Laser Additive Manufacturing of Pure Copper by F.M. Sciammarella PhD, M. Gonser PhD, M. Styrcula Mechanical Engineering Department of the College of Engineering & Engineering Technology, Northern Illinois University Keywords: Copper, laser additive manufacturing, process parameters, composition, thermal conductivity Abstract The deposition of pure copper via laser additive manufacturing (LAM) is generally limited due to its high reflectivity and high thermal conductivity. The development of process parameters to ensure the sustainable deposition of C11000 copper powder onto 4142 steel substrates was accomplished using thermal requirement calculations which are based on physical properties of the copper. Energy absorption was maximized, allowing for the building of 1mm thick layers of C11000 material onto the 4142 steel substrate. C11000 components were built up to a height of 12 mm with minimal defects. The developed process resulted in minimal dilution compared to conventional LAM with a 10 increase in deposition rate. In addition, the average grain size of the C11000 components was 20 microns. The hardness of the deposits was within the range of hardness values of wrought C11000. This research shows that it is possible to deposit pure copper onto ferrous substrates without having to use excessive laser power. Introduction The proliferation of copper components in the energy, automotive, and industrial sectors has been limited by the difficulties in processing and fabrication. With the introduction of laser materials processing over the last 20 years, the difficulties in utilizing copper have been slowly overcome. However, the fact that copper has a high reflectivity (thus low absorptivity) and high thermal conductivity makes deposition difficult compared to other nonferrous materials. The thermal conductivity of pure copper and its alloys (compared to other alloys) makes it the best candidate for thermal and electrical management applications. With the constant demand to reduce packaging size of electrical components, cycle times in molding operations, and maintain or reduce weight sizes of components (i.e. cars) copper is becoming the clear choice. For example the electric hybrid industry is considering replacing internal permanent magnet motors (IPM) with induction motors (IM) for their future vehicles [1]. They have comparable torque density and peak efficiency. Furthermore, they do not suffer a drag loss when the motor turns and do not lose efficiency during high speed, low torque conditions. Replacing the rotor from aluminum to copper will further increase the advantages. The copper rotor bars and end rings reduce electrical losses, making it a smaller, more efficient motor. Although the raw material costs for aluminum are less than copper, a motor with a copper rotor can be 15 25% smaller so cost may not be a factor. The superior heat conduction properties of copper also make thermal management easier, which can be another cost reduction for the IM. Work done by Whirlpool shows that when using copper alloys in the molds for plastic injection, cycle times are reduced by 20% since they can cool faster and warpage is reduced by more than 50% since the heat exposure is reduced [1].
In the electronic packaging industry extensive work using metal injection molding has shown that it is possible to design very complex heat sinks using copper alloys to satisfy the demand of smaller packaging [2, 3]. Others are researching the use of copper anodes for highly efficient direct liquid fuel cells capable of operating down to 25 o C. Injection molding research has focused on laser deposition of smaller amounts of copper to obtain all the qualities copper can offer in terms of reducing cycle (cooling) times but not suffer on overall performance and or increase manufacturing costs. From a manufacturing standpoint the main challenge is that when trying to use laser cladding or additive manufacturing, the very same properties making copper so great make it extremely difficult to process. In short these are just a few of the many potential applications in which copper is considered a strong candidate. The main focus of the laser additive manufacturing research has been to address the issues of cracking and porosity [4,5]. In [4] the authors used the Laser Engineered Net Shaping (LENS) system to deposit copper alloys on H13 tool steel and found that the range for solidification cracking was between 7.5-55%. This was also confirmed a few years later in [5] when using a similar laser cladding system. Further studies on the effects of cracking when depositing copper onto a steel alloy can be found [6,7]. The main arguments for porosity were to do with the powder particle morphology. When investigating these effects it is also very important to understand the role of heat input and thermal history which will also have a direct effect in cracking and porosity. Since copper has a very high thermal conductivity it is important to ensure there is enough heat in the process to maintain a stable temperature that can allow diffusion of copper into steel so as to avoid cracking. Furthermore, if there is insufficient energy in the melt pool, then regardless of powder morphology there will not be enough power or time to melt the powders which leads to porosity. This article will look into the processing conditions required to successfully deposit pure copper onto steel, while maintaining physical and mechanical properties of the copper within acceptable levels. Experimental Procedure For this investigation, 11000 copper powder was deposited with an Optomec LENS TM 850R system. Samples were produced via laser metal deposition under an argon atmosphere with oxygen content less than 10 ppm. An active ytterbium fiber laser was used as the heat source with beam waist diameter of 1mm. Figure 1 shows the types of deposit configurations that were produced with the process. Single layers (Figure 1(a), 1mm thick) were clad over the surface of 4142 steel plate and taller deposits (Figure 1(b), up to 12 mm in height) were also created. (a) (b) Figure 1. Examples of components made with process. (a) Clad layers and (b) 3D build.
The copper powder was melted and deposited onto 3 mm thick 4142 steel plates. The plate thickness ensured heat input control during processing. The size distribution of copper powder was -100/+325 mesh. CAD files were interpreted by the control software and converted into systems of coordinates for positioning. A Cartesian linear positioning system was used to traverse the laser heat source over the surface of the substrate. The operator specified layer thicknesses, layer spacing, and other parameters based on the method used. Laser power, travel speed of the deposition head, powder flow rate, and ultimately heat input were all variables in the analysis. A consistent and repeatable method (denoted as Process (I) and Process (II) in the text) was utilized to provide necessary heat input for deposition. A k-type (chromel-alumel) thermocouple was used to determine the temperature of the surface of the plate. An Hitachi TM-1000 SEM equipped with X-Ray energy dispersive spectroscopy (XEDS) capability was used to observe microstructure and quantify chemical composition of the copper deposits, respectively. A Leco microhardness tester equipped with a Vicker s needle was used with a load of 300g to quantify hardness of the copper deposits. Image analysis software was used to determine the ratio of phases in the eutectic as well as grain size. Specimens used for grain size analysis were etched with a reagent containing 1 g FeCl 3, 10 ml HCl, and 100 ml water. Image analysis software was used to determine grain size (microns) and to quantify porosity (volume percent) of the deposited copper samples. Process Results Initially the processing parameters (Power P, Travel speed V, and Powder flow rate pfr) were varied to maximize deposition efficiency. The first deposited layer was done so using 700W. Upon completion of the first layer, the laser power was increased to the values listed in Table 1. Maximized deposition efficiency is defined as achieving a relatively stable melt pool and able to deposit several repeatable layers. Once maximized, this was defined as the control process for conventional laser additive manufacturing (LAM) as shown in Figure 1b. As expected, the initial control process, yielded low deposition rates (see Table 1). The Control process did not produce a melt pool diameter that was conducive to copper deposition. The subsequent processes shown in Table 1only differ in thermal energy provided in comparison to control process resulting in a ~10 increase in deposition rate. The corresponding deposition rates of the Control, Process (I), and Process (II) were 0.14, 1.25, and 1.55 g/min, respectively. Table 1. Process parameters used during C11000 deposition Method Power, W V, mm/s pfr, g/min Layer thickness, mm Spacing, mm Control 900 2.1 1.6 0.25 0.50 Process (I) 900 2.1 3.3 0.75 1.50 Process (II) 990 2.1 3.3 0.75 1.50 To improve deposition rate, thermal energy was added to the substrate to increase heat input. The heat required for melting a given volume of copper metal was estimated. This equation is: Q = ρc p (T m -T o ) + L, where ρ is density (g/cm 3 ), C p is heat capacity (J/g-K), T m is the melting temperature (K), T o is the initial temperature of the substrate (K), and L is the latent heat of fusion. The value of Q represents the minimum energy required for stable melting. If T o is
higher than room temperature at the beginning of the process, then the energy contribution by the laser per unit length, given as H In = P/V, is required to push the total energy above the threshold Q for melting, as seen in Figure 2. The units for H In are J/cm, where P is laser power (W), and V is travel speed (mm/min). The blue portion of the bar indicates the experimental Q contributed by the laser beam, as derived from the heat required for melting equation. The red bar indicates the theoretical H In based on the operating parameters given in Table 1. In fact if we multiply Q by the cross-sectional area of a single deposited copper bead, the red Q quantity is very close to the blue H In quantity. The gray bars shown for each case indicates a rough calculation of the added energy so that process was above the threshold Q level required for melting. It was this added energy that made it possible for deposition to occur at relatively low laser powers for the initial layer. Another contributing factor was the dilution of the copper by the steel and the fact that steel has a higher heat capacity than copper, so more energy could be absorbed during the first layer deposition. However, as the build progressed (2 nd, 3 rd, and 4 th layers produced), and the fraction of copper in the deposit (in wt%) increased, the conductivity and reflectivity of the deposit increased, which warranted the higher laser powers of 900 and 900 W for Process (I) and Process (II), respectively. The 990W used in Process (II) was done to increase deposition rate. If the added energy provided in Process (I) is not supplied, then the heat generated by the laser is equal to that calculated for the Control (345 J/cm). When the Control process utilizes process parameters (P, V, pfr) that are identical to that of Process (I), then the energy required for melting per unit length is not reached since most of the energy is reflected due to the low absorptivity. Therefore, there is a minimum added energy input that is required for the substrate at which deposition is sustainable for copper powders. Again this is seen by the gray bar and changed as a result of changes in the processing parameters. Above this threshold, deposition is controllable. Based on absorptivity calculations for copper, the percentage of energy absorbed is near 10% at room temperature for a laser wavelength of 1.06 um. This quantity increases to 25% as the temperature of the copper approaches its melting temperature. Furthermore, the high reflectivity of copper makes it so that nearly 75 percent of the energy is reflected when the temperature of the copper is at its melting temperature. The H In calculation is, therefore, multiplied by a factor of 0.25. In addition, near 0.9T m of pure copper, the thermal conductivity decreases to 340 W/m- K. This allows for more energy absorption at the beam/workpiece junction. Just as heat capacity increases with temperature, the thermal conductivity decreases. Therefore, energy absorption will increase as the temperature increases. This is true for laser processes as well, ones that may be adversely affected by high reflectivity surfaces, such as copper. The aim was to ensure enough thermal energy into the build to encourage continuous deposition of the C11000 powder into the melt pool that forms on the surface of the copper deposit.
Heat required for melting (Q), J/cm 400 350 300 250 200 150 100 50 0 Threshold Q for melting 270 Q H In Q H In Q Process (I) Process (II) Control Processing Condition Added thermal energy Figure 2. Effect of improving heat input on build height during C11000 deposition. From this data, it is clear that the Control process has insufficient energy to sustain melting beyond the deposition rate of 0.14 g/min. With the ability to increase the heat input in the experiments this enabled an increase in powder flow rate resulting in much higher deposition rates. Again a minimum thermal energy threshold has to be exceeded in order for the C11000 powder to melt and consolidate on the surface of the build. Above this threshold, deposition of the powder was sustainable and consistent. Below this threshold, deposition was inconsistent and low (~80% decrease), as was the case for the Control process. Scanning Electron Microscopy (SEM) Results With the energy analysis done in the previous section it was important to perform a microstructural analysis and determine if there were any deleterious phases in the deposit. As mentioned in the introduction there is a range of Cu (7.5-55%) that when mixed with steel can cause solidification cracking. So even though the outward appearance of the deposits seemed acceptable an investigation was necessary. The SEM was used to determine both the microstructure and chemical composition of the copper deposits. For the Control process there is a distinct transition region that forms at the Fe/Cu interface (see Figure 3). This transition region showed evidence of eutectic formation which is a result of the phase separation as predicted by the Iron-Copper phase diagram (see Figure 4). Figure 3 also shows the composition results obtained from X-Ray energy dispersive spectroscopy (XEDS), which confirm the phase separation. This transition zone is not acceptable as it can severely affect mechanical properties as well the thermal conductivity.
Figure 3. Effect of Control process on transition layer thickness and microstructure. Macro back-scatter electron (BSE) image (left) showing transition layer between copper and steel materials. BSE image showing microstructure of transition layer (top right) showing evidence of eutectic phase separation. XEDS spectrum (bottom right) of eutectic microstructure. As shown above a eutectic forms in the transition region, with the light phase having a copper rich composition, and the dark phase, an iron-rich composition. The miscibility gap in the equilibrium binary Fe/Cu phase diagram in Figure 4 confirms that a phase separation occurs. The amount of copper (in wt%) increases the quantity of copper-rich phase in the microstructure based on the tie line rule [R/(S+R)], where R represents the amount of copper in wt% and the S represents the amount of iron in wt%. Until the solubility limit is reached at 96 wt% Cu, there is a phase separation. Above the transition layer and after 2 deposition layers are produced, the composition of the deposit is greater than 96 wt%, indicating that up to 4 wt% iron is dissolved in the copper microstructure. Figure 4. Equilibrium binary Fe-Cu phase diagram. Dashed lines represent S in tie line analysis.
Interestingly for Process (I) and (II) the effects of increasing heat input on the copper deposit near the Fe/Cu interface is clearly shown in Figure 5. The first difference between these and the Control shown in Figure 3 is the lack of a transition region (see Figure 5a). In most cases, there is a microscopic transition layer, but the thickness is significantly reduced. There was still dilution in the C11000 deposit (55 wt% Cu 45 wt% Fe directly above the Fe/Cu interface), but phase separation does not occur suggesting that the processing temperature is lowered or the kinetics do not support phase separation. The deposition of C11000 powder was possible at much lower laser powers (~700W) when the heat input increased, resulting in a microsctructure consisting of α-cu. Figure 5. (a) BSE image of bond interface between substrate and copper deposit from Proces (I). (b) Microstructure of C11000 deposit remote from the bond interface containing α-cu. The increase in heat input resulted in less dilution near the Fe/Cu interface resulting in purification of the C11000 deposit as evidenced in the XEDS spectrums in Figure 6. Even with considerable dilution by the 4142 steel substrate (55 wt% Cu 45 wt% Fe), the copper deposits, as well as their transition regions, did not show evidence of cracking. Lack of Fe-K α peak Figure 6. Control XEDS (left) and Process (I) XEDS (right). Evidence of iron contamination in deposit on left and copper purification on right.
Hardness Results The thermal effects on the steel resulted in the formation of a heat-affected zone adjacent to the deposited copper layer in all cases. The resulting cooling rates resulted in the formation of martensite, a common occurrence in laser material processing. This resulted in a hardness increase in the steel, but not a deleterious one. Vicker s hardness results in Figure 7 show that the changes in heat input only reduces the average hardness (~73 HV/300) of the deposited copper material by 2.5% compared to average wrought values (~75 HV/300). The ΔHV represents the range of hardness values for wrought C11000. The hardness values of the various processes (Control, Process I, and Process II) are within the range of hardness values typically seen in wrought C11000 (50-100 HV). Although slightly lower HV, the grain size of the laser deposited is still relatively small compared to wrought. As grain size decreases, yield strength, and ultimately, tensile strength, increases. This is explained by the Hall-Petch relationship [8]. The average grain size for Process (II) was 20 microns. Figure 7. Microhardness results for laser additive processing conditions and wrought C11000 products. Effect of Process on Thermal Conductivity Laser additive manufacturing often results in components that have a small quantity of porosity. Quantitative metallography indicates that the volume fraction of porosity in the copper deposits was in the range of 0.1 0.15 volume percent. The effect of porosity on the thermal conductivity of copper is marginal at this level. The reduction will be approximately 0.1 percent of the value of thermal conductivity at room temperature. The dilution effect on the copper deposit, as well as an optical image of the C11000 deposit microstructure are shown in Figure 8. The thermal conductivity was calculated based on the rule of mixtures starting from the Fe/Cu interface up to a 6 mm height in the C11000 build. At a height of 3mm, the dilution level of iron in copper falls to zero and the thermal conductivity is nearly recovered.
400 450 Composition, wt% 350 300 250 200 150 100 50 0 0 2 4 6 Height in build, mm 400 350 300 250 200 150 100 50 0 Thermal conductivity, w/m-k Cu, wt% Fe, wt% k, W/m-K Figure 8. Effect of dilution on the thermal conductivity of C11000 deposits (left) in Process (II). Optical micrograph of C11000 deposit from Process (II) showing evidence of porosity (200 magnification). Conclusions The work presented in this paper shows that it is possible to control and manage heat input during laser additive manufacturing making it an ideal candidate for depositing copper material onto ferrous and non-ferrous substrates for use in thermal management or corrosion applications. In this study it was possible to build a 12 cubic mm volume of C1100 copper onto a 4142 steel substrate. However, builds that exceed that height can potentially be built with reasonable cycle times (a 10 increase in deposition rate was achieved). Furthermore, dilution levels can be kept below threshold levels for cracking by increasing the thermal energy of the substrate. Though dilution occurs near the bonding interface with Process (I), this only slightly diminishes the thermal conductivity (300 W/m-K after the first layer deposits). More improvements in processing conditions can be made with the introduction of a calorimeter [9] to carefully develop processing window for deposition of copper using laser additive manufacturing. Finishing operations are needed to meet dimensional specifications, as is typical of any additive manufacturing processes in industry. The combination of increased deposition rate (10 ), comparable grain size, low dilution, and intermediate hardness compared to wrought components make the deposition of pure copper by LAM more feasible from an industrial perspective. References 1 http://www.copper.org/applications/industrial/cumolds/whirl.html 2 R. M. German and A. Bose, Injection Molding of Metals and Ceramics, 1997, Princeton, NJ, MPIF. 3 J.L. Johnson and Lye- King Tan, Electronics Cooling, 2004. 4 F.F Noecker II and J.N. DuPont, Functionally Graded Copper-Steel Using LENS Process, Rapid Prototyping of Materials, Symposium Proceedings, Columbus, OH,
United States, Oct. 7-10, 2002 (2002), 139-147 5 Vinay Kadekar, Sashikanth Prakash and Frank Liou, Experimental investigation of laser metal deposition of functonally graded copper and steel, 2004 6 Vainerman, A.E., Osetnik, A.A. The Formation of Cracks During the Deposition of Copper Alloys to Steel, Automatic Welding, Vol 21, June, 1968, pp 22-25. 7 Rick Noecker Cracking Susceptibility of Steel Copper Alloy, Advanced Material & Processes, Feb 2003. 8 Hansen N., Hall-Petch relation and boundary strengthening, Scripta Mateiralia, Vol. 51, Issue 8, pp. 801-806, October 2008. 9 Malin V., Sciammarella F.M., Controlling Heat Input by Measuring Net Power, Welding Journal, pp. 44-50, July 2006.