Substrate Glass Defect Classification Analysis: Methods, Standards, and Quality Evaluation

Thin and Flexible Glass

Ultra-thin glass (100 micrometers and below) intended for flexible or bendable applications requires heightened attention to handling damage, micro-cracks, and waviness, since mechanical margins are minimal.

Specialty and Packaging Glass

Glass carriers, glass cores, and interposer substrates used in semiconductor packaging, as well as borosilicate and fused silica substrates for optical and photonic devices, are analyzed for inclusions, bubbles, striae, and flatness deviations that could impair alignment or thermal performance.

Testing Items

Defect classification schemes generally organize findings into three geometric zones—surface, bulk/internal, and edge—with dimensional and form tolerances evaluated in parallel. Representative items include:

• Surface scratches: linear abrasion marks characterized by length, width, and depth; classified by visibility under defined illumination rather than width alone.
• Digs, pits, and craters: localized material removal or impact damage, sized by equivalent diameter and depth.
• Particles and residue: loose or adherent contamination such as dust, glass chips, or processing residue, identified by size and composition.
• Stains, smears, and coating marks: films or discoloration from cleaning, etching, or handling, evaluated for area and contrast.
• Bubbles and seeds: internal gaseous voids characterized by diameter, density, and distribution.
• Inclusions and stones: non-melted or crystallized material within the glass body, assessed for size and optical disturbance.
• Striae and cords: refractive index inhomogeneity appearing as thread-like distortion, graded by shadow-graph contrast.
• Edge chips and micro-cracks: shell-shaped fractures or fracture lines along cut or ground edges, measured by width and depth perpendicular to the edge.
• Form defects: warpage, bow, thickness variation, and surface waviness measured across the substrate.
• Pattern-transfer and handling mura: localized brightness or texture non-uniformity caused by contact marks, suction marks, or roller marks.

Testing Methods

Classification analysis combines complementary techniques because no single method captures all defect types and size regimes. Typical workflows proceed from rapid full-surface scanning to targeted high-resolution characterization.

• Controlled visual inspection: trained examiners evaluate samples under standardized illuminance, background, and viewing geometry as specified in ASTM C1036 and comparable GB/T or JIS procedures, establishing baseline acceptance decisions.
• Automated optical inspection (AOI): line-scan or area-scan systems with dark-field and bright-field channels map every defect coordinate across the substrate, generating size distributions and density maps.
• Laser surface scanning: laser scattering instruments detect and size particles, scratches, and pits down to the sub-micron scale, widely used for polished and pristine surfaces.
• Optical microscopy: bright-field, dark-field, and differential interference contrast (DIC) imaging resolve defect morphology; calibrated measurement reticles or software provide dimensional data consistent with ISO 10110-7 dig and scratch notation.
• Surface profilometry: stylus profilometers and white-light interferometers quantify scratch depth, pit depth, and step height, distinguishing removable contamination from permanent damage.
• Interferometry: Fizeau and Shack-Hartmann methods measure flatness, bow, and waviness for form classification.
• Polarized light and shadow-graph methods: schlieren or shadow-graph projection reveals striae, inclusions, and internal stress that scatter or refract light unevenly.
• Scanning electron microscopy with EDS (SEM/EDS): identifies the elemental composition of inclusions, stones, and particles, enabling root-cause attribution to tin, refractory, alumina, or environmental sources.
• Cross-sectional analysis: polished or cleaved cross sections expose sub-surface damage (SSD) and crack depth beneath chips and digs.

Results are compiled into a defect map with per-category counts, worst-case dimensions, and comparisons against contract-specified limits. Critical defects such as through-cracks, metallic inclusions, or chips exceeding depth thresholds are flagged regardless of quantity.

Applications

Defect classification analysis supports multiple stages of the glass and electronics value chain:

• Incoming material qualification: panel manufacturers verify that glass lots meet agreed defect density and size limits before committing substrates to expensive TFT processes.
• Supplier benchmarking and audits: comparative classification data quantifies quality differences among substrate suppliers and validates process improvement claims.
• Process improvement and root-cause analysis: compositional and morphological signatures link defects to specific melting, forming, cutting, grinding, or packaging operations.
• Yield engineering: defect maps correlated with downstream electrical or optical failures identify which defect classes are truly yield-limiting, refining acceptance criteria.
• Reliability screening: edge chip and micro-crack depth data feed strength models used to predict handling survival and in-service durability for cover and flexible glass.
• Dispute resolution and certification: accredited third-party laboratories provide impartial classification reports referenced in commercial settlements and regulatory submissions.

FAQ

Q1: What size defects can be detected during classification analysis?

Visual methods typically resolve features of a few tens of micrometers, while laser scattering and microscopy extend detection to sub-micron particles and nanometer-scale scratch depths. Practical reporting thresholds are agreed between buyer and supplier based on application criticality.

Q2: How are defects classified as critical, major, or minor?

Classification depends on defect type, size, location relative to active areas and edges, and functional consequence. Through-cracks, deep edge chips, and electrically conductive inclusions are generally critical; cosmetic scratches within size limits may be minor. Criteria are defined in product specifications referencing ASTM, ISO, SEMI, GB, or JIS frameworks.

Q3: Why is scratch visibility used instead of scratch width for acceptance?

Visual appearance correlates with the light-scattering cross-section of a scratch, which depends on depth, morphology, and orientation—not width alone. Standards therefore grade scratches by comparative visibility under controlled lighting, supplemented by measured dimensions when needed.

Q4: Can contamination be distinguished from actual glass damage?

Yes. Combination of microscopy, profilometry, and EDS elemental analysis determines whether a feature is removable residue, adherent foreign material, or permanent material loss or deformation of the glass itself.

Q5: How many samples and inspection points are required?

Sampling plans follow agreed statistical protocols, commonly based on lot size and acceptable quality levels (AQL). Full-surface automated scanning is preferred for high-value large-generation substrates because defect distributions are often non-uniform.