Function in the Complete Flow
In a modern 40nm backside illuminated (BSI) complementary metal-oxide-semiconductor (CMOS) image sensor (CIS), the optical stack above the silicon substrate dictates how efficiently incident photons are collected and converted into electrical charge. Sitting at the very top of this optical stack, the microlens array serves as the primary light-gathering optic. Its sole functional objective is to focus incoming illumination through the color filter array (CFA) and upper optical clear-layer (UOCL) directly into the light-sensitive region of the deep-lying pinned photodiode (PPD).
The upper OCL reflow process operates as a critical shape-forming transformation step within the back-end optical module. Before reflow occurs, photolithography patterns isolated cylindrical or rectangular polymer blocks on top of the planarized underlying optical layers. However, sharp-cornered polymer blocks cannot bend or focus light efficiently; unshaped structures induce significant optical diffraction, back-reflection, and light loss.
The upper OCL reflow step receives these lithographically defined polymer blocks, subjects them to a controlled thermal cycle, and converts the sharp rectangular islands into smooth, spherical or hyper-spherical refractive lenses through surface-tension-driven fluid flow. Thermal reflow occurs because the material reorganizes to minimize the total surface energy of the system . Once formed, this optimized microlens profile is handed off to final wafer-level passivation or UV/hard bake, establishing the definitive optical collection efficiency, spatial light concentration, and chief-ray angle (CRA) response of the entire 40nm BSI CMOS image sensor.
For a broader understanding of how this step integrates into the complete manufacturing sequence, see the overview of the 40nm BSI CMOS Image Sensor process flow.
[Patterned Polymer Islands] ──> [Thermal Reflow Treatment] ──> [Hemispherical Microlenses]
(Flat Top, Sharp Edges) (Viscoelastic Surface Tension) (Smooth Refractive Profile)
Process checkpoint
Understand Upper OCL Reflow in context
Understand the mechanism and integration handoff at UOCL in the 40nm BSI CMOS Image Sensor.
Process context for “Mastering the Upper OCL Reflow Mechanism in 40nm BSI CMOS Image Sensors: Physics, Surface Free Energy, and Optical Integration”: 40nm BSI CMOS Image Sensor · UOCL · Step 378
Explore Chapter 6 in the foundation guideUpstream Input State
The reflow mechanism does not operate in isolation; its success depends entirely on the structural, chemical, and topographical properties inherited from preceding process steps. When the wafer arrives at the thermal reflow module, it exhibits a complex, pre-conditioned surface state:
- Substrate Topography and Planarization: The underlying layer—typically an organic upper optical clear-layer (UOCL) deposited over the patterned color filter array—must present a flat, planarized surface. A smooth surface topography has less total surface area and fewer unsatisfied bonds than a rough structure with sharp steps or edges . Any local height variation or micro-tilt in the UOCL propagates directly into the overlying lens profile, causing localized focal length shifts or optical axis distortion.
- Surface Energy and Wettability: The chemical interface between the patterned microlens polymer and the underlying UOCL material governs the equilibrium contact angle. Prior surface treatments (such as chemical cleans, vacuum bakes, or plasma treatments) tune the substrate surface energy to establish an exact wetting balance.
- Polymer Volume and Aspect Ratio: Photolithography defines the initial block volume, width, and height of the polymer islands. Because volume is conserved during fluid flow (excluding minor solvent outgassing), the initial photoresist film thickness and critical dimension (CD) directly set the ultimate lens sag height and curvature radius.
- Capillary Pinning Boundaries: The bottom perimeter of each patterned block forms a contact line with the underlying substrate. Precise sidewall angles from the lithography step are required so that capillary pinning forces lock the wetting edge in place during thermal softening, preventing uncontrolled lateral spreading into adjacent pixel spaces.
Defects in any of these upstream inputs—such as residual topography, inconsistent surface free energy, or non-uniform lithographic footprint sizes—severely compromise the symmetry and radius of curvature formed during thermal reflow. Details regarding these preceding layer preparations can be explored in the 40nm BSI CMOS Image Sensor upper optical clear-layer and microlens integration process flow.
Physical and Chemical Mechanisms
The transformation of a flat-topped polymer block into a curved optical lens during the upper OCL reflow process relies on fundamental principles of polymer physics, thermodynamics, and fluid mechanics.
Surface Tension Forces (γ_LV)
↓ ↓ ↓ ↓
. - - - - - - - .
. .
/ \
| Viscoelastic Polymer |
______\____/______
\\\\\\ (Pinned Base) (Pinned Base) //////
=================================================
UOCL Substrate
Thermal Glass Transition and Viscoelastic Flow
When the patterned polymer structures are subjected to an elevated thermal treatment, the material temperature surpasses its glass transition temperature (Tg). Below Tg, the polymer exists in a rigid, glassy state where long-chain molecular segments are frozen in place. As the temperature rises above Tg, thermal energy overcomes secondary intermolecular bonds, enabling large-scale segmental motion of the polymer chains.
The polymer transitions from a glassy solid into a viscoelastic liquid with significantly reduced shear viscosity. For amorphous material systems, viscous flow acts as a primary mass transport mechanism during thermal softening . In this viscous regime, polymer chains gain sufficient mobility to move driven by chemical potential gradients. Simultaneously, residual solvent molecules trapped within the matrix outgas into the ambient environment, causing a minor, predictable volume shrinkage of the lens structure during fluid flow.
Surface Energy Minimization and Capillary Pinning
The primary physical driving force behind microlens shape formation in CIS is the minimization of total surface free energy. High-curvature edges and sharp right-angled corners contain a high density of unsatisfied intermolecular bonds, creating a high surface energy state. To reach thermodynamic equilibrium, the viscous fluid spontaneously redistributes its mass to minimize its total exposed surface area for a given enclosed volume.
The mathematical condition governing equilibrium at the three-phase contact line (solid substrate, liquid polymer, and gas ambient) is expressed by Young's relation:
γ_SV = γ_SL + γ_LV * cos(θ_c)
Where γ_SV is the solid-vapor surface energy, γ_SL is the solid-liquid interfacial energy, γ_LV is the liquid-vapor surface tension, and θ_c is the contact angle.
However, complete wetting to the natural contact angle does not occur because of capillary edge pinning. The sharp geometric step at the base of the lithographically patterned block acts as a physical boundary obstacle. The liquid boundary becomes locked—or pinned—at this outer perimeter. Because the base area is held constant by pinning while surface tension drives the free liquid surface into an energy-minimizing shape, the liquid naturally adopts a smooth, constant-curvature spherical or hyper-spherical cap profile.
Profile Tuning and Aspherical Radius Control
The final optical performance of the lens depends on its exact three-dimensional shape, which governs how light rays refract at the interface. For a thin parabolic or spherical microlens, the focal length f is related to the lens radius r_lens, the central sag height d_lens, and the material refractive index n by the geometric relation:
f = (d_lens^2 + r_lens^2) / (2 * d_lens * (n - 1))
During the reflow step, engineers control the sag height d_lens and focal distance by tuning the thermal profile, ambient atmosphere, and heating duration. Extended thermal treatment at higher temperatures lowers the melt viscosity, allowing surface tension to fully round the top center of the lens.
When volume shrinkage occurs concurrently with boundary pinning, the lens surface profile subtly deviates from a simple sphere, adopting a hyper-spherical or hyperbolic distribution. This hyperbolic profile reduces spherical aberration, focusing light more tightly into the active region of the BSI pixel.
Downstream Impact and Failure Propagation
Because the microlens is the absolute front-door interface for light entering the image sensor, minor structural or geometric flaws introduced during the reflow process propagate severe optical and electrical failures down the device chain.
Reflow Deviation Physical Manifestation Device-Level Impact
──────────────── ────────────────────── ───────────────────
Incomplete Reflow ──> Flat top / Low Sag Height ──> Focal point behind silicon; Low QE
Over-Reflow ──> Bridging / Gap Closure ──> Severe optical crosstalk between pixels
Asymmetrical Heating ──> Lens Tilt / Decentered Peak ──> CRA mismatch; Spatial shading artifacts
Optical Focusing and Chief-Ray Angle Integration
In BSI sensor architectures, light enters directly through the thinned silicon substrate toward the embedded pinned photodiode. Photons arriving at the outer edges of the sensor array hit the surface at steep angles, known as high chief-ray angles (CRA).
To compensate for high CRA, microlenses across the sensor periphery must be spatially shifted relative to the underlying photodiode centers—a technique called CRA shift-reflow optimization. If the reflow process distorts the lens curvature or fails to maintain precise radial symmetry, the focal spot drifts off target, missing the active photodiode area and landing on isolation structures or adjacent pixel channels.
Failure Modes and Device-Level Consequences
- Under-Reflow (Incomplete Reflow): If the thermal energy or reflow duration is insufficient, the polymer fails to reach lower viscosity, leaving a flat plateau at the lens apex. This results in an extremely long focal length f, placing the focal point deep beneath the active silicon region. The resulting loss of light concentration dramatically degrades the sensor's quantum efficiency (QE) and sensitivity under low-light conditions.
- Over-Reflow and Microlens Bridging: Excessive thermal exposure or overly high temperature reduces polymer viscosity too far, overwhelming the capillary pinning forces at the base. The liquid polymer breaks past the pinning boundary and flows into adjacent pixel channels, causing neighboring lenses to touch and bridge. Lens bridging introduces optical crosstalk, where light intended for one pixel spills into neighboring color channels, causing severe color contamination and loss of spatial resolution.
- Lens Asymmetry and Thermal Gradient Distortion: Non-uniform heating across the wafer causes localized variations in polymer viscosity and surface tension. This produces asymmetric lenses with tilted optical axes. Tilted lenses cause direction-dependent light collection, manifesting as spatial shading, color cast across the image frame, and reduced dynamic range.
Walk the Real Step
To see where this specific process step resides within the complete industrial manufacturing flow for a 40nm BSI CIS module, navigate to the step link below:
Within Step 378, the wafer enters the thermal processing chamber following the lithographic patterning of the microlens polymer blocks. The thermal sequence applies a strictly controlled bake to induce viscoelastic melting, driving surface-tension rounding while maintaining base contact pinning on the underlying upper optical clear-layer. The output of this step defines the final curvature, focal length, and optical aperture of the sensor array.
Related Learning Paths
To further expand your understanding of optical stack integration and advanced sensor architecture, explore these closely related integration guides:
- 40nm BSI CMOS Image Sensor Process Flow: Integration Principles, Device Physics, and Module Dependencies A comprehensive deep dive into the end-to-end integration logic, pixel isolation, and silicon processing of 40nm BSI image sensors.
- 40nm BSI CMOS Image Sensor Upper Optical Clear-Layer and Microlens Integration: Process Flow Principles and Physics Detailed coverage of the underlying dielectric planarization layers, color filter processing, and interface engineering required directly beneath the reflowed microlens module.
Future Outlook
As consumer and industrial applications demand smaller pixel pitches below sub-micron scales, traditional polymer thermal reflow encounters fundamental physical limits. At ultra-scaled dimensions, the gap between adjacent microlenses vanishes, leaving an ultra-narrow margin where traditional thermal reflow inevitably causes lens bridging and severe optical crosstalk.
To overcome these physical constraints, several advanced manufacturing alternatives are emerging:
- Grayscale Hard-Mask Pattern Transfer: Instead of relying solely on polymer reflow to define the final refractive shape, low-contrast grayscale lithography creates continuous 3D surface profiles in resist, which are subsequently transferred into high-refractive-index inorganic materials (e.g., silicon nitride or oxide) via reactive ion etching (RIE) with precise etch-selectivity tuning. This decoupling eliminates thermal boundary runout and provides ultra-stable optical properties.
- Planar Sub-Wavelength Metalenses: Replacing curved refractive lenses entirely, planar metasurfaces employ sub-wavelength inorganic nanostructures (such as TiO2 or SiN nanopillars) to modulate light phase, polarization, and wavefront shape directly. Metalenses eliminate the need for thermal reflow processing while completely removing chromatic aberration and enabling flat optical profiles for 3D-stacked sensor architectures.
- Zero-Gap Microlens Arrays: Advanced dry-etch transfer techniques utilize sacrificial polymer reflow combined with highly anisotropic etch steps to form zero-gap microlens arrays, maximizing the optical collection fill factor up to total area coverage without risk of fluidic bridging.
References
Silicon VLSI Technology - Full
James D. Plummer, Michael D. Deal, Peter B. Griffin
Silicon VLSI Technology · ISBN 978-0130850379