RGRonaldo Giro

Ronaldo Giro

*This is an algorithmically generated author record based on publicly available information.

Ronaldo Giro has a total of 53 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    HTA - An open-source software for assigning head and tail positions to monomer SMILES in polymerization reactions

    by Brenda de Souza Ferrari, Ronaldo Giro, Mathias B. Steiner

    Artificial Intelligence (AI) techniques are transforming the computational discovery and design of polymers. The key enablers for polymer informatics are machine-readable molecular string representations of the building blocks of a polymer, i.e., the monomers. In monomer strings, such as SMILES, symbols at the head and tail atoms indicate the locations of bond formation during polymerization. Since the linking of monomers determines a polymer’s properties, the performance of AI prediction models will, ultimately, be limited by the accuracy of the head and tail assignments in the monomer SMILES. Considering the large number of polymer precursors available in chemical data bases, reliable methods for the automated assignment of head and tail atoms are needed. Here, we report a method for assigning head and tail atoms in monomer SMILES by analyzing the reactivity of their functional groups based on the atomic index of nucleophilicity. In a reference data set containing 206 polymer precursors, the HeadTailAssign (HTA) algorithm correctly predicted the polymer class of 204 monomer SMILES, achieving an accuracy of 99%. The head and tail atoms were correctly assigned to 187 monomer SMILES, representing an accuracy of 91%. The HTA code is available for validation and reuse at https://github.com/IBM/HeadTailAssign . Scientific contribution: The algorithm was successfully applied to data pre-processing by tagging the linkage bonds in monomers for defining the repeat units in polymerization reactions.


  • Open Access

    HTA - An open-source software for assigning heads and tails to SMILES in polymerization reactions

    by Brenda de Souza Ferrari, Ronaldo Giro, Mathias Steiner

    Artificial Intelligence (AI) techniques are transforming the computational discovery and design of polymers. The key enablers for polymer informatics are machine-readable molecular string representations of the building blocks of a polymer, i.e., the monomers. In monomer strings, such as SMILES, symbols at the head and tail atoms indicate the locations of bond formation during polymerization. Since the linking of monomers determines a polymer’s properties, the performance of AI prediction models will, ultimately, be limited by the accuracy of the head and tail assignments in the monomer SMILES. Considering the large number of polymer precursors available in chemical data bases, reliable methods for the automated assignment of head and tail atoms are needed. Here, we report a method for assigning head and tail atoms in monomer SMILES by analyzing the reactivity of their functional groups. In a reference data set containing 206 polymer precursors, the HeadTailAssign (HTA) algorithm has correctly predicted the polymer class of 204 monomer SMILES, representing an accuracy of 99%. The head and tail atoms were correctly assigned to 187 monomer SMILES, representing an accuracy of 91%. The HTA code is available for validation and reuse at https://github.com/IBM/HeadTailAssign

  • Open Access

    Emergence of the Robeson bound in non-equilibrium molecular dynamics simulations

    by Tianmu Yuan, Ronaldo Giro, Mathias B. Steiner, Hsianghan Hsu, Lev Sarkisov

    Theoretical understanding of multicomponent transport processes in polymer membranes at the molecular level remains a significant scientific challenge due to the high computational cost and poor reproducibility of non-equilibrium molecular simulation methods, aiming at replicating experimental conditions and gradient-driven processes. In this study, we employ Dual Control Volume Molecular Dynamics (DCVMD) with fully atomistic, microsecond-scale simulations to elucidate gas separation mechanisms in porous polymer membranes. Using PIM-PI-8 as a case study, we develop a comprehensive workflow that includes sensitivity analysis and calculations of key transport and structural characteristics of the model membranes. This approach allows us to thoroughly characterize the uncertainties inherent in non-equilibrium simulations and explore the challenges associated with making quantitative predictions using this method. Expanding our analysis to a family of PIM polymers, our simulations successfully replicate several experimentally observed trends. These include a linear correlation between permeability and fractional free volume, as well as the relationship between perm-selectivity and sorption selectivity. Most importantly, we demonstrate that trade-offs between perm-selectivity and permeability emerge in non-equilibrium molecular simulations, qualitatively capturing one of the key patterns in experiments, known as the Robeson upper bound.

  • Open Access

    Automated, LLM enabled extraction of synthesis details for reticular materials from scientific literature

    by Viviane Torres da Silva, Alexandre Rademaker, Krystelle Lionti, Ronaldo Giro, Geisa Lima, Sandro Fiorini, Marcelo Archanjo, Breno W. Carvalho, Rodrigo Neumann, Anaximandro Souza, et al.

    Automated knowledge extraction from scientific literature can potentially accelerate materials discovery. We have investigated an approach for extracting synthesis protocols for reticular materials from scientific literature using large language models (LLMs). To that end, we introduce a Knowledge Extraction Pipeline (KEP) that automatizes LLM-assisted paragraph classification and information extraction. By applying prompt engineering with in-context learning (ICL) to a set of open-source LLMs, we demonstrate that LLMs can retrieve chemical information from PDF documents, without the need for fine-tuning or training and at a reduced risk of hallucination. By comparing the performance of five open-source families of LLMs in both paragraph classification and information extraction tasks, we observe excellent model performance even if only few example paragraphs are included in the ICL prompts. The results show the potential of the KEP approach for reducing human annotations and data curation efforts in automated scientific knowledge extraction.


  • Open Access

    Predicting polymerization reactions via transfer learning using chemical language models

    by Brenda S. Ferrari, Matteo Manica, Ronaldo Giro, Teodoro Laino, Mathias B. Steiner

    Polymers are candidate materials for a wide range of sustainability applications such as carbon capture and energy storage. However, computational polymer discovery lacks automated analysis of reaction pathways and stability assessment through retro-synthesis. Here, we report an extension of transformer-based language models to polymerization for both reaction and retrosynthesis tasks. To that end, we have curated a polymerization dataset for vinyl polymers covering reactions and retrosynthesis for representative homo-polymers and co-polymers. Overall, we obtain a forward model Top-4 accuracy of 80% and a backward model Top-4 accuracy of 60%. We further analyze the model performance with representative polymerization examples and evaluate its prediction quality from a materials science perspective. To enable validation and reuse, we have made our models and data available in public repositories.


  • Open Access

    Predicting polymerization reactions via transfer learning using chemical language models

    by Ronaldo Giro, Brenda Ferrari, Matteo Manica, Teodoro Laino, Mathias Steiner

    Polymers are candidate materials for a wide range of sustainability applications such as carbon capture and energy storage. However, computational polymer discovery lacks automated analysis of reaction pathways and stability assessment through retro-synthesis. Here, we report the first extension of transformer-based language models to polymerization reactions for both forward and retrosynthesis tasks. To that end, we have curated a polymerization dataset for vinyl polymers covering reactions and retrosynthesis for representative homo-polymers and co-polymers. Overall, we obtain a forward model Top-4 accuracy of 80\% and a backward model Top-4 accuracy of 60\%. We further analyze the model performance with representative polymerization and retro-synthesis examples and evaluate its prediction quality from a materials science perspective.

  • Open Access

    Emission Redshift in DCM2-Doped Alq$_{3}$ Caused by Non-Linear Stark Shifts and Förster-Mediated Exciton Diffusion

    by Graziâni Candiotto, Ronaldo Giro, Bruno A. C. Horta, Flávia P. Rosselli, Marcelo de Cicco, Carlos A. Achete, Marco Cremona, Rodrigo B. Capaz

    Organic light-emitting diodes (OLEDs) devices in the archetype small molecule fluorescent guest-host system tris(8-hydroxyquinolinato) aluminum (Alq$_{3}$) doped with 4-(dicyanomethylene)-2-methyl-6-julolidyl-9-enyl-4H-pyran (DCM2) displays a redshift in light-emission frequency which is extremely sensitive to the dopant concentration. This effect can be used to tune the emission frequency in this particular class of OLEDs. In this work, a model is proposed to describe this effect using a combination of density functional theory (DFT) quantum-chemical calculations and stochastic simulations of exciton diffusion via a Förster mechanism. The results show that the permanent dipole moments of the Alq$_{3}$ molecules generate random electric fields that are large enough to cause a non-linear Stark shift in the band gap of neighboring DCM2 molecules. As a consequence of these non-linear shifts, a non-Gaussian probability distribution of highest-occupied molecular orbital to lowest-unoccupied molecular orbital (HOMO-LUMO) gaps for the DCM2 molecules in the Alq$_{3}$ matrix is observed, with long exponential tails to the low-energy side. Surprisingly, this probability distribution of DCM2 HOMO-LUMO gaps is virtually independent of DCM2 concentration into Alq$_{3}$ matrix, at least up to a fraction of 10%. This study shows that this distribution of gaps, combined with out-of-equilibrium exciton diffusion among DCM2 molecules, are sufficient to explain the experimentally observed emission redshift.

  • Open Access

    AI powered, automated discovery of polymer membranes for carbon capture

    by Ronaldo Giro, Hsianghan Hsu, Akihiro Kishimoto, Toshiyuki Hama, Rodrigo F. Neumann, Binquan Luan, Seiji Takeda, Lisa Hamada, Mathias B. Steiner

    The generation of molecules with artificial intelligence (AI) or, more specifically, machine learning (ML), is poised to revolutionize materials discovery. Potential applications range from development of potent drugs to efficient carbon capture and separation technologies. However, existing computational discovery frameworks for polymer membranes lack automated training data creation, generative design, and physical performance validation at meso-scale where complex properties of amorphous materials emerge. The methodological gaps are less relevant to the ML design of individual molecules such as the monomers which constitute the building blocks of polymers. Here, we report automated discovery of complex materials through inverse molecular design which is informed by meso-scale target features and process figures-of-merit. We have explored the multi-scale discovery regime by computationally generating and validating hundreds of polymer candidates designed for application in post-combustion carbon dioxide filtration. Specifically, we have validated each discovery step, from training dataset creation, via graph-based generative design of optimized monomer units, to molecular dynamics simulation of gas permeation through the polymer membranes. For the latter, we have devised a representative elementary volume (REV) enabling permeability simulations at about 1000× the volume of an individual, ML-generated monomer, obtaining quantitative agreement. The discovery-to-validation time per polymer candidate is on the order of 100 h using one CPU and one GPU, offering a computational screening alternative prior to lab validation.


  • Emission redshift in DCM2-doped Alq3 caused by nonlinear Stark shifts and Förster-mediated exciton diffusion

    by Graziâni Candiotto, Ronaldo Giro, Bruno A. C. Horta, Flávia P. Rosselli, Marcelo de Cicco, Carlos A. Achete, Marco Cremona, Rodrigo B. Capaz

    Organic light-emitting diode (OLED) devices in the archetype small-molecule fluorescent guest-host system tris(8-hydroxyquinolinato) aluminum (Alq3) doped with 4-(dicyanomethylene)-2-methyl-6-julolidyl-9-enyl-4H-pyran (DCM2) displays a redshift in light-emission frequency which is extremely sensitive to the dopant concentration. This effect can be used to tune the emission frequency in this particular class of OLEDs. In this work, a model is proposed to describe this effect using a combination of density functional theory quantum-chemical calculations and stochastic simulations of exciton diffusion via a Förster mechanism. The results show that the permanent dipole moments of the Alq3 molecules generate random electric fields that are large enough to cause a nonlinear Stark shift in the band gap of neighboring DCM2 molecules. As a consequence of these nonlinear shifts, a non-Gaussian probability distribution of highest occupied molecular orbital to lowest unoccupied molecular orbital (HOMO-LUMO) gaps for the DCM2 molecules in the Alq3 matrix is observed, with long exponential tails to the low-energy side. Surprisingly, this probability distribution of DCM2 HOMO-LUMO gaps is virtually independent of DCM2 concentration into the Alq3 matrix, at least up to a fraction of 10%. This study shows that this distribution of gaps, combined with out-of-equilibrium exciton diffusion among DCM2 molecules, is sufficient to explain the experimentally observed emission redshift.


  • Open Access

    Adsorption energy as a metric for wettability at the nanoscale

    by Ronaldo Giro, Peter W. Bryant, Michael Engel, Rodrigo F. Neumann, Mathias B. Steiner

    Wettability is the affinity of a liquid for a solid surface. For energetic reasons, macroscopic drops of liquid form nearly spherical caps. The degree of wettability is then captured by the contact angle where the liquid-vapor interface meets the solid-liquid interface. As droplet volumes shrink to the scale of attoliters, however, surface interactions become significant, and droplets assume distorted shapes. In this regime, the contact angle becomes ambiguous, and a scalable metric for quantifying wettability is needed, especially given the emergence of technologies exploiting liquid-solid interactions at the nanoscale. Here we combine nanoscale experiments with molecular-level simulation to study the breakdown of spherical droplet shapes at small length scales. We demonstrate how measured droplet topographies increasingly reveal non-spherical features as volumes shrink. Ultimately, the nanoscale droplets flatten out to form layer-like molecular assemblies at the solid surface. For the lack of an identifiable contact angle at small scales, we introduce a droplet’s adsorption energy density as a new metric for a liquid’s affinity for a surface. We discover that extrapolating the macroscopic idealization of a drop to the nanoscale, though it does not geometrically resemble a realistic droplet, can nonetheless recover its adsorption energy if line tension is included.