FT-NIR Analysis of the Hock Process for the Production of Phenol and Acetone

Significance of the topic

The Hock process (cumene oxidation and acid-catalyzed cleavage) remains a cornerstone industrial route to produce phenol and acetone from low-cost feedstocks. Monitoring the key intermediate, cumene hydroperoxide (CHP), is critical for product quality, safety (CHP is a reactive peroxide) and process efficiency. Rapid, robust in-line analytics enable real-time control of oxidation towers and distillation steps, reduce out-of-spec production, and shorten decision times compared to conventional wet-chemistry assays.

Objectives and study overview

This application note demonstrates the use of FT-NIR spectroscopy (Thermo Scientific Antaris analyzers) for rapid multi-point monitoring of CHP and cumene in the cumene-to-phenol Hock process. Goals were to establish spectral markers, develop chemometric calibrations for CHP and cumene across industrially relevant concentration ranges, evaluate model performance (accuracy, robustness, number of PLS factors), and compare analysis times versus titration and laboratory NIR.

Methodology and calibration strategy

Samples were measured with a fiber-optic probe having a 1 mm optical pathlength to avoid total absorption at lower-frequency CHP bands. Spectra acquisition: 32 co-averaged scans at 8 cm-1 resolution (≈20 s per spectrum). A background scan of clean, dry probe air removed environmental/probe effects. Spectral preprocessing used a 2nd derivative plus a Norris derivative filter (segment length = 11, gap = 0) to enhance hidden spectral features and reduce baseline drift.

Key spectral regions selected for PLS calibration:

• CHP: 5272–4671 cm-1 (notable OH-related peaks near 6800 and 4800 cm-1 observed in raw spectra)
• Cumene: 6000–5457 cm-1 (distinct differences vs. CHP in 6000–5600 cm-1 region)

Calibration models employed partial least squares (PLS) regression with cross-validation and PRESS analysis to determine optimal factor number.

Used instrumentation

The study used the Thermo Scientific Antaris FT-NIR family (Antaris EX highlighted for in-line, hazard-certified operation). Measurements were performed via a fiber-optic immersion probe (1 mm pathlength) compatible with in-line deployment in oxidation towers and distillation streams. Antaris EX/MX platforms support multi-point simultaneous monitoring and streaming of results to process control systems.

Results and discussion

Calibration performance and analytical figures:

• CHP calibration range: 0–80% (w/w). RMSEC = 0.169% and RMSECV = 0.362% — indicating strong predictive performance on unknowns.
• Cumene calibration range: 6–100% (w/w). RMSEC = 0.323% with validation samples matching calibration accuracy.
• PLS complexity: CHP required 3 factors; cumene required 4 factors. PRESS plots showed rapid initial RMSECV decline and minima at low factor numbers, indicating parsimonious models that capture concentration-related spectral variance without overfitting.

Spectral interpretation: CHP exhibits OH-related absorbances absent in cumene (peaks near 6800 and 4800 cm-1). Cumene-specific absorbance changes are most pronounced between 6000 and 5600 cm-1. The 2nd derivative preprocessing emphasized these differences and mitigated baseline variation, improving calibration selectivity.

Speed comparison (key practical result):

• Classical titration (grab sample, lab): ~30 minutes total turnaround.
• Lab NIR analysis (offline): ~5 minutes.
• In-line FT-NIR (real-time probe): ~20 seconds per spectrum.

These time gains enable near real-time trending, alarms, and closed-loop control to prevent extended production of off-spec material.

Benefits and practical applications

Primary advantages demonstrated:

• Rapid, accurate multi-component quantification (CHP and cumene) suitable for continuous high-throughput plants.
• In-line deployment reduces sample handling and measurement delay, improving safety when monitoring reactive peroxides.
• Low model complexity (few PLS factors) provides robust predictions and simpler maintenance/calibration transfer.
• Streaming data supports process control integration for trending, automated adjustments, and alarm generation.
• Economic benefits include reduced laboratory workload, faster process optimization, decreased off-spec product, and rapid ROI for automated analyzers.

Future trends and opportunities

Opportunities to expand and enhance this approach include:

• Wider multi-point networks (simultaneous monitoring across all oxidation towers and distillation stages) for holistic process snapshots.
• Integration of FT-NIR outputs into advanced multivariate process control and model predictive control systems for automated closed-loop optimization.
• Transfer-learning and robust calibration transfer strategies to simplify model deployment across different plants and probe geometries.
• Combining FT-NIR with complementary sensors (e.g., process chromatography or Raman) to strengthen detection of minor impurities and support comprehensive quality assurance.

Conclusion

FT-NIR spectroscopy, implemented with hazard-rated Antaris in-line analyzers and chemometric PLS models, provides fast, accurate monitoring of CHP and cumene in the Hock phenol/acetone process. The method achieves percent-level (sub-percent) calibration errors over industrial concentration ranges with minimal model complexity and enables real-time process control that is impractical with conventional titration. Adoption of in-line FT-NIR delivers safety, quality, and economic benefits for continuous phenol production.

References

1. Heil C. FT-NIR Analysis of the Hock Process for the Production of Phenol and Acetone. Thermo Fisher Scientific Application Note 51711, 2008.