# Infrared spectral data of natural and man-made textile fibres for material identification and classification

## TL;DR

This paper introduces a dataset of infrared spectra from natural and man-made textile fibers to aid in material identification and classification for forensic and environmental research.

## Contribution

The paper provides a structured, openly accessible dataset of ATR-FTIR spectra from verified textile samples with multiple processing levels for analytical use.

## Key findings

- The dataset includes 160 spectra from 137 verified textile samples, covering both natural and man-made fibers.
- Multiple processing levels are provided, including raw data and pre-processed spectra for machine learning workflows.
- The dataset supports material identification, spectral comparison, and algorithm benchmarking in forensic and environmental science.

## Abstract

This article presents a structured dataset of attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectra acquired from natural and man-made textile fibres, compiled to support research in forensic, analytical, and environmental science. The collection comprises 160 spectra obtained from 137 verified textile samples originating from multiple sources, including industry reference collections and academic textile archives. The dataset includes several processing levels: (1) raw instrument files in PerkinElmer .sp format; and (2–5) tabular feature matrices containing unprocessed transmission spectra, baseline-corrected spectra, averaged baseline-corrected spectra grouped by fibre subtype, and fully pre-processed spectra prepared for chemometric and machine-learning workflows. All spectra were recorded using a Frontier FT-IR spectrometer by PerkinElmer equipped with a single-reflection diamond ATR accessory. Measurements were performed over the 4000–550 cm⁻¹ range at 4 cm⁻¹ resolution with four co-added scans per acquisition. Metadata describing sample identity and fibre characteristics are supplied. The resulting collection provides an openly accessible resource for material identification, spectral comparison, and the development or benchmarking of classification algorithms.