The benefits of continuous manufacturing include greener and safer processes, better control over reaction conditions, and smaller, more agile systems.[1] Of particular focus is the rational implementation of process analytical technologies (PAT) to monitor and control the process. Online PAT enables the simultaneous monitoring of the system while providing a) robust response to changes and, b) capability to extract fast process information.

The production of diphenhydramine, the most used antihistamine, primarily follows one of two batch processes in industry.[2] Therefore, it is an excellent target for process intensification via continuous flow synthesis.[3] It has been chosen as a model reaction in this work for showcasing the application of PAT to determine the kinetics of the synthesis reaction network, which has not been shown in the literature in the scenario of a nonpolar solvent. In this work, we evaluate low-field flow NMR for PAT applications and was employed in-line with a microfluidic system.

Simulated reaction mixtures have yielded calibration NMR spectra. Figure 1a shows the spectra and calibration curve for chlorodiphenylmethane (DPC). The methane bridge hydrogen is chemically distinct allowing for further analysis and modelling. The DPC simple calibration model yielded an R² of 0.91 and for the DPH model 0.97. Efforts on PLS modelling are placed for the development of a more robust predictive model.

The synthesis of DPH has been evaluated at various temperatures. We also conduct experiments varying initial concentrations of reaction mixtures to unravel the apparent order of the reaction which in turn is used for developing a kinetic model. Using the concentration predictions, we successfully estimate rate constants.

References

1. Burcham, C.L., Florence, A.J., and Johnson, M.D. Annu. Rev. Chem. Biomol. Eng. 9,253 (2018)
2. Snead, D.R. and Jamison, T.F. Chem. Sci. 4, 2822 (2013)
3. Loren, B.P., et. al., R.G. Chem. Sci. 8, 4363 (2017)

Low-field flow NMR enables PAT analysis of microfluidic reaction mixtures and determination of kinetics in the synthesis of active pharmaceutical ingredients. This is shown using diphenhydramine, a pharmaceutical ingredient that has some interest in the literature for continuous manufacturing, but kinetics has not yet been shown.

The isomerization of sugars is challenging to track because of the many byproducts that can form, such as humins. The developed strategy allows the detailed estimation of kinetic parameters while can be used as a means for in-line Process Analytical Control of biomass conversion processes.

Hydroxymethylfurfural (HMF) remains a target of interest in biomass upgrading as a critical steppingstone towards useful biomass-derived products. A unique challenge in HMF generation is understanding the impact of its starting material towards kinetics. Sugars take on a variety of tautomeric forms, with glucose and fructose each having five tautomers. Additionally, the literature has shown the relative concentration of the tautomers directly impacts yield and selectivity towards HMF. Therefore, there is a need to accurately monitor the transformations of sugars in situ in biomass reactions to fully understand their impact. The role of tautomer concentrations may further elucidate another facet of the highly convoluted impact of solvent interactions in such reactions.

Spectroscopy is a powerful tool that facilitates the ability to monitor concentration profiles of reactants in situ, providing detailed kinetics information. Additionally, the simultaneous usage of both Raman and infrared (IR) spectroscopies can provide a robust methodology for concentration tracking as well as molecular interaction insights that are not viable with other methods. The dual-spectroscopic approach also overcomes limitations that is present in each technique. In this work, we present the simultaneous in situ Raman and IR spectroscopy of glucose transformations. First, we present the kinetic experiments of glucose transformations in the absence and presence of catalysts. Next, chemometric analysis of the data and the critical importance of preprocessing spectroscopic data are discussed. Then, we can describe how to obtain kinetic information from the spectra. Finally, we will show correlation between the two spectroscopies via heterospectral two-dimensional correlation spectroscopy (2D-COS) to obtain further insights in the reaction system.

Spectroscopic techniques are unquestionably valuable for understanding structural and compositional changes in complex chemical, biochemical and biological systems. Consequently, spectroscopic techniques are increasingly incorporated to undergraduate/graduate curricula to prepare the next generation of engineers for successful careers in industry and academia where applications include reaction monitoring, quality control, and process control etc. Spectroscopic pedagogy excels in theory and interpretation of spectra and some educators are shifting towards programming-based lessons to introduce certain topics. The interactivity and exposure of computer science in the field of chemistry are advantageous in a future scientist’s career.

Among the challenges in traditional spectroscopic pedagogy, two stand out: the practical aspect of collecting meaningful and reproducible data along with the relevant and necessary pre-processing prior to any interpretation. Lab-based sessions are expensive and certain techniques may be prohibitively long or sensitive for students to broach topics such as instrumental parameters or background drifts. Through programming-based modules developed for education, these topics can be introduced earlier to make higher education level students aware of the reality of spectroscopic techniques in addition to the theory. Continuing our previous work, we have transitioned our previous application to Python to better utilize an open-sourced framework. In addition, the application is further enhanced in this transition for the educator by incorporating chemometric analysis, data extraction, as well as advanced preprocessing techniques that will assist in teaching Raman spectroscopy, such as the Stokes – Anti-Stokes temperature ratio. The benefits and pitfalls of developing using Python, in the lens of previously using MATLAB, are also discussed.

Hydroxymethylfurfural (HMF) has been a target of interest in biomass upgrading as a critical steppingstone towards useful biomass-derived products. Despite recent progress in understanding the competitive reactions leading to low HMF yields, key challenges remain elusive. Among those challenges, one stands out: unravelling solvent effects at the molecular level. The lack of a holistic methodology that allows the simultaneous determination of both solvent-solute interactions and catalyst reactivity hampers our ability for establishing structure-reactivity relationships in liquid systems. While research has come a long way to understand the role solvents such as DMSO have in the success of sugar dehydration to form HMF, there are still questions left unanswered which may lead to the identification (or even prediction) of a more processing friendly solvent while maintaining or even improving current processes.

Spectroscopy is a powerful tool which can enable to discrimination of small but key changes to understand the role solvents play in stabilizing HMF and curating sugar dehydration. In this work, we will present an integrated methodology as a novel experimental protocol that can be utilized to develop fundamental understanding in biomass processing. First, we will discuss the critical role of data preprocessing in data interpretation. It is known in chemometrics that poor choice of data preprocessing methods can negatively impact the strength of a model, but it can even affect conclusion drawn. With generalized two-dimensional correlation spectroscopy (2D-COS), we will show how the presence of water and/or acid in polar aprotic solvents can affect the self-association of solvents which may have direct effect in stabilizing biomass derived molecules. Last, we will focus on developing robust calibration models as means for in-operando analysis of complex reactions.

Spectroscopic techniques are unquestionably valuable for understanding structural and compositional changes in complex chemical, biochemical and biological systems. Consequently, spectroscopic techniques are increasingly incorporated to undergraduate/graduate curricula to prepare the next generation of engineers for successful careers in industry and academia where applications include reaction monitoring, quality control, and process control etc. Spectroscopic pedagogy excels in theory and interpretation of spectra and some educators are shifting towards programming-based lessons to introduce certain topics. The interactivity and exposure of computer science in the field of chemistry are advantageous in a future scientist’s career.

Among the challenges in traditional spectroscopic pedagogy, two stand out: the practical aspect of collecting meaningful and reproducible data along with the relevant and necessary pre-processing prior to any interpretation. Lab-based sessions are expensive and certain techniques may be prohibitively long or sensitive for students to broach topics such as instrumental parameters or background drifts. Through programming-based modules developed for education, these topics can be introduced earlier to make higher education level students aware of the reality of spectroscopic techniques in addition to the theory. We have developed an extensible MATLAB program which allows students to explore Raman instrument parameters to maximize signal to noise ratio while minimizing collection time. Current efforts are focused in extending a MATLAB module to interactively showcase preprocessing techniques with the intent to feed into chemometric interpretation and modelling. Spectral preprocessing is applied in nearly every publication, but justification is often neglected despite the negative impact of incorrect choices in interpretation and modelling. Finally, the benefits and pitfalls of developing MATLAB for the chemical engineering educator is discussed.