Tag: Tijmen Bos

Categories
Publications

Size-Resolved Surface Charge Analysis of Polymer Nanoparticles: From Fundamental Measurement to Collaborative Innovation

Size-Resolved Surface Charge Analysis of Polymer Nanoparticles: From Fundamental Measurement to Collaborative Innovation

Polymer nanoparticles (PNPs) play an increasingly central role in contemporary materials science, with applications ranging from advanced coatings and paints to biomedical drug delivery systems. Their functional performance is governed not only by particle size, but also by surface charge density (SCD), a key parameter that determines colloidal stability, interparticle interactions, and adhesion to interfaces. Despite its importance, SCD has traditionally been assessed only as a global, averaged value, obscuring the intrinsic heterogeneity that arises during nanoparticle synthesis.

Recent analytical advances now make it possible to resolve this complexity. A notable example is the work by Kruijswijk and colleagues, who developed a capillary zone electrophoresis (CZE)-based methodology to determine particle size–resolved SCD distributions of polymer nanoparticles. Their approach provides a more nuanced understanding of nanoparticle surface chemistry and opens new opportunities for rational material design.

(A) Schematic of a nanoparticle in an electrolyte solution surrounded by an electrical double layer comprised of the inner Stern layer and the outer diffuse layer; (B) Dependency of the reduced electrophoretic mobility (Em) on the particle size for NPs with a SCD of 0.04 C∙m-2 according Ohshima’s model (C) Dependency of the reduced electrophoretic mobility (Em) on the particle SCD for NPs with a size of 100 nm according Ohshima’s model.

Moving Beyond Average Values

Conventional techniques for nanoparticle charge characterization typically report a single zeta potential value for an entire population. While useful, such measurements implicitly assume homogeneity and therefore overlook variations that may critically affect performance. In reality, polymer nanoparticles often exhibit broad distributions in both particle size and surface charge, reflecting the stochastic nature of polymerization processes and surfactant interactions.

The method introduced by Kruijswijk et al. addresses this limitation by combining CZE with theoretical electrophoretic models and chemometric deconvolution. By carefully separating the contributions of particle size distribution and intrinsic charge heterogeneity to electrophoretic peak broadening, the authors demonstrate that it is possible to extract detailed SCD distributions for industrially relevant polymer nanoparticles.

This approach was validated using polystyrene nanoparticle standards and subsequently applied to poly(methyl methacrylate–methacrylic acid) and polyurethane nanoparticles with varying monomer compositions. The results clearly showed that surface charge density is not only composition-dependent but also size-dependent, with smaller particles often exhibiting higher mean SCD values. Such insights cannot be obtained from bulk measurements alone.

Implications for Industrial and Applied Research

The ability to determine size-resolved SCD distributions has important implications across multiple application domains. In coatings and paints, for example, subtle differences in nanoparticle charge can influence dispersion stability, film formation, and substrate adhesion. In biomedical contexts, surface charge plays a critical role in cellular uptake, biodistribution, and protein adsorption.

Equally important is the observation that adsorbed ions and surfactants can significantly distort apparent surface charge. By introducing a neutral surfactant to displace adsorbed ionic species, the authors were able to distinguish intrinsic polymer charge from extrinsic effects. This highlights the necessity of carefully controlled analytical environments when translating laboratory measurements to real-world formulations.

Overall, the study illustrates how advanced separation science can provide actionable knowledge for materials engineering, enabling the optimization of nanoparticle systems based on their true physicochemical properties rather than averaged proxies.

 

Electropherograms with electrophoretic mobility axis of three PUR NPs of different percentage DMPA (percentage DMPA and mean NP size indicated in color) obtained with CZE analysis using a BGE of 3.75 mM sodium tetraborate (pH 9.2) in (A) absence of Brij-35 and (B) presence of 0.10 mM Brij-35. (C+D) Calculated size-resolved SCD distributions for the PUR4 NP in (C) absence of Brij-35 and (D) presence of 0.10 mM Brij-35 obtained by applying Ohshima’s model followed by deconvolution in which the contributions of PSD and sample injection to the CZE peak width are negotiated. (E+F) Global SCD distributions for the PUR4 NP in (E) absence of Brij-35 and (F) presence of 0.10 mM Brij-35 obtained from projecting the data from figures C and D onto the SCD axis, respectively.

Collaboration as a Prerequisite for Impact

Research of this nature sits at the intersection of analytical chemistry, polymer science, data analysis, and industrial application. Successfully translating such methodologies from the laboratory into industrial practice requires more than technical excellence; it demands structured collaboration between disciplines and sectors.

This is precisely where IDEAS plays a crucial role. As a collaborative company and innovation environment, IDEAS provides a platform where academic researchers, industrial partners, and analytical experts can work together on complex projects such as advanced nanoparticle characterization. By fostering shared access to expertise, instrumentation, and data-driven methodologies, IDEAS enables the co-development of analytical solutions that are both scientifically rigorous and industrially relevant.

Within such a collaborative setting, techniques like size-resolved SCD analysis can be further refined, validated across different material classes, and integrated into quality control or product development workflows. Moreover, IDEAS offers a context in which fundamental research questions, such as charge heterogeneity and surface chemistry, can be directly linked to application-driven challenges faced by industry.

Towards Data-Informed Materials Design

The work discussed here exemplifies a broader shift towards data-rich, distribution-aware characterization of functional materials. Rather than relying on single-value descriptors, researchers and engineers are increasingly equipped to consider the full complexity of nanoparticle populations.

By combining advanced analytical techniques with collaborative innovation frameworks such as those provided by IDEAS, this knowledge can be translated into smarter materials, more robust processes, and ultimately more sustainable and high-performance products. For initiatives like CAST Amsterdam, such developments underscore the value of connecting fundamental science with collaborative infrastructures that support real-world impact.

Citation

Citation

Jordy D. Kruijswijk 1, Tijmen S. Bos 1, Billy van Zanten, Ton Brooijmans, Ron A.H. Peters, Kevin Jooß, and Govert W. Somsen.(2025). Assessment of Particle Size-Resolved Surface-Charge Density Distributions of Polymer Nanoparticles by Capillary Zone Electrophoresis. Analytical Chemistry, https://doi.org/10.1021/acs.analchem.5c05189

1 = Equal contributions

Categories
Publications

How Data processing and Peak Shape Affects What We See: Understanding Detection in Two-Dimensional Chromatography

As multidimensional chromatography continues to expand its reach across complex analytical challenges, the precision of data interpretation increasingly depends on what happens after separation. In comprehensive two-dimensional chromatography (2D-LC and 2D-GC), even subtle differences in how peaks appear, how wide they are, how they tail, or how they overlap, can determine whether an analyte is detected accurately or missed entirely.

A new study by Nino Milani, Nard Schellekens, Alan Garcia Cicourel, Rob Edam, Gabriel Vivo Truyols, Bob Pirok, and Tijmen Bos, published in Analytica Chimica Acta (2025), takes on this fundamental question: how do peak characteristics affect the performance of peak detection algorithms in 2D chromatography?

 

Simulating 700,000 Chromatograms to Understand Detection

Peak detection lies at the heart of chromatographic data processing. It converts raw signals into measurable information, retention times, areas, and ultimately, quantitative results. In 2D chromatography, where each measurement spans two retention-time axes with vastly different information densities, this process becomes far more complex.

Milani and colleagues developed a data simulator capable of generating highly realistic chromatograms derived from experimental probability distributions. By systematically varying peak properties, such as width, asymmetry, shape, relative intensity, and modulation shifts, they created an extensive virtual dataset encompassing over a terabyte of data.

Two distinct detection algorithms were then evaluated:

  • The two-step algorithm, which detects peaks in the second-dimension traces and subsequently groups them across modulations.
  • The watershed algorithm, which treats the chromatogram as a 2D landscape, detecting peaks through image segmentation.

Revealing the Subtle Effects of Peak Properties

Through this systematic approach, the researchers could visualize the recovery of true peak areas across thousands of configurations. The results were expressed as heatmaps, each pixel representing how accurately a simulated peak was detected under a given scenario.

Several trends emerged:

  • Peak width strongly affected both algorithms. As peaks became broader, recovery decreased, especially for the two-step approach, due to modulation “exclusion,” where low-height modulations fell below detection thresholds.
  • Modulation shifts, common in comprehensive GC×GC and shifting-gradient LC×LC, proved particularly disruptive. Both algorithms struggled when retention times drifted between modulations, with the watershed method especially prone to splitting peaks.
  • Asymmetry and shape (tailing or fronting) affected both methods modestly, although the watershed algorithm’s topographical nature made it more sensitive to skewed peaks.
  • Peak ratios (relative intensities) influenced results through thresholding effects, small peaks beside large ones were often underestimated or missed entirely.

No Perfect Algorithm—But Better Understanding

Perhaps the most insightful conclusion was that there is no universally best peak detection strategy. Instead, the choice depends on the chromatographic context:

  • The watershed algorithm is advantageous for broader, more symmetrical peaks, typical of polymer or group-type analyses.
  • The two-step algorithm performs better for sharper, complex chromatograms where peaks are closely spaced and moderately deformed, such as in natural product or petroleum analyses.

The study also underscores how pre-processing choices, such as interpolation or threshold settings, can profoundly shape results, sometimes more than the algorithm itself.

Towards More Reliable Data Interpretation

By disentangling the relationships between signal characteristics and detection performance, this work provides a quantitative foundation for algorithm selection and design. The methodology, openly shared through the GitHub repository CAST-Amsterdam/PeakDetectionBenchmark, offers a new benchmark for evaluating detection strategies objectively and reproducibly.

Ultimately, Milani et al. remind us that in advanced chromatography, understanding the data is as critical as generating it. As automation and machine learning increasingly drive method development, robust and transparent detection remains the cornerstone of analytical reliability.

Citation

Milani, N.B.L., Schellekens, N.C.A., Garcia Cicourel, A.R., Edam, R., Vivo Truyols, G., Pirok, B.W.J., & Bos, T.S. (2025). Evaluation of the relationship between peak characteristics and detection performance in two-dimensional chromatographic data. Analytica Chimica Acta, 1377, 344634. https://doi.org/10.1016/j.aca.2025.344634

Categories
Publications

Block-Length Distributions Using Fragmentation Data Obtained from Tandem Mass Spectrometry

Researchers Rick van den Hurk, Dr. Tijmen Bos, and Dr. Bob Pirok have, together with scientists Dr. Ynze Mengerink (Brightlands Chemelot Campus) and Prof. Dr. Ron Peters (Covestro, HIMS), developed a new algorithm that can analyze copolymers and determine their block structure, something that was previously out of reach with existing techniques.

Polymers are all around us, from the coatings on your phone and the materials in your running shoes to life-saving drug-delivery systems and medical implants. Many of these advanced materials are copolymers, which are made by combining different types of chemical building blocks.

Interestingly, even when two copolymers have the same overall composition, the way these building blocks are arranged can lead to drastically different properties. For example, one polymer might be rigid while another is flexible, or one transparent while another is opaque. Within a single batch, this arrangement can vary from molecule to molecule. This variability can be described using a concept called the block-length distribution (BLD), which captures how frequently different block arrangements occur. This distribution plays a key role in determining a material’s performance characteristics, including its flexibility, strength, and biodegradability.

Until now, accurately measuring these distributions at the molecular level has been a major challenge. Traditional techniques like nuclear magnetic resonance could only offer averaged information. The team’s newly developed algorithm changes that by combining tandem mass spectrometry (MS/MS) data with a smart computational approach that takes fragmentation behavior into account. The algorithm allows researchers to reconstruct how blocks are distributed within a copolymer sample, giving a much more detailed picture of the material’s internal structure.

This method has already been successfully applied to study polyamides and polyurethanes, important industrial polymers found in everything from textiles to insulation foams. Notably, the findings showed that even polymers with the same chemical makeup can have very different block distributions, depending on how they were synthesized. These subtle differences can explain variations in material performance that would otherwise remain hidden.

 

The ability to determine BLDs with such precision not only improves our understanding of polymer chemistry but also opens the door to the rational design of next-generation materials. By fine-tuning the block arrangement, scientists and engineers can tailor materials more precisely to specific applications. It could also support the development of more sustainable materials, as better control over structure may lead to improved recyclability or allow for the use of bio-based feedstocks.

This work is part of the PARADISE project, a collaboration between academic institutions (VU Amsterdam and the University of Amsterdam) and industrial partners including Covestro, DSM, Shell, and Genentech, aimed at driving forward innovation in polymer research.

Relevant article: https://doi.org/10.1021/acs.macromol.5c00297,

Categories
Conferences

CAST group participates in full force at HPLC 2025 with posters and lectures

CAST Group 2025

It is a tradition within our CAST group to attend one major scientific conference each year as a unified team. This provides an opportunity to present our work, strengthen internal cohesion, and engage with the broader research community. With liquid-phase separations as a common thread across our projects, all scientists from the CAST group, as well as those involved in related projects embedded within the Centre for Analytical Sciences Amsterdam, were therefore present at the HPLC 2025 symposium in Bruges, Belgium.

The full CAST group with fellow CASA members at the conference dinner party on Wednesday at HPLC2025 in Bruges, Belgium.

Four poster nominations for the Gargano team

The CAST group contributed to the symposium with a large number of posters. On Monday and Tuesday, our young scientists presented and discussed their work with the delegates. All four posters of the team of Andrea Gargano were nominated for the Best Poster Award.

Improving IgG Glycoform Analysis with Advanced HILIC-MS

Thomas Holmark presented his latest advances in improving the sensitivity and reliability of IgG glycoform analysis. Their work focuses on developing capillary-based HILIC-MS methods to better characterize glycosylation patterns in immunoglobulin G (IgG), which are important biomarkers for disease progression. By optimizing monolithic column synthesis and implementing self-packed trap columns, the team significantly reduced sample carryover and enhanced detection sensitivity—paving the way for more accurate, low-input analysis of IgGs secreted by in-vitro stimulated B-cells.

Masa Sawisawa, Thomas Holmark, Annika van der Zon and Ziran Zhai were nominated for a Best Poster Award.

Annika van der Zon receiving her best poster pitch prize.

Third poster-pitch prize for monolithic HILIC-MS for Monoclonal Antibody Glycoform Profiling

Annika van der Zon introduced an acrylamide-based monolithic HILIC stationary phase, specifically designed to overcome the limitations of conventional silica-based materials. This new column enables high-resolution separation of intact mAb glycoforms, significantly outperforming traditional LC-MS methods such as RPLC. Using trastuzumab as a model, the team achieved chromatographic separation and mass spectrometric identification of up to 17 N-glycoforms—demonstrating enhanced sensitivity to minor variants. This development represents a major step forward in therapeutic antibody characterization, offering a powerful tool for quality control and drug development

NanoSEC-nMS Enhances Native Protein Complex Characterization

Ziran Zhai was also nominated for the poster-pitch prize and presented an innovative nanoflow size-exclusion chromatography-native mass spectrometry (nanoSEC-nMS) method at HPLC 2025, pushing the boundaries of intact protein and protein complex analysis. By operating at ultra-low flow rates (500 nL/min) and using capillary-format columns, the method drastically reduces sample requirements while preserving native protein structures and non-covalent interactions. This approach enables sensitive, high-resolution analysis of complex biomolecular assemblies, even from limited or biologically relevant samples like urine. With enhanced ionization stability and compatibility with high-salt buffers, nanoSEC-nMS represents a powerful advancement for native top-down proteomics and structural biology workflows.

Pharmaceutical Analysis Poster Award for Masa Serizawa

Masa Serizawa showcased a major advance in the analysis of biodegradable polymers, focusing on poly(lactic-co-glycolic acid) (PLGA)—a cornerstone material in drug delivery systems. The team developed optimized SEC-MS and normal-phase LC methods to achieve precise characterization of PLGA microstructure, including molecular weight distribution (MWD), chemical composition distribution (CCD), and functionality-type distribution (FTD). By minimizing fragmentation during electrospray ionization with cesium iodide, they enabled reliable identification of PLGA isomers and block structures. Furthermore, their new NPLC approach extends analysis to high-molecular-weight PLGAs (up to 185 kDa), offering detailed insight into both end-group functionality and lactic/glycolic acid ratios. These innovations pave the way for the design of more effective, tailored PLGA-based drug delivery platforms. He was awarded the Best Pharmaceutical Analysis Poster Award for his great work.

Masa Serizawa receiving the Best Pharmaceutical Analysis Poster Award.

Other poster contributions from PhD candidates included those of Sanne Boot, who showcased her efforts to simulate LC×LC-MS data for optimizing oligonucleotide separations, enabling the exploration of separation conditions within complex chemical spaces. PhD candidate Gerben van Henten showed his work on the evaluation of chromatographic response functions. His poster explained how some optimization methods struggle with chromatographic response functions that fundamentally were believed to be suitable.

Several CAST graduate students were also present at HPLC2025. Lonneke van Dalen presented her poster on developing a new peak integration method using neural networks in collaboration with Unilever.

Merel Konings demonstrated the optimization of light degradation reactors to conduct degradation studies in liquid-phase separations, and Rebecca Gibkes presented her work on plate-height modeling and peak parking as part of a collaboration between the University of Amsterdam, the Vrije Universiteit Brussel and Gustavus Adolphus College.

Rick van den Hurk as Csaba Horvath Finalist

Rick van den Hurk presenting his latest work on the effect of a lack of radial mixing in systems where flows are combined.

Rick van den Hurk was nominated as Csaba-Horváth finalist and addressed a key obstacle in expanding two-dimensional liquid chromatography (2D-LC): the solvent mismatch between coupled separation dimensions. His work focused on active modulation strategies, such as SPAM, ACD, and ASM, that dilute the first-dimension effluent with a weaker solvent to improve retention and peak shape in the second dimension. Using experimental setups and computational fluid dynamics simulations, the team investigated how radial mixing at T-junction interfaces impacts analyte retention. Van den Hurk explained how these findings provide valuable insights into optimizing flow conditions and mixer design, paving the way for broader and more robust application of 2D-LC in industrial and research settings. This work is conducted in collaboration with the group of Dwight Stoll at Gustavus Adolphus College.

Bob Pirok was invited to present a lecture on the state of machine learning in chromatography, and explained during his presentation the importance of equiping machine-learning workflows with suitable data processing strategies. A key part of the lecture focused on the work by CAST researcher Nino Milano and his efforts to simulate realistic 2D chromatographic data for the evaluation of peak detection algorithms.

Andrea Gargano presented a novel HILIC-MS approach using custom-designed acrylamide-based monolithic columns to improve the analysis of intact biomacromolecules. These columns offer enhanced selectivity and reduced secondary interactions, enabling the separation of closely related variants that are challenging for traditional RPLC-MS. Gargano demonstrated applications which included the resolution of oligonucleotide impurities, glycated protein isomers, and monoclonal antibody glycoforms, advancing analytical capabilities in therapeutic development and quality control. 

Luca Tutis’s contribution on the development of an ion-pair HILIC method for oligonucleotide impurity analysis was selected for a talk. This method significantly alters HILIC selectivity for ONs by emphasizing nucleobases and conjugated groups (GalNAc) over highly polar phosphate groups. This enabled the resolution of deamination, non-conjugated, and PS-PO converted impurities. Importantly, this IP-HILIC approach is fully compatible with MS, crucial for accurate impurity identification.

Honorary CAST member and emeritus professor Peter Schoenmakers lectured on the sense and nonsense of artificial intelligence in chromatography and gave a historical overview in his lecture along with applications that he placed into context using fundamental principles known to chromatographers.

Special events

Advancing Education in Separation Science

Andrea Gargano, Simone Dimartino and Martina Catani organized the Education session workshop, introducing in the HPLC program a new space to discuss innovation in education. Many delegates joined discussing innovations in separation science education (including teaching materials, software based tools, project based learning and a session discussing perspectives in skills and knowledge that the analytical scientist of the future should be educated on) including Bob Pirok.

Second prize in HPLC Tube

Dissemination of scientific work is incredible important to inform society about the latest progress. In this context, the symposium annually hosts the HPLC Tube where researchers can make brief videos to explain their work. CAST researchers Andrea Gargano, Annika van der Zon, Ziran Zhai and Thomas Holmark won the third prize in this category on their video entitled: “Stay Intact: AcryHILIC’ Glycoform Impact.”

Scientists discussed the state of current education practice of chromatography and how to improve this further.

Bob dedicating the first official copy to CASA leader and co-teacher Prof. Govert Somsen.

Launch of educational textbook Analytical Separation Science

Bob Pirok and Peter Schoenmakers launched their textbook Analytical Separation Science. The book is designed to support students, educators, and practitioners by offering a comprehensive and structured overview of modern separation techniques. With clear explanations, didactic modules, and academic exercises, it serves as a valuable resource for classroom teaching and self-study alike. The book is accompanied by an interactive website (https://ass-ets.org), featuring figures, lectures, and supporting materials to bridge the gap between theory and practice.

Short course in AI in Chromatatography

Bob Pirok and Tijmen Bos delivered a well-attended short course on Artificial Intelligence in Chromatography, introducing participants to the growing role of AI in analytical method development and data processing. The course covered foundational concepts in machine learning, modern optimization techniques, and practical applications across the chromatographic workflow, from peak detection to intelligent gradient design. Through interactive examples and real-world case studies, the session provided researchers and practitioners with the tools to understand and begin applying AI in their own laboratory environments.

Scientists discussed the state of current education practice of chromatography and how to improve this further.

Categories
Conferences

Short Course on AI in Chromatography

Screenshot

Machine learning is considered increasingly important in analytical separation science because of its potential to enable faster and more accurate interpretation of complex, high-dimensional data from techniques like chromatography and mass spectrometry.  Additionally, machine learning accelerates method development and improves reproducibility, leading to more efficient and reliable analytical workflows.

Introduction to Artificial Intelligence

To help both newcomers and experienced practitioners in the field, Bob Pirok and Tijmen Bos were invited to give a short course entitled Introduction to Artificial Intelligence in Chromatography at the 54th International Symposium on High Performance Liquid Phase Separations and Related Techniques (HPLC2025) in Bruges, Belgium.

The course, designed for both academic and industry scientists, was structured in four parts. It began with a clear introduction to the foundations of artificial intelligence, including the historical context and core concepts such as regression, optimization, and pattern recognition. In the second part, the presenters expanded into modern machine learning techniques — from support vector machines to neural networks and reinforcement learning approaches such as Q-learning and Proximal Policy Optimization (PPO).

The third part showcased real-world applications in chromatography, including predictive modeling for retention time, peak detection using neural networks, and data-driven optimization of method parameters. The session concluded with hands-on exercises, challenging participants to apply what they had learned to realistic chromatographic problems.

“AI is not a replacement for analytical expertise, it’s an extension of it,” said Dr. Pirok. “With the right understanding, these tools can help us interpret complex data faster and develop better methods with fewer experiments.”

Our goal is to demystify AI for chromatographers. These technologies are no longer futuristic — they’re ready to be applied, provided we know how to ask the right questions.

Dr. Tijmen Bos

Dr. Bos added, “Our goal is to demystify AI for chromatographers. These technologies are no longer futuristic — they’re ready to be applied, provided we know how to ask the right questions.”

The short course also served as a platform to emphasize responsible and informed application of machine learning in laboratory settings. The presenters stressed the importance of data quality, domain knowledge, and understanding the assumptions behind different models.

Building on this experience, the Introduction to Artificial Intelligence in Chromatography course is also available through IDEAS as a tailored training program for academic groups, research institutes, and industrial laboratories. Delivered either on location or in a dedicated workshop format, the course can be adapted to the specific analytical techniques, datasets, and challenges relevant to the participating organization. By combining conceptual foundations with practical, application-driven examples, IDEAS ensures that participants not only understand machine learning principles, but are also equipped to apply them directly within their own chromatographic workflows.

Categories
Publications

Introducing an algorithm to accurately determine copolymer block-length distributions

Copolymers are the foundation of many high-performance materials used in advanced applications such as medical devices, implants, electronics, and self-healing coatings for aerospace and space exploration. Their material properties—such as flexibility, toughness, or responsiveness—can be finely tuned by adjusting polymer characteristics like molecular weight, chemical composition, and block-length distribution (BLD).

While molecular weight and composition are routinely analyzed, the BLD—describing how monomer blocks are arranged along the polymer chain—remains difficult to measure, particularly for copolymers composed of more than one type of monomer. Understanding and controlling BLD is crucial because it plays a pivotal role in determining mechanical, thermal, and phase-separation behavior. However, current methods, such as NMR or pyrolysis-GC-MS, have limitations in accurately and comprehensively characterizing BLDs.

 

Our Solution
In this study, we introduce a computational approach that enables the quantitative determination of block-length distributions from copolymer fragmentation data. We developed and validated an algorithm using both simulated copolymer sequences and analytical solutions to generate ground-truth fragment data. This allowed for an objective evaluation of algorithm performance—something not previously achievable.

The algorithm incorporates a trust-region-reflective optimization strategy and was tested under various conditions, including data noise and fragment size limitations. When fragment data containing chains of up to four monomers (tetramers) were included, the algorithm consistently reconstructed BLDs with high accuracy, achieving similarity coefficients (SC) above 0.99 compared to the known distributions.

https://doi.org/10.1016/j.aca.2025.343990

 

Key Innovations

  • High Accuracy: Outperforms existing algorithms in BLD reconstruction from mass spectrometry-based data.

  • Versatile: Capable of handling complex distribution shapes, including non-unimodal and asymmetric distributions.

  • Robust to Noise: Maintains accuracy even when fragment data includes measurement noise.

  • Objective Evaluation: Enables benchmarking of BLD algorithms using simulated data with known parameters.

 

Practical Relevance
The algorithm was also applied to experimental polymer systems such as polyamides and polyurethanes, demonstrating its applicability to real-world materials. This makes it a powerful tool for synthetic chemists seeking to design materials with tailored properties by manipulating block structures.

 

Future Directions
Translating this approach from simulation to experimental data introduces new challenges. Mass spectrometry data may be affected by ionization efficiencies and fragmentation biases, while NMR may suffer from overlapping signals in complex systems. To address this, future research will focus on:

  • Incorporating fragmentation preferences based on bond type or analytical method.

  • Developing preprocessing pipelines tailored to specific instrumentation.

  • Extending the algorithm to support more than two monomer types, while managing the increased computational complexity.

 

Conclusion
This work represents a significant advancement in the field of polymer analytics. For the first time, researchers can objectively and accurately reconstruct the block-length distribution of complex copolymers from fragment data. By making this tool available, we aim to empower chemists and materials scientists in designing next-generation materials with precisely engineered microstructures.

Categories
Publications

Unsupervised LC method development with AutoLC

Unsupervised LC method development with AutoLC
Unsupervised LC method development with AutoLC

In an international and interdisciplinary collaboration, CAST members Tijmen Bos, Stef Molenaar, Jim Boelrijk, Leon Niezen and Bob Pirok have demonstrated unsupervised LC method development with AutoLC. This is the first automated LC-MS method development workflow. It was applied it to a complex antibody digest sample. The work was recently published in Analytical Chemistry as cover article [1].

The majority of liquid chromatography (LC) methods are still developed in a conventional manner, that is, by analysts who rely on their knowledge and experience to make method development decisions. To tackle this problem, several tools utilizing design-of-experiment workflows, retention modeling based on experimental data and/or chemical structure information have been developed and even commercialized.

However, these approaches are generally difficult to scale with sample complexity and require significant user input to operate. Consequently, high-resolution separation technology and multi-dimensional systems have not been economically feasible for routine use. To improve the accessibility of state-of-the-art separation technology, the Pirok group at the University of Amsterdam is developing a workflow capable of unsupervised method development.

This has led to the present demonstration of a novel, open-source algorithm for automated and interpretive method development of LC(−mass spectrometry) separations (“AutoLC”). The scientists constructed a closed-loop workflow that interacted directly with the LC system and ran unsupervised in an automated fashion. 

The first demonstration of AutoLC was published as front cover article in Analytical Chemistry
The study was published as feature article in Analytical Chemistry.
Unsupervised LC method development with AutoLC
Schematic overview of the generic workflow employed by the AutoLC algorithm using retention modeling (top, blue) or BO (bottom, pink).

The team tested the algorithm using two newly designed method development strategies. The first utilized retention modeling, whereas the second used a Bayesian-optimization machine learning approach. In both cases, the algorithm could arrive within 4–10 iterations (i.e., sets of method parameters) at an optimum of the objective function, which included resolution and analysis time as measures of performance.

Retention modeling was found to be more efficient while depending on peak tracking, whereas Bayesian optimization was more flexible but limited in scalability. We have deliberately designed the algorithm to be modular to facilitate compatibility with previous and future work (e.g., previously published data handling algorithms).

AutoLC was tested on a peptide digest mixture.
The AutoLC framework was tested on an antibody digest sample. A) example of a generic scouting measurement, B) proposed optimum at the 4th iteration. Reproduced with permission of [1].

The degree of separation is often quantified as the resolution between chromatographic peaks, which can be written as a product of retention, selectivity and chromatographic efficiency. Currently, the AutoLC framework largely focuses on retention, but contemporary efforts have shifted focus to include selectivity. Support of validation is the logical next step thereafter.

AutoLC leverages earlier studies and interdisciplinary expertise

The AutoLC framework is the product of a several years of scientific studies that were conducted within public-private partnerships by the group of Pirok. These projects focused relevant aspects such as peak tracking [2,3], machine learning [4], and gradient deformation [5]. The AutoLC framework was designed to be modular so as to leverage global initiatives by the scientific community that were published in literature. Currently, the development of the framework is supported by funding from several grants from the Dutch Research Council (NWO). It is the prime topic of the UPSTAIRS project.

The present study was conducted in collaboration with Dr. Bernd Ensing (Computational Chemistry, University of Amsterdam), Dr. Saer Samanipour (Analytical Chemistry, University of Amsterdam), Dr. Patrick Forré (Institute for Informatics, University of Amsterdam), as well as scientists from Gustavus Adolphus College.

Special acknowledgement to Peter Schoenmakers

In the article, the authors acknowledged Prof. Peter Schoenmakers for his founding contributions. In one of his first papers in 1978 on gradient selection for RPLC method development Schoenmakers already envisaged the use of scouting data to facilitate automated method development [6].

Schoenmakers was the promotor of Bob Pirok, who first published about this topic in his 2016 paper in which the theoretical possibility of leveraging these concepts for 2D-LC were investigated [7]. This study was marked the start of this research line that, ultimately, led to the present publication of AutoLC.

References

  1. Chemometric Strategies for Fully Automated Interpretive Method Development in Liquid Chromatography S. Bos, J. Boelrijk, S.R.A. Molenaar, B. van ‘t Veer, L.E. Niezen, D. van Herwerden, S. Samanipour, D.R. Stoll, P. Forré, B. Ensing, G.W. Somsen, B.W.J. Pirok, Anal. Chem. 2022, 94(46), 16060–16068, DOI: 10.1021/acs.analchem.2c03160.
  2. Peak-Tracking Algorithm for Use in Automated Interpretive Method-Development Tools in Liquid Chromatography, B.W.J. Pirok, S.R.A. Molenaar, L.S. Roca and P.J. Schoenmakers, Anal. Chem., 2018, 90(23), 14011-14019, DOI: 10.1021/acs.analchem.8b03929.
  3. Peak-tracking algorithm for use in comprehensive two-dimensional liquid chromatography – application to monoclonal antibody peptides, R.A. Molenaar, T.A. Dahlseid, G. Leme, D.R. Stoll, P.J. Schoenmakers, B.W.J. Pirok, J. Chromatogr. A, 2021, 1639, 461922, DOI: 10.1016/j.chroma.2021.461922.
  4. Bayesian Optimization of Comprehensive Two-dimensional Liquid Chromatography Separations, J. Boelrijk, B.W.J. Pirok, B. Ensing, P. Forré, Chromatogr. A, 1659, 2021, 462628, DOI: 10.1016/j.chroma.2021.462628.
  5. Reducing the influence of geometry-induced gradient deformation in liquid chromatographic retention modellingS. Bos, L.E. Niezen, M.J. den Uijl, S.R.A. Molenaar, S. Lege, P.J. Schoenmakers, G.W. Somsen, B.W.J. Pirok, J. Chromatogr. A, 2021, 1635, 461714, DOI: 10.1016/j.chroma.2020.461714.
  6. Gradient selection in reversed-phase liquid chromatography, P.J. Schoenmakers, H.A.H. Billiet, R. Tussen, L. De Galan, J. Chromatogr. A, 1978, 149,  519-537, DOI: 10.1016/S0021-9673(00)81008-0.
  7. Program for the interpretive optimization of two-dimensional resolution, B.W.J. Pirok, S. Pous-Torres, C. Ortiz-Bolsico, G. Vivó-Truyols and P.J. Schoenmakers, J. Chromatogr. A, 2016, 1450, 29–37, DOI: 10.1016/j.chroma.2016.04.061.
Categories
Publications

Reducing Effect of Gradient Deformation For LC Retention Modelling

Retention modelling is a useful technique which can be used to substantially reduce the method-development process for LC separations. One approach utilizes so-called scanning (or ‘scouting’) experiments using isocratic or gradient elution [1]. Here, a number of pre-defined methods are employed to record retention times to which empirical models are fitted. 

Isocratic experiments will generally yield reliable solid datasets that are very suitable for retention modelling. Using isocratic elution is, however, not always very practical. Indeed, scouting experiments can take rather long for the slower experiments. Moreover, some manual fine-tuning and experience with the analytes in question are needed to identify the appropriate modifier concentrations.

In contrast, gradient elution allows rather quick and easy scanning experiments at the significant cost of the usefulness of the resulting data. Where isocratic experiments directly measure the retention factor at a certain modifier (φ) fraction, the retention time in gradient elution depends on the gradient experienced by the analyte.

image_2021-01-02_083104

Figure 1. Schematic illustrating a programmed linear gradient and the experienced gradients for two different systems.

However, as the programmed change in composition produced by the pump migrates through the chromatographic system, its shape is altered. In Figure 1, above, we can see how this leads to the familiar difference between the programmed (dark blue) and effective (purple, light blue) gradients for two different systems.

This deviation is the product of an array of effects, such as the morphology and inefficiencies in the pump components, chromatographic system volumes and the accuracy of pumped mobile-phase composition (A vs. B). The latter can rather easy deviate if the pump does not take into account the change in density as φ increases.

Figure 2. Response functions of systems 1 and 2. The shape essentially represents the differences between the programmed and effective gradients shown in Figure 1.

The overall effect can be represented by response functions. These functions essentially describe the difference between the programmed and measured gradient. Two examples for two different systems are shown above in Figure 2. Indeed, depending on the pump characteristics, dramatic changes can be observed.

 

The problem is only complicated further as the recorded dwell curve may also in itself represent an inaccurate depiction. Depending on the detector, solvatochromic effects and  the presence of other mobile-phase components can severely convolute the true depicted of the experienced gradient.

For modelling, deformation is a problem because

  • Published retention parameters include the system effects, rendering the values useless to other users without detailed knowledge of their chromatographic system.
  • Optimized methods cannot be directly translated to other systems.
  • The usefulness of gradients to obtain scanning experiments is severely weakened.

As part of a larger collaboration with Agilent Technologies in the “DAS PRETSEL” project, Tijmen Bos, with assistance of other CAST members Mimi den Uijl, Leon Niezen and Stef Molenaar, developed an algorithm to reduce partially the effects of gradient deformation.

In their work, Bos et al. showed that the impact of the gradient deformation significantly impacts retention parameters. By modelling so-called Stable distribution functions to the measured dwell curves, the authors were able to significantly reduce the prediction errors for water-water systems (Figure 3). Conveniently, the Stable parameters turned out to be related to physical parameters of the chromatographic system.

Figure 3. Relative errors (%) in the predicted retention times of the test compounds on Instruments 2 (top) and 3 (bottom) obtained when using retention parameters determined for the test compounds on Instrument 1 at different flow rates. Please see the publication for details about the instruments. Reproduced with permission from [2].

This work is part of a larger project. In this first stage, we mainly targeted the geometric-influences. Now, we shift our focus to more complicated solvent systems and also the effect on larger molecular systems.

The work was recently published open-access in Journal of Chromatography A and can be downloaded for free here. An accompanying video pitch can be viewed below. Readers interested in learning more about retention modelling and its application areas are referred elsewhere. 

References

[1] Recent applications of retention modelling in liquid chromatography, M.J. den Uijl,  P.J. Schoenmakers,  B.W.J. Pirok, and  M.R. van Bommel, J. Sep. Sci., 2020, DOI: 10.1002/jssc.202000905.

[2] Reducing the influence of geometry-induced gradient deformation in liquid chromatographic retention modelling, T.S. Bos, L.E. Niezen, M.J. den Uijl, S.R.A. Molenaar, S. Lege, P.J. Schoenmakers, G.W. Somsen, B.W.J. Pirok, J. Chromatogr. A, 2021, 1635, 461714, DOI: 10.1016/j.chroma.2020.461714.

The Authors

Tijmen Bos

Mimi den Uijl

Leon Niezen

Stef Molenaar

Researchers Bos, Niezen and Molenaar were part of the UNMATCHED project, which was supported by BASF, DSM and Nouryon, and received funding from the Dutch Research Council (NWO). Den Uijl was part of the TooCOLD project, which was supported by Unilever and NWO. You can read more about them and find their contact info on the Team page.