Oxygen and aluminum diffusion in α-Al2O3: How much do we really understand?

https://doi.org/10.1016/j.jeurceramsoc.2007.12.020Get rights and content

Abstract

The major diffusion processes in α-Al2O3 have been reviewed, including oxygen and aluminum lattice diffusion, Doxy and DAl, respectively; oxygen grain boundary diffusion, Db-oxy; and pipe diffusion, Dp. Their relevance to creep of Al2O3 and oxidation of Al2O3 scale-forming alloys is briefly considered.

It is concluded that the answer to the question included in the title is “not a great deal”. Eight major questions and issues have been identified that would constitute fitting subjects for future research:

  • (i)

    What is the nature of the “buffering” that appears to control Doxy?

  • (ii)

    What process(es) are occurring during annealing that eliminate non-Fickian behavior when measuring Doxy?

  • (iii)

    How should the magnitude of the measured activation energies for Doxy be interpreted?

  • (iv)

    Have all the defect types involved in oxygen diffusion been identified?

  • (v)

    Does oxygen diffusion occur by an interstitialcy mechanism?

  • (vi)

    What is the order of magnitude of DAl/Doxy?

  • (vii)

    What is the order of magnitude of Db-Al/Db-oxy?

  • (viii)

    It appears that the activation energy for Db-oxy is greater than that for Doxy. How should this be understood?

Introduction

Self-diffusion of oxygen and aluminum in Al2O3 has been of scientific and technological interest for nearly half a century. Alumina in both single crystal and polycrystalline form is arguably the most important structural ceramic, and knowledge of lattice, grain boundary, and pipe diffusion is crucial for understanding a host of important high temperature processes: plastic deformation of single crystals, diffusional creep and sintering of polycrystals, oxidation of Al2O3-scale forming alloys, etc. (Al2O3 scales provide the very best oxidation resistance for structural alloys in oxidizing environments.) Furthermore, Al2O3 is often considered a model material for many oxide ceramics and other refractory non-metallic substances, and the database of its properties, fabrication, and areas of application is probably unequaled among modern ceramics.

This writer had earlier pointed out1 the difficulty of understanding the activation energy for oxygen lattice diffusion in Al2O3 in the context of traditional point defect chemistry considerations (“Oxygen diffusion in corundum (α-Al2O3): a conundrum”, Philos. Mag. Lett. 79 p. 629 (1999)). In this paper, a tribute to Sir Richard Brook on the occasion of his 70th birthday, this issue is revisited. Some considerations of pipe diffusion and grain boundary diffusion are also included, as well as some brief comments on the applicability of these data to diffusional creep of alumina and oxidation of Al2O3-scale forming alloys. Attention is restricted to the stable corundum polymorph, α-Al2O3, with space group R3¯c.

In addition to the conundrum regarding the activation energy for lattice diffusion1 (the values deduced for the migration energy, about 5 eV, are very much larger than the calculated migration energies, which are of the order of 1.5–2.0 eV2, 3); it is concluded that there are major difficulties in understanding: (i) the processes that occur during annealing of single crystal specimens prior to diffusion experiments that eliminate non-Fickian behavior; (ii) whether the lattice diffusivities of oxygen and aluminum are similar or not; (iii) the magnitude of the pre-exponential and activation energy for oxygen grain boundary diffusion in alumina polycrystals; and (iv) the defects and defect equilibria controlling oxygen diffusion in this scientifically interesting and technologically important material. It turns out that we do not understand a great deal about diffusion in α-Al2O3!

Section snippets

Lattice diffusion

Oxygen self-diffusion in Al2O3 single crystals and polycrystals was first measured in 1960 using gaseous exchange of 18O-enriched oxygen by Oishi and Kingery,4 who annealed Verneuil-grown single crystals spheres, crushed Verneuil rods, and crushed polycrystals which had been sintered at 1900 °C. Some of the latter were made into spherical particles by dropping through a three-phase carbon arc. After the 18O diffusion anneal, the amount of 18O exchange was determined from mass spectrometric

Aluminum lattice diffusion

In contrast to the extensive literature just reviewed dealing with oxygen diffusion in Al2O3, there have been only two reports of aluminum self-diffusion in Al2O3. The first was Paladino and Kingery's early work55 on coarse-grained 130–200 μm polycrystals, and the second a more recent study by Le Gall et al.,21 who used Verneuil single crystals. The paucity of data reflects the difficulty involved in determination of aluminum self-diffusion coefficients. The only available tracer is 26Al, which

Diffusional creep in polycrystalline Al2O3 and oxidation of Al2O3-scale-forming alloys

Some years ago, Cannon et al.56 deduced Dbδ values for hot pressed and sintered MgO-doped Al2O3 from diffusional creep data (Table 1 and Fig. 5); they were convinced that the rate controlling process involved aluminum, rather than oxygen grain boundary diffusion, because the latter was deemed to be very rapid. The comparison of Db-oxyδ data with the creep-derived values of Dbδ (Fig. 5) casts doubt on this assumption.

What else can be said about the atomistic processes dominating creep and

Concluding remarks

We really do not understand a great deal about oxygen and aluminum diffusion in aluminum oxide. The following questions and issues have been identified in this review:

  • (i)

    What is the nature of the buffering that makes Doxy in Al2O3 single crystals so (relatively) insensitive to impurities, even intentionally added aliovalent dopants?

  • (ii)

    What process(es) are occurring during annealing of single crystal specimens that eliminate non-Fickian behavior? (Although the effect of annealing has been demonstrated

Note added in proof

Fielitz et al.61 performed simultaneous 18O and 26Al tracer diffusion measurements in the temperature range 1230–1500 °C on Ti-doped Al2O3 single crystals and demonstrated that DAl is orders of magnitude higher than Doxy. Extrapolation of their data to the temperature range used by Paladino and Kingery (1670–1905 °C) reveals very good agreement with the absolute magnitude of DAl, although the Do and Q parameters differed. Fielitz, et al. report Do and Q values of 7.2 × 10−6 m2/s and 375 kJ/mol,

Acknowledgements

The writer acknowledges long friendship with Sir Richard Brook, useful conversations with J.D. Cawley on non-Fickian oxygen diffusion, D.B. Hovis, who prepared the figures and Table 1, and J. Castaing, M. P. Harmer, and B. Lesage, who provided useful criticism of an earlier draft of this manuscript. K.P.D. Lagerlöf suggested that peroxide ions might be involved in oxygen transport, while G. Wei (Osram Sylvania) suggested that (OH) ions might be important. K.P.R. Reddy encouraged publication of

References (61)

  • C.R.A. Catlow et al.

    Defect energetics in α-Al2O3 and rutile TiO2

    Phys. Rev. B

    (1982)
  • P.W.M. Jacobs et al.

    Defect energies for pure corundum and for corundum doped with transition metal ions

    Philos. Mag.

    (1993)
  • Y. Oishi et al.

    Self-diffusion of oxygen in single crystal and poly-crystalline aluminum oxide

    J. Chem. Phys.

    (1960)
  • Y. Oishi et al.

    Self-diffusion of oxygen in single crystal alumina

    J. Chem. Phys.

    (1980)
  • Y. Oishi et al.

    Effect of surface condition on oxygen self-diffusion coefficients for single crystal Al2O3

    J. Am. Ceram. Soc.

    (1983)
  • D.J. Reed et al.

    Ion-probe measurement of oxygen self-diffusion in single crystal Al2O3

    J. Am. Ceram. Soc.

    (1980)
  • J.D. Cawley et al.

    Oxygen tracer diffusion in single crystal alumina

    J. Am. Ceram. Soc.

    (1991)
  • D. Prot et al.

    Self-diffusion in Al2O3. II. Oxygen diffusion in “undoped” single crystals

    Philos. Mag. A

    (1996)
  • T. Nakagawa et al.

    Oxygen pipe diffusion in sapphire basal dislocations

    J. Ceram. Soc. Jpn.

    (2006)
  • K.P.R. Reddy et al.

    Oxygen diffusion in sapphire

    J. Am. Ceram. Soc.

    (1982)
  • K.P.D. Lagerlöf et al.

    Lattice diffusion kinetics in undoped and impurity-doped sapphire (α-Al2O3): a dislocation loop annealing study

    J. Am. Ceram. Soc.

    (1989)
  • J.D. Cawley et al.

    Dopant distribution in nominally yttrium-doped sapphire

    J. Am. Ceram. Soc.

    (1986)
  • M. Le Gall et al.

    Self-diffusion in α-Al2O3 III oxygen diffusion in single crystals doped with Y2O3

    Philos. Mag.

    (1996)
  • H. Haneda et al.

    Oxygen self-diffusion in Magnesium-or Titanium-doped alumina single crystals

    J. Am. Ceram. Soc.

    (1989)
  • Reddy, K. P. R., Oxygen diffusion in close packed oxides, Ph.D. thesis, Case Western Reserve Univ., Cleveland, OH,...
  • J.H. Harding et al.

    Experiment and theory in alumina

    J. Am. Ceram. Soc.

    (2003)
  • R.H. Doremus

    Oxidations of alloys containing aluminum and diffusion in Al2O3

    J. Appl. Phys.

    (2004)
  • R.H. Doremus

    Diffusion in alumina

    J. Appl. Phys.

    (2006)
  • M. Le Gall et al.

    Self-diffusion in Al2O3. I. Aluminum diffusion in single crystals

    Philos. Mag.

    (1994)
  • B.A. Pint et al.

    Comment on “Oxidation of alloys containing aluminum and diffusion in Al2O3” [J. Appl. Phys. 95, 3217 (2004)]

    J. Appl. Phys.

    (2005)
  • Cited by (246)

    View all citing articles on Scopus
    View full text