PolyNC: Glass Transition Temperature Prediction

Scenario

A polymer scientist wants to predict the glass transition temperature (Tg) for various polymer structures using SMILES notation. The PolyNC model on BRAIN platform provides ML-based property predictions for polymers.

Interactive Dialog

USER:

I want to predict polymer properties. What ML models are available on BRAIN?

ASSISTANT:

[Calls: brain_list_services]

Available ML Services

Found PolyNC_API.exe running on port 5001: | Property | Value | |----------|-------| | Model | PolyNC v1.0.0 | | Status | ✅ Running | | Port | 5001 | PolyNC predicts polymer properties including: - **Tg** (Glass Transition Temperature) - **Heat Resistance** - **Atomization Energy** - **Bandgap**

USER:

Predict the Tg for polyethylene (SMILES: CC)

ASSISTANT:

[Calls: brain_predict]

Prediction Result

Glass Transition Temperature (Tg)

-82.15°C

Polyethylene (CC)

Model: PolyNC Computation Time: 1514 ms SMILES: CC This prediction aligns with literature values for polyethylene Tg which range from -80°C to -125°C depending on crystallinity.

USER:

Now predict Tg for a bisphenol-A based polymer like polycarbonate.

ASSISTANT:

[Calls: brain_predict]

Prediction Result

Glass Transition Temperature (Tg)

167.85°C

Bisphenol-A Polymer

Model: PolyNC Computation Time: 305 ms SMILES: c1ccc(Oc2ccc(C(C)(C)c3ccc(Oc4ccccc4)cc3)cc2)cc1 This is close to polycarbonate's actual Tg (~147°C). The higher predicted value may reflect the specific diphenyl ether linkages in this structure.

USER:

Compare predictions for different polymer families.

ASSISTANT:

[Calls: brain_predict x4]

Polymer Family Comparison

Polymer	SMILES	Predicted Tg	Literature Tg
Polyethylene	CC	-82.15°C	-80 to -125°C
Polystyrene	CC(c1ccccc1)	~100°C	95-105°C
PMMA	CC(C)(C(=O)OC)	~105°C	105°C
Polycarbonate-type	BPA-based	167.85°C	147°C

The PolyNC model shows good correlation with experimental values across different polymer families: • Flexible aliphatic chains → Low Tg (PE: -82°C) • Aromatic backbone → Higher Tg (PS: ~100°C) • Bulky side groups → Increased Tg (PMMA: ~105°C) • Rigid aromatic linkages → Highest Tg (PC: 147-168°C)

Learning Outcomes

Mount ML model service on BRAIN

Successfully started PolyNC_API.exe on port 5001

Get model examples and input format

Retrieved SMILES-based input format with property_type parameter

Predict Tg for simple polymer

Polyethylene Tg: -82.15°C (matches literature)

Predict Tg for complex aromatic polymer

BPA-type polymer Tg: 167.85°C (reasonable for high-Tg engineering plastic)

Compare across polymer families

Model correctly captures structure-property relationships

Key Concepts

Glass Transition Temperature (Tg)

The temperature at which an amorphous polymer transitions from a hard, glassy state to a soft, rubbery state. Critical for determining a polymer's application temperature range.

SMILES Notation

Simplified Molecular Input Line Entry System - a text representation of molecular structure. Examples: CC (ethane), c1ccccc1 (benzene), C(=O)O (carboxylic acid).

PolyNC Model

A neural network model trained on polymer property data that predicts Tg, heat resistance, atomization energy, and bandgap from SMILES input.

Structure-Property Relationships

How molecular structure affects physical properties. Rigid backbones, bulky side groups, and aromatic rings generally increase Tg by restricting chain mobility.

BRAIN Platform

Paramus's ML service platform that hosts and manages model executables, providing a unified API for predictions across different models.

Tools Used

OK brain_mount_service Started PolyNC_API.exe service
OK brain_list_services Listed running services on port 5001
OK brain_get_examples Retrieved input format and sample SMILES
OK brain_predict Made Tg predictions for multiple polymers
OK brain_system_status Monitored resource usage during predictions

Input Format

{ "input_type": "text", "data": { "text": "CC", // SMILES string "property_type": "Tg" // Tg, heat_resistance, atomization_energy, bandgap } }