Nine technology families, each answering a different question, none suiting everything. Our role: choose the one that suits your matrix and your measurement point, including when another route serves your need better.
The nine families
NIR spectroscopy
Moisture, identity, blend homogeneity, content. The EMA devotes a dedicated guideline to it. It is also the family that asks for the most modelling.
NIR-HPTLS spectroscopy
Composition of bioproduction media, and of that field alone. Pure-component calibration: quick to deploy when the composition is known.
MIR spectroscopy
Fine molecular signature in the liquid phase, supersaturation in crystallisation. A short optical path, which is what makes aqueous media measurable.
RAMAN spectroscopy
Molecular identity, polymorphism, reaction progress, polymers, black plastics included. Highly selective, responsive to fluorescence.
Imaging and machine vision
Counting, measuring a shape, mapping a constituent: a map where the other families return a value.
TERAHERTZ imaging and spectroscopy
Checking through non-metallic packaging, with the pack closed. Non-ionising radiation, so deployable with no controlled area and no radiation protection.
Electrical tomography ECT and ERT
Phase distribution in opaque vessels and pipes. No chemometric model to build for the image.
Optical coherence tomography OCT
Cross-sections of coatings, micrometric resolution, contact-free: a thickness measured directly, with no calibration.
SR-DLS particle sizing
In-line particle size, from the nanometre to the micron, in a concentrated medium and with no dilution.
What each page contains
The same plan for each technology: principle, what can be measured, installation, model, evidence and conditions of application, with manufacturers’ sources.
Two reading axes
These technologies are usually classed by physical family, optical, electrical, imaging. That is true, and it does little to help decide. Two other axes are more useful.
Pure-component model or model calibrated on samples
Some measurements call for a model built on a representative population, with reference analyses. That is calibration on samples, the route of NIR on a natural matrix. Others decompose the signal onto known constituents, or read a physical quantity with no calibration. The lead time and the upkeep have nothing in common. What each route costs.
What you actually get
A content, a size, a thickness, a distribution, an identity. A technology that gives an image gives no concentration. A technology that gives a concentration says nothing about the distribution in the volume. Many projects turn on that distinction.
Going further
Nine technologies compared
The table reads by column: the quantity followed, the matrix, the calibration burden accepted.
Measurement fundamentals
Mass actually analysed, sampling plan, method validation, data integrity. Twelve pages for analytical teams.
Where your need falls outside these families
These nine families are the ones we implement, and the ones we hold a founded view on. Other measurement physics exist. It happens that the answer to a process question is one of them, and in that case we say so.
Describe the operation and the quantity at the heart of your problem. You will hear what answers it, including when the answer sits outside our catalogue.
Looking for a technology, or for an answer to a problem?
If you already know what you are after, these nine pages answer it. If you start from a process problem, a drift, a cycle time, a non-quality, entering by the parameter is more direct. And if you would rather talk it through, forty-five minutes are enough.