提出了以纳米色敏传感器捕获挥发气体为中间介质的可见/近红外光谱检测霉菌感染小麦的新方法。 以霉菌感染的小麦为研究对象, 在恒温恒湿条件下储藏不同的时间; 通过纳米化处理过的色敏传感作为中间介质捕获小麦霉变后产生的特征挥发气体, 采用可见/近红外光谱检测分析挥发气体与纳米色敏传感结合前后的可见和近红外区域光谱变化特征, 结合多变量分析模型预测小麦的霉菌菌落数。 以灰绿曲霉和白曲霉为目标菌种, 分别接种至无菌小麦以培养出不同带菌量的小麦样本, 分别储藏3~9 d。 根据前期的预实验, 采用对特征霉变挥发气体敏感的色敏材料(8-(4-硝基苯基)-4,4-二氟硼二吡咯甲烷(NO₂BDP)和8-(4-硝基苯基)-4,4-二氟-6-溴硼二吡咯甲烷(NO₂BrBDP))组成的纳米传感器捕获特征挥发气体。 采用无皂乳化法合成纳米级微球聚苯乙烯-丙烯酸处理色敏材料, 提高色敏传感器捕获特征挥发气体的能力, 以纳米化前后的NO₂BDP和NO₂BrBDP四种色敏材料构建传感器阵列。 采集不同带菌量的小麦样本的可见/近红外光谱, 运用多变量分析方法对光谱信息进行变量筛选, 用平板菌落计数法测定菌落总数分别建立灰绿曲霉和白曲霉菌落总数的定量预测模型。 实验结果表明, 以纳米化NO₂BDP和NO₂BrBDP组成的传感器阵列为中介采集的带菌小麦的特征光谱波段建立的联合区间-无信息变量消除法偏最小二乘模型(Si-UVE-PLS)效果最佳, 此时预测集的交叉验证平方根(RMSECV)为0.444 4 lgcfu, 实测值与预测值的相关系数 R_p为0.981 1。 采用2个纳米化和2个非纳米材料组成的色敏传感阵列结合联合区间-遗传算法偏最小二乘模型(Si-GA-PLS)检测白曲霉菌落总数取得最佳检测结果, RMSECV为0.434 9 lgcfu, R_p为0.977 2。 表明通过可见/近红外光谱与纳米色敏传感器技术结合的方法, 能够良好的完成对挥发性气体的定量检测研究, 实现霉变小麦的品质监控。

This paper innovatively proposes a new method for detecting mold-infected wheat by visible/near-infrared spectroscopy, and it was employed with a nanoscale colorimetric sensor as an intermediate medium to detect volatile organic compounds (VOCs). The mold-infected wheatwas stored under constant temperature and humidity conditions for different time duration to prepare experimental samples. The wheat after mold infection had different characteristic volatile gases, and visible/near-infrared spectroscopy was used to collect the spectrum information before and after the combination of volatile gas and nanosized colorimetric sensor respectively. The multi-variable analysis model was combined to predict the number of mold colonies of wheat. The Aspergillus glaucus and Aspergillus candidus were inoculated into sterile wheat to cultivate, and wheat samples were prepared by storing for 0~9 days. According to the pre-experimental study, colorimetric material 8-(4-nitrophenyl)-4,4-difluorobora-dipyrromemethane (NO₂BDP) and 8-(4-nitrophenyl)-4,4-difluoro-6-bromoborodipyrrolethane (NO₂BrBDP) sensitive to volatile gases of wheat was used. A nanosized sensor array was fabricated to detect these characteristic volatile gases. The experiment used the soap-free emulsification method to synthesize a nanoscaled microsphere polystyrene-acrylic acid, and it was used to couple NO₂BDP and NO₂BrBDP dyes for producing nanoscale colorimetric sensors with high specific sensitivity. The spectral information of each wheat sample with different mold amount was collected by visible/near-infrared technology, as well as pre-processed by multivariate analysis. The number of colonies were determined by plate colony counting method, and quantitative prediction models were respectively established for the total number of Aspergillus glaucus and Aspergillus candidus colonies.The experimental results showed that the Si-UVE-PLS model for predicting the total number of Aspergillus glaucus colonies is the best in the characteristic spectral collected by two nanosized sensor.And the square root of the cross-validation of the prediction set was 0.444 4 lgcfu; the correlation coefficient between the measured value and the predicted value was 0.981 1. The optimal model for detecting the total number of Aspergillus candidus) colonies was the Si-GA-PLS model with the spectral data collected by the sensor consisting of two nonnanosized and two nanosized sensors. The RMSECV of the prediction set was 0.554 5 lgcfu, and the R_p in the predicted set was 0.977 2. The research results showed that the combination of visible/near-infrared spectroscopy and nanoscaled sensor technology could well complete the quantitative detection of volatile gases and realize the quality monitoring of mildew wheat.