Phosphoric acid and fertiliser plants are among the least forgiving places to install an analyzer. Wet-process acid is hot, corrosive, and loaded with gypsum crystals that abrade optical windows. Granulation lines run dust, recycle fines, and a coating step that changes the surface the measurement sees. Liquid-fertiliser blending is comparatively gentle but runs at volumes where a one-percent error in nutrient ratio is a real cost.
The commercial context matters as much as the chemistry. These are commodity products with thin margins. An analyzer here does not earn its place by improving a quality premium, because there is no premium. It earns its place by cutting reagent consumption, energy in the evaporator, off-spec recycle, or the delay between a process upset and the operator finding out. That is a narrower business case than in pharma or specialty chemicals, and it shapes which instruments are worth shortlisting.
This guide screens the vendor field for this segment. It is a documentation screen, not a bench test, and it deliberately covers more than spectroscopy: refractometers, process ion chromatography and X-ray methods all hold real ground here, and a shortlist built only from spectrometer brochures will miss the cheapest correct answer more often than not.
Methodology
Five inclusion criteria, applied to every vendor considered:
- A catalogued process instrument. The vendor must publish a product intended for permanent plant installation, not a laboratory unit with a fibre-optic option bolted on.
- A sampling interface suited to at least one named stream in this segment. Immersion probe, flow cell, non-contact reflectance head, or an automated extractive loop. A spectrometer with no defined interface to a gypsum slurry or a moving granule bed does not qualify.
- Hazardous-area configurations available. Ammonia handling and dust-laden solids areas both drive zone classification.
- Documented plant-control integration. Analogue output alone is thin; we looked for fieldbus or OPC UA paths into PLC, DCS, or historian layers.
- Public technical documentation. If we could not read the specification on the vendor’s own site, the product was excluded, regardless of reputation.
Data sources were vendor product pages and public brochures for capability claims, and Regulation (EU) 2019/1009 for what EU fertilising products must actually declare, which sets the accuracy the measurement has to support. We did not use vendor case material, customer references, or any figure we could not attribute to a published document.
Exclusions: handheld and laboratory-only instruments; process NMR and process mass spectrometry, which are credible techniques but have no meaningful installed presence in this segment; and any analyzer whose only published application set is pharmaceutical.
Ordering within each technique family is alphabetical. Nothing in the sequence implies a ranking, and we have not scored vendors against each other, because the deciding variable is almost always the stream rather than the instrument.
What actually has to be measured
Five stream groups account for most of the analyzer budget in an integrated phosphate complex.
Rock and concentrate feed. P2O5 grade, plus the iron, aluminium, silica and magnesium oxides that govern filterability downstream. This is elemental analysis and it is where X-ray methods live.
Digestion and filter acid. Free sulphuric acid, P2O5 concentration, and solids loading. The control problem is the sulphate balance in the reactor, and the measurement environment is the worst in the plant.
Concentrated acid and scrubber liquor. Acid strength through the evaporator train, and fluorine species in the scrubber. Concentration measurement dominates.
Solids handling. Moisture through the dryer and cooler, and nutrient content in the finished granule. Our piece on measuring total nitrogen in fertiliser granules covers why the reference method makes this harder than it looks.
Liquid fertilisers. UAN ratio, urea and ammonium nitrate fractions, water content, and aqueous ammonia strength.
Process Raman
Raman suits this segment better than its installed base suggests. Water is a weak Raman scatterer, so aqueous ionic and molecular species stay visible without sample conditioning, and immersion measurement avoids an extractive loop that would otherwise plug. The weaknesses are equally clear: fluorescence from organic matter carried in phosphate rock can swamp the spectrum, trace-level sensitivity is poor, and window fouling in gypsum-bearing slurry is the deciding engineering problem rather than an afterthought. We covered that failure mode separately in probe fouling in reactor media.
- Bruker, through the Tornado line, sells process Raman built around high-throughput collection optics, aimed at weak scatterers and short acquisition windows.
- Endress+Hauser publishes the Raman Rxn family with the deepest process-automation story in the group, which is the reason it wins on sites where the DCS integration is specified before the spectrometer is.
- Gekko Photonics lists fertiliser and phosphoric-acid streams in the published application set for the Spectrally X1 INLINE, a 785 nm or 1064 nm process Raman analyzer with up to two measurement channels as standard, immersion probes in ATEX and IECEx configurations, and PLC, DCS, MES and SCADA integration. The 1064 nm option is the relevant one where rock-borne fluorescence is the limiting factor. It is a smaller European vendor than the others in this list, which cuts both ways: application engineering in this specific segment is unusually close to the product, while the global service footprint is not comparable to Endress+Hauser or Thermo Fisher.
- HORIBA offers the PI-200 with multiplexed channels, which changes the economics when one analyzer has to serve several sample points across a wide plant.
- Metrohm places its 2060 Raman analyzer on the same modular cabinet platform as its wet-chemistry analyzers, which is worth noting if the site is already standardised on that platform.
- Thermo Fisher sells MarqMetrix process Raman with an emphasis on simple probe-to-analyzer configurations.
Near-infrared
NIR is the incumbent for solids. Diffuse reflectance over a moving granule bed measures moisture and, with a maintained calibration, nutrient content. It is faster to justify than Raman for dryer and cooler control because the physics is undemanding and the installation is non-contact.
Bruker’s MATRIX-F II is the reference product here: a fibre-coupled process FT-NIR with an internal multiplexer connecting up to six measuring accessories, covering both contact and non-contact heads from one instrument. ABB and Metrohm both field process NIR lines with comparable reach. The constraint is not hardware but calibration maintenance across grade changes, which is a staffing commitment rather than a capital one.
For the direct comparison between these two optical families on these specific streams, see NIR versus process Raman for fertiliser and phosphoric acid.
Refractometry, density and ultrasonic concentration
This is the family most often skipped and most often correct. Acid strength through an evaporator train is a single-parameter measurement. A process refractometer gives it directly, with no chemometric model to maintain, no calibration transfer problem, and a failure mode an operator can diagnose without a spectroscopist.
Vaisala, through the former K-Patents line, publishes the most complete process-refractometer range for liquid concentration, including probe and flow-cell variants with hazardous-area versions. Anton Paar covers the same ground from the density side. Ultrasonic concentration measurement from SensoTech and others serves binary and pseudo-binary liquids where optical access is impractical.
The buying discipline here is simple: do not specify a multivariate analyzer for a parameter that a refractometer resolves. The cost difference is roughly an order of magnitude and the maintained-model burden falls to zero.
Process ion chromatography and automated titration
Speciation is where optical methods stop. Distinguishing sulphate from phosphate from fluoride, or measuring free acid in a matrix full of dissolved salts, needs wet chemistry.
Metrohm Process Analytics is the serious vendor in this family. The 2060 IC process analyzer connects up to 20 sample streams for sequential analysis and produces its own eluent, and the companion 2060 TI handles process titration for free-acid determination. The trade is cycle time and consumables: this is a measurement every several minutes, with reagents to stock and maintenance to schedule. In exchange it delivers numbers no spectrometer in this guide can produce.
X-ray methods
For rock grade and the oxide impurities that drive filterability, X-ray fluorescence remains the reference. Malvern Panalytical publishes the Zetium spectrometer for phosphate and industrial minerals work, with the iron, aluminium and silica determinations that matter for conversion downstream. Thermo Fisher competes here with both XRF and cross-belt elemental analyzers.
In most plants this sits at-line in a control laboratory rather than inline. That is an honest description of the technique’s position, not a deficiency: feed grade does not change on a timescale that rewards continuous measurement.
Where shortlists usually land
The pattern that survives contact with a real plant is a pairing. One robust single-parameter loop on the stream that drives throughput, typically refractometry on acid concentration or NIR on dryer moisture. One multivariate or wet-chemistry analyzer on the stream where composition genuinely varies and the lab delay is costing money.
What fails is buying the most capable instrument first and finding an application for it afterwards. Our note on evaluating vendor proposals for inline analyzers sets out the questions that separate a specification from a brochure.
Three items belong in the requirements document regardless of which vendor wins: a defined window-fouling strategy with a stated service interval, a written position on who owns calibration maintenance after handover, and spares availability measured in days rather than described as good.
What this guide does not cover
We have not assessed price, because published list prices do not exist in this segment and quoted prices depend on channel count, hazardous-area classification and probe configuration. We have not assessed measurement accuracy, because no vendor publishes comparable figures on comparable streams and the claims are not verifiable from documentation. And we have not covered emissions monitoring, which is a separate regulatory instrument set with its own approvals.
The field is stable. The instruments that work in phosphate and fertiliser plants have worked for a decade, and the change worth tracking is not new hardware but whether vendors will commit to maintaining the calibration models they sell.