Modern analyzers and AI algorithms that solve seemingly unsolvable measurement problems.
Spectrally™ measures composition and quality directly in the process using AI models and advanced chemometrics with a CNN core, developed individually for each client's chemistry and process. We support AI transformation in production plants, from nonlinear phenomena to multicomponent mixtures.
CNN + chemometrics
Analysis of a wide range of parameters, not only component concentrations
Nonlinear phenomena
And multicomponent mixtures, where the capabilities of PLS/PCA end.
Single ppm
Accuracy achieved through AI algorithms tailored to the analyte and medium
Model for your process
Individually tailored
Comprehensive process analytics
X1 INLINE, X1 LAB / LAB+, X1 PORTABLE, X1 PROBE, and Spectrally OS
Spectrally™ X1 INLINE
Real-time quality control, directly in the process
Measurement without sampling and without downtime
Designed for 24/7 operation, even in demanding conditions
Integration of results with control systems / QA reports
Spectrally™ X1 LAB / LAB+
The bridge between the laboratory and production
Rapid verification of batch composition without tedious sample preparation
Support for scaling formulations from R&D to production
Real-time data visualization and analysis
Spectrally™ X1 PORTABLE
Mobile raw material identification and incoming quality control (IQC)
Fast PASS/FAIL, even without opening the packaging (for transparent packaging)
A tool for raw material control at warehouse intake and in the field
Non-destructive measurement with results in seconds
Spectrally™ OS
Dedicated software with proprietary analytical models
Analyte signal converted into a user-readable result
Rapid development and deployment of models in response to specific process issues
Real data for automating and optimizing the production process
Our algorithms solve measurement problems that classical methods cannot handle
Dozens of feasibility studies have confirmed that our AI models and advanced chemometrics can handle measurement problems that classical analytical methods solve with difficulty or not at all:
- Spectrum disturbed by background or fluorescence → dedicated model distinguishes analyte signal from noise
- Multicomponent mixture with variable matrix → neural networks (CNN) separate overlapping signals
- Nonlinear phenomenon, e.g., viscosity → chemometric model detects a relationship that PLS/PCA cannot capture
- Process without a previously developed measurement method → feasibility study on your samples, model built from scratch
- Compounds with very similar spectra → model recognizes differences that the operator cannot visually detect
- Measurement through opaque or colored packaging → algorithm separates the material signal from the packaging signal
- Variable process conditions: temperature, vibrations, humidity → model calibrated on real data from your line
Industries
Tailored for various industry branches
Cosmetics and Detergents
Petrochemicals
Fertilizers
Chemicals and Polymers
Environmental Monitoring
Water and Wastewater
Why
Gekko Photonics?
We start by analyzing the client's process: variability, risks, bottlenecks. Only then do we select the probe configuration, spectral range, and analytical model.
Our models are based on neural networks (CNN) and advanced chemometrics. They handle problems that classical PLS/PCA approaches solve with difficulty: nonlinear phenomena, multicomponent mixtures, matrices that change over time. Thanks to this, our algorithms achieve accuracy down to single ppm even where standard Raman spectroscopy fails.
A model for a new process problem is created quickly, because the path from notification to an updated algorithm is short: direct access to our R&D. The probe has a built-in reference for autocalibration and autodiagnostics, so the device itself signals deviations.
See the use cases of Spectrally™ analyzers.
Concentration of urea and ammonium nitrate in urea-ammonium nitrate fertilizer (UAN): inline composition control
The plant produces UAN 28, 30, or 32 to customer order on a single continuous mixing line. Hitting the specification window for the selected grade and reaching it quickly during grade changes determine the amount of transitional material and the nitrogen above declaration. We demonstrate how to control composition directly in the mixing line.
Monomer conversion and copolymer profile in acrylic dispersions and resins: inline measurement
In acrylic dispersions and resins, the copolymer composition and its homogeneity determine the glass transition temperature (Tg) and coating properties, while residual monomer is subject to customer regulations. We demonstrate how to monitor monomer conversion and copolymer composition directly in the reactor.
Verification of the F/U ratio and formaldehyde concentration in UFC concentrate upon receipt
The molar F/U ratio in UF resin is a fundamental parameter determining formaldehyde emissions from the finished board, and it directly results from the composition of the received UFC concentrate. We demonstrate how to verify the F/U ratio and formaldehyde content in UFC upon delivery – within 1–2 minutes and without reagents.
