the presence of a gas barrier at the surface.

A three-dimensional, time-dependent numerical model of free-surface flows and heat transfer including phase change has been used to simulate the impact of a liquid droplet on a solid particle and to predict the size of the void under the solid particle caused by incomplete filling by liquid landing on top of it. The solid particle was considered to be a protrusion on the substrate. Fluid flow in the impacting liquid droplet was modeled using a finite difference solution of the Navier-Stokes equations in a 3D Cartesian coordinates assuming laminar, incompressible flow. Heat transfer in the liquid droplet was modeled by solving the energy equation, assuming densities of liquid and solid to be constant and equal to each other. The free surface of the liquid droplet was assumed to be adiabatic. The porosity in this simulation was defined as the volume of the incompletely filled void under the solid particle to the volume of the solid particle. The simulation was repeated with different process parameters, and the results showed that process parameters play significant roles in determining the amount of porosity. A correlation is found to express the porosity as a function of the process parameters.

The impact and solidification of 4 mm molten aluminum alloy 380 droplets on a tool steel substrate was studied both analytically and experimentally. Temperature histories at different radial location on the substrate surface under impacting droplets were recorded using an array of thin film thermocouples with response times less than 1 µs. Photographs were taken of droplet impact onto the substrate. Initial substrate temperature was varied from room temperature to 300°C and average surface roughness from 0.5 to 5.0 µm. Estimates of thermal contact resistance were made by matching measured substrate temperatures with an analytical solution for surface temperature variation. A model of the true area of contact between molten metal and a rough surface was developed in order to predict how contact resistance changes with surface roughness and contact pressure. Impact of molten aluminum alloy droplets was simulated using a three-dimensional numerical. Using values of thermal contact resistance predicted by the model gave good agreement between computed and observed droplet shapes during impact.