1 Introduction

Titanium-based alloys [1] find widespread use in various applications due to their outstanding properties, including a high melting point, low density, excellent corrosion resistance, and high strength [2,3,4,5,6,7]. Among these alloys, two-phase α/β Ti alloys stand out for their diverse applications, ranging from aerospace components to implants [8,9,10,11,12,13,14]. The microstructural features of α/β titanium alloys vary significantly based on the thermomechanical processing involved. One commonly used α/β Ti alloy is Ti64, renowned for its superior mechanical and biocompatible properties [15,16,17,18,19,20]. It can exhibit diverse microstructures depending on the specific thermomechanical processes it undergoes, resulting in equiaxed α-grains in a transformed β-matrix under equilibrium conditions and a needle-like α′ martensitic structure under non-equilibrium conditions [21,22,23,24,25,26,27].

Lately, there has been a growing interest in additive manufacturing (AM) from both industry and academia. In AM, parts are built layer by layer, allowing for the creation of intricate shapes and minimizing the need for material removal processes like milling and drilling to achieve the desired part shape [28,29,30]. Additive manufacturing can reduce the material wastage during manufacturing which is significantly higher in traditional manufacturing methods, subtractive manufacturing. In the past, these layer-wise production techniques were primarily used for prototyping mainly because they could only process polymers [31,32,33]. While polymer components could replicate the part’s shape, they fell short in matching the mechanical properties required for various applications. However, over the last decade, several AM techniques for processing metals have been introduced, making it feasible to produce components with properties similar to conventionally manufactured parts [34,35,36]. Some of these techniques use wires as the initial material, while others employ metallic powders. For example, Electron Beam Powder Bed Fusion (EB-PBF), often referred to as Electron Beam Melting (EBM), and Laser Powder Bed Fusion (L-PBF), commonly known as Selective Laser Melting (SLM), have emerged in this regard [34,35,36].

The L-PBF system has medium productivity, good repeatability and medium to high surface quality [37,38,39]. It utilizes high-energy laser sources, in order to be able to melt metallic powders such as Ti64 used in this study [40, 41]. However, it is often reported that due to the steep temperature gradients and the subsequent higher cooling rates during the process, internal stresses are developed [42,43,44]. Some research has suggested that the only way (impossible way) to negate all the residual stresses is to maintain the temperature inside the printing chamber closer to the melting point of the powder during irradiation and hence preventing the high thermal gradients from developing during the solidification of melt pool [44,45,46,47]. However, it is extremely hard to maintain this much of a higher temperature for an alloy like Ti64 with a melting point of over 1605 °C [48]. Moreover, in L-PBF, the choice of waveform can significantly impact the final product [49,50,51,52,53,54]. Most commercially available L-PBF machines utilize either continuous waveform (Continuous Waveform Laser Powder Bed Fusion—CW L-PBF) or pulsed waveform (Pulsed Waveform Laser Powder Bed Fusion—PW L-PBF) [53,54,55,56]. CW L-PBF employs a constant, uninterrupted laser beam throughout the melting process, ensuring a consistent thermal energy supply but potentially leading to part distortion, higher residual stresses, and defects [53,54,55,56]. In contrast, PW L-PBF delivers the laser beam in discrete pulses, allowing precise control over energy input. However, it can result in non-uniform melt pool conditions and rougher surface quality if process parameters are not optimized [53,54,55,56].

Recent research has focused on employing L-PBF to manufacture Ti64 implants, with the goal of reducing stress shielding effects and enhancing implant fixation [57,58,59,60,61,62,63,64,65,66,67,68,69]. Stress shielding involves a reduction in mechanical stress on the surrounding bone due to the implant’s stiffness, potentially leading to bone atrophy or weakening [58,59,60,61,62,63,64,65,66,67,68,69]. The ability to design and print intricate geometries with L-PBF is instrumental in creating patient-specific implants, addressing these issues and encouraging the exploration of functionally graded lattice materials. However, it can widely be observed that the mechanical properties from simulations and actual tests have a wide difference. This emphasizes the importance of understanding the complex physics involved during the solidification of the metal powders during L-PBF [58, 60, 63].

To close this wide difference between the simulations and the actual prints, it is important to closely study the thin structural components called ‘struts’ that make up these lattice structures. Based upon the specific geometry of the lattice structure, these strut’s length and diameter are varied. This aids them to meet at certain points called nodes. These changes can change the overall geometry and characteristics of the lattice. This article has studied struts manufactured by both CW L-PBF and PW L-PBF and aims to systematically address the following issues:

  1. 1.

    Feasibility study The minimum manufacturable limits for circular struts were determined to be 0.2–0.3 mm diameter and 10°–20° inclination including the current authors [70,71,72,73]. However, the variation in the feasibility limits with respect to the laser irradiation mode and the reason for failure below the generalized limits has not been explored and the possibility of changing laser irradiation mode and its subsequent effect has not been explored.

  2. 2.

    Effect on size and shape Though it has been previously reported that the number of defects imparted in printed parts increases with reduction in dimensions [72,73,74,75,76], a proper detailed investigation has not been provided. Moreover, with reduction in the angle of inclination, the circular cross sections are reported to elongate into ellipsoids [72,73,74,75,76]. However, the effect of the laser waveform over these factors has not been compared.

  3. 3.

    Roughness The major reason stated in literature for using different laser irradiation modes is to optimize the surface quality. Though this aspect has been explored on flat surface for bulk samples and for single scan layers [73,74,75, 77,78,79,80,81], it has not been explored for a structurally miniature component like the struts.

  4. 4.

    Defects: Though some of the common L-PBF defects are explored extensively [82,83,84,85,86,87,88], there is no study that contrasts the defects that these thin sections might incur under different laser irradiation manufacturing modes.

2 Experimental procedure

2.1 Design of experiments

The struts samples were not directly printed on to the base plate of the L-PBF machines to avoid difficulty of removing them from the base plate. A layout in Fig. 1a depicts the additional plate of dimensions 43 mm × 55 mm (in gray) on which these miniature strut elements were printed. This additional plate which will be referred to as ‘strut plate’ was designed to be 3 mm thick. During the L-PBF manufacturing, the supports were first deposited on the machine’s base plate, followed by the strut plate and then the struts—the components needed for analysis. These struts were arranged in the order of increasing angle of inclination (left to right in the design layout—Fig. 1a) from 10° to 90° with an incremental interval of 10 and in the order of increasing diameter from 0.1 to 1 mm with an incremental interval of 0.1 mm (top to bottom in the design layout, Fig. 1a). Each of these struts was designed to a constant length of 10 mm. These geometric parameters were selected based upon the common dimensions that were followed to manufacture lattice-based implant biomaterials. The layout depicted in the Fig. 1a was designed as a.stl file by AutoDesk Netfabb® [89] and Materialise Minimagics® [90]. It has to be noted that in previous studies on strut units, the angle of inclination is either along the XZ or YZ planes [70, 71, 75, 85, 91]. A summary of these studies is given in Table 1.

Fig. 1
figure 1

Layout of the design a strut sample design arrangement on the strut plate and b arrangement of struts

Full size image
Table 1 Summary of studies on struts
Full size table

However, this study has employed a distinct approach. The design of these struts was based on the BCC lattices as these are the lattices that will be compared further ahead as a continuation of this study. To have the same type of inclination as in BCC lattices, all the struts have a consistent inclination angle of 45° in both X and Y directions. The previously mentioned variation in the angle of inclination is applied only on Z direction. This deliberate variation is crucial because diverse BCC lattice configurations and other strut and node-based configurations may require different Z-axis inclination angles. During the design process, care has been taken to have sufficient distance between the struts due to the complex way of inclination of these struts (Fig. 1b). Moreover, each row of particular diameter is offset by 2.5 mm to ensure that no overlap with previous row happens during the printing process.

2.2 L-PBF equipment and powder details

The samples were manufactured in two different machines with distinct laser irradiation type:

  • Continuous Wave L-PBF (CW L-PBF), the widely used L-PBF machine with a continuous source of laser irradiation.

  • Pulsed Wave L-PBF (PW L-PBF), that use discrete pulses of laser irradiation in various waveforms.

For the CW L-PBF, a 3D systems Pro-X 200 DMP with a maximum rated laser power of 300W at the Institute of Photonics and Advanced Studies (IPAS) at the University of Adelaide was used. This machine uses a continuously irradiating laser with a range of power up to 300 W, wavelength 1070 nm and a Focal Offset Distance (FOD) of 2 mm, producing an incident spot size of 70 µm. The CW L-PBF machine utilizes pre-alloyed powder from TLS Technik GmbH & Co with a median size of 30 µm. The CW L-PBF machine uses a bi-directional (zigzag) scan strategy with a 90° rotation after completing each layer (Fig. 2). The hatch angle rotations are implemented as they are identified to be effective in reducing the anisotropic mechanical behavior of L-PBF parts due to directional heat extraction [101].

Fig. 2
figure 2

Scan patterns used in this study. a Zigzag scanning CW-LPBF and b meander scanning with contour scanning in PW L-PBF

Full size image

The energy input of the CW mode L-PBF can be calculated using the formula [50, 55, 102],

$$\text{Energy density }\left({E}_{\text{CW}}\right)= \frac{P}{h\times {v}_{\text{CW}}\times z},$$
(1)

where P is the power (W), \({v}_{\text{CW}}\) is the scan speed (mm/s), z is the layer thickness (mm). The energy density for the process parameters given in Table 2 is calculated to be 58.82 J/mm3.

Table 2 Process parameters of L-PBF used in this study
Full size table

For the PW L-PBF, a Renishaw AM400 with a maximum rated laser power of 400 W was used. This machine uses an intermittently irradiating laser of power up to 400 W, wavelength 1070 nm and produces an incident spot size of 70 µm. The PW L-PBF machine utilizes pre-alloyed powder from Tekna Pty. Ltd., Canada with a median size of 45 µm. The PW L-PBF machine uses a bi-directional (meander) scan strategy with an additional contour scan. A 67° rotation of the scanning is applied after completing each layer (Fig. 2). The energy input of the PW mode LPBF is calculated as 84.75 J/mm3 using the process parameters listed in Table 2 and the formula [50, 55, 102] below,

$$\text{Energy density }\left({E}_{\text{PW}}\right)= \frac{{P}_{\text{peak}}\times\updelta }{h\times {v}_{(\text{PW})}\times z},$$
(2)

wherein \({P}_{\text{peak}}\) is the peak power (W) for a laser signal with a fixed pulse repetition rate (PRR) and duty cycle (\(\delta )\), for a pulsed wave with a definitive pulse duration (\({t}_{\text{exp}}\)) and point distance (\(x).\) The duty cycle is defined as a fraction of total laser movement time for printing and is calculated from the exposure time (\({t}_{\text{exp}}\)) and the idle time (\({t}_{\text{idle}}\)) as [55],

$$\delta = \frac{{t}_{\text{exp}}}{{t}_{\text{exp}}+{t}_{\text{idle}}}.$$
(3)

Moreover, it is important to note that the process parameters listed in Table 2 are the recommended process parameters set by the manufacturers—3D systems®Footnote 1 and Renishaw®Footnote 2 and no alterations has been done to them. Preliminary work explored the process parameters optimization route and established the recommended manufacturer’s parameters as the best optimized parameter set [103, 104]. Additionally, precautions were taken to keep the Oxygen in the chamber below 500 ppm to avoid unnecessary oxidation during the printing.

The size and the distribution of the alloy powder were assessed using a laser particle size analyser, Mastersizer 2000 (Fig. 3). To confirm the powders had the expected spherical morphology, a field emission scanning electron microscope (FEG-SEM), namely the FEI Quanta 450, was employed. The alloy’s composition was determined through Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES), while the presence of gaseous impurities, such as oxygen, nitrogen, and hydrogen, was analyzed using the LECO ONH836 elemental analyser. The carbon impurity content was verified using the LECO CS200, a carbon and sulfur analyser designed for metals. The primary alloying elements, namely titanium (Ti), aluminum (Al), and vanadium (V), as well as the levels of impurities, were found to fall within the acceptable range according to ASTM F2924-14 standards [105] (Table 3).

Fig. 3
figure 3

Powder size distribution and powder morphology

Full size image
Table 3 Chemical composition of the powders used
Full size table

2.3 Analysis techniques

To achieve the objectives set for this study, different metrological and microscopic techniques were employed. The samples manufactured were initially observed visually and some stereomicroscopic observations were done to analyze the struts that are built successfully and also to observe the struts that had failed to print, especially the ones at the extremities of angle of inclination and lower diameters. An approximate measure of the length of the struts and the angle of inclination of the struts were measured using these techniques. However, to ensure the accuracy of these measurements, and to view the surface morphology of the printed struts, the struts were examined by a FEI Quanta 450 FEG-SEM.

Once the geometric parameters, such as the angle of inclination of the struts and the length of the struts, were measured, these struts were sectioned off the strut plate for further analysis. They were carefully sectioned off using a hand-held rotary cutter. A low-frequency ultrasonic bath cleaning was done to remove any cutting residue off the surfaces before starting surface roughness measurements. The surface roughness measurements were done using an optical profilometer, Olympus LEXT OLS5000. This device employed a 405nm wavelength laser beam as its incident source, enabling it to detect profile changes and generate height maps from which roughness parameters of Ra and Rz were calculated. Due to the curved surface of the struts, the Areal roughness parameters are not considered for analysis as these might increase the error percentage and only profile roughness parameters, Ra and Rz are considered.

In order to prepare sectioned samples for further analysis including metallographic examination, the samples were vacuum cold mounted in epoxy resin. The samples were positioned upright, disregarding their initial angles of inclination, to ensure optimal transverse section viewing as outlined in Fig. 4. These epoxy-mounted samples then underwent conventional grinding and polishing procedures, using the Struers Tegramin-25 machine. The grinding process started from a plane grinding (MD Mezzo—equivalent to 220 grit) and continued until a 1 μm diamond slurry polishing.

Fig. 4
figure 4

Sample mounting for measurement of circularity and for microstructural analysis

Full size image

The freshly polished unetched samples were examined under the SEM, FEI Quanta 450 FEG-SEM at lower magnification to assess the circularity and porosity content. The porosity content was measured to estimate the relative density of the printed samples, as the area fraction of pores. The circularity was measured to ascertain the process’s accuracy in printing finer details with greater definition. The hardness of the mounted samples was assessed using a Vickers Micro-Hardness Tester, LECO LM-700AT, with 50 g load and 10-s dwell time. To ensure accuracy, five indentations were made on each strut, and the average value was calculated.

The samples were repolished, and an additional step of a final 0.04 μm silica suspension polish (comprising 70% OP-S mixed with H2O2) was done. These samples were etched with Kroll’s reagent, which is composed of 3% HF, 5% HNO3, and 92% distilled water, for a duration of 50 s. Preliminary metallographic observations were done using the Zeiss Axio Imager 2 Light microscope to ensure the quality of the etch and also to select areas that can be analyzed in detail using an SEM. The FEI Quanta 450 FEG-SEM was used to observe the microstructural topology that has been revealed by the etching. For quantitative metallography, ImageJ software was employed, enabling the measurement of porosity content (relative density) and circularity. A thin layer of carbon coating was deposited onto the metallographic specimen to make the surface conductive and the Oxford CMOS Electron Back Scattered Diffraction (EBSD) system was utilized to examine the morphological features of the resultant phases.

3 Results and discussion

3.1 Feasibility study

To establish the limitations of making thin circular struts using different laser modes, identical strut plates with struts ranging in diameter from 0.1 to 1 mm and angles from 10° to 90° were manufactured from both the machines, Fig. 5. The struts were built on a 3 mm thick Ti64 plate with surface roughness (Ra) of around 30 µm for CW L-PBF and 23 µm for PW L-PBF, Fig. 6. A matrix to compare the success of the builds to a predetermined length of 10 mm is produced, Fig. 6. When compared, it is inferred that the strut plate printed by CW L-PBF produced finer struts compared to the PW L-PBF, Fig. 6. Specifically, the CW L-PBF machine couldn’t create struts with a diameter less than 0.2 mm, leaving no visible trace. Similarly, the PW L-PBF machine couldn’t produce struts thinner than 0.3 mm.

Fig. 5
figure 5

Printed struts on the strut plate. a CW L-PBF and b PW L-PBF

Full size image
Fig. 6
figure 6

Surface profile of the strut plate. a CW L-PBF and b PW L-PBF

Full size image

The absence of 0.1 mm struts is believed to be due to weak bonding between the struts and substrate (strut plate). Both the laser modes irradiated a very small length with an approximate travel distance of 30 µm continuously, to deposit the first layer and in case of the PW L-PBF, there could have been a maximum of only 2 pulses/points of irradiation for the entire length. The effect of roughness, consequently the wettability [106], combined with very small molten pool size (the very small contact area) with high surface tension, may result in poor wettability and thus weak bond between the solidified melt pool and the pre-deposited layer/strut plate, making it susceptible to detachment when the recoater blade distributes the next layer of powder [51, 107,108,109]. Due to this weak bonding between layers/strut plate, and subsequent detachment, smaller struts of 0.1 mm diameter could not be built.

With a diameter increase to 0.2 mm, CW L-PBF benefits from its trailing molten pool, which offers more melt pool area, better wettability and stronger layer-to-layer bonding [51, 107,108,109]. However, in case of PW L-PBF, the laser irradiation is intermittent resulting in localized melting at the vicinity of the laser irradiation and due to the pulsed nature, the laser irradiation stops after the required exposure time resulting in less extended melt pool. The resultant area of molten pool region produced by the pulses per layer is still insufficient to generate adequate bond between the layers. In particular, due to this weaker layer-to-layer bonding coupled with the motion of the recoater blade against the strut, inclination direction may have caused some struts at lower inclinations to fail (Fig. 7).

Fig. 7
figure 7

Printability of length (the values in red, yellow and green boxes refers to the length printed during the process—design length is 10 mm)

Full size image

To assess manufacturability concerning the precise replication of designed angles of inclination, a matrix based on the measured angle of inclination and diameter of the strut, as depicted in Fig. 8, was established. Despite the fact that PW L-PBF induces higher residual stresses due to intermittent laser exposure [42, 44, 110], lack of a continuous heat source to temper solidified spot in contrast to continuous laser mode, it unexpectedly demonstrated superior performance in maintaining the designed angle of inclination compared to CW L-PBF. Both the laser modes were able to produce thin struts at 90° to a greater accuracy. It is important to note that at this inclination, the scanned area for each layer is equivalent to the circular cross-sectional area, produced by the designed diameter. However, decreasing the inclination angle creates a comparable elliptical cross-sectional area, leading to a greater scan length along a particular direction. Despite this, CW L-PBF failed to reproduce the designed angles especially at lower inclinations, even for thicker struts. This may be attributed to the fraction of supported area for subsequent depositing layers and the continuous nature of laser exposure experienced by the struts. The concept of supported region and the staircase effect will be discussed in detail under the Sect. 3.2. Interestingly, higher percentage of lower diameter CW L-PBF struts was manufactured to their designed length despite a higher variation in the inclination angles. Similarly, in PW L-PBF, at 90° inclination, the scanned area matches the designed circular cross-sectional area. However, a decrease in the inclination angle from 90° creates a corresponding elliptical cross-sectional area, leading to more points of irradiation along one direction and improved manufacturability at these directions. Moreover, it is commonly expected that the PW L-PBF struts will suffer more with residual stresses due to the steeper thermal gradients [42,43,44,45, 111,112,113,114]. However, due to the implementation of the 67° alternating scanning strategy, the effect of residual stresses is minimized as the continuous rotation helps to distribute heat more evenly and prevent the build-up of aligned stress fields, resulting in a more uniform and isotropic stress distribution [115, 116]. In addition, at lower angles (< 30°), the contour scan remelts the previously deposited region along the contour of the design, resulting in melt flowing toward the direction of the gravity and hence resulting in the anchoring of the deposited struts to the strut plate surface, aiding in some sort of support to enable these struts to maintain the designed angle of inclination and even the minimal manifestation of residual stresses is not noticed. However, as the designed angle of inclination increased, the effect of gravity reduces and deviations from intended angle of inclination were observed, resulting in comparatively poorer angular accuracy, especially between 40° and 60°.

Fig. 8
figure 8

Printability of angle of inclination (the values in red, yellow and green boxes refers to the angle of the printed struts—design angle of inclination is given in row 1)

Full size image

3.2 Surface morphology

The struts’ designed angle of inclination is achieved by offsetting the layer-by-layer deposition by a specific length, thereby implementing the desired inclination. In the case of a part built perpendicularly, there is no expectation of a staircase-like topography, Fig. 9a). However, when the struts are constructed at an inclined angle of less than 90°, the successive layers do not fully overlap, and the new layer lacks complete support from the previously deposited layer. Consequently, this offsetting results in a staircase-like appearance on the surface of the inclined strut. The offsetting length, referred to as overhang (when on the dorsal side of the strut toward the build direction) or step tread (when on the ventral side toward the build direction), can increase with a decrease in the angle of inclination from 90° to 10°, Fig. 9a). The staircase effect caused by the change in inclination angle contributes to definitive minimum roughness, dependent on the applied layer thickness.

Fig. 9
figure 9

a Schematic illustration of variation in step tread due to L-PBF process. b Geometric relations to calculate theoretical roughness parameters

Full size image

It is important to note that the thickness of the steps remains constant for all angles of inclination, and independent of the angle of inclination. This layer thickness, known as the step rise, corresponds to the pre-selected powder layer thickness, which is 30 µm for the struts manufactured in this study (fixed design parameter). Figure 9b suggests that a simple trigonometric relation can be established to calculate the step tread length for a known step rise (c) and angle of inclination (θ) in degrees and is given as

$$\text{tan}\left(90-\theta \right)= \frac{\text{Step tread }(b)}{\text{Step rise }(c)}.$$
(4)

Also, from the Fig. 9b), it can be inferred that the theoretical average roughness parameters Ra and Rz can be estimated by the relation

$$\text{Arithmetic mean roughness parameter Ra}=\frac{h}{2},$$
(5)
$$\text{Average peak}-\text{to}-\text{valley roughness parameter Rz}=\text{ step tread}.$$
(6)

The value of ‘h’ can be found using the Pythagoras theorem and Thales theorem (altitude over hypotenuse in right angle triangle) and can be presented as

Step 1: The hypotenuse ‘a’ can be found using the relation \(a=\sqrt{{b}^{2}+{c}^{2}}\)

Step 2: Using trigonometric relations,

$$h= \sqrt{m\times n}=\sqrt{\left({b}^{2}/a\right)\times \left({c}^{2}/a\right)}.$$
(7)

The parameters a, b, c, m, n and h are outlined in the Fig. 9b). The theoretical values of the profile roughness parameters Ra and Rz are outlined in the Fig. 10.

Fig. 10
figure 10

Expected trend—theoretical roughness parameters

Full size image

However, it is important to note that the thermal processing during L-PBF can introduce additional effects such as the surface texture generation due to melt pool flow, and surface adhesions (sintered particles—Fig. 11), and residual stresses due to rapid cooling that can affect the surface profile roughness of the printed parts [117,118,119,120] (Table 4).

Fig. 11
figure 11

Sintered particles in CW L-PBF and PW L-PBF. a Comparison of 1 mm strut surfaces. b Comparison of 0.6 mm strut surfaces

Full size image
Table 4 Calculated theoretical value of roughness parameters
Full size table

The CW L-PBF ensures the laser’s energy is distributed evenly across the targeted area, allowing for uniform melting of the powder. This consistent energy input prevents localized overheating, which typically leads to defects such as porosity, and cracks within the material [121,122,123]. Overheating can cause excessive melting beyond the intended melt pool boundaries, resulting in dimensional inaccuracies, geometrical errors, and the adhesion of partially melted, unirradiated powders [121,122,123]. However, it is important to note that in CW L-PBF, the continuous heat input and the stable temperature gradient allow the molten pool to cool more gradually as the laser moves away, promoting more orderly solidification [50, 54, 107, 123,124,125,126,127,128,129]. Additionally, the thermal mass affected by the continuous mode is generally larger, leading to slower cooling rates [50, 54, 123,124,125,126,127,128,129]. While this gradual cooling is beneficial for reducing residual stresses and preventing rapid microstructural changes that can introduce defects [42,43,44,45, 111,112,113,114], overheating beyond the intended boundaries can partially melt adjacent powder particles and fuse them to the surface of the strut intended by the design, Fig. 11, as also reported before [88, 130]. Owing to this aspect of overheating and the challenge of precisely controlling the energy distribution, continuous mode is often preferred for bulk sections, which benefit from its consistent energy delivery that promotes uniform melting and solidification—advantageous for the mechanical properties of larger components [53,54,55,56, 120].

In contrast, PW mode operates by emitting energy in short, intense bursts, resulting in quick temperature changes within the powder bed, leading to rapid melting and solidification. The intermittent nature of the laser reduces the overall heat accumulation, resulting in having less residual thermal energy to be utilized to melt the adjacent unirradiated powder between two successive pulses [50, 54, 123,124,125,126,127,128,129]. Hence, there is a smaller number of sintered particles expected to be found on the surface of PW mode printed struts, Fig. 11.

From a theoretical perspective (Figs. 9 and 10), it is clear that with any constant layer thickness or ‘step rise,’ a change in the strut’s angle of inclination results in a change in step tread, affecting the part’s roughness. As the angle of inclination decreases, the step tread increases, leading to higher values of the roughness parameter Rz. The roughness parameter Ra, representing average roughness, also varies with the inclination angle with an optimal value between 40 and 60. The measured values from Fig. 12 show that as the angle of inclination decreases, roughness generally increases due to the staircasing effect.

Fig. 12
figure 12

Roughness of struts—a Ra of 1 mm struts, b Rz of 1 mm struts, c Ra of 0.6 mm struts and d Rz of 0.6 mm struts

Full size image

The CW (Continuous Wave) and PW (Pulsed Wave) L-PBF processes exhibit different roughness outcomes due to their respective thermal cycles, directly related to the laser mode of irradiation, in confirmation of previous report [131]. Upon revisiting the process parameters from Table 1 and comparing it with existing literature, it is evident that during CW L-PBF, the continuous energy input creates a uniform melt pool and stable temperature gradient, facilitating uniform cooling. This can help in reducing residual stresses and defects [44,45,46,47], but can result in heat accumulation, resulting in the substantial adherence of partially melted adjacent particles on the struts’ surface (Fig. 11), and consequently resulting in higher roughness values (Fig. 12) [131].

The pulsed nature of the laser minimizes heat accumulation and the affected thermal area, which can in turn benefit smaller diameter struts but may also challenge interlayer bonding [50, 51, 131, 132]. This is evident when comparing 1 mm and 0.6 mm struts (Fig. 12). From Fig. 11, we observe fewer sintered particles adhering to the strut surface, primarily because there is less heat accumulation in the PW L-PBF process, leaving no energy to melt or partially melt the nearby powders despite an additional contour scan during the process. However, Fig. 12 shows that the standard deviation of PW L-PBF, especially at lower angles of inclination, is higher. This observation led to a study of the top and bottom surfaces of these thin sections, with the 0.6 mm strut selected for analysis. It can be inferred from the study that due to gravity and the low surface tension of the molten pool, resulting from the repetitive melting of the outer layer during the contour scan, the molten pool flows over the sides and results in tail-like formations at the strut’s bottom surface, in the direction of gravity (Fig. 13) [71, 75]. But as the angle of inclination increases, this effect stabilizes, and the strut assumes a near-circular shape at 90 degrees, resulting in uniform roughness around the strut’s circumferential region [71, 75].

Fig. 13
figure 13

Varying roughness profile along the surfaces of the strut—PW L-PBF 0.6 mm with varying angle of inclination

Full size image

3.3 Dimensional accuracy

3.3.1 Effect of laser irradiation mode on dimensional accuracy

CW irradiation and PW irradiation exhibit distinct laser-material interaction dynamics, influencing the dimensional accuracy of printed sections in LPBF processes. The schematic in Fig. 14 serves as a visual synthesis from literature and empirical observations, highlighting the contrasts in melt pool geometry and thermal mass concerning processing conditions [50, 51, 53,54,55,56, 132,133,134]. As previously stated, the CW L-PBF process is featured by a constant laser output that maintains a uniform and elongated melt pool. This uninterrupted energy delivery causes the melt pool to extend beyond the planned scan boundaries, leading to broader overlaps with adjacent scan lines, evident as oversizing in the final part dimensions [51, 107,108,109]. The generated thermal mass in CW-LPBF is considerable due to the continuous irradiation, maintaining a higher temperature around the melt pool. The excess heat is conducted beyond the target region, possibly causing over-melting of adjacent layers/scan lines and inadvertently resulting in the sintering of adjacent powders, which are later observed as unintended adhesions on the surfaces of the final prints [51, 107,108,109].

Fig. 14
figure 14

Schematic illustrating the differences with respect to CW L-PBF and PW L-PBF

Full size image

Additionally, the thermal input and the heated zone in CW-LPBF are notably larger (with respect to PW L-PBF) due to the sustained input of energy. This extra thermal energy is responsible for significant thermal stresses and can induce warping as the part cools down, a phenomenon that aggravates oversizing [51, 107,108,109]. A practical manifestation of this is seen in the strut thickness of 754 ± 15 µm, surpassing the design goal of 700 µm (0.7 mm) as demonstrated in Fig. 15.

Fig. 15
figure 15

Strut diameter measurements. a 0.7 mm, 10° and b 0.7 mm, 90°

Full size image

In contrast, the PW-LPBF approach engenders a series of discrete, transient melt pools. These melt pools are smaller and more defined, shaped by the laser’s pulsing regime [51, 107,108,109]. The pulse parameters are finely tuned to ensure that the molten regions overlap adequately, securing full fusion within the intended design perimeters. Despite this overlap, the pulsed laser’s intermittent energy input curtails the overall thermal mass, allowing for sufficient cooling between pulses and shrinking the heat-affected zone. Despite the pulsed heating strategy, the associated 67° scan rotation helps in reducing the residual stresses build-up [115, 116] and, thereby reducing the likelihood of warping or distortion and enabling the production of components that meet or are slightly below the intended dimensions. The precise melt pool size in PW-LPBF, closely matching the desired scan width, results in parts that adhere more closely to the intended dimensions. This is observed in struts with a thickness of 678 ± 8 μm, aligning with the design specification of 700 µm (0.7 mm), albeit slightly undersized, as noted in Fig. 15. This under-sizing is a consequence of the controlled melt pool geometry, the effective dissipation of thermal mass, and the strategic overlap of the melt pools without the excess heat that characterizes the CW process.

In short, the fundamental difference in the kinetics of the molten pool influences the part accuracy of these two processes. In CW-LPBF, the melt pool is more dynamic, with a tendency to remain molten for longer periods, which can lead to the mentioned oversizing. PW-LPBF, on the other hand, experiences faster solidification due to the intermittent nature of the energy supply, resulting in a melt pool that solidifies quickly, thereby reducing the chances of melt pool spread and resulting in the part dimensions not exceeding the intended size.

3.3.2 Effect on shape

As previously stated, challenges may arise in fabricating cylindrical struts with a circular cross section due to the step formation and variations in overhang/tread length with the angle of inclination. To assess the manufacturing capability of circular cross sections, it becomes crucial to determine the degree of circularity. To estimate this, the circularity of the struts was examined. Circularity (f2) often referred to as sphericity, quantifies how closely the shape of the cross section of the struts resembles a perfect circle and hence assessing the degree to which the strut’s shape deviates from being perfectly circular. According to the ASM Metals handbook, this parameter can be calculated by comparing ‘A’ (the area of the strut transverse cross section) to ‘L (the perimeter of the strut’s transverse cross section), and is given by the formula [135],

$${f}_{2}= \frac{4\pi A}{{L}^{2}}.$$
(8)

Precision is required in measuring the circularity of the struts, ensuring that the measured section represents a true cross section without any deviation from a 90-degree angle. To achieve this, the samples were positioned upright (Fig. 4) and polished.

The higher laser power combined with the continuous laser irradiation during CW-LPBF contributes to a more thorough melting of the powder layer and inducing some remelting or partial remelting of previous layers, promoting comprehensive layer-to-layer adhesion. This deep and uniform melting allows for a stable melt pool, which is less prone to the formation of defects such as porosity or balling [51, 65, 107,108,109, 121, 122, 128]. The consequence of such a stable thermal profile is a reduction in the surface irregularities that can degrade the circularity of the struts. However, the CW process is not without its challenges. The combination of higher laser power and continuous irradiation can lead to heat accumulation, with the potential for wider and deeper molten pools than intended. While this might suggest a risk for increased roughness, the CW process benefits from a slower cooling rate, allowing the molten material more time to flow and conform to the desired geometrical boundaries before solidification. The result is a more consistent and accurate replication of the designed cross-sectional shape, contributing to the higher roundness observed in CW L-PBF-manufactured struts (Fig. 16).

Fig. 16
figure 16

Circularity of 0.9 mm struts (top row—CW L-PBF struts from 10° to 90°, bottom row—PW L-PBF struts from 10° to 90°)

Full size image

In contrast, PW L-PBF’s intermittent laser operation introduces a different set of thermomechanical dynamics. The rapid on–off action of the laser leads to a series of quick heating and cooling cycles, producing distinct yet overlapping molten pools. Though the energy density is closer to that of CW L-PBF, the pulsed nature of the laser prevents excessive spreading of the melt pool, which could otherwise heat and inadvertently melt surrounding powder. This results in a series of solidified pools that are merged together. Moreover, this pulsed nature of the laser irradiation in PW L-PBF accelerates the cooling rate, leading to a faster solidification that may not afford sufficient time to remelt the previously deposited layers [44, 45, 71, 75, 110], hence resulting poorer surface finish. Though a contour scan in PW-LPBF aims to remelt the edges of the strut, improving circularity and definition, the gravitational overflow can counteract this effort, leading to the formation of tail-like structures as the molten metal flows over the edges and solidifies under the influence of gravity, especially in case of struts at lower angle of inclination (Fig. 16), also reported before [71, 75]. Hence, the overall roundness and shape retention of PW L-PBF struts vary with the angle of inclination. Higher angles benefit from the additional contour scan, which helps to define the edges and improve the structural outline. Yet, as the angle decreases, the gravity-induced overflow and the rapid solidification challenge the maintenance of the circular cross section. This is particularly evident in thin struts, where the overhang region is more susceptible to distortion due to lack of support from previous layer [71, 75].

3.4 Microstructural analysis

The general microstructure in all the samples is α′ martensite as a result of a diffusionless shear-type transformation process due to the rapid solidification from the liquid region featured by the high cooling rates [111, 122, 136]. The hierarchical formation of α′ martensite is attributed to the rapid reheating, and partial remelting of previously deposited layers coupled with high cooling cycles inherent in the L-PBF process, where the material undergoes rapid solidification and cooling rates [126, 137, 138]. This process results in the formation of acicular (needle-like) primary α′ martensite that extend across the entire parent β phase, finer secondary α′ martensite form, parallel or perpendicular to the primary α′ martensite, tertiary α′ martensite, even finer, and further elements of hierarchical α′ martensitic evolution [126, 137].

The primary difference between CW and PW L-PBF lies in their thermal input and energy distribution, which directly affects solidification kinetics [56]. CW L-PBF typically results in a more uniform thermal field and slower cooling rates, promoting coarser microstructures, Fig. 17a) in support of previous reports [108]. In contrast, PW L-PBF can achieve higher cooling rates due to intermittent energy input, and smaller volume of molten metal at a time leading to finer microstructures, Fig. 17b) [108]. Though the hierarchical formation of α′ martensite can be observed that in the samples of both these methods of L-PBF, signifying that they both undergo multiple thermal cycles, it is important to note that measurement of their lath widths provide a clearer insight on this effect of thermal cycles experienced by these samples.

Fig. 17
figure 17

a Microstructure of CW L-PBF manufactured 0.9 mm struts. b Microstructure of PW L-PBF manufactured 0.9 mm struts

Full size image

In CW-LPBF, the laser emits a continuous beam, providing a steady heat source. This results in more uniform and potentially slower cooling rates compared to PW-LPBF, as the material is continuously subjected to the laser's heat during the build process. Though the initial transformation is a rapid solidification, the extended exposure to heat, the heat accumulation due to the excess thermal energy being conducted away to the presolidified layers from the site of irradiation, results in formation of several hierarchy of martensite and potentially leading to a coarser microstructure [52, 125, 139, 140]. However, due to the low diffusion potential, only short-range diffusion of Vanadium atoms happens and this is the reason for small white dots in CW L-PBF samples [141]. Furthermore, it can be noticed qualitatively that the 0.9 mm 90° has wider α′ martensitic needles than 0.9 mm 10° as 90° samples have a perfect layer over layer deposition, whereas 10° has overhangs and hence less layer overlap regions and more overhang regions. Hence, these overhang regions do not experience the repetitive heating–cooling thermal cycling required to form the hierarchy of martensitic needles [139, 142].

The pulsed irradiation in PW-LPBF utilizes introduces rapid heating and cooling cycles to the material, leading to faster cooling rates, compared to CW L-PBF [52], hence resulting in finer α′ martensitic needles, confirming the predictions of microstructural modeling [52]. Though thermal cycling takes place due to the oncoming layer (the same way as in CW L-PBF), the intermittent pulsing ensures the effective use of all the thermal energy to melt the powder particles and no extra heat to assist diffusion to take place (no white Vanadium-rich particles are noticeable in the microstructure, Fig. 17b).

To further analyze the effect of the thermal cycling induced by the laser source, EBSD studies were conducted, Fig. 18. The lath width measurements reveal that the PW L-PBF contains many needles below the 1µm size range, whereas CW L-PBF samples have comparatively fewer needle widths in this range. The primary reason for this occurrence is the melt pool dimensions and cooling rates. CW L-PBF, with its slower cooling rate and larger melt pool size, allows for more significant heat accumulation within the part, influencing the overall thermal history and resulting in wider α′ martensitic structures. In contrast, PW L-PBF, characterized by smaller melt pools due to the intermittent nature of the laser exposure, leads to faster solidification rate that favors the formation of finer martensitic needles.

Fig. 18
figure 18

Comparison of martensitic needle width measurements using EBSD. a CW L-PBF, 0.9 mm strut at 10°, b PW L-PBF, 0.9 mm strut at 10° and c comparison of the martensitic needle widths

Full size image

The process parameters further highlight these differences. CW L-PBF’s higher laser power of 270 W creates a thermal environment conducive to larger heat-affected zones. This larger heat source facilitates the growth of α′ martensite with a broader width, correlating with the observed larger lath widths. On the other hand, PW L-PBF's parameters though it possess a lower power of 200 W and a lower scan speed of 750 mm/s, the pulsed nature of the laser generates less residual heat input per unit area and higher cooling rates, which are instrumental in the development of a finer martensitic structure.

3.5 Defects

Several defects were identified in both the CW and the PW L-PBF manufactured struts. Lack of Fusion (LoF) defects, which generally occur due to insufficient energy to melt the powders, is not noticed in either of the samples confirming the efficacy of the optimized parameters. However, in case of struts of lower diameter made by CW L-PBF, the combination of overhangs, unsupported sections, and particle adhesions can lead to a distinct form of porosity, similar to LoF defects, known as flow gaps [131, 143], Fig. 19a). Gravity causes the unsupported melt pool to flow down strut sides, but the steep temperature gradients and subsequent increase in surface tension prevent complete fusion over the previously solidified part, resulting in irregular gaps or porosity. No such feature was observed in PW L-PBF due to the contour scan in the scanning strategy that remelts the external surfaces and renders a smoother surface finish, despite resulting in elongation of the strut dimension along the direction of gravity, Fig. 19c).

Fig. 19
figure 19

Common defects in struts a defects in CW L-PBF, 0.5 mm strut at 10° b defects in CW L-PBF, 0.5 mm strut at 90° c defects in PW L-PBF, 0. 5 mm strut at 10° d defects in PW L-PBF, 0.5 mm strut at 90°

Full size image

The rapid heating and cooling during L-PBF process can cause a vapor depression called keyhole which forms due to recoil pressure. This vapor depression can become unstable, and one potential cause of this instability is a change in the energy incident on the powder, which is prevalent in PW mode due to the pulsed nature of the laser source. This instability may lead to the trapping of vapor within the depressed region, known as keyhole entrapment, Fig. 19.

Besides these defects, the occurrence of the balling phenomenon is more pronounced in PW L-PBF than CW L-PBF. In CW L-PBF, they are noticeable only in struts manufactured at lower angles of inclination and more prevalent near the periphery of the struts. However, in PW L-PBF, the balling effect is noticeable equally regardless of the angle of inclination (Fig. 19a, b, d). The end of the laser irradiation in case of CW L-PBF and the intermittent laser irradiation cycles in PW L-PBF can lead to the ejection of molten pool material due to the sudden release of vapor trapped within the powder bed. Consequently, this ejected molten pool material can splatter from the melting site to the surrounding region. This splattering often solidifies and settles as a globular particle and appears as in unlike CW L-PBF, where balling can impact the final strut’s surface roughness due to its presence near to periphery. In PW L-PBF, this issue is mitigated as the contour scan remelts and solidifies these types of defects along the edges of the design. Some residual stress-induced cracks (Fig. 19d) and gas entrapment porosities were also noticed. The gas entrapment porosities are very small, with an average size of less than 3µm and are noticed only at higher magnifications. A detailed account of defects in thin struts will be published as a separate work.

4 Conclusions

This study has examined the feasibility of manufacturing struts using both CW L-PBF and PW L-PBF methods. This study has also focused on comparing the theoretical estimations versus experimental measurements of roughness, effect of laser source on the size and shape of these thin struts at various angles of inclination, and microstructural features of both CW L-PBF and PW L-PBF samples. The results of this work can be summarized as follows:

  • Printability of struts This study demonstrates that both CW L-PBF and PW L-PBF are capable of producing thin cylindrical struts, with CW L-PBF able to produce smaller diameter struts and PW L-PBF showing better retention of the designed angles of inclination as its pulsed nature allows for better control over heat input.

  • Roughness measurement, theoretical vs practical It was found that theoretical estimations of surface roughness parameters were optimistic when compared to experimental measurements. PW L-PBF, in particular, showed a closer alignment between theory and practice, attributed to its intermittent energy input reducing thermal stresses and distortions.

    1. o

      Discrepancies in theoretical and experimental roughness parameters The discrepancies were largely attributed to the dynamic nature of the melt pool in CW L-PBF and the presence of partially sintered particles in both processes. These factors were not fully accounted for in the developed theoretical models, leading to deviations in expected versus observed roughness. It is also worthwhile to note that CW L-PBF accumulated enormous number of sintered particles due to the heat accumulation as a virtue of the continuous laser input.

  • Effect of laser source on strut size and shape The size and the shape of the struts were significantly influenced by the chosen laser source. Due to the nature of the heat source and the subsequent molten pool characteristics, CW L-PBF produced slightly oversized struts and PW L-PBF produced slightly undersized struts. Moreover, CW L-PBF tended to produce struts with more consistent cross sections despite the angle of inclination, whereas though PW L-PBF exhibited better shape retention at higher angle of inclination (> 60°), at lower inclination angles, a greater tendency toward irregularities and variance in strut dimensions is noticed.

  • Microstructural examination Distinct differences in the lath width and overall microstructure between the two methods were identified. In this case, the pulsed nature of PW L-PBF resulted in finer microstructures and narrower lath widths, attributed to comparatively faster cooling rates and reduced heat accumulation. In contrast, CW L-PBF leads to the coarser microstructures and slightly wider lath widths, attributed to the comparatively slower cooling rate and more heat accumulation due to the continuous heat input.