Monitoring Multiphase Processes in Biomass Chemistry

Preservation of the molecular complexity, which is naturally occurring in biomass, is one key element of innovative (lignocellulosic) biorefinery processes to provide renewable and sustainable building block molecules for the chemical industry.

This can be achieved by employing multiphase solvent systems for homogeneously catalyzed reactions which promise high selectivity and catalytic activity to produce biobased building block chemicals such as 5-hydroxymethylfurfural (5-HMF) and 5-methylfurfural (5-MFF).

2-methyltetrahydrofuran (2-MTHF) is an interesting solvent candidate for in-situ product extraction in multiphase reaction systems since it is biomass derived, non-toxic, and stable to acids and bases.

Moreover, the physico-chemical properties of the solvent system 2-MTHF/water can be tailored by the addition of functional gases such as carbon dioxide (CO₂) creating a gas-expanded liquid with beneficial effects for the reaction conducted in such a system, in this content, it has been shown that by-product formation could be suppressed, and extraction kinetics could be improved [1,2].

However, detailed knowledge of the pressure and temperature dependent vapor-liquid-liquid equilibrium (VLLE), e.g., expressed in a thermodynamic model, is necessary for optimal adjustment of the multiphase solvent system and, thus, for the envisioned optimal process design and operation at a technical stage. To collect the required VLLE data for thermodynamic modeling, methods of inline process spectroscopy coupled with powerful chemometrics come into play as they overcome disadvantages of sample withdrawal such as phase equilibrium disturbance or sample decomposition, but allow fast non-invasive, comprehensive, and quantitative phase monitoring.

In this work, VLLE measurements of the multiphase solvent system CO₂/2-MTHF/water are conducted. Mid-infrared (MIR) spectroscopy is used to monitor for the upper organic phase and Raman spectroscopy for the lower aqueous phase to obtain direct insight into the phase composition at different CO₂ gas phase pressures. The recorded data paves the way for thermodynamic modeling of the solvent system [3-5].

Measurements

VLLE measurements are conducted in a stainless steel Nitsch cell that is equipped with two stirrers, rotating in opposite directions for the upper and lower phase, and fiber-optic probes for MIR and Raman spectroscopy (cf. Figure 1). MIR spectroscopy with high sensitivity for water is advantageous for organic phase observation, while Raman spectroscopy is well suited for aqueous phase monitoring due to its low sensitivity for water and high sensitivity for organic molecules. Both measurement techniques are capable for detection of dissolved CO₂.

Fig. 1: Left part: Experimental setup for VLLE measurements comprising an autoclave (1), pressure sensors (2), CO₂ containers (3), stirrers (4),
inline MIR probe (5), inline Raman probe (6), temperature sensor (7), multi-gate valve (8), HPLC pump (9), and temperature controller (10).
Right part: Schematic depiction of the VLLE experiment in the with inline monitoring of the organic and aqueous phase in the autoclave.

Calibration measurements are conducted at 20°C. For the liquid mixtures of 2-MTHF in water (aq. phase, Raman) and water in 2-MTHF (org. phase, MIR), samples are prepared and measured in separate vials. For CO₂ in water (aq. phase, Raman) and CO₂ in 2-MTHF (org. phase, MIR), defined amounts of water or 2-MTHF are added to the Nitsch cell, followed by addition of defined amounts of gaseous compressed CO₂ from the CO₂ containers (cf. Figure 1, mark 3) with subsequent spectra acquisition in equilibrium.

In course of the VLLE measurements, the CO₂/2-MTHF/water-system is measured in equilibrium in a temperature range of 20°C to 90°C and at CO₂ pressures of 0.3 bar to 35.1 bar. For these experiments, defined amounts of water and 2-MTHF are filled into the Nitsch cell. Upon liquid phase equilibration at target temperature, compressed CO₂ is added to the system directly from a CO₂ gas bottle at a defined pressure. MIR and Raman spectra are continuously recorded to detect stable phase equilibrium indicated by acquisition of invariant spectral signals.

Modeling and calibration

For each liquid phase a chemometric model based on Spectral Hard Modeling is created. This is achieved by following the principles of Indirect and Complemental Hard Modeling, employing spectra of the pure solvents 2-MTHF and water, plus the binary mixture spectra of CO₂ in 2-MTHF (MIR) and water (Raman), respectively [6,7]. The resulting mixture Hard Models (HM), fitted to measurements of a ternary system containing CO₂, 2-MTHF, and water, are shown in Figure 2.

MIR and Raman spectrum
Fig. 2: Left: MIR spectrum of the organic phase (solid black line), HM fit (solid red line), pure chemical species models (solid blue line), and baseline (dotted blue line) of the upper organic phase containing CO₂ and H₂O dissolved in 2-MTHF.
Right: Raman spectrum of the lower aqueous phase containing CO₂ and 2-MTHF dissolved in water. Both spectra are recoded at 90°C and 35.1 bar.

As all components in both phases are known and determined, ratiometric calibration is feasible, forcing the total sum of all mole fractions equal to 1. The mixture HM of the organic phase is calibrated with 59 training samples whereas the mixture HM of the aqueous phase is calibrated with 17 training samples. To assess the quality of the model and to quantify its predictivity a leave-10-out cross-validation is performed. The results are provided in Table 1 for the upper organic phase (MIR) and lower aqueous phase (Raman). All components show good calibration results with RMSECV values of one order of magnitude below the adjusted calibration concentrations, proving good model predictability.

Tab.1: Calibration results for the organic (org.,
MIR) and aqueous phase (aq., Raman) given
in terms of RMSECV for CO₂/2-MTHF/water.
Results and extension to bio-molecules

Spectral Hard Modeling informed inline MIR- and Raman spectroscopy to quantify the composition of the upper organic and lower aqueous phase of the ternary system CO₂/2-MTHF/water provide a profound data basis for characterization and thermodynamic modeling of the VLLE in the analyzed temperature and pressure range. An increased phase segregation with increasing CO₂ pressure is observed and can be explained by a less polar gas-expanded organic phase upon CO₂ addition.

These findings motivate the extension of the investigated chemical system by addition of target molecules, sourced from biomass, such as 5-HMF and 5-MFF. After completing the mixture Hard Models by additional pure component models of 5-HMF and 5-MFF and associated calibration experiments, further VLLE measurements are performed at 25°C and different CO₂ pressures to study the effect of CO₂ addition on the distribution of those target bio-molecules in the multiphase system.

The results for the organic phase are exemplarily shown in Figure 3 for 5-HMF (left) and 5-MFF (right). Upon equilibration of 2-MTHF/water (after approx. 20 min experiment time), the components are added to the aqueous phase by pulse injection and subsequently distribute in both phases (after approx. 40 min experiment time). While 5-HMF is removed from the organic phase with increasing CO₂ pressure, the less polar 5-MFF remains in the organic phase. This effect allows to counteract the known phase mediation induced by mineral acids, often used as catalysts in biorefinery processes, to this solvent system. Hence, smart adjustment of CO₂ pressure allows to control the complex VLLE to optimally tailor the solvent system for the specific process requirements.

$Mole fractions$
Fig.3: Mole fractions of 2-MTHF, water, CO₂, and 5-HMF (left) and 5-MFF (right) in the organic phase for different CO₂ pressures of 10-20 bar at 25°C obtained from inline MIR and Spectral Hard Modeling.

Conclusion

Complementary inline MIR and Raman spectroscopy along with physically motivated Spectral Hard Modeling allow smooth and reliable VLLE experiments of the multiphase solvent system CO₂/2-MTHF/water without the disadvantages of conventional sample withdrawal. Extension of the solvent system by step-injection of bio-molecules such as 5-HMF and 5-MFF does not just allow to study their equilibrium distribution at different CO₂ pressures, as done in the present study, but also the associated mass transfer rates to derive extraction kinetics by the time resolved acquisition of the concentration profiles in both the organic and aqueous phase. This will provide further valuable information to design more efficient processes for structure-preserving biomass conversion.

Acknowledgements

The presented applications are developed in the Center for Next Generation Processes and Products (NGP²) of Aachener Verfahrenstechnik (AVT) at RWTH Aachen University, Aachen, Germany. Experiments are performed at Fluid Process Engineering (AVT.FVT) and Process Systems Engineering (AVT.SVT).

Images

Reproduced with courtesy of AVT RWTH Aachen University.

References

Geilen et int. Leitner, Angew. Chem. Int. Edit., 2011, 50(30):6831-6834.
Fu et al., Green Chem., 2017, 19(14):3334-3343.
Aigner et int. Jupke, J. Chem. Eng. Data, 2019, 65(3):993-1004.
Echtermeyer et int. Mitsos, EuroPACT Conference, 2017.
Echtermeyer, Doctoral Dissertation, RWTH Aachen University, 2022.
S-PACT GmbH, Getting Started with Indirect Hard Modeling, (Accessed: 25.04.2025).
S-PACT GmbH, Getting Started with Complemental Hard Modeling (CHM), (Accessed: 25.04.2025).

» Download article (514 KByte)