An Additively Manufactured Titanium Alloy in the Focus of Metallography


 Additive manufacturing processes allow the production of geometrically complex lightweight structures with specific material properties. However, by contrast with ingot metallurgy methods, the manufacture of components using this process also brings about some challenges. In the field of microstructural characterization, where mostly very fine structures are analyzed, it is thus indispensable to optimize the classic sample preparation process and to furthermore implement additional preparation steps. This work focuses on the metallography of additively manufactured Ti‑6Al‑4V components produced in a selective laser melting process. It offers a guideline for the metallographic preparation along the process chain of additive manufacturing from the metal powder characterization to the macro- and microstructural analysis of the laser melted sample. Apart from developing preparation parameters, selected etching methods were examined with regard to their practicality.


Introduction
The production of metallic structural components by additive manufacturing, AM, has become established in the last two decades. Universities and research institutions have considerably contributed to its use in industrial applications. This method is also already established in the manufacture of lightweight construction materials such as titanium and titanium alloys. It is used in the aviation sector, in medical technology, and in the automotive and ship building industry in the production of partially complex components that are, for instance, subjected to elevated temperatures and corrosive media [1]. Titanium alloys are classified as α, near-α, α+β and β alloys depending on their alloying elements [2]. The α+β alloy Ti-6Al-4V (m.%) is the most frequently used titanium alloy [3,4]. Components made of this alloy are, for instance, manufactured using a powder bed-based selective laser melting technique (Laser Powder Bed Fusion, LPBF). Against the backdrop of a process and product quality assurance process, the material is characterized -from the initial metallic powder to the additively manufactured solid material sample. The quality of the end product is closely linked with the powder quality. Parameters such as chemical composition, flow behavior, tap and bulk density, sphericity, porosity, and particle size distribution are therefore analyzed [5]. The main focus of the examination of the additively manufactured Ti-6Al-4V samples is on microstructural characteristics such as structural homogeneity, appearance of the hexagonal (hex) martensitic α´ phase, and phase fractions [6,7]. The AM process generates microstructures that differ fundamentally from those generated based on klassischen schmelzmetallurgischen Herstellprozess unterscheiden. Beispielsweise kommt es bei der additiven Fertigung von Ti-6Al-4V zu sogenannten "Layer-Bands" bzw. zur Ausbildung einer schichtartigen Struktur. Diese Schichtstruktur entsteht durch Segregationen der Legierungsbestandteile während des Erstarrungsprozesses sowie der Reaktion der Oberfläche mit der Umgebungsatmosphäre und repräsentiert somit die Schmelzbadgrenzen des Fertigungsprozesses [8,9]. Im Rahmen dieser Arbeit wird auf die Kontrastierung dieser Schichtstrukturen eingegangen.
Nach Abschluss des additiven Fertigungsprozesses wird das Bauteil meist einem Nach-classical melting metallurgy manufacturing processes. For instance, a layered structure with so-called "layer bands" is formed when Ti-6Al-4V is additively manufactured. This layered structure forms owing to segregations of the alloy constituents during the solidification process and the reaction of the surface with the ambient atmosphere and thus represents the molten pool boundaries of the manufacturing process [8,9]. This work also discusses how these layer structures are contrasted.
From a microstructural point of view, SLM manufactured α+β alloys, such as Ti-6Al-4V, exhibit an epitaxially β solidification process through several applied layers inducing an anisotropic material behavior [8]. During the solidification process, the bodycentered cubic (bcc) β phase, the high temperature phase of Ti and its alloys, is characterized by a preferred orientation in 〈100〉 direction. Given that the size and shape of this columnar β grain structure are a function of the SLM manufacturing parameters and subsequently impact on the component properties, it is particularly relevant to reveal this process-microstructure-property relationship. Knowing these relationships, SLM manufacturing parameters can be optimally adapted in order to achieve desired mechanical properties [8]. The β grain structure formed from the melt in the epitaxial solidification process can either be visualized by etching (owing to the allotropic phase transformation (β → α´)) or by reconstruction using a program-based process. The program-based back-calculation is based on data from backscatter diffraction experiments. The data analysis is, for instance, performed based on ARPGE data [10,11] (Automatic Reconstruction of Parent Grains from EBSD).
Components made of titanium alloys are usually prepared in a multi-stage grinding and polishing process subsequent to which they are directly examined or etched. In this context, Kroll etching is a frequently applied method for revealing the microstructure of the castings and forgings [7, 14 -17]. The suitability of this preparation technique for additively manufactured titanium components is assessed in this work and the optimization of the metallographic preparation along the process chain, from the powdery raw material to the additively manufactured component, is outlined taking the example of the two-phase alloy Ti-6Al-4V.

Experimental Procedure
The metallographic preparation was carried out on additively manufactured components made of gas-atomized titanium alloy powder Ti-6Al-4V, grade 23 ELI (Extra Low Interstitials). For comparison purposes, the powder's particle size determination was performed applying two different methods. In addition to dry screening with mesh widths of 100, 63, 40, and 25 µm, a Sympatec HELOS laser diffraction spectrometer was used to determine the grain size. The chemical composition of the powder was determined by atomic emission spectroscopy for the elements Al (6.27 m.%), V (3.9 m.%), and Fe (0.2 m.%), and by carrier gas hot extraction for the ele-ments O (0.083 m.%) and N (0.0205 m.%), respectively. To characterize the powder, the powder flowability was determined by measuring the flow rate (ASTM B964 [18]) using a Carney funnel, while the powder bulk density was measured according to ASTM B212 [19] using a Hall flow meter. The sphericity measurements were carried out using a Retsch Technology particle size analyzer CAMSIZER XT.
The samples for the test series were manufactured on an EOS machine of the type M290. SLM manufactured density cubes with the dimensions 22 ×22 ×25 mm 3 were manufactured for the characterization of the microstructure. Not just the as-built condition, i. e. the condition immediately subsequent to the manufacturing process, but also a heat-treated variant was analyzed. The heat treatment was performed by applying temperatures ranging from 800 °C to 850 °C for 1 -3 hours in Ar atmosphere. Due to non-disclosure agreements, the exact parameters of the heat treatment must not be specified. The samples were cut in a cutting machine ATM Brillant 221 using a resin-bonded, abrasive diamond cut-off wheel of the type BOD13 (Struers). It should be noted here that more detailed information about the preparation methods and parameters can be found in the next chapter. Subsequent to the cutting process, the sample material was embedded using the Struers hot mounting press CitoPress-20. For the sample mounting process, we had the bakelite-based mounting resin PolyFast (Struers) and the Cu-based mounting compound ProbeMet (Buehler) at our disposal. Grinding and polishing of the mounted samples were performed on a semi-automatic grinding and polishing machine Struers Tegramin-30. The vibratory polishing machine VibroMet-2 was used in combination with a chemically resistant polishing cloth of the type MicroCloth (Buehler) for further polishing. Electrolytic polishing and etching steps were performed on a extraktion für die Elemente O (0,083 m.%) und N (0,0205 m.%). Im Zuge der Pulvercharakterisierung wurde die Pulver-Fließfähigkeit mittels Messung der Durchflussrate (ASTM B964 [18]) unter Verwendung eines Carney-Trichters bestimmt und die Pulver-Schüttdichte nach ASTM B212 [19]  Struers LectroPol-5 machine using the electrolytes listed in Tab. 1. The lower part of the table also lists the etchants used to reveal the macro and microstructure. The present phases were also analyzed in X-ray diffraction experiments (XRD). For this purpose, a Bruker D8 Advance diffractometer was used in Bragg-Brentano geometry with a parallel beam configuration applying Cu-K α radiation (λ=1.54 Å). The diffractograms were acquired using a Sol-X detector applying the following measurement parameters: step size of 0.02 °, measuring time of 2 s.
secondary electron detector was used for high resolution examinations. The electron backscatter diffraction (EBSD) measurements for a determination of the crystallographic structure and the orientation of the crystalline phases were also performed on the above-mentioned FEI device using a Hikari XP EBSD camera and an EDAX EBSD detector. Data were acquired and analyzed using the EDAX software OIM Data Collection and OIM Analysis 7 based on the following EBSD measurement parameters: accelerating voltage of 20 kV, step size of 100 nm and a 4×4 binning mode of the EBSD camera.

Material Sampling and Mounting
Fig . 1 shows the metallographic sample preparation workflow. In a first step, the sample material is taken. First, the powder is sampled. It should preferably be taken at several container positions. It is thus ensured that no individual powder fractions of a particular particle size are preferably examined, such as due to segregations.  Die Partikelmorphologie des Pulvers ist in Bild 3a) durch eine REM-Aufnahme im Sekundärelektronen-Modus dargestellt. Anhand der REM-Aufnahmen ist ersichtlich, dass die Par-Subsequent to the powder sampling process, the particles are characterized. For this purpose, particle size distribution, morphology, sphericity, flow rate, and bulk density are determined. In the diagram in Fig. 2, the cumulative powder particle distribution is plotted against the particle size. As is apparent from the diagram, there is a significant difference between the particle size distribution determined by laser diffraction and the distribution determined by sieve analysis. It should be taken into consideration that the sieve analysis was performed with a limited number of mesh sizes. As compared to the distribution obtained by laser diffraction-based particle sizing which allows for a finer gradation, this limited number of mesh sizes as well as sieve clogging by fine particles therefore resulted in a coarser distribution and over-estimated d values (d 10 , d 50 , d 90 ). Due to the mentioned problem, dry sieving should therefore be dispensed with for an examination of particles with a diameter of < 40 µm.   [20]. The bulk density of the powder is 2.49 g/cm 3 . It thus amounts to 56 % of the theoretical density of Ti-6Al-4V [21]. The Carney flow rate of the powder according to ASTM B964 [18] is 6.4 s/50 g.
A metallographic section of the powder has to be prepared for the purpose of examining any existing gas pores within the powder particles and for the structural analysis. As is apparent from Fig. 3b), individual powder particles may be removed during the metal powder mounting process (marked in red).
To avoid this, it is recommended to either mix the metal powder particles with a sieved fine fraction (< 25 µm) of the mounting medium before mounting or to use a very fine mounting material with a low grain size. The use of a fine mounting material allows for reducing the fraction of removed powder particles. Die Entnahme der Vollmaterialproben erfolgt über einen Probenzuschnitt durch abrasiv wirkende Schneidscheiben. Diese bieten den Vorteil ein breites Spektrum an Werkstoffen trennen zu können und darüber hinaus eine gute Oberflächenqualität nach dem Trennprozess zu gewährleisten [16]. Bei unzureichender Kühlung oder durch einen zu hohen Vorschub der Trennscheibe kann es jedoch zum Überhitzen des Werkstoffs kommen. Während des Schneidvorgangs sollte daher stets auf eine ausreichende Kühlung sowie die richtige Vorschubgeschwindigkeit geachtet werden, da derartige Probenschädigungen nur sehr aufwendig wieder entfernt werden können bzw. zu Fehlinterpretationen des Gefüges führen können. Bei einer Vorschubgeschwindigkeit von 1 mm/min ist es zweckmäßig eine Umdrehungsgeschwindigkeit der Trennscheibe von 2500 -3600 min -1 einzuhalten. Das verwendete Kühlmittel sollte eine Mixtur aus Wasser, Schmiermittel und einem Korrosionsschutz in Form eines Additivs sein. Das Einbetten der additiv gefertigten Vollmaterialproben erfolgte entweder durch ein kunststoff-oder ein Cu-basiertes Einbettmittel. Bei der Verwendung des Cu-basierten Probe-Met ist der Härteunterschied zwischen Probe und Einbettmittel geringer, was eine bessere mechanische Präparation, auch im Bereich der Randschichten, ermöglicht. Gerade bei additiv gefertigten Bauteilen ist eine Oberflächen-und Randschichtuntersuchung von besonderer Bedeutung. Aufgrund der guten elektrischen Leitfähigkeit des Kupfers ist zusätzlich für einen hohen und stabilen Sekundärelektronenstrom (und damit einem guten Kontrast) im Zuge der elektronenmikroskopischen Untersuchung gesorgt. Nachteile dieses Einbettmittels sind jedoch eine höhere Aushärtezeit sowie höhere Materialkosten bei der Anschaffung. Im Zuge dieser Arbeit wurde für Untersuchungen im REM das Cu-basierte Einbettmittel ProbeMet verwendet. Für alle anderen Untersuchungen fand das kunststoffbasierte Einbettmittel seine Anwendung. Zur Herstellung der Schliffe wurden folgende Einstellungen an der Einbettpresse vorgenommen: The solid material samples are taken by sample cutting with the aid of abrasive cut-off wheels. They offer the advantage of being able to cut a wide range of materials and furthermore ensure a good surface quality after the cutting process [16]. However, in case of insufficient cooling or when the feed rate of the cut-off wheel is too high, the material may overheat. Such sample damages may lead to a misinterpretation of the microstructures and, once they have occurred, it is rather complicated to remove them. Sufficient cooling and the appropriate feed rate should therefore be ensured during the cut-off process. At a feed rate of 1 mm/min, it is appropriate to maintain a rotational speed of the cut-off wheel of 2500 -3600 min -1 . The used coolant should be composed of water, lubricant, and an anticorrosive agent additive. The additively manufactured solid material samples were either mounted in a bakelite mounting resin or a Cu-based mounting compound. The difference in hardness between the sample and the mounting medium is smaller using the Cu-based ProbeMet compound, thus allowing an enhanced mechanical preparation, including the surface layers. The examination of the surface and the surface layer is particularly important when analyzing additively manufactured components. During the electron microscope examination, the good electrical conductivity of copper additionally ensures a high and stable flow of secondary electrons (and thus a good contrast). However, longer curing times and higher acquisition costs are the drawbacks of this mounting material. For this work, the Cu-based mounting compound ProbeMet was used for SEM examinations. For all other examinations, the bakelite-based mounting resin was used. The sections were manufactured applying the following mounting press settings: • Kunststoffeinbettmittel Polyfast: Heizphase bei 180 °C, 250 bar für 3,0 min, gefolgt von 1,5 min Wasserspülung.

Grinding and Polishing
Subsequent to cutting and mounting, the sample preparation workflow involves a wet grinding and polishing step. The rough and deformed zone resulting from the cutoff process is unfavorable for the analysis. It is removed by plane grinding ensuring a plane-parallel sample surface [16,17]. Subsequently, successive fine grinding steps are performed using progressively finer grits. It should be taken into consideration here, that severe damages such as scratches, relief, break-offs, and cracks cannot be removed anymore in the following polishing steps [16,17].
The dense, thin, and stable oxide layer on the surface with its high affinity for oxygen presents a challenge in the preparation of Ti alloys. Although it provides effective corrosion protection [22], it needs to be removed in the course of sample preparation. A chemical-mechanical polishing process is appropriate to suppress the formation of an oxide layer, especially during the last polishing steps. The formation of a surface layer is suppressed, i. e. the removal rate is increased and the formation of a cover layer is prevented by adding hydrogen peroxide H 2 O 2 to the colloidal silicon dioxide SiO 2 polishing suspension (OPS) [16,23]. Tab. 2 shows the parameters of the grinding and polishing steps applied based on which reproducible results and an optimal metallographic sample preparation providing excellent surface quality could be achieved. The difference in hardness between the investment material and the sample and the thus generated sample elevation may provoke an undesired rocking motion of the sample zu einer unerwünschten Wippbewegung der Probe kommen [24].  [24]. It is particularly important to counter these rocking motions. Therefore, after the mounting process, all sample surfaces to be ground were given a chamfered edge. The sample motion can further be reduced by firmly fixing all samples in a sample holder instead of using a non-fixing sample holder with single piston force transmission. However, it should be taken into consideration here, that, owing to the fact that an uneven force application can occur induced by the central contact pressure, at least three samples are prepared simultaneously when using a fixing sample holder. In the course of the preparation optimization process, the removal of scratches was also analyzed with regard to the rotational motion of the sample holder relative to the grinding and polishing pad. It was found that the deformation zone was smaller when sample holder and working disk work in synchronization than when they are configured in counter rotation. Synchronism of the sample and the working disk reduces the relative speed and simultaneously suppresses rocking motions of the sections.
As is shown in Tab. 2, the plastic bonded diamond grinding wheel MD-Piano 220 (Struers) was used in preparation step 1 to provide plane-parallelism and to ensure the flatness of the section's surface. As this procedure may take up to 4 minutes, it is advantageous to use a manually resharpenable grinding wheel. It ensures a consistent removal rate and causes less wear as compared to SiC paper. When using SiC paper, the material removal rate is significantly reduced within a very short time. Therefore, the grinding time was limited to 15 to 20 seconds for the preparation steps 2 to 5, respectively. The last preparation step, step 6, is an OPS polishing process. For this procedure, a mixture of OPS and H 2 O 2 with a ratio of 90 ml OPS:15 ml H 2 O 2 is applied to a chemically resistant polishing cloth (Struers). It may take up to 8 minutes to remove the deformation zone of step 5 by OPS polishing.
It is equally important to perform intermediate sample cleaning after each preparation step. This ensures that no particles dislodged from the roughness and deformation zone are transferred to the next preparation step and, as a result, damage the surface. The sample cleaning processes during the preparation were carried out by thoroughly washing the sample holder and the clamped samples with soap. Once the preparation steps are completed, it is appropriate to remove the metallographic samples from the sample holder in order to clean them in an ultrasonic bath, for instance in ethanol solution, for several minutes. Subsequent to the ultrasonic bath, highly volatile substances durch Aufbringung von hochflüchtigen Substanzen, wie z. B. Isopropanol oder Petrolether, verhindert, dass sich Trocknungsartefakte auf der finalen, spiegelglatten Oberfläche bilden.

Porenanalyse
Bild 4b) zeigt eine für die Porenanalyse präparierte Aufnahme der Oberfläche einer as-built such as isopropyl or petroleum ether are furthermore applied to avoid the formation of drying artifacts on the final, glassy surface.

Microstructural Examination by Light Microscopy
In the SLM process, powder layers are selectively heated and molten. Once the molten pool is solidified, further cooling results in α lamellae growth at the β grain boundaries from the moment of reaching the β-transus temperature. Given sufficient undercooling, a diffusion-less martensitic transformation to acicular α´ martensite takes place induced by Umklapp processes once the temperature falls below the martensite start temperature. Moreover, the repeated thermal cycles of the AM process lead to an epitaxial growth of the β grains induced by the steep temperature gradient in the molten pool and the resulting lower nucleation rate, as is described in [6]. The aforementioned microstructure formation can be evidenced after the basic sample preparation. The crystallographic orientations of the material can be revealed as early as after preparation step 6 by using a polarization filter in the LM without recourse to specific surface-modifying methods such as for selective etching. Fig. 4a) is a LM image of an additively manufactured Ti-6Al-4V sample in the as-built condition using a polarization filter combined with increased light intensity. Not only the acicular α´ martensite structure, but also the former β grain boundaries can be recognized (dashed lines). It is apparent that the columnar β grains extend over several hundred micrometers in the build direction, BD. Fig. 4b)  , but also a lack of fusion defect (dashed red ellipse) could be detected in this sample during the 2D pore analysis. It is important to make sure the section's surface is dust-free, especially when determining the porosity in AM manufactured samples of a very high density (> 99.95 %), as otherwise the result may be distorted. In addition to that, a preferably large part of the surface should be considered and evaluated in order to ensure a statistically relevant result. The magnification for the LM images should not be less than 200× to ensure that smaller pores can be detected. The porosity was evaluated by phase fraction analysis using the software Olympus Stream Motion 1.9.3. The pores must additionally be manually sorted as a function of the surface area, and smaller pores (< 5 µm 2 ) must be removed to avoid that pixels appearing dark are erroneously assigned to the pore fraction. By adhering to the mentioned process steps, a very precise

b) a)
and reproducible porosity can be represented. In this work, a surface area of at least 100 mm 2 covering different positions of the section represented in different images was accounted for during the pore analysis. For this purpose, a section from the center of the component was taken for examination. Care should be taken that the examined site is not located too near the component's edge, as, due to the high roughness in the edge areas, a non-representative porosity may be accounted for. A porosity of 0.028 vol.% was measured for the as-built sample shown in Fig. 4b). By comparison, the heat-treated variant has an even lower porosity of 0.008 vol.%, i. e. the sample is virtually completely dense.

Phase Analysis
One of the methods used to determine the phases and phase fractions, is the X-ray diffraction examination. As it is a surface sensitive method, a deformed sample surface may lead to distorted measurement results and thus to misinterpretations. Turk et al. [25] showed that preparation-related residual compressive stresses induced by the grinding and polishing processes result in shifted and broadened XRD peaks. It is thus indispensable to ensure a deformation-free plane surface. Fig. 5 shows an extract of XRD measurements performed on laser melted samples and on the powder sample. The solid material samples were treated following the preparation method presented in Tab. 2, removed from the mounting material, and analyzed. The powder samples were poured on a monocrystalline Si wafer plate for their examination and, as opposed to the solid material samples, the measurement was carried out without sample rotation. In Bild 5 ist der Auszug aus XRD-Messungen an lasergeschmolzenen Proben sowie an der Pulverprobe dargestellt. Die Vollmaterialproben wurden nach der in Tab. 2 angegebenen Präparationsmethode behandelt und im ausgebetteten Zustand analysiert. Die Pulverproben wurden für die Untersuchung auf eine einkristalline Si-Waverplatte aufgeschüttet und die Messung erfolgte im Gegensatz zu den Vollmaterialproben ohne Probenrotation.
The obtained diffraction spectra show that the α/α´ peaks (dashed peak position) are shifted from the powder samples to the heat-treated sample via the as-built sample. The diagram furthermore reveals that the bcc β phase (dash-dotted position) occurs in the heat-treated condition. The shift of the hex 0002 peak and the precipitation of the β phase are an indication of a phase transformation towards the thermodynamic equilibrium (α´ → α+β). The atomic radii of Al (0.143 nm) and V (0.132 nm), smaller as compared to those of Ti (0.147 nm), thus lead to an increasing α lattice parameter (decreasing 2θ angle) for an enrichment of V in the β phase [7].

Enhanced Sample Preparation Electrochemical Etching
Electrochemical etching is a corrosive process generating a metallographic microstructural contrast based on a potential difference of different surface areas [26]. Potential differences occur owing to physical, structural, and chemical inhomogeneities such as those generated by the presence of multiple phases of varying density, with varying crystallographic orientations, or segregations [16]. At room temperature, the hex α phase and the bcc β phase are present in the thermodynamic equilibrium in Ti-6Al-4V. A potential difference is already present here due to the different chemical compositions of these phases. An etchant frequently used for Ti-6Al-4V is the so-called Kroll's reagent. It contains hydrofluoric acid for a selective attack of the less noble Al-rich α-phase, and nitric acid to brighten the surface [16]. The exact composition of Kroll's reagent is given in Tab. 1. Fig. 6 is a LM image of a lasermolten Ti-6Al-4V sample after ten-second Kroll etching. For the purpose of representing this 3D cube, three metallographic  Kroll-Ätzung. Zur Darstellung dieses 3D-Würfels wurden drei metallographische Schliffe angefertigt, wobei die Z-Achse die BD im SLM-Prozess repräsentiert. Durch diese dreidimensionale Darstellung kann ein Überblick über verschiedene Bereiche des additiv gefertigten Bauteils geschaffen werden. Neben der kolumnaren Kornstruktur (gestrichelte Linien) im X-Z-Querschnitt ist auch die β-Kornstruktur im X-Y-Querschnitt zu erkennen.
Another way of representing the microstructure is Weck's etching [27]. Figs. 7a-c) show the results of etching by immersion according to Weck. During the test series, it was found that an optimal microstructural contrast is achieved after a period of 35 seconds. Fig. 7a) shows the LM image (BF mode) of an as-built sample. This magnification reveals the layer structure (marked by black arrows). Owing to their different coloration, columnar grain structures (dashed lines) can also be observed normal to the BD. Fig. 7b) is a light microscope image of the same image detail at a higher magnification. It is apparent that this etching method provides an excellent contrast revealing the martensitic α´ needles.
In order to compensate the microstructural defects induced by the manufacturing process, components are subjected to a heattreatment or to HIP [28]. HIP reduces the porosity of the components and increases their density. Fig. 7c) shows the structure after Weck's etching of a heat-treated sample (1 -3 hours at 800 -850 °C). Compared to the as-built sample, the needles appear broader, indicating a transformation of the martensite towards the thermodynamic equilibrium (α´→α+β). The gray and black appearance of the α´ needles in Fig. 7c) indicates, moreover, that the needles belong to the former mother phase, the β phase.

Thermal Etching
Another possibility to visualize the microstructure is a thermal etching process. During thermal etching, a physical etching method, an oxide layer is generated on the surface by elevated temperatures. In the LM, this layer appears in colors varying as a function of the present phases and their respective crystallographic orientation. The colors vary from brown (for shorter etching times) to red and blue (for longer etching times) [29]. Fig. 8 shows the result of a 120 min thermal etching process at 540 °C in air atmosphere. The image detail clearly shows the α needles. Given that elevated temperatures are required to generate the oxide layers, it must be noted at this point that thermal wendung einer thermischen Ätzung mikrostrukturelle Veränderungen auftreten können, da erhöhte Temperaturen zur Erzeugung der Oxidschichten notwendig sind.

Electrolytic Preparation
Owing to a an electric current flow between the sample (anode) and the counter electrode (cathode) and the use of an electrolyte, the surface is anodically dissolved during the electrolytic preparation. In addition, the potential difference results in precipitations and a varying surface layer thickness caused by the electrolyte [26,30]. Fig. 9 shows the results of an electrolytic sample preparation using the electrolyte Ti Em3 on an as-built component. Owing to the varying surface layer coverage, the columnar β grain structure is particularly well contrasted applying a voltage of 100 V and a polishing time of 10 s. As shown in Fig. 9a), it is also possible to capture a large number of β grains, thus subsequently allowing a quantitative evaluation. The higher magnification of Fig. 9b) furthermore reveals the martensitic α´ needles, marked by arrows. They appear considerably brighter.
As opposed to anodic etching, where a surface layer is produced, a surface layer formation must be suppressed in case of a  preparation intended for electron backscatter diffraction, as the layer acts passively and impedes an emission of electrons from the surface. When samples are prepared for EBSD measurements, it should therefore be ensured that no surface layers are present on the surface and that the material removal process leaves a deformation-free surface [15]. Polishing parameters such as voltage, flow rate, and polishing time, were therefore varied aiming at improving the preparation result. Providing constant environmental conditions in the SEM such as sample type, accelerating voltage, sample current, and magnification, an average Confidence Index, CI, was determined for the respective EBSD measured area using the evaluation software OIM. The CI indicates the level of accuracy of crystallographic orientation indexing. In this work, this value is used as a reference value to evaluate the preparation outcome. Given

EBSD-Untersuchungen
Bilder 10 a) und b) zeigen die Kornorientierungs-und Qualitätsindex-(engl.: Image-Quality, IQ) Abbildung einer EBSD-Messung an einer as-built Probe nach einer Vibrationspolitur von 8 Stunden. Die Z-Achse des EBSD-Koordinatensystems ist parallel zu BD. Es sind deutlich die α´-Nadeln der martensitischen Mikrostruktur und deren kristallographische Orientierungen zu erkennen. Im Vergleich dazu the same environmental conditions, higher CI values indicate a better sample surface quality. The preparation studies have shown that the best surface quality could be obtained using the electrolyte A3 at a voltage of 30 V and a flow rate of 10 for a polishing time of 40 s. The highest CI values were reached based on these parameters for both, the as-built samples and the heat-treated samples.

Vibratory Polish
Apart from the described electrolytic preparation methods, it is also possible to prepare the surface of a sample for an EBSD measurement by subjecting it to a vibratory polish. However, in this case, the polishing rate is very low and the polishing process generally takes several hours [16]. For vibratory polishing, the mounted samples are clamped into a sample holder, thus providing the required contact pressure on the polishing disk. The holder is subsequently placed on a polishing cloth moistened with a polishing suspension and moved over the cloth by vibrations. For the vibratory polish, a chemically resistant MicroCloth polishing cloth (Buehler) was used in combination with an OPS. It was found that the highest CI index for as-built samples was obtained after a polishing time of 8 hours. Owing to their lower hardness, the vibratory polishing time was reduced to 4 hours for the heattreated Ti-6Al-4V samples.

EBSD Examinations
Figs. 10 a) and b) show the grain orientation and the Image-Quality map (IQ) of an EBSD measurement on an as-built sample after 8 hours of vibratory polishing, respectively. The Z axis of the EBSD coordinate system runs parallel to the BD. The martensitic microstructure's α´ needles and their crystallographic orientations can clearly be recognized, whereas, due to the heat-treat-erscheint die Mikrostruktur durch die Wärmebehandlung in den Bildern 10 c) und d) deutlich gröber, was auf ein Kornwachstum aufgrund der verwendeten Temperatur schließen lässt. Die dunklen Bereiche in den IQ-Abbildungen sind Korngrenzen, an denen die CI-Werte, im Gegensatz zum Korninneren, gering sind. Der durch die Wärmebehandlung entstandene geringe β-Phasenanteil (< 5 vol.%) befindet sich an den α-Korngrenzen und kann aufgrund des geringen Anteils und einer zu geringen Ortsauflösung der EBSD Messung bildlich nicht erfasst werden.
Durch die optimierte Probenpräparation ist es nun möglich, auch EBSD-Messungen an größeren Bereichen durchzuführen. Bild 11 zeigt das Ergebnis einer EBSD-Messung einer wärmebehandelten Probe nach einer Vi-ment, the microstructure in the figs. 10 c) and d) appears considerably coarser suggesting a temperature-induced grain growth. As opposed to the grain interior presenting higher values, the areas in the IQ images appearing darker are grain boundaries with low CI values. The heat treatment induced low β phase fraction (< 5 vol.%) is located at the α grain boundaries. As the fraction is too low and the spatial resolution of the EBSD measurement is insufficient, it cannot be imaged.

Schlussfolgerungen
In dieser Arbeit wurde grundlegend die Probenpräparation für additiv gefertigte Ti-6Al-4V-Bauteile untersucht. Hierfür wurde die gesamte Prozesskette, beginnend bei der Präparation von Pulverproben bis hin zur Probenvorbereitung von SLM-gefertigten Bauteilen, beleuchtet. Eine optimierte Probenpräparation ermöglicht, neben den tory polishing. As is apparent in Fig. 11a), the columnar β grain structure could be reconstructed based on the so-called Burger's orientation relationship using the software ARPGE [10,11]. b) presents a superposition of the ARPGE file and the original file. This superposition reveals the β grain structure in the EBSD grain orientation image. In this way, not only an image detail can be examined. Crystallographic information of individual β grains can also be obtained by singling out and cutting along grain boundaries.

Conclusions
In this work, the sample preparation process for additively manufactured Ti-6Al-4V components was fundamentally analyzed. With this in mind, the entire process chain, from the preparation of powder samples to the sample preparation of SLM manufactured components, was outlined. An optimized sample preparation not just allows stand-Figs. 11 a and b: a) β grain structure of a heat-treated Ti-6Al-4V sample with Y orientation calculated using the software ARPGE [10,11]; b) superposition of the calculated β grain structure and the respective EBSD grain orientation image indicating the respective orientation legends. [10,11] berechnete β-Kornstruktur einer wärmebehandelten Ti-6Al-4V-Probe mit Y-Ausrichtung; b) Überlagerung der berechneten β-Kornstruktur mit der zugehörigen EBSD-Kornorientierungsabbildung und der entsprechenden Orientierungslegenden.
ard examinations by LM and SEM. It also permits further crystallography investigations by EBSD of additively manufactured components. The preparations carried out on Ti-6Al-4V powder samples and AM components defined in this work lead to the following findings: • During mounting, the hardness difference between a SLM manufactured solid material sample and the mounting material should be minimal in order to ensure an optimal preparation including the surface layer.
• The parameters listed in Tab. 2 allowed the manufacture of reproducible sections providing an excellent surface quality -a requirement and thus a precondition for a variety of examination methods.
• The β grain structures formed during the SLM process based on which the microstructural evolution can be interpreted can be contrasted by selective etching or electrolytic preparation methods. Another possibility to reveal the β grains is using a polarization filter in the LM.
• Specific etching processes, such as etching by immersion according to Weck, do not only reveal the microstructure. They also visualize the layer structure and the molten pool boundaries in the SLM manufactured components.
• High-resolution EBSD analyses presuppose an artifact and a deformation-free surface. Not only electrolytic polishing, but also vibratory polishing can be carried out to ensure a sufficient surface quality.
• Superposing numerically calculated β grain structures and EBSD measurement data allows a targeted examination of the crystallography of individual β grains.