高场不对称波形离子迁移谱非线性函数系数误差分析

离子迁移率非线性函数系数α₂和α₄是高场不对称波形离子迁移谱(FAIMS)实现物质识别的基础。现有的α₂和α₄缺少先验值和误差分析方法，因此有必要建立关于α₂和α₄求解结果误差的评估标准，进而在此基础上提高α₂和α₄的求解精度。通过自制FAIMS分别对丙酮、异丙醇和1,2二氯苯三种物质在不同分离电压下的检测实验，获取样本在不同分离电压下的谱图和谱图特征值，运用组合的方法从多组分离电压值和相应补偿电压值的数据中选取指定组数，计算出大量的α₂和α₄数据。通过对α₂和α₄的数值分析，探究了α₂和α₄的分布特点和二者之间的相关性，研究了分离电压取点数量和取点方式对其求解结果误差的影响。在拟合α₂和α₄数据不同范围的频数后，发现α₂和α₄符合正态分布，其拟合度均在0.96以上，可以利用α₂和α₄分布的标准差来评估其求解结果的误差。通过对(α₂, α₄)散点进行拟合，发现α₂和α₄之间具有很强的负相关性，三样本的相关度分别为-0.977，-0.968，-0.992。随着分离电压选取点数的增加，其相应的求解结果误差在不断减少。通过不同分离电压取点方式的对比，发现当分离电压取V_Dmax和0.7 V_Dmax时求解结果最优。在保证α₂和α₄求解的准确性的前提下有效降低了检测次数，为FAIMS实现快速现场检测和精确的谱图解析创造了有利的条件。

Study on Error Analysis of Nonlinear Function Coefficient of FAIMS

The solution of ion mobility’s nonlinear function coefficients α₂ and α₄ is the basis for achieving substance identification of High Field Asymmetric waveform Ion Mobility Spectrometry (FAIMS). Currently, nonlinear function coefficients α₂ and α₄ lack priors, meanwhile, existed solving results about α₂ and α₄ are deficient in error evaluation standard. In this article, acetone, isopropanol and 1,2-dichlorobenzene were detected under different dispersion voltage by homemade FAIMS. In general, the spectrum peak of same sample at different dispersion voltage value is unique. Different dispersion voltage and corresponding compensation voltage value determines the value of α₂ and α₄. According to sample spectra at different dispersion voltage value, groups of spectral characteristics were obtained. Affirmatory number of data which were selected from multiple sets of compensation voltage value and dispersion voltage value, so that they were utilized to solved out lots of α₂ and α₄. Lots of factor have an effect on the accuracy of the solving results of α₂ and α₄, for instance, value of compensation voltage and dispersion voltage, style of fetching points of dispersion voltage, and so on. Comparing to other factors, style and amount of dispersion voltage is likely to control. By data analyzing huge amounts of α₂ and α₄ data, this paper explored their characteristic of distribution and correlation about them，research influence of number and method to fetch dispersion voltage detected points for error of solving results. After fitting frequency of α₂ and α₄, it was found that they conform to normal distribution, goodness of fitting exceed 0.96, thus standard deviation of their distribution are able to evaluate error of solving results. In addition, a strong correlation exists between them, relevance of sample is -0.977, -0.968, -0.992 respectively. With increasing of computing selected points, the corresponding error of solving results decrease. By comparing the standard deviation of method to fetch dispersion voltage detected points, found that detecting frequency in case of detecting maximum and the 70% of maximum of dispersion voltage value is lower at approximately same standard deviation, solving effect was optimized in unique fetching points style. Based on the premise of ensuring the accuracy of solving results of α₂ and α₄, it is obvious that reducing the frequency of detections for FAIMS effectively. It created favorable conditions for rapid field detection and precise spectral analysis.