Article pubs.acs.org/IC
Synthesis of Bulk BC8 Silicon Allotrope by Direct Transformation and Reduced-Pressure Chemical Pathways Oleksandr O. Kurakevych,*,† Yann Le Godec,† Wilson A. Crichton,‡ Jérémy Guignard,‡,∇ Timothy A. Strobel,§ Haidong Zhang,§ Hanyu Liu,§ Cristina Coelho Diogo,∥,⊥ Alain Polian,† Nicolas Menguy,† Stephen J. Juhl,# and Christel Gervais⊥ †
IMPMC, UPMC Sorbonne Universités, UMR CNRS 7590, Muséum National d’Histoire Naturelle, IRD UMR 206, F-75005 Paris, France ‡ The European Synchrotron Radiation Facility, 71 av. des Martyrs, F-38000 Grenoble, France § Geophysical Laboratory, Carnegie Institution of Washington, Washington, D.C. 20015, United States ∥ Institut des Matériaux de Paris Centre FR 2482, F-75252 cedex 05 Paris, France ⊥ LCMCP, UPMC Sorbonne Universités, UMR CNRS 7574, F-75005 Paris, France # Department of Chemistry, The Pennsylvania State University, University Park, Pennsylvania 16802, United States ABSTRACT: Phase-pure samples of a metastable allotrope of silicon, Si−III or BC8, were synthesized by direct elemental transformation at 14 GPa and ∼900 K and also at significantly reduced pressure in the Na−Si system at 9.5 GPa by quenching from high temperatures ∼1000 K. Pure sintered polycrystalline ingots with dimensions ranging from 0.5 to 2 mm can be easily recovered at ambient conditions. The chemical route also allowed us to decrease the synthetic pressures to as low as 7 GPa, while pressures required for direct phase transition in elemental silicon are significantly higher. In situ control of the synthetic protocol, using synchrotron radiation, allowed us to observe the underlying mechanism of chemical interactions and phase transformations in the Na−Si system. Detailed characterization of Si−III using X-ray diffraction, Raman spectroscopy, 29Si NMR spectroscopy, and transmission electron microscopy are discussed. These large-volume syntheses at significantly reduced pressures extend the range of possible future bulk characterization methods and applications.
1. INTRODUCTION The allotropy of silicon is both rich and very promising for tackling immediate challenges in electronic and photovoltaic applications. At present, a large number of metastable crystal structures have been predicted,1−8 and their electronic structures and possible stability/metastability have been studied; yet still only five allotropes of Si may be obtained at ambient conditions in a pure state: cubic Si−I (conventional silicon with diamond structure, the only thermodynamically stable form at ambient pressure); metastable open-framework cubic Si136 (clathrate II structure, known also for zeolites)9 and orthorhombic Si24 (zeolite structure);10 and metastable dense cubic Si−III (BC8 structure) 11 and hexagonal Si−IV (hexagonal diamond structure).11 These phases can be recovered at ambient conditions as pure bulk ingots, while a number of other phases (Si-XII or R8,12 9R and 27T polytypes of Si−I,13 and even some new structures locally observed by TEM/SAED14) can also exist at ambient conditions. The electronic structures of most allotropes are quite wellunderstood; however, there remain numerous controversies, in particular, regarding the Si−III phase. Early electrical measurements (Hall-effect and low-temperature resistivity)15 suggested that Si−III is a hole semimetal with low charge carrier density. The observed temperature variation of conductivity was © XXXX American Chemical Society
believed to conform to the expected behavior of a highly disordered microcrystalline conductor with indirect overlap (∼0.3 eV) of valence and conduction bands. In spite of the lack of observable superconductivity to 1.2 K, this possibility has not been precluded for ordered, particularly single-crystal, forms of Si−III allotrope.15 Although the experimental data are both reasonable and reproducible, attempts to understand the electronic band structure have met less success. Surely, with some reasonable assumptions one can explain the experiments, but still there remains a more significant and fundamental question: which properties are intrinsic for Si−III? Previous experimental observations indicate non-reproducible electrical properties from samples obtained at similar conditions,11 possible formation of a mixture of phases,12 stacking faults,16 and other features are inferred.17 Therefore, a significant gap remains in producing well-characterized phase-pure material to understand the intrinsic properties. Originally, ab initio calculations (mainly) supplemented by electrical resistivity measurements suggested that Si−III was semimetallic with negative indirect bandgap (Γ-X points). Recent results have indicated that the optical gap is close to 2.8 Received: June 15, 2016
A
DOI: 10.1021/acs.inorgchem.6b01443 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry
quenched experiments.23 The sample was compressed between 13.5 and 14 GPa at a rate of 0.7 GPa/h at RT, then heated to ∼873 K, and kept for 6 h. Then temperature was quenched to RT by shutting down the electrical power. The pressure was held still for one more hour, then slowly decompressed to atmospheric pressure over 3 d. HP ex situ synthesis using Na4Si4/Si mixture (Na/Si = 1:5) at 7 GPa and 1400(50) K (sample EP10) was performed in ParisEdinburgh cell 7/2 (opposite anvil geometry) using a 7 mm diameter pyrophyllite gasket with 2 mm hole and tungsten carbide (WC) anvils. The sample was placed directly into a Ta capsule/heater (foil of 25 μm thickness). Compression and decompression were performed during 2 and 4 h, respectively, and a total heating time of ∼1 h. The pressure calibration was made in previous synchrotron experiments24 using the equation of state of hBN.25 Temperatures were calibrated against fixed points of the melting curve of Si.26 It is interesting to note that at higher and lower heating powers (and temperatures as well), no Si−III formation was observed, just NaSi627 and HP clathrate s-II Na24+ySi136, with y ≈ 6 according to the lattice parameter.28,29 The formation of Si−III has been also observed in the recovery experiments with Na4Si4/Si mixture at 7.7 GPa and 1020(30) K (samples T1438 & T1439) in Toroid apparatus T-20 (opposite anvil geometry, the cells and calibration methods were described elsewhere)30,31 at the 5−10 min time scale, that is, at the initial stages of HPHT transformations (the completion of chemical reactions requires hour-long time scales). At lower temperatures, ∼900(20) K, only Na24+ySi136 (y ≈ 6)28,29 was observed in the sample recovered from similar conditions. In situ experiments at 7.5 (a mixture of samples of two experiments, denoted as HT R6 & R8) and 9.5 GPa (samples HT R1 and HT R4) were performed using the 20 MN Voggenreiter press at beamline ID06 of the European Synchrotron Radiation Facility (ESRF).32 For the HT R4 experiment, a Na/Si mixture with 15 mol % Na was ground in a ceramic mortar for 1 h inside a high-purity Ar glovebox and loaded into an hBN capsule. For the HT R1 experiment, we used previously synthesized HP clathrate s-II Na24+ySi136 (y ≈ 6). The sample was then introduced (under Ar atmosphere) into a 14/8 multianvil assembly (MgO octahedron with 14 mm side compressed with eight WC cubic anvils with 8 mm side triangular truncations), equipped with graphite furnaces. Temperatures and pressures were monitored using the HPHT equations of state of Si−I33 and Na34 or s-II35 (in parallel with estimation from previous calibration curves).36,37 Monochromatic Xray diffraction data were also taken using a wavelength corresponding to 33 or 55 keV. The beam was collimated to define a horizontal beam size of ∼1 mm to ensure the probing of the whole sample. Diffraction measurements were collected on an azimuthally scanning Detection Technology X-Scan c series GOS linear detector. After synthesis the power was switched off, and after that the pressure was slowly released. In situ probing of phase composition was also performed on decompression. After HPHT synthesis in Na−Si system, the recovered samples were washed four times with demineralized water in an ultrasonic bath, which allowed not only removing the excess of sodium (or, more precisely, of Na4Si4) and grain separation, but would also help to facilitate reduction of residual strains, which could be introduced during fast quenching from HT. The largest grains were chosen under an optical microscope and characterized by in house X-ray diffraction (rotating Mo anode source, two-dimensional (2D) image plate, four circle goniometer in κ-geometry, diffractometer Agilent Xcalibur S). Approximately half of the recuperated grains, both for HT R4 and HT R1 experiments, contained only pure Si−III phase. The linear size was up to 500 μm, weight of ∼1 mg. For direct elemental transformations, ∼2 mm phase-pure ingots, as measured by XRD (Bruker D8 Discover, Cu Kα radiation and Vantec 500 area detector), were obtained directly with no further processing. In situ observations of phase transformations/stability at high temperatures under vacuum condition (typically 1−5 Pa) were performed using HTK 1200N high-temperature oven chamber (Anton Paar) for conventional diffractometry (Cu anode source, goniometer two circles θ−θ, diffractometer Panalytical X’Pert Pro MTD).
eV for Si−III nanoparticles, which have very close lattice parameter a, but quite different intensities of hkl reflections compared to bulk microcrystalline samples.18 The change in optical gap might be explained by grain-size effect (quantum confinement),19 although processing methods and surface termination effects may also contribute. Moreover, later quasiparticle ab initio results indicated that the direct bandgap (H point) depends not only on the lattice parameter but also on the only independent atomic coordinate, x of the unique atomic site in the structure; the 16c Wyckoff position of space group Ia3̅, No 206, with coordinates (x x x).20 Expecting to be able to resolve the problem with wellcrystallized pure samples, we make use of the assumption that if the electronic structure is sensitive to lattice parameter a and order parameter x, it should be visible by changes in X-ray diffraction (XRD), Raman, and NMR spectra of samples obtained by various methods. In the present work, we explored a number of chemical high-pressure (HP) routes as well as direct HP and high pressure−high-temperature (HPHT) transformations to produce and compare the crystalline Si− III products. We have shown that pure, well-crystallized Si−III samples may be synthesized in quasi-hydrostatic HPHT conditions in the Na−Si system at pressures as low as 9.5 GPa, while its formation can be observed at pressures down to 7 GPa. A number of pure polycrystalline samples was also prepared by direct HPHT transition of Si−I. The detailed study of samples obtained by various methods indicates that, although some minor variations in a and x parameters take place, as well as the concentration of stacking faults, no significant change in the structure (XRD), fundamental frequencies (Raman), or chemical shift (NMR) can be observed.
2. EXPERIMENTAL AND COMPUTATIONS For synthesis we used commercial Si (Prolabo, 99.9% purity) and Na (Aldrich, 99.5% purity). Na4Si4 was synthesized by previously reported procedure with some modifications.21 Sodium hydride (Sigma-Aldrich, 95%) and silicon (Sigma-Aldrich, 99%) powders were used as received. Both powders were manipulated in an argon-filled glovebox. One millimole of silicon and 2.5 mmol of sodium hydride were loaded in a Retsch MM400 ball-miller (airtight stainless steel vials of 50 mL; one steel ball of 62.3 g and a diameter of 23 mm) and mixed for 2 min at 20 Hz. The mixture was then transferred under inert atmosphere in an alumina crucible with an alumina cap. The crucible was loaded into a vertical quartz tube. The tube was then heated at 693 K in a vertical tube oven under continuous argon flow for 3 d. After it cooled, the product with a ductile aspect was treated at 333 K under vacuum for one night. The resulting gray powder (according to XRD, a = 12.17 Å, b = 6.55 Å, c = 11.15 Å, β = 119.0°, space group No. 15, C12/c1) was transferred and stored under argon. Room-temperature (RT) synthesis of Si−III by direct transformation at 13 GPa was performed in Paris-Edinburgh (PE) press using a 5 mm diameter boron-epoxy gasket with 1.5 mm hole (cell 5/ 1.5) and sintered diamond anvils (opposite anvil geometry, RT uniaxial compression/decompression). Pressure calibration was made in previous synchrotron experiments using equation of state of MgO.22 Bulk Si−III samples at HPHT conditions were synthesized by direct elemental transformation using a 1500 ton multianvil press. Approximately 15 mg of pure Si powder (Sigma-Aldrich, 99.999%) was filled inside a cylindrical hBN capsule (∼3 mm diameter and 4 mm height). The hBN capsule was then inserted into the center of a cylindrical graphite heater, a cylindrical ZrO2 sleeve, and a 5% Cr2O3doped MgO octahedra with MgO plugs on the ends. The octahedra with 14 mm edge lengths were centered against eight tungsten carbide anvils with 8 mm truncation edge lengths. The pressure was calibrated through monitoring phase transitions of quartz−coesite, CaGeO3 garnet−perovskite, coesite−stishovite, and olivine−wadsleyite from B
DOI: 10.1021/acs.inorgchem.6b01443 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry Raman measurements were performed using spectrometer T64000 (Jobin-Yvon/Horiba) with nitrogen-cooled CCD detector. The scattering was excited using green Ar laser. Additional measurements were performed in the backscatter geometry using Raman systems with Princeton Instruments spectrographs (SP2750/SP2500). 532 and 660 nm lasers were used as excitation sources, focused through a 20× long working distance objective (NA = 0.40), and Raman light was collected in the backscatter geometry through a 50 μm confocal pinhole. The Raman signal was dispersed off a 1800 or 1200 gr/mm grating on to a liquid nitrogen-cooled CCD, providing a spectral resolution of
