Electronic Theses and Dissertation

Daun kelor (Moringa oleifera) merupakan tanaman tropis yang dikenal sebagai miracle tree karena kandungan nutrisi dan senyawa bioaktifnya yang tinggi, termasuk mineral magnesium (Mg) dan mangan (Mn). Metode analisis konvensional seperti Atomic Absorption Spectroscopy (AAS) memiliki akurasi tinggi, tetapi bersifat destruktif, memerlukan preparasi sampel yang rumit, dan menghasilkan limbah kimia. Sebagai alternatif, Spektroskopi Fourier Transform Infrared (FTIR) menawarkan analisis yang cepat, non-destruktif, dan tanpa preparasi sampel yang kompleks. Penelitian ini bertujuan menguji pengaruh pretreatment Multiplicative Scatter Correction (MSC) dan Standard Normal Variate (SNV) serta menilai performa Partial Least Square (PLS) Regression dalam memprediksi kadar magnesium dan mangan pada bubuk daun kelor.
Daun kelor (Moringa oleifera) diperoleh dari 12 pohon berbeda di Kota Banda Aceh. Daun dikeringkan menggunakan food dehydrator pada suhu 40°C selama 7 jam hingga kadar air mencapai 8,57%, kemudian dihaluskan dan diayak menggunakan ayakan 80 mesh. Penelitian diawali dengan pengumpulan sampel daun kelor dari 12 lokasi di Banda Aceh, Aceh Besar, dan sekitarnya yang kemudian diolah menjadi bubuk daun kelor. Kandungan magnesium dan mangan aktual ditentukan menggunakan AAS sebagai data referensi, sedangkan spektrum FTIR diukur menggunakan Spectrum Two FT-IR Spectrometer (PerkinElmer) pada rentang 4000–400 cm⁻¹. Seluruh spektrum diverifikasi menggunakan Principal Component Analysis (PCA) dan Hotelling’s T² Ellipse yang menunjukkan tidak terdapat outlier. Data kemudian diproses menggunakan pretreatment MSC dan SNV, lalu model PLS dengan 5 latent variable dievaluasi menggunakan Leave-One-Out Cross Validation (LOOCV) berdasarkan nilai r, R², RMSEC, dan RPD.
Hasil penelitian menunjukkan bahwa FTIR yang dikombinasikan dengan PLS mampu memprediksi kandungan magnesium dan mangan bubuk daun kelor dengan akurasi sangat baik. Untuk magnesium, pretreatment MSC memberikan performa terbaik dengan nilai r = 0,9776, R² = 0,9557, RMSEC = 31,8001 mg/kg, dan RPD = 4,9649. Untuk mangan, MSC juga memberikan performa terbaik dengan nilai r = 0,9500, R² = 0,9024, RMSEC = 8,8805 mg/kg, dan RPD = 3,3439, lebih baik dibandingkan tanpa pretreatment. Hasil ini menunjukkan bahwa MSC meningkatkan prediksi mangan dan menegaskan bahwa FTIR-PLS merupakan metode yang valid, cepat, dan non-destruktif untuk pengendalian mutu kandungan mineral pada produk berbasis daun kelor.

Moringa (Moringa oleifera) is a tropical plant widely recognized as the miracle tree due to its high nutritional value and abundance of bioactive compounds, including the essential minerals magnesium (Mg) and manganese (Mn). Conventional analytical methods, such as Atomic Absorption Spectroscopy (AAS), provide high accuracy but are destructive, require complex sample preparation, and generate chemical waste. As an alternative, Fourier Transform Infrared (FTIR) spectroscopy offers rapid, non-destructive analysis with minimal sample preparation. This study aimed to evaluate the effects of Multiplicative Scatter Correction (MSC) and Standard Normal Variate (SNV) pretreatments and to assess the performance of Partial Least Squares (PLS) regression in predicting magnesium and manganese contents in Moringa oleifera leaf powder. Moringa oleifera leaves were collected from 12 different trees in Banda Aceh, Indonesia. The leaves were dried using a food dehydrator at 40°C for 7 hours until the moisture content reached 8.57%, then ground into powder and sieved through an 80-mesh sieve. The study began with the collection of leaf samples from 12 locations in Banda Aceh, Aceh Besar, and surrounding areas, followed by processing into leaf powder. The actual magnesium and manganese contents were determined using AAS as the reference method, while FTIR spectra were acquired using a Spectrum Two FT-IR Spectrometer (PerkinElmer) within the spectral range of 4000–400 cm⁻¹. All spectra were verified using Principal Component Analysis (PCA) and Hotelling’s T² ellipse, confirming the absence of outliers. The spectral data were subsequently pretreated using MSC and SNV, and PLS regression models with five latent variables were evaluated using Leave-One-Out Cross-Validation (LOOCV) based on the correlation coefficient (r), coefficient of determination (R²), root mean square error of calibration (RMSEC), and residual predictive deviation (RPD). The results demonstrated that FTIR spectroscopy combined with PLS regression accurately predicted the magnesium and manganese contents of Moringa oleifera leaf powder. For magnesium, the MSC pretreatment produced the best performance, yielding r = 0.9776, R² = 0.9557, RMSEC = 31.8001 mg/kg, and RPD = 4.9649. For manganese, MSC also achieved the highest predictive performance, with r = 0.9500, R² = 0.9024, RMSEC = 8.8805 mg/kg, and RPD = 3.3439, outperforming both the SNV-pretreated and untreated models. These findings indicate that MSC enhances the prediction accuracy of manganese and confirm that the FTIR–PLS approach is a valid, rapid, and non-destructive method for quality control of mineral content in Moringa oleifera-based products.