Solution

Particle Characterization Solutions for the Semiconductor Industry: From Slurries to Packaging Materials – How Precise Particle Size & Shape Control Empowers Chip Manufacturing Yield

In semiconductor manufacturing, a single particle invisible to the naked eye can destroy a chip worth hundreds of dollars. A few oversized particles in a polishing slurry can scratch the wafer surface; irregular particles mixed into molding compound can cause delamination or lead deformation; insufficient sphericity of thermally conductive fillers significantly reduces the thermal efficiency of thermal interface materials. As line widths continue to shrink and chip integration densities rise, the control of particle size, shape, distribution, and cleanliness has evolved from a "quality metric" to a "process red line." Mipu Technology integrates laser diffraction, static/dynamic image analysis, microflow imaging, and nanoparticle size analysis into the front lines of semiconductor materials R&D and production, using real particle data to help you answer: Where do the oversized particles in my slurry come from? How do I ensure batch-to-batch consistency of packaging materials? Does my thermally conductive powder meet sphericity specifications? The following sections explore core semiconductor materials and demonstrate how particle characterization safeguards chip manufacturing yield.

Unique Challenges and Testing Logic for Semiconductor Particle Characterization

1.1 Ultra-Narrow Distribution and Trace Oversized Particles: The Core Pain Point of Slurries

Chemical Mechanical Planarization (CMP) is a critical process for wafer global planarization. The abrasive particles in CMP slurries (e.g., silica, ceria, alumina) must satisfy two stringent requirements: first, an extremely narrow particle size distribution (Span < 0.8) to ensure uniform polishing rate and consistent surface roughness; second, absolutely no particles exceeding a certain threshold (typically 1–5 μm). A single oversized particle acts like a "grinding wheel" during polishing, leaving deep scratches on the wafer surface and scrapping the chip. However, conventional laser particle size analyzers are extremely insensitive to trace oversized particles (ppm level) — in volume distribution, a few oversized particles contribute negligibly and are easily masked by the main peak. The solution: combined laser diffraction and microflow imaging, obtaining both volume distribution and true particle count in a single injection.

1.2 Sphericity and Particle Size Distribution: Key Metrics for Thermally Conductive Fillers

Thermal Interface Materials (TIMs) fill microscopic gaps between chips and heat sinks. The particle morphology of thermally conductive fillers (e.g., alumina, aluminum nitride, spherical silica) directly determines thermal resistance. Spherical particles achieve maximum packing density with minimal contact thermal resistance; acicular or flake-like particles tend to bridge within the matrix, creating voids and increasing thermal resistance by over 30%. Static image analysis can statistically evaluate sphericity distribution across at least 100,000 particles (sphericity > 0.9 is excellent), guiding optimization of spray granulation, flame fusion, or jet milling processes.

1.3 Metallic Contaminants and Irregular Particles: Zero Tolerance in Packaging Materials

In epoxy molding compounds (EMC) and underfill materials, foreign matter such as metallic contaminants (copper, iron, zinc), fibers, and hard agglomerates can cause lead short circuits, delamination, electrochemical corrosion, or stress concentration. These foreign particles typically range from 1–100 μm (sub-visible range). Traditional light obscuration methods cannot distinguish bubbles from solid particles, let alone identify particle types. Microflow imaging automatically captures particle images and classifies foreign matter types via AI algorithms, achieving ppm-level detection and providing critical data support for packaging reliability.

1.4 Nanoparticle Size and Dispersibility: Photoresists and Slurries

Pigment or photosensitive resin particles in photoresists, and nano-silica in colloidal silica slurries, typically range from 10–300 nm. Oversized or agglomerated particles cause coating defects, non-uniform exposure, or wafer scratching. Nanoparticle size analyzers (Dynamic Light Scattering, DLS) precisely measure hydrodynamic particle size (D10, D50, D90) and polydispersity index (PDI), with PDI < 0.1 indicating excellent monodispersity and uniform particle distribution.

Testing Logic at a Glance: Laser diffraction for primary particle size of slurries; microflow imaging in tandem for trace oversized particles; static image analysis for sphericity of thermally conductive powders; microflow imaging for foreign matter in packaging materials; nanoparticle size analysis for nano-scale particles.

CMP Slurries: Comprehensive Control from Primary Particle Size to Trace Oversized Particles

2.1 Material Characteristics and Testing Challenges

CMP slurries typically consist of abrasive particles, oxidizers, complexing agents, and surfactants. The particle size distribution of abrasive particles determines Polishing Rate (PR) and wafer surface roughness (Ra). For silica slurries, typical requirements are D50 of 80–120 nm, D90 < 150 nm, Span < 0.8, and particle count >1 μm must be below 100 particles/mL. The pain point: during laser diffraction testing, a small number of oversized particles contribute negligibly to the volume distribution and are easily overlooked; however, microflow imaging alone cannot provide accurate volume distribution. Only by combining both techniques can the dual requirements of "precise main peak" and "resolved tail" be simultaneously satisfied.

2.2 Mipu Solution: Laser Particle Size Analyzer + Microflow Imaging in Tandem

• Primary Particle Size Distribution: Laser particle size analyzer with wet dispersion, using filtered particle-free deionized water or dedicated diluent, with ultrasonication to break agglomerates. Rapidly measures D10, D50, D90, and Span with repeatability error ≤ ±1%.

• Trace Oversized Particle Detection: The same dispersed sample stream is split to the microflow imaging analyzer, which automatically captures images of all particles >0.5 μm, counts oversized particles per milliliter (particles/mL), and identifies particle morphology (spherical, irregular, agglomerate, metal debris).

• Data Integration: Software automatically integrates laser diffraction volume distribution with microflow imaging particle count, generating a complete particle size QC report with oversized particle alarm thresholds.

• Advantages: No additional sample preparation, no increased testing time, ppm-level sensitivity, meeting ultra-clean requirements of advanced nodes (e.g., 14nm, sub-7nm).

2.3 Case Study: Tracing Oversized Particles in Ceria Slurry

A semiconductor materials manufacturer producing ceria slurry for oxide CMP received customer feedback that wafer scratch rate had suddenly risen to 1.2%. Internal QC showed laser diffraction results of D90 = 0.85 μm (passing), but combined microflow imaging revealed 20 particles >2 μm per milliliter, with irregular blocky morphology and a few agglomerates. Further traceback identified these oversized particles as hard agglomerates from the raw ceria powder that were not broken during dispersion due to insufficient ultrasonication power. After increasing ultrasonication power from 300W to 500W and extending dispersion time, oversized particles dropped to 2 particles/mL, wafer scratch rate fell to 0.1%, and annual losses of millions of yuan were recovered.

2.4 Standard Compliance

Compliant with SEMI standards (e.g., SEMI C23, C29), GB/T 19077, ISO 13320, etc.

Thermally Conductive Fillers: How Sphericity Determines TIM Performance

3.1 Material Characteristics and Testing Challenges

Thermally conductive fillers (spherical alumina, aluminum nitride, spherical silica) are widely used in thermal pads, thermal greases, and thermal encapsulants. Particle sphericity directly affects packing density and interfacial thermal resistance. Studies show that increasing sphericity from 0.85 to 0.95 can increase maximum packing fraction from 0.65 to 0.75, improving thermal conductivity by 30–50%. Industry pain points: traditional microscopy observes only dozens to hundreds of particles per view, offering poor statistical representation; manual visual assessment is subjective and cannot quantify sphericity distribution.

3.2 Mipu Solution: Static Image Analysis (Particle Size & Shape Analyzer)

• Sampling and Spreading: A small amount of dry powder (approximately 10–20 mg) is uniformly spread on a glass slide using a vibratory feeder to avoid particle overlap.

• Testing: A high-resolution microscope camera captures images of at least 100,000 particles; AI automatically calculates sphericity (ratio of projected area to perimeter of equivalent circle), circularity, aspect ratio, and convexity for each particle.

• Output Parameters: Sphericity distribution histogram, mean sphericity, proportion of particles with sphericity <0.85 (non-conformance rate), proportion of particles with aspect ratio >2.

• Process Guidance: When sphericity is low, adjust flame fusion temperature (higher temperature yields rounder particles), spray granulation parameters (optimize slurry concentration and atomization pressure), or jet milling pressure (avoid excessive breakage).

3.3 Case Study: Thermal Conductivity Improvement of Spherical Silica Powder

A manufacturer producing spherical silica powder for 5G base station thermal gels received customer feedback that thermal conductivity was only 2.5 W/m·K, below the target of 3.5 W/m·K. Static image analysis showed mean sphericity of only 0.82, with 15% of particles having aspect ratio >2 (elliptical or rod-shaped). After CFD simulation optimization of the flame fusion nozzle and adjustment of flame temperature distribution and particle residence time, the new process achieved mean sphericity of 0.94, aspect ratio >2 proportion reduced to 2%, thermal conductivity increased to 3.8 W/m·K, and the product successfully entered the Huawei supply chain.

3.4 Standard Compliance

Compliant with GB/T 31057.2 (Determination of Particle Sphericity), ASTM F1877, JIS R 1639, etc.

Packaging Materials: Foreign Matter Control in EMC and Underfill

4.1 Material Characteristics and Testing Challenges

In epoxy molding compounds (EMC) and underfill materials, foreign matter such as metallic contaminants (copper, iron, zinc), fibers, and hard agglomerates can cause multiple failure modes: metal particles may cause short circuits between leads; fibers may impede resin flow, creating voids; hard agglomerates cause stress concentration leading to delamination. These foreign particles typically range from 1–100 μm at extremely low concentrations (ppm level). Traditional light obscuration methods cannot distinguish bubbles from solid particles nor identify particle types, leading to false positives or missed detections.

4.2 Mipu Solution: Microflow Imaging + Static Image Analysis

• Liquid Samples (Underfill, Epoxy Resin): Direct or diluted sample is injected into the microflow imaging microfluidic chip, automatically capturing images of all particles in the 1–100 μm range. AI deep learning models classify particles by morphological features: Metals (bright, irregular, reflective), Fibers (aspect ratio >5, uniform width), Bubbles (circular, transparent, bright edge), Agglomerates (multiple particles adhered), Other foreign matter.

• Powder Samples (EMC Powder): Small amount of dry powder is analyzed by static image analysis for rapid screening, focusing on metallic-luster and off-color particles; can detect foreign matter >10 μm.

• Combined Strategy: For high-value products (e.g., advanced packaging EMC), pairing laser particle size analysis with microflow imaging is recommended — two functions, one instrument, no additional sample preparation.

• Data Traceability: All foreign matter images are automatically saved for later review and source tracing.

4.3 Case Study: Eliminating Metal Particles in Underfill

An underfill manufacturer received customer feedback regarding occasional needle clogging during dispensing (approximately once per 1,000 chips). Microflow imaging analysis revealed approximately 50 metal particles >10 μm per milliliter of underfill; EDS energy spectroscopy confirmed the particles were primarily copper debris. After investigating the raw material transfer line, a section of stainless steel corrugated pipe was found to have internal wall wear and flaking. After replacing with a ceramic-lined pipe and adding an in-line 5 μm filter, metal particles dropped to zero, clogging was completely resolved, and customer orders were restored.

4.4 Standard Compliance

Compliant with IPC/JEDEC J-STD-020, SEMI G74, IEC 62321, etc.

Photoresists and Slurries: Dispersion Stability of Nanoparticles

5.1 Material Characteristics and Testing Challenges

Pigment or photosensitive resin particles in photoresists, and nano-silica in colloidal silica slurries, typically range from 10–300 nm. Oversized or agglomerated particles cause: streaks or pinholes during photoresist coating; increased light scattering during exposure, degrading line-width uniformity; wafer scratching from oversized particles in slurries. Precise measurement of D50, D90, and PDI (polydispersity index) is required, with PDI < 0.1 indicating excellent monodispersity and PDI > 0.2 indicating agglomeration or multimodal distribution.

5.2 Mipu Solution: Nanoparticle Size Analyzer (Dynamic Light Scattering)

• Particle Size and Distribution: DLS measures hydrodynamic particle size (D10, D50, D90) and PDI. Samples are diluted to appropriate concentration (typically 1–10 mg/mL) with filtered solvent (0.1 μm filter membrane) to avoid multiple scattering and particle interactions.

• Sample Cell: High-precision quartz cuvette, temperature controlled at 25 ± 0.1°C to eliminate temperature effects on viscosity.

• Repeatability: Deviation <2% across three measurements.

• Process Guidance: When PDI is high, optimize ultrasonication time, increase dispersant concentration, or adjust pH.

5.3 Case Study: Eliminating Agglomerates in Photoresist

An i-line photoresist manufacturer observed tens-of-micrometer-sized spots after coating, causing yield loss. Nanoparticle size analysis showed D50 = 120 nm (passing), but PDI = 0.28 (distribution too broad), indicating the presence of agglomerates. Further analysis revealed insufficient ultrasonication time (only 5 minutes). After extending ultrasonication to 15 minutes, PDI dropped to 0.09, coating defect rate fell from 5% to 1%, and the product passed customer qualification.

5.4 Standard Compliance

Compliant with SEMI P19 (Photoresist Particle Size Testing), ISO 22412, etc.

Other Semiconductor Materials

Combined Solution Note: For samples such as diamond slurries that require both particle size distribution and oversized particle detection, pairing laser diffraction with microflow imaging is recommended for simultaneous volume distribution and large-particle imaging, preventing missed detections. For solder paste, particles with insufficient sphericity cause short circuits or cold solder joints — static image analysis is a critical QC tool.

Full-Process Services: From Method Selection to Particle Source Tracing

Mipu Technology provides a closed-loop "Consultation – Testing – Analysis – Improvement" service:

1. Free Pre-Testing: For your specific samples (slurries, thermal powders, EMC, photoresists, etc.), we recommend the optimal testing method (laser/image/nano/tandem) and provide a preliminary report.

2. Method Development and Validation: Establish Standard Operating Procedures (SOPs), determine dispersion conditions (ultrasonication power, time, dilution ratio, dispersant type), and validate against SEMI standards or customer internal specifications.

3. Instrument Delivery and Training: On-site installation, commissioning, and operator training, ensuring data complies with cleanroom requirements (Class 1000 or higher). IQ/OQ/PQ validation services provided.

4. In-Depth Data Interpretation: Beyond particle size data, we provide process correlation analysis — for example: how does polishing rate change with every 10 nm increase in D90? How much does thermal resistance increase with every 0.05 decrease in sphericity? Foreign matter images can be used directly for supplier audits and internal accountability.

5. Online System Integration: For continuous production lines of slurries or grinding fluids, online laser particle size analyzers can be integrated with MES systems for real-time monitoring, automatic alarm, and historical data traceability.

Conclusion: From Particle Size to Morphology, From Distribution to Foreign Matter