Silica Monitoring: Where It Matters and How It’s Measured

Method guide · Technical perspective · Published: October 12, 2026

Silica is the quiet chemistry of the water industry. It is not toxic at drinking-water concentrations and it rarely appears on consumer-facing reports — yet it is one of the most expensive dissolved substances a plant can ignore. In boilers it rides steam into turbines and leaves glass-like deposits on blades. In ultrapure water systems it defines the last speciation battle at the point of use. In reverse osmosis it sets the recovery ceiling. In cooling towers it accumulates until heat-transfer surfaces pay the bill.

This article maps where silica monitoring decides outcomes, separates reactive from colloidal silica — the distinction that determines what your method can actually see — and explains how online molybdenum-blue analyzers do their work.

Key Takeaways

• Boiler pressure sets the silica ceiling. Allowable boiler-water silica falls from 150 ppm at low pressures to 90 ppm at 300–449 psig, 40 ppm at 450–599 psig, 30 ppm at 600–749 psig, and 20 ppm at 750 psig and above [1].
• Steam-side targets run far tighter than boiler-water targets. Most turbines stay deposit-free below 0.02 ppm silica in steam, and some turbine vendors specify less than 0.01 ppm [2].
• The standard colorimetric measurement sees only molybdate-reactive silica — monomeric silicic acid. Colloidal and polymeric silica are invisible to it; total silica requires a digestion step [3].
• Reactive-silica methods reach down to about 20 µg/L, which makes trace-grade analysis feasible at demineralizer outlets and polishing loops [4].
• Semiconductor ultrapure water puts dissolved silica at roughly 0.05–10 ppb at the point of use, alongside an 18.2 MΩ·cm resistivity requirement [5].
• In RO, amorphous silica solubility is only about 120–150 mg/L at neutral pH and 25 °C. Exceed it in the concentrate and you grow a glassy scale that resists normal cleaning [6].

Where Silica Decides Outcomes

Boilers and steam turbines

Silica volatility is a pressure story. At low pressures silica stays mostly in the boiler water; as drum pressure rises, it carries over into steam mechanically and, at high pressures, vapourously. United States facilities criteria tabulate the consequence: allowable boiler-water silica drops from 150 ppm in the low-pressure bands to 90 ppm at 300–449 psig, 40 ppm at 450–599 psig, 30 ppm at 600–749 psig, and 20 ppm at 750 psig [1].

The reason for the tightening is downstream, not upstream. Silica deposits on turbine blades are hard, adherent, and expensive, so the customary steam-side limit sits below 0.02 ppm for most turbines, with some vendors specifying below 0.01 ppm [2]. That gap — a 150 ppm boiler-water allowance at low pressure against a 20 ppm allowance at 750 psig — is why boiler chemistry programs track silica continuously rather than annually.

Semiconductor ultrapure water

In a fabrication plant, dissolved silica competes with particles as a yield killer at the point of use. Reference specifications for ultrapure water put dissolved silica in the 0.05-to-10-ppb band alongside an 18.2 MΩ·cm resistivity requirement [5]. Those are numbers a colorimetric method has to reach in its most sensitive configuration — and numbers at which a loop’s polishing step (mixed-bed ion exchange or electrodeionization) gets judged. Monitoring silica at the polishing loop outlet, before it reaches wafer surfaces, is the difference between a water system that protects yield and one that ends it.

Reverse osmosis and desalination pretreatment

Amorphous silica solubility sits at only about 120–150 mg/L at neutral pH and 25 °C [6]. An RO array concentrates the feed, so dissolved silica concentrates with everything else; when the concentrate crosses that solubility ceiling, silica polymerises into a glassy deposit that standard cleaning chemistry struggles to remove [6]. The margin between “concentrating normally” and “polymerising on the membrane” is thin, which is why silica is usually the parameter that sets the recovery limit on high-silica feeds. Track it in the concentrate, not just the feed — that is what keeps recovery decisions honest.

Cooling towers and recirculating systems

Recirculating systems enrich whatever enters them. Cycles of concentration push silica upward in the basin water, and above its solubility it deposits alongside hardness and magnesium scales on heat-transfer surfaces — a combination that insulates exchangers gradually, which is exactly what makes it easy to miss. Utilities in geothermal regions, mining districts, and other high-silica basins treat the basin silica trend as a cycles-of-concentration governor, backing off blowdown control when the trend climbs. The numbers are site-specific — make-up chemistry, temperature, and the treatment program all shift the practical ceiling — which is why a trend line, not a single laboratory result, is the operative deliverable for the operations team.

Where to Sample, and How Often

The silica question changes character at every sample point, so draw the sampling plan from the failure modes, not from convenience:

• Boiler drum water — the control point for carryover. Sampling frequency rises with operating pressure, because the allowable margins shrink from the 150 ppm-scale allowances at low pressure to the 20 ppm band at 750 psig [1].
• Saturated and main steam — the verification point. The customary sub-0.02 ppm steam-side limit [2] means measurements here sit at trace concentrations, where grab-sample handling errors become significant and continuous or high-frequency sampling earns its keep.
• Demineralizer and mixed-bed outlets — the early-warning point. Reactive silica leakage here forecasts resin exhaustion long before the cycle does.
• RO concentrate — the recovery-governance point. The 120–150 mg/L solubility context [6] makes this the sample that decides whether the recovery setpoint can move up.
• Cooling basin and make-up — the trend point. The difference between the two is the enrichment the tower is imposing.

A defensible rule: sample continuously where the limit is tight and the consequence is equipment damage (steam, UPW polishing loops), and sample on a fixed operational cadence where the margin is wide and trends move slowly (basin water). Either way, the sampling plan — point, frequency, and method — belongs in the same document as the limit it defends.

Reactive vs Colloidal: The Split That Decides the Method

Silica in water is not one analyte but a family. The colorimetric technique standardized for water analysis measures only molybdate-reactive silica — the monosilicic acid molecule and closely related monomeric species. The remaining polymerised material, loosely called colloidal or unreactive silica, is not detected in the colorimetric test. The sum of reactive and colloidal is total silica, and converting the colloidal fraction into a measurable form requires treating the sample with hot sulfuric acid digestion first [3]. In the classic procedure, the reactive fraction develops a yellow silicomolybdate complex that a reducing agent — commonly ascorbic acid — converts to molybdenum blue for photometric readout [3][4].

Which fraction matters depends on the question. For boiler carryover and steam purity, reactive silica is the operative variable, because it is the dissolved form that volatilises. For RO scaling, total silica drives the mass balance while reactive silica drives the solubility model. For ultrapure water, both fractions matter — colloidal silica behaves as a particle defect source, reactive silica as an ionic contamination source. An online analyzer that reports reactive silica alone is not wrong; it is answering a specific question, and the operator has to know which one.

The Scenario Table

Scenario	Why silica is monitored	What to measure
Boiler water and steam	Carryover deposits on turbine blades; limits tighten with pressure, down to 20 ppm boiler-water silica at 750 psig [1]	Reactive silica in boiler water and steam; routine online or high-frequency grab sampling at the drum
Demineralizer and mixed-bed outlet	The last barrier before the steam cycle; trace leakage foretells exhaustion [2]	Reactive silica at trace levels — methods to about 20 µg/L are standard [4]
Semiconductor UPW	Point-of-use dissolved silica spec of roughly 0.05–10 ppb beside 18.2 MΩ·cm resistivity [5]	Reactive silica online at the polishing loop; periodic total silica for colloidal assessment
RO concentrate	Amorphous silica solubility of about 120–150 mg/L at neutral pH, 25 °C; above it, glassy scale [6]	Reactive silica for the scaling model; periodic total silica for the mass balance
Cooling tower basin	Cycles of concentration enrich silica toward deposition on heat-transfer surfaces	Reactive silica trend in make-up and basin water; site-specific ceiling

How Online Molybdenum-Blue Analyzers Work

An online reactive-silica analyzer is a wet-chemistry laboratory compressed into a wall-mounted cabinet, running the classic sequence on a schedule. A sample is drawn and conditioned; molybdate reagent is added to form the yellow silicomolybdate complex; a reducing agent — commonly ascorbic acid — converts it to molybdenum blue; a photometer reads the blue intensity against a calibration [3][4]. The measurement is the ASTM D859-family colorimetric chemistry — determination of molybdate-reactive silica — applied automatically and repeatedly, with the method useful for concentrations as low as 20 µg/L [4].

That automation converts chemistry into three recurring maintenance items. Reagents come first: the analyzer consumes molybdate and reducing reagent continuously, so inventory planning and expiry control are operating costs, not accessories. Then optics and flow: the colorimeter window and sample tubing must stay clean, because turbidity, colour, and fouling bias the photometric read. Finally calibration: zero and span checks against silica standards anchor the traceability a plant audit will ask for. None of this is exotic — but it explains why analyzer selection is a five-year conversation, not a purchase-order line. “Buy the sensor, own the outcome. We quote 5-year cost, not just unit price.” Reagent logistics, optical spares, and calibration service dominate the out-years of that quote.

Setting Range and Detection Expectations

Match the instrument to the point of use, and the published limits do the rest:

• Boiler water needs a ppm-range instrument for most bands — the 150 ppm low-pressure allowance and the 20 ppm high-pressure ceiling bracket the useful span [1].
• Steam and condensate live at tens of ppb, against a customary 0.02 ppm (20 ppb) steam-side limit and tighter vendor specs [2] — squarely inside what a trace-configured colorimetric analyzer resolves at its 20 µg/L floor [4].
• UPW polishing loops operate below or near that floor. Point-of-use dissolved silica expectations of 0.05–10 ppb [5] demand the most sensitive configuration and meticulous zero control.
• RO concentrate and cooling basins are milligram-per-litre territory, where the 120–150 mg/L solubility context [6] and site-specific trends set the instrument’s span.

Before buying, ask five questions. What is the detection limit at the lowest point of use, and under what conditions? Does the analyzer report reactive only, or offer a total-silica digestion option [3]? What is the reagent consumption per month at my sampling frequency? Does the output integrate digitally with the plant historian? And can the vendor supply calibration documentation you can hand to your auditor without translation work? The answers decide whether the analyzer pays for itself in prevented turbine deposits, protected membranes, or preserved UPW yield — or becomes one more cabinet with an expired reagent.

Silica rewards the plants that watch it and bills the ones that do not. The chemistry is well understood, the methods are standardized down to trace levels, and the sampling points are few and well defined. What separates a mature program from an occasional laboratory check is simple: the limit, the sampling point, and the measurement written into one operating document, with the instrument chosen against its five-year chemistry bill — not just its purchase price.

References

1. UFC 3-240-13FN (US Unified Facilities Criteria) — “Determining Water Carryover in Steam”; tabulates allowable boiler-water silica by pressure band, from 150 ppm at low pressure to 20 ppm at 750 psig. https://www.ufgscriteria.tpub.com/3_240_13/3_240_130085.htm
2. Process Group (Australia) — Water Treatment Handbook, Chapter 18 on steam turbine deposition, erosion and corrosion; notes the customary steam silica limit below 0.02 ppm, vendor specs below 0.01 ppm, and vaporous carryover behaviour. http://processgroup.com.au/handbook/chapter-18-steam-turbine-deposition-erosion-and-corrosion
3. US Nuclear Regulatory Commission — technical letter report on Boraflex, RACKLIFE and BADGER; explains the molybdenum-blue colorimetric technique (ASTM D859), reactive versus colloidal versus total silica, and the digestion step for total silica. https://ww2.nrc.gov/docs/ML1221/ML12216A307.pdf
4. ANSI Webstore — ASTM D859, Standard Test Method for Silica in Water; colorimetric determination of molybdate-reactive silica, useful for concentrations as low as 20 µg/L. https://webstore.ansi.org/standards/astm/astmd85905
5. Aguato — semiconductor water treatment reference; ultrapure water specifications per SEMI F63 with 18.2 MΩ·cm resistivity and dissolved silica of 0.05–10 ppb at the point of use. https://aguato.com/resources/semiconductor-water-treatment
6. Vcycletech — RO silica scaling control overview; amorphous silica solubility of about 120–150 mg/L at neutral pH and 25 °C and polymerisation into difficult glassy scale. https://www.vcycletech.com/ro-silica-scaling-control

About the author: Written by the Shanghai ChiMay Technical Editorial Team — the editorial group behind Shanghai ChiMay’s in-line water quality analyzer line, covering photometric and electrochemical process analysis for boiler, RO, ultrapure water and cooling applications.