Accepted Practice for Metallographic Preparation of Thermal Spray Coating Samples

Since thermal spray coatings are formed by stacking of flattened splats, the coating external surface (the top view) and its cross-section (the view perpendicular to the coating’s surface) will take on different microstructural profiles.^[2] Thus, observations of thermal spray coating microstructures both at the top surface and on the cross-section are required to reveal the anisotropic nature of the coating. The observation of a coating surface can typically be performed without additional preparation; however, analysis of the cross-section requires the sample to undergo metallographic preparation.

The aim of metallographic preparation is to reveal an undisturbed thermal spray microstructure for analytical or microscopy work, while minimizing the amount of damage introduced to the coated sample.^[3] The preparation process presented here was adapted from the ASTM E1920, Standard Guide for Metallographic Preparation of Thermal Sprayed Coatings (2014)^[1] that specifically covers the preparation and examination of thermal spray coatings.

The test specimen must be sectioned to obtain the cross-section profile of a coating. Sectioning is a destructive process that, unless care is taken, is likely to cause damage to the thermal spray coatings before observation. Thus, prior to any sectioning procedures, if a coating is deemed to be brittle, highly porous, friable, or otherwise particularly susceptible to damage, the entire witness coupon may be impregnated with a low viscosity resin under vacuum. For later observation of pore and void features, a fluorescent dye can be added to the transparent epoxy, which adds contrast under UV illumination, effectively highlighting those features for optical observation. The vacuum chamber should be held at 80 kPa for between two to ten minutes to allow the epoxy to penetrate and occupy the pore and crack network of the thermal spray coated specimen. Longer durations and multiple vacuum applications may be required for thicker coatings.

Upon hardening, the epoxy resin helps to support the microstructure against harsh cutting procedure, and so may reduce or prevent propagation or enlargement of existing microstructural artifacts. The fluorescent dye, meanwhile, is visible under UV light and so assessment of the coating can be performed to distinguish inherent pores and equivalent features from any subsequent metallographic preparation-induced features. An example of a resin-impregnated coupon is shown prior to sectioning in Fig. 1.

It should be noted that while this pre-sectioning impregnation process is recommended as a preliminary step for the preparation of fragile coatings, it may not be necessary nor desirable for denser, stronger coatings. The limitation of this method is primarily time based, as the curing time for a two-part epoxy resin may range from anywhere between 0.5 to over 8 hours, slowing analysis in areas where a rapid turn-around is desired (e.g., quality control processes). Therefore, the materials scientist or engineer should consider carefully which practice is appropriate for their desired application.

To prepare a sample cross-section for examination, the sample must undergo sectioning, or cutting, to reduce the specimen to a size suitable for mounting and subsequent metallographic preparation of the surface of interest. At this point, the metallographer must choose the desired area for analysis. Typically, a section near the center of the test coupon is selected for analysis, with the polished face therefore lying along the sectioning plane, as shown in Fig. 2a. This has the benefit of avoiding edge effects arising from the interaction of the sprayed material with the coupon holder, but it means that care must be taken to avoid damage during the sectioning process. Conversely, some metallographers may prefer to mount specimens such that the free edge is exposed, as in Fig. 2c. While this reduces impact of the chosen sectioning method, it can be extremely difficult to identify artifacts arising due to edge effects.

The end goal is to obtain a planar (flat) surface with as little deformation/damage as possible introduced, to help minimize the duration of surface grinding operations required later in the sample’s preparation. A simple and effective technique for sample sectioning is to use an abrasive wet cutting procedure, with care given to select cutting parameters to minimize sample damage. The abrasive wet cutting process may typically be conducted using a rotating cut-off wheel in which the impregnated specimen is first clamped firmly in the machine grips. As there is a substantial force exerted by the cut-off blade through the specimen, rigid clamping on the coating sample is necessary to avoid sample walking and coating damage that may be triggered by sample vibration. However, care should be taken in clamping the specimen, as high clamping forces may cause cracking of the coating. This can be a particular issue with mismatches in the coefficient of thermal expansion between the substrate and coating, where residual stresses may produce a curved sample. The use of customized sample holders may reduce sample deformation and thus coating damage in clamping these cases. Alternatively, rubber pads or flat wooden sticks may be used to cushion the coating from direct contact with the metal clamps, as in Fig. 3.

The clamping orientation of the sample with respect to the rotation direction of the cut-off wheel is an important consideration when sectioning thermal spray coupons. The primary concern is to keep the coating in compression to prevent coating delamination and/or chipping. Therefore, the specimen orientation should be such that sectioning commences through the outer layer of the coating and the cut-off wheel descends into the substrate. The direction of cut-off wheel rotation (either clockwise or anticlockwise) must also be selected so that the abrasive wheel engages the coating in an inward cutting motion to exert a compressive cutting stress. Failure to correctly orient the sample will lead to tensile forces on the coating system, which may cause delamination of the coating and separation at the coating/substrate interface and/or at interlayer(s). As delamination during sectioning may be mistaken for poor adhesion during coating, specimens cut under tension must be excluded. Figure 4 shows a schematic of the correct orientation between the sectioning sample and cut-off wheel. Another point to be mentioned about clamping orientation is to ensure that the direction of cut is (as close as possible to the) perpendicular to the coated surface. Not doing so, may lead to poor micrographs and misleading results such as overestimating the thickness measurements.

Following correct alignment of the sample, sectioning is performed by translation of self-sharpening wheel across the sample while maintaining a fixed speed of rotation. A diamond impregnated wheel, SiC, AlO_x or cubic boron nitride (CBN) wheel can be used for sectioning. Wheel type, thickness, and cutting speed needs to be customized for coating type and individual use. A thicker, slow cutting wheel can impart greater damage as its deformation zone is greater and contact time higher. While a thinner wheel with faster speed has the possibility of wheel breakage, which may lead to coating damage. A cooling liquid, usually water, is used to flush the wheel to avoid damaging the sample and cutting wheel due to frictional heat. The coolant also removes debris from the cutting area. It is recommended to avoid oil-based coolants because residual oil compounds may penetrate the coating and are difficult to remove. There is a substantial force exerted by the cut-off blade through the specimen.

Even if no damage is produced through clamping, sectioning of the coupon will still cause alteration of the thermal spray coating structure adjacent to the cut. This includes localized cracking, spallation, delamination, separation, and other false indicators/artifacts in the coating structure. Proper cutting speed and selection of wheel composition and other properties should be considered to avoid these gross defects, and produce a straight and flat cut, which will reduce planar grinding time and steps. Even best cutting surface leaves grooves, tear out, and other surface damage. Subsequently, any evaluation of the coating microstructure must be carried out at a point beyond the plane of induced damage. To remove sectioning damage, planar grinding of the sectioned surface of the coating is required.

The minimum planar grinding depth required after mounting depends on the sectioning parameters. Namely, uniform sectioning conditions can bring about a high surface quality and eliminate the need for lengthy planar grinding. Further, a high wheel speed and high feed rate will apply greater pressure to the surrounding material, and thus damage more of the coating. While high speed cutting may be desired for reducing sample size, this should be avoided near features of interest, and a low-speed method with automated feed control is preferred for final cuts.

The choice of cut-off wheel also influences the amount of sectioning damage induced into the coating sample. Assuming optimum parameters, abrasive machining with a diamond cut-off wheel can produce samples with less than 100 μm sectioning damage. However, if a silicon carbide cut-off wheel is used, a minimum planar grinding depth of at least 1.5 times the thickness of blade is necessary to reveal the undisturbed microstructure.

An abrasive cut-off blade capable of cutting the substrate effectively should be used for the combination of thermal spray coating and substrate, as the substrate thickness would be much higher than the thermal spray coating. A low- speed diamond cut-off wheel is the most suitable for processing of thermal spray coatings because it provides a clean cut with minimal sectioning damage to the coating. The typical grinding step required after sectioning with a low-speed diamond sectioning machine is less than 100 μm. A diamond cut-off wheel is made of metal-bonded diamond particles. The polycrystalline diamond particles are responsible for the effective microcutting of the coating specimen, and the wheel must be “dressed” periodically by a ceramic dressing stick to expose fresh diamond particles. The thickness of diamond cut-off wheels can be less than 300 μm; therefore, there will be less material loss, reduced coating deformation and less heat generated during sectioning. Subsequent grinding steps after the use of non-diamond wheel sectioning can require removal of 1300 – 2000 μm. So, it can significantly increase processing time and cost, and for smaller specimens there might not be enough coating left.

Other suitable cut-off wheels include those made of resin-bonded silicon carbide or aluminum oxide, where the soft resin will break down and supply fresh cutting particles to section the sample. Examples of these, along with a metal-bonded diamond blade, are shown in Fig. 5. Blade choice will depend on the hardness and ductility of the substrate, but the damage to the coating is typically greater due to the blade thickness. It is important to select the correct density (also referred to as hardness) of the wheel, as it affects the self-dressing attributes of the wheel as it breaks down during cutting. Failure to cut the coating material can cause not only surface damage of the face being cut, but cracks and chips well into the coating surface that could show up in microstructural analysis even after significant material removal during grinding. In some cases, introduced coating delamination and interface separation cannot be removed by subsequent steps. The wear of the cut-off disc increases with the hardness of the material being sectioned. Thus, to provide an effective cutting action and longer useful lifespan, the thickness of the silicon carbide blade is usually more than 0.5 mm. The use of thicker silicon carbide blades thus implies that the sectioning damage to the thermal spray coating will be greater.

Due to their superior cutting ability, it is recommended that samples be sectioned using a low-speed diamond cut-off wheel machine; blade speed, sample feed rate, and degree of coolant flow must be chosen to minimize frictional heating and excessive cutting pressure that could cause “burning” on the cut face, spallation of the coating, or wheel breakage. Exact parameters may be varied with operator experience, but selection of these parameters is advised to start in accordance with cut-off wheel manufacturer recommendations for substrate hardness and ductility. In certain cases, where the sample size is too large to be processed with a low-speed diamond saw, a larger sectioning machine equipped with less suitable blades may be used to reduce the sample size, but a low-speed diamond wheel should be used for final sectioning where possible.

Improper dressing, cutting speeds of wafer diamonds and other wheel types can cause the blade to wander. Flatness of the cut, especially of significantly hard materials such as carbide composites will significantly affect the grinding procedure and time, so flatter specimens are also goal of sectioning the coating.

Following sectioning, it is necessary to mount the specimens into round resin molds, to facilitate both manual and/or machine gripping of samples in the ensuing steps of grinding and polishing. Due to the greater control and reliability of automated polishing, it is recommended to choose a compatible mounting geometry; typically, a 25 mm (1 to 1-1/4 in.) diameter circular mount may be used (see Fig. 6), although other geometries may be desired depending on specimen size and available machine polishing plates.

Although hot compression mounting is both a time effective and economical method, it is typically preferable to do cold mounting of thermal spray coatings, as the heat and/or pressure exerted during hot mounting could affect the coating integrity, and so substantially change the observed morphology by, for example, collapsing coating porosity. An exception is high density, non-friable coatings which exhibit high strength such that the compression during mounting does not change the coating properties. Such coatings are applied by high velocity processes, such as HVOF, HVAF, and cold spray. The advantage of hot compression mounting materials is they provide good edge retention to the specimen and contain hard particles that reduce the relief between a mount material and specimen, which helps maintain a flat surface of the coating edges. Depending upon subsequent processes, a thermosetting or thermoplastic resin can be used for hot mounting. While using hot compression mounting, users need to ensure that the panel substrate are strong and thick enough, so that it does not deform during compression. Typically, a minimum of 2 mm thick substrate is suggested.

Before insertion into the cold mounting cups or placing on hot mount stage, the sectioned samples should be cleaned with an appropriate cleaning agent; for water insensitive materials, an effective method is cleaning with soap and water using cotton swabs, then rinsing with ethanol. The purpose of cleaning and rinsing the sectioned samples was to remove dirt or cutting fluids trapped within the pores and crack network. Excess moisture within the coating may be removed with a hot air dryer. The potential fragile nature of some thermal spray coatings precludes ultrasonic cleaning techniques; such cleaning might propagate the existing crack network or cause cavitation damage within the coating microstructure.

The specimen samples should be placed into the cold mounting cups or onto the hot mounting stage with the sectioned face placed downwards. If the sample is too thin to support itself, it may be necessary to use a plastic clip or similar support to keep the sample vertical, perpendicular to the final mount face that will later be ground and polished. For cold mounting, it is recommended to use two sectioned pieces of the sample to a single mount, positioned with coating facing inwards; this provides greater coating edge retention, and ensures a flatter and more uniform surface finish due to a more even distribution of the harder sample within the softer mounting resin. An example of a typical specimen configuration prior to mounting is shown in Fig. 7.

For cold mounting samples, similar resin choices, including incorporation of a fluorescent dye, may be used as in the preliminary resin infiltration step, with use of a vacuum chamber again recommended to eliminate porosity in the resin. Furthermore, vacuum impregnation can be used for cold mounting friable, porous abradable, and non-metallic coatings. Low viscosity epoxies are recommended for vacuum impregnation. It is vital that curing be undertaken on a flat surface, to ensure that the back of the mounted sample is flat in order to avoid uneven polishing when gripped by the automatic polisher, and to avoid the shifting of the specimens within the mounting cups.

Common issues to be wary of during mounting are setting clips too high on the sample, and/or knocking the resin mounts as they set, which can lead to the specimens falling over within the resin, as in Fig. 8. This can occur for both cold mounted and hot mounted samples and often leads to difficulty in polishing the material, and in the extreme case may render the sample unusable.

Planar and fine grinding play a major role in the preparation process and, therefore, these will be discussed in the following section. Planar grinding serves two primary purposes; first leveling the specimen mount surface such that whole cross section is on same focal plane, and subsequently removing excess resin and gross sectioning damage. The process works as micromachining to remove deformed material in subsequent steps to reach virgin, artifact-free, as-sprayed coating.

Subsequent grinding steps, called fine grinding, involve applying finer abrasives to remove the scratch and deformation artifacts caused by the previous grinding step. Fine grinding is thus the process used for establishing a specimen surface suited for the first polishing step from the relatively rough surface formed from sectioning or plane grinding. These intermediate fine grinding steps are necessary because the material removal is still relatively high, as contrasted to the low material removal during polishing.

Regardless of which of grinding or polishing is being performed, it is important to consider the way samples are manipulated during material removal operations. While specimens can be polished individually, with proper style of sample holder, the comparatively narrow face of the sample mount can hinder efforts to achieve a flat surface. Instead, it is recommended practice to polish multiple samples concurrently whenever possible, using a multi-specimen holder. The number of specimens to a holder will vary depending on the chosen sample size, instrument manufacturer, and instrument model, but a common geometry is 3, 4, or 6 equidistant specimens. These may further be shaped to accommodate a variety of specimen geometries, as required. Examples of these holders are given in Fig. 9.

A holder may be used with less than its full complement of specimens; however, it is recommended to space samples in an axially symmetric manner around the holder, as shown in Fig. 10. Depending on the instrument, samples may either be restrained only during the polishing process, or rigidly affixed to the sample holder via screws or clamps. In the latter case, it is important not to add, remove or otherwise adjust the position of samples wherever possible following the planar grinding steps, as this will change the loading across the holder and may lead to uneven grinding and polishing of the specimen mounts.

In the past, planar grinding has traditionally been accomplished using coarse silicon carbide (SiC) papers, at least for softer materials; harder materials, particularly those containing carbide and equivalent phases, have necessitated diamond disks and slurries. The common sequence of using SiC papers is P120, P180, P240, P320, P400, P600, P800, P1200 grit with the machine thoroughly cleaned during each paper changeover if dedicated machines are not available. The grinding duration for each paper should be around 60 to 90 seconds and is water cooled (i) to prevent frictional heating that might damage the coating, as well as (ii) to remove grinding debris that can quickly load or dull the paper. The rotation of the grinding disc should not exceed 300 rpm and samples held with a force no more than 50 N per sample. An example of the traditional preparation method used for ceramic; thermally sprayed specimens is given in Table 1. The speed and load combination prevented a smearing effect in the coating, which results in filling of voids within the coating by grinding debris or pulling out less-bonded splats or phases. Quality of post polished coatings may depend on parameters other than the ones listed in Table 1, such as coating process, coating parameters, and quality of consumables; thus, readers are recommended to use these parameters a as starting point and develop their own grinding and polishing procedures.

However, more recently, the recommended practice has shifted to deemphasize the use of multiple papers. Although a single initial planar grinding step on a SiC foil may be advised for certain materials, contemporary planar grinding methods increasingly use a grinding disc made of resin-bonded diamond particles with a moderately coarse grit for this initial stage. These discs are designed for repeated instead of single use, unlike the silicon carbide papers that have a useful lifespan of about 90 seconds. The disks have a relatively high, constant removal rate that is due to the high hardness of the diamonds and the ability of the bond to break down, releasing fresh abrasive grains. The diamond particles, being of a more regular shape and with a closer tolerance of the grain size than silicon carbide particles, will also produce less deformation. These fixed diamond disks, however, may need dressing to relieve the diamond depending on the coating and substrate being ground. The edge retention is also good, as the combination of a rigid resin bond and hard diamond abrasive is able to remove material from very hard phases in the specimen surface, thereby avoiding relief effects that result from preferential removal rates. However, users should be aware of using the disc across various coatings families as the end life of the disc depends on the frequency of use and coating/substrate hardness. Flatness of used disks is also a factor when disks have been used inconsistently. The path which the specimens rotate on the disks should cover the disk as much as possible to improve retention flatness of disk and sample being polished.

Beyond resin-bonded diamond discs, lapping methods are also becoming popular for planar grinding; in this method, the sample is moved against a hard lapping surface to which an abrasive suspension (typically diamond) is introduced. Here, abrasion is produced by two mechanisms: in one, the abrasive particles may embed into the hard lapping surface, and so act as fixed points ploughing into the sample surface. Alternatively, free rolling particles may puncture the surface as they are pulled between the lap and sample surface. The ‘fixed’ particles will typically pull free under high loads, transitioning to the less aggressive, more even rolling mechanism; as such, lapping is notable for providing a smooth surface with excellent edge retention, and in particular may show better results for softer specimens than rigid discs.

A contemporary metallographic preparation method using fixed diamond bonded discs for ceramic, thermally sprayed specimens is provided in Table 2. Similar methods applicable for thermally sprayed coatings of soft metals and composites/hard metals are given in Table 3 and Table 4, respectively.

Most importantly, the advantage of using the contemporary grinding method is that the entire grinding sequence can take place in one operation. This avoids the need for time consuming and costly sequential changeover of silicon carbide papers. Only a surface dressing is needed if the surface of the disk is clogged. In the authors’ experience, grinding using a P220 grit, resin-bonded diamond disc provides an ideal balance of material removal with final surface finish. This contemporary preparation procedure is particularly preferred for thermal spray coatings, since grinding with a bonded diamond grinding disk causes less smearing and reduces pull-outs and cavities. These artifacts, which can be mistakenly considered to be pores belonging to the true coating microstructure, must be removed during the final fine grinding step on the rigid composite disks.

Regardless of the grinding media and method used, the ensuing step is a final fine grinding step on rigid composite disks (RCD) or very hard polishing cloths along with a polycrystalline diamond suspension. The RCD replaces several grinding steps on silicon carbide papers by one single step. Diamond suspensions are available with both polycrystalline and monocrystalline diamonds. The polycrystalline diamond suspension is preferred because it can achieve a high removal rate while maintaining a quality surface finish. This is due to the fact that the polycrystalline grains break down during the fine grinding process, creating new cutting edges. Monocrystalline grains are stronger, having a blocky form with relatively few cutting edges, and will not easily break down. Therefore, although monocrystalline diamond suspensions are normally less expensive than polycrystalline, their removal rate is known to be much lower.

The surface of the RCD consists of a resin with mixed-in abrasive powders of varying grain sizes. Together with the 9 μm polycrystalline diamond suspension that should be continuously added, a consistent material removal rate can be achieved, producing flat specimens with little deformation and good edge retention. After only one fine grinding step on a suitable RCD, the specimens are primed for polishing. Example micrographs of a thermally sprayed coating after planar grinding (P220) and fine grinding (9 μm diamond suspension) are shown in Fig. 11.

Grinding steps plays a crucial role in revealing the true microstructure of a coating. Not all consumables from suppliers are the same, even though they are graded the same. Figures 12 (a) and (b) show the difference in end microstructure after grinding through a series of grinding papers obtained from Supplier A and Supplier B. Both samples were ground through a series of grinding paper that had abrasive particle size starting from 82 μm down to 5 μm, other grinding parameters were unchanged. Figure 12(a) shows significantly higher number of deeper scratches compared to the one shown in Fig. 12(b).

Polishing is the final metallographic step to obtain the cross-sectional microstructure of thermal spray coatings. The polishing process must remove any smearing or deformation that has taken place in previous preparation steps but must also retain phases inherent to the coating structure. Figure 13(a) shows examples of grinding and polishing cloths.

The key difference of the polishing cloths used compared to those cloths in the grinding steps is their resilience. Resilience is the cloth elasticity in the vertical direction and dictates whether a cloth is hard or soft. All cloths can be compressed when subject to pressure from the specimen. A cloth having a low resilience is hard and a cloth with a high resilience is soft. A hard cloth, which can only be slightly compressed, will usually give high material removal, and it will create deep scratches and more damage to the specimen than a soft cloth. When paired with fine abrasives such as 3 μm polycrystalline diamond and 0.05 μm colloidal silica suspensions, it is therefore possible to obtain a surface that is truer to the microstructure using a softer cloth. When conducting the final polishing step for a ceramic thermal spray coating, spherical silica is less abrasive than diamond, which produces a smoother surface. Most colloidal silica suspensions have a carrier liquid with a pH in the range of 8.5 to 11. This pH level alongside the fine grain size, will create a combined low level of mechanical and chemical material removal.

Thus, a two-step polishing procedure is recommended for all samples. First, a medium resilience satin woven acetate cloth should be paired with a 3 μm diamond suspension, see Fig. 13(b). The second polishing step uses a high resilience, porous neoprene cloth, Fig. 13(c), and a 0.05 μm colloidal silica suspension. These colloidal suspensions can typically be purchased in formulations that span a range of chemical activity; modifications can also be made by the metallographer by adding reagents such as H₂O₂ to the suspension. As a rule, the specific suspension used should be chosen with consideration of the specimen material. Harsher suspensions are appropriate for ceramics and hard, homogenous metals; however, for more heterogenous materials, such as composites or carbon/low-alloy steels, a less aggressive polishing suspension, such as gamma alumina, may be preferred.

The disc rotation speed, process time and specimen loading required for the respective grinding and polishing steps for various classes of material are summarized in Table 2 to Table 4. Grinding and polishing should be performed with a co-rotating stage and head for all stages bar the final oxide polish, for which counter-rotation is preferred. Generally, the polishing time and load are lower than the equivalent grinding parameters to minimize the possibility of any mechanical deformation of the surface. Samples should be cleaned after each step with an appropriate agent. For example, if using a glycol-based polishing suspension, soap and water is recommended to clear the residue, while cleaning mounts with running cold water and wiping specimen with cotton pads may be suitable for removing water-based suspensions residue. In either case, a clean (not reused) soft brush may be used to assist in clearing debris, followed by rinsing with alcohol. Finally, the sample may then be wiped down with cotton wool pad and dried using warm air dryer; this rapid drying minimizes the formation of drying streaks and stains.

Inspection of the surface at each stage of grinding and polishing is vital and ensures that previous deformation artifacts are completely removed before proceeding. Example micrographs of a thermally sprayed coating after 3 μm (diamond) and 0.05 μm (oxide) polishing steps are shown in Fig. 14.

After being ground and polished, the various sectioned and mounted thermal spray coating samples are ready for characterization experiments, such as microscopy and indentation hardness tests. For particularly sensitive applications, such as electron back-scatter diffraction (EBSD) analysis, additional polishing steps such as vibratory polishing may also be desirable, as this can remove the remaining minor blemishes with the minimum of applied forces. A vibratory polisher provides a mellow polishing action that produces a stress-free surface by oscillating horizontally (typically 7200 cycles/minute) to increase the contact time between samples and the polishing cloth. As the result, surfaces with excellent flatness and minimal deformation can be achieved. It is unusual to chemically etch or use other hazardous or difficult methods, such as electro-polishing and typically not recommended.

Errors in metallographic preparation of thermal spray coatings may lead to observation of a number of common issues during subsequent specimen analysis. To aid in the identification and resolution of these issues, common issues are presented below, along with causes and mitigation strategies.

Edge rounding is a phenomenon where the edge of a mounted sample will be worn to a greater extent than the rest of the sample, appearing rounded under the optical microscope. This can occur for a number of reasons, most frequently when the mounted sample has a slower wear rate than that of the mounting resin; as a result, a vertical relief is formed between the sample and the resin, with the edge of the sample undergoing greater wear. Edge rounding may also occur when the mounting material does not torch or bond well to the sample, leaving a gap between sample and resin. An example of edge rounding can be seen in Fig. 15a, with good edge retention depicted in Fig. 15b for comparison. Edge rounding can also occur at voids in coatings. This typically occurs when a nap cloth is used during the polishing step, which can cause relief in coatings but also catch edges and round them. The use of DIC (Differential Image Contrast) also referred to Nomarski microscopy to aid in capturing rounding of edges and other artifacts that can influence the coating interpretation.

Typical causes of edge rounding include: (i), incorrect selection of mounting resin or technique, causing poor adhesion between the sample and the resin; (ii), a polishing cloth with an overly high resilience; (iii), an overlong polishing time; and (iv), excessively high forces that accelerate wear of the resin. To address (i), it is important to choose the right resin and right mixing ratio between the resin and the hardener to create good adhesion between sample and resin. To resolve (ii), where possible, the selection of polishing discs with lower resilience should be prioritized, along with selection of lubricants with appropriate viscosity. If using high resilience cloths during fine polishing, higher viscosity lubricants are recommended. In the case of low-resilience cloths, and for softer materials, it is better to use lower viscosity lubricants to mitigate the risk of edge rounding. For (iii), it is advised to reduce the polishing time, and frequently check the sample surface via microscope examination. Meanwhile, (iv) may be addressed through use of relatively low forces during the polishing process.

Pull-out refers to a type of flaw or irregularity that may occur during various sample preparation steps, including sectioning, mounting, and coarse grinding. Pull-out phenomena may be characterized by issues such as the loss of elemental structure, holes from the breaking-down or removal of inclusions (i.e., oxides and carbide particles), cavities or pits left in the surface from dissolution of water-sensitive phases, and damage remaining from the previous coarse grinding. Figure 16 shows an example of pull-out of inclusions, with scratches formed by these inclusions.

To prevent pull-out from happening, it is important not to induce too much stress during the sectioning and mounting steps. This may be achieved by use of appropriate cutting wheels, ideally a metal-bonded diamond wheel; these provide rapid and effective material removal with less force and heating generated in the surrounding material, which may limit damage. A preliminary mounting step using an epoxy resin applied under vacuum may further support the retention of microstructural features; incorporation of a fluorescent dye into this resin can also aid in distinguishing natural porosity from pull-out artifacts. During the polishing steps, it is better to use less aggressive polishing disks, with a finer abrasive, to prevent pull-out. In addition, a low force is recommended for plane grinding and fine grinding. Finally, it remains vital to check the sample at every polishing step and ensure that the damage from the previous step has been eliminated; this mitigates the risk of introducing potential damage in the later steps.

Smearing is a form of plastic deformation where, rather than being removed via abrasion, the sample material is forced across the surface; in effect, the action of the abrasive agent appears “blunted.” This may obscure small voids and similar microstructural features, hindering determination of coating quality. An example of smearing can be seen in Fig. 17, the apparent porosity in the smeared sample can be seen to be substantially less than that visible following removal of the smeared layer. Although smearing can be removed via extended diamond polishing times, this both delays sample preparation, and risks edge rounding of the sample. Smearing is most common in softer coating materials and substrates but may affect all materials depending on process. Using incorrect polishing cloths, improper lubricant volumes, or inappropriate abrasives may all induce smearing. These issues can be addressed as follows: the polishing cloth may be substituted for a lower resilience, which minimizes the depth a given abrasive particle sinks into the cloth; the abrasive grain size may be increased, increasing the degree of cutting by each abrasive grain; and the degree of lubrication may be increased.

Scratches are grooves caused by the action of abrasive particles on the sample’s surface. Examples of scratches persisting through a) fine grinding, and b) diamond polishing, are shown in Fig. 18. Although this abrasive action is a fundamental part of the grinding and polishing process, the presence of scratches in the final finish is undesirable, as they may obscure or damage microstructural features of interest. Persistent scratches are principally a result of either the incomplete removal of deformation from an earlier grinding/polishing step, or the presence of abrasive particles of greater size than that chosen for the current grinding/polishing level. To ensure the removal of pre-existing scratches, it is important to maintain the uniformity of deformation across the whole surface during the plane grinding. This is done by targeting a consistent, flat contact between the sample and grinding surface, and repeating the plane grinding until the scratch patterns are uniform. If the observed scratch pattern is uniform across all samples, increasing the polishing time for previous steps may assist in removing the scratches. Scratches can also be a result of pullout (i.e., hard carbide grains or brittle oxides) from the coating. Soft coatings or substrates also can imbed abrasive particles that loosen during subsequent polishing step and cause scratches. Cracks in coating or separation of the sample and mounting resin, and pores in coating or substrate can release abrasive and debris from previous steps, so care must be taken to ensure thorough cleaning. In limited situations the use of ultrasonic cleaner can help to remove debris stuck in these cracks and pore network. However, it is important to note that the use of ultrasonic baths for thermal spray coatings propagate the existing crack network or cause cavitation damage and is not recommended.

Uneven or isolated scratches across an otherwise evenly ground or polished surface may also be seen, and typically arise following contamination of the grinding/polishing surface with coarser abrasive agents. To resolve this form of scratch formation, it is important to carefully clean the surface after finishing the grinding/polishing for every step to completely remove large abrasive particles or inclusions that may cause damage in the later finer polishing steps.

When interpreting micrographs of thermal spray coatings, it is helpful to be aware of the key microstructural features common to the process; to aid in this, a schematic diagram showing the cross section of a thermal spray coating is presented in Fig. 19, with key features labelled. These features include splats, voids such as pores and cracks, oxides, un-and partially melted particles, and inclusions such as contaminants/impurities.^[2]

Splats are the fundamental features of a thermal spray coating and are formed from the impact of molten particles onto a surface. Deformation and adhesion of these splats onto coating surface creates lamellar structure of thermal spray coatings.^[2,4] An ideal splat is one that is spread out, deformed, and adhered to the mating particles or surface. If the impinging particles have insufficient temperature and/or velocity on impact, particles may not undergo significant deformation, instead embedding in the coating in solid form, with similar geometry to the powder feedstock, or simply bouncing of the surface. The common phenomenon of the failure on surface originated from the splats is “spalling,” which is the detachment or flaking of particles/layers from a surface. Unmelted particles have insufficient thermal energy to deform adequately and are undesirable due to their poor adhesion strength. Excessive presence of these so-called “unmelts” is a sign of poorly optimized processes. Unmelts can be identified by the aspect ratio (horizontal vs vertical length of splat) that is close to or similar to its original feedstock morphology. This issue is particularly common with large particle size of feedstock and lower flame temperatures, but factors such as improper powder injection or insufficient dwell times may also contribute.

In addition to splats and unmelts, it is also possible to find evidence of oxidation and other contaminants. Oxidation and/or nitridation of the feedstock may occur during the heating process, as well as during its flight towards the substrate, with the oxides forming a thin film around the exterior of the particle. Upon impacting the substrate, particle splatters, and the oxide film is fractured and spread in “stringers” along the edge of the splat. These oxide stringers contribute to the hardness of the coating, and so in small amounts may improve wear properties, but in excessive amounts can render coating brittle and interfere with adhesion between splats. They typically occur in presence of reactive environments, excessive flame temperatures, and overlong dwell times, but factors such as fine particle size, coating surface and/or substrate temperatures may also influence their formation. Contamination may also come from other sources, such as impurities in the feedstock, or residues on improperly cleaned surfaces. These may have a variety of effects depending on nature of contaminant, such as hindering adhesion between splats and contaminated surfaces.

Apart from the material make-up of the coating, it is also important to consider the voids of the microstructure. These voids in thermal spray coatings typically appear as one of two sorts of coating features: porosity and cracks. Porosity is an innate property of a thermal spray coating that occurs in both open- and closed-pore configurations; the former referring to interconnected porosity extending to the sample surface, the latter to internal porosity without interconnection to the external environment. Porosity can be desirable, for instance in abradable coatings, reduces heat transfer such as thermal barrier coatings and friction reduction applications where lubricants can infiltrate the coating pores and reduces sliding resistance. However, in many situations’ pores are undesirable as it reduces the coating’s corrosion barrier to the substrate and limits its use in mechanically demanding roles, especially where wear resistance is critical. Similarly, the type of porosity desired may differ with the role, with e.g., open porosity is desirable in lubricated bearings, as it permits the diffusion of liquids such as oil through the coating, while undesirable in oxidation-resistant coatings, where it permits diffusion of oxidizing gases through to the substrate. Pores may commonly be found between splats, particularly in the presence of high amounts of unmelts, because thermal spray is a line-of-sight process, and these features may shield voids from subsequent impinging particles. In addition, depending on the feedstock and spray parameters, intra-splat porosity may also be observed.

In addition to pores, voids may also be formed through cracking. Cracks may be broadly divided into two categories: those that extend within a splat, i.e., “intra-lamellar” cracks, and those that lie along the boundaries between splats, i.e., “inter-lamellar” cracks. These may be caused by issues such as excessively high thermal strains, or the presence of voids, oxides, and contamination that provide sites of crack nucleation due to poor inter-splat cohesion and can lead to coating delamination or spallation. Cracks may also be labelled depending on their geometry, with transverse cracks appearing in the vertical direction, perpendicular to the substrate, while horizontal cracks extend parallel to the substrate, as in delamination events. Reducing and eliminating the formation of cracks is possible by minimizing thermal stresses during thermal spray, limiting the presence of pores, oxides and unmelts in the spray coating, and ensuring a clean substrate with good bonding to the spray coating material. This final parameter may be aided by using an intermediate bond coat between the substrate and spray coating, and/or roughening the surface to provide greater mechanical interlocking on subsequent coating.

The anisotropic nature of the thermal spray microstructural profile requires special attention in order to achieve the goal of revealing an undisturbed thermal spray microstructure for analytical or microscopy work. Various aspects on the metallographic preparation of thermal spray coatings, such as the cutting orientation, sample placement, mounting conditions as well as choice of grinding and polishing recipes, have been presented in this document.

Subsequently, typical thermal spray coating cross-section microstructure that is rich with various features such as splats, crack and porosity network, defects and contaminations can be revealed without additional preparation.

This Accepted Practice is intended to be used as a baseline, but it does not replace local test or laboratory instructions. Additional requirements may apply based on the available equipment, testing materials, customer requirements, and other criteria.

This document was prepared for publication in 2022 by the ASM Thermal Spray Society Accepted Practices Committee. Contributing authors:

• Duy Quang Pham, Arne Biesiekierski, Ashok Meghwal, Samuel Pinches, and Andrew S.M. Ang, Swinburne University of Technology, Melbourne, Australia

• David Lee, David Lee Consulting LLC, Indiana, USA

• Robert Miller, Inovati, California, USA

• Triratna Shrestha, Metcut Research Inc, Ohio, USA

• Atin Sharma, Siemens Energy, New Jersey, USA

• George Vander Voort, Vander Voort Consulting LLC, Illinois, USA

Unless otherwise noted, the figure images are courtesy of the Swinburne University of Technology, ARC Training Centre in Surface Engineering for Advanced Materials (SEAM), Melbourne, Australia.

Andrew S.M. Ang acknowledges the funding support from the Australian Research Council (ARC) under the Industrial Transformation Training Centre project IC180100005 that is titled “Surface Engineering for Advanced Materials”