Characterization of 3D metal printed cutting tool with transpiration cooling channels

This study characterizes the 3D-printed transpiration cooling (TC) channels in a single-point cutting tool to enhance lubrication and cooling at the cutting zones. Five different TC channels, namely two circular profiled channels (Designs 1-A & 1-B), two hexagonal profiled channel (Designs 2-A & 2-B), and one bio-inspired blood vessel (Designs 3), are designed inside a single-point turning tool and 3D printed using AISI-1.2709 in powder bed fusion (PBF). From the materials and mechanical characterisation, fine cellular microstructure and high hardness are achieved. X-ray microcomputed tomography (XRμCT) has been used as a non-destructive inspection strategy to analyse the built structures. The results of XRμCT showed that the TC channel built is highly orientation-dependent, steeper angles deviate highly, and nominal angles such as 0° and 90° (to the build platform) provide the best dimensional accuracy. The average dimensional deviations of the five designs are −35.8%, −19.42%, −19.45%, −15.85%, and −5.02%, respectively, from the as-designed. The best designs are circular free-form (Design 1-B), hexagonal free-form (Design 2-B), and bio-inspired blood vessel (Design 3), which have the least dimensional deviation and highest accuracy.


Introduction
Turning hard-to-cut materials with better cooling and lubrication enhances the overall machinability.In Machining hard-to-cut materials, the existing conventional cooling and lubrication methods are insufficient as it becomes difficult for the cutting fluid to reach the cutting zone, leading to increased tool wear and failures [1,2].AM allows for the creation of cutting tools and holders with internal and external fluid-flow channels to enhance this need.AISI-1.2709material is often used for this purpose due to its classification as tool steel.Advancements in AM have made it possible to 3D print these tools and holders for various machining operations, providing a more efficient cooling and lubricating system for conventional machining processes.Using this material, an indexable milling tool holder is 3D printed [3], the AM tool with an L-shaped nozzle performed better than the conventional tool with a round nozzle of diameter 2.5 mm.The AM tool's focused cutting-fluid channels on the rake face resulted in reduced tool wear, cutting forces, and vibrations.They also experimented with optimising the shape of the nozzle through CFD analysis [4].New nozzle designs with different channel designs (sharp edge, curved, and helical) and angles (0°and 30°) are tested on milling cutters.The curved design showed a 23% increase in flow rate compared to the sharp-edged design, and the helical design increased it by 9%.Unconventional designs for internal geometries can be achieved easily.In an experiment [5], an AISI-1.2709end miller cutter with four-flute design is 3D printed using LS technique in PBF method and characterised for dimensional accuracies and surface quality.Results showed deviation in the range of 0.25 mm, highlighting AM's ability to produce monolithic metal cutting tools.
It is realised from the past research is that internal cooling methodologies provides better machining results.In the present work a new internal cooling methodology called Transpiration cooling (TC) is designed, printed and characterized.Many researchers studied the performance of TC effectiveness for cooling high heat-flux surfaces with optimal injection ratio and cooling area and reported TC to be effective for efficient cooling and lubrication applications [6][7][8][9][10].Investigations on TC on a wedge-shaped nose are performed by fabricating a cone with unequal thickness and porosity.The wind tunnel testing at 1600 K reveals higher cooling effectiveness at a specific optimal injection ratio, and the gravitational force causes more cooling at the bottom than the top [6].Researchers [11] developed a self-pumping TC using a porous bronze plate and investigated.The porous plate acts as the TC structure and automatically pulls water from a connected tank via capillary force.They also experimented using tree-like micro channels for the same self-pumping TC [9].A tree-like channel made via Laser Sintering (LS) technique using PBF method studied the effect of angle (0°, 45°, 90°, 135°, and 180°) on TC.No significant change in cooling effectiveness is observed, indicating TC can be used for varying angle requirements, such as spacecraft leading edges.In an experiment [8], a TC test coupon is made via LS technique using PBF method, with five designs (perforated plates, spheres, wire meshing, nerves) studied.Results show TC has higher cooling potential than film cooling, with peak effectiveness at array centre and complex structures can be produced via AM methods.In the study [10], a multiple-hole array of 40 μm diameter is created in a C3X turbine blade through AM.TC is found to have 34% and 25% higher cooling effectiveness than effusion and internal cooling, respectively.This technique is being implemented in areas of higher significance and applications such as military and civil aviation.Researches reveal that TC method enhances the cooling and lubrication process for various problems of interest in metallic samples.This work adopted the advantages TC and incorporated in the single point cutting tool.To realise the purpose the complex TC geometries are designed printed and characterized.The AM methods provide greater flexibility to design complex geometries and consequently the dimensional deviations has to be examined for better cooling and lubrication.
Here, this work examines the printed geometries and suggest suitable orientation for cooling through TC method.Traditional methods like optical microscopy, CMM, and 3D scanning are used to verify dimensions of the 3D printed component [12][13][14][15].The surface profilometers (contact and non-contact) measure external features and the conventional methods have limitations, unable to measure interior features with higher accuracy.XRμCT which used x-rays for 3D scanning the as built part is now a widely-used technique for inspecting the AM parts, providing precise dimensional measurements.With AM, which enables complex shapes, internal features, lattice structures, and foams can be easily 3D printed and XRμCT helps study porosity, cell morphology, surface geometry, roughness, and detecting internal defects [16,17] of the as built part.In experiments performed [18,19], the effect of build direction on internal micro-channel geometrical tolerance and surface roughness is studied.The impact of surface roughness for coolant flow and heat transfer of AM coupons [20] is examined by performing an XRμCT scan on the AM test coupon of Inconel-718 using LS technique of the PBF method.Channels built vertically in AM showed the best quality regarding dimensional accuracy and surface roughness.Dimensional deviations are minimal when build angle varied from 0°to 45°.To improve accuracy, a tear-drop shape is suggested to compensate for circular microchannel variations.The results indicate that the microchannel angle should not exceed 45°for as-designed part to be built with highest accuracy.The test artefacts [21][22][23] made via LS technique in PBF method using 17-4PH stainless steel had internal holes at orientations from 0°−90°.XRμCT scanning revealed 90°channels had the least deviation from the design, while deviation increased as build orientation angle decreased.XRμCT scans are conducted on AISI-1.2709conformal cooling channels [24] produced through LS technique which melted the material on the build surface.Results revealed XRμCT's usefulness in measuring freeform-shaped internal micro channels.XRμCT scans are used to study the dimensional deviations of lattice structures made of PBF-LB/M/Ti 6 Al 4 V [25] and the results reveals the actual part quality, confirming XRμCT direct application for dimensional evaluation.This study focuses on producing a single-point turning tool with novel TC channels, 3D printed in AISI 1.2709 using LS technique of the PBF method and characterised for materials and mechanical attributes and dimensional accuracies by XRμCT scanning.

Experimental methodology 2.1. Experimental design
The monolithic single-point turning tool is designed according to the American Standards Association (ASA) system.The designated tool nomenclature is 10, 15, 7, 7, 15, 7, 0.5 mm.The size of the tool is 3/8 th of an inch square tool, and the total length of the tool is 50 mm.
From figure 1, in Design 1, the TC channels are circular in profile and contain a storage tank-like chamber of 5 mm × 5 mm × 4 mm.This chamber also has the overall fluid inlet from the back end of the shank connected with it.In Design 2, the TC channels are hexagonal in profile and contain a chamber of the exact dimensions mentioned in Design 1. Finally, in Design 3, novel bio-inspired blood vessel (BV) type TC channels are designed.
The 500 μm diameter circular profile channels in Design 1-A travel through an angled loft-type trajectory from the top of the chamber to the rake face land.The 500 μm diameter circular profile in Design 1-B is a freeform loft-type path extending from the chamber's front face to the rake face land.The Design 2-A's hexagonal profile follows an angled loft-type direction from the top of the chamber to the rake face land, with the hexagon inscribed within a circle of 500 μm diameter.The hexagonal profile in Design 2-B is an inscribed hexagon within a circle of 500 μm diameter, and it follows a free-form loft-type path from the front face of the chamber till the rake face land.Nine TC channels are set up in a 3 × 3 array with a 1500 μm spacing between them.This array is positioned 500 μm from the tool's cutting edge.The size of the holes and the spacing between them are decided from pilot experiments and previous research [8] and EOS material data sheet [26].
The specifications of the novel bio-inspired BV type TC channels are the bottom most trunk of the BV is circular in profile with a of 3500 μm diameter.The bottommost trunk splits into four intermediate branches, circular in profile, with a of 1500 μm diameter each.Then, each of the four intermediate branches splits into four more branches circular in profile with a 500 μm diameter each, making up to 16 branches at the topmost section of the BV.

Part manufacturing plan
The feedstock used for the manufacture of the single point cutting tool is gas atomised spherical AISI 1.2709, with the commercial name of Maraging Steel MS1.The morphology and chemical composition of the feedstock is confirmed through a field emission scanning electron microscope FESEM-EDS (Energy Dispersive Spectrometer) analysis.The FESEM image of the feedstock is shown in figure 2 below.The average chemical compositions of the feedstock are mentioned in table 1. and it matches with the datasheet provided from the feedstock supplier [26].
From the FESEM analysis, the particle size distribution is measured.The average particle size of 35 μm and particle size distribution D10-18 μm, D50-35.69μm and D90-61.41μm is obtained.The part is printed in EOS M290 (EOS GmbH, Germany) printer, which uses a 400 W Yb-fiber laser source with a 100 μm focus diameter to selectively sinter the powder particles during the build cycle.The parts are printed using optimised process parameters after several experiments conducted for printing first article, are mentioned below in the table 2.
The layer thickness is maintained at 40 μm and a hatch spacing of 100 μm is fixed throughout.The first three layers of the print is performed using parameters specified as Down skin in the table.The core layers of the part is printed using the parameters specified as Core.Contour parameters are used to form the shape and surface roughness of the overall part.A 67°angle shift in hatching is used.A protective atmosphere of Nitrogen gas is used for preventing the oxidation in the build chamber.The build platform is also made up of the same AISI 1.2709 and is a part cake at 200 °C throughout the experiment.

Materials characterisation
Sample preparation for optical microscopy to study the microstructure of the 3D printed sample is done by grinding the sample followed by polishing the surface using silica solution.The polished samples are then etched with 2-Nital solution (2 mL HNO 3 + 98 ml C 2 H 6 O).Optical microscope (Make: Carl Zeiss; Model: Axio Lab A1) is used for analysing the microstructure and a field emission scanning electron microscope (FESEM) (Make: Thermo Fisher; Model: FEI QUANTA 250 FEG) is performed along with energy dispersive x-ray spectrometer (EDS) for studying the chemical compositions.To identify the phases of the as printed part, x-ray diffractometer (XRD) (Make: Bruker; Model: D8 Advanced) with Cu anode 2.2 KW operated at 40 mA, 35 kV, scanning step of m PBF water is the weight of the as printed sample in water, r water is the water density.The relative density measured in this analytical method is also cross verified by XRμCT scan by characterising the porosity of the part.The as printed tool is examined for surface roughness measurement and analysed to understand the effect of the optimised process parameters on the final quality of the part.This experiment is performed using 3D digital microscope (Make: Keyence; Model VHX-7000N).The surface roughness is measured and analysed at various location on the rake face of the tool, using 3D images.

Mechanical characterisation
The hardness of the as built part is measured to understand the mechanical characteristic of the tool.An automated microhardness testing machine (Make: Mitutoyo; Model: HM-200) is used at 500gf load with dwell time of 10 s as specified in ASTM E384.The microhardness is measured along and across the build platform (perpendicular to z axis).Numerous measurements at a distance of 0.1 mm is taken for this study.

X-ray micro computed tomography
X-ray micro computed tomography (XRμCT) scans, an advanced 3D scanning method is performed for each cutting tool separately using identical parameters on SKYSCAN 2214 (Bruker) scanner.The images are obtained  using LaB 6 source x-ray tube operated at 140 kV and 68 μA with 1 mm Aluminium foil for reducing beam hardening.The samples are rotated 360°(full scan) during scanning, with images being captured every 0.4°, i.e., 901 images, with an exposure time of 6 s per image, resulting in a scan time of 5 h: 30 min per sample, including scan performed for compensating shifts errors.The appropriate geometrical settings of the object position from the source, object position from the detector, and detector system are selected to achieve a voxel size of 8.0 μm.NRecon (Bruker) software is used to reconstruct the TIFF images from the CT scan data.Dataviewer (Bruker) software visualised the reconstructed scan data in 3D.The greyscale histogram is changed after applying all filtering techniques to get an optimal contrast between the object and the background.The rendered volumes of the ROI displaying the segmentations of the 3D images are pictured using CTVox (Bruker) software.Finally, using CTan (Bruker) software, the required measurements are taken and recorded.

Results and discussions
This section discusses the results of the materials and mechanical characterisation and XRμCT scans for each design.

Materials characterisation
The micrographs for the polished and etched surfaces of the as printed part using LS technique of the PBF method using AISI 1.2709 is shown in figure 3. When considering the micrographs of the part across the build platform (perpendicular to z axis), as shown in figure 3(a), the laser scan tracks width and its corresponding 67°r otation angle in successive layer is clearly visible.This rotation in scan tracks is due to the scan strategy used for printing the part.The width of the scan track measured is in the 110-150 μm range when the laser beam diameter of 100 μm is used.The difference in the width is due to the laser process parameters used during 3D printing.When considering the micrographs along the build platform (z axis), as shown in figure 3(b), melt pools are distinctively visible, separated by melt pool boundaries in the shape of semi-ellipse.The height of the semi-ellipse melt pool measured is in the 47-71 μm range.The reason for the distinctive visibility of the melt pool with boundaries is due to the non-homogeneous cooling rate within a melt pool, with maximum cooling rate located at the boundaries of the melt pool.This higher cooling rate at the boundaries results in much faster cooling of the material than at the inner regions of the melt pool.This region when etched will distinctively separate the boundary and the inner regions.From the figure 3(c), the higher magnification of the micrographs across the build platform (perpendicular to z axis) reveals the grain structure of the as printed part.The microstructure is in the form of very fine cellular structure.This fine nature of the microstructure is due to the extremely high cooling rate, in the range of 10 6 K s −1 achieved in the PBF process.This fine cellular structure achieved will directly have an impact on the mechanical properties such as hardness and strength.From the figure 3(d), the higher magnification of the micrographs along the build platform, reveals columnar dendrites oriented along the direction of cooling, i.e., along the build platform as the cooling occurs first in the first built layer.
A detailed elemental distribution of the as printed part is analysed by performing FESEM-EDS analysis.The FESEM image and EDS element distribution maps are shown below in figure 4.
From the FESEM-EDS analysis, it can be confirm that all the major elements such as Fe, Ni, Mo, Ti and Co are all distributed evenly throughout the part after printing.The wt% of all the elements can be found from the EDS spectrum analysis as shown below in figure 5.
From the spectrum analysis it is seen that the wt% of each element present in the as printed tool are within the specified limits as per the feedstock manufacturer and supplier (EOS GmbH, Germany).
The XRD pattern of the as printed part is shown below in figure 6, with the diffraction peaks α(110), α(200) and α(211) of the martensitic phase are identified at diffraction angles of 45.2°, 65.3°and 82.6°respectively.There is a weak austenite peak γ(111) identified at 44.3° [27][28][29][30].As analysed here, the phases are majorly composed of martensitic phases with only a minimal amount of retained austenite.This retained austenite phase is due to the micro-segregation of the solute element (Nickel) occurring at the boundaries of the melt pool and its subsequent enrichment during curing due to higher cooling rate makes this austenite phase stable even upon cooling to room temperature and get detected during analysis.
The relative density of the as printed part when measured using the Archimedes principle by using the density measurement setup (Mettler Toledo) is 98.9% (±0.32%), is very close to fully dense part for AISI 1.2709 steel.This density measured by analytical method is also confirmed by porosity analysis by XRμCT scan.
Figure 7 shows the areas in which the surface attributes is measured.Two random areas (1&2) are selected on the rake face of the tool.
Figure 8 shows the surface in three dimensions and its maximum height of the profile measured (R t ) for the selected ROI.From this 3D analysis, the average surface roughness (R a ) of the area 1 is 3.47 μm with the average maximum height of the roughness profile (R z ) being 11.37 μm.For area 2, the R a is 3.57 μm and R z is 16.81 μm.

Mechanical characterisation
The microhardness of the as printed part in two directions, along and across the build platform is shown below in the figure 9.The average microhardness of the as built part along the build platform is 405.5 ± 7.3 HV and across the build platform is 420.9 ± 5.3 HV.
The variation in average hardness is due to the formation of fine cellular microstructure after curing across the build platform (cross section), which is missing along the build platform.Although a variation occurs, this difference is minimal, and it is concluded that there is no anisotropic behaviour in the materials as 67°scan strategy was used.Thus, this high microhardness achieved will help in maintaining the cutting edge for a longer time during machining and it can be considered for machining various steels and alloys which has a lower hardness than the AISI 1.2709.
The materials and mechanical properties observed here is same for all the five designs, as all the parts are printed using the same process parameters.The discussed results above are common to the five parts.

X-ray micro computed tomography 3.3.1. Porosity measurement
From the XRμCT scan conducted for measuring the internal features dimensions, the porosity is also measured and there is no measurable pores formed in the part during the PBF process.Thus it can be concluded that the density achieved is very high, close to fully dense and the values produced by analytical method is validated.

Circular angled loft (Design 1-A)
Figure 10 shows the collective data from XRμCT scan for Design 1-A, (a) scanned 3D model of the part, (b) separately represented internal TC channels inside the tool, (c) the cross-section data of the scans from the CTVox software used for measurements in CTan software.From figure 10(a), we can see the quality of the XRμCT scan is very high with minimal noise, as no data points are lost during the scan and reconstruction.But from figure 10(b), a minor loss of data is seen when the greyscale histogram adjustments are made to visualise the internal TC structures separately.The reason for this data loss is the reflection of the small channels captured by the detector and the inability of the image processing software equipped with the system to distinguish between the object and the background.
From figure 10(c), the cross-section of the top view of the 3D printed monolithic single-point cutting tool is used for measuring the diameter of the TC channels.The channel near the cutting edge is named 1, and the  It is seen that the channels 1, 2, 3 and 8 are almost entirely blocked, with their average diameters being 240 μm (deviation −52%), 256.66 μm (deviation −48.66%), 242.66 μm (deviation −51.46%) and 314.66 μm (deviation −37.06%) throughout the build.The rest channels are built with an average diameter of 400 μm (deviation −20%).The bottom layer measurement is low due to the surface and the support structures of the chamber and the shift of layer caused by the unsupported area.Channels 1, 2, and 3 made a steeper angle (>45°) to the build platform (z axis) than the other channels, whose angles are 30°−50°.This dimensional deviation occurring due to the build angle is confirmed by [21][22][23][24]31].All the channels till the top are circular in profile, which complies with the intended design.The profile at the rake face of the tool is ovular due to the angles of the tool.

Circular free-form (Design 1-B)
Figure 12 shows the collective data representation for Design 1-B, (a) the scanned 3D model of the part, (b) the separately represented internal TC channels, (c) the cross-section data of the scans from the CTVox software.From figure 12(a), it is seen that the quality of the XRμCT scan is very high with minimal noise, as no data points are lost during the scan and reconstruction.But from figure 12(b), a minor loss of data is seen when the greyscale histogram adjustments are made to visualise the internal TC structures separately.The main reason for this data loss is the reflection of the small channels captured by the detector and the inability of the image processing software equipped with the system to distinguish between the object and the background.
From figure 12(c), the front view of the scan is used for measuring the bottom position diameter and the top view for the other two positions as the design is free-form.The diameter of the TC channels is measured three times for each channel at three different positions: bottom, middle and top and is represented as a graph shown in figure 13.From the graph, the mean diameter of the channels measured in the bottom position is 436 μm, the middle position is 366.67 μm, and the top position is 406 μm.The percentage deviation of the diameters is −12.8%,−26.67% and −18.8%, respectively, with an overall average deviation of −19.42%.
It is also seen that all the TC channels are built with high dimensional accuracy when measured at the bottom position because they make a 0°angle to the build platform, but when it is measured at the middle position, the deviations increased significantly.In channels 7, 8 and 9, the dimensions achieved are 360 μm (deviation −28%), 300 μm (deviation −40%) and 312 μm (deviation −37.6%), respectively, as the angle of the free-form increases steeply.Significantly the same channels are affected even at the top position, with the dimensions achieved being 360 μm (deviation −28%), 376 μm (deviation −25.6%) and 312 μm (deviation −37.6%).This deviation is mainly due to the angle at which it is printed, and the trend it follows is confirmed by [21][22][23][24]31].The free-form path's angle is severe (>45°) for channels 7, 8 and 9.All the channels till the top are circular in profile, which complies with the as designed dimensions.The profile at the rake face of the tool is ovular due to the angles of the tool.From figure 14(a), it is seen that the quality of the XRμCT scan is very high with minimal noise, as no data points are lost during the scan and reconstruction.But from figure 14(b), a minor loss of data is seen when the greyscale histogram adjustments are made to visualise the internal TC structures separately.The main reason for this data loss is the reflection of the small channels captured by the detector and the inability of the image processing software equipped with the system to distinguish between the object and the background.
From figure 14(c), the top view of the scan is used for measuring all the position's diameters.Since the profile design is an inscribed hexagon, the diameter of the inscription circle will fully justify the shape and size of the channel.The diameter of the TC channels is measured three times at three different positions: bottom, middle and top and is represented as a graph shown in figure 15.From the graph, the mean diameter of all the channels measured in the bottom position is 405.33 μm, the middle position is 405.33 μm, and the top position is 397.44 μm.The percentage deviation of the diameters is −18.93%,−18.93% and −20.51%, respectively, with an overall average deviation of −19.45%.
Considering channels 4 and 5, the average dimensions achieved at the three measuring positions are 369 μm (deviation −26.2%) and 362.66 μm (deviation −27.46%), respectively.All other channels are built with an average of 413.24 μm (deviation −17.35%).The first row of channels made a steeper angle (>45°) than the other channels with angles of 30°−50°.This dimensional deviation occurring due to the build angle is confirmed by [21][22][23][24]31].The dimensional deviation in the built channel is mainly due to the adhesion of unsintered powder particles at the profile edges, decreasing the diameter and the step effect caused due to steeper angles causing overhangs.All the channels built in this design are uneven circular in profile, sometimes even making it difficult to define a shape.

Hexagonal free-form (Design 2-B)
Figure 16 shows the collective data representation for Design 2-B, (a) the scanned 3D model of the part, (b) the separately represented internal TC channels, (c) the cross-section data of the scans from the CTVox software.From figure 16(a), it is seen the quality of the XRμCT scan is very high with minimal noise, as no data points are lost during the scan and reconstruction.But from figure 16(b), a minor loss of data is seen when the greyscale histogram adjustments are made to visualise the internal TC structures separately.The main reason for this data loss is the reflection of the small channels captured by the detector and the inability of the image processing software equipped with the system to distinguish between the object and the background.
From figure 16(c), the front view of the scan is used for measuring the bottom position diameter and the top view for the other two positions as the design is free-form.Since the profile design is an inscribed hexagon, the diameter of the inscription circle will fully justify the shape and size of the channel.The diameter of the TC channels is measured three times at three different positions: bottom, middle and top and is represented as a graph in figure 17.From the graph, the mean diameter of all the channels measured in the bottom position is 469.33 μm, the middle position is 366.22 μm, and the top position is 426.66 μm.The percentage deviation of the diameters is −6.13%, −26.75% and −14.67%, respectively, with an overall average deviation of −15.85%.
The channels 5, 7, 8 and 9 in the middle position of measurement, the dimensions achieved are 336 μm (deviation −32.8%), 336 μm (deviation −32.8%), 352 μm (deviation −29.6%) and 288 μm (deviation −42.4%).This is because the angle at which the channels are built in the middle regions is very steep (>45°), resulting in overhangs and step effects.This dimensional deviation occurring due to the build angle is confirmed by [21][22][23][24]31].This trend in the dimensional deviation of channels 7, 8 and 9 follows the same pattern as in Design 1-B, which is similar in free-form.The severity (>45°) of the free-form angle of channels 7, 8 and 9 causes  higher deviation in the middle measurement position.All the channels built in this design are uneven circular in profile, sometimes even making it difficult to define a shape.

Blood vessel (Design 3)
Figure 18 shows the collective data representation for Design 3, (a) the scanned 3D model of the part, (b) the separately represented internal TC channels, (c) the cross-section data of the scans from the CTVox software.From figure 18(a), it is seen that the quality of the XRμCT scan is very high with minimal noise, as no data points are lost during the scan and reconstruction.But from figure 18(b), a minor loss of data is seen when the greyscale histogram adjustments are made to visualise the internal TC structures separately.The main reason for this data loss is the reflection of the small channels captured by the detector and the inability of the image processing software equipped with the system to distinguish between the object and the background.
From figure 18(c), the top view is used for measuring all diameters.The four-channel array nearest to the cutting edge is named 1-4, and the one opposite is 12-16, following an array of 4 × 4 patterns repeated four times.The diameter of the TC channels is measured three times for the trunk and branches of the design and is represented as a graph in figure 19.From the graph, the mean diameter of the bottom trunk (Bottom) is 3457.7 μm (deviation −1.2%), the intermediate four branches (Middle) are 1458.39μm (deviation −2.77%), and the top 16 branches are 444.5 μm (deviation −11.1%).This design's overall deviation of the TC channels is −5.02%.
The main reason for the higher dimensional accuracies achieved is that the channels are built at a 0°to the build platform.As confirmed earlier by [21][22][23][24]31], the angle at which the channels are built plays a significant role in dimensional accuracies.

Conclusions
In this study, five different TC channels, namely two circular, two hexagonal profiled channels and a bioinspired blood vessel channel are successfully designed and 3D printed.The AM manufactured parts are then characterised for its material, mechanical attributes and non-destructively characterised using XRμCT scan (3d scan) to study the dimensional accuracy achieved.The key findings from the studies are listed below.
• The as printed part has a fine cellular microstructure with martensitic phases, resulting in high hardness even without any post processing, qualifying the entity for consideration in machining steels and alloys.• The TC channels are built with higher accuracy in free-form designs because the angles pertaining to the channels are not harsh, and the layer shift and step effect in the build is minimum, thus concluding the best designs are 1-B, 2-B and 3.
• The profile of hexagonal channels is very irregular for inscribed hexagons.To achieve a better profile throughout the part, it is recommended to design a circumscribed hexagon.In addition, hexagonal profiles increase the structural stability of the channels.
• To achieve a better dimensionally accurate part via the PBF method, it is suggested to design parts with angles ranging from 30°to 45°and 0°or 90°with 0°and 90°, the most recommended.
• With the advancement of the AM technology, commercially used tool steels (high-speed steel and tungsten carbide) can be 3D printed with complex TC channels.

Figure 3 .
Figure 3. Optical micrographs of the part (a) Across the build platform in 500× (b) Along the build platform in 500× (c) Across the build platform in 1000× (d) Along the build platform in 1000×.

Figure 4 .
Figure 4. FESEM image and EDS elemental distribution mapping of as printed AISI 1.2709 tool.

Figure 6 .
Figure 6.XRD pattern and identification of Braggs peaks on as printed AISI 1.2709 tool.

Figure 7 .
Figure 7. Areas on the rake face of the tool selected for surface roughness measurement.

Figure 8 .
Figure 8. 3D maximum height of the profile measured (R t ) for the selected ROI.

Figure 9 .
Figure 9. Microhardness plot of as printed part along and across the build platform.

Figure 10 .
Figure 10.Scan data for Design 1-A (a) Scanned 3D model (b) Separate representation of internal channels (c) Cross-section image from CTVox.

Figure 11 .
Figure 11.Graphical representation of all the TC channels for Design 1-A.

Figure 12
Figure 12 Scan data for Design 1-B (a) Scanned 3D model (b) Separate representation of internal channels (c) Cross-section image from CTVox.

Figure 13 .
Figure 13.Graphical representation of all the TC channels for Design 1-B.

3. 3 . 4 .
Hexagonal angled loft (Design 2-A) Figure 14 shows the collective data representation for Design 2-A, (a) the scanned 3D model of the part, (b) the separately represented internal TC channels, (c) the cross-section data of the scans from the CTVox software.

Figure 14 .
Figure 14.Scan data for Design 2-A (a) Scanned 3D model (b) Separate representation of internal channels (c) Cross-section image from CTVox.

Figure 15 .
Figure 15.Graphical representation of all the TC channels for Design 2-A.

Figure 16 .
Figure 16.Scan data for Design 2-B (a) Scanned 3D model (b) Separate representation of internal channels (c) Cross-section image from CTVox.

Figure 17 .
Figure 17.Graphical representation of all the TC channels for Design 2-B.

Figure 18 .
Figure 18.Scan data for Design 3 (a) Scanned 3D model (b) Separate representation of internal channels (c) Cross-section image from CTVox.

Figure 19 .
Figure 19.Graphical representation of all the TC channels for Design 3.

Table 2 .
Optimised process parameters used for metal 3D printing.