Ice exists in essentially the most excessive environments of the universe, from comets travelling by way of the chilly vacuum of area, to the recent, dense interiors of big planets. Terrestrial experiments proceed to find new types of ice at excessive stress–temperature situations1,2,3,4,5,6,7,8,9, reminiscent of superionic states, which bridge the soundness fields of stable and fluid water5,10. Concurrently, state-of-the-art theoretical works predict a wealthy panorama of unexplored structural phases of ice, which seemingly are past the present technical limitations of high-pressure experimentation11,12,13,14,15,16,17.
Regardless of exhibiting a posh stress–temperature section diagram, upon room-temperature compression, the section sequence of ice is comparatively easy, with solely three recognized stable phases (VI, VII, X), all of which have been extensively studied18,19,20,21. Ices VII and X share the identical body-centred cubic crystal lattice of oxygen atoms and solely differ within the association of their hydrogen atoms22. Upon compression, the remoted H2O molecules of section VII turn out to be indistinguishable, remodeling to hydrogen-ordered section X, inside which hydrogen atoms are halfway between nearest-neighbour oxygen atoms23,24,25,26,27,28.
For nearly 30 years, density practical concept (DFT) calculations have predicted that cubic ice X ought to rework to an orthorhombic crystal construction (with Pbcm symmetry) at greater pressures12,17,28,29,30. This orthorhombic modification of crystalline ice is predicted to be secure over an enormous stress regime, exceeding that of some other experimentally noticed section8,12,17. The transition to Pbcm ice has by no means been experimentally realized, and, up till now, remained probably the most vital unresolved issues in essentially the most plentiful molecule on Earth.
Regardless of the significance for planetary science, the research of H2O at such ultrahigh pressures are restricted owing to the experimental problem. This is because of a mix of constrained small pattern volumes, weak Raman scattering cross part and ice inherently being a poor X-ray scatterer. Nonetheless, huge enhancements have been made in diagnostic strategies, specifically the brilliance of synchrotron sources, mixed with the focusing capabilities at X-ray diffraction beamlines31,32.
To research the properties of ice at ultrahigh stress situations, we now have carried out a sequence of static compression experiments, reaching pressures in extra of 330 GPa. By synchrotron X-ray diffraction and Raman spectroscopy measurements, we reveal that post-symmetrization of hydrogen bonds, the crystal lattice of ice X constantly distorts upon compression. Above 308(5) GPa, we observe a first-order section transition from distorted ice X to ice XXII, which is in line with the beforehand predicted Pbcm section.
Compression of ice X
The molecular constructions of all at present recognized ices obey the so-called H2O ice guidelines—every oxygen atom is sure to 4 neighbouring oxygen atoms by hydrogen bonds, and every bond is realized by one hydrogen atom2,5,33,34,35,36. At room temperature, fluid H2O crystallizes into ice VI with P42/nmc area group at 1.0 GPa, earlier than remodeling to ice VII with (Pnbar{3}m) symmetry at 2.2 GPa (refs. 37,38) (see section diagram in Fig. 1). At pressures above 70 GPa, ice VII constantly transforms to ice X and it’s indistinguishable from a supercritical crossover25,26,39. Ices VII and X share the identical body-centred cubic crystal lattice of oxygen atoms and solely differ within the association of their hydrogen atoms22,40.
Section strains proven in black are taken from ref. 42 and references therein. The dashed line represents the transition stress to section XXII. Inset: construction of cubic ice X (area group (Pnoverline{3}m)) and ice XXII (area group Pbcm).
Whereas the transition from ice VII to ice X is unidentifiable in X-ray diffraction experiments, the Raman spectra present clear signatures of the altering crystal symmetry, arising from shifts in hydrogen atomic positions. These modifications usually are not essentially centred on the section boundary or Widom line however are nonetheless attribute of the section change39. Upon compression, it has been beforehand proven that the symmetric stretching mode of H2O ice VII is just not noticed above 60 GPa, whereas above 80 GPa the translational modes are outmoded by a single band in section X26. In line with group concept evaluation, there is just one allowed Raman lively phonon (T2g) in (Pnbar{3}m) H2O ice X, akin to the vibrations of oxygen atoms surrounded by stationary hydrogen atoms26,28.
We’ve carried out high-resolution Raman spectroscopy measurements of ice X as much as pressures of 253 GPa (Fig. 2). We observe the looks of the T2g mode above 80 GPa, which is in settlement with a earlier examine that tracked the frequency dependency as much as a most stress of 128 GPa (ref. 26). Nonetheless, we discover that the T2g mode continues to broaden till it splits into two distinct and resolvable modes above 160 GPa. Whereas each modes harden with stress, the splitting will increase (for instance, Δν ≃ 52 cm−1 and 69 cm−1 at 167 GPa and 253 GPa, respectively). This was noticed in all experimental runs and the frequency of every mode at a given stress was constant between experiments. The splitting of the T2g mode is indicative of a change within the symmetry of the oxygen lattice. Concomitantly, the X-ray diffraction measurements in all experimental runs reveal a marked asymmetry within the (110) reflection of ice X at pressures above ~100 GPa (Fig. 3a and Supplementary Figs. 1, 5, 10 and 11), coinciding with the splitting of the T2g Raman mode. We attribute this to a weak tetragonal distortion of the crystal lattice. Sadly, inside this stress regime, the (200) and (211) reflections are too weak to discern any asymmetry.
a, Consultant Raman spectra upon compression to 253 GPa. b, Raman frequency as a perform of stress of the modes noticed in ice X. Crammed symbols characterize the frequencies taken from becoming two modes to the experimental information. Completely different symbols characterize completely different experimental runs. Black dashed strains characterize extrapolations of the experimental information as a information to the attention. Gray open stars characterize the beforehand reported experimental frequencies of ice X26. The open blue symbols characterize the calculated shifts for tetragonally distorted (Pnbar{3}m) ice X (Methods), and black stable strains characterize the DFT calculated modes of Pbcm ice XXII with thicker strains used for essentially the most intense modes. The gray shaded space represents the frequency regime the place the first-order Raman band from the diamond anvil would obscure the pattern bands. The gradient represents the tail of the diamond band.
a, Consultant X-ray diffraction patterns upon compression from 105 GPa to 335 GPa. Ices X and XXII are represented by darkish blue and black, respectively. Additional diffraction patterns may be discovered within the Supplementary Information. b, 2D diffraction picture of section XXII at 338 GPa. c, Rietveld refinement of ice XXII (Pbcm) at 338 GPa. The black, purple and gray curves present the experimental construction issue, the Rietveld match and the match residual, respectively. The place of the oxygen atom was taken from DFT calculations at 400 GPa (Supplementary Desk 1) and stuck.
Inside calculations, any distorted constructions of ice X are mechanically unstable beneath 300 GPa and chill out again to cubic ((Pnbar{3}m)) symmetry. In actual fact, a mechanical instability (c12 > c11) of the (Pnbar{3}m) mannequin for ice X solely develops round 500 GPa, nicely inside the energetic stability area of the expected Pbcm section (see Supplementary Fig. 2 for the calculated elastic stiffness coefficients of ice X). This instability results in a tetragonal variant of ice X (area group P42/nnm) that has been predicted lately as a metastable section17. As the one out there candidate construction, we now have used the P42/nnm structural mannequin within the Rietveld refinements above 100 GPa to account for a doable distortion. Whereas the expected section is elongated alongside c at 500 GPa (c/a ~1.20), our measurements discover a small compression of c, which will increase with stress (for instance, c/a ~0.99 at 140 GPa to c/a ~0.97 at 300 GPa). We’ve simulated Raman spectra of various tetragonally distorted variants of ice X at a sequence of densities. We discover good settlement between the experimental and calculated Raman frequencies on the experimentally decided distortions (Fig. 2 and Supplementary Fig. 3). Very comparable behaviour is seen in NH4F, an ionic ice analogue, the place NH4F-III (a structural analogue of ice X) develops a tetragonal distortion throughout a large stress vary, earlier than remodeling to a brand new high-pressure section41. Equally, it might appear that this tetragonal distortion is intrinsic to ice X and never as a result of pressure results inside the pattern chamber. An evaluation of whether or not these observations may very well be attributed to deviatoric stress situations of the pattern is included within the Supplementary Information.
Transition to ice XXII
Upon additional compression, we observe distorted ice X as much as 308 GPa, above which there are distinct modifications within the X-ray diffraction sample (Fig. 3a). Over a stress regime of ~10 GPa, we observe additional splitting of the primary ice X reflections into 3 distinct parts, indicative of a change to a different polymorph, henceforth referred to as ice XXII.
The presence of three distinct parts in the primary group of peaks means that the symmetry of ice XXII is not any greater than orthorhombic. The upper-angle peak at 10° is extra intense than the lower-angle peaks. This means that the orthorhombic distortion is utilized to the (1times sqrt{2}occasions sqrt{2}) supercell of the cubic construction slightly than to the unique cubic unit cell, which might in any other case end in a splitting into three parts with equal intensities. Such a distortion corresponds to an Abmm area group. In among the ice XXII patterns (reminiscent of that proven in Supplementary Fig. 5), we observe a weak characteristic at 10.5°, which may very well be listed as (021). Among the many maximal subgroups of Abmm, solely Pbcm and Pbmb permit the presence of the (021) reflection. Pbmb may be rejected on the grounds that it must also have an considerable depth of the (001) peak at 4.7°, which isn’t noticed experimentally. Thus, the Pbcm area group gives the best symmetry appropriate with the experimentally noticed X-ray diffraction patterns of ice XXII.
DFT-based computational quantum chemistry has beforehand predicted that H2O ought to rework into an orthorhombic construction with Pbcm area group upon compression above 250–350 GPa (refs. 12,17,28,29,30). Determine 3c exhibits a comparability of the crystal construction issue of ice XXII at 338 GPa with the simulated construction issue of the expected Pbcm mannequin, relaxed with DFT at 400 GPa, with the refined lattice parameters a = 2.279 Å, b = 3.562 Å and c = 3.495 Å. Regardless of the weak pattern sign and robust grain impact, we observe a great settlement relating to the depth distribution in the primary peak group. The intensities of the (100), (110) and (021) reflections of the Pbcm mannequin, anticipated at 7.3°, 8.7° and 10.5°, respectively, depend upon the atomic coordinates of the oxygen atom, O (0.251, −0.054, 1/4), which have been taken from the DFT calculation at 400 GPa. These coordinates have been fastened within the Rietveld refinement to immediately examine the construction issue of the expected Pbcm mannequin with the experiment. Refining these coordinates doesn’t considerably enhance the residual R-factors, that are largely restricted by the grain impact.
Gentle blue symbols, darkish blue symbols and black symbols characterize phases VII, X and XXII, respectively. The dashed darkish blue line corresponds to the Birch–Murnaghan suits of phases X (between 108 GPa and 308 GPa) and the dashed black line is a Birch–Murnaghan match to section XXII. The outcomes of the DFT P–V equations of state (EoS) of (Pnbar{3}m) section X and Pbcm section XXII are proven as purple and black stable strains, respectively. The gray dotted line exhibits the beforehand reported EoS of ice to 170 GPa (ref. 25). The cross, plus and open star symbols characterize experimental information from refs. 5,6,8, respectively. Left inset: zoomed view of the amount as a perform of stress on the ice X to XXII transition area. Proper inset: the b/a (hexagon symbols) and c/a ratios (sq. symbols) for section X (darkish blue) and section XXII (black). The stable purple and black strains characterize the b/a and c/a ratios for the DFT phases X and XXII, respectively. The error bars characterize the mannequin uncertainty within the Rietveld evaluation of X-ray diffraction information, which largely originates from the refinement of the height full-width at half-maximum and most well-liked orientation.
There’s additionally good settlement relating to the magnitude of the orthorhombic splitting (proper inset in Fig. 3), and the stress at which the height splits, which permits us to assign a Pbcm crystal construction mannequin to ice XXII. Structurally, the O–H–O bonds stay the identical in Pbcm in contrast with ice X, however adjoining layers within the ab aircraft are sheared with respect to one another (inset in Fig. 1). Each the construction and transition stress are in good settlement with earlier predictions12,17,28,29,30. Upon the transition to ice XXII, we observe a discontinuity in quantity of −0.04 Å3 per H2O, which is in line with our DFT calculated worth of −0.02 Å3 per H2O and indicative of a first-order section transition (Fig. 4). In a single experimental run, stress dropped from 338 GPa to ~39 GPa owing to the failure of one of many diamond anvils, which resulted in a change again into ice VII (Supplementary Fig. 7).
To evaluate doable non-hydrostatic results, we now have in contrast the X-ray diffraction patterns collected at every stress from the pattern and the gold stress marker (Supplementary Figs. 4–6 and 8). As gold is surrounded by the pattern, the doable non-hydrostatic results ought to have an effect on each patterns. Significantly, if the noticed peak splitting within the X-ray diffraction patterns of the pattern at round 308 GPa is to be related to non-hydrostaticity, the comparable modifications must be moreover noticed within the gold sample. Nonetheless, no pressure-induced modifications within the profile of the gold reflections are noticed, which guidelines out such an interpretation.
In a subsequent experimental run, we have been in a position to reproduce the transition to ice XXII. Nonetheless, the information high quality was a lot poorer, giving extra information scatter within the extracted volumes, with the stress overestimated owing to the gold calibrant being compressed in opposition to the gasket. This information may be present in Supplementary Figs. 9, 10 and 12. Taking this overestimation under consideration, this experimental run demonstrated ice XXII to be secure to at the least 355(10) GPa.
We have been unable to measure the Raman spectra of ice XXII owing to the weak Raman scattering of H2O, mixed with elevated pressure-induced fluorescence engulfing the pattern sign within the low frequency regime at pressures above 250 GPa. Nonetheless, extrapolating the Raman frequencies of the modes noticed in distorted ice X, the values are in line with two of essentially the most intense modes calculated by way of DFT of Pbcm ice (Fig. 2 and Supplementary Figs. 3 and 14). This additional means that the distorted ice X is an middleman state resulting in section XXII. Whereas complementary Raman and IR spectroscopy research exploring the ice X to XXII transition may present invaluable info, inherent limitations of present diamond anvil cell experiments would hinder such research. Within the case of Raman research, all however probably the most intense Raman modes for the Pbcm construction are in a frequency regime that will be obscured by the confused diamond anvil (gray shaded space in Fig. 2b). Though the IR stretching band of ice X is predicted to separate upon the transition to ice XXII, essentially the most intense modes lie at extrapolated frequencies of ice X and would likely be indiscernible (Supplementary Figs. 3, 15 and 16).
Ice XXII is predicted to be secure over an enormous vary of stress–temperature area, with one other orthorhombic modification, with Pbca symmetry, anticipated to emerge at 760 GPa (ref. 11). Nonetheless, that is at present past the experimental limitations of standard static high-pressure experiments and poses the subsequent generational problem. Shock research have proven that at temperatures exceeding 2,000 Okay and within the stress regime the place we observe section XXII, ice adopts an fcc construction, ice XVIII3. Bridging the hole between static and dynamic compression experiments to grasp the XXII–XVIII transition, in addition to how these relate to the superionic states, would be the topic of nice future curiosity.
In abstract, we observe a distortion of ice X above pressures of 160 GPa, resulting in the formation of ice XXII, at 308(5) GPa; to our data, that is essentially the most dense H2O polymorph ever noticed (Fig. 1). Ice XXII possesses an orthorhombic crystal construction, which is in settlement with quantum chemistry calculations predicting a Pbcm section29. Our examine is an development in direction of understanding the inner construction of big planets, and gives proof that the state-of-the-art quantum chemistry calculations based mostly on DFT precisely reproduce the behaviour of H2O below such excessive environments.
