MODEL PENDUGAAN KANDUNGAN AIR, LEMAK DAN ASAM LEMAK BEBAS PADA TIGA PROVENAN BIJI JARAK PAGAR (Jatropha curcas L.) MENGGUNAKAN SPEKTROSKOPI INFRAMERAH DEKAT DENGAN METODE PARTIAL LEAST SQUARE (PLS)

Abstract

ABSTRAK Jarak pagar berpotensi sebagai sumber biodiesel karena kandungan lemak yang tinggi (>40%) dan belum ada penggunaan lainnya. Spektroskopi (Near Infrared) NIR adalah metode yang cepat untuk mengukur spektrum sampel dan tidak terdapat limbah kimia. Tujuan penelitian adalah mengembangkan metode pendugaan komposisi kimia beberapa  provenan  jarak  pagar  berdasarkan  spektroskopi  NIR menggunakan kalibrasi PLS. Pengujian dilakukan menggunakan tiga provenan jarak pagar yaitu IP-3A, IP-3M, dan IP-3P masing-masing 85 sampel. Spektrum reflektansi diukur menggunakan alat NIRFlex Solids Petri pada panjang gelombang 1000–2500 nm. Sekitar ⅔ jumlah sampel digunakan untuk mengembangkan persamaan kalibrasi dan ⅓ jumlah sampel untuk validasi. Pra perlakuan data spektrum dilakukan dengan normalisasi antara 0-1, turunan pertama Savitzky-Golay 9 titik dan gabungan keduanya. Hasil penelitian menunjukkan spektroskopi NIR dapat menduga kadar air, lemak, dan asam lemak bebas . Koefisien korelasi (r) antara komponen kimia metode acuan dengan dugaan NIR 0,83 menunjukkan ketepatan model cukup baik (r kadar air=0,96, r kadar lemak=0,92, dan r ALB=0,89 ). Konsistensi model kalibrasi kadar air=94,85%, lemak=82,56%, dan ALB=87,80%. Koefisien keragaman dugaan (Prediction Coeficient Variability/PCV) ketiga model <10% menunjukkan model yang dibangun cukup handal. Ratio of standard error prediction to deviation (RPD) menunjukkan metode spektroskopi NIR dapat digunakan untuk menentukan kadar air (RPD=3,30) dan lemak (RPD=2,06). Model-model yang dikembangkan secara umum layak untuk menentukan kadar air dan lemak biji jarak pagar, tetapi belum optimal untuk penentuan kadar ALB biji jarak pagar. Kata kunci: NIR , jarak pagar, kadar air, kadar lemak, kadar asam lemak bebas ABSTRACT Physic nut is a potential source of biodiesel. It is high in fat content, above 40% and has not been usesed for other purposes. Moisture, free fatty acid, and fat content are the chemical compounds and determinant factor for physic nut seed quality. The objective of this study was to develop a method to predict chemical composition of physic nut by NIR spectroscopy and PLS calibration. The study was conducted using three provenances of physic nut, i.e. IP-3A, IP-3M, and IP-3P, with 85 samples each. The wavelengths of near infrared reflectance ranged from 1000 to 2500 nm, and measured by NIR Flex Solids Petri Apparatus. Approximately ⅔ of total samples were used for developing calibration equation, while ⅓ of total samples for performing validation. Pre-treatment of spectrum data was done by applying normalization, first derivative of Savitzky–Golay 9 points, and as well as their combination. The results showed that NIR spectroscopy performed acceptable prediction for moisture and fat content. Correlation coefficients (r) between the reference method and NIR prediction were 0.96 for moisture content, 0.92 for fat content, and 0.89 for FFA and the consistency of the model were 94.85% for moisture content, 82.56% for fat, and 87.80% for FFA. Prediction of coefficient of variability (PCV) of the three models ≤10 % shows that the models are reliable. Ratio of standard error prediction to deviation (RPD) for moisture content has the potential to be used for screening (RPD=3.30) though the fat content model has rough screening (RPD=2.06). Key words: NIR, physic nut, moisture, fat, free fatty acid contents.