Ultrahigh-pressure generation above 50 GPa in a Kawai-type large-volume press

High-pressure techniques in large-volume presses (LVPs) are widely used in physics, chemistry, materials science, and Earth and planetary science.^1–4 The attainment of higher pressures within a large sample volume facilitates the discovery of novel functional materials, desirable properties, and unprecedented phenomena.^5–7 Among LVPs, the Kawai type (KLVP) is one of the most user-friendly. It is capable of easily generating pressures exceeding 10 GPa within a substantial sample volume ranging from millimeters to centimeters under precisely controlled pressure and temperature conditions, owing to its straightforward double-stage compression system and the Kawai-type cell assembly, which consists of an octahedral pressure medium and eight truncated second-stage anvils.^8,9 The pressure generation capability of KLVPs using conventional anvils and commercial assemblies is generally limited to 25 GPa, which corresponds to the uppermost region of Earth’s lower mantle.¹⁰ Ultrahigh-pressure generation, particularly at high temperatures, in KLVPs remains challenging owing to the frequent occurrence of blowouts and uneven compression in the Kawai-type cell assembly resulting from elastic bending of the first-stage anvil and limitations imposed by the mechanical properties of the materials used.

Mechanical properties of second-stage anvils and pressure media, along with the geometric design of cell assemblies, are crucial for achieving ultrahigh pressures in KLVPs. Recent technical developments in anvil materials have allowed KLVPs to reach pressures above 30 GPa by employing hard tungsten carbide (WC) and sintered diamond (SD) second-stage anvils.^11,12 Although pressures above 30 GPa can be routinely generated using hard SD anvils with a hardness two times higher than that of WC anvils in KLVPs,^13,14 high-pressure and high-temperature (HPHT) experiments with SD anvils remain limited because of their high costs and the restricted edge lengths below 25 mm. Moreover, the heating efficiency of SD anvils is generally lower than that of WC anvils under high pressure because of the poorer electrical resistance of the former compared with the latter. The high hardness of SD anvils also makes them prone to blowouts, rendering them impractical for ultrahigh-pressure studies and the synthesis of large-sized samples under very HPHT conditions. By contrast, WC anvils are widely used in KLVPs owing to their practicality and ease of manufacture compared with SD anvils. The development of ultrahard TF05 (HV = 2200) and TJS01 (HV = 2700) WC anvils has enabled the generation of high pressures of 35 and 48 GPa, respectively, at room temperature in KLVPs equipped with a sophisticated Osugi-type guide block system.¹⁵ By tapering 1° on the surfaces of these ultrahard WC anvils, it was possible to reach much higher pressures of 45–50 GPa at high temperatures of 1600–2000 K.¹⁶ More recently, Hou and co-workers^17,18 have developed more practical ZK01F WC anvils (HV = 2200) and achieved ultrahigh pressures exceeding 40 GPa at 1900–2000 K in a KLVP with a Walker-type module. However, these ultrahigh-pressure experiments using either SD or ultrahard WC anvils generally have a limited sample volume of less than 1 mm in diameter. Furthermore, the ultrahard WC anvils often blow out at high press loads or during decompression, owing to the high stress concentrations near the truncation, which limits their use to one or two cycles.¹⁶ Therefore, to facilitate advances in high-pressure science and technology, it is necessary to develop practical and routine ultrahigh-pressure techniques at high temperatures within a large sample volume using practical WC anvils in KLVPs.

In this study, we expand the attainable pressure range in a millimeter-sized sample volume at both room temperature and high temperatures by introducing electrically conductive boron-doped polycrystalline diamonds and hard Al₂O₃ materials into conventional cell assemblies combined with practical WC anvils in a newly constructed KLVP. We systematically compare the efficiency of pressure generation with different types of cell assemblies and WC anvils. We also discuss the generation of ultrahigh pressures above 50 GPa in KLVPs and their potential applications in both science and technology.

A new KLVP (Precision Industry Revolution Equipment Technology Co., Ltd., China), with a maximum press load of 15 MN (Fig. 1), has been constructed and installed at the Synergetic Extreme Condition User Facility (SECUF) at Jilin University, China. Three of the first-stage steel anvils are embedded within the upper guide block, with the other three being arranged in the bottom guide block. The edge lengths of the first-stage hard steel anvils and second-stage WC anvils are 60 and 32 mm, respectively. Additionally, six hardened steel spacers of thickness 6 mm are placed on the top and bottom square surfaces of the first-stage anvils, allowing us to use second-stage anvils with an edge length of 25.4 mm. The uniaxial ram is operated using oil pressures up to 70 MPa, corresponding to a maximum load of 15 MN, and an electric power source with a capacity of 6 kW is supplied.

Three types of WC anvils (ZK10F, ZK01F, and TJS01) are used in the present HPHT experiments. The first two types were manufactured by Heyuan Zhengxin Hard-Metal Carbide Pte Ltd. Co. in China, while the latter was produced by Fuji Die Co., Ltd. in Japan. Table S1 (supplementary material) provides details of their mechanical properties. Among these three types of WC anvils, the TJS01 WC anvil exhibits the highest hardness on both the Vickers and Rockwell scales. The grain sizes are 0.8, 0.6, and 0.3 μm for the TJS01, ZK01F, and ZK10F types, respectively, as determined using a Magellan 400 field emission scanning electron microscope (SEM) equipped with a backscattered electron (BSE) detector [Figs. S1(a)–S1(f), supplementary material]. The hardness of the WC anvils decreases with increasing cobalt (Co) binder content and grain size.

The mechanical properties and geometric configurations of cell assemblies significantly influence the HPHT capabilities of KLVPs. Semisintered Cr-doped MgO serves as the octahedral pressure medium. Rhenium (Re) is utilized as the heater element. For both pressure calibrations and HPHT experiments, conventional cell assemblies are employed by adjusting the ratio of the octahedral edge length (OEL) to the truncated edge length (TEL) to give, for example, 14/8 (OEL/TEL), 10/4, and 8/3 assemblies. The efficiency of pressure generation and the sample capsule size can thus be modulated by altering the OEL/TEL ratio.

To overcome the limitation of small sample sizes below 1 mm³ in previous assemblies, such as the 4/1, 5.7/1.5, and 6/1.5 assemblies for ultrahigh-pressure experiments exceeding 40 GPa,^17,19,20 we modified a conventional 8/3 cell assembly [Figs. 2(a) and 2(b)] to increase the sample volume to ∼1.8 mm³. This was achieved by incorporating polycrystalline boron-doped diamond disks of thickness 0.5 mm, hard alumina, boron nitride, and magnesia tubes [Figs. 2(c) and 2(d)]. Cross-sectional views of two examples quenched from HP and HPHT are shown in Figs. S2(a) and S2(b) (supplementary material). It can be seen that the sample is well sandwiched by two hard boron-doped diamond disks. The polycrystalline boron-doped diamond disks contain ∼0.46 wt. % boron and have a high hardness of 128 GPa.²¹ The sample chamber is made of hard alumina ceramics, with hard magnesia cylinders wrapped with Cu foils serving as electrodes and to support the hard alumina and diamond disks. A pad made of hexagonal boron nitride separates the alumina and boron-doped diamond disks. Rhenium foils with a thickness of 100 μm serve as the heater, and zirconia is employed as the thermal insulator. It should be noted that before these cell assemblies, we also tested several cell assemblies as shown in Fig. S3 (supplementary material), but they were found to easily blow out at low press loads or were unable to generate a high pressure. A D-type (W₉₇Re₃–W₇₅Re₂₅) thermocouple (TC) is positioned near one end of the diamond disks within the sample chamber.

The pressure calibration of conventional cell assemblies at room temperature is achieved by placing a pressure standard material in the center of the pressure medium, with two Mo or Cu rods serving as electrodes. By contrast, our new cell assemblies utilize conductive and superhard boron-doped diamond disks and dense alumina, with Cu foils acting as electrodes [Fig. 2(c)]. Although pure diamond is an excellent insulator, it becomes conductive when doped with a certain amount of boron and exhibits stable electrical conductivity even under high-pressure conditions.^22,23 The incorporation of boron-doped diamond in our cell assemblies thus ensures accurate resistance measurements upon compression. Pressure calibration is based on the resistance changes of the material associated with phase transitions at specific pressure points using a two-wire method. The pressure standard materials used at room temperature include zinc telluride (ZnTe, with I–II, II–III, and III–IV phase transitions at ∼6.6, ∼8.9, and ∼12.9 GPa, respectively), gallium arsenide (GaAs, with a semiconductor–metal transition at ∼18 GPa), and gallium phosphide (GaP, with a semiconductor–metal transition at ∼23 GPa).^2,5,24–28 We have also adopted the resistance variations associated with a Mott transition or phase transition of Fe₂O₃ hematite from ∼50 to ∼60 GPa at room temperature from in situ X-ray diffraction observations combined with resistance measurements.²⁸ 

Pressure generation at high temperatures is calibrated by the derived Al₂O₃ content in MgSiO₃ bridgmanite using an MgAl₂SiO₆ oxide mixture as the starting material in the HPHT experiments.^29,30 Such HPHT experiments were conducted using the new 8/3 assembly and ZK10F WC anvils in the present KLVP. The sample was compressed to the target pressure and then heated to the target temperature and kept there for 12 h. After that, the run was quenched by turning off the electric power, and the pressure was released over 12 h. Recovered samples were mounted into epoxy resin, which was then polished with diamond paste to expose the sample. The texture and composition of bridgmanite in the run products were analyzed using a Magellan 400 field emission SEM equipped with a BSE and a JEOL JXA-8230 electron probe microanalyzer (EPMA) equipped with a wavelength dispersive spectrometer (WDS). Forsterite was used as a standard for Mg and Si, and pyrope for Al.

Table I lists the experimental details for pressure calibrations using different types of WC anvils and cell assemblies at room and high temperatures in the present study. The survival times of WC anvils were evaluated by the number of experimental runs at HP or HPHT until the anvils broke or cracked. It should be noted that these survival times include the present runs and other HPHT runs.

As shown in Fig. 3, we evaluated several representative behaviors of pressure generation and heating performance for different cell assemblies and WC anvils. Figures 3(a) and 3(b) display the resistance changes for ZnTe and CaP, respectively. It can be observed that the pressures of the three phase transitions (∼6.6, ∼8.9, and ∼12.9 GPa) for ZnTe correspond to press loads of 1, 1.5, and 1.8 MN, respectively, whereas the phase transition pressure (23 GPa) for GaP occurs at a press load of 7.3 MN. Other results are presented in Figs. S4–S8 (supplementary material). Each pressure curve was fitted and extrapolated on the basis of the phase transition results for pressure materials, as shown in Fig. 3(c). The equation used for fitting these data is $y=Aexp−x/B+C$⁠, where y is the pressure, x is the press load, and A, B, and C are parameters. When the 8/3 assembly is used, the pressure generation efficiency of ZK01F WC anvils is higher than that of ZK10F anvils, because the former are harder than the latter. The capacity of these conventional 8/3 assemblies combined with ZK01F and ZK10F anvils can generally reach ∼25 GPa, as extrapolated by the present pressure calibrations. The limitation of pressure generation of the large 14/8 assemblies is significantly lower than those of the 8/3 assemblies when using the same type of WC anvil. Figure 3(d) illustrates typical heating behaviors of Re heaters in selected cell assemblies under high pressure. It can be observed that the Re heaters in the 8/3, 10/4, and 14/8 assemblies can reach temperatures of 1700–2100 K. In the 10/4 and 14/8 assemblies, the use of a large-sized Re heater leads to a decrease in heating efficiency. Notably, the highest recorded values shown in the figures are not the limiting values, and even higher temperatures can be achieved with higher output power.

The electrical resistance variations of GaP and GaAs, as depicted in Figs. 4(a) and S9 (supplementary material), were observed under compression using our new 8/3 cell assembly. The resistance jumps associated with the phase transitions of GaAs and GaP from a semiconducting to a metallic phase occurred at ∼0.9 and 1.8 MN, respectively, corresponding to 18 and 23 GPa. Figure S2(a) (supplementary material) shows a microscopic cross-sectional view of a recovered sample after decompression from 23 GPa. The pressure standard of GaP is well sandwiched by two diamond disks that touched the Cu electrodes in our new assembly.

Figure 4(b) illustrates the resistance variation of Fe₂O₃ hematite upon compression. It is clearly observed that the resistance decreases sharply from several megaohms to several tens of ohms at a press load of 9.2 MN, and then drops to several ohms and remains stable above 9.5 MN. These observations are almost consistent with those of Ito et al.,²⁸ who measured the electrical resistance of Fe₂O₃ at accurate high pressures using sintered diamond anvils in a six-axis LVP through in situ synchrotron X-ray diffraction. They found that the resistance of Fe₂O₃ rapidly decreased from tens of megaohms to a few ohms at pressures of 50–55 GPa, owing to the occurrence of a Mott transition or phase transition, finally reaching the lowest plateau at 58 GPa. The slightly difference in resistance variation of Fe₂O₃ with press load between Ito et al.²⁸ and our study could be related to different types of cell assemblies, anvil materials, and LVPs. The press load at which the resistance reaches its lowest plateau value can be used for the pressure calibration of 58 GPa as suggested by Ito et al.²⁸ We therefore adopted the press load at 9.5 MN reaching the lowest resistance as the calibrated pressure of 58 GPa.

Pressure generation at high temperatures will differ significantly from that at room temperature owing to the effect of thermal pressure on samples. To further constrain the pressure at high temperatures, we synthesized aluminous bridgmanite together with corundum using our new 8/3 cell assembly combined with ZK10F WC anvils in the present KLVP. We derived the actual pressure through the relationship between aluminum content and pressure at high temperatures.^20,21 The experiment was conducted at a press load of 6.4 MN and a high temperature of 1900 K [Fig. 4(c)] for a duration of 12 h. Figures 5(a) and 5(b) show a representative Raman spectrum and a backscattered electron image of the run product, respectively. The peaks in the Raman spectrum can clearly be assigned to those from bridgmanite and corundum crystals.^31–33 Bridgmanite crystals exhibit a bright color in Fig. 5(b), while corundum crystals are slightly darker. The grain size of corundum is around 3–5 μm, while that of bridgmanite is 1–3 μm. The nonuniform grain sizes and partially amorphous texture of the bridgmanite crystals may be caused by the high stress generated by the superhard diamond disks, and have been observed in previous deformation studies on bridgmanite using hard alumina pistons.^34,35 The Al₂O₃ content in the bridgmanite was found by the EPMA to be 24.4 ± 0.23 mol. % (Table S2, supplementary material). The derived pressure within the sample capsule was around 50 GPa, with an uncertainty of ±1 GPa at 1900 K^29,30 at a press load of 6.4 MN. Our new cell assemblies combined with conventional ZK10F WC anvils not only produce an ultrahigh pressure of up to 60 GPa at room temperature, but also stably reach ∼50 GPa at high temperatures.

We observed no occurrences of blowouts upon compression, heating, or decompression in the ultrahigh-pressure experiments using the present new assembly and ZK10F anvils. Blowouts mostly occur for pressures above 40 GPa, especially upon decompression, using ultrahard but fragile WC anvils such as TJS01 and TF05 WC, owing due to accumulated large strains.^15–20 The present new assembly and WC anvils exhibit better performance and stability than previous studies.^15–20 

Kunimoto et al.³⁶ developed a KVLP using third-stage tapered anvils similar to those in diamond anvil cells and tested the performance of single-crystal diamond, sintered diamond, and nano-polycrystalline diamond as anvil materials. They found that, in contrast to single-crystal and sintered diamond, nano-polycrystalline diamond anvils were able to maintain high pressures above 50 GPa, even at high temperatures. Compared to their apparatus, our cell assembly incorporates much thinner polycrystalline boron-doped diamond disks (0.5 mm in thickness) inside hard alumina chambers and possesses a larger sample volume [see Figs. 2(d) and S2(b) (supplementary material)]. We believe that the hard alumina chambers confine the pressure generated by the diamond disks and surrounding materials during compression and even under heating. Possible evidence for this is provided by the smaller thickness variation of the gaskets used in the new cell assembly upon decompression from high pressure compared with those in a conventional cell assembly. Furthermore, the conductive boron-doped diamond disks allow resistance measurements on pressure markers such as ZnTe and GaP for LVP experiments. The smaller pressure drops upon heating in our study compared with those observed by Kunimoto et al.³⁶ might be due to our use of much thinner boron-doped diamond disks in a hard alumina chamber. Our HPHT experiment shows that the pressure can reach 50 GPa at 1900 K, as indicated by the alumina content in bridgmanite at a press load of 6.4 MN, which is slightly lower than that (∼52 GPa) derived at room temperature under the same press load. Further in situ X-ray diffraction studies are required to clarify the precise pressure variation upon heating in LVPs.

Figure 6(a) displays the fitting curves of pressure generation vs press load based on the phase transitions of pressure markers for our new assembly and conventional cell assemblies with WC anvils. Extrapolation from the fitting indicates that our new 8/3 cell assembly with ZK10F WC anvils can achieve pressures exceeding 50 GPa without compromising the sample volume at press loads above 10 MN in our KLVP. By contrast, conventional 8/3 cell assemblies with ZK10F or ZK01F WC anvils plateau at ∼27 GPa under the same press load. Consequently, the pressure generation efficiency of our new cell assembly is significantly higher than that of conventional 8/3 assemblies. This remarkable ultrahigh-pressure generation capability in a large cell assembly is likely due to the high hardness of the polycrystalline boron-doped diamond and alumina materials.^37,38 The hard alumina chambers are able to confine the pressure generated by the diamond disks and surrounding materials during compression and even under heating. Surrounding the diamond disks with hard ductile boron nitride and semisintered alumina materials further enhances the stability and reliability of the cell assembly under HPHT conditions.

A systematic comparison of the ultrahigh-pressure generation capacity between the present study and previous studies on KLVPs with different guide-block systems is shown in Fig. 6(b). Traditional KLVPs that use harder WC anvils with a TEL of 3 mm, such as TF05 with a 7/3 configuration,³⁹ TF05 with Trim. Oct./3,¹⁵ and TJS01 with Trim. Oct./3,¹⁵ can generate a maximum pressure of 31–36 GPa. By contrast, KLVPs with or without an Osugi-type guide block system using the same type of anvils but with a smaller TEL of 1–1.5 mm can achieve much higher pressures of 33–53 GPa for smaller assemblies such as ZK01F with a 6/1.5 configuration,¹⁸ TF05 with 5.7/1.5,²⁰ and TJS01 with 4/1.0.¹⁹ The pressure generation in these previous studies is significantly lower than that achieved in the present study, which employs a new cell assembly and ZK10F WC anvils. Consequently, the efficiency of our assembly is substantially higher than that of the smaller assemblies in previous studies with 7/3, 6/1.5, 5.7/1.5, and 4/1.0 configurations, which used harder TF05, ZK01F, and TJS01 WC anvils.^{5,15,19,20,39} Even though the surfaces of ZK01F and TF05 WC anvils are tapered (indicated by “Tap.”) by 1° near the truncation, the attainable pressure (40–45 GPa) under high temperatures^17,20 is still lower than that in our study. The pressure efficiency of assemblies with 6/1.5 and 5.7/1.5 configurations is also lower than that achieved in the present study, which uses conventional ZK10F WC anvils and a new cell assembly. Although the ultrahard Tap. TJS01 WC anvils and 5.7/1.5 cell assembly in Osugi-type KLVPs¹⁹ show higher pressure generation (50–64 GPa) and efficiency than those in our study of traditional KLVPs, the former assembly often fails catastrophically, and the WC anvils can only be used once. By contrast, our assembly can operate stably even under extreme HPHT conditions, and the WC anvils can be reused three to five times for these HPHT experiments.

Figure 7 depicts the capacities of pressure generation at high temperatures in KLVPs with various WC anvils. Traditional KLVPs using conventional WC anvils typically achieve pressures below 25–27 GPa at temperatures below 3000 K.^21,40,41 The development of ultrahard and fragile WC anvils, such as TF05, ZK01F, and TJS01, has enabled the extension of pressures to 40–50 GPa at 1600–2000 K in KLVPs equipped with Osugi-type guide blocks or Walker-type modules.^17,19,20 As discussed above, the use of these ultrahard or tapered WC anvils in KLVPs is very limited and is not practical for the synthesis of large samples in extreme HPHT experiments. The techniques that we have developed, which involve modifications to assemblies with conventional and practical WC anvils, enable the expansion of pressure up to 50 GPa at 1900 K within a millimeter-sized sample chamber in traditional KLVPs. Our study thus provides a more practical and routine ultrahigh-pressure technique for the exploration of novel materials at extreme high pressures as well as for studying the Earth’s and planetary interiors.

We have achieved significant advances in ultrahigh-pressure generation via new assemblies and conventional WC anvils in a new KLVP. We have successfully reached ultrahigh pressures of ∼60 GPa at room temperature and 50 GPa at a high temperature of 1900 K within a substantially larger sample volume than that in previous studies. The utilization of advanced new assemblies and WC anvils has demonstrated a significant improvement in pressure generation in KLVPs. The integration of electrically conductive polycrystalline boron-doped diamond and alumina has allowed the monitoring of resistance changes in the Fe₂O₃ pressure standard upon compression. This capability to generate pressures above 50 GPa in a larger sample chamber volume represents a breakthrough that has been unattainable in traditional LVPs. This accomplishment not only signifies a significant step forward in high-pressure science and technology, but also paves the way for new investigations of the Earth’s deep interior and the synthesis of novel materials under extreme high-pressure conditions.

See the supplementary material for the following: characterization of WC anvils; SEM images of WC and their particle size distributions; cross-sectional views of two representative recovered assemblies; cross-sectional views of failure cell assemblies for ultrahigh-pressure experiments; resistance variations of pressure standards; chemical compositions of recovered samples.

This work is financially supported by the National Key R&D Program of China (Grant No. 2023YFA1406200), the National Natural Science Foundation of China (Grant Nos. 42272041 and 52302043), the National Natural Science Foundation of China (Grant No. U23A20561), the Jilin University High-level Innovation Team Foundation (Grant No. 2021TD–05), and the Shanghai Synchrotron Radiation Facility (Grant Nos. 2024-SSRF-PT-510031 and 505511). The manuscript was greatly improved by constructive comments from two anonymous reviewers and the editor.

The authors have no conflicts to disclose.

Xinyu Zhao: Conceptualization (equal); Formal analysis (equal); Investigation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Fenglin Ren: Investigation (equal). Jinze He: Investigation (equal). Yue Pan: Investigation (equal). Hu Tang: Investigation (equal); Resources (equal). Xiaoming Zhang: Investigation (equal). Di Yao: Investigation (equal). Ran Liu: Investigation (equal). Kuo Hu: Funding acquisition (equal); Investigation (equal); Project administration (equal); Resources (equal); Writing – original draft (equal). Zhaodong Liu: Conceptualization (equal); Funding acquisition (equal); Investigation (equal); Resources (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal). Bingbing Liu: Conceptualization (equal); Funding acquisition (equal); Project administration (equal); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding authors upon reasonable request.