Motivation

Although the multi-anvil press is a reliable high-pressure-temperature apparatus, the conventional multi-anvil press has two disadvantages. First, we cannot directly see the sample under high pressure and high temperature. We do not know what changes are occurring with the sample. Second, we do not know the sample pressure. In conventional multi-anvil experiments, the sample pressure is calibrated against the press load by detecting several known high-pressure phase transitions. However, the sample pressure varies in a complicated and often unpredictable way by heating.  

In situ X-ray observation solves the above problems. In this method, highly brilliant, directional, and penetrative X-rays from synchrotron radiation reach the sample through the gasket and pressure medium. The X-ray signals generated by reacting with the sample are detected out of the press to provide real-time information about the sample. X-ray diffraction provides information about the sample's phase transition and chemical reaction. Particularly, the change in diffraction intensities of existing phases indicates the direction of the reaction. The radiograph by transmitted X-rays provides information about the sample geometry and dimension. By placing a pressure standard material next to the sample, we can estimate the sample pressure from the change in the unit cell volume of the pressure standard material together with the temperature measured by a thermocouple.

Energy-dispersive X-ray diffraction

In situ X-ray observation is a modern and relatively routine experimental technique for multi-anvil workers. Although there are many multi-anvil beamlines in synchrotron radiation facilities worldwide, we preferably use the following two facilities:

• Beamline P61B, PETRA-III, DESY, Germany
• Beamline BL04B1, SPring-8, Japan

We adopt the energy-dispersive X-ray diffraction (EDXD), where the incident beam is high-energy white X-rays, and diffracted X-rays are collected using a solid-state detector (SSD) at a fixed diffraction angle. The main advantage of EDXD is high counts against the strong absorption by the gaskets, pressure medium, and sample itself.

Figure 1. The concept of the energy-dispersive X-ray diffraction with a multi-anvil press.

We adopt in situ X-ray observation exclusively for determining phase relations although many other workers conduct rock and mineral deformation and viscosity measurement. Below, we explain our advanced experimental technique for phase-relation studies.

High-pressure cell assembly

Figure 2. Schematic cross-sections of the cell assembly. Two sintered samples and a sintered MgO pressure marker (P.M.) were set at the center of the cell assembly. Fo100 and Fo70 denote the samples with bulk compositions of Mg2SiO4 and (Mg0.7Fe0.3)2SiO4, respectively, for example.

Figure 3. Radiographic images of the sample part. (a) before compression, (b) after compression to 7 MN.

Precise pressure determination

 We usually adopt MgO as a pressure calibrant for the following reasons. (1) no phase occurs transitions in the Earth's interior. (2) The melting temperature is high, 1800℃ even at ambient pressure. (3) The crystal structure is simple (B1). Figure 4 shows a typical diffraction pattern of the MgO pressure calibrant. Except for very tiny diamond peaks, the diffraction pattern consists of MgO peaks only. Since we can read more than seven peaks, (111), (200), (220), (311), (222), (400), (331), (420), and (422), the precision in pressure determination is very high: 0.05 GPa at 25 GPa. This high precision enables us to determine phase relations reliably and in detail. 

Figure 4. A typical diffraction of MgO pressure calibrant. The right is the expansion of the red rectangle in the left figure. The peaks up to (422) can clearly be observed.

Challenge for determining phase relations

There are several challenges in determining phase boundaries by combining multi-anvil experiments with in situ X-ray observation. Let us consider the phase boundary of two high-pressure phases.

• Mantle minerals are inert. It is relatively easy to form a high-pressure phase from a starting material because of high-density defects during cold compression. However, the formation of one phase from the starting material does not provide any information about the stability of the two high-pressure phases. It just suggests the stability of the formed phase compared to the starting material. On the other hand, it isn't easy to convert from one high-pressure phase to another high-pressure phase because an in situ formed phase has low-defect density. High temperature does not solve this inertness or even strengthens it due to decreased defect density such as grain growth.
• Pressure changes upon heating. Pressure rapidly and then slowly decreases due to stress release and gasket and pressure-medium softening. In addition, pressure instantaneously increases when reaching a temperature due to thermal pressure. Thus it is difficult to identify "experimental pressure".
• The temperature of the reaction is also uncertain. Samples sometimes react before reaching the target temperature and remain without transforming to actual stable phases.

Solution for accurate determination of phase relations

To overcome the above challenges, we adopt the following strategy.

1. The starting material is the HP-phase pre-synthesized at the lowest temperature, where the HP-phase can be synthesized.
2. We load the starting material of the HP-phase in a multi-anvil cell, compress it using a multi-anvil press, and heat it to the lowest possible temperature where the HP-phase can transform to LP-phase to convert the HP-phase partially to LP-phase and obtain the coexistence of the HP- and LP-phases.
3. After the LP-phase becomes enough to observe using X-ray diffraction and to trace the relative intensity change of the phases, we cool the sample by several hundred K to stop the transition and bring it to higher pressures than the supposed phase boundary.
4. We heat the sample to the lowest temperature, where the HP-phase starts to become the LP-phase.
5. We take diffraction patterns of the pressure calibrant and sample alternatively by keeping the press load and temperature. The pressure spontaneously drops gradually. We observe the change in the ratio of the HP-phase to the LP-phaseat constant temperature and monotonically decreasing pressure. LP-phase should slowly transform into HP-phase at the beginning.
6. At some time, the HP-phase/LP-phase ratio should start to decrease because it enters the LP-phase stability field. Therefore, we bracket the phase boundary with the lowest pressure where the HP-phase/LP-phase ratio increases and the highest pressure where this ratio decreases.
7. We decrease the sample temperature by several hundred K to stop the proceeding of the transition, compress the sample to a pressure sufficiently above the phase boundary determined in the previous temperature condition, and heat it to a higher temperature by 50 or 100 K than in the previous stage (Step 5 in Fig. 4).
8. We repeat the procedure from Step (5) to (7) to bracket the phase boundary every 50 or 100 K until the phase transition does not proceed due to the long time annealing.

Figure 5. A schematic drawing explaining the novel strategy to determine a phase boundary accurately and precisely. The pressure spontaneously and gradually decreases while keeping the temperature and press load constant. In the HP- and LP-phase stability fields, HP- and LP-phases increase with time, respectively, enabling to bracket the phase boundary. The numbers in the circles indicate the steps in the novel strategy explained above.

Examples

Chanyshev, A., Ishii, T., Bondar, D, Bhat, S. Kim, E.-J., Farla, R., Nishida, K., Liu, Z., Wang, L., Nakajima, A., Yan, B., Tang, H., Chen, Z., Higo, Y., Tange, Y., Katsura, T., Depressed 660-km discontinuity caused by akimotoite-bridgmanite transition, Nature, 601, 69-73, 2022. 10.1038/s41586-021-04157-z

Ishii, T., Huang, R., Myhill, R., Fei, H., Koemets, I., Liu, Z., Maeda, F., Yuan, L., Wang, L., Druzhbin, D., Yamamoto, T., Bhat, S., Farla, R., Kawazoe, T., Tsujino, N., Kulik, E., Higo, Y., Tange, Y., Katsura, T., Sharp 660-km discontinuity controlled by extremely narrow binary post-spinel transition. Nature Geosci., 12, 869-872, 2019.  10.1038/s41561-019-0452-1

Ishii, T., Huang, R., Fei, H., Koemets, I., Liu, Z., Maeda, F., Yuan, L., Wang, L., Druzhbin, D., Yamamoto, T., Bhat, S., Farla, R., Kawazoe, T., Tsujino, N., Kulik, E., Higo, Y., Tange, Y., Katsura, T., Complete agreement of the post-spinel transition with the 660-km seismic discontinuity. Sci. Rep. 8, 6358, 2018. 10.1038/s41598-018-24832-y