Machine-learning-assisted dispersive micro-solid-phase extraction of heavy metals in cosmetics using nickel chromite-functionalized activated nanodiamonds prior to microwave plasma-atomic emission spectrometry


Shirani M., AHMED H. E. H., Jahromi M. S. B., Teimouri M., SOYLAK M.

Microchemical Journal, cilt.231, 2026 (SCI-Expanded, Scopus)

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 231
  • Basım Tarihi: 2026
  • Doi Numarası: 10.1016/j.microc.2026.120077
  • Dergi Adı: Microchemical Journal
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, BIOSIS, Chemical Abstracts Core, Chimica, Index Islamicus, Academic Search Ultimate (EBSCO), Engineering Source (EBSCO)
  • Anahtar Kelimeler: Carboxylated nanodiamonds (ND-COOH), Cosmetics, Dispersive μ-solid-phase extraction (D-μ-SPE), Green metrics evaluation, Heavy metals
  • Erciyes Üniversitesi Adresli: Evet

Özet

In this study, nickel chromite-modified activated nanodiamonds (NiCr₂O₄@ND-COOH), as an advanced category of carbon-based nanomaterials, were synthesized and used for dispersive micro-solid-phase extraction of trace/ultra-trace levels of cadmium, chromium, lead, manganese, nickel, and cobalt ions prior to flow injection analysis coupled with microwave plasma atomic emission spectrometry. The accuracy of the prepared NiCr₂O₄@ND-COOH structure was assessed by techniques such as Fourier-transform infrared spectroscopy, field emission scanning electron microscopy, energy dispersive X-ray spectroscopy, and X-ray diffraction. The linear calibration curves were obtained via a matrix-matched method in which a certified amount of each analyte was spiked into each matrix. The linear ranges are 0.02–500 μg kg−1 for lipstick, mascara, eye shadow, lip liner, and sunscreen and 0.02–750 μg kg−1 for deodorant and body lotion for the target analytes. The limit of detection values of 0.005–0.007 μg kg−1 and the limit of quantification values of 0.018–0.022 μg kg−1 were obtained for the analyte in real samples. The intra-day precision and the inter-day precision were ≤ 2.2% and ≤ 2.8%, respectively. The applicability of the method was studied in cosmetic samples, which represented recoveries of 89.00–99.91%. Moreover, the consideration and evaluation of the proposed method with green metric tools of AGREE, AGREEprep, BAGI, CaFRI, VIGI, and CACI corroborated the greenness of the method based on the green analytical chemistry attitudes. Machine-learning models based on artificial neural network, evolutionary polynomial regression (EPR), and k-nearest neighbors were subsequently employed to model and predict the experimentally observed extraction efficiencies across the investigated operating variables. The three models showed generally good agreement with the experimental responses; however, their relative performance depended on the target metal and operating parameter. Given that the experimental data were generated using a one-factor-at-a-time strategy, the machine-learning models were used for prediction and identification of favorable operating conditions within the experimentally investigated domain rather than for independently validated multidimensional process optimization.