远程激光诱导击穿光谱技术分析岩石元素成分

远程激光诱导击穿光谱技术是一种利用脉冲激光和聚焦光路对远距离目标烧蚀击穿，获取目标等离子体光谱，定性或定量分析物质元素组成的光谱探测技术。设计并搭建了一套远程激光诱导击穿光谱系统。该系统结合卡式望远镜光学结构，实现探测2～10 m距离的目标、并可自动变焦。基于该系统提出一种远程探测岩石主要元素含量方法。通过对比实验，研究了脉冲能量、采集延时、积分时间、探测点累计探测次数对光谱信号的影响，确定了岩石谱线获得的最佳条件。选择48块岩石标本和6种常见国标岩石样品(页岩、花岗岩、安山岩、玄武岩、片麻岩、伟晶岩)进行LIBS实验。以原子光谱数据库为参考，根据岩石的主要元素提取特征谱线(SiⅠ390.55 nm，AlⅠ394.40 nm，AlⅠ396.15 nm，CaⅡ396.85 nm，FeⅠ404. 60 nm，SiⅠ500.60 nm，MgⅠ518.36 nm，NaⅠ589.59 nm)。利用偏最小二乘算法(PLS)建立岩石成分定量分析模型，将48块岩石标本作为训练集进行求解，并用六种国标岩石对模型进行检测，预测岩石Si和Al元素含量，平均误差分别为9.4%和9.6%。

Elemental Analysis of Rock with Remote Laser Induced Breakdown Spectroscopy

As a kind of spectroscopic technique,the remote laser-induced breakdown spectroscopy (Remote LIBS) can qualitatively or quanti tatively measure the elemental compositions of remote targets with high-power l aser and focusing optical path.In this work,a Remote LIBS system was designed and established to probe the target from 2 to 10 meters.It is characterized wit h Cassegrain telescope structure and automatic focusing technique.Based on this system,a method to remotely retrieve the major elemental abundance of rocks is presented.With comparative experiment,the influences on spectral signal by pu lse laser energy,acquisition delay time,integration time,accumulative pulse n umber are analyzed to find out the optimum parameters:the wavelength of laser i s 1 064 nm,pulse energy is 120 mJ,delay time is 1.5 μs,integration time is 1 ms,and each spectrum data acquired with averaging 30 times detection.48 piec es of rock specimens and 6 kinds of standard rock samples (shale,granite,andes ite,basalt,gneiss and pegmatite) are selected for the experiment.As to the at omic spectra database,8 characteristic spectral lines of the major elements (SiⅠ390.55 nm,AlⅠ394.40 nm,AlⅠ396.15 nm,CaⅡ396.85 nm,FeⅠ404.6 nm,SiⅠ500.60nm,MgⅠ518.36nm,NaⅠ589.59 nm) were extracted for analyses.Then a PLS model is constructed to quantitatively analyze the rock elements.48 rock sp ecimens were selected as the training sets to serve the model.The 6 standard sa mples were used to test the solved model.The testing results shows that the ele mental abundance of Si and Al can be predicted accurately with average relative error of only 9.4% and 9.6% respectively.