| چکیده انگلیسی مقاله |
The goal of this study is to identify sources and error structures for a smartphone-based spectrophotometer. Here, we introduced a smartphone-based diffuse reflectance spectrophotometer (smartDRS) with the advantage of rapid, simple, and cost-effective to acquire spectral information from a Cadmium Sulfide (CdS) pigment. Using a grating containing accessory, and the camera of the smartphone a simple smartDRS, with a resolution of 5 nm, was manufactured in our lab [1]. Error structure information is important to identifying and minimizing error sources, as well as to separate chemical meaningful variance from noise variance [2]. To analyze the error structure of smartDRS, we used the error covariance and error correlation matrices obtained through replication [3]. Multivariate analysis techniques as maximum likelihood common factor analysis (MLCFA) and multivariate curve resolution alternative least square (MCR-ALS) methods were used to the analysis of experimental error covariance matrices (ECMs) and obtain the number and contribution of error sources. We analyzed experimental ECMs using the MLCFA and the MCR-ALS methods and obtained three correlated error sources and three independent error sources. Correlated error sources were included signal-proportional noise (Up,s), first proportional noise similar to signal (Up,1), and second proportional noise similar to signal (Up,2). Independent error sources were included iid noise (Viid), signal-proportional noise (Vp,s), and proportional noise similar to signal (Vp). Also, we applied the maximum likelihood criterion based on the Wishart distribution for fitting the calculated ECM to the experimental ECMs [4]. Obtained confidence intervals based on resampling, at the confidence level of 95% and regarding the distribution of the Wishart's objective functions, did not illustrate a considerable shift as the number of replicates to estimated experimental ECMs was increased. The results show that the difference between the calculated ECM and experimental ECM was negligible, and the error sources used for the calculation of ECM were all valid. |
