Griffinite, Al₂TiO₅: A New Oxide Mineral from Inclusions in Corundum Xenocrysts from the Mount Carmel Area, Israel

Abstract

Griffinite (IMA 2021-110), ideally Al₂TiO₅, is a new mineral from inclusions in corundum xenocrysts from the Mount Carmel area, Israel. It occurs as subhedral crystals, ~1–4 μm in size, together with Zr-rich rutile within a corundum grain. In this study, a mean of eight electron probe microanalyses gave TiO₂ 44.41 (24), Al₂O₃ 55.13 (18), FeO 0.47 (5), and MgO 0.37 (2), totaling 100.38 wt%, which corresponded, on the basis of a total of five oxygen atoms, to (Al_1.97Mg_0.02Fe_0.01)Ti_1.01O_5. Electron back-scatter diffraction studies revealed that griffinite is orthorhombic and in the space group Cmcm, with a = 3.58 (2) Å, b = 9.44 (1) Å, c = 9.65 (1) Å, and V = 326 (2) Å3 with Z = 4. The six strongest calculated powder diffraction lines [d in Å (I/I₀) (hkl)] are 3.347 (100) (110); 2.658 (90) (023); 4.720 (77) (020); 1.903 (57) (043); 1.790 (55) (200); and 1.688 (44) (134). In the crystal structure, Al³⁺ and Ti⁴⁺ are disordered into two distinct distorted octahedra, which form edge-sharing double chains. Griffinite is a high-temperature oxide mineral, formed in melt pockets in corundum-aggregate xenoliths derived from the upper mantle beneath Mount Carmel, Israel. The new mineral is named after William L. Griffin, a geologist at Macquarie University, Australia.

1. Introduction

Small Cretaceous volcanoes exposed on Mount Carmel (Northern Israel) and associated Plio-Pleistocene paleoplacers in the adjacent Kishon Valley contain xenoliths (up to cm size) comprising aggregates of corundum crystals with intercrystalline to interstitial melt pockets. The melt pockets and individual mineral inclusions in corundum contain a remarkable assemblage of minerals crystallized under highly reducing conditions [1].

During a study of melt inclusions in the corundum xenocrysts, we have identified seven IMA-approved new minerals since 2021: griffinite (Al₂TiO₅), magnéliite (Ti³⁺₂Ti⁴⁺₂O₇), ziroite (ZrO₂), sassite (Ti³⁺₂Ti⁴⁺O₅), mizraite-(Ce) (Ce(Al₁₁Mg)O₁₉), toledoite (TiFeSi), and yeite (TiSi) [2,3,4,5,6,7,8]. Reported here is the new mineral griffinite, the first natural occurrence of Al₂TiO₅ in a corundum xenocryst from Mount Carmel (Figure 1), providing more insights into the origin of high-temperature minerals from the upper mantle.

The new oxide mineral has been named in honor of William L. Griffin (b. 1941), a geologist at Macquarie University, Australia, for his outstanding contributions to mineralogy, petrology, and geochemistry of the deep crust and lithospheric mantle, including intense investigations of materials from the Mount Carmel area. Griffinite was approved as a new mineral by the IMA-CNMNC (2021-110) [2]. Noteworthily, synthetic ceramic Al₂TiO₅ is known in the scientific literature as an excellent thermal-shock-resistant material referred to as tialite and tielieite (e.g., [9,10]). Al₂TiO₅ has a very low thermal expansion (1.5 × 10⁻⁶ K⁻¹), low Young’s modulus, and high-temperature resistance (melting point 1860 ± 10 °C; [11]).

2. Materials and Methods

The corundum xenolith in which we found the type griffinite occurs in a Plio-Pleistocene placer gemstone deposit in the Kishon River, which drains Mount Carmel and the adjacent Yizre’el Valley and enters the sea near Haifa in Northern Israel [12]. It is part of a xenolith assemblage that includes aggregates of skeletal corundum crystals with melt pockets containing reduced mineral assemblages [12,13,14,15,16].

The material used in the present study came from both volcanic centers on Mount Carmel, N. Israel, and alluvial deposits derived from these and Miocene to Pliocene basalts exposed in the Yisre’el Valley. The holotype material in one corundum grain (No. 198c) from Mount Carmel mount Corundum-18-1 is deposited in the mineralogy collection of the Università degli Studi di Milano, Via Mangiagalli, 34—20133 Milano, Italy, registration number MCMGPG-H2022-002.

A ZEISS 1550VP Field-Emission Scanning Electron Microscope (SEM) (ZEISS Group, Jena, Germany) with an Oxford X-Max energy-dispersive spectroscopy (EDS) (Oxford Instruments, Abingdon, UK) device was used for backscatter electron (BSE) imaging and fast elemental analysis. A preliminary chemical analysis using EDS performed on the crystal fragment used for the structural study did not indicate the presence of elements (Z > 9) other than Ti, Al, O, and minor Fe and Mg. Electron probe microanalyses (EPMA) were carried out using a JEOL 8200 Superprobe (JEOL Ltd., Tokyo, Japan) (WDS mode, 15 kV, 10 nA, focused beam = ~150 nm in diameter) on griffinite crystals in polished corundum grain 198c with a 25 nm carbon coating, as shown in Figure 1. The following lines were used: Ti Kα, Al Kα, Mg Kα, and Fe Kα. The standards employed were synthetic TiO₂ (Ti), anorthite (Al), fayalite (Fe), and forsterite (Mg). Quantitative elemental microanalyses were processed with the CITZAF correction procedure [17]. The crystal fragment was found to be homogeneous within analytical error. The analytical results of griffinite are given in

Table 1. Electron microprobe analysis (wt% of oxides) of griffinite.

Constituent (wt%)	Mean (n = 8)	Range	SD	Probe Standard
TiO2	44.41	44.15–44.71	0.24	TiO2
Al2O3	55.13	54.90–55.34	0.18	anorthite
FeO	0.47	0.40–0.58	0.05	fayalite
MgO	0.37	0.36–0.40	0.02	forsterite
Total	100.38

.

Conventional X-ray studies could not be carried out because of the small crystal size. Electron backscatter diffraction (EBSD) analyses were performed using the methods described in [18,19] for micron-sized new mineral studies. An HKL EBSD system on a ZEISS 1550VP Field-Emission SEM was operated at 20 kV and 6 nA in focused beam mode with a 70°-tilted stage and in a variable pressure mode (25 Pa) on griffinite crystals in polished corundum grain 198c without any coating. The focused electron beam was several nanometers in diameter. The spatial resolution for diffracted backscatter electrons was ~30 nm in size [20]. The EBSD system was calibrated using a single-crystal silicon standard. Structural information and cell constants were obtained by matching the experimental EBSD patterns with structures of Al-Ti-O and Ti-O phases from the ICSD [21,22].

3. Results

Griffinite in the type material occurs as subhedral crystals, ~1–4 μm in size, together with Zr-rich rutile within corundum grain 198c in the polished 1-inch mount Corundum-18-1 (Figure 1). This association forms pseudomorphs and appears to reflect the oxidation breakdown of carmeltazite (ZrAl₂Ti₄O₁₁). Other inclusions in this corundum grain contain yeite (TiSi), baddeleyite, hibonite, osbornite (TiN), khamrabaevite (TiC), Ti,Al,Zr-oxide, and zirconolite. Griffinite is transparent and, given the size, most of the physical and optical properties could not be obtained. Griffinite has also been observed to crystallize directly from melts trapped between corundum grains.

The average chemical compositions (eight analyses of different spots on four larger crystals in the same inclusion in Figure 1) together with the wt% ranges of elements are reported in Table 1. Fe and Ti were considered di- and tetravalent, respectively. On the basis of five oxygen atoms, the empirical formula of griffinite is (Al_1.97Mg_0.02Fe_0.01)Ti_1.01O₅. The simplified ideal formula is (Al,Mg,Fe,Ti)₂TiO₅, and the ideal formula is Al₂TiO₅ (Z = 4), which requires Al₂O₃ 56.07 and TiO₂ 43.93, totaling 100 wt%.

The EBSD patterns of griffinite can be indexed only by the Cmcm pseudobrookite-type structure and match to the cell values reported for synthetic Al₂TiO₅ cells by [21,22] (Figure 2), with a mean angular deviation of 0.32°–0.38°, revealing the following cell parameters: a = 3.58(2) Å, b = 9.44 (1) Å, c = 9.65 (1) Å, and V = 326 (2) ų with Z = 4.

X-ray powder diffraction data (

Table 2. Calculated X-ray powder diffraction data for griffinite (I_rel > 1). Reflections with I_rel > 25% are evidenced in bold.

h	k	l	d (Ả)	Irel
0	0	2	4.825	1
0	2	0	4.720	77
0	2	1	4.240	2
0	2	2	3.374	24
1	1	0	3.347	100
1	1	1	3.163	39
0	2	3	2.658	90
0	0	4	2.413	1
1	3	0	2.364	13
0	4	0	2.360	1
1	1	3	2.319	15
0	2	4	2.148	26
1	3	2	2.123	6
0	4	2	2.120	17
0	4	3	1.903	57
2	0	0	1.790	55
0	2	5	1.786	5
1	3	4	1.688	44
2	2	0	1.674	9
1	1	5	1.672	1
1	5	0	1.670	2
0	0	6	1.608	21
2	2	2	1.581	5
1	5	2	1.578	19
0	6	0	1.573	8
0	6	1	1.553	7
0	2	6	1.522	10
0	6	2	1.496	1
1	3	5	1.495	7
0	4	5	1.494	1
2	2	3	1.485	32
1	5	3	1.482	26
1	1	6	1.450	4
2	0	4	1.438	1
2	4	0	1.426	1
2	2	4	1.375	11
1	5	4	1.373	8
2	4	2	1.368	8
1	3	6	1.330	7
0	2	7	1.323	4
2	4	3	1.304	30
1	1	7	1.275	13
2	2	5	1.265	3
1	7	0	1.262	1
1	7	1	1.251	14
1	7	2	1.221	10
0	6	5	1.220	3
2	0	6	1.196	14
3	1	0	1.184	4
2	6	0	1.182	6

, in Å for CuKα, Bragg–Brentano geometry) were calculated with the unit cell parameters above, the crystallographic data of synthetic Al₂TiO₅ [23], and the empirical formula, using Powder Cell version 2.4.

4. Discussion

Griffinite is a new member of the pseudobrookite group. Griffinite (Al₂TiO₅) is the second Al-Ti-oxide mineral, joining machiite (Al₂Ti₃O₉) [24]. Both are high-temperature minerals. Machiite is an ultrarefractory phase from the solar nebula. There is also a chance that griffinite might occur as a refractory phase in the solar nebula.

By analogy with synthetic Al₂TiO₅, griffinite is isostructural with pseudobrookite (Fe³⁺₂TiO₅), armalcolite ((Fe²⁺,Mg)Ti₂O₅), and sassite ((Ti³⁺₂Ti⁴⁺O₅) [5]). In the synthetic Al₂TiO₅ structure, Al³⁺ and Ti⁴⁺ are disordered into two distinct distorted octahedra (Wyckoff positions 4c and 8f). Such (Al,Ti)O₆ octahedra form edge-sharing double chains [21,22]. The crystal structure is shown in Figure 3. Studies on synthetic Al₂TiO₅ indicate that griffinite may show significant cation disorder among octahedra [23,25]. Synthetic Al₂TiO₅ is thermodynamically stable at T > 1280 °C [26]. At lower temperatures, it decomposes to corundum + rutile. The formation temperature of synthetic Al₂TiO₅ is lowered to <1280° in the presence of Mg [27,28] and low diffusion. However, the instability temperature range is restricted to T > 1100 °C [29]. The process of decomposition of synthetic Al₂TiO₅ is also strongly affected by the oxygen partial pressure [29].

Upon heating, synthetic Al₂TiO₅ tends to contract in the direction of its stronger bonding apex-sharing oxygens, i.e., the a-axis, meaning negative thermal expansion in this direction [23] showing α_a = −2.38 (32) × 10⁻⁶ K⁻¹ [22].

Griffinite is a high-temperature (>1300–ca. 1150 °C) phase crystallized from melts trapped in voids of corundum crystals [12,31]. The mineralogical assemblage demonstrates an oxygen fugacity (fO₂) below the levels usually observed in Earth’s crust or upper mantle (IW to IW-9; [32]). The extreme reduction observed requires a hydrogen-dominated environment, as proved by the presence of natural hydrides and hydrogen in vacancies in hibonite [14,32,33,34], and reflects the reduction and desilication of differentiated syenitic melts, through interaction with mantle-derived CH₄ + H₂ fluids [32,35,36]. The desilication of the melts as well as the separation of Si⁰ melts and Fe-Ti-Si melts drive the supersaturation of Al₂O₃ in the silicate melts and lead to the crystallization of large corundum crystals. Other minerals present in the melt inclusions in corundum xenocrysts from the Mount Carmel area are reported in Figure 4.

Among the corundum aggregates, three types of paragenesis can be recognized [31].

Crn-A: Strongly Ti-zoned hopper crystals [36]. The trapped melts are Ca-Mg-Al silicates with high sulfur content and incompatible elements. The phase assemblages (Ti³⁺ in oxides and Ti²⁺ in carbides and borides) reflect fO₂ ≤ IW-6.

Crn-B: Large Ti-poor corundum crystals, with trapped pockets exhibiting small amounts of Ca-Al-Mg silicate glass, typically high in REE, Zr, Th, U, and other incompatible elements. Ti is present as both Ti³⁺ and Ti⁴⁺.

Crn-C: Similar to Crn-B but with lower Ti contents. Rare Ca-Al-Na-K silicate glasses are rich in LREE and Ba. The presence of more abundant Ti⁴⁺ phases indicates a higher mean fO² than in Crn-A and Crn-B.

The identification of the different valence states of the diverse phases in the three parageneses indicates that Crn-B and Crn-C are cumulates from immiscible Fe-rich melts (subsequently depleted in Fe by the separation of Fe-Ti silicide melts; [1]) and Si-Al-Na-K melts, respectively. It is likely that these melts have similar but divergent histories, separated into volumes that were affected to different extents by interaction with the reducing (CH₄ + H₂) fluids before being entrained in the host basalt.