Least Absolute Shrinkage and Selection Operator as a Multivariate Calibration Tool for Simultaneous Determination of Diphenylamine and its Nitro Derivatives in Propellants. Issue 4 (2nd March 2018)

• Record Type: Journal Article
• Authors: Mani‐Varnosfaderani, Ahmad; Soleimani, Masoud; Alizadeh, Naader
• Is Part Of: Propellants, explosives, pyrotechnics. Volume 43:Issue 4(2018)
• Page Start: 379
• Page End: 389
• Publication Date: 2018-03-02
• Subjects: Diphenylamine -- multivariate calibration -- least absolute shrinkage and selection operator -- partial least square -- UV-Vis; Propellants -- Periodicals; Explosives -- Periodicals; 662.2
• DOI: 10.1002/prep.201700250
• Languages: English
• ISSNs: 0721-3115

Abstract

Abstract: Determination of diphenylamine (DPA) and its nitro derivatives received great attention for storing, deposition and on‐time usage of propellants. Herein, we present a novel and simple method for simultaneous determination of DPA and its nitro derivatives in solid propellants using UV‐Vis spectroscopy and chemometrics techniques. The UV‐Vis spectra of the analytes revealed strong overlap and it was difficult to determine them individually in their mixture without any separation and purification. To tackle the overlapping problem in collected spectra, analysis of first‐order UV‐Vis data was performed using multivariate calibration techniques. In this way, principle component regression (PCR), different modes of partial least square (PLS) and least absolute shrinkage and selection operator (LASSO) have been used for correlating the collected spectra to the concentration of DPAs in synthetic and real samples. The important variables were selected by confining the L1 ‐norm of the regression coefficients in multivariate model via the shrinkage and selection operator in LASSO approach. The results obtained by LASSO regression technique in this work were superior to those obtained by different modes of PLS algorithm. Moreover, it is shown that LASSO can be used as a reliable variable selection and modeling technique in multivariate calibration studies. Generally, the proposed strategy in this work is simple, non‐destructive, low‐cost and rapid and can be effectively applied for simultaneous determination of DPA and its nitro derivatives in solid propellants. Abstract :