Bitumen Refractive Index and ATR-FTIR baseline correction (script)

ATR-FTIR Baseline Correction Pipeline for Bituminous Materials

Overview

This repository contains a complete processing pipeline for ATR-FTIR spectra of bituminous materials, with a focus on:

• Physics-based ATR correction
• Baseline correction
• Spectral normalization
• Quantitative peak integration
• Comparative visualization

The correction approach is based on refractive index–dependent ATR penetration depth and characteristic aliphatic band vibrations, allowing improved comparison between spectra and more physically meaningful interpretation of band areas.

The pipeline is designed to process raw FTIR spectra in a predefined CSV format and automatically generate corrected datasets and publication-ready plots.

Scientific Background

ATR-FTIR measurements of bituminous materials are strongly influenced by:

• Wavenumber-dependent penetration depth
• Refractive index mismatch
• Baseline distortions
• Band intensity scaling effects

This script applies:

1. ATR correction based on refractive index assumptions
2. Baseline correction
3. Spectral normalization
4. Quantitative integration of key functional group regions

Special attention is given to aliphatic CH stretching and rocking vibrations, which are used as internal references in the correction workflow.

Pipeline Structure

The main pipeline performs the following steps:

1. Extraction and Normalization

• Reads raw CSV spectra
• Averages replicates (if present)
• Stores extracted spectra
• Generates normalized spectra

2. ATR + Baseline Correction

• Applies ATR correction based on penetration depth
• Applies baseline correction
• Generates corrected and normalized corrected spectra

3. Spectral Parameter Extraction

Integrates characteristic peak regions:

Areas are extracted for:

• Raw spectra
• Raw ATR-corrected spectra
• Normalized raw spectra
• Normalized ATR-corrected spectra

4. Visualization

The script automatically generates:

• Raw vs corrected spectra plots
• Difference spectra
• Aliphatic rocking vs stretching comparison plots
• Bar plots of integrated peak areas (dual y-axis comparison)
• Fitted ATR parameter visualization

All figures are saved to disk for further analysis.

Input Requirements

Raw Data Format

The script expects raw spectra as CSV files with:

• Column containing wavenumber (cm⁻¹)
• Column(s) containing absorbance values
• Wavenumber axis in descending or ascending order (consistent format)
• Uniform spacing across spectra

Improper formatting will prevent correct parsing.

It is recommended that:

• All spectra share the same spectral range
• Units are in cm⁻¹ and absorbance

Output

The pipeline generates:

• Extracted spectra JSON files
• Normalized spectra JSON files
• ATR-corrected spectra JSON files
• Integrated spectral parameter JSON file
• PNG figures for:
  • Spectra comparison
  • Difference spectra
  • Peak area barplots
  • ATR fitting diagnostics

Goal of the Project

The purpose of this project is to provide a reproducible and physically informed correction workflow for ATR-FTIR spectra of bituminous materials, enabling:

• Reliable comparison between samples
• Reduction of ATR-induced distortions
• Quantitative analysis of ageing and oxidation markers
• Improved interpretation of aliphatic band behaviour

The pipeline is intended for research applications in:

• Bitumen ageing studies
• Oxidation monitoring
• Chemical fingerprint analysis
• Material comparison studies

Running the Pipeline

python main.py --raw data/raw --extracted data/extracted --corrected data/corrected --plots data/spectra_plots

Default directories are provided but can be modified via CLI arguments.

Notes

• The ATR correction relies on assumed refractive index behaviour.
• Results should be interpreted within the limits of the physical model.
• The script is designed specifically for bituminous materials and may require adaptation for other material systems.
• The zipped folder "data.zip", should be extracted before the script is run and it contains a small series of exemplary data