Magnetic inclusions in single silicate crystals

Tinghong Zhou

University of Rochester, Department of Earth and Environmental Sciences

1. Introduction

Most paleomagnetic researches are based on the characteristic remanence carried by rocks. However, during the long period of evolution, the original characteristic remanence carried by grains with short relaxation time might be altered due to thermal or chemical events. Single silicate grains often contain minute magnetic inclusions with ideal recording characteristics having relaxation times on billion-year time scales (Feinberg et al., 2005; Tarduno et al., 2006). These magnetic inclusions, with size of hundreds nm to μm, provide a possible way to explore primary information about the ancient magnetic field. To determine the structure and component of these magnetic inclusions, Scanning Electron Microscopy (SEM) research is necessary.

The aim of this project is to recognize the magnetic inclusions and their distributions in clinopyroxene crystals from a certain time in geological history, as well as define the compositions of them. Two clinopyroxene samples were used in this project, which are named cpx01 and cpx02 respectively. Technics supporting this project include sample polish, light microscopy, evaporation coating, backscattered electron imaging, secondary electron imaging, X-ray analysis, image analysis, and interaction volume modeling. Most contexts reported here are based on cpx01.

2. Sample Preparation

Before SEM study, sample preparation is required for the clinopyroxene grains. The purpose of this project requires sample have a relatively flat surface. Two steps of polish were conducted for the sample. The first step was putting the sample in epoxy with smooth surface. The second step was using the colloidal silica to polish the grain again. On the other hand, since clinopyroxene is insulated, coating is demanded for the sample to make it conducted. By using LADD Vacuum Evaporator, ~ 5 nm thick carbon layer was coated on surface of sample.

Figure 1 Two prepared sample for SEM

3. Light Microscopy

Light microscope is very helpful to catch the general characteristics of the sample and find target areas before we conduct any elaborate SEM work. Figs 2a-e shows light microscope images under differential interference contrast (DIC) mode. This mode is very sensitive to the variation of topography. Two kinds of needles in perpendicular directions are recognized. The smaller ones are parallel to the long axis of the grain (Fig 2d), while the larger one is vertical to the long axis (Fig 2d).

a) 5X                                                                                                        b) 5X

c)10X                                                                                                     d) 50X

Fig 2 Light microscope images with different magnifications for cpx01. All images are under the DIC mode. a) is after putting in epoxy but before colloidal silicate polishing; b), c), d) are after colloidal silicate polishing. a) 5X; b) 5X; c)10X; d) 50X.

4. Backscattered Electron Microscopy

Contrast of backscattered electron (BSD) image is mainly caused by various elements with different atomic numbers, despite edge effect still exists. In BSD image, areas with heavier element will be brighter. Fig 3 shows BSD images for cpx01. Orientations of the sample are not exactly same due to repeated experiments, but they are similar.

Dark triangle area in the right top of grain indicate it may have different chemical composition (Fig 3a). Two kinds of bright inclusions are recognized in BSD images -- arbitrary shaped inclusions (Fig 3b, c), and needle shaped inclusions (Fig 3d, e, f). The short linear holes in Fig 3e and small holes in Fig 3f might be caused by over polish, which damaged the original needles.

a)                                                                                                                  b)

c)                                                                                                                  d)

e)                                                                                                                 f)

Fig 3 BSD image for cpx01. a) is the whole crystal; b) and c) are arbitrary shaped inclusions; d), e), and f) are needle shaped inclusions.

5. Secondary Electron Microscopy

Secondary electron (SE) images, including SE2 and InLens images, are sensitive to variation of topography. They have better surface resolution than BSD images. InLens image even has better resolution when the working distance is short. Three SE images are shown in Fig 4. Fig 4a shows a needle which might be damaged by polish. Fig 4b, c present the pattern which is perpendicular to the long axis of cpx01 crystal.

a) Needle inclusion

b) Horizontal patterns

c) Horizontal pattern

Fig 4 SE images for cpx01. a) is needle inclusion. b) and c) show patterns perpendicular to the long axis of the crystal.

6. Energy-Dispersive X-ray Spectroscopy analysis

To determine the chemical component of different phases in the grain,energy-dispersive X-ray spectroscopy (EDS) analysis was conducted on the sample. Fig 5 shows the location of EDS spots and corresponding spectra. Spot 1 and spot 2 are on large, bright, arbitrary shaped inclusions; spot 3 is on bright needles; spot 4 is on the background. From these spectra, we can know that spot 1 and spot 2 are of similar chemical components, which have Ti/Fe ratio greater than 1. For spot 3, even though the interaction volume might beyond the range of the needle (see Discussion), combined with the background component shown by spot 4, it is highly possibly that its Ti/Fe ratio is less than 1, which is different from spot 1 and spot 2.

a) EDS result of Spot 1, which locates in arbitrary shaped inclusion

b) EDS result of Spot 2, which locates in arbitrary shaped inclusion

c) EDS result of Spot 3, which locates in needle shaped inclusion

d) EDS result of Spot 4, which locates in background

Fig 5 EDS spots (left column) and corresponding spectra (right column)

An EDS map was also measured for the cpx01 crystal (Fig 6). The crystal is enriched Ca, Fe, Mg, O, Si, and Al, which is common for a clinopyroxene crystal. The dark triangle area mentioned previously are rich in AI, O, Si, and lack Fe and Mg compared to other areas of the grain.

Fig 6 EDS map for cpx01

7. Image Analysis

One way to describe the distribution of magnetic inclusions in the grain is to estimate the spatial density of them. To do this, particle analysis was done for a BSD image of sample cpx02 (Fig 7). Results are shown in Table 1. Compared Fig 7a and Fig 7b, the density presented here is a conservative value.  All analyses were conducted using Image J software.

Fig 7 Particle analysis for cpx02. Left (a): BSD image for particle analysis; right (b): particle analysis in Image J

Slice Count Total Area (μm2) Density (count/μm2)
cpx02 30 1001.46 0.03

Table 1 The result of particle analysis for cpx02

8. Interaction volume modeling

To better understand SEM results, especially EDS results, interaction volume modeling was conducted. Fig 8 shows a two-layer model with a silicon dioxide substrate under 20 kV accelerating voltage. The top layer is 5 nm thick carbon. The second layer is 100 nm thick Fe2O3. A single spot on the surface actually reflects the information from a region of ~5*5*5μm3.

Fig 8 Interaction volume simulation


Two kinds of needles in orthogonal directions were recognized in light microscope images (Fig 2d). However, only one direction magnetic needle, which parallel to the long axis of the grain, was observed in BSD images (Fig 3d, e, f). SE images (Fig 4b, c) illustrate that the needles perpendicular to the long axis of the grain are tinny topography variations, rather than magnetic inclusions.

From BSD and SE images, magnetic mineral needles in cpx01 are usually of hundreds of nanometers width (Fig 3 and Fig 4). EDS analysis for such a narrow needle (Fig 5c) is hard to be accurate. The interaction volume tends to be much larger instead of being confined to the needle (Fig 8), which means the spectrum is a mix result of the needle and background.

10. Conclusion

In cpx01 crystal, two kinds of magnetic inclusions with different Ti, Fe concentration are found. One is magnetic mineral needle with low Ti/Fe ratio. The other is arbitrary shaped inclusion with high Ti/Fe ratio. By analyzing BSD and SE images, sizes and distributions of these inclusions are presented.


I would like to thank Brian McIntyre for his patient, kind and "magic" help all the time. I also want to thank my TA Nursah, my classmates Yaad and Martha, for helping me with the SEM classes and labs. Thank Paleomagnetic Research Group in University of Rochester, especially Josh, Rory, John and Richard, for helping me conduct this project.


Feinberg JM, Scott GR, Renne PR, Wenk HR. Exsolved magnetite inclusions in silicates: Features determining their remanence behavior. Geology. 2005 Jun 1;33(6):513-6.

Tarduno JA, Cottrell RD, Smirnov AV. The paleomagnetism of single silicate crystals: Recording geomagnetic field strength during mixed polarity intervals, superchrons, and inner core growth. Reviews of Geophysics. 2006 Mar 1;44(1).

Note: Some data and images presented here may have been processed in order to protect originality. For citation, please contact Paleomagnetic Research Group of University of Rochester. Thanks.



Please enter any comments, criticisms, questions, etc. below.

Your name:

Email address: