In the production of RTV silicone reactors, MS-polymer sealants, and reactive structural adhesives, quality is determined not by whether a single batch falls within specification, but by how tightly that parameter is maintained over the entire 24 months of production. Batch stability directly translates into adhesion and curing time for the end customer — and a conventional QC laboratory cannot keep pace with the reactor’s speed. This is precisely one of the chemistries for which we most frequently configure probes and chemometric models directly tailored to a specific formulation.
In this article, we describe why inline measurement using Raman spectroscopy delivers the highest return of all PAT techniques on silicone, adhesive, and sealant lines — and how to implement it practically without rebuilding the reactor.
Why batch-to-batch variation in silicone, adhesive, and sealant production is painful
In RTV and HTV silicone reactors, two variables that the laboratory cannot keep up with are decisive: the final degree of conversion (residual vinyl or hydrogen monomer in hydrosilylation crosslinking) and the intermediate viscosity during polycondensation. A small difference in temperature hold time — five or ten minutes — shifts the chain length distribution and yields a batch that formally meets the COA but behaves differently in the customer's application.
In acrylic and epoxy adhesives, the analogous problem is incomplete monomer conversion: residual methacrylate reduces joint strength and causes migration, while unreacted epoxy groups alter the curing profile in application. In silane and MS-polymer sealants, critical parameters are the moisture content of the batch and the degree of precursor hydrolysis — parameters that, in a classic workflow, are only visible after an FT-IR laboratory sample, i.e., with a 2–4 hour delay.
The effect? The line operates within a safe but wide process window. Rework, reblends, and downgraded product cost more on an annual basis than the CAPEX for an inline analyzer.
What Raman sees in silicones, adhesives, and sealants
Raman spectroscopy reads covalent bond vibrations — making it an ideal tool for three families of materials whose chemistry differs fundamentally:
Silicones (PDMS, RTV-1, RTV-2, HTV). The symmetric stretching of the Si–O–Si backbone gives a distinct band around 490 cm⁻¹, Si–CH₃ deformation extends around ~700 cm⁻¹, and the symmetric bending of the Si–CH₃ group lies at ~1260 cm⁻¹. These are bands with strong signal and low fluorescence — silicone is practically an „ideal analyte” for 785 nm.
Acrylic adhesives and composites. The methacrylate monomer has a clear C=C stretching signal around 1640 cm⁻¹, whose decay directly indicates the degree of conversion. A quantitative PLS model on this feature allows real-time tracking of the reaction endpoint.
Epoxy adhesives. The oxirane ring gives a breathing signal around 1250 cm⁻¹, and the ether bridge of the epoxy backbone appears at ~915 cm⁻¹. The decrease in intensity of these bands during curing allows conversion to be calculated.
Silane and MS-polymer sealants. Polyether backbone vibrations, alkoxysilane crosslinking, and residual alkoxy groups are visible simultaneously — Raman captures all these structures in a single spectrum, without phase separation or sampling.
In our Spectrally X1 INLINE standard range of 300–1650 cm⁻¹ covers practically all diagnostic bands for this group of materials. A resolution of 8 cm⁻¹ safely separates overlapping Si–CH₃ and backbone deformation signals — without the risk of aliasing in the chemometric model.
Inline vs at-line — where measurement delivers the greatest effect on CPK
The decision to invest in inline measurement in the reactor or to set up an at-line station next to the line comes down to how quickly the process drifts. In RTV silicones, the reaction proceeds in minutes, so a sample taken every hour will not stop the deviation — here, inline is the only sensible choice. In the case of sealant compounds, where the crosslinking reaction is slow and the mixer itself operates in 2–4 hour cycles, at-line measurement every 15 minutes delivers a comparable operational effect and is cheaper to implement.
In practice, we combine both levels: Spectrally X1 INLINE monitors the critical reactor, and Spectrally X1 LAB handles samples from mixers, raw material storage, and outgoing quality control. The chemometric models share the same library in the Spectrally OS, platform, so the inline operating point can be validated one-to-one against a laboratory spectrum — without double calibration.
In our implementations in reactive chemistry, the reduction in batch variability (CPK) typically translates to 8–12% fewer reworks and a reactor cycle time reduction of approximately 10% — without formulation changes, solely through more precise endpoint targeting.
Implementation: probe, fluorescence, calibration
Three elements that determine the success of implementation in silicones and adhesives:
1. Probe selection. Silicone reactors often involve viscous and slurry media, with a tendency for product deposition on the optical window. For such processes, we configure the Spectrally X1 INLINE with a self-cleaning Retractex module — the probe cyclically retracts to a flushing position, cleans the window with inert gas or flushing medium, and returns to the process. This provides signal stability for thousands of operating hours without manual intervention.
2. Fluorescence. Contrary to common opinion, pure silicones and most acrylic adhesives have low fluorescence under 785 nm excitation. The problem arises with compounds containing dyes, carbon black, or stabilizers — in such cases, we apply suppression techniques detailed in our guide on fluorescence suppression. In most cases, optimization of acquisition time and baseline correction is sufficient; in extreme cases, we propose a configuration with an extended spectral range, where the signal escapes from under the fluorescence tail.
3. Chemometric calibration. PLS models for residual monomer, degree of conversion, or mixture composition are built on customer samples — in a feasibility cycle, before the equipment is shipped to the site. The RMSECV achieved on typical silicone formulations is on the order of 0.1–0.3% by weight of monomer in the calibration range of 0.5–8%; for residual epoxy, typically below 0.5% by weight. The model lives in Spectrally OS, which monitors the drift of the reference spectrum and signals the need for recalibration before it affects the line result.
Integration with DCS, MES, and historian is described in a separate article — the IT/OT patterns and checklist apply directly to these installations as well.
Gekko Photonics solutions for silicones, adhesives, and sealants
In the Spectrally X1 analyzer family family, we have four complementary layers that we implement in the production of silicones, reactive adhesives, and sealant compounds:
- Spectrally X1 INLINE — process analyzer with immersion probe, 785 nm 600 mW laser (30 mW in ATEX version), TE-cooled back-thinned CCD, PROFIBUS / PROFINET / GSM communication, fiber optic cable up to 100 m. Configurable with the Retractex module for viscous media.
- Spectrally X1 LAB — laboratory analyzer with a carousel for up to 25 sample positions, through-package analysis via glass vials and quartz cuvettes. For incoming inspection of silicone raw materials, acrylic monomers, epoxy prepolymers, and hardeners.
- Spectrally X1 PORTABLE — portable analyzer for the warehouse gate and mobile raw material verification at suppliers. Touchscreen, IP54, spectral libraries compatible with Spectrally OS.
- Spectrally OS — common software layer: PLS, PCA, and CNN models, library of ~28,000 reference spectra, RBAC, audit trail, CSV/PDF/RAW export, integration with DCS and MES.
For plants operating silicone and adhesive lines in hazardous areas, we configure versions with reduced laser power; details of the selection principles are described in our ATEX / IECEx guide. Implementations are delivered as a full stack: we select the probe, calibrate the model on your samples, integrate with DCS, and provide service support. The average time from workshop to operational inline system is 3–5.5 months.
FAQ — frequently asked questions
Will Raman replace our FT-IR in silicone QC?
For quantitative control of the degree of conversion and inline monitoring — yes, Raman has higher sensitivity to Si–O–Si backbone and C=C vibrations and does not require an ATR window. FT-IR remains complementary for analyses of moisture and hydroxyl groups, which Raman sees less strongly. In most of our silicone implementations, FT-IR shifts to less frequent laboratory control, while Raman takes over daily line monitoring.
Do we need to rebuild the reactor for an inline probe?
Most often, no. The immersion probe is installed in a standard NPT port or DN flange available on typical silicone reactors. For retrofits, we coordinate installation during a planned maintenance shutdown — full probe installation typically takes a few hours.
What about fluorescence from dyes and fillers in adhesives?
In most formulations, it can be suppressed without changing the wavelength. We use baseline correction, acquisition time modulation, or an extended spectral range — we described five methods in a separate guide. If the compound contains strongly fluorescent dyes, the feasibility study verifies whether a 1064 nm laser configuration is more sensible for the given portfolio.
Do you have implementations in silicone and adhesive production?
We have the most implementations in process chemistry — phenolic and urea resins, cosmetics, fertilizers,, adhesives, hydrocarbons. Silicones and structural adhesives are an area where we are actively developing our portfolio; we enter every project through a feasibility study on customer samples, verifying the specific formulation chemistry before the client commits CAPEX.
How quickly from first contact to inline startup on the line?
Typically 3–5.5 months from engineering workshop to operating system. The first 4–6 weeks are the feasibility study on samples, the next 6–10 weeks are hardware configuration and model calibration, and the final weeks are site installation, validation, and acceptance.
Let’s talk about your process
If you operate an RTV/HTV silicone line, MS-polymer sealant line, or structural adhesive line and see a place where the lack of continuous measurement costs you rework or underestimation of the reactor cycle — let's talk. Here at Gekko Photonics, we select the probe configuration and chemometric model during a 30-minute conversation with an application engineer, and a test measurement on your samples is typically completed within 2 weeks of receiving the material. Write to us with a brief description of the process — we will respond with a feasibility proposal within a few business days.