Combined resistive and laser heating technique for in situ radial X-ray diffraction in the diamond anvil cell at high pressure and temperature

Combined resistive and laser heating technique for in situ radial X-ray diffraction in thediamond anvil cell at high pressure and temperatureLowell Miyagi, Waruntorn Kanitpanyacharoen, Selva Vennila Raju, Pamela Kaercher, Jason Knight, AlastairMacDowell, Hans-Rudolf Wenk, Quentin Williams, and Eloisa Zepeda Alarcon Citation: Review of Scientific Instruments 84, 025118 (2013); doi: 10.1063/1.4793398 View online: http://dx.doi.org/10.1063/1.4793398 View Table of Contents: http://scitation.aip.org/content/aip/journal/rsi/84/2?ver=pdfcov Published by the AIP Publishing Articles you may be interested in 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); 10.1063/1.4772458 BX90: A new diamond anvil cell design for X-ray diffraction and optical measurements Rev. Sci. Instrum. 83, 125102 (2012); 10.1063/1.4768541 A diamond anvil cell with resistive heating for high pressure and high temperature x-ray diffraction and absorptionstudies Rev. Sci. Instrum. 79, 085103 (2008); 10.1063/1.2968199 In situ laser heating and radial synchrotron x-ray diffraction in a diamond anvil cell Rev. Sci. Instrum. 78, 063907 (2007); 10.1063/1.2749443 Melting of indium at high pressure determined by monochromatic x-ray diffraction in an externally-heateddiamond anvil cell Appl. Phys. Lett. 78, 3208 (2001); 10.1063/1.1374497

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

131.94.220.125 On: Fri, 10 Oct 2014 15:52:01

REVIEW OF SCIENTIFIC INSTRUMENTS 84, 025118 (2013)

Combined resistive and laser heating technique for in situ radial X-raydiffraction in the diamond anvil cell at high pressure and temperature

Lowell Miyagi,1,2 Waruntorn Kanitpanyacharoen,3 Selva Vennila Raju,4,5

Pamela Kaercher,3 Jason Knight,4 Alastair MacDowell,4 Hans-Rudolf Wenk,3

Quentin Williams,6 and Eloisa Zepeda Alarcon3

1Department of Geology and Geophysics, University of Utah, Salt Lake City, Utah 84112, USA2Department of Earth Sciences, Montana State University, Bozeman, Montana 59717, USA3Department of Earth and Planetary Science, University of California, Berkeley, California 94720, USA4Advanced Light Source, Lawrence Berkeley Laboratory, Berkeley, California 94720, USA5HiPSEC, Department of Physics, University of Nevada, Las Vegas, Nevada 89154, USA6Department of Earth and Planetary Science, University of California, Santa Cruz, California 95064, USA

(Received 17 October 2012; accepted 10 February 2013; published online 28 February 2013)

To extend the range of high-temperature, high-pressure studies within the diamond anvil cell, aLiermann-type diamond anvil cell with radial diffraction geometry (rDAC) was redesigned anddeveloped for synchrotron X-ray diffraction experiments at beamline 12.2.2 of the AdvancedLight Source. The rDAC, equipped with graphite heating arrays, allows simultaneous resistiveand laser heating while the material is subjected to high pressure. The goals are both to ex-tend the temperature range of external (resistive) heating and to produce environments withlower temperature gradients in a simultaneously resistive- and laser-heated rDAC. Three dif-ferent geomaterials were used as pilot samples to calibrate and optimize conditions for com-bined resistive and laser heating. For example, in Run#1, FeO was loaded in a boron-mica gas-ket and compressed to 11 GPa then gradually resistively heated to 1007 K (1073 K at the di-amond side). The laser heating was further applied to FeO to raise temperature to 2273 K.In Run#2, Fe–Ni alloy was compressed to 18 GPa and resistively heated to 1785 K (1973 Kat the diamond side). The combined resistive and laser heating was successfully performed again on(Mg0.9Fe0.1)O in Run#3. In this instance, the sample was loaded in a boron-kapton gasket, compressedto 29 GPa, resistive-heated up to 1007 K (1073 K at the diamond side), and further simultaneouslylaser-heated to achieve a temperature in excess of 2273 K at the sample position. Diffraction patternsobtained from the experiments were deconvoluted using the Rietveld method and quantified for lat-tice preferred orientation of each material under extreme conditions and during phase transformation.© 2013 American Institute of Physics. [http://dx.doi.org/10.1063/1.4793398]

I. INTRODUCTION

The diamond anvil cell in radial geometry (rDAC)consists of two opposed diamonds that are used to compressa small polycrystalline sample between the diamond tips (orculets) while synchrotron X-rays are brought in orthogonalto the compression direction. The diamond culets applyuniaxial stress to the sample, which is enclosed in an X-raytransparent gasket. This stress serves to increase the pressurewhile deforming the sample elastically and plastically. Forover a decade, X-ray diffraction patterns obtained fromrDAC experiments have been used to study lattice preferredorientation (or texture) and lattice strain development undernon-hydrostatic stresses in a wide range of materials at highpressure and temperature.1–5

One major drawback to DAC experiments has been thedifficulties associated with simultaneously heating and de-forming material at high pressure. While large volume defor-mation devices such as the deformation-DIA6 and the rota-tional Drickamer apparatus7 can heat and deform samples atpressures equivalent to the upper mantle and transition zoneof the Earth (which extend to pressures of ∼25 GPa), theDAC remains the only device that can achieve the entire pres-

sure range of the deep earth (to ∼360 GPa). Because of thechallenges associated with heating, the majority of textureand lattice strain measurements in the rDAC on lower man-tle minerals have been performed at room temperature andhigh pressure.4, 8–11 Room temperature deformational resultscan be difficult to extrapolate to conditions in the deep earthwhere materials deform at both high pressure and tempera-ture. Similarly, the properties of advanced materials for en-gineering applications vary with pressure and temperature.12

Consequently, in recent years several attempts have beenmade to develop high-temperature systems to measure de-formational behavior at high pressures in the rDAC using ei-ther in situ laser heating3, 13 or resistive heaters external to theanvils.14

Laser heating, which typically utilizes infrared lasers di-rected down the compression direction of the DAC, has theadvantage of being able to achieve higher temperatures thanresistive heating. However, it has the disadvantage that largetemperature gradients are generally present within the sam-ple. These temperature gradients are particularly substantialperpendicular to the compression direction, in the radial di-rection. In recent years significant advances have been madeto reduce temperature gradients in the laser-heated diamond

0034-6748/2013/84(2)/025118/9/$30.00 © 2013 American Institute of Physics84, 025118-1

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

131.94.220.125 On: Fri, 10 Oct 2014 15:52:01

025118-2 Miyagi et al. Rev. Sci. Instrum. 84, 025118 (2013)

anvil cell through the use of double-sided heating, beam shap-ing optics, and flat-topped laser power distributions.15, 16 Inaxial geometry, where the X-ray beam travels parallel tothe compression direction of the DAC, the effects of radialtemperature gradients are mitigated by using an X-ray beamsize that is small relative to the size of the laser-heated spot.The X-rays thus probe only a small region of the samplewithin the laser-heated hot spot, and the temperature gradi-ent in the region sampled is minimized. Minimizing the ef-fects of temperature gradients is much more difficult for radialdiffraction experiments. In radial diffraction, the X-ray beamtransverses across the sample in the radial direction and thussamples the large laser spot imposed temperature gradient.3

One solution is to use the combined resistive and laser heat-ing technique in the rDAC. This approach arises from the ob-servations that (1) hot samples typically couple better withinfrared laser light, which may be a consequence ofvibrational- and/or defect-mediated processes, and (2) havinga high background temperature is likely to reduce the temper-ature gradients associated with laser-heated spots. A study byLi et al.17 on temperature profile measurements both from ex-periment and theory shows that lower temperature gradientsare present when the sample is held at 600 K through the useof external resistive heating.

External resistive heating within the DAC 18–21 has theadvantages of both a stable thermal environment and a lackof temperature gradients within the sample chamber. Indeed,the importance of these advantages has led to continuing de-sign developments for external heating in the rDAC. Previ-ous studies by Liermann et al.14 and Petitgirard et al.22 usedan external heater of graphite sheets, similar to the design ofRekhi et al.,20 but modified for radial geometry. These studiesshowed that their rDAC designs using graphite heating sheetswere able to reach 1100 K and 1273 K, respectively. In therecent past, experiments were limited to temperatures below∼1200 K due to the instability of the diamond anvils relativeto graphite at high temperature conditions, particularly in en-vironments that are even slightly oxidizing. Based on the re-sults described below, considerably higher temperatures canbe achieved by relying on the metastability of diamond in ex-tremely anoxic environments.

The present work is oriented towards three goals: (1) ex-tending the temperature capabilities of the externally heateddiamond cell through design modifications; this includesthe combination of resistive heating and laser heating in therDAC, (2) calibrating the temperature of the sample to thehighest temperature attained, and (3) distributing the designand modifications of our high-pressure, high-temperature ap-paratus to allow these conditions to be reproduced and furtherimproved.

A Liermann-type rDAC for combined laser and resistiveheating technique was redesigned and developed at beam-line 12.2.2 of the Advanced Light Source (ALS) at LawrenceBerkeley National Lab. The new rDAC was equipped witha more flexible and more efficient membrane frame as wellas smaller molybdenum heating rods and graphite heatingsheets. Three different geomaterials, FeO, Fe–Ni alloy, and(Mg0.9Fe0.1)O, were used to test the capability of the tech-nique and optimize the experimental conditions. Rietveld

analysis23 was applied to diffraction patterns obtained fromthe different experiments to quantify crystal structure and lat-tice preferred orientation under different pressure and temper-ature conditions.

II. EXPERIMENTAL SETUP

A. The rDAC for resistive heating

The rDAC consists of two opposing diamond anvils gluedon tungsten carbide seats. The seat angles were modified fromthe Liermann design14 by increasing the angles from the topplane by 10◦ which was helpful in collecting high-angle peaksfrom powder diffraction data. Two graphite sheets (Alfa Ae-sar, 97% purity, initially of 1 mm thickness) were placedaround the diamonds and used as heaters that sandwich thegasket. The heaters were custom carved to form fit the gasketdimension as well as the incoming and outgoing X-ray beampaths (Fig. 1(a)). Heating power was supplied through twomolybdenum (Mo) rods, which were held tightly between thetwo graphite sheets in order to ensure good electrical contact.The geometry of the Mo rods has been significantly modi-fied, since both the power input and the mechanical stabilityof the furnace hinges on these rods. Thus, they are criticalfor the success of very high-temperature external heating ex-periments. To improve the stability of the electrical contactbetween the rods and heater, the Mo rods were redesignedto have sharp tips instead of a perpendicular extension end.14

The rods were also enclosed in an alumina sleeve to insulatethe cell body from both the electrical current and heat. In con-trast to the Liermann design,14 the Mo rods were brought into the rDAC from the sides, perpendicular to the compres-sion direction, allowing unrestricted access to the top of therDAC for laser heating. Two 30 μm diameter type B ther-mocouple wires (containing Pt with 30% Rh (+) vs Pt with6% Rh (−)) were used to measure the resistive heating fur-nace temperature. To prevent shorting, one wire was attachedclose to the diamond culet (within 1 mm) using ceramic ce-ment while the other was embedded in the graphite heater nextto the gasket. A Model BPAN-O-PLUS power supply fromMycropyretics Heaters International, Inc., was used to deliverup to 10 V of power with 200 A of current. The power sup-ply has a UL microprocessor (PID) programmable temper-ature controller (Eurotherm 2416) with an over-temperaturecontroller and a current limiter. A silicon controlled rectifier(SCR) power controller has a current limiter (extended softstart) interfaced power supply (220 V, 30 A) with a step downtransformer that delivers a fixed voltage of 10 V and a variablecurrent (up to 250 A). During the experiment, the temperaturecontroller receives feedback from the thermocouple placed onthe diamond side, which then triggers the SCR to reach andmaintain the set temperature. The system was cold junctioncompensated, and the calibration curve was preset for type Bthermocouples.

The rDAC and resistive heating array were then assem-bled in a gas membrane for external pressure control duringthe experiment. The entire assembly was put inside an alu-minum (Al) vessel into which an Ar + 1% H2 atmosphere waspumped to provide a reducing atmosphere, thus preventing

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

131.94.220.125 On: Fri, 10 Oct 2014 15:52:01

025118-3 Miyagi et al. Rev. Sci. Instrum. 84, 025118 (2013)

FIG. 1. Schematic of (a) a graphite heater carved for gasket dimension and X-ray beam paths, (b) the redesigned rDAC assembled in the Al-vessel, (c) theexperimental setup for combined resistive and laser heating, and (d) resistive heating images on Fe–Ni alloy in Run#3 at 1007 K (left) and 1713 K (right).

oxidation of the heaters, the rDAC components, and the di-amonds. The aluminum vessel (Al-vessel) was redesignedto be more compact, significantly lighter, and hence moregeometrically suitable for deployment and usage at beam-line 12.2.2 of the ALS than the original Liermann et al.14

design. The modified Liermann setup at ALS is shown inFig. 1(b). The modifications in the Al-vessel render it substan-tially more portable and easier to mount within size-limitedexperimental setups. And, as we show below, the tempera-ture range of the redesigned assembly is also substantiallyaugmented, representing a key improvement in performance.The redesigned Al-vessel of the rDAC assembly contains ademountable roof that makes this design of rDAC easier toassemble than the original Liermann design.14 The Al-vesselwas sealed with kapton tape and water-cooled to preventover-heating of the sample stage. The WC seat angle wasfurther increased from the original Liermann design to al-low collection of data to higher 2θ angle. Two copper (Cu)rods support restraints were attached to either side on thetop of the outer vessel. These restraints add mechanical sta-bility and particularly prevent twisting of the Mo rods dueto stresses associated with their attachment to heavy power-delivery cables. A spring or glass fiber twine was used torestrain the Cu rods so that they gently forced the Mo rodson to the heater to establish good electrical contact. This ge-ometry, with the power and associated supports entering thecell from the side, allows us to perform combined resistiveand single-sided laser heating. The Al-vessel was mountedperpendicular to the incoming X-ray path in order to mea-sure lattice preferred orientation of the sample (Fig. 1(c)).Previously, the optical view of the sample along the axialdirection of force (through the anvils) was blurred due to akapton window. In the current setup, a sapphire window isused which provides a clearer view of the sample withoutcompromising temperature capabilities. The sample within

the new assembly can be imaged during heating by remov-ing the carbon mirror from the downstream laser and po-sitioning the laser optics above the sample: these changeswere partially enabled by an upgrade to beamline 12.2.2 atthe ALS.

B. Temperature calibration of the resistive heaters

Previously, Liermann et al.14 tested the accuracy of thetemperature measurement of the resistive heater in the rDACup to 1077 K by melting of the standards Pb, Al, and NaCl atambient pressure. Up to 1077 K, the temperature differencebetween the sample and a thermocouple position inside thegraphite heater was ∼30 K. However, the difference betweenthe sample and the heater temperature becomes progressivelylarger at higher temperatures. A temperature calibration wasconducted using type B thermocouples with a 30 μm tip size.The comparison is between (1) a thermocouple put directlybetween the diamond culet where the sample is located (di-amond tip in Fig. 2(b)) and (2) another thermocouple ce-mented onto one of the diamond anvils within 1 mm of theculet.

In Fig. 2(a), the power and temperature measured at thediamond’s side were monitored and compared with resultsfrom Liermann et al.14 The combination of the shifts in ge-ometry of the power delivery system and reducing the sizeof the assembly container has produced a markedly moreefficient system. For example, at a power of 120 W, ourpresent system produces temperatures at the diamond tip∼750 K higher than the original Liermann-type assembly, andthe difference becomes progressively larger at greater power(Fig. 2(a)).

Offsets between input and output temperatures were alsocharacterized. Input temperature indicates the input from the

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

131.94.220.125 On: Fri, 10 Oct 2014 15:52:01

025118-4 Miyagi et al. Rev. Sci. Instrum. 84, 025118 (2013)

FIG. 2. (a) A comparison of the power-temperature relationship between the Liermann-rDAC14 and the redesigned rDAC in this study. (b) Temperature cal-ibration between input temperature (measured on the diamond side) and temperature measured at the sample position. Part (c) shows the difference in inputand sample position temperature (i.e., T input- T sample) vs input temperature for six calibration runs. Errors in measurement are equivalent to the size of thesymbols.

power supply, which is monitored using the thermocouple po-sitioned on one of the diamonds sides. Offset between theapplied temperature at the diamond side and the tempera-ture between the culets is shown in Fig. 2(b). All tempera-tures reported below are corrected to the temperature at thesample position based on this calibration, so, for example, atemperature of 1775 K at the sample corresponds to an inputtemperature of 2093 K (at the diamond side).

Figure 2(c) shows the difference between input temper-ature and temperature measured at the sample position (i.e.,T input – T sample) for six different calibration runs. Therelative magnitude of variation between the temperature mea-sured at the diamond side and on the diamond culet is repro-ducible (Fig. 2(c)). Due to the risk that very high tempera-tures pose to the diamond anvils, most of these calibrationruns were performed below ∼1500 K. One calibration runwas taken to an input temperature of 2093 K at the diamondside, corresponding to a temperature of 1775 K at the sam-ple position (Fig. 2(b)). Below ∼1500 K the average standarddeviation of the temperature difference shown in Fig. 2(c)is 18 K, and the largest standard deviation is 38 K. A sig-nificant source of the scatter in the temperature difference(Fig. 2(c)) is likely due to the reproducibility of placementof the diamond side thermocouple: although this is nominally1 mm from the diamond tip, small deviations in its positionwould be expected to produce the observed scatter. Since thefeedback system on the PID maintains the input (diamondside) temperature to within ±1 K, the main source of errorin our temperature measurement is due to scatter in the tem-perature differences in our calibration runs. Thus, we conser-vatively estimate the errors in temperature measurement tobe ∼±40 K. Furthermore, it is possible that there could betemperature gradients across the sample. At the highest tem-peratures attained (input temperature of 2093 K), the differ-ence between the diamond side and the sample position is 288K. Since the diamond side thermocouple is ∼1 mm from thediamond tip, this translates to a temperature gradient or 0.3K/μm. For an 80 μm diameter sample, this corresponds to a

12 K temperature difference between the center of the sampleand the edge of the sample. However, we expect that temper-ature gradients at the tip of the anvils should be lower sincethe diamond tips are fully surrounded by the graphite heater.Therefore, the temperature gradient across the sample itselfcreated by external heating is likely to be insignificant rela-tive to other uncertainties (less than 10 K).

C. Combined resistive and laser heating

A Nd: YLF laser was used for locally heating the hotsample in the externally heated rDAC assembly. A doubleparallel YLF mirror was installed at 12.2.2 for convenientswitching of the configuration from a double-sided laser heat-ing setup to a single-sided laser heating setup. To allow si-multaneous laser- and resistive-heating, it is important to iso-late the diamonds completely from the air to avoid traces ofgraphitization that typically commence at around 973 K in air.Any graphitization results in the inability of the laser to passthrough the diamond anvil to the sample and absorption of thelaser (and local heating) at the location of graphitization. Weused a sapphire window that was attached to the top aluminumvessel to prevent the diamond table from being exposed to air.The controlled atmosphere that surrounded the entirety of thediamonds slows down the rate of conversion to graphite whenthe sample is resistively heated to high temperatures. The ini-tial alignment of the laser hotspot to the X-ray beam was doneusing a few grains of LaB6 calibrant mounted on the tip of aglass fiber, which was aligned to the center of the goniome-ter stage using a 10 × 10 μm2 X-ray beam. A Gig-E cameramounted along the laser heating stream allows in situ sam-ple viewing and was used to bring the alignment laser intoalignment with the LaB6. The LaB6 was replaced by the ex-ternally heated rDAC assembly and aligned to the goniometerstage using the X-ray beam. Finally, to obtain a clear imageof the sample, the focus of the top lens in the laser streamwas adjusted. As is conventional for spectroradiometry of

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

131.94.220.125 On: Fri, 10 Oct 2014 15:52:01

025118-5 Miyagi et al. Rev. Sci. Instrum. 84, 025118 (2013)

FIG. 3. Diffraction spectra or “unrolled” images of (a) FeO in Run#1, (b) Fe-Ni in Run#2, and (c) (Mg0.9Fe0.1)O in Run#3.

laser-heated hot spots, Wien’s approximation was used to cal-culate the temperature, as described in Yan et al.24

D. Experiments

In our first demonstration run (Run #1), wüstite (Fe0.95O)from the same sample used by Jeanloz and Sato-Sorenson25

was loaded in a 30 μm hole drilled in a boron disk. This50 μm thick and 400 μm diameter disk was then embeddedin a small rectangular mica sheet as in Liermann et al.14 (thisis modified from a previous boron-kapton gasket geometry26).A tiny piece of Pt (Alfa Aesar foil, 99.95% purity), less than5 μm diameter, was deposited on the top of the FeO. The x-ray diffraction pattern of this Pt was used as a pressure stan-dard through the equation of state of Pt from Fei et al.27 TherDAC was positioned ∼300 mm away from a MAR345 im-age plate detector. The X-ray beam was collimated to 10 × 10μm2 and tuned to 25 keV, or a wavelength of 0.49594 Å, fordata collection. A diffraction pattern of LaB6 obtained withthe same X-ray energy and instrument geometry was used todetermine the exact sample-to-detector distance and detectororientation and to calibrate instrumental parameters such asdiffraction peak shape. Diffraction patterns at each temper-ature and pressure step were collected for 120 s. FeO wascompressed up to 11 GPa and gradually resistively heatedto 1007 K (Fig. 3(a)). Once the pressure and temperaturewere stabilized, the YLF laser (at 1097 nm) was focused onthe sample and used to simultaneously heat the sample to2273 K, as measured by spectroradiometry. The pressure in-creased to 20 GPa from thermal effects associated with thecombined heating. Both the thermal expansion of the cell andthermal pressure are likely to play a role in this pressure in-crease, as the estimated thermal pressure associated with a∼1500 K increase in temperature in FeO is near 10 GPa.25

In Run#2, Fe–Ni alloy was loaded in a cubic boron nitride(cBN) disk28 with 50 μm thickness and a sample chamber of

50 μm diameter. The cBN disk was embedded in a small rect-angular kapton sheet.26 In addition, two small discs of MgO(Alfa Aesar, 99.9% metal basis), 5 μm thick and 50 μm indiameter, were placed on the top and bottom of the sampleto serve as thermal insulators during single-sided laser heat-ing. The MgO was also used as a pressure standard insteadof Pt.29 The Fe–Ni alloy (bcc-phase at ambient pressure andtemperature) was first compressed to 13 GPa and transformedto the hcp-phase. After completely transforming to the hcp-phase, the sample was gradually resistively heated to 1007 K(Fig. 3(b)). The laser heating spot was aligned on to the sam-ple; however, the sample did not couple very well with thelaser. Thus, the sample was solely resistively heated up to amaximum temperature of 1785 K (Fig. 3(b)); this rise in tem-perature induced an increase in pressure to 40 GPa and trans-formed the sample to the fcc-phase. The temperature was de-creased to 1335 K (1473 K at the diamond side) as the vesselbecame quite hot after being maintained at 1785 K (1973 Kat the diamond side) for 30 min. Pressure was then incre-mentally increased up to 52 GPa along the 1335 K isotherm.There was a malfunction of the detector while we were col-lecting Fe–Ni alloy data in the fcc-phase. Thus, only data fromthe bcc- and hcp-phases were analyzed for lattice preferredorientation.

(Mg0.9Fe0.1)O was loaded into an amorphous boron-kapton gasket26 in Run#3. A Pt flake was used as the pressurecalibrant. The sample was first brought to 3 GPa pressure andgradually compressed to 29 GPa. After reaching 29 GPa, thesample was resistively heated to 1007 K (1073 K at the di-amond side). The laser spot was focused on the sample andused to simultaneously heat while collecting diffraction pat-terns (Fig. 3(c)). The combined resistive and laser heating onthis (Mg0.9Fe0.1)O sample reached a temperature of 2273 K.During resistive heating, the pressure increased to 66 GPa,due to the combined effects of thermal expansion of the DACand thermal pressure in the sample.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

131.94.220.125 On: Fri, 10 Oct 2014 15:52:01

025118-6 Miyagi et al. Rev. Sci. Instrum. 84, 025118 (2013)

E. Data analysis

Because of interference from mica peaks in the diffrac-tion images of FeO in Run#1 (Fig. 3(a)), a Rietveldrefinement23 was not possible. Instead, intensity values weretaken every 15◦ along the azimuth for the 111, 200, and 220peaks using the Fit2D code,30 and smoothed by finding theaverage intensity within a 10◦ range. Values for every 5◦

were then interpolated, and the orientation distribution func-tion (ODF) was calculated from the resulting table of an-gles versus their corresponding diffracted intensities using theBEARTEX software31 and the WIMV algorithm.32 The ODFcan be represented as an inverse pole figure (IPF) of the com-pression direction, which shows the probability of finding thenormal to the lattice plane (or the pole) in the compression di-rection. The pole densities are expressed as multiples of ran-dom distribution (m.r.d.). Examples are shown in Fig. 4.

Diffraction patterns of the Fe–Ni alloy and (Mg0.9Fe0.1)Owere analyzed by the Rietveld method23 as implemented inthe MAUD software33 in order to extract information aboutlattice preferred orientation. The sample-to-detector distance,wavelength, and instrument geometry were first calibratedwith a LaB6 standard. The diffraction rings were then “un-rolled” or integrated along the azimuth over 10◦ incrementsto produce 36 spectra. A stack of the observed spectra showazimuthal variations in peak intensity suggesting the pres-ence of lattice preferred orientation (Fig 3). Sinusoidal vari-ations of d-spacing also indicate lattice strain in the sample(Fig. 3). Several factors such as instrumental parameters, scat-tering background, lattice parameters, microstructure, volumefraction, and lattice preferred orientation were refined to ob-tain the calculated spectra. Crystal structures of the sampleswere obtained from the American Mineralogist Database,34

and preferred orientation was computed using EWIMV, a to-mographic algorithm similar to WIMV,32 using 10◦ resolu-tion for the ODF determination and imposing sample symme-try. The BEARTEX software was used to further smooth theODF with a 7.5◦ filter to minimize artifacts from ODF cellstructures.

FIG. 4. (a) The P-T path followed the rDAC experiments and selected IPFsobtained from (b) Run#1: FeO and (c) Run#3: (Mg0.9Fe0.1)O at various pres-sure and temperature ranges. Temperatures are corrected to the sample posi-tion using the calibration in Fig. 2(b).

III. RESULTS AND DISCUSSION

A. Diamond stability at high temperature

Run#3 accessed extremely high temperatures via externalheating to 1973 K as measured on the side of the diamond.Here, the likely origins for the persistence of diamond un-der these high temperature conditions are examined. Diamondbegins to oxidize and graphitize in air at temperatures above∼970 K35, 36—a fundamental limitation on non-controlled at-mosphere environments in externally heated diamond cell ex-periments. The graphitization occurs entirely at the diamondsurface and is crystallographically controlled, with the (100)surface beginning graphitization in air at ∼1120 K, and the(111) and (110) surfaces beginning to graphitize at 970 K.37

Within oxygen-poor environments (whether in reduced-gas or under vacuum), the metastability range of diamond atambient pressure is markedly enhanced. For example, trans-mission electron diffraction is required to detect the minimaldegree of graphitization of diamond that occurs over a 45 mintime span when held at 1773 K, with the rate of graphitizationmarkedly accelerating up to 2273 K.35, 38 Davies and Evans39

report that graphitization is not detectable in an inert atmo-sphere until 1800 K. The relatively large volume change as-sociated with the reversion of diamond to graphite (and thecommensurate requirement that a high stress environment beproduced by conversion to graphite within the bulk of a dia-mond) typically confines graphitization to the surface of thediamonds, unless there are micro-cracks or inclusions withinthe diamond.39,35

Several studies have extensively documented high-temperature metastability of diamond without graphitization,including Raman studies to 1850 K in vacuum,40 SEM anal-yses in a N2 atmosphere to 1573 K,41 and SEM of diamondseed crystal on fibers to 2473 K in a H2 atmosphere.36 Mod-est pressures of 1–8 GPa, such as those that are likely tobe associated with diamond tips in externally heated exper-iments, appear to weakly elevate the temperature of graphi-tization of bulk diamond, in accord with thermodynamicexpectations.42, 43

In our experiments, the diamonds were pressed into andthus fully encased in the graphite heater. Hence, the dia-monds, which survived these experiments, were directly jux-taposed with a large-scale oxygen sink (the graphite fur-naces). Furthermore, the entire system is stabilized by a con-stant rapid flow of Ar + 1% H2 gas. The presence of 1% H2

causes any oxygen that enters the system to react to form wa-ter or water vapor. As a result, it is likely that no graphitizationof the diamonds could occur, as they were present in an en-tirely reduced environment. Thus, externally heated diamondcells, within a controlled, anoxic environment, would not beexpected to encounter significant graphitization until temper-atures exceed ∼1750–2000 K over experimental timescales.

Perhaps, more importantly for the operation of exter-nally heated diamond anvil cells, three-point bending experi-ments on diamond have shown that plastic flow of diamondsinitiates above 1773 K, with relatively modest (from a di-amond cell perspective) critical shear stress values of 0.5–1.2 GPa measured at 2073 K.44 This result is in general ac-cord with semi-theoretic constraints on the yield strength of

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

131.94.220.125 On: Fri, 10 Oct 2014 15:52:01

025118-7 Miyagi et al. Rev. Sci. Instrum. 84, 025118 (2013)

TABLE I. A summary of cell parameters and texture strength obtained from diffraction data in Run#1-3 as afunction of pressure and temperature.

Pole densitiesP Temp. Cell parameters (m.r.d.)

Run# Sample (GPa) (K) (Å) Min Max

1 FeO 5 300 4.271 (1) 0.78 1.791 FeO 11 300 4.270 (1) 0.31 1.571 FeO 13 1005 4.290 (2) 0.45 1.711 FeO 20 2273 4.293 (1) 0.48 1.882 Fe–Ni (bcc) 2 300 2.851 (1) 0.60 1.672 Fe–Ni (hcp) 13 300 a = 2.426 (1) 0.77 1.49

c = 4.081 (1)2 Fe–Ni (fcc) 40 1715 . . . . . . . . .3 (Mg0.9Fe0.1)O 3 300 4.212 (2) 0.27 5.563 (Mg0.9Fe0.1)O 29 300 4.055 (1) 0.19 7.043 (Mg0.9Fe0.1)O 57 1005 3.923 (1) 0.17 3.773 (Mg0.9Fe0.1)O 69 2273 3.919 (2) 0.29 4.16

diamond at high temperatures.45 Larger critical shear stresses(∼3.2 GPa) in a point-loading geometry may induce ductil-ity in diamond at temperatures as low as 1273 K.46 Hence,the strong possibility exists that the ultimate control on thepressure-temperature range of the externally heated diamondcell may lie not in the metastability of diamond relativeto graphite but rather in the high-temperature brittle-ductiletransition of diamond itself. The Knoop hardness of dia-mond decreases by almost a factor of four between 300 and1473 K,47 and we would therefore anticipate that ultra-hightemperature externally heated cells would be substantiallypressure-limited relative to their ambient temperature capa-bilities. One solution to this issue may be recently devel-oped nano-polycrystalline diamonds.48 Nano-polycrystallinediamonds are significantly harder and exhibit higher fracturetoughness than single crystal diamond.49 Furthermore, thismaterial retains its strength to high temperatures50 and thushas the potential to considerably extend the pressure and tem-perature range of externally resistive heated diamond anvilcells.

B. Gasket materials

For experiments at the highest externally heated tem-perature conditions, the strength and brittle-ductile transitionof the gasket material is likely to be important for the suc-cess of the experimental assembly: thus, ceramics such as c-BN or amorphous materials such as amorphous boron mayprove to be more useful than metals such as rhenium at theseconditions.

In Run#1 on FeO, a boron (amorphous)-mica gasket waschosen, rather than an amorphous boron-kapton combination,because of the high melting temperature of mica. However,mica is not ideal as a gasket material as it is has a complex x-ray pattern, which results in many additional peaks/streaks inthe diffraction patterns (Fig. 3(a)). The boron-mica gasket wasreplaced with cBN-kapton material for greater gasket strengthat high pressures and temperatures in Run#2. The kapton de-composed to a glass at high temperatures, but this thermally

altered material continued to provide mechanical support forthe cBN insert and remained X-ray transparent to our mostextreme experimental conditions. cBN has some advantagesover boron as a gasket material at high temperatures: it is me-chanically more stable than boron at high temperature, andit also does not couple with the infrared laser during heat-ing. Thus, the laser beam can be defocused to cover the entiresample in cBN-gasketted samples without the laser couplingto the gasket material. However, cBN diffraction peaks canoverlap with sample peaks (the straight lines in Fig. 3(b)). InRun#3, a boron-kapton gasket was used for (Mg0.9Fe0.1)O asthis minimized diffraction peak overlaps, and is both X-raytransparent and reasonably stable at high pressure and tem-perature. Thus, for external resistive heating, the use of a kap-ton jacket is significantly better than mica, but the choice ofan amorphous boron or c-BN insert depends on the applica-tion. A boron insert is generally more useful; however, c-BNis mechanically more stable and appropriate for usage in com-bination with laser heating.

C. Texture development

Overall the graphite heaters, particularly in Run#2-3,were quite stable and able to heat the samples in excess of1000 K over an extended period of time (10 h or more),and the sample assembly was stable in terms of its align-ment throughout the heating cycles. The pressure-temperaturepaths of Run#1-3 as well as selected IPFs are displayed inFig. 4. The information about cell parameters, pressure, tem-perature, texture strength, and differential stress of samples isalso summarized in Table I.

1. Run#1 on FeO

The diffraction patterns in Run#1 contained additionalpeaks of Fe and Fe2O3, possibly due to high-temperature dis-proportionation or oxidation of FeO. Multiple peaks fromunidentified contaminants, mica (from the gasket), and car-bon (from the heaters) are also present (Fig. 3(a)). FeO was

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

131.94.220.125 On: Fri, 10 Oct 2014 15:52:01

025118-8 Miyagi et al. Rev. Sci. Instrum. 84, 025118 (2013)

first compressed to 5 GPa at ambient temperature and devel-oped weak texture at [001] of 1.79 m.r.d. The sample was fur-ther compressed to 11 GPa and the texture strength slightlydecreased to 1.57 m.r.d. The texture strength increased to1.71 m.r.d. as resistive heating was applied to 1007 K. FeOstill remained in a cubic phase and was laser heated to a com-bined temperature of 2273 K (measured by spectroradiome-try), resulting in a texture strength of 1.88 m.r.d at a pressureof 20 GPa. Figure 4 displays IPFs of FeO at different pres-sures and temperatures.

2. Run#2 on Fe–Ni alloy

The Fe–Ni alloy was initially in a body-centered-cubic(bcc) phase. When a relatively small pressure (2 GPa) wasapplied to the sample, weak texture (1.67 m.r.d.) started toevolve. The sample was further compressed to 13 GPa atwhich pressure the Fe–Ni alloy transformed to a hexagonalclose packed (hcp) phase,51 and the texture strength decreasedto 1.49 m.r.d. The reduced strength in texture is probably dueto a larger degree of accommodation of deformation via climbor diffusion-related creep at high temperatures and less viadislocation propagation. Resistive heating was then appliedto the sample in small incremental steps up to 1713 K, whichwas the highest resistively heated temperature recorded. Aspreviously mentioned, a detector malfunction occurred dur-ing collection of data in the fcc-phase, and therefore theseresults could not be analyzed for lattice preferred orientation.Stress and textures generated in fcc Fe have been previouslyanalyzed at pressure up to 36 GPa and 1000 K.14, 52

3. Run#3 on (Mg0.9Fe0.1)O

At pressures as low as 3 GPa, weak texture in(Mg0.9Fe0.1)O started to develop a maxima near the [001]with 5.56 m.r.d. The sample was then compressed to 29 GPa,resulting in a texture strength of 7.04 m.r.d. The texture ofthe (Mg0.9Fe0.1)O was reduced to 3.77 m.r.d. when apply-ing resistive heating at 1007 K. When pressure is increasedat high temperature stress remains low, indicating a decreasein strength likely due to activation of dislocation climb and/ordiffusion creep. After the resistive heating temperature wasstabilized, the downstream laser was focused and used to si-multaneously heat the sample. The temperature was deter-mined from spectroradiometry, resulting in a total tempera-ture reading of 2273 K. Texture strength also increased to4.16 m.r.d. after laser heating. Textures in (Mg0.9Fe0.1)O arequite similar to room temperature texture measurements on(Mg0.83Fe0.17)O53 and we do not observe any change in tex-ture type with temperature. It appears that at least for this min-eral; in the regime of pressure-induced plasticity, temperaturedoes not change deformation mechanisms appreciably, whichis significant for applications to the deep Earth.

IV. CONCLUSION

We have achieved a maximum temperature of 1713 Kat the sample using the modified Liermann-type externally

heated rDAC. At these temperatures, graphitization of dia-mond was not detectable. The temperature limit of the typeB thermocouple restricted further testing to higher tempera-tures. Furthermore, simultaneous resistive and laser heatingcan be readily conducted with the new design. Thus: (1) thisapparatus can achieve temperatures of the sample on the dia-mond tip that are substantially higher than any reported pre-viously and (2) the results from our apparatus can be robustlycalibrated at extreme temperature conditions. The dramaticextension of the thermal stability range of externally heateddiamond anvil cell experiments enhances the range of condi-tions, and hence experiments that can be conducted using thisapparatus. This extends the pressure-temperature capabilitiesof the externally heated diamond anvil cell through the con-ditions present within the Earth’s mantle to depths in excessof 1200 km. This allows in situ externally heated diamondanvil studies on topics as diverse as interrogating the kineticsof mantle phase transitions, exploring the deformation mech-anisms of mantle materials, and conducting petrologic exper-iments at mid-mantle conditions. Applications also extend tomaterials science and include mapping phase transitions inpressure-temperature space, exploring orientation variant se-lections during phase transformations and probing chemistryat extreme conditions. Indeed, this extended range means thata large portion of the pressure-temperature domain primarilyoccupied by multianvil presses can now be accessed by ex-ternally heated diamond cells as well as extended to muchhigher pressures. Furthermore, the combination of laser heat-ing with the resistively heated rDAC likely provides a way tofurther extend the available pressure and temperature range,while maintaining relatively small thermal gradients withinthe sample. Furthermore, we have observed that the tempera-tures were stable to within a few degrees over a period of 16to 20 h during our experimental runs.

ACKNOWLEDGMENTS

We thank the U.S. Department of Energy, Director, Of-fice of Science, and Office of Basic Energy Sciences, underContract No. DEAC02-05CH11231, and COMPRES, underNational Science Foundation (NSF) Cooperative AgreementNo. EAR 10-43050 for supporting this project. L.M. wouldlike to acknowledge support from a Bateman fellowship atYale University and H.R.W. for support from the Carnegie-DOE CDAC program and NSF EAR 0836402. The authorswould also like to thank sample providers, including Ray-mond Jeanloz at UC-Berkeley for FeO, Daniel Reaman atUniversity of Chicago for Fe-Ni alloy, and Hauke Marquardtfor (Mg0.9Fe0.1)O. The authors are appreciative for access tobeamline 12.2.2 at the Advanced Light Source of LawrenceBerkeley National Laboratory. The authors would also like tothank an anonymous reviewer whose comments greatly im-proved the paper.

1R. J. Hemley, H. Mao, G. Shen, J. Badro, P. Gillet, M. Hanfland, and D.Hausermann, Science 276, 1242 (1997).

2A. K. Singh, J. Appl. Phys. 73, 4278 (1993).3L. Miyagi, M. Kunz, J. Knight, J. Nasiatka, M. Voltolini, and H.-R. Wenk,J. Appl. Phys. 104, 103510 (2008).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

131.94.220.125 On: Fri, 10 Oct 2014 15:52:01

025118-9 Miyagi et al. Rev. Sci. Instrum. 84, 025118 (2013)

4L. Miyagi, W. Kanitpanyacharoen, P. Kaercher, K. K. M. Lee, and H.-R.Wenk, Science 329, 1639 (2010).

5W. Kanitpanyacharoen, S. Merkel, L. Miyagi, P. Kaercher, C. N. Tomé, Y.Wang, and H.-R. Wenk, Acta Mater. 60, 430 (2012).

6Y. Wang, W. B. Durham, I. C. Getting, and D. J. Weidner, Rev. Sci. Instrum.74, 3002 (2003).

7D. Yamazaki and S. Karato, Rev. Sci. Instrum. 72, 4207 (2001).8H.-R. Wenk, I. Lonardelli, J. Pehl, J. M. Devine, V. B. Prakapenka, G. Shen,and H. Mao, Earth Planet. Sci. Lett. 226, 507 (2004).

9H.-R. Wenk, I. Lonardelli, S. Merkel, L. Miyagi, J. Pehl, S. Speziale, andC. E. Tommaseo, J. Phys.: Condens. Matter 18, S933 (2006).

10S. Merkel, A. Kubo, L. Miyagi, S. Speziale, T. S. Duffy, H. Mao, and H.-R.Wenk, Science 311, 644 (2006).

11S. Merkel, A. K. McNamara, A. Kubo, S. Speziale, L. Miyagi, Y. Meng, T.S. Duffy, and H.-R. Wenk, Science 316, 1729 (2007).

12R. Hrubiak, V. Drozd, A. Karbasi, and S. K. Saxena, J. Appl. Phys. 111,112612 (2012).

13M. Kunz, W. A. Caldwell, L. Miyagi, and H.-R. Wenk, Rev. Sci. Instrum.78, 063907 (2007).

14H.-P. Liermann, S. Merkel, L. Miyagi, H.-R. Wenk, G. Shen, H. Cynn, andW. J. Evans, Rev. Sci. Instrum. 80, 104501 (2009).

15G. Y. Shen, M. L. Rivers, Y. B. Wang, and S. R. Sutton, Rev. Sci. Instrum.72, 1273 (2001).

16V. B. Prakapenka, A. Kubo, A. Kuznetsov, A. Laskin, O. Shkurikhin,P. Dera, M. L. Rivers, and S. R. Sutton, High Pressure Res. 28, 225(2008).

17X. Y. Li, M. Manga, J. H. Nguyen, and R. Jeanloz, Geophys. Res. Lett. 23,3775, doi:10.1029/96GL03648 (1996).

18D. Schiferl, Rev. Sci. Instrum. 58, 1316 (1987).19W. Bassett, A. Shen, M. Bucknum, and I. Chou, Pure Appl. Geophys. 141,

487 (1993).20S. Rekhi, L. S. Dubrovinsky, and S. K. Saxena, High Temp.-High Press.

31, 299 (1999).21N. Dubrovinskaia and L. Dubrovinsky, in Advances in High-Pressure Tech-

nology for Geophysical Applications, edited by J. Chen, Y. Wang, T. S.Duffy, Y. W. G. Shen, and L. F. Dobrzhinetskaya (Elsevier, Amsterdam,2005), pp. 487–501.

22S. Petitgirard, I. Daniel, Y. Dabin, H. Cardon, R. Tucoulou, and J. Susini,Rev. Sci. Instrum. 80, 033906 (2009).

23H. Rietveld, J. Appl. Crystallogr. 2, 65 (1969).24J. Yan, J. Knight, M. Kunz, S. V. Raju, B. Chen, A. E. Gleason, B. K.

Godwal, Z. Geballe, R. Jeanloz, and S. M. Clark, J. Phys. Chem. Solids 71,1179 (2010).

25R. Jeanloz and Y. Sato-Sorensen, J. Geophys. Res. 91, 4665,doi:10.1029/JB091iB05p04665 (1986).

26S. Merkel and T. Yagi, Rev. Sci. Instrum. 76, 046109 (2005).27Y. Fei, A. Ricolleau, M. Frank, K. Mibe, G. Shen, and V. Prakapenka, Proc.

Natl. Acad. Sci. U.S.A. 104, 9182 (2007).28N. Funamori and T. Sato, Rev. Sci. Instrum. 79, 053903 (2008).29C. S. Zha, H. K. Mao, and R. J. Hemley, Proc. Natl. Acad. Sci. U.S.A. 97,

13494 (2000).30A. P. Hammersley, ESRF Internal Report, Report No. ESRF97HA02T

(1997).31H. R. Wenk, S. Matthies, J. Donovan, and D. Chateigner, J. Appl. Crystal-

logr. 31, 262 (1998).32S. Matthies and G. W. Vinel, Phys. Statatus Solidi B 112, K111 (1982).33L. Lutterotti, S. Matthies, H.-R. Wenk, A. S. Schultz, and J. W. Richardson,

J. Appl. Phys. 81, 594 (1997).34R. T. Downs and M. Hall-Wallace, Am. Miner. 88, 247 (2003).35T. Evans, Contemp. Phys. 17, 45 (1976).36S. Kumar, P. Ravindranathan, H. S. Dewan, and R. Roy, Diamond Relat.

Mater. 5, 1246 (1996).37T. Evans and C. Phaal, in Proceedings of the 5th Biennial Conference on

Carbon (University of Pennsylvania, Pennsylvania, 1962), pp. 147–153.38T. Evans and P. James, Proc. R. Soc. London, Ser. A 277, 260 (1964).39G. Davies and T. Evans, Proc. R. Soc. London, Ser. A 328, 413 (1972).40H. Herchen and M. A. Cappelli, Phys. Rev. B 43, 11740 (1991).41J.-C. Pu, S.-F. Wang, and J. C. Sung, Appl. Surf. Sci. 256, 668 (2009).42C. Pantea, J. Qian, G. A. Voronin, and T. W. Zerda, J. Appl. Phys. 91, 1957

(2002).43J. Qian, C. Pantea, J. Huang, T. W. Zerda, and Y. Zhao, Carbon 42, 2691

(2004).44T. Evans and R. K. Wild, Philos. Magn. 12, 479 (1965).45T. Suzuki, I. Yonenaga, and H. O. K. Kirchner, Phys. Rev. Lett. 75, 3470

(1995).46C. A. Brookes, V. R. Howes, and A. R. Parry, Nature (London) 332, 139

(1988).47C. A. Brookes, E. J. Brookes, and V. R. Howes, Science and Technology of

New Diamond, edited by S. Saito, O. Fukunaga, and M. Yoshikawa (TerraScientific Publishing, Tokyo, 1990), pp. 261–272.

48T. Irifune, A. Kurio, S. Sakamoto, T. Inoue, and H. Sumiya, Nature (Lon-don) 421, 599 (2003).

49H. Sumiya and T. Irifune, J. Mater. Res. 22, 2345 (2007).50H. Sumiya, K. Harano, and T. Irifune, Rev. Sci. Instrum. 79, 056102 (2008).51W. Bassett and E. Huang, Science 238, 780 (1987).52S. Merkel, H.-P. Liermann, L. Miyagi, and H.-R. Wenk, “In-situ, radial x-

ray diffraction study of texture and stress during phase transformations inbcc-, fcc-, and hcp-iron up to 36 GPa and 1000 K,” Acta Mater. (submitted).

53J.-F. Lin, H.-R. Wenk, M. Voltolini, S. Speziale, J. Shu, and T. Duffy, Phys.Chem. Miner. 36, 585 (2009).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

131.94.220.125 On: Fri, 10 Oct 2014 15:52:01