原位分析和在线检测是激光诱导击穿光谱(LIBS)技术的一大优势, 但是, 在野外环境中, 人们无法对样品进行统一预处理, 面对各种形态的待测样品如何保证LIBS的检测精度是函待解决的一大难题。 提出一种多谱线内定标的方法来解决上述问题, 即通过求解多条分析谱线的强度和与内标元素谱线的强度比值来建立定标曲线, 进而降低光谱信号波动带来的误差, 提高线性相关性和检测精度。 实验中以铅黄铜合金样品为例, 采用LIBS对厚度不一(最大变化值为±2 mm)的铅黄铜样品中的Pb元素进行了定量检测研究, 并分别采用传统定标法和多谱线内定标法对这种不规则样品进行校正和建立定标曲线。 实验发现, 对于不规则样品, 传统定标法的检测精度大大降低, 定标曲线没有明显的线性关系。 当采用单条谱线的内定标方法时, 定标曲线线性相关度大大提高, 校正决定系数达到0.724 89。 而采用多条谱线内标方法(考虑多条分析谱线的相对强度总和)计算发现, 当选取5条Pb谱线(Pb 261.42 nm, Pb 280.20 nm, Pb 368.35 nm, Pb 405.78 nm和Pb 520.14 nm)进行计算时, 定标曲线线性拟合度达到0.984 6, 由此可见该方法消除了样品不规则所带来的光谱强度波动误差, 显著提高了测量精度。 虽然继续增加分析谱线数目可以进一步提升线性相关度, 但是也会增加计算的复杂度, 所以选择合适的分析谱线是十分重要的。 此外, 通过多谱线内标法也能一定程度上消除基体效应和光谱干扰等影响, 是一种简单有效且具备普适性的数据处理方法。 当然, 该方法也存在一定的局限性(如样品成分分布极不均匀、 样品表面极不规则致使激光能量低于击穿阈值等), 不过通过调整和优化检测装置方案(例如增大激光能量、 增大聚焦光斑、 采用长焦距聚焦透镜等)可以更好的发挥该方法的优势。 该研究内容可以为LIBS原位检测和在线检测的应用提供一种新思路。

In-situ analysis or on-line detection is a major advantage of laser-induced breakdown spectroscopy (LIBS) technology. However, in the outdoor environment, people cannot uniformly pre-process samples, then it is difficult to ensure the detection accuracy when facing the various types of samples. In this paper, a multi-line calibration method is proposed to solve the above problem, that is, the calibration curve is established by calculating the intensity ratio of multiple analytical lines and the internal standard element lines, which can reduce the error caused by spectral signal fluctuation and improve linear correlation and detection accuracy. In this experiment, the lead brass alloy samples were taken as an example. The quantitative detection of Pb elements in lead brass samples with different thicknesses (the maximum variation is 2 mm) was carried out by LIBS, and the traditional calibration method and multi-line calibration method were used respectively to establish the calibration curves. The experiment found that for irregular samples, the detection accuracy of the traditional calibration method is very poor, which has no obvious linear relationship from the calibration curve. When the internal calibration method of a single line is adopted, the linear correlation of the calibration curve is greatly improved, and the adjusted determination coefficient reaches 0.724 89. While using the multi-line calibration method (considering the sum of the intensities of multiple analytical lines), it is found that the adjusted determination coefficient of the calibration curve reached 0.984 6 when five Pb lines (Pb 261.42 nm, Pb 280.20 nm, Pb 368.35 nm, Pb 405.78 nm and Pb 520.14 nm) were selected. It can be seen that this method eliminates the spectral intensity fluctuation error caused by sample irregularity and significantly improves the measurement accuracy. While increasing the number of analytical lines can further increase linear correlation, but it also increases the computational complexity, so it is important to choose the appropriate analytical lines. In addition, the multi-line calibration method can also eliminate the matrix effects and spectral interference to a certain extent, which is a simple, effective and universal data processing method. Of course, this method also has limitations (such as extremely heterogeneous sample, extremely irregular sample surface which results in the laser energy below the breakdown threshold, etc.), but by adjusting and optimizing the detection device scheme (for example, increasing laser energy, increasing the diameter of focus spot, using a long-focus lens, etc.), we can improve the advantages of this method. This research content of this paper can provide a new idea for the application of LIBS in-situ analysis and on-line detection.