Analysis of the morphology, composition, and spacing of exsolution lamellae can provide essential information to understand geological processes related to volcanism and metamorphism. We present a novel application of APT for the characterization of such lamellae and compare this approach to the conventional use of electron microscopy and FIB-based nanotomography.
Element diffusion rates and temperature/pressure control a range of fundamental volcanic and metamorphic processes. Such processes are often recorded in lamellae exsolved from host mineral phases. Thus, the analysis of the orientation, size, morphology, composition and spacing of exsolution lamellae is an area of active research in the geosciences. The conventional study of these lamellae has been conducted by scanning electron microscopy (SEM) and transmission electron microscopy (TEM), and more recently with focused ion beam (FIB)-based nanotomography, yet with limited chemical information. Here, we explore the use of atom probe tomography (APT) for the nanoscale analysis of ilmenite exsolution lamellae in igneous titanomagnetite from ash deposits erupted from the active Soufrière Hills Volcano (Montserrat, British West Indies). APT allows the precise calculation of interlamellar spacings (14–29 ± 2 nm) and reveals smooth diffusion profiles with no sharp phase boundaries during the exchange of Fe and Ti/O between the exsolved lamellae and the host crystal. Our results suggest that this novel approach permits nanoscale measurements of lamellae composition and interlamellar spacing that may provide a means to estimate the lava dome temperatures necessary to model extrusion rates and lava dome failure, both of which play a key role in volcanic hazard mitigation efforts.
The study of chemical mineralogy has been a major source of information within the field of Earth Sciences for more than a century, as minerals actively record geological processes during and after their crystallization. Physio-chemical conditions of these processes, such as temperature changes during volcanism and metamorphism, are recorded during mineral nucleation and growth in the form of chemical zonation, striations, and lamellae, among others. Exsolution lamellae form when a phase unmixes into two separate phases in the solid state. The analysis of the orientation, size, morphology, and spacing of such exsolution lamellae can provide essential information to understand temperature and pressure changes during volcanism and metamorphism1,2,3 and the formation of ore mineral deposits4.
Traditionally, the study of exsolution lamellae was conducted with the observation of micrographs by simple scanning electron imaging5. More recently, this has been substituted by the use of energy-filtered transmission electron microscopy (TEM) providing detailed observations at the nanoscale level1,2,3. Nevertheless, in both cases, the observations are made in two dimensions (2D), which is not fully adequate for three dimensional (3D) structures represented by these exsolution lamellae. Nanotomography6 is emerging as a new technique for the 3D observation of nanoscale features inside minerals grains, yet there is insufficient information about the composition of these features. An alternative to these approaches is the use of atom probe tomography (APT), representing the highest spatial resolution analytical technique in existence for the characterization of materials7. The strength of the technique lies in the possibility of combining a 3D reconstruction of nanoscale features with their chemical composition at the atomic scale with a near part-per-million analytical sensitivity7. Previous applications of APT to the analysis of geological samples have provided excellent results8,9,10,11, particularly in the chemical characterization of element diffusion and concentrations9,12,13. Yet, this application has not been used for the study of exsolution lamellae, abundant in some minerals hosted in metamorphic and igneous rocks. Here, we explore the use of APT, and its limitations, for the analysis of the size and composition of exsolution lamellae, and interlamellar spacing in volcanic titanomagnetite crystals.
1. Sourcing, selection, and preparation of mineral grains
NOTE: Samples were obtained from the catalogued collection at the Montserrat Volcano Observatory (MVO) and derived from falling deposits originating from a vigorous ash venting episode at Soufrière Hills Volcano that occurred on 5 October, 2009; this was one of 13 similar events over a course of three days14. This ash venting preceded a new phase of lava dome growth (phase 5) that commenced on 9 October. Previous analysis of this sample showed it to be a combination of dense dome rock fragments, glassy particles, and accidental lithics14.
Figure 1: Example of magnetite-rich ash grains from venting episodes at the Soufrière Hills volcano. (a, b): Backscattered electron images (BSE) of both reacted and unreacted textures in magnetite grains. (c) BSE image of a polished magnetite grain showing the presence of exsolution lamellae (light grey laths; red arrows) of potential ilmenite composition. (d) Secondary electron image of a polished magnetite grain prepared for atom probe tomography (APT) analysis, showing the location of some exsolution lamellae (dashed red lines), which are distributed all along the grain surface, and the location of the wedge extraction (blue arrow). Please click here to view a larger version of this figure.
2. Atom probe tomography (APT) sample preparation
Figure 2: Example of FIB-SEM sample preparation protocol for APT analysis. (a) Wedge (W) lift-out extraction with the nanomanipulator (Nm). (b) Lateral view of the micro-coupon array of silicon posts mounted on a copper clip. (c) Top view of the micro-coupon array of silicon posts showing the nanomanipulator for mounting the wedge sections. (d) Wedge fragment (S), showing a portion of the protective platinum cap (Ptc), mounted on a silicon post after welding with platinum (Ptw). Please click here to view a larger version of this figure.
Figure 3: Example of tips prepared for APT analysis. (Left) Image of tip after the first stage of sharpening. (Right) Image of the same tip after low kV cleaning, indicating the tip radius (67.17 nm) and the shank angle (26°). Please click here to view a larger version of this figure.
3. APT data acquisition
Specimen | 207 | 217 | 218 | 219 |
Sample Description | SHV Magnetite | SHV Magnetite | SHV Magnetite | SHV Magnetite |
Instrument Model | LEAP 5000 XS | LEAP 5000 XS | LEAP 5000 XS | LEAP 5000 XS |
Instrument Settings | ||||
Laser Wavelength | 355 nm | 355 nm | 355 nm | 355 nm |
Laser Pulse Rate | 60 pJ | 30 pJ | 30 pJ | 30 pJ |
Laser Pulse Energy | 500 kHz | 500 kHz | 500 kHz | 500 kHz |
Evaporation Control | Detection Rate | Detection Rate | Detection Rate | Detection Rate |
Target Detection Rate (%) | 0.5 | 0.5 | 0.5 | 0.5 |
Nominal Flight Path (mm) | 100 | 100 | 100 | 100 |
Temperature (K) | 50 | 50 | 50 | 50 |
Pressure (Torr) | 5.7×10-11 | 6.0×10-11 | 6.1×10-11 | 6.1×10-11 |
ToF offset, to (ns) | 279.94 | 279.94 | 279.94 | 279.94 |
Data Analysis | ||||
Software | IVAS 3.6.12 | IVAS 3.6.12 | IVAS 3.6.12 | IVAS 3.6.12 |
Total Ions: | 26,189,967 | 92,045,430 | 40,013,656 | 40,016,543 |
Single | 15,941,806 | 55,999,564 | 24,312,784 | 23,965,867 |
Multiple | 9,985,564 | 35,294,528 | 15,331,670 | 15,716,119 |
Partial | 262,597 | 751,338 | 369,202 | 334,557 |
Reconstructed Ions: | 25,173,742 | 89,915,256 | 38,415,309 | 39,120,141 |
Ranged | 16,053,253 | 61,820,803 | 25,859,574 | 26,598,745 |
Unranged | 9,120,489 | 28,094,453 | 12,555,735 | 12,521,396 |
Background (ppm/nsec) | 12 | 12 | 12 | 12 |
Reconstruction | ||||
Final tip state | Fractured | Fractured | Fractured | Fractured |
Pre-/Post-analysis Imaging | SEM/n.a. | SEM/n.a. | SEM/n.a. | SEM/n.a. |
Radius Evolution Model | “voltage” | “voltage” | “voltage” | “voltage” |
Vinitial; Vfinal | 2205 V; 6413 V | 2361 V; 7083 V | 2198 V; 6154 V | 2356 V; 6902 V |
Table 1. Atom probe tomography data acquisition settings and run summary.
4. APT data processing
Figure 4: Example of a representative APT mass-to-charge spectrum. Spectrum for the analyzed magnetite crystal with individual ranged peaks showing examples of the identification of peaks corresponding to single elements (e.g., oxygen (O) or iron (Fe)) or molecules (e.g., FeO). Please click here to view a larger version of this figure.
Specimen | 207 | 217 | 218 | 219 | ||||||||
Element | Atom count | Atomic % | 1s error | Atom count | Atomic % | 1s error | Atom count | Atomic % | 1s error | Atom count | Atomic % | 1s error |
O | 9459276 | 40.263 | 0.0155 | 36679256 | 40.724 | 0.0080 | 15396155 | 41.010 | 0.0124 | 16212281 | 41.224 | 0.0122 |
Fe | 9424298 | 40.114 | 0.0155 | 35948593 | 39.913 | 0.0079 | 14829905 | 39.502 | 0.0121 | 15006853 | 38.159 | 0.0116 |
Mn | 15954 | 0.068 | 0.0005 | 72884 | 0.081 | 0.0003 | 28166 | 0.075 | 0.0004 | 31450 | 0.080 | 0.0005 |
Mg | 123755 | 0.527 | 0.0015 | 486732 | 0.540 | 0.0008 | 203596 | 0.542 | 0.0012 | 234231 | 0.596 | 0.0012 |
Al | 85598 | 0.364 | 0.0013 | 329602 | 0.366 | 0.0006 | 134637 | 0.359 | 0.0010 | 154779 | 0.394 | 0.0010 |
Si | 13855 | 0.059 | 0.0005 | 39307 | 0.044 | 0.0002 | 16278 | 0.043 | 0.0003 | 25750 | 0.065 | 0.0004 |
Na | 166 | 0.001 | 0.0001 | 1254 | 0.001 | 0.0000 | 447 | 0.001 | 0.0001 | 1468 | 0.004 | 0.0001 |
Ti | 4360052 | 18.558 | 0.0097 | 16478946 | 18.296 | 0.0049 | 6920481 | 18.434 | 0.0076 | 7645849 | 19.442 | 0.0077 |
H | 10657 | 0.045 | 0.0004 | 30522 | 0.034 | 0.0002 | 12899 | 0.034 | 0.0003 | 14478 | 0.037 | 0.0003 |
Total | 23493611 | 100.00 | 0.04 | 90067097 | 100.00 | 0.02 | 37542563 | 100.00 | 0.04 | 39327140 | 100.00 | 0.03 |
Fe+Ti+O | 98.94 | 98.93 | 98.95 | 98.82 | ||||||||
Fe/Ti | 2.16 | 2.18 | 2.14 | 1.96 |
Table 2. Atom probe tomography bulk compositional data for all analyzed specimens.
Like many titanomagnetite crystals from various stages of the Soufrière Hills Volcano (SHV) eruption, the crystal analyzed here contains exsolution lamellae <10 µm in thickness, visible in secondary SEM images (Figure 1d), which separate zones of Ti-rich magnetite, indicating a C2 stage of oxidation18. Based upon the SEM images, spacing between these lamellae ranges from 2 to 6 µm (n = 15). Four titanomagnetite specimen tips, referred as 207, 217, 218, and 219, were successfully extracted from this single crystal and analyzed by APT (Figure 5). Two of the specimens (207 and 218) showed homogenous concentrations of both Fe and Ti throughout (Figure 5a), indicating that lamellae were not intersected. The other two specimens (217 and 219) showed zones with variable concentrations in Fe, O, and Ti (Figure 5b-e). These features are parallel to each other and have tapered terminations, indicative of trellis ilmenite18. Specimen 219 contains a larger proportion of the intersected lamellae than specimen 217. 3D reconstructions of the APT data (Figure 5c-e, Movie 1) permit a precise measurement of the interlamellar spacing (λ) and provide length scales that average 29 nm for specimen 219 (n = 30) and 14 nm for specimen 217 (n = 15), with a 1s value of 2 nm for both. In addition to these measurements, APT permits extraction of chemical information across these lamellae at high spatial resolution (nanoscale) through the analysis of proxigrams, taking the point 0 as the intersection between the lamella and the host mineral (Figure 6). Proxigram diffusion profiles through these zones are smooth. Atomic concentrations of Ti in the crystal, 17% in specimen 217 and 16.5% in specimen 219 (Figure 6), confirm that it is a titanomagnetite and are consistent with previous petrologic analyses of SHV eruptive products18. These proxigrams also confirm that the composition of lamellae matches that of ilmenite (Figure 6).
Figure 5: Example of 3D atomic reconstructions of magnetite tips analyzed (z values in nm). (a) Specimen 218, showing homogenous concentrations of O, Fe, and Ti. (b) Specimen 219, displaying regions depleted in Fe and enriched in O and Ti (lighter areas). (c) Elemental isoconcentration surfaces of specimen 219, showing intralamellar concentrations of Fe < 30.0% in blue, Ti > 25.0% in yellow, and O < 51.5% in purple. (d) View of same area in panel c, but rotated 90° to reveal the measured lamellar spacing (l). (e) Specimen 217, showing a relatively smaller portion of exsolution lamellae in the lower left (intralamellar concentrations: Fe < 19.0%, Ti > 33.0%, O < 49.0%). Please click here to view a larger version of this figure.
Figure 6: Proxigrams of APT data showing atomic concentrations of O, Fe and Ti in specimens 217 and 219, respectively. A distance of zero on the x-axis (dashed vertical line) represents the boundary between the titanomagnetite on the left and the ilmenite lamellae on the right. Across this boundary, Ti concentrations increase from 17.0% to 44.0% and from 16.5% to 42.5% in specimens 217 and 219, respectively. Fe contents decrease from 37.7% to 0.8% and from 38.5% to 0.6% in specimens 217 and 219, respectively. O concentrations increase from 40.0% to 50.2% and from 38.8% to 50.6% in specimens 217 and 219, respectively. Ti and O diffuse toward the lamellae (to the right) and Fe is diffusing away from the lamellae (to the left). Error in all measurements is < 2 atomic %. Please click here to view a larger version of this figure.
Movie 1. 3D elemental isoconcentration reconstruction of specimen 219, showing intralamellar concentrations of Fe < 30.0% in blue, Ti > 25.0% in yellow, and O < 51.5% in purple, and rotated 360° along the longitudinal (z) axis. Please click here to download this file.
3D APT data reconstructions allow a precise measurement of the interlamellar spacing in the analyzed crystal at a resolution three orders of magnitude higher than those measured from conventional SEM images. This indicates that atomic variations in chemistry occur over a spatial extent three orders of magnitude smaller than optically observable mineralogical changes. Also, the measured interlamellar distances (29 nm and 14 nm), are consistent with the length scale for oxyexsolution as opposed to that for nucleation and growth of a separate phase, the latter of which occurs over length scales an order of magnitude greater19,20. Oxyexsolution of the titanomagnetite examined here may have resulted from either heating of the magma within the chamber due to mingling events, or oxidation induced by atmospheric exposure during ascent through the conduit and emplacement at the vent. The titanomagnetite crystal used in this study displays numerous lamellae on the order of 10 µm or less in diameter, indicating sufficient time for partial exsolution of the entire 0.06 mm2 crystal. That observation, in conjunction with the ash-venting origin of the sample, suggests that the lamellae were formed from oxidation of the lava dome following emplacement at the vent.
Figure 7: Schematic representation of different possibilities for capturing exsolution lamellae in APT tips, and the implications for measuring lamellar spacing (λ). Please click here to view a larger version of this figure.
These measurements of the interlamellar spacing were only possible in two of the four tips analyzed and this sampling bias could induce error. As we only analyze nanoscale volumes of the crystals, each tip could capture partial representation of the distribution of lamellae, potentially leading to an erroneous calculation of interlamellar spacing (λ) (Figure 7). Nevertheless, with the analysis of sufficient tips and the 3D tomography capabilities of APT, our approach provides a novel method to precisely measure λ, which can be fundamental to understand a variety of igneous and metamorphic processes. For example, analysis of Fe-Ti diffusion in magnetite crystals has been used to characterize magma mingling events and changes in magma temperature at Soufrière Hills Volcano19. The interdiffusion rate of Fe and Ti could be modeled from accurate measurements of the interlamellar spacing if the timescale of diffusion can be constrained. Previous studies have utilized exsolution lamellae to deduce cooling rates in igneous intrusions21,22, and the novel approach of using APT can potentially be utilized to constrain element interdiffusion and improve calculations of magma and lava dome temperatures in active volcanic systems23,24. Besides the precise measurement of 3D structures at nanoscale, APT provides chemical information at the atomic scale for the same spatial resolution. This has resulted in demonstrating that the transition from titanomagnetite to ilmenite is gradual and smooth (Figure 6). This is in contrast to previous studies suggesting sharp and distinct contacts between the exsolving lamellae and the host phase, based on microscopy observations (SEM and TEM) without the chemical information. Consequently, only the addition of APT can accurately provide the geochemical characteristics of these phase transitions.
Atom probe tomography (APT) is still an emerging technique for geological applications25, yet our case study demonstrates the usefulness of applying this approach for the study of exsolution lamellae, quite common in minerals hosted in igneous and metamorphic rocks. In particular for understanding volcanic processes, future studies can examine multiple samples with exsolution lamellae throughout lava domes (derived from either ash venting episodes or dome collapse events) in order to better constrain variations in temperature during eruptive episodes with calculated diffusion rates.
The authors have nothing to disclose.
This work was supported by funding from the National Science Foundation (NSF) through grants EAR-1560779 and EAR-1647012, the Office of the VP for Research and Economic Development, the College of Arts and Sciences, and the Department of Geological Sciences. Authors also acknowledge Chiara Cappelli, Rich Martens and Johnny Goodwin for technical assistance and the Montserrat Volcano Observatory for providing the ash samples.
InTouchScope Secondary Electron Microscope (SEM) | JEOL | JSM-6010PLUS/LA | |
Focus Ion Beam (FIB) Secondary Electron Microscope (SEM) | TESCAN | LYRA XMU | |
Local Electrode Atom Probe (LEAP) | CAMECA | 5000 XS | |
Integrated Visualization and Analysis Software (IVAS, version 3.6.12). | processing software |