激光探针高精度分析方法及其应用研究

发布时间：2017-12-30 19:35

本文关键词：激光探针高精度分析方法及其应用研究　出处：《华中科技大学》2016年博士论文　论文类型：学位论文

【摘要】：激光探针技术,即激光诱导击穿光谱技术,具有制样简单、微损、快速、多元素同时分析和远程检测等优点,特别适合实时、在线、原位检测,因而成为众多领域生产和质量检测极具潜力的分析技术。然而,由于受光谱干扰、基体效应、自吸收效应和仪器漂移等因素的影响,激光探针在分析灵敏度、精确度和准确度上亟待进一步提高。基于此,本文以钢铁和铁矿石为研究对象,研究了几种提高激光探针定量分析灵敏度、精确度和准确度的新方法,并基于自主研发的激光探针成分分析仪为平台解决了仪器分析中建立稳健定标模型和仪器漂移校准的两个关键问题。具体研究内容如下：提出采用可移动式空间约束技术用于激光等离子体光谱增强,克服了固定式空间约束技术的深坑效应,实现了对钢铁成分定量分析精密度和准确度的提高。结果表明,空间约束光谱增强效果主要受激光波长、激光能量和腔体尺寸的影响,即波长越短增强效果越好,通过优化激光能量和腔体尺寸获得了最大增强因子为5.22。进一步研究结果表明,可移动式空间约束技术可使分析精密度最大提高一倍,使得定量分析准确度提升1-5%。其原因在于空间约束通过反射冲击波压缩等离子体,造成等离子体温度和电子数密度的增加,使得等离子体更容易达到局部热力学平衡状态,从而可改善光谱精密度和定量分析准确度。首次采用环形磁铁约束激光诱导等离子体,实现了对等离子体空间约束与磁约束的双重作用,并将其用于钢铁的定量分析。采用最大磁场强度为4090 G的环形磁铁,获得谱线强度的最大增强因子为3.1,信噪比的最大增强因子为2.9,使钢铁中V. Mn、Cr、Ti等痕量元素定标曲线的决定系数较无约束时均提升至0.99以上,检测极灵敏度提高了2～-4倍。提出一种改进的偏最小二乘回归(PLSR)算法建立了铁矿石激光探针检测的多元回归模型,实现了对铁矿石粉末中8种氧化物、品味和酸碱度的精确定量分析。研究了一种适合于激光探针检测的粉末压片制样的方法,解决了直接压片难退模、易破碎和开裂的问题。为解决铁矿石定量分析的基体效应、自吸收效应和主量元素分析难的问题,只采用10～20 nm一段包含多条元素特征谱线的光谱数据建立PLSR多元回归模型,实现了铁矿石粉末中CaO、MgO、Al2O3、SiO2、MnO、TiO2、Na2O、K2O等8种氧化物的精确定量分析,平均相对误差在2.9-15.46%之间：铁矿石品位的预测均方根误差(RMSE)和平均相对误差(ARE)分别达到0.5134 wt.%和0.69%(低于1%的行业允许误差)：在此基础上计算的铁矿石酸碱度RMSE和ARE分别达到0.0048和3.65%。采用自主研发的JGTZ-2型激光探针成分分析仪对仪器的高精度检测方法进行了研究,主要解决了激光探针仪器分析中建立稳健定标模型和仪器漂移校准两个关键问题。基于分析谱线和内标线选择、异常值剔除、分段定标、高阶多项式拟合等方法建立了JGTZ-2对低合金钢9种成分的定标曲线,提出采用高低标校准对仪器漂移进行修正的方法,可有效改善JGTZ-2检测的可重复性。结果表明,JGTZ-2对低合金钢中Mn、Cr、Ni、Cu、Si、Mo、Ti、V、Al等元素的分析误差均达到或优于国家标准,多次测量值的相对标准差RSD均小于2%,Mn、Cr、Ni、Cu、Si、Mo和Ti等元素的相对误差均小于8%,其检测准确度达到与ARL-4460型火花直读光谱仪相当的水平。
[Abstract]:Laser probe technique, the laser induced breakdown spectroscopy, with sample preparation is simple, rapid, and micro damage analysis and remote detection etc. multiple elements, especially suitable for real-time, on-line, in situ detection, and thus become the analysis technology in many areas of production and quality testing potential. However, due to spectral interference and matrix effect effect of self absorption factors effect and instrument drift, laser probe in the analysis of sensitivity, precision and accuracy should be further improved. Based on this, this paper takes the steel and iron ore as the research object, studied several kinds of improved laser probe quantitative sensitivity analysis, a new method of precision and accuracy, and the laser microprobe analyzer independently based on the development of the platform to solve two key problems to establish robust calibration model and calibration of the instrument drift in instrumental analysis. The specific research contents are as follows: put forward by Mobile space constraint technology for laser plasma spectral enhancement, overcomes the effect of pit fixed space constraint technology, to realize the quantitative analysis of the components of steel to improve the precision and accuracy. The results show that the spectral space constraint enhancement effect mainly by laser wavelength, laser energy and the effect of cavity size, namely, the shorter the wavelength enhancement effect the better, by optimizing the laser energy and the cavity size obtained the maximum enhancement factor of 5.22. further research results show that the mobile space constraint technology can make analysis of the maximum density doubled, making the quantitative analysis accuracy improvement 1-5%. the reason is that the space constraints through the reflection of shock wave compression caused by the increase of plasma, plasma temperature and electron number density that makes the plasma to reach local thermodynamic equilibrium, which can improve the precision and quantitative spectra Accuracy analysis for the first time. The ring magnet constraints of laser induced plasma, the dual role of constraints on the plasma space and magnetic confinement, and used for quantitative analysis of iron and steel. The ring magnet maximum magnetic field strength of 4090 G, obtained the maximum spectral line intensity enhancement factor is 3.1, the signal-to-noise ratio of the maximum enhancement factor 2.9, the V. Mn Cr in iron and steel, and the coefficient of determination Ti trace elements calibration curve with no constraints were increased to more than 0.99, detection sensitivity is increased by 2 ~ -4 times. Proposed a modified partial least squares (PLSR) algorithm to establish the multiple regression model of iron ore laser probe detection and the 8 kinds of iron ore powder oxide, accurate quantitative analysis of taste and pH were studied. A suitable laser probe detection method of powder preparation, solve the difficult direct compression casting, easy Crushing and cracking problems. In order to solve the matrix effect of quantitative analysis of iron ore, the self absorption effect and the main elements of the difficult problem of using only 10 ~ 20 nm a PLSR multiple regression model includes the establishment of a plurality of spectral line spectral data, the iron ore powder in CaO, MgO, Al2O3. SiO2, MnO, TiO2, Na2O, accurate quantitative analysis K2O 8 oxides, the average relative error between 2.9-15.46% in the prediction of iron ore grade of the root mean square error (RMSE) and the average relative error (ARE) reached 0.5134 wt.% and 0.69% respectively (less than 1% of the industry: the calculation of allowable error) on the basis of iron ore pH RMSE and ARE were studied to achieve high precision detection method of JGTZ-2 type laser microprobe analyzer and 0.0048 3.65%. using self-developed instrument, mainly to solve the analysis of laser probe instrument to build robust calibration model And the instrument drift calibration two key problems. The analysis of spectral line and line selection based on abnormal value elimination, subsection calibration, high order polynomial fitting method is established for JGTZ-2 low alloy steel 9 components of the calibration curve, the height calibration correction method for the instrument drift, can effectively improve the JGTZ-2 detection repeatability. The results showed that JGTZ-2 of low alloy steel Mn, Cr, Ni, Cu, Si, Mo, Ti, V, Al and other elements of the error analysis are reached or exceeded national standard, the relative standard deviation of RSD values of repeated measurements were less than 2%, Mn, Cr, Ni, Cu. Si, relative error of Mo and Ti elements were less than 8%, the detection accuracy of ARL-4460 type spark spectrometer equivalent level.

【学位授予单位】：华中科技大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TN249

【相似文献】