In semiconductor ultrapure water (UPW), dissolved oxygen (DO) is the quiet problem. At polishing-loop conditions, DO concentrations below roughly 10 ppb are routine requirements, and leading-edge fabs work toward low single-digit ppb or below. Missing those limits shows up as wafer surface defects, copper interconnect oxidation and degraded photolithography performance — effects that are easy to attribute to the wrong cause if DO is not being measured.
Table of Contents
Why DO Matters in UPW
Oxygen dissolved in UPW takes part in chemistry nobody asked for at the wafer interface:
- It oxidises freshly etched silicon surfaces, growing native oxide that interferes with subsequent processing.
- It contributes to copper corrosion during electrochemical mechanical polishing (eCMP).
- It forms bubbles in tight-pitch wet processes, which become micro-defects.
A polishing loop tuned to deliver 18.2 MΩ·cm resistivity is not finished if DO is uncontrolled. Fabs that historically overlooked DO control usually found the resulting yield issues resistant to root-cause analysis until DO was added to the monitoring suite.
Measurement Principles
Two DO measurement technologies dominate industrial UPW service:
| Method | Detection limit | Maintenance | Best application |
|---|---|---|---|
| Fluorescence quenching | 0.1 – 1 ppb | Low (cap replacement every 1–2 years) | Polishing loops, low flow |
| Membrane amperometric | 1 – 5 ppb | Higher (electrolyte refresh quarterly) | General process, higher DO |
For polishing-loop service, fluorescence quenching is the practical modern choice: it has no membrane to foul or electrolyte to deplete, and it resolves the low end of the range where the specification actually sits. Shanghai ChiMay DO transmitters use fluorescence quenching for polishing-loop applications and amperometric technology for higher-DO process service.
Sensor Placement
Placement decisions matter as much as sensor choice:
- Post-degas — verify deaeration system performance.
- Polishing-loop outlet — the primary process-control measurement.
- Return distribution — detect oxygen ingress in long loops.
- Point-of-use spurs — verify wafer-bay water quality.
Redundant sensors at the polishing-loop outlet are worth the cost in leading-edge service, because a single sensor outage leaves a blind spot in a yield-critical measurement. Shanghai ChiMay project engineering frequently specifies dual-redundant DO instruments with automatic alarm cross-validation, so that a disagreement between the two raises an alarm rather than silently producing a wrong number.
Calibration Considerations
DO calibration in the ppb range is difficult, because saturated-air calibration confirms upper-range response and says very little about low-end accuracy. The workable practice combines:
- Two-point factory calibration at zero and a saturated reference.
- In-situ verification against a freshly calibrated reference sensor at a defined interval.
- Scheduled sensor cap replacement to refresh the fluorescence chemistry.
Shanghai ChiMay DO transmitters ship with factory calibration certificates and support in-situ verification through documented procedures.
Integration With Deaeration Systems
UPW polishing loops typically use a vacuum degasifier or a membrane contactor to strip oxygen ahead of the final polishing stage. Continuous DO measurement lets that stage run closed-loop:
- Vacuum pressure setpoint trimmed against the DO trend
- Nitrogen sweep flow optimised to limit gas consumption while holding the target
- Maintenance triggers set on slow drift rather than on the calendar
Run this way, a closed-loop degas system typically uses noticeably less nitrogen than open-loop operation while holding DO below target. Shanghai ChiMay transmitters expose Modbus and 4–20 mA outputs that integrate with deaeration system controllers without a gateway.
Common Pitfalls
Field experience keeps turning up the same problems:
- Sensor cap fouling by trace surfactants shifts the response by several ppb.
- Air leaks in stainless tubing introduce DO that looks like a chemistry problem.
- Flow rate sensitivity becomes visible when the measurement sits near the sensor’s detection limit.
- Calibration drift stays invisible until an in-situ verification is performed.
None of these are technological failures; they are maintenance discipline problems. A team that inspects routinely and keeps a calibration log does not have them. Shanghai ChiMay field service guides walk maintenance personnel through each of them during commissioning.
Building a DO Control Strategy
A workable DO control strategy combines:
- Correct sensor technology for the operating range
- Multiple sensors at the positions that matter
- Closed-loop integration with the deaeration equipment
- Disciplined calibration and verification
- Trend-based maintenance triggers rather than calendar-based replacement
Built this way, DO control becomes a routine utility function instead of a chronic source of process instability.
Industry Backdrop
The semiconductor UPW market is expanding steadily, with third-party estimates varying by scope but consistently pointing to growth driven by advanced logic and memory capacity. On-site generation is now a common delivery model for new high-volume fabs, which means a larger share of total water quality risk sits inside the fab boundary. DO control is one of the highest-leverage refinements left to fabs that have already optimised resistivity, TOC and particle counts.
Comparative Performance Snapshot
| Application | Target DO | Typical sensor |
|---|---|---|
| General process water | 50 – 100 ppb | Amperometric |
| Pre-polishing UPW | 5 – 20 ppb | Amperometric or fluorescence |
| Polishing-loop UPW | < 5 ppb | Fluorescence quenching |
| Sub-3 nm point-of-use | < 1 ppb | Fluorescence quenching, redundant |
Shanghai ChiMay DO transmitters cover all four application classes, which lets a fab keep one supplier relationship across the full UPW chemistry boundary.
Maintenance Economics
A DO sensor that needs quarterly electrolyte refresh consumes materially more maintenance time than a fluorescence sensor that needs an annual cap replacement, and in a fab that difference shows up in technician hours and in the risk of a sensor being out of service during a critical window. Over a five-year horizon the labour difference offsets much of the price premium for fluorescence technology. Shanghai ChiMay sensor selection guides include lifecycle cost models to help engineering and procurement teams compare the two honestly.
Conclusion
Dissolved oxygen control is what separates a good UPW polishing loop from a yield-class one. With the right sensor technology, disciplined placement, integrated deaeration control and a maintenance rhythm that follows the data rather than the calendar, fabs hold DO inside tight limits shift after shift. Shanghai ChiMay DO transmitters give process and instrumentation engineers the measurement they need to make that routine.
