基于底物内标的蜂蜜中硝基呋喃妥因拉曼信号校正方法

针对表面增强拉曼光谱信号重复性欠佳的问题，利用实验室自行搭建的拉曼点检测系统，以蜂蜜中硝基呋喃妥因兽药为检测对象，探讨了基于蜂蜜固有内标的硝基呋喃妥因表面增强拉曼峰强校正方法。首先通过含不同浓度硝基呋喃妥因蜂蜜样品及硝基呋喃妥因标准品的拉曼光谱对比分析，确定739 cm-1处蜂蜜拉曼特征位移作为底物蜂蜜的内标峰，用比值法校正硝基呋喃妥因1 353和1 612 cm-1处拉曼特征峰强用于蜂蜜中硝基呋喃妥因定量分析。相同条件下分别采集了浓度为20 mg·kg-1的硝基呋喃妥因蜂蜜样品表面增强拉曼光谱30次，1 353和1 612 cm-1处硝基呋喃妥因特征峰强相对标准偏差（RSD）分别为11.515 6%和11.162 5%，利用739 cm-1处蜂蜜拉曼特征峰强作为内标分别校正1 353和1 612 cm-1处硝基呋喃妥因拉曼特征峰强后相对标准偏差分别降为4.852 6%和4.733 2%，显著提升了表面增强拉曼特征峰强的重复性和稳定性。因为仪器系统误差及表面增强过程中不可控因素引起的人为误差等对样品表面增强光谱中739 cm-1处蜂蜜特征峰强和1 353和1 612 cm-1处硝基呋喃妥因特征峰强的影响是完全相同的，所以通过内标比值法可以有效消除和减少拉曼信号稳定性和重复性差的问题。最后采集硝基呋喃妥因浓度范围为0.4~20 mg·kg-1的69个蜂蜜样品，基于硝基呋喃妥因1 353和1 612 cm-1处拉曼特征峰强和蜂蜜739 cm-1处拉曼特征峰强比值，分别建立了一元线性回归预测模型和多元线性回归模型，其中基于蜂蜜739 cm-1处内标校正硝基呋喃妥因1 612 cm-1处拉曼特征峰强的一元线性回归模型效果最佳，与校正前相比具有更高的精度和预测能力。该模型校正集决定系数（R2_C）和验证集决定系数（R2_Ｖ）分别为0.971 2和0.969 6，校正集均方根误差（RMSEC）和验证集均方根误差（RMSEP） 分别为1.115 1和1.242 2，相对分析误差（RPD）为4.306 0。结果表明，被测底物本身持有固有内标的样品可无需加入额外的内标物，简单用内标比值法可以有效消除仪器的系统误差以及表面增强剂与样品的混合时间等对拉曼信号强度的影响，显著提高了拉曼特征信号的重复性和稳定性，为表面增强拉曼光谱定量分析提供了技术参考。

A Method for Correcting Nitrofurantoin Raman Signal in Honey Based on Internal Standard of Substrate

In order to solve the problem of poor reproducibility of surface-enhanced Raman spectroscopy signals,this paper explored the surface-enhanced Raman spectroscopy correction method based on internal honey standard using the Raman point detection system built by the laboratory.Firstly,the Raman characteristic displacement at 739 cm^-1 was determined to be the standard internal peak of honey by surface enhanced Raman spectra of honey samples and Raman spectra of standard products of nitrofurantoin.The ratio method was used to correct the Raman characteristic peak strength of nitrofurantoin at 1353 and 1612 cm^-1 for quantitative analysis in honey.The surface-enhanced Raman spectra of nitrofurantoin honey samples with a concentration of 20 mg·kg^-1 were collected for 30 times under the same conditions,and the peak intensity relative standard deviations of nitrofurantoin at 1353 and 1612 cm^-1 was 11.5156%and 11.1625%,respectively.However,after using the Raman characteristic peak intensity at 739 cm^-1 as the internal standard to correct the peak intensity of nitrofurantoin at 1353 and 1612 cm^-1,the relative standard deviations were reduced to 4.8526%and 4.7332%,respectively.The repeatability and stability of the surface-enhanced Raman characteristic peaks are significantly improved.Because the instrument system errors and surface enhanced during uncontrollable factors and errors human-induced surface of the sample enhanced 739 cm^-1 honey characteristic peak intensity spectrum and at 1353 and 1612 cm-1 nitro nitrofurantoin characteristic peak intensity the effects are the same,it is possible to effectively eliminate the difference between the Raman signals and decrease the stability or poor repeatability problem by internal standard ratio method.Because the error of the instrument system and the artificial error caused by uncontrollable factors in the surface enhancement process have the same effect on the peak strength of 739 cm-1 honey characteristic peak and the peak strength of 1353 and 1612 cm-1 nitrofurantoin characteristic peak in the sample surface enhancement spectrum,therefore,the internal label ratio method can effectively eliminate or reduce the problem of Raman signal stability and poor repeatability.Finally,69 honey samples with a nitrofurantoin concentration range of 0.4~20 mg·kg^-1 were collected.Based on Raman characteristic peak intensity at 1353 and 1612 cm^-1 of nitrofurantoin and Raman characteristics at 739 cm^-1 of honey,The linear regression prediction model and the multi-linear regression model were established respectively,in which the unary linear regression model based on the internal standard of honey at 739 cm^-1 to correct the Raman characteristic peak intensity at 1612 cm^-1 of nitrofurantoin with higher precision and predictability.In this model,the determination coefficient of the calibration set and validation set is 0.9712 and 0.9696 respectively,and the root means square error of calibration set and validation set are 1.1151 and 1.2422 respectively,and the relative analysis error is 4.3060.The results show that the sample of the underlying itself holds the inherent internal criteria without adding additional internal markers,and the simple internal standard ratio method can effectively eliminate the effect of the instrument’s system error and the mixing time between surface enhancers and samples on the Raman signal strength,and significantly improve the repeatability and stability of the Raman characteristic signal.It provides a technical reference for the quantitative analysis of surface-enhanced Raman spectra.