Multi-element quantification of ancient/historic glasses by laser ablation inductively coupled plasma mass spectrometry using sum normalization calibration

Abstract

Laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) for quantitative analysis of ancient/historic glasses is subject to calibration issues which have been addressed in this work. Since ancient/historic glasses have widely ranging matrix compositions, a complementary analysis by an alternative method is generally employed to determine at least one major element which can be used as an internal standard. We demonstrate that such a complementary analysis is unnecessary using a so-called sum normalization calibration technique (mathematically formulated) by simultaneous measurement of 54 elements and normalizing them to 100% [w/w] based on their corresponding oxide concentrations. The crux of this approach is that by assuming a random internal standard concentration of a particular major oxide, e.g. SiO₂, the normalization algorithm varies the internal standard concentration until the cumulated concentrations of all 54 elemental oxides reach 100% [w/w]. The fact that 54 elements are measured simultaneously predetermines the laser ablation mode to rastering. Nine glass standards, some replicating historic compositions, were used for calibration. The linearity of the calibration graphs (forced through the origin) represented by the relative standard deviations in the slope were between 0.1 and 6.6% using SiO₂ as an internal standard. This allows high-accuracy determination of elemental oxides as confirmed by good agreement between found and reported values for major and minor elemental oxides in some synthetic glasses with typical medieval composition (European Science Foundation 151 and 158). Also for trace elemental concentrations of lanthanides in a reference glass (P&H Developments Ltd. DLH7, a base glass composition with nominally 75 μg g⁻¹ elements added) accurate data were obtained.

Interferences from polyatomic species and doubly charged species on the masses of trace elements are possible, depending on the base composition of the glass, with Ba and Sb glasses showing potential interferences on some lanthanides. We showed that they may be reduced to a great extent by using an Octopole Reaction System although the overall sensitivity decreases which may be a problem for some low-level determinations.

Introduction

For an understanding of glass compositions, major, minor and trace elements can all be important in archaeometric investigations. These data are generally obtained with the motivation to elucidate past technologies, establish provenance or understand deterioration processes. Techniques for elemental bulk analysis of glass are mainly comprised of atomic absorption spectrometry (AAS), inductively coupled plasma optical emission spectrometry or mass spectrometry (ICP-OES or -MS) and instrumental neutron activation analysis (INAA). For surface analysis of glass, techniques such as X-ray fluorescence (XRF) and proton-induced X-ray emission spectrometry (PIXE) are applied; these techniques may also be used for bulk analysis if a homogeneous matrix is assumed. Laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) is an alternative technique which involves direct solid sampling of the artifact with subsequent real-time detection. Glass is particularly suited to be studied with LA-ICP-MS as the conventional bulk analysis techniques mentioned above suffer from elaborate digestion protocols (AAS and ICP-OES or -MS) or low sample throughput (INAA), whereas the surface analysis techniques mentioned above are limited in sensitivity.

In LA-ICP-MS practice a glass sample is placed in a chamber where ablation takes place and particles are released as a fine, dense aerosol. The aerosol is then transported to the ICP-MS by an argon or helium carrier gas. The ablation characteristics depend on the type of laser used, deep-UV lasers are the norm in the analysis of glass, with 193 nm lasers producing smaller particles than 213 nm lasers [1]. Furthermore, a suite of parameters may be adjusted (fluence, laser beam diameter, pulse rate, ablation time, ablation mode, ablation chamber gas type/flow rate) to optimize the ablation characteristics (and generate sub-micrometer particles with a composition matching these of the glass) and effectively transport and atomize/ionize them in the plasma [2], [3]. The microanalysis capabilities of the technique are particularly effective in overcoming the limitations associated with very small samples (sub-millimeter dimensions). LA-ICP-MS is not yet frequently applied to analysis of ancient/historic glass or glazes, although it has been used to reveal, for example, compositional differences between Egyptian and Mesopotamian blue and colorless glasses [4], for trace analysis of Venetian and facon-de-Venise glass vessels of the 16th and 17th century [5], for characterization of cobalt pigments in Valencian ceramics [6], and for complementary analysis of historical glass by scanning electron microscopy with energy dispersive X-ray spectroscopy [7]. In contrast, in forensic science the technique is widely utilised, especially for the analysis of float glass [e.g. [8], [9]], probably because of the limited matrix variation and the availability of good calibration standards.

Various quantification strategies exist in the field of LA-ICP-MS as the measurements are prone to ablation rate variations depending on the composition of the matrix. For this reason matrix-matched external standards and/or (multi-element) internal standards are generally used for calibration purposes [10]. However, these approaches do not resolve issues related to ancient/historic glasses with poorly-characterized matrices and so an alternative, complementary method of analysis is necessary in order to determine at least one major element which can be used as an internal standard. To circumvent such complementary analysis several approaches have been mentioned in the literature based on some kind of sum normalization of elemental oxides to 100% [w/w]. Halicz and Günther [11] reported an oxide scaling method applied to silicate reference materials using a desolvated solution standard to determine a scaling factor for Ca, which was then used for all other elements. Gagnon et al. [12] quantitatively analyzed silicate reference materials based on the same principle but using scaling factors for all element abundances simultaneously and applying silicate reference materials for calibration. Gratuze et al. [13] described an elegant method for analysis of ancient/historic glass by measuring element response factors assuming a random concentration for one of the major elements. An application of this method to study trade routes of ancient glass beads has been recently published [14].

In this work the method introduced by Gratuze et al. [13] was adapted using raster sampling instead of crater sampling, allowing long acquisition times and therefore normalization on all (54) elemental oxides instead of ca. 10 major elemental oxides. By avoiding crater sampling with inherent deep craters, potentially less elemental fractionation will be encountered resulting in higher measurement accuracy [15]. The protocol was evaluated by analyzing several synthetic glass samples with known elemental composition, both for major, minor and trace elements. Furthermore, spectroscopic interferences from polyatomic species and doubly charged species caused by the base elements in glass on specific masses at trace element analysis were studied.