Although sintered-diamond (SD) anvils can generate very high pressures, it is still useful to generate much higher pressures than usual multi-anvil press can (25 GPa) using carbide (WC) anvils. One reason is high electrical resistivity of SD, which makes it difficult to use anvils for heater electrode, and as a result, makes heating difficult. Another reason is that the sample chambers after compression by SD anvils are much smaller than those by WC anvils due to the higher hardness of SD than WC. For these reasons, we have made efforts to expand the pressure range that can be generated using a multi-anvil press with WC anvils. The ultrahigh-pressure generation by a multi-anvil press using carbide anvils is explained by Ishii et al. , and its essence is given below.
There are the following key points for generation of ultrahigh pressures by multi-anvil presses.
In Kawai-type multi-anvil presses, which can generate the highest pressures among various kinds of multi-anvil presses, an octahedral pressure medium is compressed by eight inner anvils whose one corner is truncated, and the assemblage made of the inner anvils with the pressure medium is compressed by six outer anvils. In order to generate ultrahigh pressures, the compression should be balanced, and therefore, the space made by the outer anvils must remain exactly cubic at any press load. In many Kawai-type apparatus including the Walker modules, 3 and 3 outer anvils are supported by the upper and lower guide blocks, which cause rhombohedral distortion. Therefore, we adopt the Osugi-type (DIA-type) guide block systems, in which an upper and lower outer anvils are fixed in the upper and lower guide block, respectively, and the other four outer anvils are individually placed on four sliding wedges that move ahead with increasing press load by 45-degree slopes in the guide blocks (Fig. 1). This system prevents rhombohedral distortion.
Fig. 1. Schematic drawings of the Osugi (DIA)-type compression system. A uniaxial press compresses guide blocks and finally creates a cubic compression space surrounded by six first-stage anvils, each of which compresses the cubic space in the  directions. (From Ishii et al. )
We installed a Kawai-type multi-anvil press with the Osugi-type guide block system in the Bayerisches Geoinstitut, University of Bayreuth, called "IRIS-15" (Fig. 2).
Fig. 2A. Overview of IRIS-15. (1) load control system, (2) lower guide block, (3) upper guide block, (4) heating system, (5) outer (first-stage) anvils, (6) Teflon sheet and (7) the sliding wedges placed on the 45° slope.
Fig. 2B. The lower guide block with two sliding wedges of IRIS-15. The other two sliding wedges are not placed to show the bottom outer anvil.
However, the space compressed by the outer anvils are tetrahedrally distorted with changing press load, because the supporting strengths to the upper and lower anvils are different from the horizontal four anvils. Therefore, we weakened the strength to support the upper and lower anvils so that the compression space shrinks with keeping an exact cubic shape with increasing press load.
The hardness of inner anvils are of course directly related to the maximum pressure. Engineering development of carbide by some companies allows manufacturing carbide anvils that can generate over 30 GPa. The below is one example.
Fig. 3. Comparison of generated pressures using carbide with hardness HRC = 95.5 (blue, Hv = 2200) and HRC = 94.0 (red, Hv = 1800) with truncation 2.0 mm at ambient teperature. The pressure generation by anvils with HRC = 94.0 has a limit of 26 GPa. On the other hand, anvils with HRC = 95.5 can generate 32 GPa and could be more if more press load had been applied.
Even though hard carbide is used for anvil material, yielding of anvils around the truncation anyway limits pressure generation. To reduce the yielding, three anvil faces around a truncation of inner anvils were inclined typically by 1.0°(Fig. 4). This is essentially the same technology as bevel processing in DACs, by which pressures greater than 200 GPa were first achieved in a DAC. Fig. 5 demonstrates the effect of anvil tapering on ultrahigh-pressure generation.
(Above) Fig. 4. Explanation of anvil tapering.
(Left) Fig. 5. Comparison of high-pressure generation using flat (red) and tapered (blue) anvils with truncation of 1.5 mm. The flat anvils generate pressures to 32 GPa, whereas the tapered anvil generated pressures to 41 GPa at a press load of 15 GPa.
Even though ultrahigh pressure is generated by cold compression, a significant part of generated pressure is lost by heating to mantle temperatures (Fig. 6). The reasons for this pressure drop may be (1) decrease in anvil hardness, (2) softening of gasket and pressure medium, (3) sintering of sample and pressure medium, and (5) volume decrease upon sample phase transition. In order to suppress the pressure drop by the possible reasons (1) to (4), very effective thermal insulation is indispensable. In order to suppress the pressure drop by the reasons (4) and (5), we pre-cooked the sample to convert it to an aggregate of medium-high pressre (20 GPa) and medium-high temperature (1200 K) phase. Fig. 7 demonstrate such examples of ultrahigh-pressure generation at mantle temperature (2000 K).
Fig. 6. An example of pressure drop by heating. Although ultrahigh pressure of 63 GPa was generated using carbide with very high hardness (Hv = 2700) at a temperature of 300 K, it decreased to 48 GPa by heating to 2000 K.
Fig. 7. An example of ultrahigh-pressure generation at a high temperature of 2000 K. The generated pressure was estimated by Al2O3 content in bridgamanite coexisting with corundum. The starting material was pre-converted to sintered aggregate of aluminous akimotoite. The Re heater was surrounded by LaCrO3 thermal insulator. The highest pressure at 2000 K was even higher than that at 300 K at a press load of 15 MN.
Thus, we can generate pressures to 48 GPa at a temperature of 2000 K using a multi-anvil press with carbide anvils. Because experimental costs are very high due to easy breakage of anvils, however, generation of 48 GPa at 2000 K is impractical. Even generation of 40 GPa requires high costs, and pressures to 35 GPa may be the limit for routine works. Therefore we are making efforts to decrease anvil costs in this kind of experiments.