Metallographic Polishing (MP) Techniques

• Metallographic polishing (MP) is a surface preparation process that employs mechanical and chemo-mechanical techniques to yield flat, mirror-like surfaces with controlled topography.
• It ensures microstructural integrity by effectively removing deformation layers and artifacts to support applications in microscopy, SRF niobium, nuclear graphite, and oxidation studies.
• MP is often integrated into hybrid workflows—combined with processes like ion milling and thermal treatments—to minimize defects, reduce contamination, and enhance performance metrics.

Metallographic polishing (MP) is a family of mechanical and chemo-mechanical surface-finishing methods used to create flat, mirror-like, and preparation-artifact free surfaces for microscopy, diffraction, oxidation studies, and superconducting radio-frequency (SRF) applications. Across the literature, MP may denote a conventional final polishing stage, a deliberately adapted acid-light mechanical finishing route, or a component of a hybrid preparation chain that also includes ion milling, oxidation, annealing, or rinsing. Its technical objective is material-dependent: in bulk niobium for SRF, MP is designed to recover a damage-free, chemically benign, and microstructurally sound RF surface while reducing ecological footprint, safety risks, and cost relative to HF-based electropolishing (EP) and buffered chemical polishing (BCP); in soft, porous nuclear graphite, mechanical polishing is only the first stage of a three-stage workflow because MP alone introduces artifacts that obscure submicron micro-cracks (Hryhorenko et al., 19 Sep 2025, Huang et al., 2021, Hryhorenko et al., 2023).

1. Conceptual scope and relation to neighboring preparation routes

MP is conventionally situated between coarse material-removal operations and application-specific validation. In metallography and microscopy, it follows grinding and is expected to reduce roughness, remove deformation layers, and preserve microstructural fidelity. In SRF niobium, it is being advanced by IJCLab, CEA/Irfu, HZB, and KEK as a substrate-preparation route for ultra-smooth, defect-minimized niobium and copper surfaces with reduced ecological and safety risks relative to hydrofluoric-acid-based EP, and with explicit relevance to thin-film coatings and multilayer structures designed for efficient operation at 4.2 K (Hryhorenko et al., 2023).

The relation of MP to adjacent methods is not uniform across materials. For niobium SRF surfaces, MP is contrasted with BCP and EP, which are diffusion- or etch-limited and can exhibit grain-orientation sensitivity, uneven finish, and persistent subsurface strain. For Fe–5 wt% Al, MP is contrasted with 220 grit grinding and grit-blasting; polishing and grinding preserve the primary grains, whereas grit-blasting produces pronounced grain refinement in the near-surface region and markedly alters oxidation kinetics. For Zircaloy-4 prepared for EBSD, mechanical preparation is followed by broad ion beam (BIB) milling because EBSD is sensitive to the top $\sim 20$ nm of the surface and requires removal of the mechanically induced damaged layer. For nuclear graphite, MP must be complemented by ion milling and rapid oxidation to obtain a surface confirmed by Raman spectroscopy and SEM to be free from artificial defects (Nowak et al., 2020, Fang et al., 2022, Huang et al., 2021).

A recurrent misconception is that a low roughness value alone establishes a preparation as artifact-free. The data do not support that simplification. In SRF niobium, the observed RF gains require a combination of reduced roughness amplitudes and improved microstructural integrity; topography alone is explicitly stated to be insufficient. In graphite, a visually smooth mechanically polished surface still contains broken graphite pieces, elevated Raman $I_D/I_G$, muted filler-binder contrast, and obscured micro-cracks. In oxidation studies, mirror-like polishing minimizes roughness but does not by itself determine the subsequent oxide-scale morphology; the mechanically induced near-surface state remains consequential (Hryhorenko et al., 19 Sep 2025, Huang et al., 2021, Nowak et al., 2020).

2. Polishing mechanisms and near-surface transformation

MP is often discussed in terms of micro-cutting and micro-plowing by hard abrasives, but direct electron microscopy on rough Ti–6Al–4V with controlled spherical asperities shows that much of the smoothing can proceed by plastic flow that is fluid-like at sliding asperity–abrasive contacts. After 90 s of polishing, SEM images show severe shear at the tops of asperities and redistribution of material toward their edges in the form of thin layers with a molten-like appearance. Continued polishing produces repeated stacking of such layers as asperities progressively flatten. At around 180 s, interconnected flat “islands” appear, and neighboring asperities become connected by bridges of smeared material once the edge-to-edge distance falls below approximately $30\ \mu\text{m}$. By 450 s, the network of flat areas becomes strongly connected and yields a uniformly smooth surface with average roughness $S_a \approx 30$ nm (Iquebal et al., 2016).

The same work interprets the observed behavior as viscous flow at asperity–abrasive contacts, while noting that a more precise description is viscoplastic flow: thermally softened, rate-dependent plastic flow under high local shear and normal stress. Flash-temperature modeling uses the frictional/plastic heat-flux relation

$q = \mu H V,$

with $H$ the workpiece surface hardness, $\mu$ the friction coefficient, and $V$ the sliding speed. Under the reported conditions, peak flash temperatures exceed $700$ K for $\sim 30\%$ of sliding contacts on Ti–6Al–4V, within the dynamic recrystallization range. This provides a mechanistic basis for the fluid-like layer formation and bridging that dominate smoothening in that study (Iquebal et al., 2016).

In chemo-mechanical niobium polishing for SRF, the mechanism is described differently but is conceptually related. MP tailors both topography and near-surface crystalline integrity through controlled abrasion and benign chemical passivation. The colloidal silica step activated by $I_D/I_G$0 and $I_D/I_G$1 is intended to remove the mechanically affected layer from the prior abrasion step while passivating the surface and promoting controlled oxide growth. The reported consequence is lower residual RF losses and a lower temperature-dependent BCS component under cryogenic RF (Hryhorenko et al., 19 Sep 2025).

These results suggest that MP should not be reduced to a single removal mechanism. Depending on material, abrasive geometry, surface compliance, chemistry, and thermal conditions, the operative process can involve redistribution, passivation-assisted layer removal, damage-layer elimination, or a combination of these.

3. Process architectures, consumables, and hybrid workflows

The implemented process chain varies strongly with material class and application. In SRF niobium, the MP protocol reported for a QPR sample is a two-step sequence on a LAM PLAN Masterlam 1.0 system: a lapping step with a rigid composite disk charged with $I_D/I_G$2 polycrystalline diamond abrasives, removing $I_D/I_G$3, followed by a polishing step using a polyurethane cloth with a colloidal silica suspension ($I_D/I_G$4, $I_D/I_G$5 nm) activated by $I_D/I_G$6 and $I_D/I_G$7, diluted in deionized water up to $I_D/I_G$8, removing $I_D/I_G$9. Post-polishing contamination control and hydrogen mitigation consist of a $30\ \mu\text{m}$0 vacuum anneal for $30\ \mu\text{m}$1 h and high-pressure rinsing (HPR). This route removes $30\ \mu\text{m}$2 in total and uses no HF acids (Hryhorenko et al., 19 Sep 2025).

The broader SRF development program reports a five-step MP recipe for copper RF disks and a two-step MP route for planar niobium disks. For copper choke-cavity disks, the sequence uses diamond abrasives of $30\ \mu\text{m}$3, $30\ \mu\text{m}$4, $30\ \mu\text{m}$5, $30\ \mu\text{m}$6, and $30\ \mu\text{m}$7 on successive pads and disks, with pressures between $30\ \mu\text{m}$8 and $30\ \mu\text{m}$9 kPa and total polishing time $S_a \approx 30$0 min. For planar niobium, a pre-grind removes $S_a \approx 30$1 to establish flatness, then a $S_a \approx 30$2 diamond step removes the damaged layer, followed by a $S_a \approx 30$3 min chemo-mechanical finish on micro-porous polyurethane using $S_a \approx 30$4 nm $S_a \approx 30$5 diluted in water to $S_a \approx 30$6. Cleaning between steps uses ultrapure-water rinsing, ultrasonic cleaning in deionized water, and $S_a \approx 30$7 drying; an innovation in the final copper step is replacement of abrasive supply by DI water to purge residuals in situ (Hryhorenko et al., 2023).

For soft, porous IG-110 nuclear graphite, the validated route is explicitly hybrid. Stage 1 is mechanical polishing with pore filling to limit blocking and smearing: $S_a \approx 30$8-grit SiC grinding, then cloth polishing with $S_a \approx 30$9 diamond suspension and $q = \mu H V,$0 nm silica suspension, each for “several minutes,” followed by acetone washing to remove thermoplastic resin. Stage 2 is glancing-angle argon ion milling at $q = \mu H V,$1 with $q = \mu H V,$2 keV followed by $q = \mu H V,$3 keV. Stage 3 is rapid oxidation in air at an actual calibrated temperature of $q = \mu H V,$4 for $q = \mu H V,$5 min. Mechanical polishing alone leaves broken graphite pieces and elevated Raman $q = \mu H V,$6; ion milling removes debris but forms a thin irradiation-damaged carbon layer; rapid oxidation removes that layer without altering the underlying microstructure (Huang et al., 2021).

For Fe–5 wt% Al, metallographic polishing is reported as mirror-like polishing to a $q = \mu H V,$7 finish using a colloidal $q = \mu H V,$8 suspension; for cross-sections after oxidation, the final step uses $q = \mu H V,$9 of $H$0 granulation. For high-purity copper, a conventional metallographic route uses SiC pre-grinding through $H$1 grit followed by $H$2 suspensions on woven cloths decreasing from $H$3 to $H$4, then ultrasonic ethanol cleaning (Nowak et al., 2020, Serafin et al., 2019).

Material/application	Reported route	Reported purpose
Bulk Nb for SRF QPR	$H$5 diamond lapping + $H$6 nm $H$7 polishing + $H$8 anneal + HPR	RF-ready surface without HF acids
Cu and planar Nb for SRF substrates	Multi-step diamond or two-step chemo-mechanical MP	Ultra-smooth substrates for thin films
IG-110 nuclear graphite	MP + Ar ion milling + rapid oxidation	Artifact-free crack and pore imaging
Fe–5 wt% Al	$H$9 colloidal $\mu$0 finish	Controlled surface state for oxidation studies
High-purity Cu	SiC to $\mu$1 grit + $\mu$2 $\mu$3	Low initial roughness before annealing/oxidation

A general kinetics framework used in the SRF overview is Preston’s equation,

$\mu$4

where $\mu$5 is material removal rate, $\mu$6 is Preston’s coefficient, $\mu$7 is applied pressure, and $\mu$8 is relative velocity. The reported data specify pressures and rotation speeds, but do not report $\mu$9, $V$0 in m/s, or $V$1 explicitly (Hryhorenko et al., 2023).

4. Metrology, figures of merit, and validation criteria

MP is evaluated by a mixture of geometric, microstructural, spectroscopic, and functional metrics. Roughness metrics include line quantities such as

$V$2

and

$V$3

as well as areal analogs such as $V$4 and $V$5. In the SRF program, $V$6 and $V$7 were measured by laser confocal microscopy over $V$8 spots per sample with scan area $V$9. In copper oxidation studies, contact profilometry used a $700$0 mm traverse length, $700$1 mm/s scan speed, and five measurements per sample. In Zircaloy-4 BIB studies, white-light interferometry reported $700$2, $700$3, and $700$4, with $700$5 emphasized in the optimization (Hryhorenko et al., 2023, Serafin et al., 2019, Fang et al., 2022).

Spectroscopic and image-based criteria are indispensable when roughness alone is ambiguous. In nuclear graphite, Raman spectroscopy tracks the D, G, and D′ bands, with the ratio

$700$6

used as a defect indicator. Mechanical polishing yields the highest $700$7, ion milling broadens the G peak most strongly and adds an amorphous bump, and rapid oxidation reduces the G peak FWHM to $700$8 and $700$9 to $\sim 30\%$0, which the study recommends as an acceptance benchmark for a surface essentially free of artificial defects. Digital image processing then defines porosity as

$\sim 30\%$1

and uses ellipse-equivalent $\sim 30\%$2 and $\sim 30\%$3 to characterize pores and cracks (Huang et al., 2021).

In SRF niobium, validation is functional rather than purely topographic. The HZB quadrupole resonator (QPR) directly measures the RF surface resistance $\sim 30\%$4 calorimetrically with nanohm resolution. The analysis uses the decomposition

$\sim 30\%$5

and, for $\sim 30\%$6 K, the phenomenological fit

$\sim 30\%$7

The commonly used relation $\sim 30\%$8 is provided for context, but $\sim 30\%$9 and $I_D/I_G$00 are not used in the calorimetric analysis. This distinction is important because MP in SRF is ultimately judged by reductions in $I_D/I_G$01, lower $I_D/I_G$02, and accessible field level, not only by optical reflectivity or $I_D/I_G$03 (Hryhorenko et al., 19 Sep 2025).

Fractal analysis adds a scale-sensitive descriptor where buried interfaces or highly convoluted surfaces are important. In high-purity copper exposed at high temperature, relative length $I_D/I_G$04 is extracted from profile stepping, and the fractal dimension is given by

$I_D/I_G$05

This was used to evaluate the roughness of the buried $I_D/I_G$06 boundary under oxide, a surface inaccessible to conventional profilometry (Serafin et al., 2019).

5. Material-specific outcomes and performance regimes

In bulk niobium for SRF, the reported quantitative changes are substantial. After BCP, the QPR sample had $I_D/I_G$07 and $I_D/I_G$08; after the baseline EP sequence, $I_D/I_G$09 and $I_D/I_G$10; after MP, $I_D/I_G$11 and $I_D/I_G$12. Power spectral density analysis from ten $I_D/I_G$13 scans showed that MP reduced surface feature intensities across spatial frequencies more effectively than EP or BCP. Under RF at 415 and 847 MHz, MP reduced $I_D/I_G$14 across the temperature range, reduced $I_D/I_G$15 across the entire field range, achieved a minimum $I_D/I_G$16 n$I_D/I_G$17 at 415 MHz, and reduced residual resistance by $I_D/I_G$18–$I_D/I_G$19 n$I_D/I_G$20 relative to the baseline EP-treated state. Owing to the lowered dissipation, the MP sample sustained fields in excess of $I_D/I_G$21 mT at 415 MHz and $I_D/I_G$22 K without quench, and no field emission was reported (Hryhorenko et al., 19 Sep 2025).

The collaborative SRF overview extends this picture to substrate preparation. Copper choke-cavity disks were improved from $I_D/I_G$23, $I_D/I_G$24 before MP to $I_D/I_G$25 and $I_D/I_G$26 after MP. For niobium disks with $I_D/I_G$27, initial $I_D/I_G$28 and $I_D/I_G$29 were reduced to $I_D/I_G$30 nm and $I_D/I_G$31 after pre-grinding and two-step MP. For $I_D/I_G$32 niobium, a shorter cycle yielded $I_D/I_G$33 nm. In QPR tests, the MP-processed sample showed $I_D/I_G$34 n$I_D/I_G$35, versus an EP-processed baseline with $I_D/I_G$36 n$I_D/I_G$37, and MP enabled RF fields $I_D/I_G$38 mT with reduced dissipated power (Hryhorenko et al., 2023).

For IG-110 nuclear graphite, the central outcome is not that MP alone performs well, but that it is insufficient for artifact-free microstructure revelation. Mechanical polishing produces the highest Raman D-mode intensity, $I_D/I_G$39, and a surface where widespread micro-cracks are not visible. After ion milling and rapid oxidation, micro-cracks become open and distinct in SEM, EDX shows no Ar, and Raman reaches G-peak FWHM $I_D/I_G$40 and $I_D/I_G$41. Quantitative image analysis then identifies over $I_D/I_G$42 pores and $I_D/I_G$43 cracks, with porosity distributions spanning from about $I_D/I_G$44 nm to $I_D/I_G$45 (Huang et al., 2021).

For Fe–5 wt% Al, metallographic polishing yields the smoothest initial state among the reported mechanical treatments: $I_D/I_G$46, $I_D/I_G$47, $I_D/I_G$48, and $I_D/I_G$49. After oxidation in air at $I_D/I_G$50 and $I_D/I_G$51, polished and ground surfaces develop local Fe-rich oxide nodules with a continuous sub-layer of $I_D/I_G$52 at the oxide/metal interface, whereas grit-blasted surfaces form multilayer Fe-rich scales and an internal nitridation zone of AlN. The polished sample shows 24 h mass gains of $I_D/I_G$53 mg cm$I_D/I_G$54 at $I_D/I_G$55 and $I_D/I_G$56 mg cm$I_D/I_G$57 at $I_D/I_G$58, compared with $I_D/I_G$59 and $I_D/I_G$60 mg cm$I_D/I_G$61 for grit-blasted surfaces. The study attributes the accelerated oxidation after grit-blasting to grain refinement in the near-surface region and increased easy diffusion paths (Nowak et al., 2020).

For high-purity copper, a polished surface prepared by SiC grinding to $I_D/I_G$62 grit and final $I_D/I_G$63 polishing to $I_D/I_G$64 achieves initial $I_D/I_G$65, compared with $I_D/I_G$66 for a 220-grit ground surface and $I_D/I_G$67 for a sand-blasted surface. After annealing in argon, the polished surface roughens to $I_D/I_G$68 at $I_D/I_G$69 but remains near baseline at $I_D/I_G$70 with $I_D/I_G$71. Under oxidation in air, fractal analysis shows that the buried $I_D/I_G$72 interface smooths with temperature; the most pronounced smoothing occurs for initially rough sand-blasted surfaces, which reach $I_D/I_G$73 and $I_D/I_G$74 at $I_D/I_G$75 (Serafin et al., 2019).

6. Artifacts, limitations, and unresolved questions

The dominant practical limitation of MP is that the surface can become smooth while retaining preparation-induced defects, contamination, or buried damage. In graphite, conventional MP tends to smear and fracture near-surface lamellae, block pores with debris, and inflate the Raman D band; ultrasonic cleaning was intentionally avoided because it can induce damage to graphite structures. In SRF niobium, abrasive or slurry residue contamination and hydrogen uptake are explicit risks, mitigated respectively by HPR and a $I_D/I_G$76 h vacuum anneal. In BIB-polished Zircaloy-4, parameter coupling among angle, voltage, duration, and temperature determines whether the damaged layer is removed or whether ripple roughness, curtaining, pitting, or grain-boundary shadowing become dominant (Huang et al., 2021, Hryhorenko et al., 19 Sep 2025, Fang et al., 2022).

A second limitation is geometric transferability. The SRF results reported for niobium and copper were obtained primarily on planar disks and QPR coupons. The next step identified by the collaboration is transfer of MP protocols to the curved interiors of $I_D/I_G$77 GHz elliptical cavities, where contact uniformity, pressure distribution, velocity control, fluid management, and fixturing become decisive. This suggests that coupon-level success does not automatically imply cavity-level uniformity, even when the underlying chemistry and consumables are unchanged (Hryhorenko et al., 2023).

A third issue is interpretation. Topography and performance can correlate strongly, but the data repeatedly show that roughness is not the sole state variable. In SRF niobium, the combination of reduced roughness amplitudes and improved microstructural integrity is stated to be necessary to explain the concurrent reductions in $I_D/I_G$78 and $I_D/I_G$79; topography alone is insufficient because purely mechanical polishing approaches such as CBP can yield low roughness but leave a damage layer that requires subsequent EP to restore RF performance. In oxidation studies, polishing and grinding may have similar near-surface grain size while still generating different diffusion paths and oxide-scale development. In EBSD preparation, low $I_D/I_G$80 must still be reconciled with indexing quality and orientation-dependent sputter relief (Hryhorenko et al., 19 Sep 2025, Nowak et al., 2020, Fang et al., 2022).

Several open questions remain explicit. In the niobium SRF study, trapped flux sensitivity was not assessed, and the trapped-flux relation

$I_D/I_G$81

was therefore not parameterized. The study also identifies the quantitative separation of MP-induced changes in $I_D/I_G$82 parameters, such as $I_D/I_G$83 and mean free path, as unresolved. Future work is stated to include flux studies and microscopic spectroscopy to link MP processing variables to superconducting parameters. For graphite and other soft porous materials, several polishing parameters such as load, speed, slurry flow, ion-milling current, and milling time were not reported and must therefore be tuned experimentally. For curved SRF cavities, industrialization will require fixtures and automation that preserve uniform $I_D/I_G$84 and $I_D/I_G$85 while preventing edge roll-off and uneven dwell (Hryhorenko et al., 19 Sep 2025, Huang et al., 2021, Hryhorenko et al., 2023).

Taken together, these results define MP not as a single universal recipe but as a material- and application-specific strategy for controlling topography, damage, contamination, and near-surface structure. In some systems, such as SRF niobium and copper substrates, it functions as a credible alternative to hazardous acid-based finishing. In others, such as graphite or EBSD-sensitive zirconium alloys, it is best understood as one stage within a broader preparation architecture whose success depends on complementary removal of polishing-induced artifacts.