AMS Ideas in Action: Screening Polymer Dielectric Loss with QSPR Descriptors

In this first AMS Ideas in Action post, we share an executable Python notebook inspired by a 2008 QSPR study on polymer dielectric dissipation by Yu and coworkers.

The notebook revisits the idea of connecting polymer structure to tangent-loss behavior using quantum-chemical descriptors and machine-learning models. With AMS Ideas in Action, we want to share applied, literature-inspired notebooks earlier and more often: practical starting points that can be explored, modified, extended, and discussed with SCM experts.

Background: Dielectric loss

Dielectric loss, often expressed as tan delta or tangent loss, is an important property for polymers used in electrical insulation, electronic packaging, interlayer dielectrics, printed wiring boards, and related applications.

It is also a challenging property to work with. Experimental values can depend strongly on frequency, temperature, and material details. For industrial materials research, this makes early-stage screening especially valuable: before investing in synthesis, processing, and measurement, can we use molecular simulation and data-driven models to help prioritize promising polymer candidates?

From Literature Idea to Executable Notebook

The notebook uses AMS/PLAMS and DFTB calculations to generate molecular descriptors for monomer and repeat-unit proxy structures. These descriptors are then used in a QSPR workflow for polymer tangent-loss data.

The notebook walks through:

• preparing the literature dataset and molecular proxy structures;
• running live DFTB descriptor calculations with AMS/PLAMS;
• assembling the descriptor table;
• comparing simple baseline and nonlinear models;
• exploring practical refinements such as logarithmic frequency scaling and additional cheminformatics descriptors;
• outlining possible next steps for a more rigorous study.

This is the kind of rapid prototyping workflow where notebooks are especially useful: fast to evaluate, transparent in each step, and highly customizable for different datasets, polymer families, descriptors, or modeling choices.

Why This Is Interesting for Polymer Screening

For applications where dielectric loss matters, even an early-stage computational workflow can be useful if it helps researchers ask better questions earlier:

• Which structural features may be worth investigating further?
• Can available experimental data be used to build an initial screening model?
• Which descriptors are fast enough for broad exploration?
• Where would higher-level DFT, COSMO-RS descriptors, conformer sampling, or more careful validation add value?
• What would be needed to adapt the workflow to a proprietary polymer dataset?

This is where AMS can be especially useful: not as a black-box replacement for experiments, but as part of an expert-guided workflow for generating molecular descriptors, testing hypotheses, and designing better follow-up studies.

Discuss Your Polymer Screening Challenge with SCM

Are you working on polymer dielectrics, electrical insulation, packaging materials, or structure-property screening?

We invite you to schedule a discussion with SCM experts. Together we can look at your target property, available data, chemistry space, and the level of simulation or validation that would be appropriate for your research question.

Contact us to discuss how AMS simulations and descriptor-based workflows could support your materials discovery project.

New to Python notebooks? See our short introduction to running AMS notebooks.