Development of a Portable Near Infrared Hadamard Photometer Based on Light Emitting Diodes for Diesel/Biodiesel Blends Analysis

Diesel blended with 7.0 ± 0.5% (v/v) biodiesel is the prevalent fuel commercialized in Brazil. These blends can be adulterated with vegetable oil. Therefore, it is important to monitor the quality of diesel blends according to biodiesel content or vegetable oil adulteration.
Literature reports use of benchtop FT-NIR spectrophotometers to analyze quality parameters of diesel/biodiesel blends. These instruments offer high spectral resolution, signal-to-noise ratio and better wavelength reproducibility. However, they are expensive and less suitable for field analysis.
The present work reports on the development of a portable, low cost NIR photometer, based on Light Emitting Diodes (LEDs), interference filters and Hadamard multiplexing, for quantifying biodiesel and vegetable oil in diesel/biodiesel blends.
Initially, it was established that the instrument would be based on a Hadamard matrix with seven elements. Therefore, a preliminary study was performed to select seven sets of LED/Filters on the NIR region (1100-1580 nm). For this, a data set with 132 NIR spectra of diesel blends with both biodiesel and vegetable oil contents varying from 0.0 to 10% v/v acquired using a FT-NIR spectrophotometer (FTLA 2000-160 Bomem), with a quartz cell of 20 mm optical path length, was used. In order to simulate the spectral resolution obtained with the interference filters, which have a half-width band of approximately 20 nm, these spectra were mathematically processed, resulting in a data set of 24 variables. With this data set, MLR regression models using spectral variables selected by Successive Projections Algorithm (SPA) were developed to determine both biodiesel and vegetable oil in the blends. From these results, the seven best variables for quantifying both biodiesel and vegetable oil were assigned to the interference filters with maximum of transmittance at: 1240, 1260, 1320, 1340, 1360, 1440, 1480 nm. Radiation sources were provided by three different LEDs, with maximum of emission at 1200, 1300 and 1450 nm. The radiation from the sets of LEDs/filters were collimated by a single biconvex lens 50 mm in diameter onto the sample cell of 20 mm optical path length and then focused on an InGaAs detector employing a spherical lens. A microcontroller (Arduino MEGA 2560) was used for pulsing the LEDs, according to the lines of a Hadamard matrix, to obtain the multiplexed signals, for signal acquisition and data storage. With this arrangement, a portable dedicated photometer was built without any moving parts.
The performance of the instrument was evaluated using 118 additional samples varying both biodiesel and vegetable oil content from 0.0 to 15.0% v/v. MLR regression models were developed for quantifying both biodiesel and vegetable oil and the RMSEP of 1.96% v/v and 1.87% v/v, respectively, were obtained. Although these results are worse than those described in the literature for benchtop spectrophotometers, they showed that the developed instrument is able to quantify biodiesel and vegetable oil in diesel blends, even with the high similarities among their spectra (differences in absorbance intensities around 20 mAbs). The instrument set-up should be further improved in order to increase its analytical sensitivity.