Laser directed energy deposition process
A robot arm and a fibre laser are being used together to help achieve a higher degree of accuracy with adding materials.
Metal additive manufacturing (AM) processes are finding applications in industrial sectors such as aerospace, automotive, biomedical and tooling. This technology is capable of reducing production time, eliminating the need for expensive tooling and fixtures, and can achieve complex geometries that are difficult to manufacture.
Although there is a variety of metal AM processes, many parts are made by either powder bed fusion (PBF) or direct energy deposition (DED) techniques. In PBF processes, a bed of powder is selectively sintered or melted via laser or electron beam heat source, while in DED, the feedstock material in the form of powder or wire is continuously fed into the melt pool, formed by a highly focused energy source such as a laser, electron beam or arc.
A robotised laser DED process, developed at the Research Center for Advanced Manufacturing (RCAM), Dallas, USA, uses an industrial KUKA robot arm to provide flexible motion for the path planning of complex geometries along with a 4kW continuous wave (CW) fibre laser, from supplier IPG Photonics, that generates the heat source to melt the material. The result is less wasted material and a higher degree of part complexity.
Generally, powder and wire are the two forms of feedstock material in laser DED process and the appropriate selection of feedstock form depends upon different factors. Deposition rate and material usage efficiency are the key factors when it comes to the size of part. Material usage efficiency is typically less than 50% in powder-fed deposition, while it goes up to almost 100% when wire is used. Parts larger than 50cm should be fabricated by using wire as feedstock material.
In contrast, relatively smaller parts with more intricate geometries and small features could be printed by using powder-fed deposition. Feedstock selection also depends on the raw material availability. In general, a wider selection of wire feedstock is available in the market compared with powder. Titanium alloys, Inconel, nickel alloys, stainless steels, aluminium alloys, tantalum, niobium and molybdenum are wire materials that have been used in laser wire DED processes.
The choice of feedstock form could greatly influence the formation of defects such as porosity in the bulk of laser DED parts. Porosity is recognised as one of the main issues associated with AM parts and is under investigation by researchers in this field, trying to identify the different source of formation of pores.
RCAM has found that the source of porosity in laser powder-fed metal deposition is different than wire-fed process. Entrapment of gas, porous powder and lack of fusion (LoF) are the three major sources of formation of porosity in laser powder-fed DED parts. LoF defects are formed mainly due to insufficient laser power input to fully melt the material. In this case, the proper adjustment of the processing parameters such as laser power, travel speed and degree of overlap could mitigate the existence of unmelted material or LoF. The LoF porosity exists at interfacial boundaries of the beads or layers. Another common type of porosity, due to the gas entrapment, is formed because the laser powder-fed process uses a gas such as argon as the carrier to assist metal powder transfer to the melt pool.
Shielding gas protects the melt pool from oxidation, and thus promotes the gas entrapment inside the buildup. This type of porosity appears at intralayer of the deposition in a spherical shape, and generally does not form at a specific location. Moreover, the interaction of metallic powders and stability of powder flow are correlated with intralayer pores. Another type of porosity comes from pores inside the powder particles that are formed during the gas atomisation process. These are a potential source of intralayer porosity in the process. All aforementioned porosity types act as stress-raisers that can nucleate under loading and grow with increasing local plastic deformation. Voids start interacting that lead to increased porosity volume fraction, then local necking and failure occur.
In the laser wire-fed process, RCAM found the main source of pores is the LoF at the inter-layer/bead boundaries. These defects in laser wire deposition are produced due to the insufficient laser power or inappropriate setting of vertical overlap between layers. This type of porosity is observed in the first layers of deposition. The porosity could be attributed to the high heat transfer rate at the initial layers when the substrate is cold. By increasing the number of layers, as the cooling rate decreases, the density of pores is usually reduced significantly.
The pores are observed in elongated shapes with very sharp edges that typically result in high local stress. Also, pores range from 50μm–200μm in laser wire DED parts. The LoF pores might have large impact on weakening the tensile properties of the samples. Another type of porosity observed in the microstructure of samples is microvoids with the size of less than 1μm. The existence of such microvoids is the consequence of shrinkage from solidification. During the liquid to solid phase change, similar to welding process, microvoids could be nucleated and grown.
Metamaterials have properties such as zero or negative Poisson’s ratio, a bi-stable structure with a negative stiffness, a non-positive coefficient of thermal expansion and a controllable frequency band gap. These abnormal properties hold great potential for developing novel products for use in wave filtering, acoustic cloaking, vibration control and energy harvesting. Most existing fabrication processes for metallic metamaterials requires decomposing a structure into small units and join them with pins, adhesive, welding, or pressure-fit joints. Such a process often requires manual work and can result in stress concentrations.
Metallic metamaterials previously difficult to produce are possible because of AM technology. One approach is to use robotised laser DED process to produce metallic metamaterials. Five different structure designs of metamaterials are fabricated in RCAM – structures with enlarged positive Poisson’s ratio, negative Poisson’s ratio, bi-stable property, non-positive coefficient of thermal expansion, and frequency band gap. In a sample of a co-continuous metamaterial with a body-centered cubic structure, such a material can exhibit simultaneous wave filtering capability coupled with enhanced mechanical properties.
The frequency of complete band gaps can be attained by tailoring the geometrical arrangements and volume fraction of the co-continuous metamaterials. The interface between phases is controlled by the equation: f (x, y, z) = 3 (cos (x) + cos(y) + cos(z)) + 4 cos(x) cos(y) cos(z) – t = 0
where x, y, z are the Cartesian coordinates of points inside the cube and t denotes the volume fraction of the composite. In the current design, the space with f (x, y, z) > 0 is filled by invar, while the space with f (x, y, z) ≤ 0 is filled by stainless steel. The value of t is chosen to be -0.7, for which the stainless steel phase achieved a volume fraction of 35%. In a sample of a printed metamaterial composed of two materials by using robotised laser DED process, a microstructural analysis revealed a good metallurgical bonding was achieved at the interface of the two materials.
Simultaneous wire and powder feeding
Robotised laser DED processes can require multiple materials feeding nozzles at the same time. The benefit of this is producing functionally graded materials. There is an increasing demand to extend the lifespan of high-value parts such as drilling tools in the oil and gas industry or agricultural equipment, that are exposed to extreme wear and corrosive conditions. Using wire feedstock as the matrix material and adding powder particles such as tungsten carbide (WC) or titanium carbide (TiC) reinforcement, allowed us to fabricate composite components.
This way, the wear, corrosion and hardness properties of parts have been improved significantly. In one sample, clay was deposited on a tooth surface of a tool used for oil well drilling applications. The wire used was stainless steel 316L, which is compatible with the tool material, allowing a good metallurgical bonding between the clad and tool and WC powder particl
es and increased the wear-resistance and hardness value. The powder particles were uniformly distributed along the cross section of clad, producing its homogeneous mechanical and tribological properties.
In robotised laser powder DED, the main processing variables such as powder flow rate, laser power, layer thickness, travel speed, and step-over value, affect the quality of deposition and stability of the process. Optimisation and tuning of these parameters are necessary prior to the process in order to achieve consistency and stability.
However, due to the existence of disturbances such as change in the thermal condition, in material properties or process variables throughout the process, the implementation of an in-situ sensing and control system is required. There is a wide variety of monitoring techniques for laser DED process, such as thermal-based, vision-based and acoustic-based sensing methods. Among all, the vision-based are the most reliable techniques.
RCAM uses charge-coupled device (CCD) and complementary metal-oxide semiconductor (CMOS) cameras to capture the real-time image of melt pool in order to measure its length, width and area. But in order to extract useful information, image processing techniques are required. In the first step, a raw image of melt pool is acquired by a camera equipped with an infrared filter. Then, the image is converted to binary black and white by applying a user-defined threshold value. The threshold is obtained by depositing several tracks with different widths. Then the measurement of track widths is taken and compared with the corresponding images to get the appropriate threshold value.
Usually, the melt pool is surrounded by some flare from the hot spatter particles, which can cause considerable noise and error in measuring the melt pool width, so a low pass fast fourier transform (FFT) filter is employed. The FFT of an image represents the frequencies of occurrence of pixel intensity variations in the original image.
The consistent and smooth intensity variations in the image correspond to low frequencies in the FFT, while the abrupt and fast intensity variations in the image such as flare or noisy pixels at the edge of melt pool correspond to high frequencies in FFT.
A low pass filter with a cut-off frequency of 5% is used to remove the noise, while preserving the melt pool boundary. The frequency of each pixel is set to zero if it is higher than cut-off frequency, and remains unchanged if it is less than the cut-off frequency. In the next step, the outline of melt pool is extracted. Ultimately, in order to measure the widest section of melt pool, all possible circles contained inside the outline are detected. The largest diameter of circles is recognised as the melt pool width, therefore, the melt pool width as the control variable is obtainable in real-time.
Further, an adaptive, closed loop control algorithm is developed. The input variable to the control system is melt pool width, that is obtained after image processing, and the output of controller is laser power. A controller that relies on proportional-integral-derivative (PID) algorithm adjusts the laser power in real-time based on the feedback from camera. Using this approach, it is possible to maintain the melt pool width throughout the building process.
Analysis of the microstructure of the two thin-walled 316L samples built with and without closed loop control revealed that, while there is a sharp increase in the grain size of sample with constant laser power from bottom layers towards the top, there is only a slight increase in the grain size of the control sample. The control sample also shows more uniform, homogenous and finer microstructure. As the process progresses from bottom layers to top, due to the heat saturation and change in the heat transfer mode from 3D to 2D, the constant laser power in the system causes an increase in the melt pool size and temperature, causing lower cooling of the melt pool.
Eventually, grains have more time to grow and in some locations, their morphology changes from cellular to dendritic structures. However in the case of the controlled laser power sample, the controller tries to maintain the size of melt pool by compensating the laser power. Therefore, a gradual decrease in laser power results in lowest variation in grain size and morphology. The coarse columnar and dendritic grains elongated in the direction of thermal gradient are seen in the 316L sample with constant laser power, whereas fine cellular grains are seen in the sample with controlled laser power. Hence, tailoring the microstructural properties is feasible while the part is printed.
Experiments show that laser DED is a challenging process. Many precautions must be taken before the process starts, for instance, parameter optimisation and tuning must be taken to ensure a stable process. However, due to in-process, unexpected disturbances such as change in heat transfer mode or in thermal conductivity of material, pre-optimised parameters might not lead to a consistent and repeatable process. Therefore, process monitoring and feedback system is essential to ensure the quality of final buildup. Moreover, due to the non-linearity of the laser DED process, RCAM aims to develop an advanced adaptive feedback system that corresponds to the varying dynamic response between melt pool size and laser power as the process continues.
Meysam Akbari is Southern Methodist University Graduate Research Assistant and Radovan Kovacevic is RCAM Director.