Plant operators are increasingly treating industrial wastewater not as a „secondary circuit after the process,” but as a separate measurement stream requiring continuous monitoring. This is driven by regulatory pressure, the cost of reagents in the treatment plant, and the risk of production line downtime when the load of nitrates, phosphates, or hydrocarbons exceeds the contractual limits with the wastewater receiver. This is one of those measurements that chemical plants, refineries, and wastewater treatment operators turn to when they need a real-time result without laboratory sampling — and it is for this scenario that we most often configure an analyzer installed directly in the pipeline.
In this guide, we show how Raman spectroscopy measures nitrates, phosphates, and hydrocarbons directly in wastewater, which diagnostic bands we use, how we build the system architecture (probe in the pipeline, measurement unit, chemometric layer), and what a typical implementation timeline looks like in the water and wastewater industry.
Why classical laboratory analysis cannot keep up
The traditional wastewater monitoring model is based on manual or automatic sampling (ISCO samplers), laboratory analysis (colorimetry, IC, GC-FID for hydrocarbons), and reporting with a delay of several hours to several days. During this time, the wastewater stream is already flowing to the municipal treatment plant, to a receiver, or to recycling, and the response to a deviation occurs after the fact.
Three consequences we most frequently observe at our clients:
- Penalties for exceedances — when the nitrogen or phosphorus load in wastewater exceeds the contractual threshold with the municipal treatment plant, the plant pays an increased rate per m³ or an administrative fine.
- Excess reagent usage — the treatment plant operator doses coagulant „blindly” with a safety margin because there is no online measurement. This is a direct path to excessive consumption of iron sulfate or polyelectrolytes.
- Product loss — in the petrochemical and organic chemistry industries, hydrocarbons in wastewater are simply product that has left the process via an uncollected stream. Without real-time measurement, an alarm is triggered only after laboratory analysis.
Process Raman closes this loop. The measurement typically takes 5 seconds (spectrum acquisition from 5 to 300 s depending on the matrix), and the chemometric model converts the spectrum into a concentration in a fraction of a second. The operator sees the value on the DCS alongside flow and pH.
What Raman „sees” in wastewater
Raman spectroscopy in industrial wastewater operates in the range of approx. 300–1650 cm⁻¹, which covers most key diagnostic bands for dissolved anions and hydrocarbons. Below are three groups of analytes we most frequently work with.
Nitrates (NO₃⁻)
The nitrate anion has a strong symmetric stretching band ν₁ at approx. 1048 cm⁻¹. This band is well separated from most other anions and provides a linear relationship between intensity and concentration in the range of several tens of mg/L in classical Raman, and lower — using SERS or UV-resonance Raman configurations. In industrial wastewater from plants, fertilizer plants we typically see nitrates together with urea and biuret — the PLS model separates these fractions without difficulty.
Phosphates (PO₄³⁻ / HPO₄²⁻ / H₂PO₄⁻)
The symmetric stretching ν₁ of phosphate yields a band around 960 cm⁻¹ (with sidebands at 590, 435, and approximately 1075 cm⁻¹ from the ν₃ mode). The ionic form of phosphate depends on pH, so we supplement the Raman measurement with simultaneous online pH measurement to convert the dominant form into total phosphorus. In wastewater from fertilizer production or detergents, phosphates are a key eutrophication parameter.
Dissolved and emulsified hydrocarbons
Hydrocarbons (hexane, xylene, aliphatic and aromatic fractions) produce characteristic bands in the C-C skeletal vibration and aromatic ring ranges below 1650 cm⁻¹. The C-H stretching bands around 2800–3000 cm⁻¹ lie outside the standard X1 INLINE range, so in practice for streams, petrochemical we use low-frequency bands and ring modes (e.g., for xylene around 720–800 cm⁻¹, for toluene around 1005 cm⁻¹). In heavily oil-laden wastewater, we configure the probe with a self-cleaning module (Retractex) to avoid hydrophobic deposits on the probe window.
In addition to these three groups, Raman allows monitoring in industrial wastewater of Sulfates (approx. 980 cm⁻¹), Biuret (characteristic C–N bands), Urea (approx. 1003 cm⁻¹) and selected organic salts — the full list depends on the specific stream matrix and is determined during feasibility studies on client samples.
Measurement system architecture
For industrial wastewater, we typically implement one of three configurations:
1. Inline in pipeline or channel
An immersion probe is mounted directly in the main wastewater stream or on a side-stream, so that the stream flows over the optical window. The signal returns via fiber optic cable (up to 100 m) to the Spectrally X1 INLINE, which is housed in a free-standing cabinet or mounted on the wall of the control room. This solution is suitable for relatively clean streams — produced water, wastewater after preliminary separation, cooling water with potential contamination.
2. At-line in a sampling loop
When the wastewater stream contains a high amount of suspended solids or sediment, we route a side loop from a constricted pipe, pass it through a large-surface filter, and only then place the probe in this loop. The measurement occurs in a continuous loop, but „inline” from an analytical perspective is performed on a derived stream, not the main one. This allows maintaining the optics in good condition for longer and reduces cleaning cycles.
3. Lab and portable verification
For spot samples — daily monitoring, alarms from environmental monitoring., external audits — we use Spectrally X1 LAB in the plant laboratory (analysis in a vial, 25-sample carousel, same chemometric model as inline) or the Spectrally X1 PORTABLE at the warehouse gate or directly at the discharge point. The same spectral library works across all three devices, so a model trained once works everywhere.
The entire „brain” of the system is the Spectrally OS — PLS, PCA, and CNN models over a library of approx. 28,000 reference spectra, drift monitoring, export to CSV/PDF/RAW, role-based access control (RBAC), full audit trail. Integration with the plant DCS via PROFIBUS, PROFINET, or GSM modem (when the plant does not want to connect the analyzer to the ICS layer).
Business value — what pays off and in what timeframe
From our implementations at chemical, fertilizer, and petrochemical clients, we observe the following typical indicators:
- Measurement time — typically 5 seconds from acquisition to value on the operator screen (vs. several hours for classical laboratory);
- Implementation time — 3 to 5.5 months from the kick-off workshop to an operational system on the line (depending on sample availability for feasibility and the mechanical schedule);
- ROI — 6 to 10 months in projects where the main driver is savings on treatment plant reagents or avoidance of administrative fines;
- Reduction in analytical costs — up to 80% compared to classical sampling laboratory (while maintaining control validation tests once daily instead of every hour).
We treat these values as orders of magnitude — in a specific case, they depend on the matrix, concentration range, and probe installation conditions. A detailed business plan is prepared during the feasibility study on client samples.
Gekko Photonics solutions for water and wastewater
For industrial wastewater treatment operators, environmental departments, and plants with their own wastewater discharge, we configure systems based on the Spectrally X1 family:
- Spectrally X1 INLINE — process analyzer with a 785 nm laser and 600 mW power (30 mW in ATEX version), thermoelectrically cooled CCD detector, two measurement channels as standard, immersion probe or self-cleaning Retractex for difficult media, fiber optic cable up to 100 m, communication via PROFIBUS, PROFINET, GSM.
- Spectrally X1 LAB — stationary laboratory analyzer for model validation, through-package analysis in glass vials or quartz cuvettes, carousel for up to 25 samples, same chemometric engine as inline.
- Spectrally X1 PORTABLE — portable analyzer case for spot measurements in the field, at the warehouse gate, or during environmental audits, IP54.
- Spectrally OS — software layer, PLS/PCA/CNN models, library of approx. 28,000 spectra, audit trail, role-based access control, CSV/PDF/RAW export, DCS integration.
Our most concentrated real-world implementations are in process chemistry (resins, cosmetics, fertilizers, adhesives, hydrocarbons). We have been working in industrial water and wastewater for several years — this is a public tier of our portfolio. A full list of possible applications is available in the analyzers section and in our industry-specific guides.
Test measurement and engineering consultation
At Gekko Photonics, we select the probe configuration, wavelength, acquisition time, and chemometric model tailored to the specific matrix of your wastewater. The standard schedule for the first step is as follows:
- 30-minute discussion with an applications engineer — review of your process, analytes, concentration ranges, and probe installation conditions.
- Test measurement within 2 weeks — we acquire spectra from your samples (3–5 samples representative of the operating range) using our laboratory Spectrally X1 LAB and report the reliability of analyte vs. matrix separation.
- Feasibility report within 10 business days — recommendation: whether Raman is the appropriate method, what hardware configuration is suitable, what calibration ranges are realistic, and what CAPEX/OPEX expenditures should be anticipated.
We invite you to contact us. If you would like to read about the technology first, we encourage you to review Raman spectroscopy in process chemistry and the main guide on process analyzers.
Frequently asked questions
Can Raman detect nitrates in wastewater at concentrations typical for municipal discharge?
Classical Raman spectroscopy has a detection limit for nitrates in the millimolar range, corresponding to several tens of mg/L. For lower concentrations (at the mg/L level or below), enhancement techniques are used — UV-resonance Raman or SERS. For industrial wastewater, the typical measurement range (several tens to several hundred mg/L) falls within the window of classical process Raman, without the need for signal enhancement.
What about turbidity and suspended solids in wastewater?
High turbidity or dense suspended solids scatter the laser beam and can reduce the signal-to-background ratio. For such streams, we use an at-line configuration with a sampling loop and a pre-filter, or a probe with a Retractex self-cleaning module. The decision is made after feasibility — if the matrix is too complex for direct measurement, we recommend an at-line configuration or a combination of Raman with a complementary reference method.
How does Raman compare to NIR and FT-IR for wastewater?
Raman is less sensitive to water than FT-IR and NIR — this is a practical advantage in water-based wastewater, as water does not obscure the bands of the analytes of interest. For polar mixtures in an aqueous background (nitrates, phosphates, sulfates, urea, biuret), Raman provides cleaner diagnostic bands. NIR and FT-IR perform better for non-aqueous matrices or when C-H/O-H bands are of interest — there they serve as a complement, not a replacement. A broader comparison can be found in the guide Gekko Photonics blog.
Does Gekko Photonics supply systems for municipal wastewater treatment plants?
Most of our implementations are in process chemistry and industrial wastewater from specific industrial operators. In municipal treatment plants, we engage on a project basis, after feasibility testing on samples from a specific monitoring point — we check whether the concentration range falls within the measurement window of classical process Raman, whether an enhanced configuration is necessary, and what level of service availability we can offer the operator.
How long does implementation take?
Typically 3 to 5.5 months from the kickoff workshop to an operational system on the line. The first month covers feasibility and chemometric model building, the second and third months involve mechanical and electrical integration at the site, and the remaining weeks are dedicated to validation and model tuning on the actual stream. ROI is typically observed within 6 to 10 months.