Additive manufacturing processes and additively manufactured products

Abstract

A technique to additively print onto a dissimilar material, especially ceramics and glasses (e.g., semiconductors, graphite, diamond, other metals) is disclosed herein. The technique enables manufacture of heat removal devices and other deposited structures, especially on heat sensitive substrates. It also enables novel composites through additive manufacturing. The process enables rapid bonding, orders-of-magnitude faster than conventional techniques.

Claims

1. A device, comprising: a substrate having a supporting surface; a solid layer; a fused adhesion layer, formed between the solid layer and the supporting surface of the substrate by a process comprising: depositing a powder on a supporting surface of the substrate; locally heating a portion of the powder on the supporting surface of the substrate with a focused energy beam, to fuse the locally heated portion of the powder without achieving thermal equilibrium between the fused locally heated portion of the powder and the substrate, to thereby form the fused adhesion layer having a different chemical composition than the fused powder and the substrate, which increases an adhesion of the fused powder to the substrate; and cooling the fused powder to form the solid layer.

2. The device according to claim 1, wherein the adhesion layer is formed substantially without concurrently fusing a non-locally heated portion of the powder, configured as a region-specific pattern.

3. The device according to claim 2, wherein the adhesion interlayer comprises a heating-induced chemical reaction product of the supporting surface of the substrate and the powder.

4. A manufactured structure, comprising: a substrate; a solid layer, formed over the substrate from fused portions of a powder, in a regional pattern; and an adhesion layer disposed between the solid layer and the substrate configured to increase an adhesion of the solid layer to the substrate, the adhesion layer comprising a metallic chemical reaction product of a surface of the substrate and the powder having a different chemical composition than the solid layer and the surface of the substrate, formed by locally heating a portion of the powder on the substrate with an energy beam to a fusion temperature, without achieving thermal equilibrium between the fusion temperature of the powder and a bulk of the substrate beneath the surface of the substrate.

5. The manufactured structure according to claim 4, wherein the powder further comprises particles which form inclusions in the solid layer, an exposed surface of the solid layer comprising a plurality of surface re-entrant cavities.

6. The manufactured structure according to claim 4, wherein the solid layer on the interface layer is configured as a heat sink configured to shed heat from the substrate.

7. A manufactured structure, comprising: a substrate; an interface layer, comprising a metallic chemical reaction product of a surface of the substrate and a powder; and a solid layer, formed over the interface layer from fused portions of the powder, in a regional pattern, wherein the solid layer on the interface layer is configured as a heat sink configured to shed heat from the substrate, and wherein the regional pattern of the solid layer defines a surface with adjacent regions of heterogeneous aqueous fluid wetting, wherein regions with high wettability promote aqueous liquid flow to bubble generation sites, and regions with low wettability promote bubble liftoff.

8. The manufactured structure according to claim 6, wherein the regional pattern defines a circumferential wall of a microchannel configured to guide fluid flow.

9. The manufactured structure according to claim 4, wherein the powder comprises a metal alloy, and between about 1% to about 10% aggregate weight, per weight of the metal alloy, of a reactive element that after heating, actively bonds to the substrate surface, the reactive element being selected from the group consisting of Ti, Ta, Zr, V, Hf, Cr, Mo, W, and Nb.

10. The manufactured structure according to claim 9, wherein the reactive element facilitates bonding to the substrate, the substrate being selected from the group consisting of Si, SiC, SiN, graphite, diamond, carbon nanotubes, graphene, fullerenes, GaN, GaAs, βGa.sub.2O.sub.3, Al.sub.2O.sub.3, SiO.sub.2, silica glass, quartz, borosilicate glass, aluminosilicate glass, beryllia, ceria, zirconia, lithium aluminum silicates, gallium oxide, indium tin oxide, indium zinc oxide, molybdenum disilicide, tungsten borides, gallium arsenide, zinc sulfide.

11. The manufactured structure according to claim 4, wherein: the substrate comprises an integrated circuit having a deposited metal layer, the solid layer is metallic, the interface layer comprises a fused intermetallic composition, and the regional pattern is configured as a heatsink for the integrated circuit.

12. The manufactured structure according to claim 4, further comprising an additional solid layer formed on the solid layer by regional heating of an additional powder.

13. The manufactured structure according to claim 12, wherein the additional solid layer is formed in a three dimensional pattern having a void region comprising a gap between selectively formed regions of the regional pattern.

14. The manufactured structure according to claim 4, wherein the powder that chemically reacts with the surface of the substrate to form the adhesion layer comprising at least one of an intermetallic compound, a metal carbide compound, a metal nitride compound, a metal boride compound, and a metal silicide compound.

15. The manufactured structure according to claim 4, wherein the substrate comprises a semiconductor.

16. The manufactured structure according to claim 15, wherein the semiconductor is configured as an integrated electronic circuit, and wherein the solid layer is configured as a set of electrically-isolated electrical interconnects to the integrated electronic circuit.

17. The device according to claim 1, wherein the powder comprises a metal alloy powder; and the fused adhesion layer comprises at least one of a metal silicide, a metal carbide, a metal boride, and a metal nitride, the chemical reaction product being shear resistant and facilitating adhesion of the fused metal alloy powder to the substrate.

18. The device according to claim 17, wherein the powder comprises the metal alloy, and between about 1% to about 10% aggregate weight, per weight of the metal alloy, of a reactive element that actively bonds to the substrate surface.

19. The device according to claim 18, wherein the reactive element is one of more element selected from the group consisting of Ti, Ta, Zr, V, Hf, Cr, Mo, W, and Nb.

20. The device according to claim 18, wherein the substrate is selected from the group consisting of Si, SiC, SiN, pyrolytic graphite, diamond, carbon nanotubes, graphene, fullerene, GaN, GaAs, βGa.sub.2O.sub.3, Gallium oxide, Al.sub.2O.sub.3, SiO.sub.2, silica glass, quartz, borosilicate glass, aluminosilicate glass, beryllia, ceria, zirconia, lithium aluminum silicates, indium tin oxide, indium zinc oxide, molybdenum disilicide, tungsten boride, gallium arsenide, and zinc sulfide.

21. The device according to claim 17, wherein the different chemical composition is selected from one or more of the group consisting of an intermetallic compound, an intermetallic-like compound, and a chalcogen bond compound.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 shows an overview of the additive manufacturing process.

(2) FIG. 2 shows an SEM image (8.6 mm WD, 9 KV EHT, SE2 detector) of mixed Ag (3 wt %), Ti (4 wt %) and Sn (93 wt %) powders.

(3) FIGS. 3A and 3B show SEM images (8.5 mm WD, 18 kV EHT, SE2 detector) of two sample at the interface of alloy and silicon with different exposure parameters, where FIG. 3A shows partial melting in sample one, and FIG. 3B shows full melting.

(4) FIG. 4 shows quantitative measurement of elemental composition of alloy using EDS (DT between 30% to 40%). SEM image (8.2 mm WD, 15 kV EHT, SE2 detector). Polished up to 20 nm roughness.

(5) FIGS. 5A and 5B show Sn.sub.3Ag.sub.4Ti alloy deposited on silicon by SLM process.

(6) FIG. 5C shows a stainless steel logo built on 4″ silicon wafer.

(7) FIG. 6 shows stainless steel parts fabricated by SLM on silicon wafer after thermal cycling test (100 cycles from −40° C. to 130° C.). Note, the steel plate used to secure the silicon to the build platform has rusted over the course of the thermal cycling.

(8) FIG. 7 shows a sample with different patterns printed under varying conditions. Surface property and mechanical strength were studied versus processing parameters, including number of exposures. Silicon is held in a wafer holder. Wafer can be held by a vacuum chuck or adhesive and the chuck can include powder skirts to prevent powder ingress into undesired areas.

(9) FIG. 8 shows the bond strength of the Sn.sub.3Ag.sub.4Ti-silicon fabricated by SLM technique (laser power 120 W, laser scan velocity 1100-2300 mm/s, double exposure).

(10) FIGS. 9A and 9B show examples of optical profilometry for a line of alloy on silicon created by 120 W, 1500 mm/s exposure parameter.

(11) FIGS. 9C and 9D show equivalent contact angle measurement versus laser scan velocity based on optical profilometry.

(12) FIG. 10A shows representative EDS spectrum of the alloy-silicon interface (8.7 mm WD, 10 kV EHT, DT 20%). The scalebar is 1 μm.

(13) FIG. 10B shows back scatter detector scanning electron microscopy image of the interface shows light elements (i.e., Ti) darker than heavy elements (i.e., Sn, Ag) (8.7 mm WD, 10 kV EHT, RBSD detector). The scalebar is 1 μm.

(14) FIG. 10C-10F show EDS mapping of the SiK peak (10C), AgL peak (10D), SnL (10E) peak, and TiK (10F) peak.

(15) FIGS. 11A-11C show surface roughness study by optical profilometry on three samples with different number of rasters (120 W, 1700 mm/s, 0.09 hatch distance).

(16) FIG. 12 shows phase identification of Sn—Ti—Ag alloy fabricated by SLM process using GIXRD (PANalytical X'Pert PRO High-Resolution X-ray Diffractometer).

(17) FIGS. 13A-13B show optical image of stainless steel 316L deposited on graphite exhibiting poor wetting (13A). Fixture designed to hold graphite substrate on build platform of EOS M290 (13B).

(18) FIG. 14 shows optical profilometry of ablated graphite.

(19) FIGS. 15A and 15B show a stainless steel logo printed on graphite (15A). Removing printed part using hot plate (15B).

(20) FIGS. 16A and 16B show spreading of the alloy on graphite substrate based on laser exposure parameters.

(21) FIG. 17 shows a schematic of the designed shear lap test.

(22) FIGS. 18A, 18B, and 18C show the interfacial strength of the graphite-metal bonding at various laser scan velocities (18A, 18B). Sample preparation for tensile test (18C).

(23) FIG. 19 shows an optical image of the Sn—Ag—Ti alloy laser bonded onto graphite.

(24) FIG. 20 shows an unpolished scanning electron microscopy of graphite-alloy interface.

(25) FIGS. 21A and 21B show the formation of carbide near the interface between graphite (bottom) and alloy (top).

(26) FIGS. 22A and 22B show two stainless steel parts after thermal cycling test.

(27) FIG. 23 shows Sn3Ag4Ti alloy additively deposited on HT grade high conductivity (1700 W/m-K at room temperature) pyrolytic graphite by laser processing.

(28) FIG. 24 shows ball shear testing is performed to measure interfacial strength of Sn3Ag4Ti alloy additively deposited on HT grade high conductivity (˜1700 W/m-K at room temperature) pyrolytic graphite by laser processing.

(29) FIG. 25 shows a photograph of Sn.sub.3Ag.sub.4Ti alloy deposited on Borosilicate glass.

(30) FIG. 26A shows an SEM image of Sn.sub.3Ag.sub.4Ti alloy deposited on Borosilicate glass (5.6 mm WD, 10 kV EHT, SE2 detector).

(31) FIG. 26B shows an SEM image of Sn.sub.3Ag.sub.4Ti alloy deposited on Borosilicate glass (8.1 mm WD, 10 kV EHT, SE2 detector).

(32) FIGS. 27A-27F shows EDS mapping of Sn3Ag4Ti alloy deposited on Borosilicate glass 8.1 mm WD, 10 kV EHT, SE2 detector). FIG. 27A: SE2 image, FIG. 27B: AgL peak, FIG. 27C: SnL peak, FIG. 27D: O K peak, FIG. 27E: TiK peak, FIG. 27F: SiK peak.

(33) FIGS. 28A-28E shows the printing onto sapphire process mapping. FIG. 28A shows a D50×0.43 mm Sapphire wafer with <0001> orientation and single side polish is purchased from a supplier and installed on a steel build platform using Kapton tape. FIG. 28B shows a single layer of Sn3Ag4Ti powder was deposited on the unpolished surface with manual deposition. The layer thickness is ˜54 μm which is the thickness of the Kapton tape. FIG. 28C shows the exposure parameters were set to 150 W with scanning rates varying from 4400 mm/s to 5600 mm/s. Two different geometries are considered lines and pillars. FIG. 28D shows a bright field optical imaging at 5× of two lines, top with 150 W and 4400 mm/s and bottom with 150 W and 4600 mm/s. FIG. 28E shows a bright field optical imaging at 20× magnification of 150 W 4400 mm/s line with Z-stack. Line thickness of 100 um is achieved with edge exposure (single line raster).

(34) FIG. 29 shows schematic showing layers of a 3D printed heat removal device made on a metallized silicon substrate.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

(35) The present technology provides a method for bonding metal powder to various types of dissimilar substrates, employing process technology that forms an adhesion layer from components of the substrate and the metal powder, substantially without causing heat damage to nearby or underlying structures.

Example 1

(36) Metal powder may be deposited in fused layers a non-metal substrate by directed-energy powder fusion manufacturing. The energy source for this process can be from any focused energy sources, and popular sources include laser and electron beams. The powder can be deposited uniformly over an entire layer and then selectively melted, as is done in powder bed fusion processes, or it can be made with powder that is directly deposited onto the area of the part that is being exposed to the focused energy source, in the process typically referred to as directed energy deposition. The directed energy source may be steered or the sample stage may move the sample to selectively fuse powder where desired. This process is conducted in an inert environment, such as an Argon, N.sub.2, or vacuum environment to reduce oxidation risk.

(37) A prototype of the technology employs a laser-powder bed fusion EOS M290 machine that possesses a 500 W peak 1064 nm Yb-fiber laser and high speed galvanometer mirror system (maximum raster speed of 7000 mm/s), which presents results which are broadly applicable to other powder bed fusion and directed energy deposition tools, regardless of the focused energy source.

(38) This technique can involve selective melting to a high density solid, or selective sintering to a porous solid depending on the processing parameters. Differences in the processing parameters (powder deposition mechanism, powder composition, substrate properties, energy source (e.g., wavelength of laser), energy source diameter) results in changes in the acceptable range of energy delivery parameters, though the methodology to identify near-ideal processing conditions will be unchanged.

(39) One benefit of this technology is that it enables localized high temperature bonding without damaging the sensitive circuits on the other side of the electronic wafer or sample. Even with the high temperatures of laser melting, the short exposure time of the laser limits the thermal penetration depth to avoid exceeding thermal limits of nearby elements. The thermally effected zone is quite small, as the exposure time for typical focused energy powder fusion is only ˜100 μs. The zone that is thermally affected can be estimated by the thermal penetration depth, L.sub.p∝√{square root over (αt)}. [64] This depth is less than the typical electronic wafer thickness (e.g., 300 μm for silicon), and so can be safely performed without damaging sensitive devices. This short exposure also enables building of higher melting temperature metals onto low-melting point alloys (e.g., stainless steel 316L on Sn.sub.3Ag.sub.4Ti on Si or graphite).

(40) In particular, during selective laser melting, the metals are typically only molten for times on the order of 75 μs (assuming typical scan rates of 1 m/s, and focused energy spot sizes of 75 μm). This period is much shorter than the typical time for conventional bonding of dissimilar materials (tens of milliseconds to tens of minutes). To overcome this kinetic limitation, the focused energy locally heats the sample beyond the melting point to thermally overcome the energy barrier for the active metal species to more easily diffuse, as the diffusivity, D, has an Arrhenius rate term so that the D∝exp(−E.sub.a,D/K.sub.B T), where E.sub.a,D is the energy barrier for the active alloy element to diffuse. The barrier to nucleate intermetallic phases can also be overcome faster at higher temperature, as the nucleation rate, N, will also be N∝exp(−E.sub.a,N/K.sub.B T), where E.sub.a,N is the energy barrier for intermetallic phases to nucleate at the interface.

(41) The contact angle and mechanical strength of the interface is not at its highest after a single exposure. Therefore, another key insight provides exposure of a single layer of brazing alloy to energy pulses multiple times. This provides greater time for diffusion of the active metal species and nucleation of the intermetallic phase. The benefit of a greater number of exposure diminishes with each successive exposure. For Sn.sub.3Ag.sub.4Ti on silicon and graphite, the difference between single and double exposures is extreme, but negligible additional improvements for three and more exposures.

(42) FIG. 1 shows an overview of this embodiment of the additive manufacturing process.

(43) Additive manufacturing on electronic substrates can be performed at the wafer level, after dicing, or after reflow onto a polymeric substrate (heterogeneously integrated circuits). The wafer or chips can be mounted in a chuck that maintains the device steady by such means as a vacuum chuck or heat release adhesive. The powder can be kept away from areas where powder ingress is undesired by a powder skirt. The powder skirt can be made out of silicone or elastomeric material. The bonding can be performed on an unpolished side or polished side of the wafer. It can also be used to bond onto diamond films that are used for heat spreading. For bonding, roughening is advantageous for mechanical bonding. This same technique can also be used to make electrical contacts on electronic devices like microprocessors and power electronic devices.

(44) Another embodiment uses this same technique to bond dissimilar metals for composite applications. The alloy may be printed directly on the metal substrate, and then the printed alloy is then bonded this metal substrate by heating to an elevated temperature, which for Sn.sub.3Ag.sub.4Ti is ˜220° C. This embodiment can bond to carbon materials, like graphite, that are of interest for composites, nuclear, heat transfer and medical applications. For instance, it can be used to make a pyrolytic graphite based heat spreader or heat sink. The benefit of making in pyrolytic graphite stems from pyrolytic graphite's high thermal conductivity.

(45) Another embodiment uses a bonding alloy for one or more layers and then transitions to printing a different metal. This can be desirable when different properties of a different metal are desired. For instance, copper or aluminum may be preferred to the bonding alloy for their transport properties. For Sn containing braze will form Cu—Sn intermetallics (Cu.sub.3Sn and Cu.sub.6Sn.sub.5)[65] and a braze with Ti element can bond to Al via formation of Al.sub.3Ti intermetallics.[21]

(46) Another embodiment uses this alloy to bond to metal as a thermally removable support material. This can help save the cost of refinishing build platforms that are currently used to build metal additively formed parts. This process works by printing the metal mixture or alloy, which forms high melting point intermetallics on the substrate, but the bulk metal melts at temperatures lower than the metal used to form the part. This bonding metal can then be used to bond to the part metal. The bonding metal can also be used as removable support internal to the part, like an overhang feature in the part. The low-melt alloy can be softened or melted in an oven.

(47) Subsequent sub-sections detail the particular embodiment of heterogeneous Sn.sub.3Ag.sub.4Ti powder bonding to silicon and graphite substrates. These subsections also detail subsequent bonding of this Sn.sub.3Ag.sub.4Ti alloy to stainless steel (316L). This same approach and material can also bond to SiC, SiN by forming silicides, and diamond and graphite by forming carbides, and GaN, GaAs, βGa.sub.2O.sub.3 by forming Ga—Ti intermetallic compounds, and to silicon containing oxides of crystalline and non-crystalline types (e.g., silica glass, quartz, borosilicate glass, lithium aluminum silicates) via the formation of oxides at the interface.

(48) In an alternative embodiment, this braze can alternatively replace the Ti with other elements that react actively with the substrate, such as Ta, Zr, V, Hf, Cr, Mo, W. The purpose of the Sn.sub.3Ag part of the alloy is to lower the melting point and increase the activity of the active metal elements.

(49) One embodiment of this invention uses Sn.sub.3Ag.sub.4Ti that reacts to form a Titanium Silicide on Si that promotes good bonding, and has a melting point around 250° C. This low melting point increases the activity of Ti. The low melting point also reduces the thermal stresses experienced on cool down from the molten state, hence preventing interfacial mechanical failure when printed. The interfacial mechanical failure is observed to be dependent on the size of the feature made and on the shape of the edges, as sharp corners are prone to greater thermal stresses.

(50) Bonding to Silicon with Sn.sub.3Ag.sub.4Ti:

(51) All experiments were conducted using a laser-powder bed fusion EOS M290 tool having a 500 W 1064 nm Yb-fiber laser and high speed galvanometer mirror system (maximum raster speed of 7000 mm/s). The machine has a N.sub.2 generator that maintains an N.sub.2 gas atmosphere with O.sub.2 concentrations of less than 1.5%. A special wafer tray is used to securely hold the silicon wafer.

(52) A Sn.sub.3Ag.sub.4Ti powder mixture is used as the bonding metal mixture, as it provides good bonding with Si via the Ti silicide formation, and also bonds to a broad range of other materials including graphite, Al, Cu, and Fe-based metals. This low melting point of this alloy (˜250° C.) reduces the thermal stresses experienced after solidification upon cooling. Sn, Ag, and Ti powders (>99% purity) with respective average diameters of 40 μm, 5 μm, 10 μm were mixed with weight percentages of 93%, 3%, and 4% (Sn.sub.87.55Ag.sub.3.11Ti.sub.9.34) (FIG. 2). Smaller powder sizes were chosen for the lower concentration metals to facilitate mixing and reduce diffusional resistance of the Ag and Ti in the Sn melt. FIG. 2 shows an SEM image (8.6 mm WD, 9 KV EHT, SE2 detector) of mixed Ag (3 wt %), Ti (4 wt %) and Sn (93 wt %) powders.

(53) The process parameters for the first layer of bonding and subsequent layers were determined via conventional process mapping techniques [66-71]. The process parameters were identified in a high-throughput manner, where various powers ranging from 30 to 200 watts and scanning rates from 800 to 6500 mm/s with fixed hatch distance of 0.09 mm were printed in each build, to identify the exposure parameters that provide the best wetting and bonding strength between silicon and the alloy. At the extremes, it was observed that at a fixed scanning speed, low power will result in partial sintering behavior and high power causes high vapor flux and large recoil pressure which leads to splattering. One key difference with conventional processing is that the transformation is assisted by multiple laser exposers for each layer.

(54) FIGS. 3A and 3B show SEM images (8.5 mm WD, 18 kV EHT, SE2 detector) of two samples at the interface of alloy and silicon with different exposure parameters. FIG. 3A shows partial melting, as opposed to FIG. 3B that shows complete melting and solidification. Sample 2 possesses an average composition as expected from the mixing weights, indicating sufficient mixing (FIG. 4) as quantified by electron dispersive spectroscopy (EDS). The properties used for the sample of FIG. 3B were close to ideal.

(55) TABLE-US-00001 TABLE 1 Scan Hatch Layer Energy Power speed distance thickness density Sample [W] [mm/s] Exposure [mm] [μm] [J/mm.sup.3] 1 150 5250 Double 0.09 20 15.87 2 120 1700 Double 0.09 20 39.21

(56) TABLE-US-00002 TABLE 2 Element Wt % - EDS At % - EDS Sn 91.13 (L shell)  83.14 (L shell) Ti 6.33 (K shell) 14.30 (K shell) Ag 2.55 (L shell)  2.56 (L shell)

(57) FIG. 4 shows quantitative measurement of elemental composition of alloy using EDS (DT between 30% to 40%). SEM image (8.2 mm WD, 15 kV EHT, SE2 detector). The specimen was polished up to 20 nm roughness.

(58) This alloy-silicon bond can survive subsequent printing of high temperature metals, like stainless steel 316L (melting point≈1400° C.). The Sn.sub.3Ag.sub.4Ti alloy forms an interlayer between silicon and the stainless steel that wets and bonds to both. A good bond was achieved by depositing three layers the alloy and from then on printing the stainless steel. Parts produced by this method are shown in FIGS. 5A and 5B, which show Sn.sub.3Ag.sub.4Ti alloy deposited on silicon by SLM process. FIG. 5C Stainless steel Binghamton University logo built on 4″ silicon wafer. 316L was selected because it demonstrates the temperature survivability. For heat transfer applications, high thermal conductivity metals, like Al and Cu can be used instead [72]. The choice of 316L stainless steel is to demonstrate robustness to hotter melting temperatures.

(59) The reliability of the bond to thermal cycling was probed via thermal cycling. The Si—Sn.sub.3Ag.sub.4Ti-SS316L samples were thermal cycled in Thermotron chamber between −40° C. and 130° C. for 100 cycles (over 1 week of testing). The temperatures of the parts are monitored to confirm the parts reach the chamber setpoints each cycle. After the 100 cycles, the parts were visually examined for defects, such as delamination, along the perimeter where thermal stresses are greatest. All parts passed the visual test without any noticeable failure or defect. FIG. 1 shows stainless steel parts fabricated by SLM on silicon wafer after thermal cycling test (100 cycles from −40° C. to 130° C.). (Note, the steel plate used to secure the silicon to the build platform has rusted over the course of the thermal cycling.)

(60) Characterization of Si-Alloy Interface

(61) FIG. 2 shows a sample with different patterns printed under varying conditions. Surface property and mechanical strength were studied versus processing parameters, including number of exposures. Silicon is held in a wafer holder. Wafer can be held by a vacuum chuck or adhesive and the chuck can include powder skirts to prevent powder ingress into undesired areas The strength of the Sn.sub.3Ag.sub.4Ti—Si bond was destructive shear tested with a Nordson DAGE 4000 Plus on hemispherical balls with a mean diameter of 432. μm. The stylus is positioned 1 μm from the interface with top raster speed of 100 μm/s. During the test it is observed that the balls did not fracture at the interface and rather scraped from the alloy. This indicates that the bond at the interface is stronger than the shear strength of the alloy.

(62) FIG. 8 shows the bond strength of the Sn.sub.3Ag.sub.4Ti-silicon fabricated by SLM technique (laser power 120 W, laser scan velocity 1100-2300 mm/s, double exposure). Consequently, what is being reported in FIG. 8 is the lower limit of the interfacial strength. The obtained data show shear strengths between 14.7 to 22.1 MPa. This value was expected as shear strength of pure Sn is 17.7 MPa and Sn-3.5 Ag is 31.7 MPa [73]. Each sample is repeated ten times with standard deviation of ˜10%. A control test on the silicon wafer was also performed with 0 μm liftoff, and the frictional force was negligible compared to the force due to the alloy. The slower laser scan velocities have higher lower limits of the interfacial strength, indicating stronger alloy at these processing conditions. Balls fabricated by single exposure generally obtained lower interfacial strength compared to double exposure.

(63) The solidified metal contact angle is a proxy for the interfacial bonding strength between the alloy and the silicon substrate. Better wetting implies a lower energy interface and stronger interfacial bonding [19]. In lieu of instantaneous contact angle measurements during selective laser melting, the contact angles of a solidified metal lines are studied (FIGS. 9A-9D) [74,75]. While the system is not in equilibrium during the transient cooling, the contact angle still provides insight and approximates the liquid contact angle [19]. The exposure parameters were varied, by varying scan rate from 1500 mm/s to 2300 mm/s while holding power at 120 W. Each solidified metal line was formed at a different laser exposure ranging from 1500 mm/s to 2300 mm/s and power of 120 watts. Afterwards the contact angles of each line were studied by optical profilometry and the height-to-diameter (H/D) ratio is used to extract the contact angle [19], θ.sub.equiv=2 a tan(2H/D). By lowering the velocity of scan, the laser heating period and maximum temperature increases, which improves the diffusion of reactants, assisting spreading. FIGS. 9A and 9B show examples of optical profilometry for a line of alloy on silicon created by 120 W, 1500 mm/s exposure parameter. FIGS. 9C and 9D show equivalent contact angle measurement versus laser scan velocity based on optical profilometry.

(64) Based on the Si—Ti phase diagram, it is expected to observe titanium-silicides at the interface as they are energetically favored [76,77]. The growth and formation of the titanium silicide rapidly forms over the limited molten metal phase and is limited by diffusion of Ti. Hence, a very thin silicide forms just at the interface. This is visible form the electron dispersive spectroscopy (EDS) mapping of the interface. The concentration of Ti is seen to have a bright line near the interface, as pointed out in FIG. 10F. This titanium-silicide band appears about 1 μm wide, but this is an artifact of the EDS interaction volume being on the order of a 1 μm.sup.3 [78]. The actual titanium-silicide layer formed in this process is sub-μm. Noteworthy, the laser-processed silicide thickness and EDS maps are comparable to conventional brazing to silicon [18].

(65) FIG. 10A shows a representative EDS spectrum of the alloy-silicon interface (8.7 mm WD, 10 kV EHT, DT 20%). FIG. 10B shows a back scatter detector scanning electron microscopy image of the interface shows light elements (i.e., Ti) darker than heavy elements (i.e., Sn, Ag) (8.7 mm WD, 10 kV EHT, RBSD detector). FIG. 10C shows EDS mapping of the SiK peak. FIG. 10D shows EDS mapping of the, AgL peak. FIG. 10E shows EDS mapping of the SnL peak. FIG. 10F shows EDS mapping of the TiK peak. The scalebar is 1 μm.

(66) The energy barrier for diffusion and crystallization must be overcome with thermal energy, so the reaction rate is expected to have an Arrhenius rate equation, exp(−E.sub.a/k.sub.B T), where N is the nucleation rate, E.sub.a is the activation energy, k.sub.B is the Boltzmann constant, and T is absolute temperature [62]. Due to the rapid laser processing temperature, diffusion and nucleation can overcome the energy barrier more rapidly than conventional brazing and proceed more rapidly than reactions occurring just above the melting point of the metal, which explains why the bonding is rapid. Despite faster reaction rates, the rapid cooling rates are on the order of millions of degrees Celsius per second, which leads to quenching and non-equilibrium properties.

(67) Properties of the Alloy on Si

(68) On the test designed for surface and interface properties, six rectangles were created with exposure parameters similar to sample 2 in terms of hatch distance, power and scan speed but with single, double and triple exposures (FIGS. 11A, 11B, and 11C). Surface optical profilometry reviled improvement on surface roughness from first to second exposure. However, the third exposure did not improve the surface properties any further. FIGS. 11A-11C show a surface roughness study by optical profilometry on three samples with different number of rasters (120 W, 1700 mm/s, 0.09 hatch distance).

(69) Grazing incidence X-ray diffraction (GIXRD) was performed on a bulk Sn.sub.3Ag.sub.4Ti sample to identify the compounds that appear in the laser processed alloy, and compare them to the conventionally processed braze. See FIG. 12, which shows phase identification of Sn—Ti—Ag alloy fabricated by SLM process using GIXRD (PANalytical X'Pert PRO High-Resolution X-ray Diffractometer). The dominant signal is from the β-Sn phase, as expected from the large concentration of Sn in the alloy. There are also weak, yet distinct peaks from the intermetallic compounds Sn.sub.5Ti.sub.6 and Ag.sub.3Sn. These formations agree with what is thermodynamically expected from phase diagrams [79,80]. Moreover, this agrees with prior conventionally processed Sn.sub.3Ag.sub.4Ti alloy phases, showing that laser processing and conventional brazing results in similar phases being formed [18, 21].

(70) Bonding to Graphite with Sn.sub.3Ag.sub.4Ti:

(71) One of the key challenges in additively manufacturing metal parts is removing the printed part from build platform. Conventional platforms are made of the same material as the metal powder to achieve reliable bonding and resistance to thermal stress. Support or solid structures are used to provide required offset for the cutting blade to detach the part from building platform followed by detailed machining to achieve proper surface finish. This process adds additional machining, particularly challenging for difficult to machine metals, like titanium or stainless steel. Another downside of the current build plate paradigm is that current building platforms used in commercial metal printing devices are heavy. These blocks need to be removed each time the process completes. Since graphite is lightweight, graphite buildplates can help to make handling and post-machining easier and safer.

(72) The graphite crystal structure consists of strong Graphene planes (sp.sup.2 hybridization of carbon atoms) which are connected by π-orbitals resulting in weaker interplanar Van der Waals forces [81,82]. The thermal conductivity in the graphene plane is extraordinary (>1700 W/m-K) over eight times better than aluminum. Furthermore, graphite is 16% lighter than aluminum. Graphite also possesses beneficial properties for nuclear applications. The thermal conductivity of pyrolytic graphite makes it an attractive material for use as a heat spreader, especially when coupled with a heat pipe or heat exchanger that is additively printed.

(73) Studying wetting of metal alloys on substrates of dissimilar material can provide valuable insights regarding bonding. However, there are debates regarding the mechanism of wetting in metal alloy systems at high temperatures. In non-reactive systems, it is hypothesized that wetting occurs by displacement of molecules of advancing front on adsorption sites of the substrate originated from surface tension. Interdiffusion and dissolution of the substrate directly affects the composition of the liquid and as a result solid-liquid interfacial energy [9]. For reactive systems, wetting is limited by the diffusion of reactive elements and local reaction kinetics [10].

(74) The joining of dissimilar materials such as graphite to common metals employed in additive manufacturing is challenging due to these metals having contact angles greater than 130° (i.e., non-wetting) [81]. Reactive metals like Ti, Ta, Zr, and Nb have been bonded to graphite by employing carbides of the same metal as an interlayer. This process is done in furnace at temperatures around 1000° C. for 90 minutes [83]. The reason for wetting behavior is carbide formation at the interface by diffusive transport of reactive species which result in modification of specific surface energies of the system [81]. The bonding time in these conventional bonding techniques is very long (minutes to hours), compared to laser exposure time in additive manufacturing (˜100 μs).

(75) Previously, rods of Graphite and SUS304 Stainless Steel were joined using solid state diffusion bonding method under compressive pressure of 25 MPa, annealing temperature of 664° C. and duration of 120 minutes. Clearly, this methodology is not applicable in the present method as basic requirements such as annealing time and pressure cannot be satisfied [23]. In another study, brazed carbon-carbon composite was brazed to Stainless Steel 316L at a temperature of 1050° C. using BNi-7 (Ni-14Cr-10.5P-0.1Si) filler alloy. Observation of the microstructure confirmed the carbide formation. Joint shear strength of approximately 16 MPa reported at 1 mm thickness of Nb interlayer [24]. Sn—Ag—Ti alloys are considered as “active” alloys which makes them suitable candidate for bonding of metals to various materials such as ceramics and carbon materials at temperatures between 250° C. to 450° C. [15] Activity of elements under required thermodynamic condition is of interest as it influences the effectiveness of the bond between alloy and substrate. Based on a thermodynamics study on Sn—Ag—Ti filler alloys, the activity of Ti increases with weight percentage between 0 and 2.25 wt % at 450° C. The addition of silver at low mass fractions (e.g., 3 wt %) tends to increase the titanium activity at 450° C.[59]

(76) The contact angle of Sn—Ti and Ag—Ti alloys on graphite was reduced to under 10° by adding approximately 2-5 wt % of titanium to the systems. As a result, it is expected to see acceptable wetting on graphite. Also in the case of using the alloy as interlayer to bond steel to graphite, the reactive wetting mechanism of Sn/Fe system can is described by dissolution of Fe and precipitation of FeSn.sub.2 at the interface [25]. Similarly, intermetallics form with this alloy and Cu (e.g., Cu.sub.3Sn, Cu.sub.6Sn.sub.5). Sn is also known to wet aluminum at 350-450° C.[25]

(77) Superfine isomolded graphite with grade of GM-10 was purchased from Graphitestore.com and machined into samples 1″×1″×0.125″ to fit experimental configuration. Similar performance can also be achieved on higher quality pyrolytic graphite with slightly adjusted processing parameters. Building on pyrolytic graphite may be desired for certain applications that demand high thermal conductivity and excellent heat spreading, as the in-plane thermal conductivity exceeding 1,700 W/m-K [84].

(78) As initial test, stainless steel 316L was deposited on the graphite surface by various exposure parameters. The idea was that 316L steel has minimum 17 percent chromium content which might assist the bond by creating carbides at the interface [85]. The results indicated poor wetting in all tests, as shown in FIG. 13A, which presents an optical image of stainless steel 316L deposited on graphite exhibiting poor wetting. The fixture designed to hold graphite substrate on build platform of EOS M290 is shown in FIG. 13B.

(79) As mentioned above, various elements are investigated to find a metal or alloy that can produce proper bond with carbon materials which can be applied in additive manufacturing. Sn—Ti—Ag alloy is chosen next as it provides good bonding between broad range of materials including carbon, aluminum and iron based metals in conventional brazing. Silver, titanium and tin powders with respective average diameters of 5 μm, 10 μm and 40 μm and purities of above 99% were mixed. Working with various diameter of powders show that size distribution has great importance in homogeneity and quality of uniform deposition on substrate.

(80) The exposure power was set to 200 watts and a scan rate of 1000 mm/s was used to ablate a rectangular volume at the interface of two graphite pieces with RMS depth of 29.83 μm (measured by Veeco NT1100 optical profilometer). Three-dimensional Topography of the surface at the edge of exposed area to laser scan can be seen in FIG. 14, which shows optical profilometry of ablated graphite. The area scan size is 1.7×2.3 mm. Laser ablation was used to create a roughened surface for better adhesion, and to expose clean graphite surface. Also, for wetting liquids, increases in surface roughness can improve wetting behavior [86]. This step could possibly be omitted, though with weaker bonding.

(81) Phase transformation and reactive wetting are dictated by laser matter interactions, transport and reactions during this process. Various powers ranging from 30 to 200 watts and scanning rates from 800 to 6500 mm/s and constant hatch distance of 0.09 mm were tested to find the exposure parameters that provide the highest wetting and bonding strength between graphite and the alloy. During this process, it is observed that at a fixed scanning speed, low power will result in partial sintering behavior and high power causes high vapor flux and large recoil pressure which leads to splattering. These process parameter trends with scan rate and power mostly agree with the general process parameters seen for other materials [61]. One key difference with regular stainless steel processing is that the mixed powder phase transformation is assisted by a double laser exposing each layer. The bonding appears to be a two-step process that requires a double exposure

(82) In an experiment, three layers of the alloy with thickness of 40 um each are deposited on graphite surface by laser melting. Afterwards, two layers of powder mixed with 50 wt % of stainless steel 316L deposited with same exposure parameters on the surface to create an elemental gradient in the build normal direction. Afterwards, a “Binghamton University” logo was built on the alloy with SS316L material using support structure exposure parameters to study the applicability of the interlayer method. Since the alloy meting temperature is under 450° C., the graphite with the structure built on it then removed from the build platform and placed on a hot plate. By increasing the temperature of the hot plate to 400° C. for one minute, the part can be separated from the graphite substrate. FIG. 15A shows the stainless steel logo printed on graphite. FIG. 15B shows removal of the printed part using hot plate.

(83) A sessile drop experiment is generally used for molten metal contact angle measurements. However, it is hard to accurately study instantaneous spreading process in case of selective laser melting, as phase change occurs fast. On the other hand, solidified sessile drop measurements can be used to give an estimate on the contact angle. However, since in SLM the molten metal drop solidifies before reaching its equilibrium state [74,75]. The contact angle can be estimated from the solidified droplet shape [87]. Studying the post-solidified triple contact line can in various exposure parameter provides information regarding the wetting mechanism during laser processing. Based on this ideology, an experiment is designed to create 5 lines of 120 μm width on graphite. Each line was scanned by a different laser exposure ranging from 5000 mm/s to 6000 mm/s and power of 150 watts. Afterwards the contact angles of each line are studied by optical profilometry. FIGS. 16A and 16B show spreading of the alloy on graphite substrate based on laser exposure parameters. The height-to-width (H/D) ratio is provided for each scanned line in order to have a better comparison. By lowering the velocity of scan, the laser heating period and maximum temperature increase, which improves the diffusion of reactants, assisting spreading.

(84) The energy barrier for crystallization must be overcome with thermal energy, so the reaction rate is expected to have an Arrhenius rate equation, N∝exp(−E.sub.a/k.sub.BT), where N is the nucleation rate, E.sub.a is the activation energy, k.sub.B is the Boltzmann constant, and T is absolute temperature [62]. Due to the rapid laser processing temperature, nucleation can overcome the energy barrier and proceed more rapidly than reactions just above the melting point of the metal, which explains why bonding occurs so rapidly. Despite faster reaction rates, the rapid cooling rates are on the order of millions of degrees Celsius per second, which leads to imperfect quenching of the interfacial layer with defects with respect to microstructure and surface coverage due to reaction and diffusion kinetics.

(85) Since the bond is under combined normal and shear stresses imposed by the thermal stress inherent to SLM and the recoating blade of SLM machine, it is necessary to measure the bonding strength of the joint. The common approach is to perform tensile tests to calculate yield tensile strength and shear yield strength of the joint. However, these tests are designed for isotropic or homogenous materials, which are not the case at the interface [88,89]. Direct measurement of tensile load at the joint is not straight forward since there is not enough space to grip the deposited layer of metal. To characterize the joint, a lap shear test according to FIG. 17 is performed, where the alloy layer functions as a weld connecting two graphite blocks that are loaded under tensile load.

(86) The designed mechanical testing reveals the lower limit of interfacial strength for the alloy in FIGS. 18A, 18B and 18C, which shows the Interfacial strength of the graphite-metal bonding at various laser scan velocities (FIGS. 18A and 18B). Sample preparation for tensile test (FIG. 18C). Since the fracture occurred in alloy for most of the samples rather than at the graphite-metal interface, only a lower limit of the interfacial strength can be estimated (shown by arrow).

(87) FIG. 19 shows an optical image using a polarized light microscope of the Sn—Ag—Ti alloy laser bonded onto graphite.

(88) FIG. 20 shows an unpolished scanning electron microscopy of graphite-alloy interface.

(89) Similar processing will work for printing onto pyrolytic graphite. FIG. 23 shows a print on a pyrolytic graphite sample. Pyroid® HT pyrolytic graphite substrate is thermally annealed in the manufacturing process to increase its thermal conductivity up to 1700 W/m-K at room temperature. Owing to the higher thermal conductivity of pyrolytic graphite, a greater energy density is required for bonding. This sample was made at laser scan speeds ranging from 4400 mm/s to 5600 mm/s and 150 W power and 20 μm layer thickness. Double exposure was applied to all samples. Ball shear testing is performed to measure interfacial strength of Sn.sub.3Ag.sub.4Ti alloy deposited by different laser processing parameters on pyrolytic graphite as shown on FIG. 24. 4400 mm/s and 150 W process parameters provided highest bond strength.

(90) Energy-dispersive X-ray spectroscopy (EDS)/SEM 15 kV MAG 300 performed on the interface of the joint is shown in FIGS. 21A and 21B. A scan across the interface shows diffusion of Carbon into the alloy. Elemental compositions observed at the interface show a titanium-tin-carbide interlayer [90]. Carbide formation is observed near the interface between graphite (bottom) and alloy (top).

(91) TABLE-US-00003 TABLE 3 Element C Ag Sn Ti At % 25.85 00.00 61.92 12.22 Wt % 03.77 00.00 89.13 07.10

(92) To examine the reliability of the bond in various thermal conditions, a thermal cycling test is performed. The bulk SS316L samples printed on graphite were tested using a Thermotron temperature chamber. During the test, the chamber temperature varied from −40° C. to 130° C. for 100 cycles. The temperatures of the parts are monitored. After the 100 cycles (over 1 week of testing), the parts were visually examined for defects, such as delamination, along the perimeter where thermal stresses are greatest. All parts passed the visual test without any noticeable failure or defects. FIGS. 22A and 22B show two stainless steel parts after the thermal cycling test.

(93) This technology efficiently and effectively employs a focused energy additive manufacturing process capable of bonding metals onto to graphite. Parts produced by this method can help effectively spread heat and take advantage of the high in-plane thermal conductivity of pyrolytic graphite. Furthermore, graphite can be used as an alternative build platform which decreases the cost of printing with stainless steel or titanium since current metal additively manufactured parts require machining to remove support, which requires either manually machining the parts or computer-controlled machining (CNC), which are both expensive and time-consuming. There is no need to print support structures as the part can be printed directly on the build platform and separated from it after print by melting the alloy connecting the stainless to graphite, or by brittle fracturing the graphite plate.

(94) Detailed Bonding to Silica-Based Compounds with Sn.sub.3Ag.sub.4Ti

(95) Experiments were conducted using a laser-powder bed fusion EOS M290 tool having a 500 W 1064 nm Yb-fiber laser and high speed galvanometer mirror system (maximum raster speed of 7000 mm/s). The machine has a N.sub.2 generator that maintains an N.sub.2 gas atmosphere with O.sub.2 concentrations of less than 1.5%. A special wafer tray is used to securely hold the silicon wafer.

(96) A Sn.sub.3Ag.sub.4Ti powder mixture is used as the bonding metal mixture, as it provides good bonding with silica and alumina via the Ti silicide and Ti oxide formation [58, 91-93], and also bonds to a broad range of other materials including graphite, Al, Cu, and Fe-based metals. This low melting point of this alloy (˜250° C.) reduces the thermal stresses experienced after solidification upon cooling. Sn, Ag, and Ti powders (>99% purity) with respective average diameters of 40 μm, 5 μm, 10 μm were mixed with weight percentages of 93%, 3%, and 4% (Sn.sub.87.55Ag.sub.3.11Ti.sub.9.34) (FIG. 2). Smaller powder sizes were chosen for the lower concentration metals to facilitate mixing and reduce diffusional resistance of the Ag and Ti in the Sn melt. FIG. 2 shows an SEM image (8.6 mm WD, 9 KV EHT, SE2 detector) of mixed Ag (3 wt %), Ti (4 wt %) and Sn (93 wt %) powders.

(97) FIGS. 25 and 26A and 26B show optical and scanning electron microscopy images of the first layer of printing Sn.sub.3Ag.sub.4Ti alloy onto a borosilicate glass substrate fractured at the interface. The laser processing parameters used for this deposition are given in table 4.

(98) TABLE-US-00004 TABLE 4 Scan Hatch Layer Energy Power speed distance thickness density Sample [W] [mm/s] Exposure [mm] [μm] [J/mm.sup.3] 1 120 1700 Double 0.09 20 39.21

(99) Energy dispersive X-ray spectroscopy mapping was applied at the interface of the sample with the Sn.sub.3Ag.sub.4Ti alloy deposited onto borosilicate glass to investigate the bonding, as shown in FIGS. 27A-27F. During directed energy deposition by the laser, the titanium in the alloy reacts to silica and create titanium oxide and titanium silicide. Presence of similar patterns for Ti, Si and O in EDS elemental mapping supports titanium oxides forming at the interface.

(100) Detailed Bonding to Al.sub.2O.sub.3-Based Compounds with Sn.sub.3Ag.sub.4Ti

(101) An EOS m290 system and Sn.sub.3Ag.sub.4Ti alloy powder was used to print on D50×0.43 mm Sapphire wafer with <0001> orientation and single side polish, installed on a steel build platform using Kapton tape. FIG. 28A shows the powder mounted on the build platform. The sapphire is unpolished on the side that was printed on. FIG. 28B shows a thin single layer of approximately 54 μm (the thickness of the Kapton tape) of Sn.sub.3Ag.sub.4Ti alloy powder, deposited on the unpolished surface with manual deposition. FIG. 28C shows the pattern of pillars and lines printed with this technique using a laser power of 150 W with scanning rate from 4400 to 5600 mm/s and a spot size of approximately 100 μm. As can be seen in the brightfield optical microscope images of FIG. 28D and FIG. 28E, good bonding can be achieved with a scan rate of 4400 mm/s and a power of 150 W. FIG. 28D shows bright field optical imaging at 5× of two lines, top with 150 W and 4400 mm/s and bottom with 150 W and 4600 mm/s. FIG. 28E shows bright field optical imaging at 20× magnification of 150 W 4400 mm/s line with Z-stack. Line thickness of 100 μm is achieved with edge exposure (single line raster).

Example 2

(102) The present technology provides an alternate technique for direct fabrication of heat transfer devices onto the chip by additive manufacturing, by building on recent hybrid bonding literature for interconnect manufacturing [98-101]. Previously lead solders, and more recently non-lead solders, have been used extensively in the electronic packaging industry to form interconnects, etc. However, reliability concerns are pushing the industry to move away from solders, towards a concept called hybrid bonding [94-97] which does not require solders. Hybrid bonding consists of multi-stage physical and chemical depositions onto a semiconductor device by employing photolithography-based techniques. This process produces a thin bonding layer via depositions that can then have features, like fins and microchannels, additively deposited via a powder-based laser melting process.

(103) In this example, ˜50 nm of titanium is deposited onto the back of the semiconductor die by physical vapor deposition (PVD), such as sputtering or evaporation as adhesion layer. This layer could be patterned or unpatterned. Then ˜250 nm of copper is deposited on top with similar techniques. Afterwards, 5-40 μm of copper layer is grown by electrodeposition onto the seed layer. This process is illustrated in FIG. 29. After metallizing the silicon with Cu, it is possible to 3d metal print heat sink structures made of copper directly on the metalized surface. Alternative embodiments may choose to make features out of other high conductivity metals, like silver or aluminum. This technique has an interface from titanium to the silicon via titanium silicides, and the first interlayer to the copper via Ti—Cu intermetallic formation.

(104) The disclosure has been described with reference to various specific embodiments and techniques. However, many variations and modifications are possible while remaining within the scope of the disclosure.

(105) References (each of which is expressly incorporated herein by reference in its entirety): [1] S. M. Ameli, Additive Layer Manufactured Sinter-Style Aluminium/Ammonia Heat Pipes, (n.d.) 182. [2] M. Ameli, B. Agnew, P. S. Leung, B. Ng, C. J. Sutcliffe, J. Singh, R. McGlen, A novel method for manufacturing sintered aluminium heat pipes (SAHP), Applied Thermal Engineering. 52 (2013) 498-504. doi:10.1016/j.applthermaleng.2012.12.011. [3] S. M. Thompson, Z. S. Aspin, N. Shamsaei, A. Elwany, L. Bian, Additive manufacturing of heat exchangers: A case study on a multi-layered Ti-6Al-4V oscillating heat pipe, Additive Manufacturing. 8 (2015) 163-174. doi:10.1016/j.addma.2015.09.003. [4] M. Wong, S. Tsopanos, C. J. Sutcliffe, I. Owen, Selective laser melting of heat transfer devices, Rapid Prototyping Journal. 13 (2007) 291-297. doi:10.1108/13552540710824797. [5] M. Wong, I. Owen, C. J. Sutcliffe, A. Puri, Convective heat transfer and pressure losses across novel heat sinks fabricated by Selective Laser Melting, International Journal of Heat and Mass Transfer. 52 (2009) 281-288. doi:10.1016/j.ijheatmasstransfer.2008.06.002. [6] Y. Li, Y. Shen, C.-H. Hung, M. C. Leu, H.-L. Tsai, Additive manufacturing of Zr-based metallic glass structures on 304 stainless steel substrates via V/Ti/Zr intermediate layers, Materials Science and Engineering: A. 729 (2018) 185-195. doi:10.1016/j.msea.2018.05.052. [7] I. Tomashchuk, D. Grevey, P. Sallamand, Dissimilar laser welding of AISI 316L stainless steel to Ti6-Al4-6V alloy via pure vanadium interlayer, Materials Science and Engineering: A. 622 (2015) 37-45. doi:10.1016/j.msea.2014.10.084. [8] M. K. Lee, J. G. Lee, Y. H. Choi, D. W. Kim, C. K. Rhee, Y. B. Lee, S. J. Hong, Interlayer engineering for dissimilar bonding of titanium to stainless steel, Materials Letters. 64 (2010) 1105-1108. doi:10.1016/j.matlet.2010.02.024. [9] E. Saiz, A. P. Tomsia, Atomic dynamics and Marangoni films during liquid-metal spreading, Nature Materials. 3 (2004) 903-909. doi:10.1038/nmat1252. [10] N. Eustathopoulos, Progress in understanding and modeling reactive wetting of metals on ceramics, Current Opinion in Solid State and Materials Science. 9 (2005) 152-160. doi:10.1016/j.cossms.2006.04.004. [11] S. Sugihara, K. Okazaki, K. Suganuma, Wetting of silicon single crystal by silver and tin, and their interfaces, Journal of Materials Science. 28 (1993) 2455-2458. doi:10.1007/BF01151679. [12] F. Baudin, L. Di Cioccio, V. Delaye, N. Chevalier, J. Dechamp, H. Moriceau, E. Martinez, Y. Bréchet, Direct bonding of titanium layers on silicon, Microsystem Technologies. 19 (2013) 647-653. doi:10.1007/s00542-012-1664-0. [13] J. Yu, Y. Wang, J.-Q. Lu, R. J. Gutmann, Low-temperature silicon wafer bonding based on Ti/Si solid-state amorphization, Applied Physics Letters. 89 (2006) 092104. doi:10.1063/1.2338574. [14] K. Holloway, R. Sinclair, Amorphous Ti—Si alloy formed by interdiffusion of amorphous Si and crystalline Ti multilayers, Journal of Applied Physics. 61 (1987) 1359-1364. doi:10.1063/1.338114. [15] R. W. Smith, Process of using an active solder alloy, U.S. Pat. No. 6,047,876A, 2000. [16] Y. Tang, G. Li, Thermodynamic Study of Sn—Ag—Ti Active Filler Metals, Physics Procedia. 25 (2012) 30-35. doi:10.1016/j.phpro.2012.03.045. [17] R. W. Smith, R. Redd, Electronic package formed using low-temperature active solder including indium, bismuth, and/or cadmium, EP1749315B1, 2009. [18] R. Kolenak, Igor Kostolný, Martin Sahul, Direct bonding of silicon with solders type Sn—Ag—Ti, 28 (2016) 149-158. doi:doi.org/10.1108/SSMT-11-2015-0040. [19] Nicolas Eustathopoulus, Michael G. Nicolas, Béatrice Drevet, eds., Wettability at High Temperatures, Pergamon, 1999. [20] K. Holloway, K. B. Do, R. Sinclair, Interfacial reactions on annealing molybdenum-silicon multilayers, Journal of Applied Physics. 65 (1989) 474-480. doi:10.1063/1.343425. [21] W.-L. Wang, Y.-C. Tsai, Microstructural characterization and mechanical property of active soldering anodized 6061 Al alloy using Sn-3.5Ag-xTi active solders, Materials Characterization. 68 (2012) 42-48. doi:10.1016/j.matchar.2012.03.007. [22] M. Tashiro, A. Kasahara, Method of bonding graphite to metal, U.S. Pat. No. 5,904,287A, 1999. [23] H. Sueyoshi, T. Nishida, Solid State Bonding of Graphite to SUS304 Steel, Materials Transactions, JIM. 41 (2000) 414-419. doi:10.2320/matertrans1989.41.414. [24] J. Y. Liu, S. Chen, B. A. Chin, Brazing of vanadium and carbon-carbon composites to stainless steel for fusion reactor applications, Journal of Nuclear Materials. 212-215 (1994) 1590-1593. doi:10.1016/0022-3115(94)91095-2. [25] Q. Lin, W. Zhong, F. Li, W. Yu, Reactive wetting of tin/steel and tin/aluminum at 350-450° C., Journal of Alloys and Compounds. 716 (2017) 73-80. doi:10.1016/j.jallcom.2017.05.036. [26] T. H. Chuang, M. S. Yeh, Y. H. Chai, Brazing of zirconia with AgCuTi and SnAgTi active filler metals, Metall and Mat Trans A. 31 (2000) 1591-1597. doi:10.1007/s11661-000-0169-0. [27] E. Saiz, A. P. Tomsia, R. M. Cannon, Ridging effects on wetting and spreading of liquids on solids, Acta Materialia. 46 (1998) 2349-2361. doi:10.1016/S1359-6454(98)80016-5. [28] K. Landry, C. Rado, R. Voitovich, N. Eustathopoulos, MECHANISMS OF REACTIVE WETTING: THE QUESTION OF TRIPLE LINE CONFIGURATION, Acta Materialia. 45 (1997) 3079-3085. doi:10.1016/S1359-6454(96)00372-2. [29] D. Lu, C. P. Wong, eds., Materials for Advanced Packaging, Springer US, Boston, Mass., 2009. doi:10.1007/978-0-387-78219-5. [30] F. J. Mesa-Martinez, M. Brown, J. Nayfach-Battilana, J. Renau, Measuring power and temperature from real processors, in: IEEE, 2008: pp. 1-5. doi:10.1109/IPDPS.2008.4536423. [31] Y. H. Lin, C. M. Tsai, Y. C. Hu, Y. L. Lin, C. R. Kao, Electromigration-induced failure in flip-chip solder joints, Journal of Electronic Materials. 34 (2005) 27-33. doi:10.1007/s11664-005-0176-4. [32] Z. Huang, R. E. Jones, A. Jain, Experimental investigation of electromigration failure in Cu—Sn—Cu micropads in 3D integrated circuits, Microelectronic Engineering. 122 (2014) 46-51. doi:10.1016/j.mee.2014.03.003. [33] E. Le Bourhis, Glass: Mechanics and Technology, 2nd ed., Wiley-VCH, 2014. [34] W. H. Zachariasen, The Atomic Arrangement In Glass, Journal of the American Chemical Society. 54 (1932) 3841-3851. doi:10.1021/ja01349a006. [35] B. E. Warren, X-Ray Determination Of The Structure Of Glass, Journal of the American Ceramic Society. 17 (1934) 249-254. doi:10.1111/j.1151-2916.1934.tb19316.x. [36] J. E. Shelby, Introduction to Glass Science and Technology, 2nd ed., Royal Society of Chemistry, Cambridge, 2007. doi:10.1039/9781847551160. [37] M. Hasanuzzaman, A. Rafferty, M. Sajjia, A.-G. Olabi, Properties of Glass Materials, in: Reference Module in Materials Science and Materials Engineering, Elsevier, 2016. doi:10.1016/B978-0-12-803581-8.03998-9. [38] M. D. Ediger, Perspective: Highly stable vapor-deposited glasses, The Journal of Chemical Physics. 147 (2017) 210901. doi:10.1063/1.5006265. [39] L. Yang, H. Y. Li, P. W. Wang, S. Y. Wu, G. Q. Guo, B. Liao, Q. L. Guo, X. Q. Fan, P. Huang, H. B. Lou, F. M. Guo, Q. S. Zeng, T. Sun, Y. Ren, L. Y. Chen, Structural responses of metallic glasses under neutron irradiation, Scientific Reports. 7 (2017). doi:10.1038/s41598-017-17099-2. [40] D. Kundu, G. De, B. Karmakar, A. Patra, D. Ganguli, Sol-gel preparation of silica glass, Bulletin of Materials Science. 15 (1992) 453-457. doi:10.1007/BF02745295. [41] K. Kihara, An X-ray study of the temperature dependence of the quartz structure, European Journal of Mineralogy. 2 (1990) 63-78. doi:10.1127/ejm/2/1/0063. [42] K. Bettger, E. Brown, E. N. Boulos, D. H. Stark, Glass-to-metal bond structure, US20100119740A1, 2010. [43] M. P. William, H. B. Pardam, P. Graham, Process of bonding glass or ceramic to metal, U.S. Pat. No. 3,220,815A, 1965. [44] L. L. Hench, P. J. Buscemi, Method of bonding a bioglass to metal, U.S. Pat. No. 4,159,358A, 1979. [45] R. W. Bruce, D. L. III, M. Kahn, A. W. Fliflet, S. H. Gold, Microwave assisted reactive brazing of ceramic materials, U.S. Pat. No. 7,022,198B2, 2006. [46] C. Dresbach, A. Krombholz, M. Ebert, J. Bagdahn, Mechanical properties of glass frit bonded micro packages, Microsystem Technologies. 12 (2006) 473-480. doi:10.1007/s00542-005-0031-9. [47] R. Knechtel, Glass frit bonding: an universal technology for wafer level encapsulation and packaging, Microsystem Technologies. 12 (2005) 63-68. doi:10.1007/s00542-005-0022-x. [48] R. Knechtel, M. Wiemer, J. Frömel, Wafer level encapsulation of microsystems using glass frit bonding, Microsystem Technologies. 12 (2006) 468-472. doi:10.1007/s00542-005-0036-4. [49] K. Nötzold, C. Dresbach, J. Graf, B. Böttge, Temperature dependent fracture toughness of glass frit bonding layers, Microsystem Technologies. 16 (2010) 1243-1249. doi:10.1007/s00542-010-1037-5. [50] M. Petzold, C. Dresbach, M. Ebert, J. Bagdahn, M. Wiemer, K. Glien, J. Graf, R. Muller-Fiedler, H. Hofer, Fracture mechanical life-time investigation of glass frit-bonded sensors, in: Thermal and Thermomechanical Proceedings 10th Intersociety Conference on Phenomena in Electronics Systems, 2006. ITHERM 2006., IEEE, San Diego, Calif., 2006: pp. 1343-1348. doi:10.1109/ITHERM.2006.1645501. [51] Z. Sun, D. Pan, J. Wei, C. K. Wong, Ceramics bonding using solder glass frit, Journal of Electronic Materials. 33 (2004) 1516-1523. doi:10.1007/s11664-004-0093-y. [52] R. J. Churchill, U. Varshney, H. P. Groger, J. M. Glass, Laser brazing for ceramic-to-metal joining, U.S. Pat. No. 5,407,119A, 1995. [53] Materials Science International Team MSIT®, Ag—Cu—Ti (Silver-Copper-Titanium), in: G. Effenberg, S. Ilyenko (Eds.), Non-Ferrous Metal Systems. Part 3, Springer Berlin Heidelberg, Berlin, Heidelberg, 2007: pp. 63-74. doi:10.1007/978-3-540-47004-5_10. [54] H. Mizuhara, Silver-copper-titanium brazing alloy containing crust inhibiting element, U.S. Pat. No. 4,883,745A, 1989. [55] M. L. Santella, J. A. Horton, J. J. Pak, Microstructure of Alumina Brazed with a Silver-Copper-Titanium Alloy, Journal of the American Ceramic Society. 73 (1990) 1785-1787. doi:10.1111/j.1151-2916.1990.tb09835.x. [56] Y. Sechi, K. Nagatsuka, K. Nakata, Effect of composition of titanium in silver-copper-titanium braze alloy on dissimilar laser brazing of binder-less cubic boron nitride and tungsten carbide, IOP Conference Series: Materials Science and Engineering. 61 (2014) 012019. doi:10.1088/1757-899X/61/1/012019. [57] A. Gasse, N. Eustathopoulos, Brazing composition and method for brazing parts made of alumina-based materials with said composition, U.S. Pat. No. 6,616,032B1, 2003. [58] Y. S. Chaug, N. J. Chou, Y. H. Kim, Interaction of Ti with fused silica and sapphire during metallization, Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films. 5 (1987) 1288-1291. doi:10.1116/1.574792. [59] Y. Tang, G. Li, Thermodynamic Study of Sn—Ag—Ti Active Filler Metals, Physics Procedia. 25 (2012) 30-35. doi:10.1016/j.phpro.2012.03.045. [60] S. A. Khairallah, A. Anderson, Mesoscopic simulation model of selective laser melting of stainless steel powder, Journal of Materials Processing Technology. 214 (2014) 2627-2636. doi:10.1016/j.jmatprotec.2014.06.001. [61] S. A. Khairallah, A. T. Anderson, A. Rubenchik, W. E. King, Laser powder-bed fusion additive manufacturing: Physics of complex melt flow and formation mechanisms of pores, spatter, and denudation zones, Acta Materialia. 108 (2016) 36-45. doi:10.1016/j.actamat.2016.02.014. [62] R. Lacmann, Crystallization, Third Edition. J. W. MULLIN, Butterworth-Heinemann, Oxford 1997, 527 Seiten, zahlr. Abb. und ISBN 0-7506-3759-5, Chemie Ingenieur Technik-CIT. 70 (1998) 1468-1468. doi:10.1002/cite.330701126. [63] B. D. Iverson, T. W. Davis, S. V. Garimella, M. T. North, S. S. Kang, Heat and Mass Transport in Heat Pipe Wick Structures, Journal of Thermophysics and Heat Transfer. 21 (2007) 392-404. doi:10.2514/1.25809. [64] T. L. Bergman, F. P. Incropera, L. S. Adrienne, D. P. DeWitt, Introduction to heat transfer, John Wiley & Sons, 2011. [65] S. Fürtauer, D. Li, D. Cupid, H. Flandorfer, The Cu—Sn phase diagram, Part I: New experimental results, Intermetallics. 34 (2013) 142-147. doi:10.1016/j.intermet.2012.10.004. [66] J. P. Kruth, L. Froyen, J. Van Vaerenbergh, P. Mercelis, M. Rombouts, B. Lauwers, Selective laser melting of iron-based powder, Journal of Materials Processing Technology. 149 (2004) 616-622. doi:10.1016/j.jmatprotec.2003.11.051. [67] J. P. Kruth, X. Wang, T. Laoui, L. Froyen, Lasers and materials in selective laser sintering, Assembly Automation. 23 (2003) 357-371. doi:10.1108/01445150310698652. [68] L. Thijs, F. Verhaeghe, T. Craeghs, J. V. Humbeeck, J.-P. Kruth, A study of the microstructural evolution during selective laser melting of Ti-6Al-4V, Acta Materialia. 58 (2010) 3303-3312. doi:10.1016/j.actamat.2010.02.004. [69] J. Beuth, N. Klingbeil, The role of process variables in laser-based direct metal solid freeform fabrication, JOM. 53 (2001) 36-39. doi:10.1007/s11837-001-0067-y. [70] J. Gockel, J. Beuth, K. Taminger, Integrated control of solidification microstructure and melt pool dimensions in electron beam wire feed additive manufacturing of Ti-6Al-4V, Additive Manufacturing. 1-4 (2014) 119-126. doi:10.1016/j.addma.2014.09.004. [71] N. K. Roy, O. G. Dibua, W. Jou, F. He, J. Jeong, Y. Wang, M. A. Cullinan, A Comprehensive Study of the Sintering of Copper Nanoparticles Using Femtosecond, Nanosecond, and Continuous Wave Lasers, Journal of Micro and Nano-Manufacturing. 6 (2017) 010903. doi:10.1115/1.4038455. [72] D. D. Chung, Materials for thermal conduction, Applied Thermal Engineering. 21 (2001) 1593-1605. doi:10.1016/S1359-4311(01)00042-4. [73] T. Siewert, L. Stephen, S. David R., M. Juan Carlos, Database for Solder Properties with Emphasis on New Lead-free Solders, National Institute of Standards and Technology & Colorado School of Mines, 2002. [74] T. Young, III. An essay on the cohesion of fluids, Phil. Trans. R. Soc. Lond. 95 (1805) 65-87. doi:10.1098/rst1.1805.0005. [75] M. E. Schrader, Young-Dupre Revisited, Langmuir. 11 (1995) 3585-3589. doi:10.1021/la00009a049. [76] B. Predel, Si—Ti (Silicon-Titanium), in: O. Madelung (Ed.), Pu—Re—Zn—Zr, Springer-Verlag, Berlin/Heidelberg, 1998: pp. 1-3. doi:10.1007/10551312_2737. [77] R. J. Kematick, C. E. Myers, Thermodynamics of the Phase Formation of the Titanium Silicides, Chemistry of Materials. 8 (1996) 287-291. doi:10.1021/cm950386q. [78] J. I. Goldstein, D. E. Newbury, P. Echlin, D. C. Joy, C. E. Lyman, E. Lifshin, L. Sawyer, J. R. Michael, Scanning Electron Microscopy and X-ray Microanalysis, Springer US, Boston, Mass., 2003. doi:10.1007/978-1-4615-0215-9. [79] Solder Systems in Phase Diagrams & Computational Thermodynamics, (n.d.). www.metallurgy.nist.gov/phase/solder/solder.html (accessed Jul. 18, 2018). [80] I. Ansara, A. T. Dinsdale, M. H. Rand, eds., Thermochemical database for light metal alloys, EUR 18499 EN, European cooperation in the field of scientific and technical research, 1998. [81] N. Eustathopoulus, M. G. Nicolas, B. Drevet, eds., Chapter 8 Wetting properties of metal/carbon systems, in: Pergamon Materials Series, Pergamon, 1999: pp. 317-338. doi:10.1016/S1470-1804(99)80010-8. [82] H. Zabel, S. Solin, eds., Graphite Intercalation Compounds I, Springer Berlin Heidelberg, Berlin, Heidelberg, 1990. doi:10.1007/978-3-642-75270-4. [83] M. Tashiro, A. Kasahara, Method of bonding graphite to metal, U.S. Pat. No. 5,904,287 A, 1999. [84] G. A. Slack, Anisotropic Thermal Conductivity of Pyrolytic Graphite, Phys. Rev. 127 (1962) 694-701. doi:10.1103/PhysRev.127.694. [85] EOS Metal Materials for Additive Manufacturing, (n.d.). www.eos.info/material-m (accessed Nov. 28, 2017). [86] N. Eustathopoulus, M. G. Nicolas, B. Drevet, eds., Chapter 1 Fundamental equations of wetting, in: Pergamon Materials Series, Pergamon, 1999: pp. 1-53. doi:10.1016/S1470-1804(99)80003-0. [87] N. Eustathopoulus, M. G. Nicolas, B. Drevet, eds., Chapter 3 Methods of measuring wettability parameters, in: Pergamon Materials Series, Pergamon, 1999: pp. 106-147. doi:10.1016/S1470-1804(99)80005-4. [88] R. M. Christensen, A Two-Property Yield, Failure (Fracture) Criterion for Homogeneous, Isotropic Materials, J. Eng. Mater. Technol. 126 (2004) 45-52. doi:10.1115/1.1631024. [89] Y. Flom, 2—Strength and margins of brazed joints, in: D. P. Sekulić (Ed.), Advances in Brazing, Woodhead Publishing, 2013: pp. 31-54. doi:10.1533/9780857096500.1.31. [90] Y. C. Zhou, H. Y. Dong, B. H. Yu, Development of two-dimensional titanium tin carbide (Ti2SnC) plates based on the electronic structure investigation, Materials Research Innovations. 4 (2000) 36-41. doi:10.1007/s100190000065. [91] A. K. Podshivalova, I. K. Karpov, Thermodynamic analysis of the stability of titanium oxides in the TiO—TiO2 range, Russian Journal of Inorganic Chemistry. 52 (2007) 1147-1150. doi:10.1134/50036023607070273. [92] H. B. Liu, L. X. Zhang, L. Z. Wu, D. Liu, J. C. Feng, Vacuum brazing of SiO2 glass ceramic and Ti-6Al-4V alloy using AgCuTi filler foil, Materials Science and Engineering: A. 498 (2008) 321-326. doi:10.1016/j.msea.2008.08.008. [93] R. Gordon, Chemical vapor deposition of coatings on glass, Journal of Non-Crystalline Solids. 218 (1997) 81-91. doi:10.1016/S0022-3093(97)00198-1. [94] R. A. Sosa, K. Mohan, L. Nguyen, R. Tummala, A. Antoniou, V. Smet, Cu Pillar with Nanocopper Caps: The Next Interconnection Node Beyond Traditional Cu Pillar, in: 2019 IEEE 69th Electronic Components and Technology Conference (ECTC), 2019: pp. 655-660. [95] S. W. Ho, L. Ding, S. H. Lim, S. A. Sek, M. Yu, G. Q. Lo, Polymer-based fine pitch Cu RDL to enable cost-effective re-routing for 2.5D interposer and 3D-IC, in: 2013 IEEE 15th Electronics Packaging Technology Conference (EPTC 2013), IEEE, Singapore, 2013: pp. 435-439. doi:10.1109/EPTC.2013.6745758. [96] C. Nair, F. Pieralisi, F. Liu, V. Sundaram, U. Muehlfeld, M. Hanika, S. Ramaswami, R. Tummala, Sputtered Ti—Cu as a superior barrier and seed layer for panel-based high-density RDL wiring structures, in: 2015 IEEE 65th Electronic Components and Technology Conference (ECTC), IEEE, San Diego, Calif., 2015: pp. 2248-2253. doi:10.1109/ECTC.2015.7159916. [97] T. Takano, H. Kudo, M. Tanaka, M Akazawa, Submicron-Scale Cu RDL Pattering Based on Semi-Additive Process for Heterogeneous Integration, in: 2019 IEEE 69th Electronic Components and Technology Conference (ECTC), 2019: pp. 94-100. [98] Z. R. Camacho, L. C. H. Tay, H. D. Bathan, J. D. Punzalan, Method of forming a wafer level package with RDL interconnection over encapsulant between bump and semiconductor die, U.S. Pat. No. 7,888,181B2, 2011. [99] K. Suzuki, Method of forming low-resistivity recessed features in copper metallization, U.S. Pat. No. 7,704,879B2, 2010. [100] P. R. Besser, D. M. Erb, S. Lopatin, Selective deposition process for passivating top interface of damascene-type Cu interconnect lines, U.S. Pat. No. 6,455,425B1, 2002. [101] V. Dubin, Self-encapsulated copper metallization, U.S. Pat. No. 6,249,055B1, 2001. [102] K. Hoshino, Semiconductor device using copper metallization, U.S. Pat. No. 4,985,750A, 1991. [103] R. Khazaka, D. Martineau, T. Youssef, T. Long Le, S. Azzopardi, Direct Printing of Heat Sinks, Cases and Power Connectors on Insulated Substrate Using Selective Laser Melting Techniques, in: 2019 IEEE 69th Electronic Components and Technology Conference (ECTC), 2019: pp. 2173-2179. [104] R Tummala, Fundamentals of Microsystems Packaging, McGraw-Hill, New York, 2001. [105] A. Sakanova, C. F. Tong, A. Nawawi, R. Simanjorang, K. J. Tseng, A. K. Gupta, Investigation on weight consideration of liquid coolant system for power electronics converter in future aircraft, Applied Thermal Engineering, Vol. I 04, pp. 603-615, 2016. [106] E. M. Dede, S. N. Joshi, F. Zhou, Topology Optimization, Additive Layer Manufacturing, and Experimental Testing of an Air-Cooled Heat Sink, Journal of Mechanical Design, Vol. 137, 2015, DOT: I 0.1115/1.4030989 [107] M. Wong, I. Owen, C. J. Sutcliffe, A. Puri, Convective heat transfer and pressure losses across novel heat sinks fabricated by Selective Laser Melting, International Journal of Heat and Mass Transfer, Vol. 52, pp. 281-288, 2009 [108] A. Syed-Khaja, A. Perinan Freire, C. Kaestle, J. Franke, Feasibility Investigations on Selective Laser Melting for the Development of Microchannel Cooling in Power Electronics, IEEE 67th Electronic Components and Technology Conference proceeding, 2017. [109] R. Skuriat, J. F. Li, P. A. Agyakwa, N. Mattey, P. Evans, C. M. Johnson, Degradation of thermal interface materials for high-temperature power electronics applications, Microelectronics Reliability, Vol. 53, pp. 1933-1942, 2013. [110] R. Khazaka, L. Mendizabal, D. Henry, R. Hanna, “Survey of high-temperature reliability of power electronics packaging components”, IEEE Transactions on Power Electronics, Vol. 30, No. 5, 2015. [111] L Ventola, E Chiavazzo, F Calignano, D Manfredi, P Asinari, Heat Transfer Enhancement by Finned Heat Sinks with Micro-structured Roughness, Microtechnology and Thermal Problems in Electronics, Journal of Physics: Conference Series 494, 2014. doi:IO.I088/1742-6596/494/I/O 12009. [112] Bey Vrancken, Study of residual stresses in Selective laser melting, PhD thesis at KU Leuven, Faculty of engineering science, 2016. [113] N. C. DeJong, L. W. Zhang, A. M. Jacobi, S. Balachandar, O. K. Tafti, A complementary experimental and numerical study of the flow and heat transfer in offset strip-fin heat exchangers, Transactions of the ASME: Journal of Heat Transfer, Vol. 120, pp. 690-698, 1998. [114] H. M. Joshi, R. L. Webb, Heat transfer and friction in the offset strip fin heat exchanger, International Journal of Heat and Mass Transfer, Vol. 30, pp. 69-84, 1987. [115] R. S. Matos, J. V. C. Vargas, T. A. Laursen, F. E. M. Saboya, Optimization study and heat transfer comparison of staggered circular and elliptic tubes in forced convection, International Journal of Heat and Mass Transfer, Vol. 44, pp. 3953-3961, 2001.