Fast quantitative elemental mapping of highly inhomogeneous materials by micro-Laser-Induced Breakdown Spectroscopy

https://doi.org/10.1016/j.sab.2018.04.018Get rights and content

Highlights

  • Fast method for quantitative elemental mapping of highly inhomogeneous samples by μ-LIBS maps.

  • Analysis of elementary maps using a Self-Organizing Map clustering method coupled to Calibration-Free LIBS for quantification.

  • Method for a larger class of inhomogeneous materials.

Abstract

In this work, a fast method for obtaining a quantitative elemental mapping of highly inhomogeneous samples by μ-LIBS maps is proposed. The method, transportable and cheap, allows the analysis of large maps through the use of a Self-Organizing Map clustering method coupled to Calibration-Free LIBS for quantification of cluster prototypes. The method proposed has been verified on heterogeneous materials such historical lime mortars but it can be easily applied to a larger class of inhomogeneous materials for very different applications (modern building materials, biological samples, industrial materials, etc.).

Introduction

Laser-based techniques have attracted a considerable interest in the last decades for their capability of obtaining elemental images of solid samples without specific treatment, with high spatial resolution and at different depths [1,2,[4], [5], [6]]. A number of applications have been proposed in several fields, ranging from biomedical, geological and environmental research, to forensic analysis, to industrial diagnostics, to Cultural Heritage study and conservation [[7], [8], [9], [10], [11], [12], [13], [14]].

Among these techniques, applications based on the μ-LIBS technique are becoming more and more frequent to scan surfaces and obtain compositional maps, providing interesting results in a number of applications that require qualitative and quantitative analyses [[15], [16], [17], [18], [19], [20], [21], [22], [23]]. The use of μ-LIBS-scan technique has proved to be very advantageous from an economic and experimental point of view with respect to other laser-based techniques such as Laser-Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS) [24]. The method is, in fact, fast, transportable, relatively cheap and can analyse simultaneously elements with very different ionization energy, a task that can be problematic in LA-ICP-MS [25]. While the qualitative analysis of μ-LIBS elemental maps is a relatively simple task, the quantification of the elemental composition of the sample is much more challenging. In principle, a quantitative analysis based on the use of reference samples of known composition for building linear or non-linear, uni- or multi-variate calibration surfaces is applicable only when the matrix of the sample remains more or less constant in the region of analysis [26]. If the sample is characterized by strong inhomogeneities, with materials of different matrixes, or when suitable reference samples are not available, a possible approach to quantitative elemental mapping would be the use of Calibration-Free approaches [[26], [27], [28]]. An important drawback of the CF-LIBS approach, however, is the time required for the analysis: the emission lines of the elements in the samples must be individuated and their intensities calculated through their fit with a Voigt profile. The electron number density must be calculated from the Stark broadening of the hydrogen Balmer alpha line, then the electron temperature must be calculated from the Boltzmann or Saha-Boltzmann plot. Finally, the sample composition must be calculated. If automated all these operations take at least less than 30 s per spectrum; however, μ-LIBS elemental maps with megapixel spatial resolution have been obtained by different groups, and a CF-LIBS approach applied on millions of LIBS spectra is, at the moment, unrealistic. D'Andrea et al. [29] have recently proposed a hybrid Artificial Neural Network (ANN) – CF-LIBS method that can be very effective in most of the cases, but requires the variations in the material matrix to be relatively small for the ANN to work properly.

In this work, we propose a method based on the sequential application of elemental map segmentation (obtained using an automatic classification method based on the use of Self-Organizing Maps, as proposed by the authors in [30]), followed by a CF-LIBS analysis of the prototypal spectra representing the different clusters (materials) in the map. The method is presented and tested for the analysis of ancient mortars. The knowledge developed in the study of this class of highly inhomogeneous materials may have also interesting applications in the analysis and study of modern binding materials and techniques.

Section snippets

Materials and methods

To assess the analytical capability of the method proposed, two mortar fragments from the Norman Adrano Castle (Catania, Sicily) were selected as test samples. The two samples (labelled with the inventory numbers N2-2 and S2-3) have been analysed with the permission of the Soprintendenza per i Beni culturali e Ambientali di Catania. They consist of polished thin sections consolidated by epoxy resin; the mortars are characterized by a heterogeneous binder with the presence of aggregates due to

Experimental procedure

A μ-Modì double-pulse instrument [38], equipped with a collinear double-pulse Nd:YAG Laser (λ = 1064 nm) coupled with a Zeiss Axio Plan A1 microscope with 10× objective was used for the mapping of the samples. The energy of the two pulses was set to 20 mJ and 30 mJ, respectively, in 20 ns FWHM [39]. The delay between the laser pulses was set at 1 μs. The LIBS signal was collected using an optical fiber, placed at 45° with respect to the laser direction, at a distance of about 1 cm from the

Results

Based on the intensity of the lines identified, a series of elementary maps constituting the starting point for the subsequent clusterization (or segmentation) processing were obtained, as shown in Fig. 2.

Following the method described in ref. [30], for the qualitative characterization of the spatial relationship between aggregate and binder we realized for both the samples a grayscale map of the Ca/Mg line intensity ratio and a false-color map of the distribution of Si (red), Al (green) and Ca

Conclusions

We have proposed a fast method for the quantitative analysis of μ-LIBS elemental images, based on the application of the SOM method for the determination of the different classes of materials in the samples, followed by CF-LIBS analysis of the average representative spectra.

In this way, large variation of the sample matrix can be dealt with, and the textural features of the material can be obtained. The technique proposed is transportable, rapid and cheap. It has been presented and tested for

Acknowledgment

This work has been partially supported by MIUR (PRIN 2015 - 2015WBEP3H).

References (45)

  • N. Li et al.

    Microstructural changes in alkali-activated slag mortars induced by accelerated carbonation

    Cem. Concr. Res.

    (2017)
  • C.M. Belfiore et al.

    Image processing of the pozzolanic reactions in Roman mortars via X-Ray Map Analyser

    Microchem. J.

    (2016)
  • P.A. Benedetti et al.

    Effect of laser pulse energies in laser induced breakdown spectroscopy in double-pulse configuration

    Spectrochim. Acta B

    (2005)
  • T. Kohonen

    The self-organizing map

    Neurocomputing

    (1998)
  • T. Villmann et al.

    Neural maps in remote sensing image analysis

    Neural Netw.

    (2003)
  • G. Cristoforetti et al.

    Local thermodynamic equilibrium in laser-induced breakdown spectroscopy: beyond the McWhirter criterion

    Spectrochim. Acta B

    (2010)
  • E. Tognoni et al.

    A numerical study of expected accuracy and precision in Calibration-Free Laser-Induced Breakdown Spectroscopy in the assumption of ideal analytical plasma

    Spectrochim. Acta B

    (2007)
  • A. Riedo et al.

    Laser ablation/ionisation mass spectrometry: sensitive and quantitative chemical depth profiling of solid materials

    Chimia

    (2016)
  • P.K. Diwakar et al.

    Ultrafast laser ablation ICP-MS: role of spot size, laser fluence, and repetition rate in signal intensity and elemental fractionation

    J. Anal. At. Spectrom.

    (2014)
  • L. Wang et al.

    Developments of laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) in microanalysis

    Geol. Bull. China

    (2012)
  • M.L. Warburton et al.

    Validation of depth-profiling la-ICP-MS in otolith applications

    Can. J. Fish. Aquat. Sci.

    (2017)
  • K.E. Sjåstad et al.

    Studies of SRM NIST glasses by laser ablation multicollector inductively coupled plasma source mass spectrometry (LA-ICP-MS)

    J. Anal. At. Spectrom.

    (2012)
  • Cited by (0)

    Selected papers from the 9th Euro-Mediterranean Symposium on LIBS held in June 2017 in Pisa, Italy.

    View full text