FT-IR Spectroscopy to Accurately Identify Native Commercial Wood Species

According to a recent study published in the Royal Society of Chemistry journal, Fourier transform-infrared (FT-IR) spectroscopy is the technique of choice to classify native commercial wood species (1). The findings demonstrated that the technique’s effectiveness stems from its ability to perform chemical analysis on very small sample sizes in a short amount of time.

Spectroscopic techniques have been routinely used to classify wood species. For example, laser-induced breakdown spectroscopy (LIBS) has demonstrated its utility for analyzing the chemical composition of wood (2). Understanding the chemical composition of wood is important for the global economy, because it is a resource exported and imported globally (1). In this study, a multidisciplinary team of researchers from Brazil examined five wood species native to their country, rapidly and accurately identifying native commercial wood species using FT-IR spectroscopy combined with machine learning algorithms (1).

Atlantic Forest Brazil | Image Credit: © josefurlan_pissol - stock.adobe.com

The research focused on five Brazilian wood species that are commonly used in commercial applications: Angelim-pedra (Hymenolobium petraeum Ducke), Cambara (Gochnatia polymorpha), Cedrinho (Erisma uncinatum), Champagne (Dipteryx odorata), and Peroba do Norte (Goupia glabra Aubl) (1). Although these species are integral to industries such as construction and furniture production, their physical similarities can often lead to misidentification.

According to the study, traditional methods of wood classification can take up to four days to complete, a significant delay for industries that rely on fast, accurate species identification (1). Furthermore, these methods are highly dependent on the subjective assessment of human analysts, which increases the risk of error (1). FT-IR spectroscopy is designed to accelerate the classification process. The FT-IR approach offers a faster, more objective solution, providing results in as little as 15 minutes once samples are prepared (1).

FT-IR spectroscopy measures the IR spectrum of absorption, emission, or reflection of a material, providing a molecular fingerprint that can be used to identify specific compounds (1). In the case of wood, FT-IR detects the unique chemical signatures of its components, including cellulose, lignin, hemicellulose, and smaller quantities of lipids, phenolic compounds, and waxes (1).

To enhance the accuracy of FT-IR analysis, the team incorporated machine learning (ML) algorithms. Doing so allowed them to analyze a large data set and classify the wood samples with exceptional precision (1). By using multivariate analysis and machine learning models like linear support vector machines (SVM) and linear discriminant analysis (LDA), the researchers were able to accurately differentiate between the five wood species (1).

The study's results showed that the linear SVM algorithm was able to classify the wood species with an accuracy rate of 98%, whereas the LDA classifier achieved 100% accuracy during internal validation and 98% in external validation (1). These findings highlight the effectiveness of the FT-IR and ML combination in identifying wood species, even those with only subtle differences in their IR spectra.

In terms of sample preparation, the researchers obtained 52 heartwood sawdust samples from 26 different batches for each species (1). These samples were then sifted, dried, and analyzed using a PerkinElmer FT-IR spectrophotometer. The entire analysis process, from sample preparation to result generation, took only a few days, significantly reducing the time required for wood classification.

In addition to its practical applications, using FT-IR spectroscopy with ML could play a crucial role in promoting sustainable practices within the wood industry. Numerous studies, including this one, have demonstrated the technique’s potential in this space (1,3–4). By ensuring that wood species are accurately identified, the system could help prevent illegal logging and trade of protected species, supporting global efforts to protect biodiversity and promote sustainable forestry practices.

References

1. Jesus, E.; Franca, T.; Calvani, C.; et al. Making Wood Inspection Easier: FTIR Spectroscopy and Machine Learning for Brazilian Native Commercial Wood Species Identification. RSC Adv. 2024, 14, 7283–7289. DOI: 10.1039/D4RA00174E
2. Spectroscopy Staff, Handheld Laser-induced Breakdown Spectroscopy Could Improve Identification of Dalbergia spp. Spectroscopy. Available at: https://www.spectroscopyonline.com/view/handheld-laser-induced-breakdown-spectroscopy-could-improve-identification-of-dalbergia-spp (accessed 2024-09-30).
3. Wetzel, W. Examining the Role of ATR-FT-IR Spectroscopy and Machine Learning in Wood Forensics, Part 1. Spectroscopy. Available at: https://www.spectroscopyonline.com/view/examining-the-role-of-atr-ft-ir-spectroscopy-and-machine-learning-in-wood-forensics-part-1 (accessed 2024-10-01).
4. Wetzel, W. Examining the Role of ATR-FT-IR Spectroscopy and Machine Learning in Wood Forensics, Part 2. Spectroscopy. Available at: https://www.spectroscopyonline.com/view/examining-the-role-of-atr-ft-ir-spectroscopy-and-machine-learning-in-wood-forensics-part-2 (accessed 2024-10-01).