Process Raman as a comprehensive solution for downstream buffer workflow

Importance of the topic

Raman spectroscopy as an in-line Process Analytical Technology (PAT) addresses a pressing need in biopharmaceutical downstream processing: rapid, non-invasive, multi-component monitoring in aqueous streams without sample preparation. Real-time quantification of formulation excipients and buffer quality during ultrafiltration/diafiltration (UF/DF) and related unit operations reduces reliance on offline laboratory testing, enables tighter process control, and supports automation strategies that improve product quality and manufacturing efficiency.

Objectives and overview of the study

This application note demonstrates the capabilities of the Thermo Scientific MarqMetrix All‑In‑One Process Raman Analyzer to: quantify key excipients (L‑histidine, L‑arginine, sucrose) in real time during bench‑scale UF/DF of an IgG1 monoclonal antibody; assess buffer stability before use; and provide data features usable for quality assessment and potential feedback control. The study evaluates chemometric model development, cross‑instrument transferability, and an observed case of sucrose hydrolysis detected by Raman and confirmed by HPLC.

Methodology

Calibration and model building:

• Design of experiments: Uniform Design (UD) to efficiently span concentration space for L‑histidine (0–15 mg/mL), L‑arginine (0–40 mg/mL), and sucrose (0–200 mg/mL).
• Sample set included mixtures and added IgG1 mAb spectra (5–150 mg/mL in varied matrices) to improve model specificity in protein‑containing streams.
• Chemometrics: Partial Least Squares (PLS) regression using spectral window 800–3235 cm⁻¹.
• Preprocessing: spectral interpolation to common x‑axis (2048 pixels over 60–3250 cm⁻¹), relative standardization using NIST SRM fluorescence method, infinity‑norm normalization on 2900–3230 cm⁻¹, Savitzky‑Golay first derivative (polynomial order 2, window 13), mean centering.
• Model validation: leave‑one‑out cross‑validation (LOOCV) for latent variable selection and external validation with seven independent samples measured on three instruments.

Acquisition and in‑line measurement:

• Sampling optics: FlowCell Probe for calibration/UF/DF runs; BallProbe optic used for buffer stability hold experiment.
• Acquisition parameters: 785 nm excitation at 450 mW, 3000 ms integration, average of 3 spectra → ~18 s total per measurement; flow rate through FlowCell ~100 mL/min.
• UF/DF bench conditions: PES membrane equilibrated with Tris pH 7.0; IgG1 feed 10 g/L; target membrane loading 500 g/m²; UF feed rate 300 L/m²·hr; transmembrane pressure (TMP) maintained 10–15 psi; manual diafiltration by buffer addition to recirculation tank.

Used instrumentation

• Thermo Scientific MarqMetrix All‑In‑One Process Raman Analyzer.
• Thermo Scientific MarqMetrix FlowCell Sampling Optic (in‑line measurements during flow).
• MarqMetrix BallProbe Sampling Optic (hold‑time buffer stability experiment).
• Offline confirmation analyses: HPLC for sugar species and quantitation; pH and osmolarity measurements.

Main results and discussion

Calibration and validation performance:

• PLS models for histidine, arginine and sucrose showed strong correlation to reference values across instruments with R² for prediction >0.98 and prediction absolute errors <5% for final buffer exchange comparisons (table summarized in text: predicted vs reference concentrations yielding absolute errors 1.0% for histidine, 1.4% for arginine, 3.4% for sucrose).
• Reported calibration statistics included low RMSEC/RMSECV/RMSEP values and negligible biases, demonstrating accurate model fit and cross‑instrument transferability.

Real‑time UF/DF monitoring:

• During UF/DF runs the Raman models tracked L‑histidine, L‑arginine and sucrose concentration profiles in the retentate, enabling direct comparison to theoretical expectations and offline HPLC confirmation.
• Observed sucrose behavior: real‑time Raman predictions captured expected changes during diafiltration and subsequent concentration steps; however, discrepancies in some runs (predicted higher concentrations relative to HPLC) were traced to the appearance of new species not present in calibration data.

Buffer stability and unusual event detection:

• In a buffer hold study at room temperature with brief air exposure, Raman monitoring detected a decline in sucrose signal from ~86 mg/mL to ~57 mg/mL over ~14–15 days.
• Concurrent spectral changes revealed emergence of bands attributable to glucose and fructose (observed ~525 cm⁻¹ and ~640 cm⁻¹ respectively) and decrease of sucrose‑specific bands (~835 cm⁻¹ and ~550 cm⁻¹). HPLC confirmed sucrose hydrolysis to glucose and fructose.
• Physicochemical measurements showed a ~1 unit drop in pH and ~40% increase in osmolarity over the hold period; acidic hydrolysis of sucrose was considered the likely mechanism.
• Model limitation: the sucrose PLS model had not been trained with glucose/fructose spectra, causing some prediction bias when hydrolysis products were present; augmenting the training set would reduce this bias.

Quality assessment metrics and control capability:

• Projection statistics (reduced Hotelling T² and reduced Q residual) were used to detect deviation from the calibration space as hydrolysis proceeded. Raised Q residuals and low T² after day 5 indicated buffer samples would fail predefined control limits, providing a pragmatic in‑line QC flag.

Practical benefits and applications

• Enables real‑time quantification of multiple excipients simultaneously during UF/DF, supporting decisions on diafiltration volume (Vdf) and endpoint determination rather than relying on empirical assumptions (notably relevant when Gibbs‑Donnan or other retention effects alter excipient behavior).
• Reduces time and cost associated with offline analytics and potential rework from undetected buffer quality issues.
• Supports process automation: combined protein and excipient readouts allow closed‑loop or semi‑automated controls for diafiltration endpoints and concentration steps.
• Provides in‑line buffer quality screening prior to use, reducing the risk of batch failure driven by degraded or contaminated buffers.

Limitations and recommended improvements

• Model scope: models should include expected degradation or impurity species (e.g., glucose, fructose) to avoid biased quantification if conversion products appear.
• Instrument standardization: inter‑instrument transfer requires consistent standardization (NIST SRM‑based relative intensity correction) and common preprocessing pipelines.
• Process variability: cause(s) of pH drift and osmolarity increase (e.g., microbial contamination, CO2 dissolution, buffer preparation error) should be investigated and mitigated to avoid false alarms or degraded product stability.

Future trends and potential applications

• Integration of Raman PAT with process control systems to enable automated diafiltration endpoints, TMP adjustments, and adaptive feed strategies based on simultaneous protein and excipient concentrations.
• Expansion of chemometric libraries to cover common degradation products, excipient variants and broader formulation spaces to improve robustness across manufacturing campaigns.
• Hybrid monitoring strategies combining Raman with orthogonal sensors (e.g., UV‑Vis, conductivity, pH, osmolarity) and multivariate process control to strengthen detection of complex failure modes.
• Deployment at larger scale and in cGMP environments with qualification/validation workflows, and use of model maintenance approaches for long‑term drift compensation.

Conclusion

Process Raman implemented with appropriate chemometric models and sampling optics provides a powerful in‑line PAT for downstream bioprocessing. The study demonstrates accurate, cross‑instrument quantification of L‑histidine, L‑arginine and sucrose (prediction errors <5%), real‑time detection of buffer degradation (sucrose hydrolysis), and the utility of projection statistics for buffer quality assessment. When combined with protein monitoring, process Raman unlocks opportunities for tighter process control and automation of UF/DF operations, improving product quality assurance and operational efficiency. Future work should expand model coverage to degradation products and formalize integration pathways for closed‑loop control.

Reference

1. Zhang L.; Liang Y.-Z.; Jiang J.-H.; Yu R.-Q.; Fang K.-T. Uniform Design Applied to Nonlinear Multivariate Calibration by ANN. Analytica Chimica Acta 1998, 370(1), 65–77.
2. Choquette S. J.; Etz E. S.; Hurst W. S.; Blackburn D. H.; Leigh S. D. Relative Intensity Correction of Raman Spectrometers: NIST SRMs 2241–2243 for 785 nm, 532 nm, and 488.5/514.5 nm Excitation. Applied Spectroscopy 2007, 61(2), 117–129.
3. Torres A. P.; Oliveira F. A. R.; Silva C. L. M.; Fortuna S. P. The Influence of pH on the Kinetics of Acid Hydrolysis of Sucrose. Journal of Food Process Engineering 1994, 17(2), 191–208.
4. Wiercigroch E.; Szafraniec E.; Czamara K.; Pacia M. Z.; Majzner K.; Kochan K.; Kaczor A.; Baranska M.; Malek K. Raman and Infrared Spectroscopy of Carbohydrates: A Review. Spectrochimica Acta Part A 2017, 185, 317–335.
5. Kumar S.; Martin E. B.; Morris J. Detection of Process Model Change in PLS Based Performance Monitoring. IFAC Proceedings Volumes 2002, 35(1), 125–130.
6. Agrawal P.; Wilkstein K.; Guinn E.; Mason M.; Serrano Martinez C. I.; Saylae J. A Review of Tangential Flow Filtration: Process Development and Applications in the Pharmaceutical Industry. Organic Process Research & Development 2023, 27(4), 571–591.
7. Nolasco M.; Pleitt K.; Khadka N. Using a Process Raman Analyzer as an In‑Line Tool for Accurate Protein Quantification in Downstream Processes. (Application note).
8. Nolasco M.; Pleitt K.; Khadka N. Raman‑Based Accurate Protein Quantification in a Matrix That Interferes with UV‑Vis Measurement. (Application note).