Efficient graphite ring heater suitable for diamond-anvil cells to 1300 KZhixue Du, Lowell Miyagi, George Amulele, and Kanani K. M. Lee Citation: Rev. Sci. Instrum. 84, 024502 (2013); doi: 10.1063/1.4792395 View online: http://dx.doi.org/10.1063/1.4792395 View Table of Contents: http://rsi.aip.org/resource/1/RSINAK/v84/i2 Published by the American Institute of Physics. Related ArticlesAn experimental study on SrB4O7:Sm2+ as a pressure sensor J. Appl. Phys. 113, 023507 (2013) Development of a high pressure electrical conductivity probe for experimental studies of gas hydrates inelectrolytes Rev. Sci. Instrum. 84, 015110 (2013) Portable double-sided laser-heating system for Mössbauer spectroscopy and X-ray diffraction experiments atsynchrotron facilities with diamond anvil cells Rev. Sci. Instrum. 83, 124501 (2012) Ruby pressure scale in a low-temperature diamond anvil cell J. Appl. Phys. 112, 124503 (2012) BX90: A new diamond anvil cell design for X-ray diffraction and optical measurements Rev. Sci. Instrum. 83, 125102 (2012) Additional information on Rev. Sci. Instrum.Journal Homepage: http://rsi.aip.org Journal Information: http://rsi.aip.org/about/about_the_journal Top downloads: http://rsi.aip.org/features/most_downloaded Information for Authors: http://rsi.aip.org/authors

REVIEW OF SCIENTIFIC INSTRUMENTS 84, 024502 (2013)

Efficient graphite ring heater suitable for diamond-anvil cells to 1300 KZhixue Du,1,a) Lowell Miyagi,1,2 George Amulele,1 and Kanani K. M. Lee1

1Department of Geology and Geophysics, Yale University, New Haven, Connecticut 06511, USA2Department of Geology and Geophysics, University of Utah, Salt Lake City, Utah 84112, USA

(Received 2 June 2012; accepted 1 February 2013; published online 22 February 2013)

In order to generate homogeneous high temperatures at high pressures, a ring-shaped graphite heaterhas been developed to resistively heat diamond-anvil cell (DAC) samples up to 1300 K. By puttingthe heater in direct contact with the diamond anvils, this graphite heater design features the followingadvantages: (1) efficient heating: sample can be heated to 1300 K while the DAC body temperatureremains less than 800 K, eliminating the requirement of a special alloy for the DAC; (2) compactdesign: the sample can be analyzed with in situ measurements, e.g., x-ray, optical, and electricalprobes are possible. In particular, the side access of the heater allows for radial x-ray diffraction(XRD) measurements in addition to traditional axial XRD. © 2013 American Institute of Physics.[http://dx.doi.org/10.1063/1.4792395]

I. INTRODUCTION

The generation of homogeneous high temperatures atstatic high pressures has been a great challenge over the pastseveral decades.1, 2 The multi-anvil press has earned its rep-utation in achieving homogeneous high temperatures up to3000 K,3 yet is routinely limited to pressures <28 GPa, al-though remarkable efforts have gone up to 50 GPa.4 Break-throughs in diamond-anvil cell (DAC) technology have ledto high temperatures at ultrahigh pressures.5, 6 In combi-nation with laser heating, temperatures >1500 K and pres-sures >100 GPa have been achieved in DACs, however, tem-perature gradients can be extreme.1, 7, 8 Much of the recentprogress has been to minimize large temperature gradients(>100 K/μm) inherent in the laser-heated DAC,2, 9 includ-ing heating from both sides, adjusting the laser into multi-mode,2 and flat-top laser geometry,10 thus achieving a moreuniform sample heating (<20 K/μm). However, the smalluniformly heated spot (typically < 30 μm, although in prin-ciple could be as large as 50 μm) still imposes great dif-ficulty for many applications: e.g., elemental partitioning,9

high-temperature elasticity,11–13 high-temperature rheology,14

and melt structure.15 Meanwhile, internal resistive heating hasmade striking progress16–19 to achieve stable, homogeneoustemperatures greater than 2000 K up to 70 GPa, yet in prac-tice, intensive and difficult sample loading procedures dis-courage wider applications.

In contrast to laser heating, external resistive heatingaround the diamond anvils provide homogeneous heating inthe sample region under high pressure.20 Most resistivelyheated DACs employ coiled resistive wire heater assem-blies in a cavity surrounding the diamond anvils,21–24 includ-ing commercial ones designed by EasyLab or Betsa. Theseheaters are limited to temperatures less than 900 K due toinefficient radiative heating since there is no direct contactbetween the diamond and heater. Many designs have sub-sequently been proposed to extend the temperatures up to

a)Author to whom correspondence should be addressed. Electronic mail:zhixue.du@yale.edu.

1500 K, but are limited to either pressures under 10 GPa,20, 25

require a special alloy for the DAC body,26, 27 a complex heat-ing assemblage,24, 28–34 or are restricted to a particular DACdesign (e.g., HeliosDAC by EasyLab). In addition, only a fewheaters30, 35 allow access for radial XRD measurements.

In this study, we modify a previous design employ-ing graphite foils30, 31 with a ring-shaped graphite heater(Fig. 1(a)). We have successfully generated temperatures upto ∼1300 K at a pressure of ∼50 GPa. This technique willopen doors to studying material properties, particularly melt-ing, transport, and elasticity at high temperatures and pres-sures. To assess the new design, we have tested calibratedmaterials at both room pressure and high pressure as wellas confirmed the melting curve of H2O water ice VII up to23 GPa.

II. HIGH-PRESSURE, HIGH-TEMPERATURE SYSTEM

Graphite is an ideal material to use for heating as it isa good thermal and electrical conductor with conductivitiesof ∼100 W/(m · K) and ∼105 S/m, respectively. Additionally,graphite can be either soft and flexible or hard and machin-able. As the anvils are diamond, we also take advantage ofdiamond’s thermal and electrical conducting qualities: dia-monds conduct heat well (>1000 W/(m · K)), however, aregood electrical insulators (∼10−13 S/m).

The central part of this design which makes it uniqueand efficient, is the direct coupling of a machined, ring-shaped graphite heater that fits over the tip of each anvil,approximately 0.7 mm from the culet (Fig. 1). The cone-shaped ring heater (3.5 mm OD, 2 mm ID, 0.8 mm thick)is machined from a 4.7 mm rod of EC-16 grade graphite(http://www.graphitestore.com/) and is cut such that it fits di-rectly on to the pavilion of the anvil. The diamond anvil is af-fixed on the tungsten carbide (WC) seat with a high-workingtemperature graphite adhesive (Graphi-bond 551RN). Subse-quently, the heater is affixed onto the diamond with an elec-trically insulating MgO-based adhesive (Ceramabond 571).Early tests show that the cone shape of the ring heater fits

0034-6748/2013/84(2)/024502/5/$30.00 © 2013 American Institute of Physics84, 024502-1

024502-2 Du et al. Rev. Sci. Instrum. 84, 024502 (2013)

FIG. 1. (a) Ring heater assembly. (b) Top view of the ring heater. Flexiblegraphite foils on each side are pressed between the ring-shaped heaters tosupply electric current from Mo rod to the graphite heaters. (c) Photomicro-graph of ring heater assembly. Two type-R thermocouples (TC) are cementedonto the pavilion of the diamond to measure temperature.

nicely on a 16-sided culet faceted pavilion without much ef-ficiency lost as compared to a smooth (unfaceted) pavilion.To apply the current to the ring heater, a piece of soft, mal-leable graphite foil ∼1.4 mm in thickness (Alfa Aesar, 99.8%pure, metals basis) is placed at an end of two machine-notchedmolybdenum rods. The soft nature of the graphite foil moldsitself in between the two ring heaters and allows a single cur-rent to be applied to each ring heater allowing for even heatingof the sample with a single power supply. More importantly,the pressed graphite foils provide the graphite heaters withgood contact on the diamonds at high temperature, which isessential for efficient heating of the sample. Two type-R ther-mocouples are glued with graphite adhesive on to the diamondpavilion to measure the temperature. Care is taken not to shortthe thermocouple wires with the underlain graphite heater.

The heating assembly is integrated with a modified sym-metric DAC (Princeton University Machine Shop) (Fig. 2),an OmniDAC (EasyLab), and a panoramic DAC (Syntek), al-though the design can be easily substituted into a DAC ofchoice once a few minor modifications are made. To allowfor the addition of the two Mo rods that supply current andthermocouple leads, two cuts, performed by electric dischargemachining (EDM), are made into the piston of the symmetric

FIG. 2. (a) Cross section of resistively heated diamond-anvil cell assembly.The DAC is modified from a standard symmetric cell, made of Vascomax350 (Princeton Symmetric DAC). (b) Overview of the cell design. Two 5 mmwide, 7 mm deep grooves are cut by EDM on the piston side to allow the Morods to be inserted. Also note that the cut shown on the right-hand side of theDAC cylinder is also produced by EDM but is not necessary for this design.A cylindrical alumina sleeve between the DAC and the membrane can is alsoused, however, not required to thermally insulate the DAC.

DAC as shown in Fig. 2. We have not observed a decrease inthe performance or the overall alignment of the DAC from theEDM cuts.

A stainless steel gas membrane can is machined to sup-port and align the Mo rods to the center of the sample re-gion. The membrane can also seal argon with 1% hydrogen(Ar + 1%H2) or pure Ar gas to prevent the whole cell assem-bly (e.g., DAC, heater, and anvils) from oxidation or graphiti-zation. The inflatable gas membrane is used to smoothly con-trol the sample pressure at high temperatures, bypassing theusual mode of screws for compression (Fig. 2). The mem-brane can is subsequently mounted into a water-cooling jacket(not shown) during the heating experiments, which turns outto be essential in keeping the temperature of the DAC body<800 K, thus negating the need for special high-temperaturealloys.

III. TEMPERATURE CALIBRATION AT AMBIENTPRESSURE

To test the efficiency of the ring-shaped graphite heater, aseries of experiments are conducted at ambient pressure up toa temperature of ∼1300 K. A type-R thermocouple junctionis placed between the two diamond anvils to measure temper-ature of the diamond culets. In addition, another type-R ther-mocouple is cemented to the pavilion of one diamond anvil

024502-3 Du et al. Rev. Sci. Instrum. 84, 024502 (2013)

FIG. 3. Temperature as function of power at ambient pressure in externalresistively heated DAC. Diamond culet temperatures are measured by a type-R thermocouple, while the diamond table and WC seats are measured by atype-K thermocouple. The temperatures are monitored throughout and arereported here only after reaching steady state. Note the strong linear rela-tionship between temperature and power. Uncertainties in temperature are±10 K.

within 0.3 mm from the culet, and this agrees with the tem-perature of the diamond culets to within 10 K up to 700 K,after which it falls off the anvil and thus is no longer ableto measure a valid temperature. The temperature of one of thediamond’s table and tungsten carbide seat is also measured byinserting a type-K thermocouple at the back of the cell duringheating.

In separate experiments, standards of Au foil and NaClpowder are loaded into a Re gasket, and heated under ambi-ent pressure, and the melting temperatures measured. Meltingis observed visually when the standards turned into spheri-cal droplets as observed by optical microscopy. Temperatureversus power curves are shown in Fig. 3 and show a nice lin-ear relationship between the experiments, suggesting excel-lent reproducibility. Diamond culets are heated to ∼1300 Kwith a modest power input of 300 W. Meanwhile the temper-ature of the WC seats in contact with the DAC, and thus thehighest temperature of the DAC body itself, does not exceed800 K. The exterior of the DAC does not reach a temperaturegreater than 500 K at the highest temperatures in this study,provided it is water-cooled.

To attain these high temperatures, it takes ∼10–15 minto reach a steady-state temperature. By changing the electricpower (or current), the temperature is increased from, for ex-ample, 800 K to 900 K continuously and steadily. The same istrue for decreasing temperatures as well. Once the steady statetemperature is reached, the temperature fluctuates to within afew degrees. In addition, this temperature-power calibrationis reproducible to within 5% after a change of one or bothring heaters as long as the heater dimensions are the same.If the heaters are not changed and only a reapplication ofglue/epoxy occurs (due to loosening at temperatures >900 K),the temperature-power calibration remains the same to withinour measurement uncertainties.

Despite the high temperatures and long heating durations,we have not observed degradation of the graphite heater or

graphitization of the diamond anvils during our experimentalruns provided we supplied sufficient Ar or Ar + 1% H2 gas,including 2 h of heating at 1300 K and over 10 h of heatingat 800 K at pressure. Outside of the DAC, the heater contin-ues to perform well: we held the temperature at 2000 K, asmeasured by spectroradiometry, in a sealed chamber flushedwith pure Ar gas or Ar + 1%H2 for 2 h. The heater tempera-ture remained stable during these tests and did not exhibit anynoticeable degradation after heating. In fact, the only time weobserved severe degradation of the heater was when the tem-perature exceeded 2500 K after 1 h of heating.

IV. APPLICATIONS

A. Melting of H2O ice

To test the new design at high pressures, we measuredthe melting curve of water ice VII up to 23 GPa (Fig. 4).We loaded deionized water into a Re gasket or gold-lined Regasket, reloading the water sample and gasket for each melt-ing point measurement. Ruby spheres were used as pressurecalibrant36, 37 as well as a motion detector. Melting is detectedby visual observation either by disappearance of ice crystalsor sudden movement of the ruby sphere. Melting points aremeasured up to 23 GPa, and are in good agreement with pre-vious measurements using similar techniques.26, 29, 38 The ap-parent disagreement with Ref. 39 is likely due to differentcriteria of detecting melting.29

B. In situ high-pressure and high-temperaturesynchrotron XRD experiments

In order to measure the pressure simultaneously at hightemperature, we performed XRD experiments at the Ad-vanced Light Source (ALS) at Lawrence Berkeley NationalLaboratory. Diamond anvils with 300 μm culets were used inthis experiment. MgO powder (Alfa Aesar, 99.9%, metal ba-sis) was loaded into 100 μm diameter sample chamber, whichwas drilled through a Re gasket preindented to 30 μm. A piece

FIG. 4. Water ice VII melting curve as function of pressure. Four melt-ing points are plotted in comparison with melting curves of previousstudies26, 29, 38, 39 using external resistively-heated DACs.

024502-4 Du et al. Rev. Sci. Instrum. 84, 024502 (2013)

FIG. 5. X-ray diffraction patterns of a sample of Au and MgO before (thinline) and during heating (thick line) at high pressures. The hkl’s of both Auand MgO are noted.

of Au foil (Alfa Aesar, 99.9%, metal basis), 20 μm × 20 μm× 5 μm in dimension, was loaded as a high-temperature pres-sure calibrant.40 X-ray diffraction patterns were taken at eachincrement of pressure and temperature at steady state. Asshown in Fig. 5, Au and MgO peaks are clearly resolved athigh pressure and high temperature. Subsequently, pressurewas calculated based on thermal equation of state of Au.40 Asshown in Fig. 6, while maintaining the gas membrane pres-sure, the pressure of the sample remains nearly constant at∼46 GPa up to 1000 K and gradually increases to ∼51 GPaat 1300 K. We held the temperature and pressure constantfor an hour at 1300 K before further increasing the temper-ature. As we increased the temperature to 1350 K, the pres-sure dropped to 37 GPa and subsequently both diamond anvilsfailed.

FIG. 6. Pressure-temperature path of Au/MgO sample during heating. At thehighest temperature of 1300 K, a pressure of ∼51 GPa was reached and heldconstant for 1 h. The dashed line through the data points is a guide for theeye.

V. DISCUSSION

From a practical standpoint, it is convenient to put to-gether the ring-shaped graphite heater assembly. For exper-iments at temperatures below 800 K, the whole cell assem-bly remains intact and can be reloaded with different samplesrepeatedly. However, for experiments at temperatures greaterthan ∼900 K, the diamond anvils loosen upon decompres-sion thus it is necessary to realign and reaffix the anvils, reaf-fix the graphite heaters and thermocouples for subsequent ex-periments, but the rest of cell assembly, e.g., Mo electrodes,graphite foils are still in place and reusable. This is in contrastto previous graphite heater designs,30–34 in which the heatercan only be used once. Additionally, we do not observe a sig-nificant increase in pressure as temperature is increased, atleast up to 1300 K (Fig. 6) as has been reported previously.34

This may be due to smaller temperature gradients outside ofthe sample chamber or due to the steady gas membrane con-trol applying the force on the DAC.

Thermocouples on the pavilion of diamonds are recom-mended to measure sample temperatures precisely, however,we also found that the culet temperature-power curve (e.g.,Fig. 3) is reproducible to within 5% when steady state isreached, usually within ∼10 min. Therefore, a thermocou-ple may not be necessary, once a reliable temperature-powercurve is measured for a given heating assembly, which makessample preparation even easier.

The gap between the two graphite heaters (Fig. 1) is largeenough to allow x-rays to pass through, providing the oppor-tunity for radial XRD measurement at high temperatures.30, 41

Additionally, any other standard techniques, including axialXRD, can also be used with this design.

In several of our runs to test the limit of the current de-sign, in addition to the run shown in Fig. 6, we did not experi-ence gasket blowouts until temperatures exceeded 1300 K atpressure ∼50 GPa. Therefore, we indentify the temperaturelimit as 1300 K at high pressures. The pressure limit remainsto be explored and extended in future experiments.

Despite the efficiency and ease of use of the current de-sign, there are some limitations. First, strength of the WCseat is significantly reduced at high temperatures42 and even-tually the strength vanishes at its eutectic melting point at∼1500 K.43 Second, Re gaskets (and metallic gaskets in gen-eral) may become too weak to support the diamond anvilsat high temperatures. Third, the diamond anvils also be-come more ductile at increasingly high temperatures andmight eventually limit our maximum pressure and tempera-ture range.44

In order to achieve higher temperatures (>1300 K) athigh pressures, cubic boron nitride gaskets45 are essential overmetallic gaskets (e.g., Re, W) given its extreme high strengthand chemical inertness, even at the extreme conditions. Asfor anvil materials, nanopolycrystalline diamond (NPD) maybe an ideal solution for external heating. With an orderof magnitude lower thermal conductivity than single-crystaldiamond,46 this difference will further build up the temper-ature difference between the diamond culet and WC seat(Fig. 3), thereby allowing for higher sample tempera-tures. Moreover, NPD retains its high strength at high

024502-5 Du et al. Rev. Sci. Instrum. 84, 024502 (2013)

temperature.47 Additionally, this external heating can be cou-pled with laser heating to provide bigger and more uniformhot spots.

VI. CONCLUSION

We designed a compact and efficient heater that can beused with most DAC designs with minimal modifications. Thedesign is unique in the use of direct coupling of a ring shapedgraphite heater to the diamond anvil thus yielding efficient,reproducible, and reusable heating. Additionally, the compactnature provides easy access for both radial and axial XRD,electrical leads, and the ability to fit into many synchrotronbeamlines and spectroscopic setups.

ACKNOWLEDGMENTS

The authors would like to thank Bill Samela and ChrisFiederlein for technical assistance. We also thank AlastairMacDowell and Jason Knight for their help at the AdvancedLight Source, Lawrence Berkeley National Laboratory. Z.D.and K.K.M.L. acknowledge the support from CDAC and Na-tional Science Foundation (NSF) (Grant No. EAR-0955824).L.M. acknowledges support from the Bateman Fellowship atYale University. The authors would like to thank two anony-mous reviewers for a critical assessment of this paper.

1R. Boehler, Rev. Geophys. 38, 221, doi:10.1029/1998RG000053 (2000).2G. Y. Shen, M. L. Rivers, Y. B. Wang, and S. R. Sutton, Rev. Sci. Instrum.72, 1273 (2001).

3E. Ito and E. Takahashi, Nature (London) 328, 514 (1987).4T. Irifune, T. Shinmei, C. A. McCammon, N. Miyajima, D. C. Rubie, andD. J. Frost, Science 327, 193 (2010).

5H. K. Mao and P. M. Bell, Science 191, 851 (1976).6H. K. Mao and P. M. Bell, Science 200, 1145 (1978).7L. Ming and W. A. Bassett, Rev. Sci. Instrum. 45, 1115 (1974).8D. L. Heinz and R. Jeanloz, J. Geophys. Res. 92, 11437,doi:10.1029/JB092iB11p11437 (1987).

9R. Sinmyo, K. Hirose, D. Nishio-Hamane, Y. Seto, K. Fujino, N.Sata, and Y. Ohishi, J. Geophys. Res., [Solid Earth] 113, B11204,doi:10.1029/2008JB005730 (2008).

10V. B. Prakapenka, A. Kubo, A. Kuznetsov, A. Laskin, O. Shkurikhin, P.Dera, M. L. Rivers, and S. R. Sutton, High Press. Res. 28, 225 (2008).

11J. D. Bass, Techniques for Measuring High P/T Elasticity (Elsevier, 2007).12J. F. Lin, W. Sturhahn, J. Y. Zhao, G. Y. Shen, H. K. Mao, and R. J. Hemley,

Science 308, 1892 (2005).13M. Murakami, Y. Ohishi, N. Hirao, and K. Hirose, Earth Planet. Sci. Lett.

277, 123 (2009).14L. Miyagi, M. Kunz, J. Knight, J. Nasiatka, M. Voltolini, and H. R. Wenk,

J. Appl. Phys. 104, 103510 (2008).15G. Y. Shen, V. B. Prakapenka, M. L. Rivers, and S. R. Sutton, Phys. Rev.

Lett. 92, 185701 (2004).

16R. Boehler, Geophys. Res. Lett. 13, 1153, doi:10.1029/GL013i011p01153(1986).

17T. Komabayashi, Y. Fei, Y. Meng, and V. Prakapenka, Earth Planet. Sci.Lett. 282, 252 (2009).

18C. S. Zha, K. Mibe, W. A. Bassett, O. Tschauner, H. K. Mao, and R. J.Hemley, J. Appl. Phys. 103, 054908 (2008).

19S. T. Weir, D. D. Jackson, S. Falabella, G. Samudrala, and Y. K. Vohra,Rev. Sci. Instrum. 80, 013905 (2009).

20W. A. Bassett, A. H. Shen, M. Bucknum, and I. M. Chou, Rev. Sci. Instrum.64, 2340 (1993).

21M. J. Moore, D. B. Sorensen, and R. C. Devries, Rev. Sci. Instrum. 41,1665 (1970).

22W. A. Bassett, A. H. Shen, M. Bucknum, and I. M. Chou, Pure Appl. Geo-phys. 141, 487 (1993).

23D. Schiferl, Rev. Sci. Instrum. 58, 1316 (1987).24Y. W. Fei and H. K. Mao, Science 266, 1678 (1994).25N. M. Balzaretti, E. J. Gonzalez, G. J. Piermarini, and T. P. Russell, Rev.

Sci. Instrum. 70, 4316 (1999).26N. Dubrovinskaia and L. Dubrovinsky, Rev. Sci. Instrum. 74, 3433

(2003).27S. Pasternak, G. Aquilanti, S. Pascarelli, R. Poloni, B. Canny, M. V. Coulet,

and L. Zhang, Rev. Sci. Instrum. 79, 085103 (2008).28Z. Jenei, Ph.D. thesis, Stockholm University, 2009.29J.-F. Lin, B. Militzer, V. V. Struzkhin, and E. Gregoryanz, J. Chem. Phys.

121, 8423 (2004).30H. P. Liermann, S. Merkel, L. Miyagi, H. R. Wenk, G. Y. Shen, H. Cynn,

and W. J. Evans, Rev. Sci. Instrum. 80, 104501 (2009).31S. Rekhi, L. S. Dubrovinsky, and S. K. Saxena, High Temp. - High Press.

31, 299 (1999).32L. S. Dubrovinsky, N. A. Dubrovinskaia, S. K. Saxena, H. Annersten, E.

Halenius, H. Harryson, F. Tutti, S. Rekhi, and T. Le Bihan, Science 289,430 (2000).

33L. S. Dubrovinsky, S. K. Saxena, F. Tutti, S. Rekhi, and T. LeBehan, Phys.Rev. Lett. 84, 1720 (2000).

34N. Dubrovinskaia and L. Dubrovinsky, in Advances in High-Pressure Tech-nology for Geophysical Applications, edited by J. Chen, Y. Wang, T. S.Duffy, G. Shen, and L. F. Dobrzhinetskaya (Elsevier Science and Technol-ogy, Amsterdam, Netherlands, 2005), p. 487.

35S. Petitgirard, I. Daniel, Y. Dabin, H. Cardon, R. Tucoulou, and J. Susini,Rev. Sci. Instrum. 80, 033906 (2009).

36H. K. Mao, J. Xu, and P. M. Bell, J. Geophys. Res. 91, 4673,doi:10.1029/JB091iB05p04673 (1986).

37A. F. Goncharov, J. M. Zaug, J. C. Crowhurst, and E. Gregoryanz, J. Appl.Phys. 97, 094917 (2005).

38F. Datchi, P. Loubeyre, and R. LeToullec, Phys. Rev. B 61, 6535 (2000).39M. R. Frank, Y. W. Fei, and J. Z. Hu, Geochim. Cosmochim. Acta 68, 2781

(2004).40K. Hirose, N. Sata, T. Komabayashi, and Y. Ohishi, Phys. Earth. Planet.

Inter. 167, 149 (2008).41S. Merkel and T. Yagi, Rev. Sci. Instrum. 76, 046109 (2005).42M. Lee, Metall. Trans. A 14, 1625 (1983).43C. M. Fernandes and A. M. R. Senos, Int. J. Refract. Met. Hard Mater. 29,

405 (2011).44D. J. Weidner, Y. B. Wang, and M. T. Vaughan, Science 266, 419 (1994).45N. Funamori and T. Sato, Rev. Sci. Instrum. 79, 053903 (2008).46H. Ohfuji, T. Okada, T. Yagi, H. Sumiya, and T. Irifune, High Press. Res.

30, 142 (2010).47H. Sumiya, K. Harano, and T. Irifune, Rev. Sci. Instrum. 79, 056102

(2008).