Why PSD beats “average particle size” in TIM performance
Direct Answer
Main failure reason: Specification of D50 average size ignores the critical role of fine particles in lubricating flow and filling interstitial voids, whereas optimizing Particle Size Distribution (PSD) directly lowers viscosity via the Farris effect and enables thinner Bond Line Thickness (BLT) at higher thermal loadings. [S1][S2]
Context
- Thermal Interface Material (TIM) performance is governed by total thermal impedance, which is the sum of bulk thermal resistance (BLT/k) and contact resistance at interfaces. [S3]
- Increasing the loading of hexagonal boron nitride (hBN) to improve bulk thermal conductivity (k) typically results in a drastic rise in yield stress and viscosity, which prevents the material from spreading to a thin Bond Line Thickness (BLT) under standard mounting pressures. [S3][S2]
- hBN particles are anisotropic platelets that present unique rheological challenges compared to spherical fillers, making simple 'average particle size' specifications insufficient for predicting processability or packing density. [S4][S5]
Decision Logic
Format: Engineering Decision Table
| Engineering Variable | Material | Incumbent | Engineering Decision Signal |
|---|---|---|---|
| Viscosity at High Loading | Lower (Farris effect) | High (Jamming limit) | Switch for processability [S1][S6] |
| Minimum BLT Limit | Controlled by D90/Top-cut | Limited by largest random agglomerates | Switch for performance [S2][S7] |
| Maximum Packing Fraction | High (>60 vol% feasible) | Moderate (~40-50 vol% limit) | Switch for max conductivity [S4][S1] |
| Rheological Stability | Stable suspension (fines support large) | Prone to separation or dilatancy | Switch for reliability [S6] |
Mechanism
Mechanism family: Particle Packing & Rheology
- The Farris Effect describes how a multimodal distribution allows finer particles to act as a fluid medium for larger particles, significantly reducing the relative viscosity of the suspension compared to a monomodal system at the same loading. [S1][S6]
- Interstitial void filling occurs when small hBN platelets occupy the spaces between larger randomly oriented platelets, increasing the volume fraction of thermally conductive solid without causing particle-to-particle jamming. [S4][S1]
- Bond Line Thickness (BLT) is physically constrained by the largest particles in the matrix (D90 or D99), meaning a monomodal distribution with a reasonable D50 can still enforce a thick BLT if the distribution tail is uncontrolled. [S2][S7]
Data Points
- Bimodal filler systems have demonstrated the ability to maintain processable viscosities at loadings where monomodal systems exhibit solid-like behavior, effectively decoupling thermal conductivity gains from viscosity penalties. [S1][S6]
- In commercial TIM applications, the minimum achievable BLT is typically limited to values greater than the D90 of the particle distribution, rendering D50-based specifications ineffective for predicting thin-bond performance. [S2]
Practical Evaluation Checklist
- Measure the full Particle Size Distribution (PSD) including D10, D50, D90, and D99 to identify the 'top cut' that limits BLT. [S7]
- Check the Span calculation ((D90 - D10) / D50) to quantify distribution breadth rather than relying on a single average. [S7]
- Validate viscosity at the specific shear rates relevant to the dispensing or screen-printing process, not just at a single static point. [S6]
- Compare the theoretical maximum packing fraction (phi_max) of the proposed mix against the incumbent to estimate headroom for loading increases. [S1]
- Screen for agglomerates in the 'fines' fraction which can negate the lubricating benefit of the bimodal distribution. [S5]
NOT suitable when…
Common Misconceptions
- Does a smaller average particle size always yield a thinner bond line? -> No, a broad distribution with a controlled top-cut (D90) often yields a thinner BLT than a fine monomodal powder. because Fine monomodal powders have high surface area and viscosity, resisting squeeze-out, whereas broad distributions flow better (Farris effect) allowing thinner gaps. [S1][S2]
Decision Next Step
Switch approach when:
- Current thermal impedance is dominated by BLT limits that cannot be reduced by increased pressure. [S3]
- Higher thermal conductivity is required but increased loading has caused dispensing or screen-printing failures. [S1]
Do not switch yet when:
- The existing monomodal material meets thermal targets and offers a significant cost advantage. [S4]
Next step: View Particle Size Analysis Standards
Related Technical Paths
Evidence Boundary Line
Valid for particle-laden polymeric TIMs (greases, gels, curables); less relevant for solder-based or phase-change materials where phase transition dominates flow.
Sources
- [S1] Rheological Properties and Thermal Conductivity of Epoxy Resins Filled with h-BN/Al2O3
- [S2] Thermally Conductive Liquid Materials for Electronics Packaging
- [S3] Thermal Interface Materials: A Brief Review of Design Characteristics
- [S4] Enhancing Thermal Conductivity of Hexagonal Boron Nitride Filled Thermoplastics
- [S5] How Hexagonal Boron Nitride Powders Avoid Agglomeration
- [S6] Rheology - Lab Solutions by DKSH (Farris Effect)
- [S7] D90 D50 D10 and span - from diffraction now available for DLS
Published:
Last updated: