Background

Inductively coupled plasma mass spectrometry (ICP-MS) has become the most widely accepted analytical technique for simultaneously measuring a range of trace elements at variable concentration levels (Durrant 1999; Günther and Heinrich 1999; Günther et al. 2000; Sylvester 2001a; Heinrich et al. 2003). Laser ablation (LA) system is powerful and versatile for in situ microsampling of solid materials including geological samples particularly when it is attached to ICP-MS (e.g., Jackson et al. 1992; Norman et al. 1996, Günther et al. 1997; Jeffries et al. 1998; Niemax 2001; Heinrich et al. 2003; Günther and Hattendorf 2005; Large et al. 2009; Koenig et al. 2009; Ulrich et al. 2009). The development and application of laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) have grown with a dramatic rate, and in situ analyses for elemental abundances and isotopic compositions are now performed routinely in numerous laboratories worldwide.

In addition to LA-ICP-MS, both ion and electron microprobes are commonly used for in situ analysis of trace elements (Müller et al. 2003). An electron microprobe has advantages of the smallest sampling volume and well-established data reduction procedures as well as quantitative calibration methods, but extreme care is required for elements at very low abundances. In many cases, even long counting times coupled with a detailed background characterization may not achieve the required sensitivity of a trace element (Müller et al. 2003). Ion microprobe has the main strength on both high sensitivity and small sampling volume. However, the severe matrix effect in this technique commonly prevents us from quantitative analyses in the absence of appropriate standard materials (Hinton 1990). LA-ICP-MS analyses combine the strength of the above techniques: relatively straightforward calibration procedures with a low detection limit. A glass reference sample is applicable to the quantitative analyses of trace elements for different types of materials with acceptable uncertainties. However, LA-ICP-MS has relatively poorer spatial resolution and requires relatively larger sampling volumes than both electron and ion microprobes (Woodhead et al. 2007).

The quantitative mapping of trace elements in geological samples has been facilitated by LA-ICP-MS (Treble et al. 2003; Fairchild et al. 2006; Montagna et al. 2007). It is also possible with other microanalytical techniques such as ion microprobe, proton-induced X-ray emission (PIXE), and synchrotron X-ray fluorescence (Woodhead et al. 2007; Ulrich et al. 2009). In contrast to the above methods, LA-ICP-MS trace element mapping is the only technique performing at low limits of detection for a wide range of elements in a relatively short acquisition time. The application of quantitative element distribution maps using LA-ICP-MS has been accelerated with the development of analytical protocol and easy-to-use software such as Iolite (Woodhead et al. 2007; Ulrich et al. 2009).

Agate is an aggregate of various forms of microcrystalline silica, mainly chalcedony, and is generally formed by chemical precipitation in rock cavities (Götze et al. 2001). Chalcedony occurs commonly in a semi-transparent or translucent state, with waxy luster, and varies widely in color, in contrast to colorless, transparent microcrystalline quartz (Polk 2012). Transition metals incorporated during the precipitation color agates (Götze et al. 2001; Möckel et al. 2009). The species and concentration of transition metals in agate are commonly reflected by its rhythmic color banding (Heaney and Davis 1995). The shade of color is further modified when these elements work in combination. Thus, agate is an appropriate sample to test the availability, efficiency, and detection limit of LA-ICP-MS trace element mapping. In this study, the distribution patterns of trace elements in two agates were investigated, using LA-ICP-MS, for demonstrating (1) the advantages of LA-ICP-MS trace element mapping through a correlation between the colors and concentrations of trace elements in the agate samples and (2) the analytical uncertainties and detection limits on the analyses of trace elements best optimized in the study.

Sample descriptions

Two fortification agate samples were selected for elemental mapping because of their sub-radial growth towards the center. Both samples are oval in shape with approximately 50-mm-long major axes and are characterized by the colorful and bluish banding layers, respectively. The colorful agate sample is mainly divided into two growth zones: the inner and outer halves on the basis of ~0.2-mm-thick white band (Fig. 1a). The inner half shows varying colors, from pale orange to red brown whereas the outer half is generally composed of color bands in gray tones. On the other hand, the blue agate sample simply shows alternating blue and white bands (Fig. 1b). Using an optical microscope, numerous small mineral inclusions were observed in the inner and outer halves of the former sample. Two different types of mineral inclusions were identified under scanning electron microscopy. In the inner half, Fe oxide/hydroxide is mainly distributed as clusters of oriented crystallites (up to 10 μm long) (Fig. 2a) whereas mineral inclusions in the outer half are mainly ~10-μm-long hollandite (Ba(Mn4+ 6Mn3+ 2)O16) (Fig. 2b). On the other hand, no mineral inclusions were observed in the blue agate sample. The ablated area of each sample is represented by an open rectangle (Fig. 1a, b)

Fig. 1
figure 1

Two fortification agate samples. They are oval in shape with approximately 50-mm-long major axes and are characterized by a colorful and b bluish banding layers, respectively. The former varies in color, from gray through orange and red to brown towards the center. On the other hand, the latter consists of bluish chalcedony alternating with white microcrystalline quartz. Open rectangles represent the areas of ablation

Full size image
Fig. 2
figure 2

Back-scattered electron images of a Fe oxide/hydroxide and b hollandite inclusions in agate, together with the corresponding EDS spectra

Full size image

Methods

Qualitative mapping was carried out, using a Shimadzu 1600 electron microprobe equipped with wavelength-dispersive spectrometers (WDS) at the Jeonju branch of Korea Basic Science Institute (KBSI), by measuring intensities of the Al Kα, Co Kα, Cr Kα, Cu Kα, Fe Kα, Mn Kα, Ti Kα, and Ni Kα characteristic X-ray lines considered as potential chromophores in agate (Table 1). The analytical conditions were under the instrumental settings at 15 keV accelerating voltage, 1 μm beam size, and 20 nA current with 50 ms dwell time per pixel. Quantitative mapping was carried out, using a New Wave UP213 Nd:YAG laser ablation system attached to a Thermo X Series 2 quadrupole ICP-MS in the ICP-MS laboratory of KBSI. A series of parallel line scans was combined to image trace elemental maps. This method was superior to a mosaic of spot analyses because no interpolation between the spot analyses was necessary. Moreover, downhole fractionation was minimized in the method, and acquisition time was significantly saved. The 10-Hz repetition rate and 55-μm spot size of laser beam were used because the stability and intensity of ion signals as well as spatial resolution of images were best optimized at the condition (Fig. 3). Lines were scanned at 10 μm/s speed with 60-μm spacing between adjacent line scans. The common quadrupole and imaging parameters used in the study are listed in Table 2.

Table 1 The correlation between the chemical composition and color of agate bands
Full size table
Fig. 3
figure 3

Signal intensities and signal stabilities depending on the experimental conditions (repetition and spot size) during line scanning. NIST 612 glass was used for determining the optimal measurement conditions

Full size image
Table 2 LA-ICP-MS instrument settings
Full size table

NIST612 SRM glass was used as the external standard and 29Si as the internal standard assuming stoichiometric quartz. The oxidation rate was monitored using the ThO+/Th+ ratios of the external standard and was maintained below 1 % during the analyses. The data reduction and image processing of chemical data were conducted with Iolite software (Hellstrom et al. 2008; Ulrich et al. 2009; Paton et al. 2011). The main advantage of this software is its capability to render line scans into quantitative element distribution images. Outlier rejection was applied to each line scan based on a 2σ threshold. The dwell times for each trace element were 10 ms per analyte.

The accuracy of line analyses was estimated against two USGS microchemical standard materials (GSD-1G and BHVO-2G), using the above analytical conditions. The former reference material has a relatively constant concentration level (~54 ppm) for all trace elements, whereas the latter has a range of concentrations of trace elements from 0.3 up to 163 ppm. Concentrations of 20 trace elements were measured within 20 % error estimates, with respect to the reference values of GSD-1G and BHVO-2G (Fig. 4). In the case of GSD-1G, the RSD values were calculated within 5 % in almost all the elements. For BHVO-2G, however, the RSD values has increased with decreasing concentrations of the trace elements: the RSD values less than 5.5 % for trace elements over 10 ppm, the RSD range of 5.9~8.9 % for trace elements between 4 and 10 ppm, and the RSD range of 10~20 % for trace elements less than 4 ppm. The coefficient of determination (R 2) for a linear relationship between the measured and reference concentrations of GSD-1G and BHVO-2G was calculated as 0.811 and 0.9993, respectively.

Fig. 4
figure 4

Accuracy of line analyses of LA-ICP-MS in the study. USGS RMs (GSD-1G, BHVO-2G) were used for estimating the accuracy of line analyses. Vertical bars with squares are error ranges of the reference materials and those with circles are measured uncertainties

Full size image

Results

Colorful agate

Aluminum and Fe were the most abundant trace elements in the colorful agate sample, up to 5000 and 4000 ppm, respectively, whereas Cu, Mn, and Co were minor, less than 200, 3, and 5 ppm, respectively. The heterogeneous distribution of trace elements in the sample was shown in four quantitative maps of Al, Fe, Cu, and Mn (Fig. 5). Aluminum concentration was particularly high in the outer half characterized by grayish color (Fig. 5a, b). The element was also concentrated at narrow bands and small spots in the middle part although it was generally absent in the inner half. On the other hand, several Fe-rich bands as well as many Fe-rich small spots were observed in the inner half (Fig. 5a, c). The distribution of Cu and Mn in the sample was similar to that of Al although these elements were concentrated mainly as small spots (Fig. 5d, e). In the X-ray maps, only Al and Fe showed the heterogeneous elemental distribution of colorful agate sample (Fig. 5f, g). Their distribution patterns resembled those of equivalent quantitative maps, but the boundaries of narrow bands and the overall background levels were blurred and increased, respectively.

Fig. 5
figure 5

The optical image, quantitative maps, and qualitative X-ray maps of the colorful agate sample. a Photomicrograph of ablated area. b–e LA-ICP-MS quantitative maps of Al, Fe, Cu, and Mn. Scale bars represent concentrations of each element in parts per million. f, g Qualitative X-ray maps of Al and Fe. U.I. unidentified material filling cavity

Full size image

Blue agate

Aluminum and Fe were also the most abundant trace elements in the blue agate sample, up to 400 and 100 ppm, respectively, whereas Mn and Ti were minor, less than 15 and 1 ppm, respectively. Heterogeneous elemental distribution was observed in the Al, Fe, and Ti quantitative maps of the blue agate sample (Fig. 6). All the elements were concentrated as narrow bands oscillating their concentrations (Fig. 6b–d). The oscillatory zoning pattern of Al was consistent with that of Ti, but was antithetic to that of Fe (Fig. 6e). In the X-ray map, only Al showed the heterogeneous elemental distribution of the blue agate sample (Fig. 6f). Although the resolution of image was much poorer, its distribution pattern resembled that of equivalent quantitative map particularly in the outer part of the sample, but not in the inner part (Fig. 6b, f).

Fig. 6
figure 6

The optical image, quantitative maps, and qualitative X-ray maps of the blue agate sample. a Photomicrograph of ablated area. b–d LA-ICP-MS quantitative maps of Al, Fe, and Ti. Scale bars represent concentrations of each element in parts per million. e Composite quantitative maps with two reference white lines. f Qualitative X-ray map of Al

Full size image

Discussion

Aluminum was the most abundant trace element in both agate samples and also showed the widest variation in concentration ranging from several thousands of parts per million to hundreds of parts per million, with sharp boundaries between the compositionally distinct bands (Figs. 5b and 6b). The compositional variation and rhythmic banding of Al in the samples are typical for silica precipitation at low temperatures below 300 °C (Rusk et al. 2011). Agates commonly contain considerable amounts of mineral and fluid inclusions possibly responsible for their different coloration (Götze et al. 2001). The quantitative Fe and Mn elemental maps and microscopic images of the colorful agate sample clearly showed spot-shaped numeral mineral inclusions definitely changing from Fe oxide/hydroxide in the inner half to Mn-bearing hollandite in the outer half (Fig. 5c, e). The distribution of Fe ions was highlighted in either Fe oxide/hydroxide inclusions or chalcedony structure as a substitute for Si4+ particularly in the inner half (Fig. 5c), whereas that of Mn ions was mainly restricted to hollandite inclusions in the outer half (Fig. 5e). The distribution patterns indicate that the valence state of Mn remained tri- and tetra-valent during the silica precipitation of outer half and further suggest that redox conditions and/or fluid compositions might change significantly between two distinct growth zones (Barnaby and Rimstidt 1989; Götze et al. 2012). The change is also reflected by the different coloration of two growth zones in the colorful agate sample from pale orange-red brown to gray in combination with Cu and Co (Table 1; Fig. 5). On the other hand, the blue agate sample does not show conspicuous mineral inclusions. Aluminum, Fe, and Ti ions were distributed in the sample as a replacement for Si4+. The oscillatory zoning patterns of Al, Fe, and Ti were apparent in the sample (Fig. 6b–d). Such fluctuations of trace element concentrations might be associated with abundance of microcrystalline quartz (Götze et al. 2001). The partitioning of trace elements into chalcedony stronger than that into microcrystalline quartz might result in the concentration of transition metals at bluish bands.

The quantitative maps of trace elements in two agate samples show several lines of advantages on the LA-ICP-MS mapping technique. The quantitative Mn and Ti maps of colorful and blue agate samples, respectively, suggest that the compositional zonation of trace elements can be observed with their concentrations down to several parts per million levels (Figs. 5e and 6d). Analytical time spent for the quantitative mapping was ca. 18 h. Moggi-Cecchi et al. (2002) suggested that long counting times and special operating conditions are required to obtain X-ray distribution maps for trace elements below 1000 ppm. The maps can further be semi-quantified by electron microprobe spot analyses. Rusk et al. (2008) lowered the detection limits of semi-quantitative maps of trace elements down to several tens of parts per million levels, but the amount of time spent for the mapping was ca. 90 h on the 500 × 500 pixel map size. The qualitative Al Kα and Fe Kα X-ray maps of the colorful agate sample have reproduced the distribution patterns of trace elements similar to their quantitative maps (Fig. 5b, c, f, g), but the reproduction failed in other qualitative maps. Based upon the qualitative Al Kα X-ray map of the blue agate, the detection limits of qualitative mapping in the study were around 100–200 ppm (Fig. 6b, f), and it took 15 h for a common electron microprobe equipped with four WDS to map the distribution of eight trace elements in the blue agate sample with 500 × 500 pixel resolution. The potential limitations of laser ablation imaging that have commonly been mentioned lie in (1) its destructive nature with respect to other microsampling methods (Woodhead et al. 2007) and (2) its relatively large beam size resulting in relatively low spatial resolution (Ulrich et al. 2009). The ablated depth of agates has been minimized to ca. 5 μm in the study by a series of parallel line scans. It is also noted that the presence and distribution of ~10-μm-long mineral inclusions were well identified in the quantitative maps (Fig. 5c, e) although their size is much smaller than the beam size of laser ablation.

Conclusions

The quantitative distribution maps of trace elements in two agate samples were acquired by a series of equally long line scan and a subsequent merging of line scans with Iolite software. These maps have particularly shown the fluctuations and concentrations of trace elements in the agate samples which cannot be visualized by conventional spot analysis. The colorful agate sample showed a variation of trace elements from Fe-rich inner half to Mn-rich outer half. The quantitative maps of the blue agate sample showed a correlative distribution relationship between Al and Ti, but an antithetic distribution relationship between Al and Fe. The distribution patterns resulted mainly from alternating microcrystalline quartz-rich and chalcedony bands. The detection limits of quantitative maps in the study were estimated as several parts per million levels with the RSD values of ca. 10 %, whereas those of X-ray qualitative maps as 100–200 ppm. These results clearly demonstrate that LA-ICP-MS can provide valuable data for interpretations of compositional variations in association with microstructures, chemical/physical phase boundaries, and geological and environmental heterogeneous samples.