| Reports of excessive use of pesticides and chemical in agricultural products in recent years are frequently published. Cases of excessive use of pesticides in agro-products cause environment and health poisoning. Pesticide residue in fruits and vegetables causes negative impact to international agro-business. Although numbers of analytical technologies exist to detect pesticide residue in agro-product, such traditional technologies being time consuming, expensive, complex and sample destructive in nature, are not able to meet the current desire of agro-industry. This study presents development of a novel optical technology prototype for rapid, non-destructive, accurate detection of pesticide residue in agricultural product for industrial application. The current thesis presents the design and development of hardware and software system equipped with Raman spectroscopy system for detection of commercially available organophosphorous chlorpyrifos pesticide in apple as pesticide detection technology.A Charged Coupled Device (CCD) camera mounted with a Raman spectroscopy was used for development of optical system. A laser module of 785 nm was used for laser production.Bifurcated optical fiber at one end was mounted to the laser module to act as light source to hit sample with laser light, while the other end of the optical fiber was mounted to the Raman spectrometer through input slit to direct the scattered light beam from the sample into diffraction grating to analyze the Raman scattering signal.Hardware and its control module was designed and developed for collecting spectral signal from apple surface without causing any damage to sample. Apple holding and rotating device, capable of holding apple of different shape and size and rotate the sample to prescribed point for Raman spectral acquisition maintaining a fix gap between optical probe and sample surface point of detection was designed and developed.A software system in Lab View platform was developed for CCD control, Raman spectral acquisition, and spectral analysis for qualitative and quantitative modeling. The software system was equipped with easy to use Graphical Users' Interface (GUI). The software was developed with the objective of real-time analysis in industrial processing line.Experiments were performed with Fuji and Gala apple samples contaminated with commercially available organophosphorous (48% chlorpyrifos) pesticides. Raman spectral signal from pure chlorpyrifos pesticide was acquired and Raman spectral peak at 677 cm-1 was identified as the Raman fingerprint of chlorpyrifos pesticide.Fuji apple samples of approximately same shape and size were bought from local market. The samples were ensured free of pesticide by cleaning and ringing with soap water and ionic water and let to dry for few hours. Similarly, freshly harvested Gala samples of approximately same size and shape were bought from apple orchard and cleaned and ringed to ensure pesticide free sample and left to dry. Pesticide solutions of different concentrations were prepared by mixing the commercially available chlorpyrifos pesticide with acetone in a beaker. Three Fuji samples were completely dipped inside each beaker containing 4800, 2400 and 1600 mg/kg totaling in 12 Fuji samples. Samples were then left to dry for few hours. Similarly 3 Gala apple samples were dipped inside beaker containing 4800, 3200, 2800, 2400, 2000, 1600 and 1200 mg/kg totaling in 20 samples (2 samples in 1600 ppm beaker) for few seconds and left to dry in laboratory ambient temperature. Raman spectral signal were collected from 20-30 points in the equatorial region of the sample using the developed optical instrument. Laser power of 450 mW and exposure time of 3 sec was used for data acquisition. Input slit of 50 micron was used for spectral collection from Fuji apples while 100 micron slit was used for Raman spectral collection from Gala apple.The apple samples after spectral collection were tested in standard machine. The Fuji apple samples were tested in HPLC and the Gala samples were tested in the GC machine to observe the pesticide residual amount in each sample. The Raman spectral peak at 677 cm-1 was compared with the result obtained by GC and HPLC test for qualitative and quantitative analysis. The results show that the system could classify samples as chlorpyrifos contaminated or free to the minimum detectable amount of 3.25 mg/kg accurately. The Raman spectral signal were processed with SG filter for noise removal, MSC algorithm for baseline drift removal and 8th order polynomial fitting as well as 8th order iterated polynomial fitting separately for fluorescence background removal.The spectral intensity at 677 cm-1 was correlated with the standard test result to establish quantitative model. Results show a very poor linear correlation between spectral peak intensity and pesticide residue value for both Fuji and Gala samples. However,a satisfactory predictive model was developed using MLR algorithm. A correlation coefficient (R) of 0.88 and 0.96 was obtained for Fuji samples using 8th order polynomial fitting and 8th order iterated polynomial fitting for fluorescence removal respectively. Similarly, the correlation coefficient of calibration and validation (Rc and Rv) of 0.85 and 0.79 were obtained in Gala samples with 8th order polynomial fitting for fluorescence removal. For the samples Rc and Rv of 0.86 and 0.84 was achieved when fluorescence were removed by 8th order iterated polynomial fitting.Although the result by iterated polynomial fitting is precise compared to simple polynomial fitting for fluorescence removal, simple polynomial fitting algorithm was used in the software system for its computational simplicity. The results of the optical system shows satisfactory result, the rate of detection is faster than other traditional analytical machines, the system being small and movable,more over the system being simple to operate, can prove to be an efficient machine for detection of pesticide in fruit processing industry. The system was also used to detect deltamethrin and acetamiprid pesticide in apple to trace amount value. The system is easy to modify for detection of mixture of multiple numbers of pesticides in different fruits and vegetables in future for practical industrial application. |
PDF Full Text Request