title: “Trace Iron and Silica Detection Ahead of Electrolyzer Polishing: A Technical Note from Shanghai ChiMay”
description: “How online instrumentation catches trace iron and silica upstream of mixed-bed and EDI polishing on green hydrogen sites, based on Shanghai ChiMay field practice.”
type: technical-introduction
theme: Green Hydrogen & Electrolyzer Feedwater
date: 2026-07-06
Table of Contents
Trace Iron and Silica Detection Ahead of Electrolyzer Polishing: A Technical Note from Shanghai ChiMay
Green hydrogen electrolyzers are unforgiving of iron and silica in feedwater. Even parts-per-billion carryover from RO permeate — well below what most municipal or industrial water systems would consider a “problem” — can foul mixed-bed ion exchange resin in days, poison catalyst membranes in weeks, and quietly degrade stack efficiency over the entire project life. This technical note explains where the online detection layer belongs, what physical principles it should rely on, and how integrators typically implement it. It draws on Shanghai ChiMay field data from PEM and alkaline projects across four continents.
Why Iron and Silica Break Electrolyzers
Iron and silica behave very differently, but both are toxic to a well-optimized water circuit.
- Iron enters as soluble Fe²⁺ from carbon steel piping or as colloidal Fe(OH)₃ from oxidized upstream water. In the anode of a PEM stack, dissolved iron migrates to the catalyst layer and displaces platinum sites. Stack degradation as low as 5–10 µV per hour has been traced to feedwater iron above 1 ppb.
- Silica is much more mobile. Reactive silica passes through most RO membranes at 1–3 % rejection loss, then adsorbs onto anion exchange resin. Once a mixed-bed is silica-loaded, it releases silica at low ionic strength — meaning your polished water can actually get worse than your RO permeate for a period.
For these reasons, PEM plants specify feed silica typically ≤5 ppb and feed total iron ≤1 ppb, sometimes tighter. Standard laboratory analysis is far too slow to defend those numbers in real time.
Where the Detection Layer Belongs on the P&ID
There are three positions where trace metal and silica detection reasonably lives on a hydrogen feedwater P&ID:
- RO permeate — the first pass filter for what will hit the polishers
- After the degasser / prior to mixed-bed — the last chance to flag a resin-poisoning event
- Mixed-bed outlet — the final quality check before stack feed
At each position, Shanghai ChiMay recommends pairing a direct online chemistry measurement with an indirect fast measurement (conductivity, resistivity), because trace analyzers have longer response cycles and benefit from a second, fast-response check.
Online Techniques That Work
Practical online techniques for hydrogen feedwater applications include:
- Molybdate colorimetric analyzers for reactive silica. These use a molybdenum-blue reagent chemistry and read 0.5–500 ppb. They are the field standard for trace silica, run one measurement every 3–10 minutes, and need reagent replenishment every 30–60 days.
- Colorimetric TPTZ or ferrozine chemistry for iron. These read soluble iron from ~0.2 ppb up. They can be operated in the same analytical cabinet as silica for space savings.
- Conductivity / resistivity ultra-pure water probes. These will not identify the ion, but they will flag any breakthrough within seconds — much faster than the wet-chemistry loops. Shanghai ChiMay’s toroidal and low-cell-constant conductivity electrodes remain stable to ±0.005 µS/cm on ultrapure loops, which is enough to detect the earliest resin breakthrough events.
- Multi-parameter cabinets that combine pH, ORP, temperature, and conductivity provide crucial context. A shift in ORP is often the first sign that iron has begun mobilizing upstream.
Sampling Discipline Matters
The measurement is only as good as the sample. Three details make the difference between reliable data and expensive noise:
- Continuous flowing sample. Trace iron and silica samples must flow at 100–500 mL/min continuously. Stagnant sample lines contaminate within minutes at ppb levels.
- PVDF or PFA sample tubing only. Copper, brass, and stainless steel will leach at ppb levels; PVDF and PFA are effectively inert.
- Grab-sample port immediately adjacent to the online cell. This lets operators periodically confirm the online reading with an ICP-MS lab result. Correlation within ±0.5 ppb is achievable and audit-ready.
What the Signals Should Trigger
Well-instrumented plants use the detection layer for two things:
- Interlock the polishing regeneration cycle. When silica at the mixed-bed outlet rises to 60–70 % of the target limit, initiate regeneration proactively rather than at breakthrough.
- Fault the stack before contamination. Iron above 2 ppb at the stack feed should trip stack current before the contaminant reaches the catalyst.
Both interlocks are cheap in software and priceless in stack availability.
Integration Patterns Shanghai ChiMay Has Seen Working
Across recent 20–100 MW projects, three patterns show up consistently:
- A compact analyzer cabinet with silica and iron analyzers side by side at the polishing skid, sharing sample conditioning and drainage.
- A redundant conductivity/resistivity check at the mixed-bed outlet, wired to the same DCS block as the analyzers so trends can be overlaid.
- A DO transmitter co-located with the trace analyzers, since DO is a leading indicator for iron oxidation and helps discriminate real breakthrough from analyzer drift.
Bringing It Together
Trace iron and silica detection is not a nice-to-have on green hydrogen sites; it is the guardrail that protects millions of dollars of stack and polishing capital. The pattern is consistent: place a wet-chemistry analyzer at RO permeate and mixed-bed outlet, back it with fast-response resistivity or toroidal conductivity, discipline the sampling, and wire the signals into meaningful interlocks. That is what allows a PEM or alkaline plant to defend feed silica ≤5 ppb and feed iron ≤1 ppb during real operation, not just on the day commissioning was signed off.
Shanghai ChiMay maintains this stack of measurements — silica-ready analyzer cabinets, low-drift resistivity, optical DO, and multi-parameter probes — precisely because green hydrogen developers are asking for polishing-line visibility that older water plants never needed. Getting it right upstream is the cheapest way to keep the stack producing hydrogen downstream.