THE GOVERNING PRINCIPLE What you see at the end of the process is rarely what caused it. A scratch can be born in grinding, carried by a contaminated slurry loop, and revealed only after final polishing. Diagnose the process history, not only the final surface.
Olivier M. Founder, Illumine-Consultants
Technical publication | 16 October 2026
How to use this guide
This guide is designed for production managers, process engineers, quality teams, laboratory teams, machine builders, and technical distributors. It provides a disciplined way to move from an observed defect to a ranked set of causes, controlled process adjustments, and compound-selection logic. It is not a substitute for the validated operating window of your machine, substrate, pad, or chemical supplier.
SAFETY AND VALIDATION Do not change multiple variables at once. Work inside the equipment and material supplier limits. Follow the applicable Safety Data Sheet, machine lockout rules, ventilation requirements, waste procedures, and site risk assessment. Stop the process when cracks, thermal damage, unstable fixturing, or abnormal machine behaviour are observed. Any numerical setpoint must come from your validated recipe or a controlled trial approved by the responsible process owner.
The diagnostic sequence
Step
Question
Required evidence
1. Define
What exactly is the defect, where is it, and when does it appear?
Photos under fixed lighting; defect map; part orientation; first failing piece.
2. Contain
What product, machine, lot, shift, and slurry loop may be affected?
Hold boundaries; traceability; retained samples; reference parts.
3. Separate
Is the signal caused by substrate, geometry, slurry, pad/tool, machine, cleaning, or inspection?
Before/after cleaning; pre-polish inspection; witness sample; machine and consumable history.
4. Rank
Which causes best fit the defect morphology and timing?
Cause tree; change log; defect direction; spatial repeatability.
5. Recover
What is the lowest-risk single-variable adjustment?
Baseline recipe; approved operating window; short controlled run.
6. Verify
Did the defect decrease without creating a new failure mode?
Same inspection method; roughness/figure data where required; yield and rate check.
7. Lock
How will recurrence be prevented?
Updated standard work; limits; alarms; filter/pad/slurry life rules; training.
Defect families covered
Scratches, sleeks, pits, digs, micro-cracks, and subsurface damage
Wave formation, ripple, orange peel, waviness, and local figure error
Roughness drift, removal-rate loss, non-uniformity, and edge roll-off
Edge chipping, bevel damage, and handling fractures
Haze, cloudiness, stain, swirl marks, and polishing trails
Compound buildup, pad loading, residues, watermarking, and post-polish contamination
1. Diagnosis before adjustment
1.1 Read morphology as process evidence
Observed pattern
First interpretation
Do not conclude yet
Long straight or directional scratch
Entrained hard particle, damaged tool, handling contact, or prior-stage damage aligned with motion
That the polishing powder is intrinsically too coarse.
Random short scratches or sleeks
Contamination, agglomerates, debris, pad condition, poor cleaning between stages
That higher pressure will remove them.
Radial or rotational swirl
Machine motion, local dryness, pad texture, slurry starvation, or trapped debris following the kinematics
That additional polishing time alone will cure it.
Uniform haze
Fine widespread scatter from roughness, residue, micro-pitting, incomplete polishing, or cleaning film
That haze is always a polishing-rate problem.
Edge-only defects
Pressure distribution, support, bevel, part loading, tool overhang, entry/exit shock, or handling
That the centre recipe is wrong.
Defect repeats at same machine location
Tool, fixture, delivery jet, conditioning, platen, spindle, or inspection artefact
That it is a random lot issue.
Defect follows a material lot
Incoming surface, glass chemistry, coating, lot contamination, or compound variability
That the machine is stable merely because settings are unchanged.
1.2 Separate appearance, topography, and cleanliness
A visible mark can be a groove, a raised deposit, an optical contrast effect, a cleaning film, or a buried defect revealed by polishing. Use at least two independent observations whenever the decision is consequential: controlled visual inspection plus microscopy, profilometry, interferometry, roughness measurement, or a validated cleaning challenge as appropriate.
GOLDEN RULE Never repolish an unidentified defect simply because it is visible. Repolishing can deepen a pit, round an edge, worsen figure, expose subsurface damage, or erase evidence of the true cause.
1.3 Capture the minimum process history
Category
Record before changing anything
Part
Material, geometry, coating state, incoming lot, orientation, previous process, storage and handling.
Machine
Machine ID, recipe revision, speed/motion, load or pressure command, time, temperature trend, alarms, maintenance.
Stop and preserve evidence. Clean and inspect tool, carrier, loop, filters, and part path. Replace suspect consumable if pattern repeats. Re-run within baseline settings.
Choose a controlled particle-size distribution with low oversize tail and verified suspension behaviour. Do not “solve” contamination by moving directly to a finer grade.
Micro-cracks / subsurface damage
Fine crack network, edge-origin crack, pits opening during polish, latent failure after cleaning
Damage from prior grinding; excessive local stress; impact; thermal gradient; unstable support; brittle material response
Stop. Inspect pre-polish surface and edge. Check grinding progression, support, clamping and temperature history. Remove damaged layer only under an approved rework route.
Compound change cannot repair a structural crack. Use grade selection only after the upstream damage source is corrected.
Wave / ripple / orange peel
Periodic waviness, mottled texture, local hills and valleys, distorted reflection
Check spatial periodicity against machine motion. Inspect tool flatness/conformity, pad condition, support and thermal stability. Return to validated balanced motion.
Prefer a compound and suspension that deliver stable, uniform action instead of chasing maximum removal rate.
Roughness drift
Ra/Rq or scatter gradually worsens; lot-to-lot finish changes
Slurry ageing; concentration or pH drift; particle agglomeration; pad ageing; contamination; temperature; water quality; lot variability
Trend the process against time and consumable age. Refresh or segregate slurry per validated rules. Check measurement system and retained references.
Prioritise lot consistency, narrow distribution, low coarse tail, chemical compatibility, and controlled suspension.
Edge chipping / roll-off
Chips, shelling, corner breakage, rounded or low edge
Insufficient bevel; unsupported edge; fixture stress; tool overhang; entry/exit shock; high local pressure; handling
Protect and inspect edges before polish. Check bevel, support, loading, trajectory and unloading. Reduce edge aggressiveness only within approved recipe limits.
Use a grade/tool combination matched to brittle edge risk and finishing need; high cut rate is not the first priority.
Haze / cloudiness
Uniform or local loss of clarity; diffuse scatter
Incomplete removal; fine roughness; residue; micro-pits; pad glazing; poor rinse/dry; incompatible chemistry
Run a validated cleaning challenge. Compare pre- and post-clean inspection. Check rate, pad condition, slurry delivery and rinse water.
Distinguish true surface haze from residue before changing grade. Select for finish, low defectivity, cleanability and substrate compatibility.
Determine whether pit existed before polish. Inspect incoming and intermediate stages. Check chemistry exposure and particle events.
Finer compound will not close a pit. Choose low-defectivity material after source removal and assess rework feasibility.
3. Scratches, sleeks, pits, and digs
Scratches are among the most informative defects because their direction, repetition, and depth often preserve the history of the contact event.
Likely causes
Foreign hard particles from the workpiece path, carrier, pad, tank, piping, filter bypass, conditioner, or environment.
Agglomerated abrasive or a coarse tail in the working particle-size distribution.
Pad damage, embedded debris, glazed regions, exposed backing, or a worn fixture.
Residual damage from grinding or lapping that final polishing reveals but did not create.
Part-to-part, carrier-to-part, or handling contact after polishing.
Recovery sequence
Quarantine the affected loop and retain the last good and first bad parts.
Record scratch direction relative to part orientation and machine motion.
Inspect a cleaned part, tool/pad, carrier, tank, delivery nozzles, return path, and filtration.
Use a witness part to determine whether the scratch is generated before, during, or after polishing.
Restore the validated baseline with clean consumables before adjusting pressure, speed, time, or grade.
Compound logic
Ask for particle-size distribution and oversize control, not only median particle size.
Verify dispersion and suspension behaviour under the actual water, agitation, temperature, and recirculation conditions.
For finish-sensitive work, prioritise defectivity and consistency over maximum nominal removal rate.
RELEASE CRITERION A correction is complete only when the defect is reduced under the same inspection method, the required geometry/roughness is preserved, removal rate remains controlled, and the result repeats on more than one part or run according to the site validation plan.
4. Micro-cracks and subsurface damage
Micro-cracks may be inherited from cutting, grinding, edge finishing, blocking, clamping, impact, or thermal stress. Polishing can expose these cracks or reduce the remaining ligament until the defect becomes visible.
Likely causes
Grinding damage not fully removed before the polishing stage.
Brittle edge condition, insufficient bevel, sharp corners, or local contact stress.
Excessive or unstable clamping/support load.
Thermal gradients from friction, coolant/slurry temperature, or interrupted wetting.
Material, inclusion, or coating-related weakness.
Recovery sequence
Stop processing cracked parts and assess containment by lot, machine, and upstream stage.
Inspect edges and surfaces before polishing using the appropriate method for the material and risk.
Review stock-removal allowance and grinding progression.
Check support, blocking, clamping, loading and unloading, and temperature history.
Use only an approved rework route. Reject when structural integrity or optical function cannot be demonstrated.
Compound logic
A polishing compound is not a crack repair.
After upstream correction, use an application-matched grade that provides controlled removal without a high defectivity tail.
For brittle or thin parts, select the entire system as a package: compound, pad/tool, support, motion, and thermal control.
RELEASE CRITERION A correction is complete only when the defect is reduced under the same inspection method, the required geometry/roughness is preserved, removal rate remains controlled, and the result repeats on more than one part or run according to the site validation plan.
5. Wave formation, ripple, and orange peel
These defects are geometry and texture problems. They can be periodic, local, or broad-area, and may be driven by the machine-tool-workpiece system rather than by abrasive chemistry alone.
Likely causes
Tool or pad not conforming to the intended surface.
Pressure distribution or support that varies with position.
Pad wear, groove pattern, glazing, or local loading.
Unstable slurry film, local drying, or temperature drift.
Inherited waviness from generation, grinding, mould machining, or EDM preparation.
Recovery sequence
Map the frequencies and directions of the surface pattern against machine kinematics.
Check tool geometry, pad thickness/wear, support, fixture runout, and contact pattern.
Confirm stable wetting and temperature.
Return to the validated motion and contact condition before extending cycle time.
Verify figure and roughness separately; one can improve while the other deteriorates.
Compound logic
Do not select solely on removal rate. A fast grade can amplify non-uniform contact.
Prefer repeatable chemistry, stable dispersion, and a particle distribution suited to the target finish and tool.
Use controlled comparative trials with the same machine, pad, substrate, and metrology.
RELEASE CRITERION A correction is complete only when the defect is reduced under the same inspection method, the required geometry/roughness is preserved, removal rate remains controlled, and the result repeats on more than one part or run according to the site validation plan.
6. Roughness drift and removal-rate instability
Drift is a time-dependent failure. The key question is what ages, accumulates, warms, dilutes, loads, settles, or changes between the first good part and the first bad part.
Likely causes
Slurry concentration, chemistry, temperature, contamination, settling, or ageing.
Pad/tool conditioning state, glazing, wear, or saturation.
Water-quality variation or uncontrolled make-up additions.
Filter loading, bypass, flow change, dead zones, or recirculation history.
Incoming lot or glass chemistry change.
Measurement-system drift, cleaning inconsistency, or sampling bias.
Recovery sequence
Plot finish and removal rate against run time, slurry age, pad age, filter history, temperature, and additions.
Check the measurement system with a retained reference.
Compare fresh and working slurry only through an approved controlled test.
Restore one variable at a time to baseline.
Create control limits for the variables that show predictive value.
Compound logic
Demand lot-to-lot certificate data relevant to the application.
Assess median size together with distribution width, coarse tail, oxide composition where relevant, pH behaviour, sedimentation, and compatibility.
A grade that starts fast but drifts early can have a worse cost per conforming part than a slower, stable grade.
RELEASE CRITERION A correction is complete only when the defect is reduced under the same inspection method, the required geometry/roughness is preserved, removal rate remains controlled, and the result repeats on more than one part or run according to the site validation plan.
7. Edge chipping and edge roll-off
Edges see a different mechanical system than the centre: less support, changing contact area, bevel discontinuities, and entry/exit effects.
Likely causes
Missing, inconsistent, or damaged protective bevel.
Unsupported edge or excessive carrier clearance.
Tool overhang, local pressure concentration, or trajectory reversing at the edge.
Hard loading/unloading contact or debris at the fixture boundary.
Upstream micro-cracking that opens under polishing.
Recovery sequence
Inspect and document edge condition before polishing.
Check bevel geometry, carrier pocket, support film, blocking, and fixture cleanliness.
Review tool path and contact near the boundary.
Reduce local aggressiveness only within an approved recipe; verify centre rate and figure are not compromised.
Confirm edge quality after cleaning and handling, not only on-machine.
Compound logic
Prioritise controlled action and low oversize content for brittle edges.
Match compound to pad compliance and target finish.
For mould and precision-optics work, protect geometry and critical details before increasing cut.
RELEASE CRITERION A correction is complete only when the defect is reduced under the same inspection method, the required geometry/roughness is preserved, removal rate remains controlled, and the result repeats on more than one part or run according to the site validation plan.
8. Haze, swirl marks, and compound buildup
These defects often sit at the boundary between polishing and cleaning. Diagnose whether material has been removed incorrectly or whether material remains on the surface.
Likely causes
Incomplete polish or elevated fine-scale roughness.
Residue from dried slurry, pad binder, water minerals, cleaner, or reaction products.
Pad glazing/loading or surface texture transfer.
Slurry starvation, local drying, poor agitation, dead zones, or unstable concentration.
Rinse and dry process that leaves film or watermarking.
Recovery sequence
Inspect before and after a validated cleaning challenge.
Use topography or microscopy to distinguish grooves, pits, and raised residue.
Check wetting, delivery, return flow, agitation, pad state, rinse water, and drying.
Restore loop hygiene and validated slurry condition.
Only then adjust the polishing recipe or compound.
Compound logic
Select for cleanability and suspension stability as well as finish.
Verify compatibility with the actual water, pad, substrate, and post-polish cleaning chemistry.
RELEASE CRITERION A correction is complete only when the defect is reduced under the same inspection method, the required geometry/roughness is preserved, removal rate remains controlled, and the result repeats on more than one part or run according to the site validation plan.
9. Application playbooks
The same visual defect can have different priorities by application. Use the following playbooks to focus the investigation without narrowing it prematurely.
Flat glass
Primary control priorities: Edge quality, visual clarity, distortion, cycle time, wheel/pad life, recirculated slurry stability.
Signal
Where to investigate first
Residual linear scratches
Check glass handling and washer path, upstream grinding debris, edge-machine wheels/pads, tank and loop contamination.
Wave or optical distortion
Check local over-processing, dwell pattern, support, machine motion and thermal history.
Haze after polishing
Separate true roughness from dried compound or rinse residue; verify water and cleaning.
Buildup in recirculation
Review agitation, dead zones, solids control, evaporation, filtration and tank-cleaning discipline.
COMPOUND SELECTION Choose an application-grade ceria or rare-earth compound with stable recirculation behaviour, controlled oversize particles, consistent lot quality, and proven compatibility with the wheel/pad and water system.
Glass mould polishing
Primary control priorities: Detail preservation, gloss, release surface, absence of pits/drag lines, controlled edge and feature geometry.
Signal
Where to investigate first
Orange peel
Check preparation quality, incomplete removal of machining/EDM texture, pressure, tool conformity and abrasive sequence.
Pitting
Determine whether the pit is an inclusion, corrosion, pull-out, contamination impact, or prior damage before deeper polishing.
Rounded edges/details
Reduce uncontrolled access and local pressure; protect shut-offs, vents, engravings and sharp transitions.
Directional drag lines
Check tool direction, contamination, spent abrasive and stage-change cleaning.
COMPOUND SELECTION Compound selection follows mould material, preparation state, feature geometry, target gloss and cleanability. Do not publish a universal grade recommendation without a controlled trial.
Ophthalmic lenses
Primary control priorities: Optical power/geometry, cosmetic surface, scatter, coating readiness, reproducibility and traceability.
Signal
Where to investigate first
Haze or cloudiness
Check incomplete polish, residue, pad condition, incoming surface and cleaning.
Waves/orange peel
Check tool conformity, support/blocking, pressure distribution and inherited surfacing texture.
Edge chips
Check bevel, blocking/unblocking, carrier contact and handling.
Coating reveals defects
Inspect and document the polished substrate before coating; coating may replicate or amplify existing defects.
COMPOUND SELECTION Use optical-grade material with tight consistency and defectivity control. Requalification is required when changing compound, pad, substrate, or cleaning because product and process are coupled.
Precision optics
Primary control priorities: Surface quality, roughness, figure, wavefront, scatter, edge integrity, coating performance and laser-damage risk where applicable.
Signal
Where to investigate first
Scratch/dig failure
Use the applicable drawing and inspection standard; localise whether the defect is manufacturing or handling related.
Figure error with good roughness
Investigate tool influence, dwell, support and thermal stability rather than selecting a finer abrasive.
Good figure with poor roughness
Investigate particle distribution, pad/tool state, contamination, chemistry and cleaning.
Subsurface damage
Review fine-grinding removal and inspection; polishing may expose rather than create the damage.
COMPOUND SELECTION Prioritise documented particle distribution, lot traceability, low defectivity, stable suspension and substrate/tool compatibility. Validate using the actual metrology and acceptance criteria.
Electronics and glass CMP
Primary control priorities: Planarity, defect density, particle residue, scratches, contamination, uniformity and downstream cleanliness.
Signal
Where to investigate first
Microscratches
Focus on slurry, pad, conditioner, filtration, agglomeration and delivery-loop history.
Particles/residue
Investigate adhesion under process chemistry, post-polish cleaning, pad debris, dead zones and cross-contamination.
Non-uniformity / dishing / erosion
Separate chemical selectivity and pattern effects from mechanical pressure and pad effects.
Watermarks / films
Review rinse, brush/cleaning condition, drying and chemical compatibility.
COMPOUND SELECTION Electronics requires application-specific CMP validation. A general glass-polishing powder must not be presented as electronics-qualified without data from the exact substrate, stack, tool and clean sequence.
10. Compound-selection logic
Compound selection is not a catalogue lookup. It is a controlled decision that connects substrate, defect mode, removal requirement, surface requirement, tool/pad, water system, recirculation, cleaning, and inspection.
Decision gate
Questions
Evidence to request or generate
1. Application fit
Is the compound intended for this substrate and process architecture?
Technical data sheet, Safety Data Sheet, application statement, supplier discussion, controlled trial.
2. Particle system
What are the median, distribution width, oversize tail, agglomeration tendency, and working distribution after mixing/recirculation?
Particle-size data and method; coarse-particle control; retained-lot comparison.
3. Chemistry
How does the grade interact with the glass/material, water, pH window, pad/tool and cleaning chemistry?
Supplier documentation; compatibility trial; residue and corrosion/etch inspection.
4. Suspension
Does it remain distributed under the real tank, agitation, delivery, temperature and downtime pattern?
What is the cost per conforming part, including cycle, rejects, pad life, cleaning and changeover?
Production trial data. Do not decide from price per kilogram alone.
Ceria and lanthana: practical positioning
For silicate-glass polishing, ceria-based systems are widely used because polishing combines abrasive action with chemical interaction at the hydrated glass surface. Rare-earth polishing products may also contain lanthanum and other light rare-earth components. The correct decision is application-specific and must be tied to verified product composition, particle system, surface result, rate, stability and cleanability. Chemical purity alone does not establish polishing-grade performance.
ILLUMINE RULE Recommend a polishing profile, not a generic powder. The profile should state the intended application, surface objective, removal requirement, particle-system evidence, suspension behaviour, lot-control evidence, pad/tool compatibility, cleaning route, and trial acceptance criteria.
Controlled comparison protocol
Freeze the current approved recipe and reference product as the control.
Define acceptance criteria before the test: appearance, roughness, figure/geometry, removal rate, edge quality, residues, cleaning, pad/tool condition and repeatability.
Change only the compound for the first comparison unless the purpose is a full process redesign.
Use identified lots, documented mixing, the same water, machine, pad/tool state, substrate lot where possible, and the same inspection method.
Record all additions, stops, temperatures, durations, filter changes and anomalies.
Do not claim equivalence from one visually acceptable part.
11. Rapid triage tools
11.1 Stop / continue decision
Condition
Action
Crack, unstable fixturing, abnormal vibration/noise, overheating, loss of wetting, or safety deviation
STOP. Make equipment and product safe. Escalate to the responsible owner.
Sudden scratch burst across multiple parts
STOP the affected loop. Preserve parts, slurry samples, filters and consumables.
Gradual rate or roughness drift within known pattern
CONTAIN and troubleshoot against time-based variables before recipe escalation.
Single isolated cosmetic mark with no repeat
HOLD the part, inspect handling and repeatability; do not rewrite the process from one data point.
Defect appears only after cleaning/coating
Separate substrate, cleaning and downstream-process contributions with staged inspection.
11.2 Five-question shift handover
What was the last verified good part and first verified bad part?
What changed, even if it was not considered a process change?
Where is the defect on the part and how is it oriented?
What evidence has been preserved?
What single controlled action has been taken, and what was the result?
11.3 Root-cause worksheet
Field
Entry
Defect name and photo ID
Application / material / geometry
Machine / recipe revision
Part and compound lot
Pad/tool/fixture history
Slurry mix / age / filtration / additions
Last good / first bad
Defect map and orientation
Ranked suspected causes
Single-variable test
Result and metrology
Disposition / containment
Corrective and preventive action
Owner / date / approval
12. Risks and points of vigilance
Diagnostic risks
Treating the visible defect as the root cause.
Changing pressure, speed, concentration, time and pad simultaneously, which destroys learning.
Using a supplier label or nominal particle size as a substitute for working particle distribution and process evidence.
Confusing residue with roughness, or coating-amplified defects with coating-created defects.
Extending polish time until appearance improves while figure, edge geometry, or subsurface integrity worsens.
Comparing parts inspected under different cleaning, lighting, magnification or thresholds.
Commercial and publication risks
Publishing numerical recipes without substrate-, machine-, pad-, and product-specific validation.
Claiming a compound is identical or equivalent to a competitor without public evidence and controlled trials.
Presenting electronics suitability without exact stack, tool, defect and cleaning data.
Signing technical content in the founder’s name without his explicit technical validation.
Using confidential knowledge from previous employment rather than public sources, Illumine trials and customer-approved case data.
PUBLICATION GATE Before website publication, Olivier M. should validate every technical statement, especially any numerical setpoint, product recommendation, comparison, safety instruction, or performance claim. This guide deliberately provides adjustment direction and diagnostic logic rather than universal numerical recipes.
Conclusion
A polished surface is the final record of an entire process chain.
The fastest path to recovery is not aggressive adjustment. It is disciplined separation of substrate history, machine and fixture behaviour, tool/pad condition, slurry and compound state, cleaning, handling, and inspection. Once the mechanism is understood, the lowest-risk correction is usually clear.
Illumine’s role is to help industrial teams connect observable evidence to a controlled polishing profile: the right compound for the application, used inside a stable and documented process, with the result judged by the correct surface and geometry criteria.
FOR A USEFUL TECHNICAL DISCUSSION Prepare: application and substrate, machine and tool/pad, current compound and lot, defect photos under fixed lighting, location/orientation map, recipe revision, slurry history, last good/first bad part, cleaning sequence, and available roughness/figure data.
Sources and verification status
Information verified from the sources below was used to establish the guide’s process logic, defect families, inspection principles, application scope and compound-selection framework. Application-specific numerical setpoints have intentionally not been invented. They must be taken from validated recipes, supplier documentation, or controlled trials.
Source
Use in this guide
Stanford Advanced Materials
“How to Use Cerium Oxide for Glass Polishing,” accessed 30 September 2026. Ceria polishing mechanism; relevance of grade, particle size, slurry, pad, pressure, speed and temperature; limitations for cracks/chips.
BenchChem
“Application Notes and Protocols for the Use of Cerium(IV) Oxide as a Glass Polishing Agent,” May 2026. Chemical-mechanical interaction and physicochemical properties.
Edmund Optics
“Understanding Surface Quality Specifications,” accessed 30 September 2026. Scratches/digs, scatter and inspection standards context.
ZEISS
“Surface Inspection” and “Surface Finish and Roughness Solutions,” accessed 30 September 2026. Visual/tactile/optical inspection and distinction among surface deviation, waviness and roughness.
IntechOpen
Jihoon Seo, “Chemical Mechanical Planarization-Related Contaminants: Their Sources and Characteristics,” published 29 October 2020. CMP residues, particles, pad debris, metallic contamination, watermarks and post-CMP cleaning.
Electrochemical Society conference paper
J.-G. Park et al., “CMP defects; their detection and analysis on root causes.” Consumables, slurry, pad and conditioner as scratch/particle sources.
Entegris
“CMP Process Monitoring,” accessed 30 September 2026. Chemistry, working particle size and distribution as process-monitoring variables for removal and defect control.
SENKO
“Polishing Best Practices,” application note, 2023. Importance of cleaning between polishing steps and troubleshooting scratches, pits and geometry.
Yishun Optical
“Common Defects in Optical Polishing and How to Avoid Them,” 10 June 2026. Defect taxonomy and effect on optical function.
Cavity Mold
“Injection Mold Polishing Defects: Orange Peel, Pitting, and Over-Polishing,” updated 23 July 2025. Defect classification before repolishing; geometry and edge risks.
V-Optics
“Detection and Measurement of Lens Defects,” accessed 30 September 2026. Automated detection of chips, scratches and machining defects in lenses.
J. Racenstein
“How to Polish Glass with Cerium Oxide,” 8 June 2026. Haze, swirl marks, wet-process discipline and pad role in general glass polishing.
GlobalFoundries / CMP User Group
“CMP defects: Influence of CMP slurry, handling and delivery,” 2020. Slurry age, storage, filtration, flow and handling as defect variables.
