In TGV glass via wet etching, incomplete penetration and trapped bubbles are core bottlenecks limiting yield. Mass production verifications show most through failures stem from bubble blockage instead of insufficient etchant concentration. This paper classifies bubbles into entrained air, reaction-generated gas and dissolved outgassing, clarifies their different impacts on penetration yield and via profile. A practical 4-stage programmed vacuum degassing workflow, equipment specifications, yield acceptance criteria and low-cost auxiliary improvements are provided to stably lift mass yield above 99.5%.
The dominant root cause of incomplete penetration in deep TGV wet etching is trapped bubbles inside vias. Trapped gas falls into three categories: entrained air, reaction-generated gas and dissolved outgassing. Entrained air determines whether a via can be fully penetrated, while the other two mainly govern via profile and sidewall uniformity. To raise yield from the baseline of 60%~85% (without optimized degassing) to over 99.5%, the key is not to boost HF concentration, but to implement a programmed 4-stage vacuum degassing system.
Bubble Sources and Failure Mechanisms
Entrained air: When the substrate enters the bath, the via opening is instantly sealed by liquid. Internal air cannot escape and gets enclosed by etchant. Hydrophobic masks (photoresist, gold mask) aggravate this sealing effect. Air accumulates at the via bottom and forms a stable air cushion, halting etching and causing open through failure.
Reaction-generated gas: SiF₄ produced during etching hydrolyzes into microbubbles. Meanwhile, residual carbon from laser-modified layers reacts with HF and continuously generates gas. These bubbles mostly attach to the lower sidewall and via bottom corners, triggering abnormal bottom rounding and sidewall pits.
Dissolved outgassing: With rising bath temperature, dissolved air becomes supersaturated and precipitates randomly as tiny bubbles on sidewalls. This creates localized etch blocking, manifesting as pockmarks and uneven etching.
Vacuum degassing is the mainstream engineering countermeasure. Under isothermal conditions, for a bubble of radius r, when pressure drops from atmospheric pressure P₀ to absolute pressure P, bubble volume expands by a factor of P₀/P, and buoyancy rises proportionally. More importantly, stable air cushions at via bottoms can be removed via repeated evacuation – venting cycles. During evacuation, bubbles expand and part of the gas is extracted. During venting, fresh etchant is forced into vias by pressure difference to replace residual gas. One evacuation-vent cycle performs a forced liquid exchange inside deep vias. After N cycles, residual gas decays exponentially with the retention ratio per cycle. This explains why mature degassing adopts intermittent cycling rather than constant continuous evacuation.
Mass Production Equipment Specification
A circulation bypass vacuum degassing setup is preferred over direct whole-tank evacuation. An independent PFA degassing tank accounts for 5%~10% of the main etch tank volume, operating at absolute pressure 20–40 kPa. The main etch tank remains at atmospheric pressure, free from boiling-point limitations under reduced pressure. Degassing runs continuously alongside etching without consuming extra tact time.
Three non-negotiable equipment rules:
Seals contacting HF etchant shall use FFKM perfluoroelastomer. Standard FKM fluoroelastomer suffers swelling and degradation under long-term hot HF exposure.
Oil-sealed vacuum pumps are forbidden, to prevent oil back-contamination and oil degradation upon contact with HF.
The degassing tank must be equipped with a rupture disc and dedicated vent piping for negative-pressure safety protection.
4-Stage Programmed Degassing Process (Adjustable by Aperture & Aspect Ratio)
P1: Immersion & wetting: Step dipping plus pre-spraying etchant on via surfaces to break the instant sealing air cushion at via openings and reduce entrained air from the start.
P2: Pre-degassing: Evacuate to 40 kPa, hold for 20 s then vent. Repeat 5–10 cycles to remove primary entrained air at via bottom, the critical step for penetration yield.
P3: In-etch degassing: Maintain 30–40 kPa in the bypass degassing tank continuously. Insert 2-minute megasonic pulses every 30 minutes to suppress reaction bubble adhesion.
P4: Final intensive degassing: During the last 10%~15% of etch duration before breakthrough, drop pressure to 25 kPa and run 10–20 evacuation-vent cycles to eliminate residual air cushions at breakthrough threshold.
For 50 μm aperture, 10:1 aspect ratio TGV, continuous megasonic activation and ±15° tilt fixture are recommended.
Closed-Loop Yield Acceptance & Troubleshooting Criteria
Three combined indicators for process control:
Coupon penetration yield ≥ 99.5%
For 80 μm aperture, aperture variation across 5 sampling points per panel ≤ ±3 μm
Post-etch IR backlight inspection: bubble-defect via ratio ≤ 0.5%
If any indicator drifts, troubleshoot in sequence: megasonic waveform & power → vacuum stability → circulation flow rate → bath aging and contaminant buildup.
Low-Cost Auxiliary Improvements
Pre-wetting station: Prior to etching, substrates pass through DIW shower or IPA pre-wetting bath. IPA has surface tension ~22 mN/m (1/3 of water), which cuts residual air inside vias by one order of magnitude.
Tilt fixture: Adjustable tilt angle 5~15°, aligning via openings with bubble rising direction. For aspect ratio ≥10:1, tilt fixture shall work together with vacuum degassing.
Frequently Asked Questions
Q1: Why not simply raise HF concentration or extend etch time to fix penetration failure?
A1: Higher HF concentration mainly boosts etch rate and cannot eliminate bubble-induced mass transport blockage. When bubbles separate etchant from glass, longer time or stronger acid will not etch the bottom. Instead, lateral undercut increases, via profile goes out of control and mask degradation accelerates. Bubble blocking is an interfacial mass transfer issue; liquid exchange and degassing shall be prioritized before tuning etchant parameters.
Q2: What is the core difference between whole-tank evacuation and bypass independent degassing tank?
A2: Whole-tank vacuum lowers boiling point, accelerates HF volatilization and shifts bath composition. Besides, the tank and fixtures need full vacuum-rated design with higher cost and safety risk. Bypass degassing only applies vacuum in the circulation loop while the main tank stays at ambient pressure. HF loss is well controlled and no extra tact time is added for mass production.
Q3: Does IPA pre-wetting bring safety hazard or bath contamination?
A3: IPA pre-wetting is a standalone upstream bath separated from HF etch tank. Proper overflow rinsing removes residual IPA. The station requires good ventilation and anti-static control. IPA must never be carried into the HF tank, otherwise impurity accumulation and abnormal side reactions will occur.
Q4: Does higher megasonic power always deliver better degassing?
A4: No. Excessive power causes mask fatigue and peeling, glass microcracks and substrate damage. Insufficient power cannot detach bubbles from sidewalls. Megasonic waveform uniformity shall be prioritized over power setting. Acceptance criteria are coupon yield and sidewall morphology, not maximum power.
Q5: Is this full 4-stage vacuum recipe required for aspect ratio below 5:1?
A5: Entrained air escapes more easily for AR＜5:1. The recipe can be simplified: keep P1 wetting and short P2 pre-degassing; P3 and P4 intensive vacuum cycles can be removed. For small aperture with hydrophobic mask, basic degassing is still recommended to avoid random bubble defects.
Q6: Will more vacuum cycles reduce throughput?
A6: This bypass degassing design runs evacuation-vent cycles in parallel with etching without extending main process time. Only P2 pre-degassing adds upstream duration, which can be offset by multi-station loading. The yield loss from rework and low penetration rate costs far more than the small overhead of degassing.
Q7: Is aperture variation always caused by bubbles?
A7: Not necessarily. Bubbles cause localized stagnation and uneven aperture. Other root causes include bath temperature gradient, flow dead zone, mask thickness variation and glass material inconsistency. Check bubble defect ratio first, then investigate flow and temperature uniformity.
Q8: Can FFKM seals last permanently in HF service?
A8: No. FFKM is the preferred seal material for HF, but it still ages under long immersion and temperature cycling. Periodic inspection and replacement schedule are mandatory. Standard FKM is prohibited for long-term hot HF service due to rapid swelling and cracking.
