目的在近红外光谱(near infrared spectroscopy,NIR)与表面增强拉曼光谱(surface-enhanced Raman spectroscopy,SERS)特征层数据融合的基础上构建偏最小二乘回归(partial least squares regression,PLSR)模型实现花生油中黄曲霉毒素B_(1...目的在近红外光谱(near infrared spectroscopy,NIR)与表面增强拉曼光谱(surface-enhanced Raman spectroscopy,SERS)特征层数据融合的基础上构建偏最小二乘回归(partial least squares regression,PLSR)模型实现花生油中黄曲霉毒素B_(1)(aflatoxin B_(1),AFB_(1))含量的快速检测。方法首先,分别采集待测样本的NIR与SERS光谱。其次,将采集的NIR与SERS光谱分别进行光谱预处理。然后,采用基于希尔伯特-施密特独立准则的变量空间迭代优化算法(Hilbert-Schmidt independence criterion based variable space iterative optimization,HSIC-VSIO)分别筛选NIR与SERS光谱的特征变量。最后,将筛选的特征变量进行融合并构建PLSR模型用于定量检测花生油中AFB_(1)含量。结果与NIR光谱数据、SERS光谱数据以及NIR与SERS光谱直接融合数据构建的PLSR模型相比,NIR与SERS光谱特征层融合数据构建的PLSR模型具有最佳的预测性能:校正集均方根误差(root mean squared error of calibration set,RMSEC)为0.1569,校正集决定系数(coefficient of determination of calibration set,R_(c)^(2))为0.9908,预测集均方根误差(root mean squared error of prediction set,RMSEP)为0.1827,预测集决定系数(coefficient of determination of prediction set,R_(c)^(2))为0.9854,性能偏差比(ratio of performance to deviation,RPD)为8.2761。将本方法与标准方法分别检测真实含有AFB_(1)的花生油样本,结果表明两者的检测性能无显著性差异(P=0.84>0.05)。结论本方法可实现花生油中AFB_(1)含量的快速、高精度定量检测,也验证了NIR与SERS光谱融合的可行性与有效性。展开更多
Although surface-enhanced Raman spectroscopy(SERS)substrates have achieved high sensitivity,it is still difficult to apply these SERS protocols to the on-site detection of real samples due to the SERS substrate being ...Although surface-enhanced Raman spectroscopy(SERS)substrates have achieved high sensitivity,it is still difficult to apply these SERS protocols to the on-site detection of real samples due to the SERS substrate being fabrication-complexed,unstable,reproducible,or unable to be applied in batch production.Here,a large-scale ordered two-dimensional array of Au nano-hemispheres was developed through electron beam vaporization of Au onto the easy-available commercialized anodic aluminum oxide(AAO)template with two-layer nanostructures of different diameters.The uniform Au nano-hemisphere is reliable for SERS detection of malachite green(MG)due to the relative standard deviation(RSD)of the SERS intensities at different locations less than 10%.With the optimized excitation wavelength,solvent effect and pH environment,the linear range of MG detection spans from 10^(-10) to 10^(-6) mol/L with a limit of detection(LOD)of 4×10^(-10) mol/L.The enhancement factor can reach 1.2×10^(6).After extraction with acetonitrile and dichloromethane,MG in the spiked tilapia was detected with satisfactory recovery.The results indicate that the Au nano-hemisphere array can be expected to greatly facilitate SERS practical applications in detecting harmful food additives and chemicals due to the advantages of chemical inertness,physical robustness,simple fabrication,controllability,large-area uniformity,and large-batch production.展开更多
文摘目的在近红外光谱(near infrared spectroscopy,NIR)与表面增强拉曼光谱(surface-enhanced Raman spectroscopy,SERS)特征层数据融合的基础上构建偏最小二乘回归(partial least squares regression,PLSR)模型实现花生油中黄曲霉毒素B_(1)(aflatoxin B_(1),AFB_(1))含量的快速检测。方法首先,分别采集待测样本的NIR与SERS光谱。其次,将采集的NIR与SERS光谱分别进行光谱预处理。然后,采用基于希尔伯特-施密特独立准则的变量空间迭代优化算法(Hilbert-Schmidt independence criterion based variable space iterative optimization,HSIC-VSIO)分别筛选NIR与SERS光谱的特征变量。最后,将筛选的特征变量进行融合并构建PLSR模型用于定量检测花生油中AFB_(1)含量。结果与NIR光谱数据、SERS光谱数据以及NIR与SERS光谱直接融合数据构建的PLSR模型相比,NIR与SERS光谱特征层融合数据构建的PLSR模型具有最佳的预测性能:校正集均方根误差(root mean squared error of calibration set,RMSEC)为0.1569,校正集决定系数(coefficient of determination of calibration set,R_(c)^(2))为0.9908,预测集均方根误差(root mean squared error of prediction set,RMSEP)为0.1827,预测集决定系数(coefficient of determination of prediction set,R_(c)^(2))为0.9854,性能偏差比(ratio of performance to deviation,RPD)为8.2761。将本方法与标准方法分别检测真实含有AFB_(1)的花生油样本,结果表明两者的检测性能无显著性差异(P=0.84>0.05)。结论本方法可实现花生油中AFB_(1)含量的快速、高精度定量检测,也验证了NIR与SERS光谱融合的可行性与有效性。
基金the National Natural Science Foundation of China(32272417)the National Key Research and Development Program of China(2018YFC1602802)+1 种基金the Natural Science Foundation of Fujian Province,China(2020J01681)the Open Project of PCOSS,Xiamen University,China(201926).
文摘Although surface-enhanced Raman spectroscopy(SERS)substrates have achieved high sensitivity,it is still difficult to apply these SERS protocols to the on-site detection of real samples due to the SERS substrate being fabrication-complexed,unstable,reproducible,or unable to be applied in batch production.Here,a large-scale ordered two-dimensional array of Au nano-hemispheres was developed through electron beam vaporization of Au onto the easy-available commercialized anodic aluminum oxide(AAO)template with two-layer nanostructures of different diameters.The uniform Au nano-hemisphere is reliable for SERS detection of malachite green(MG)due to the relative standard deviation(RSD)of the SERS intensities at different locations less than 10%.With the optimized excitation wavelength,solvent effect and pH environment,the linear range of MG detection spans from 10^(-10) to 10^(-6) mol/L with a limit of detection(LOD)of 4×10^(-10) mol/L.The enhancement factor can reach 1.2×10^(6).After extraction with acetonitrile and dichloromethane,MG in the spiked tilapia was detected with satisfactory recovery.The results indicate that the Au nano-hemisphere array can be expected to greatly facilitate SERS practical applications in detecting harmful food additives and chemicals due to the advantages of chemical inertness,physical robustness,simple fabrication,controllability,large-area uniformity,and large-batch production.
