An Ultradense Polymorph of Rutile with Seven-Coordinated Titanium from the Ries Crater

See allHide authors and affiliations

Science  24 Aug 2001:
Vol. 293, Issue 5534, pp. 1467-1470
DOI: 10.1126/science.1062342


We report the discovery of an ultradense post-rutile polymorph of titanium dioxide in shocked gneisses of the Ries crater in Germany. The microscopic diagnostic feature is intense blue internal reflections in crossed polarizers in reflected light. X-ray diffraction studies revealed a monoclinic lattice, isostructural with the baddeleyite ZrO2 polymorph, and the titanium cation is coordinated with seven oxygen anions. The cell parameters are as follows: a = 4.606(2) angstroms, b = 4.986(3) angstroms, c = 4.933(3) angstroms, β (angle between c and a axes) = 99.17(6)°; space groupP2 1 /c; density = 4.72 grams per cubic centimeter, where the numbers in parentheses are standard deviations in the last significant digits. This phase is 11% denser than rutile. The mineral is sensitive to x-ray irradiation and tends to invert to rutile. The presence of baddeleyite-type TiO2 in the shocked rocks indicates that the peak shock pressure was between 16 and 20 gigapascals, and the post-shock temperature was much lower than 500°C.

The response of TiO2 to high pressure has been a subject of many recent experimental investigations (1–11). Because rutile is isostructural with stishovite (SiO2), its behavior at high pressures and temperatures has been considered to offer an analogy that would allow the mechanisms and path of post-stishovite phase transitions to be explored under more convenient laboratory conditions. High-pressure experiments on TiO2revealed the existence of several high-pressure polymorphs (3, 7–9). However, until recently, the nature and stability fields of some of these polymorphs and the path of the phase transitions have remained controversial (1, 3, 6, 7, 12,13). Rutile is the most abundant species in nature among the low-pressure polymorphs and is a minor constituent in igneous and metamorphic rocks (14). The response of TiO2 to static and dynamic high pressure may thus reveal fundamental information about the behavior of this and other related oxides in Earth's mantle during the subduction of crustal limbs in Earth's interior and natural dynamic events. An α-PbO2–structured polymorph has been recently reported from two entirely different petrologic settings: (i) as a nanometer slab in a rutile twin bicrystal inclusion in garnet from diamondiferous quartzofeldspathic rocks from the Saxonian Erzgebirge (15), and (ii) in shocked gneisses from the Ries crater in Germany (16).

Recent diamond anvil cell (DAC) experiments on rutile up to total pressures in excess of 55 GPa established the existence of four dense polymorphs: (i) an orthorhombic α-PbO2 phase (TiO2 II, space group Pbcn), stable below 14 GPa at 300 K (8, 9); (ii) a monoclinic baddeleyite-structured phase (M I, space groupP2 1 /c), stable above 14 GPa (8); (iii) an orthorhombic polymorph (O I, space groupPbca), stable above 28 GPa (11); and (iv) an orthorhombic cotunnite (PbCl2)–structured polymorph (O II, space group Pnma), with nine-coordinated Ti, stable above 55 GPa (10).

The monoclinic baddeleyite-structured phase was never obtained in any shock-loading experiment. Instead, the α-PbO2 phase was recovered from shock-loading experiments at pressure (P) > 20 GPa (1, 2). This result was interpreted as suggestive of inversion of the baddeleyite-structured polymorph to the α-PbO2 phase upon decompression (1, 2).

It was inferred that rutile, in analogy to MnF2, transforms by shock compression in the pressure range of 72 GPa to a distorted fluorite or fluorite-type structure (12, 13). However, recent DAC experiments and ab initio calculations refute the existence of a distorted fluorite or fluorite-type structure in this pressure range (10, 16, 17).

The heavily shocked gneisses and amphibolites in the Ries suevite collected in the Altebürg, Otting, and Seelbronn localities belong to upper stage II of the progressive shock scale (18–20). The stabilities of both the ZrO2-structured (M I, space groupP21/c) polymorphs and the ZrO2-related (O I, space group Pbca) polymorphs lie within this deformation stage (up to 30 GPa) (16). It appeared plausible that one or both polymorphs may be present and probably survived the post-shock temperature. A careful search for TiO2 polymorphs denser than the α-PbO2 phase in polished thin sections (PTSs) of the selected rock suite was conducted. Important criteria for the recognition of these phases are the expected higher reflectivity and different internal reflections in reflected light as a result of the higher densities (14). It is expected that the brightness of these polymorphs (M I and O I) in reflected light should be much higher than that of rutile, because these phases are 11.5 and 12.7% denser, respectively, than rutile.

We found a phase with optical reflectance remarkably higher than that of rutile in an assemblage of TiO2 phases,ilmenite, and minor sphene. Electron microprobe analysis of rutile grains and the denser TiO2 phase indicated that they are identical in composition: almost pure TiO2, with minor concentrations of FeO and Nb2O5 (97.69% TiO2, 0.14% FeO, and 0.20% Nb2O5). The new phase shows, in contrast to the white-to-yellow internal reflections of coexisting rutile, intense blue internal reflections in reflected polarized light and crossed polarizers (Figs. 1B and2B). This is a characteristic diagnostic optical feature that serves to distinguish this phase from the α-PbO2 polymorph, which has pink internal reflections (16). The new phase occurs in clusters of individual fine-grained polycrystalline monophase grains surrounded by large individual shock-twinned ilmenite grains and small rutile grains (Fig. 1) or as single grains surrounded by rutile, the α-PbO2phase, ilmenite, and sphene (Fig. 2). The new phase occupies the core of the assemblage, and the α-PbO2 phase, ilmenite, and rutile occupy the outer regions (Fig. 2). We see no petrographic evidence of one TiO2 polymorph replacing the other in the same grain.

Figure 1

A reflected light photograph displays a shocked opaque assemblage consisting of the new baddeleyite-structured polymorph of TiO2, rutile, and ilmenite. (A) White in the central area and on the right side is the new phase and rutile; here indistinguishable from each other. Brown at the left and bottom right is ilmenite. The photograph was taken with plane-polarized light. (B) The same as in (A) with crossed polars. The material with blue internal reflections is the new monoclinic baddeleyite-structured (ZrO2) TiO2 polymorph. The very small grains at the lower edge of the new phase with white or yellow internal reflections (arrows) are rutile. The fussy appearance of the new phase is the result of the extremely fine-grained nature of the new mineral.

Figure 2

(A) A reflected light photograph depicting a shocked opaque assemblage consisting of rutile, the α-PbO2 polymorph, and the baddeleyite-structured polymorphs of TiO2, along with ilmenite, sphene, and graphite. The photograph was taken with plane-polarized light. Length of the white bar on the bottom right is 100 μm. (B) The same as in (A) with crossed polars. The figure displays the textural relations and the internal reflections of the three TiO2 polymorphs. The baddeleyite-structured polymorph (deep blue internal reflections) occupies the core of the opaque assemblage. Both less dense polymorphs, rutile (white internal reflections), and the α-PbO2 phase (pink internal reflections) surround the new, very dense, baddeleyite-structured phase. Length of the white bar on the bottom right is 100 μm.

We refrained from subjecting the assemblages to laser micro-Raman studies because of the instability of the ZrO2-structured polymorphs (M I) and the denser orthorhombic (O I) polymorphs (10). The M I– and O I–TiO2 phases synthesized in experiments above 14 and 28 GPa, respectively, are metastable and invert instantaneously upon laser irradiation at ambient pressure to the α-PbO2–structured phase (10,11). We determined the structure of the new phase by x-ray microbeam diffraction techniques. A 0.8-mm disc containing the new phase and rutile (Fig. 1) was cored out from the PTS with a high-precision microdrill (21). We collected 21 x-ray reflections from the new phase in addition to the [110] reflection of rutile (Fig. 3 and Table 1). The new phase is sensitive to x-ray irradiation. The intensities of all x-ray reflections of this phase decreased progressively over the 18 hours of x-ray irradiation, and the intensity of the rutile reflection increased. Almost 50% of the new phase transformed within this time to rutile. The d spacings of all 21 x-ray reflections of the new phase (Fig. 3 and Table 1) are indexed in the framework of a monoclinic lattice with the following cell parameters: a = 4.606(2) Å, b = 4.986(3) Å, c = 4.933(3) Å, and β = 99.17(6)° (space group P2 1 /c) (numbers in parentheses are standard deviations in the last significant digits). The positions of the diffraction lines and their relative intensities could be explained in terms of the baddeleyite (ZrO2)–structured polymorph of TiO2 (space group P2 1 /c) (Fig. 3 and Table 1). The calculated density of the new phase is ρ = 4.72 g/cm3. The baddeleyite ZrO2–structured TiO2 phase from the Ries crater is 11% denser than rutile. This is the first natural occurrence of an ultradense TiO2 phase with Ti cations in seven-coordinated oxygen polyhedra.

Figure 3

X-ray diffraction pattern obtained from the central area of Fig. 1B. Twenty-one diffraction lines are indexed in the monoclinic baddeleyite structure (M I). One diffraction line (R 110) belongs to rutile (R) (Table 1).

Table 1

Indexed peaks of the x-ray diffraction pattern and Miller indices collected from the natural baddeleyite-structured TiO2 in the assemblage shown in Fig. 1, A and B.

View this table:

We infer that the new phase formed by direct reconstructive phase transition from rutile. Formation by back-transformation through inversion of one of the denser orthorhombic O I polymorph, the PbCl2-structured O II polymorph, or the cotunnite-structured O II polymorph would require much higher pressures [P = 28 to >60 GPa (10)]. Pressures at this range would have induced not only the inversion of quartz to stishovite but also melting in feldspar and quartz in the shocked gneiss (20, 22, 23). We find no evidence of melting in feldspar or silica grains. Raman studies of the silica inclusions in garnet show that they are dense diaplectic glass with no evidence of coesite or stishovite (18,20, 22–24). This entirely excludes the denser polymorphs as possible precursors. Consequently, the peak pressures very probably did not exceed 28 GPa. However, we cannot entirely exclude the possibility that the new phase formed by inversion of the O I phase.

The fabric setting of the monoclinic ZrO2-structured phase is indicative that the original rutile grains in the core of the cluster experienced higher peak pressures than did the grains near the exterior and that pressure attenuated in the cluster outward during the impact event. Our x-ray diffraction analysis of the assemblage depicted in Fig. 1 revealed no evidence for the presence of the orthorhombic α-PbO2–type phase (Fig. 3 and Table 1). This result supports the idea that the high-pressure polymorphs in the Ries event were preserved during the post-shock decompression period. This contradicts the results of static experiments (8–11). However, we may speculate that the presence of minor elements (0.14% FeO and 0.20% Nb2O5) could have been an important factor in preventing the expected back-transformation during decompression. Another possible reason for preservation of the new phase could be the existence of intrinsic post-shock stresses in the high-pressure assemblage in the oxide clusters. The following parameters of the Birch-Murnaghan equation of state were obtained for synthetic baddeleyite-type TiO2 (10): KT = 304(6) GPa, K′ = 3.9(2), and VO = 16.90(3) cm3/mol. We obtained a molar volume of 16.82(2) cm3/mol for the natural polymorph. This could indicate that the natural sample has a residual stress of about 1.5 GPa, which could stabilize the baddeleyite-type–structured TiO2. Diaplectic SiO2 glass in the same gneisses also shows evidence of residual stress. Raman spectra of diaplectic SiO2 glass inclusions in garnet in the shocked gneisses are very similar to spectra collected from SiO2 glass dynamically densified at 25 GPa (22, 23,28).

The survival of the new polymorph of TiO2 in the shocked gneisses of the Ries crater also provides additional evidence for very low post-shock temperatures. The post-shock temperature must have been far below the upper temperature bound of the orthorhombic α-PbO2 phase (500°C).

A search for post-rutile polymorphs in the Tagamites of Popigai crater in Russia revealed no dense TiO2 phases. This impactite was subjected to post-shock temperatures much in excess of 1000°C, thus leading to extensive melting (29) and back-transformation to polycrystalline rutile.

Rutile in subducted crust is expected to transform to the new phase in the lower regions of Earth's upper mantle or at the transition zone. Because this polymorph is isostructural with baddeleyite, it can accommodate appreciable amounts of ZrO2and HfO2 in solid solution. This may change the fractionation and partitioning processes of these elements at the transition zone (660 km depth) and the lower mantle.

  • * To whom correspondence should be addressed. E-mail: goresy{at}


View Abstract

Navigate This Article