Oxides and Hydroxides of Aluminumepsc511.wustl.edu/Aluminum_Oxides_Alcoa1987.pdf · Preface One hundred years after the invention of the Bayer Process for the production of aluminum

Oxides and Hydroxidesof Aluminum

Alcoa Technical Paper No. 19, Revised

Karl Wefers

Chanakya Misra

Alcoa Laboratories1987

Copytight” Aluminum Company of America

., _

Preface

One hundred years after the invention of the Bayer Process for the production of aluminum oxide,new uses are still being found for this material. in advanced ceramics, separations and catalysts,

the manufacture of integrated circuits as well as in composite systems.

This revised edition of Alcoa Laboratories Technical Pa~r No. 19 includes discussions of how ourunderstanding of the stmcture and properties of aluminum oxides has progressed since the originalwas published in 1972. The first Technical Paper on this subject was published in 1953 by

Dr. Allen S. Russell.

1 believe it is entirely appropriate for this revision to be published in the year in which both Alcoa

and the Bayer Process conclude their first century.

Dr. Peter R. Bridenbaugh

!

Acknowledgments

The authors wish to thank their colleagues of Alcoa Laboratories who contributed to this monograph1 by supplying data, photomicrographs, and many helpful suggestions: A. Pearson, J, J. Ptasienski,

D. R. Micholas, W. T. Evans, G. A. Nitowski, L. F. Wiesennan, C. Bates, and D. Tzeng. Special/ thanks go to D. E. Thomas and his associates for their excellent work in reproduction and printing

~ and Mrs. M. Weber for her diligent and patient typing of the many revisions of the manuscript.

Tbe encouragement and support by the directors of Alcoa Laboratories is great fully acknowledged; sois the critical review of the manuscript by Mr. W. B. Frank of Alcoa, Dr. A. S. Russell (Alcoa, ret. )

I and Dr. J. Dillard of Virginia Tech. Mr. R. S. Zhu and Dr. R. L. Snyder of Alfred Universitysupplied the computer-generated stmcture models. We thank them, and we are indebted to the manycolleagues in industry and academia who responded so well to the first edition of TP 19 and

encouraged us to undertake this revision.

Karl Josef Bayer introduced his elegant process for the industrial production of afuminum oxide 100 years ago, in 1887. Sincethen, the basic concept of the process - hydrothermal digest of bauxite, crystallization of aluminum trihydroxide from asupersaturated sodium aluminate solution, thermal conversion of the trihydroxide to anhydrous aluminum oxide has remainedunchanged.

More than 30 million tons of aluminum oxide we produced each yew, most of it being used for the smcltin8 of aluminum. Tberemaining part of the annual production finds application in areas which utilize the high melting point, excellent mechanicalstrength, electrical resistivity, or chemical inertness of aluminum oxide. Refractory, we= resistant ceramic parts, equipment forchemical processing, and high voltage insulators are the classical products. Newer applications include substrates for integrated

circuits and reinforcing fibers for pot ymer and metal-based composites.

Surface oxides on aluminum and its alloys play an important role in corrosion and protection from corrosion, in adhesive bonding,coating, and laminating of metal. They are also applied in the manufacture of capacitors and electronic devices such as MOM(Metal Oxide Metal) transistors.

The unique surface properties of the stmcturally and stoichiometrically disordered transition aluminas are utilized in catalysis and

separation technology.

Many of these applications were driven by an increasing scientific understanding of the chemical, structural, and surface

properties of aluminum hydroxides and oxides. The foundations for this fascinating field of science and engineering were laid by ageneration of researchers, many of whom were these authors’ teachers, mentors, or older colleagues. F. C. Fr~, H. Ginsberg,A. S. Russell, J. Newsome, H. Stumpf are some of the names that come to mind.

The authors were introduced to the field of alumina science at a time when refined techniques of stmctural, thermal, and electronmicroscopic analysis had made it possible to correlate chemical and physical properties with CIYSM structure andmicromorphology at a level of a few nanometers. Recent advances in theoretical solid state chemistry and, in the past decade, the

emergence of surface probes ~rmitting the study of monolayer or small clusters of atoms, further increased our understanding ofthe effect of non-ideal stmcture and composition on functional properties of aluminas.

Because of the authors’ personal experience and interest, the following monograph is somewhat biased towards a stmcturalviewpoint. An attempt has ken made to demonstrate the structural and chemical principles that xe common to the many forms in

which oxides and hydroxides of aluminum occur. Rather than trying to give an exhaustive review of the vast literature onaluminas, tbe authors have used the concept of stmcture-pro~rty relationships to deal with an extensive and still very active weaof inorganic chemistry.

Oxides and Hydroxides of Aluminum

TABLE OF CONTENTS

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Wlstory and Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2. Properties of Aluminum Hydroxides and Oxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.1

2.11

2.12

2.13

2.14

2.2

2.21

2.22

2.23

2.24

2.25

2.32.312.322.33

2.4

2.412.4112.4122.413

2.4142.4152.416

2.4172.4182.4192.4110

2.42

2.5

Alumina Gels and Gelatinous Aluminas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Gelatinous Aluminas Formed by Acid-Base Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Gelatinous Aluminas Prepared by Otber Chemical Routes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

The Structural Evolution of Gelatinous Alumina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Gelatinous Boehmite or Pseudoboehmite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Aluminum Tribydroxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Gibbsite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Bayerite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Nordstrandite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Doyleite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Interrelationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Aluminum Oxide Hydroxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Boehmite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Diaspora . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .K1-Alz030r Tobdite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Aluminum Oxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Corundum, a-Alz03 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Crystal Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Thermal Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Diffusion and Related Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Electrical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Dielectric Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Optical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Magnetic Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Mechanical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Surface Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Aluminum Suboxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Appendix to Section 2- IR Spectra and X-Ray Data

3. Binary and Ternary Alumina Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.13.113.12

3.23.213.223.23

3.24

3.3

The A1203-H20 System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Phase Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Volubility of A1203 in Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

The Na20-A1203 System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Phases Occurring in the System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Sodium Aluminate NaAIOz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Sodium Beta Aluminas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Beta Aluminas as Solid Electrolytes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

The Na20-A1203-H20 System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

iii

1

2

3’2458

101011121314

15151717

1818182020232425262s282935

36

363637

3939394042

43

.. .111

Alcoa Laboratories

TABLE OF CONTENTS (continued)

-

4. Phase Relations of Aluminum Hydroxides and Oxides UnderNonequilibrium Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.1

4.24.214.22

4.34.314.324.33

4.44.414.424.43

4.54.514.524.53

Properties and Applications of Calcined Hydroxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Thermal Decompositions of Aluminum Trihydroxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Gibhsite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bayerite and Nordstrandite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Decomposition of Aluminum Oxide Hydroxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Boehmite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Diaspore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Other Methods for Preparing Transition Aluminas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Structures of Transition Aluminas. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Chi and Kappa Alumina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Etaand Theta Alumina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Gamma and Delta Alumina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Texture and Surface Properties of Transition Aluminas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Development of Porosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Relationships Between Texture and Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Properties of Active Alumina Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5. Surface Oxides on Aluminum Metal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.1 Oxides Formed by Solid-Gas Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.2 Tbe Reaction of Aluminum Surfaces with Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.3 Oxides Formed by Anodic Polarization of Aluminum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5.31 Non-Porous Oxide Films . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5.32 Porous Anodic Oxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5.321 Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.322 Applications of Porous Anodic Films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6. Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

No.

2.12.22.32.42.52.62.72.82.92.10

iv

Title-—

2.12.112.122.132.132.132.132.212.212.25

Volubility of AI(OH)3 as Function of pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Aging of Aluminum Hydroxide Gel, X-Ray Pattern . . . . . . . . . . . . . . . . . . . . . .Fibrillar Gelatinous Aluminum Hydroxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Formation of Polynuclear Rings by Condensation . . . . . . . . . . . . . . . . . . . . . . . .DeprotonationlCondensationReactionsLeadingtoChains . . . . . . . . . . . . . . . . .Tabular Gibbsite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bayerite Somatoid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Aggregates of Technical Gibbsite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Structure of Gibbsite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Layer. Stacking in Al(OH)3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

46

46 (

464648

48485051

51525253

54546060

64

64

66

6868696973

75

3346778

1011


i

I

I

,,

I

No.

2.112.122.132.142.152.162.172.182.192.202.212.222.23

3.13.23.33.43.53.63.73.8

4.14.24.34.44.54.64.74.84.94.104.114.124.134.144.154.164.174.18

5.15.25.35.45.55.65.75.85.9A-C5.105.11

=

2.312.312.322.4112.4112.4132.4132.4132.4132.4142.4152.52.5

3.113.123.133.213.233.33.33.3

4.14.214.214.214.214.214.33/4.44.514.514.514.514.514.514.514.514.534.534.53

5.15.15.25.25.315.315.3215.3215.3215.3215.322

FIGURES (continued)

Title 1

Boehmite Crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Structures of Boehmite and Diaspore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Diaspore Crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Corundum Crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Structure of Corundum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Cell Parameters of &-A1203 as a Function of Temperature . . . . . . . . . . . . . . . .

Specific Heat of a-A1203 as a Function of Temperature . . . . . . . . . . . . . . . . . . .

Thermal Expansion of a-A1203 as a Function of Temperature . . . . . . . . . . . . .Thermal Conductivity of a-A1203 as a Function of Temperature . . . . . . . . . . .

Self Diffusion of Al and O in A1Z03 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Electrical Conductivity of a-A1203 as a Function of Temperature . . . . . . . . . .

X-Ray Diagrams of Aluminum Oxide and Hydroxides . . . . . . . . . . . . . . . . . . . .

Infrared Spectra of Aluminum Hydroxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Phase Diagram A1203-Hz0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Soluhility of Gibbsite in Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Stability Ranges of Aluminum Hydroxy Complexes in Water . . . . . . . . . . . . . .

Na20-A1203 Pbase Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Structures of ~ and ~ Aluminas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Phase Diagram Na20-A1203-H20 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Volubility of A1203 as a Function of Temperature and Concentration of Na20Solution Isotherms in the System Na20-A1203-H20 . . . . . . . . . . . . . . . . . . . . . . .

Transformation Sequence AI(OH)3 + AliOj . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Acicular Gibbsite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Acicular Corundum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Surface Area vs. Temperature of Calcined Gibbsite . . . . . . . . . . . . . . . . . . . . . .

DTA - Curves of Heated Hydroxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Tabular Alumina from Gibbsite Calcined with Fluoride . . . . . . . . . . . . . . . . . . .X-Ray Diffraction Patterns of Transition Aluminas . . . . . . . . . . . . . . . . . . . . . . .Chi Alumina lo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Chi Alumina llc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Eti Alumina llc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Kappa Alumina from AcicuIar Gibbsite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Kappa Alumina llcof Gibbsite, Reordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Alpha Alumina from Acicular Glbbsite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Alpha Alumina, Primary Crystals and Pores . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Gamma Alumina from Boebmite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Configurations of Surface Hydroxyl Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Five Coordination of Surface Hydroxyl Groups . . . . . . . . . . . . . . . . . . . . . . . . .Adsorption of First and Second Layer of Water on A1203 . . . . . . . . . . . . . . . . .

Growth and Transport Mechanisms of Thermal Oxides, Schematic . . . . . . . . .Progressive Recrystallization of Amorphous Surface Oxide on Aluminum . . .Multilayered Hydroxide Film on Aluminum . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Duplex Film on Al Formed in Boiling Water . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Voltage-Current Relationships in Anodizing Al . . . . . . . . . . . . . . . . . . . . . . . . . . .

“Tunnel’’E tchedAlw ithBarrierOxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Initial Cell Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Texture of Porous Anodtc Oxide. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Cross Section, Surface and Fragments of Porous Anodic Oxide . . . . . . . . . . . .Cells of Anodic Oxide Advancing into Metal . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Cross Section of Sealed Anodic Oxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Page No.

15161718191922242424263435

3638394041444545

474848494950525656575858595960616262

6566676868697071

72, 737374

v

Alcoa Laboratories (

Table No.

1.1

2.1

2.2

2.3

2.4

2.5

2.6

2.7

2.8

2.9

2.10

2.11

3.1

3.2

4.1

4.2

4.3

=

1

2.21

2.21

2.21

2.21

2.413

2.414

2.416

2.418

2.419

2.5

2.5

3.12

3.24

4.4

4.4

4.53

INDEX OF TABLES

Title—

Comparison of Nomenclature

Mineralogical Properties of Oxides and Hydroxides

Structural Properties

Thermodynamic Properties

Values of Coefficients for Specific Heat Equations

Selected Values of Thermal Properties of A1203

Possible Disorder Mechanisms in AlzOA

Dielectric Properties of Corundum

Magnetic Properties

Mechanical Properties of Corundum

Infrared Absorption Frequencies

X-ray Diffraction Data

Ion Activity Products

Structural Properties of ~-Aluminas

X-ray Diffraction Data for Transition Aluminas

Structural Properties of Transition Aluminas

IR Frequencies of Surface Hydroxyl Groups

-

2

12

13

21

22

23

25

27

28

29, 30

32

35

38

42

54

55

63

vi


1. History and Nomenclature

Materials with a styptic or astringent pro~rty were called“alumen” by tbe Remans. These may have included impureforms botb of afuminum sulfate and alum, which occurnaturally in volcanic areas, Al”men seems to be the source of

tbe word alumina (Bec@ann, 1846).

The recognition of alumina evolved gradually as the

composition of afum became ktter understocmi. Hoffman in1722 held that the base of alum was a distinct “earth,” and Pott(1754) called the base “thonichte erde” or “terre argilleuse”

(clay earth). Marggraf in 1754 showed that the earth of alumwas a distinct substance, that existed in natural clays, and that

it could k extracted by sulfuric acid. His memoirs gave rise to

the term “argil pur” as the name of the earthy base of alum. DeMorveau in 1786 argued that since alum was designated “se]

alumineux ,“ the pro~r name for the base should be “alumine.”This latter term has been Anglicized to alumina. In 1821,

Berthier described a sediment rich in alumina which he foundin the vicinity of Les Baux, a small town near Aries in theProvence district of France, This sediment, named bauxite, was

considered an aluminum mineral of the compositionA1203 2H20, with some iron. Not before the end of thenineteeth century was bauxite recognized as a sedimentq rockcontaining aluminum hydroxide and oxide hydroxide, as wellas various amounts of iron minerals, aluminum silicates, andtitanium dioxide.

A mineral from India, having the composition A1203, wasdescribed in 1798 by Greville under the name corundum. Themineral dlaspore was described by Hauy in 1801; it was

analyzed by Vauquelin in 1802 and shown to be A1203 H20.Hauy (1801) named it diaspore from the Greek for “scatter,”because it flew apart when heated. In 1805, Davy analyzed amineral and named it hydrargillite after the Greek words hyder

and arg ylles for water and clay. Later, however, other workersshowed this mineral to be aluminum phosphate. Dewey in 1820found a mineral he called gibbsite in honor of G. Gibbs, anAmerican mineralogist. Analyses by Torrey in 1822

corresponded with the formula Al(OH)j or A1203 3H20. Thename hydrargillite was applied later to a similar mineral foundin the Urals; both names are still employed, but gibbsite is

prefemed in the USA.

Bohm and Niclassen (1924) identified another aluminummineral by X-rays, Bobm (1925) found it to k an isomer of

diaspore and showed a purified bauxite from Les Baux toconsist predominantly of this phase, De Lapparent (1927)

suggested that the mineral k called Whmit. B6hm ( 1925) alsodiscovered an alufinum compound isomeric with gibbsite.Fricke (1928) suggested the name bayerite for this materialkcause he thought it was the product of the Bayer process. Intbe next year, Fricke (1929) recognized his error, but the namehas been continued. Gibbsite is produced in the Bayer process.

Rankin and Merwin in 1916 assigned the prefix beta to a

high-temprature alumina that, according to later experience(Ridgway et al. 1936), contains alkali or alkaline earth atoms.Corundum was differentiated as alpha alumina. As severalforms of “beta alumina” have ken distinguished, tbe foreigncation was made part of the name, i.e., sodium kta alumina.

Barlett (1932) assigned the name zeta to the phase formed withlithium.

Haber in 1925 divided the aluminum and ferric hydroxides into

two series, alpha and gamma. The apparent failure of thissystem to differentiate the aluminum trihydroxides prompted arenaming of individual phases (Edwards et al. 1930). Thesymbol alpha was applied to the form more abundant in nature.

In this nomenclature, bayerite was beta trihydrate, and thephases gibbsite, Mhmit (spelled “boehmite” in the UnitedStates), and diaspore were designated alpha trihydrate, alphamonoh ydrate, and beta monoh ydrate, respective y. Weiser and

Milligan (1934) and Weiser (1935) later adopted the Haberseries and classified bayerite as alpha uihydroxide.

The designation gamma was given originally by Ulrich (1925)

to an undescribed alumina. This term has been used in manycases for all the alumina transition forms encountered in theIow-temprature calcination of aluminum compounds and in theoxidation of aluminum. Stumpf et al. (1950) restricted thename gamma to the product obtained in the dehydrationsequence of boehmite at >800K.

As new forms have been identified, they have been assigned

Greek letters. Tbe uniqueness of many of these has beendisproved.

Table 1 shows the nomenclature systems for those aluminasmost frequently confused in tbe literature. Obviously, a

universal standard nomenclature is highly desirable, An attemptin this direction was made in 1957 when, in West Germany, asymposium on “alumina” nomenclature was held. The resultsof this meeting were reported by Ginsberg et al, (1957). Thefollowing nomenclature, was suggested

a. Use of the chemically correct term “hydroxide” instead of“hydrate,” namely: aluminum trihydroxide, instead of

alumina trihydrate; aluminum oxide hydroxide, instead ofalumina monohydrate.

b, Use of the nomenclature of Alcoa for the transition

aluminas, but avoiding the term “phase”, use “form”instead, e.g., “chi form”, “kappa form,” “gamma form”.

c. In honor of Van Nordstrand, use the name nordstrandite forthe trihydroxide discovered by him.

In the internationally accepted crystallographic nomenclature

the prefix a is generally applied to hexagonal close packed and

1

Alcoa Laboratories

Table 1.1

Comparison of Nomenclatures

Accepted

Mineral Chemical CrystallographicName

GibbsiteHydrargillite

Bayerite

Nordstrandite

Dlaspore

Composition Designation

Aluminum y-Al(OH)3

trihydroxide

Aluminum a-Al(OH)3

trihydroxide

Aluminum AI(OH)3

trihydroxide

Aluminum Y-AIOOH

oxidehydroxide

Aluminum a-AIOOH

oxidehydroxide

Aluminum CK-A1203

oxide

Alcoa (1930)

Alpha aluminatrihydrate

Beta alumina

trihydrate

Alpha aluminamonohydrate

Beta aluminamonohydrate

Alpha alumina

related stmctures, They-phases have cubic close packed

lattices or structural elements of this symmetry. This system ofclassification is used in the USA for iron and manganesecompounds isostructural with the aluminum oxides andhydroxides. Itisalso applied tothecubic V,and the hexagonal

(bigonal) a-alumina.

Toeliminatk fhe existing inconsistencies we will use themineral names and the crystallographic nomenclature

fhroughouttbis monograph.

2. Properties of Aluminum Hydroxides and Oxides

The chapters of this section are arranged according to thesequence of phases occurring in the alumina-water system withincreasing temperature or crystalline order.

2.1 Alumina Gels and Gelatinous Aiuminss

The term alumina gel covers a wide variety of two-phasesystems in which colloidal aluminum hydroxide or oxide

hydroxide isthepredominant solid phase. The second phasecan be water, a mixture of water and an organic solvent, or air

2

(aero-gel). Depnding onthemethod ofpreparation, the solidmay & present as discrete particles ranging in size from a few ,<

nanometers to micrometers, or it can form three-dimensional,~lymeric networks in which domains of solid are linked to

(

(

others via chemical bonds. Structural order ranges from X-ray

indifferent (amorphous*) to crystalline. Composition (i. e., theoxide-to-water ratio and amount of impurities) varies with the

conditions of preparation, as does thes~cific surface area.

The following chapters will emphasize the stmcture and

composition of the solid phase, the gelatinous aluminas.However, their proprfies canonly be fully understood in the

context of the chemistw and texture of the two-phase systems,the gels.

Gelatinous aluminas have ken studied for a number ofreasons. Colloidal aluminum hydroxides play an important rolein the chemistry of soils. Certain alumina gels strongly adsorbacids. Because aluminas are not toxic to humans, they ag

widely used as antacids. Also, since particle size, surface area,and water content of gels can be varied over a wide range, they

we useful precursors for the manufacture of aluminas forcatalysis md adsorption.

Polymeric network and particulate gels can be prepared bycontrolled hydrolysis of aluminum alkoxides dissolved inorganic solvents. These gels are used for the preparation ofporous ceramic hodles, thin coatings, and granulm alumina ofhigh purity.

Inorganic chemists andcolloid andsoilscicntists have

investigated the precipitation and aging khaviorof gelatinousaluminas for more fban40 years. G. C. Bye, R. Fricke,H. Ginsberg, P. H. Hsu, D. Pa@e, K. S. W. Sing,

S. Teichner, K. Torkar, H. B. Weiser, and their collaboratorsrepresent the most prominent working groups. Ceramists andpharmaceutical chemists have entered this tield more recently.E. Matijevi6 andhls coworkers extensively studied the

preparation of monodisprsed, colloidal aluminum hydroxidesby controlled precipitation and aging of alumina gels.

S. L. Hem, J. C. White, andcollaborator sinvestigatedfhecomposition and stmcture of gels and their effect on acidadsorption, neutralization, and ion exchange. B. Yoldas was

one of the early workers in the rapidly expanding field ofcontrolled hydrolysis of aluminum organic compounds for thepreparation of ceramic alumina bodies and thin films.

2.11 Gelatinous Aluminss Formsdby Acid-Bsse Reactions

Aluminum hydroxides xe amphoteric. They are soluble in

strong acids and strong bases. In aqueous solutions of

●These terms i.dicatc lackof sufficientlong-rangeorder to prcduce adiffraction pattern with X-ray$in tbe O,l-O.2nm wavelengthrange. Few solidsxc ~ly amo~ho.s i.e., witiout any shon-rmgesncmral .rderat all.


i)

intermediate pH their volubility is very low (see Chapter 3. 12)Figure 1 schematically shows the volubility of A1(OH)3 as afunction of pH. Because of the steep slope of tbe volubilitycurve, a small change in pH can cause considerablesuperraturatlon and, consequently, rapid precipitation. As aresult, the first precipitate is generally of colloidal size and of

low crystalline order. Colloidal aluminum hydroxide ishydrophilic and easily coagulates to gels.

Several factors determine the degree of crystalline order,

particle size, and chemical composition of gelatinous aluminastemperature, rate of precipitation, final pH, ionic composition,concentration of starting solutions and time of aging(Figure 2.2).

Rapid neutralization of aluminum salt solutions with basesleads to gels rich in water which contain variable amounts ofresidual acid anions, The water content may be as high as

5 moles per mole of A1203.

A first attempt to classify the solid phases in gels precipitatedfrom aluminum salt solutions was made by WiOstitter et al.

(1925). Three ty~s were distinguished: Ca, CB, and Cy.

GA 202,6.2

Oh

4h (Gelatinous Boehmite)

8h

12h

16h

I

7 I I I I I I

6‘“u

Aging of Aluminum Hydroxide Gel,

.E 5pH 9, 300K, X-Ray Patterns

= Figure 2.2~

E4= According to these authors and later investigators Kraut et al.=E

(1942), Fricke and Schmti (1948), Souza Santos et al. (1953),

l.iand Watson et al. (1955, 1957) Ca is X-ray indifferent, Ittransforms in a few hours to C13 which consists of fibrils. The

53 diffraction pattern of CD shows broad lines of boehmite, while~x

loss on ignition indicates 2.2 moles H20 per mole A1203. C13 is

g2

slowly converted to CT, a mixture of gibbsite and bayerite.

z The sequence, X-ray indifferent poorly crystallized boehmite

crystalline trihydroxide also occurs in the precipitates formed

1 by rapid neutralization of alkaline aluminate solutions.

Gelatinous aluminas prepmed by neutralization of concentrated

.0 I I I I I I aluminum salt solutions at temperature below 29OK have low

0123456789 101112Crystalline order and very small particle size. The rate of aging

(i.e., transformation of the solid phase to ordered aluminumhydroxide) is also very slow under these temperature and pH

Volubility of AI (OH)3 as a Function of pH conditions. Anions adsorbed on the solid retard the aging

(schematic) process. Sema, White, and Hem (1977) repofled the results of

Figure 2.1 an investigation of tbe interaction between anions and

3

Alcoa Laboratories(

aluminum hydroxide gels and discussed previous work by Hsuand Bates (1964), Hem et al. (1970), Matijevit and Stryker(1966), and others. They infemed from infrared analysis that

gels precipitated in the presence of nitrate, sulfate, andcarbonate ions behave “as a positively charged polymericmateri al.” While the interaction of nitrate appeared to bepredominantly electrostatic, sulfate and carbonate were found to

coordinate with aluminum ions. The carbonate ion has thestrongest retarding effect on aging of alumina gels.

Gels in which the initial molar ratio of carbonate to aluminumis as high as 1.2 can be prepared by reacting sodium carbonatesolution with aluminum nitrate. Repeated washing with neutralwater lowers this ratio to a constant value of 0.45, while

sodium can be practically eliminated from the gel by thistreatment according to Sema, White, and Hem (1977).

The easy removal of sodium indicates that such gels do not

contain dawsonite (NaAl (OH)2C03), hut an X-ray indifferentaluminum hydroxy carbonate. Dawsonite gels are precipitatedvia different routes (e. g., reaction of carbon dioxide withsodium aluminate solutions).

Other than carbonate, acid anions such as chloride or nitratecan be removed from aluminum hydroxide gels by washingwith a large excess of water or by dialysis. Green and Hem(1974) reported that the concentration of chloride in a gelwhich has been prepared by neutralization of an aluminum

chloride solution decreased in proportion to the volume of washwater used. As the chloride content decreased, viscosity of the

gel increased, while negative electrophoretic mobilitydiminished at the same rate.

Removal of acid anions accelerates aging (i.e., consolidationand crystallization of the solid). At temperatures below about350K, the end product is gibbsite if the pH of the solution islower than 5.8 and higher than about 9. Bayerite forms in the

pH range of 5.8 to 9. Nordstrandite may occur as a transitionfrom bayerite to gibbsite under some conditions (seeChapter 2.25).

2.12 Gelatinous Aluminas Prepared hy Other ChemicalRoutes

The strong interaction of freshly precipitated alumina gels withions from the precursor solutions makes it very difficult toprepare them in pure form. To avoid this complication, twoalternate synthesis routes are available: the reaction of

depassivated aluminum metal with water, and hydrolysis ofaluminum alkyls or alkoxides.

As early as 1908, H. W1slicenus observed that a slightly

amalgamated aluminum surface exposed to moist air reactsrapidly to form bundles of fibrous material. According toWatson et al. (1957), these fibers contain approximately

4

4 moles HZO per mole of Al*O,. They are composed of parallel

arrays of fibrils which we 7-8 nm in diameter and severalmicrometers long (Figure 2.3).

If depassivated aluminum is immersed in liquid water, an X-rayindifferent reaction product of very small (<1 Wm) particle sizeis formed. It converts to well crystallized bayerite, a-Al (OH)3,within hours as first reported by Schmih (1946).

Thin films of gelatinous aluminum hydroxide grow on pas3ive

(oxidized) aluminum metal that is exposed to water attemperatures below about 375K (see Chapter 5.2)

Aluminum alkyl compounds of the general composition A1(R)3

react with water to form aluminum hydroxide and therespective alkane, e. g., AI(CH3)3 + 3H20 +Al(OH)J + 3CH4.

In 1946, H. Schmih described the preparation by hydrolysis oftriethyl aluminum of very pure, X-ray indifferent aluminum

hydroxide gel for chromatographic applications. This reaction ishighly exothennic. Aluminum alkyl compounds are unstable in

the presence of oxygen, water vapor, or other oxidants andmay ignite spontaneously.

Aluminum alkoxides, AI(OR)3, (aluminum alcoholates) are lessreactive and their hydrolysis is easier to control. An early

report of the preparation of alumina gels from aluminumalcoholates was given by Adklns (1922). Harris and Sing(1958) and Torkm and Eggh~ (1961) are among the workers

who later investigated the physical and chemical properties of

gels prepared by this method.

K. Wefers

.- ,>,

: .,

., ..,2

M,ootfx

i

Fibrillar Gelatinous Aluminum HydroxideFigure 2.3

I


)

)

t

I

When a solution of aluminum alkoxide in alcohol, benzene, or

other organic solvent is mixed with water, the overall reactioncan be described by two equations:

1. AI(OR)3 + 3H20 e Al(OH)J + 3ROH

2. Al(OR), + 2H20 - Al OOH + 3ROH

In both cases the initial reaction products are X-ray indifferent.Poorly crystallized boehmite ~pse”doboebmite”) develops after

a few hours of aging the precipitate. If an excess of more than20 times the stoichiometric amount of water is used, bayerite isthe final crystalline phase that forms at temperatures below350K. At a molar ratio of H20 to A1203 ktween 5 and 6,

aluminum oxide hydroxide is the end product. Boehmite or theless ordered “pseudoboehmite” is the only phase ~cun-ing ifthe temperature exceeds 350K during the hydrolysis reaction or

during aging of an initially amorphous precipitate, A“pseudoboehmite” having particle sizes of 10-50 nm isproduced commercially as a by-product of the manufacture oflinear alcohols from long-chain aluminum alkoxides.

Several other variables, besides temperature and molarratio of water, determine the pro~rties of gels prepared fromaluminum alkoxides. The composition of the alkoxide, the ty~

of solvent and its miscibility with water, presence ofelectrolytes, and tbe pH of the solution arc factors whichinfluence tbe final product. Since S. Teicbner and coworkers(1969, 1970) reported the prepmation of alumina aerogels, andB. Yoldas (1975) showed that transparent ceramic bodies can

be obtained by pyrolyzing suitable alumina gels, interest in thissubject has increased substantially, particularly amongceramists.

Ingebrethsen and Matijevit (1984) reported hydrolysis by moist

air of aerosols of aluminum secondary butoxide to be aconvenient method for producing particles of gelatinousaluminum hydroxide which have a very uniform size of about100 nm. Crystallite 10-100 nm in size are obtained if

hydrolysis in the liquid phase of an aluminum alkoxide isfollowed by ~ptization of tbe freshly formed gel and ahydrothermal treatment.

In the procedure described first by Yoldas ( 1975), the solobtained after acid peptization of the hydrolysis product is

gelled by removing part of the liquid phase, or by adding asuitable electrolyte which neutralizes tbe positive sutiace chargeof the colloidal particles, After gelling has occurred, furtherremoval of liquid decreases the volume of the gel by one-thirdto two-thirds. The resulting transparent, ~rous body can befired to about 800K without undergoing major additiondshrinkage,

Teicbner et al. (1970) pinted out that the internal porosity of

alumina gels partially collapses if a liquid-vapor interfacedevelops inside the pores during evaporation. The system willdecrease the resulting capilkuy pressure by coalescing smallpres; i.e., increasing the average pore size. These workersproduced aerogels having s~cific sutiace areas of over

500 m2/g by extracting the liquid phase under su~rcrhicalconditions to avoid the formation of an interface.

A serious obstacle for the manufacture of lager, monolithic

ceramic bodies has been the unsolved problem of controllingshrinkage and cracking. These occur in the drying step andwhen the gel is fued to the temperatures necessary for theformation of stable aluminum oxide. Major industrial use of

sol-gel technology, therefore, has been limited to thin films,dip coats, fibers, and alumina ptiicles such as abrasives.

Much of the know-how deve[opd during the past decade in the

field of sol-gel technology bas been reflected in the patentliterature. A review of tbe state of the art was given byZ.elinski and Uhlmann in 1984.

2.13 The Structural Evolution of Gelatinous Aluminas

In the preceding chapters, it was shown that gelatinous

aluminas can be produced by a numkr of different chemicalprocesses. However, in all cases the initial precipitate lackssufficient long-range crystalline order to give an X-raydiffraction pattern. In an attempt to detenninc the short-rangeorder in the “amorphous” solid, Ceiling and Clocker (1943),

Bale and Schmidt (1959), and Petz (1968) ~rformedsmall-angle X-ray scattering experiments. Results indicate thatthe aluminum ions in these early precipitates are octahedrallycwrdinated as they are in crystalline aluminum hydroxides. Nodistinction can be made by X-ray techniques between O, OH,

and H20. Pmicle sizes are on the order of 2.5 nm, while

aggregates may measure several tens of nanometers, Neutronscattering analysis by Christensen et al. (1982) gave essentiallythe same results; high resolution electron microscopy confirms

these findings.

If the “amorphous” solid remains in contact with the liquid, itwill transfom into one of tbe ordered hydroxides or oxide

hydroxides. The rate and path of this transformation arecomplex functions of the chemistry of precipitation and tbe

environment prevailing during aging,

Much understanding of these processes was provided by soil

and water chemists who investigated the hydrolysis reactions indilute solutions of aluminum sails. Major contributions weremade by Marbe and Bentur (1961), Hsu and Bates (1964),Hem and Robemon (1967), Schoen and Robemon (1970), and

Smith and Hem (1972).

5

Alcoa Laboratories

The predominate ionic s~cies in dilute, acidic solutions ofaluminum salts is

[AI(H20),]3+

Hem and Roberson (l.c. ) assume that deprotonation can occur

due to the strong polarization of the water molecules by the Alion. As a result, a divalently charged complex forms

Al(H20)~+ = Al(OH)(H20)~+ + H+

which will dimerize by condensation

2Al(OH)(H20)~+ - A12(OH)2(H20)~+ + 2H20

The stmcture of the dlmer is a double-octahedron which isconnected via common hydroxyls.

De~nding on the chemical environment and the temperature,the condensationlply merization (polycondensation) reactioncan proceed in one of two ways:

1. Forming chains by litilng octahedra through commonedges.

2. Forming hexagonal rings which further coalesce to l~gepolynucleas complexes.

In both the chain and the palynuclear ring configuration, the

positive charges and number of water molecules per aluminumion decrease as the polycondensation progresses. mepro~rtion of hydroxyl ions increase to an upper limit of 2

OH: 1 Al for the chain polymer and approaches 3 in thehexagonal arrangement. Figure 2,4 schematically represents the

coalescence of rings, as suggested by Hsu and Bates (1964)and Hsu ( 1977). Figure 2.5 shows the transition from a linearaquo-hydroxopolymer to a fully dehydrated AIOOH-chain,together with the antiparallel double chain which constitutes the

structural element of crystalline AIOOH (see Figure 2,12 inChapter 2.32).

The positive charge carried by the solute precursors and by the

larger, solid plynuclear complexes explains the strong bondingof anions to the gel and the formation of hydroxy salts athigher anion concentrations; e.g., aluminum hydroxy chlorides.Anion complexes on the surface of the solid must k

hydrolyzed kfore further growth can occur via shared hydroxylions; hence, the observed retarding effect of anions on thecrystallization of gels (Nail, White, and S, L. Hem, 1976). If

edge hydroxyls of the polynuclear complexes me capped byhydrogen bnding of polybydroxy compounds such as sorbhol,further Pcdycondensation also is inhibited (Nail et al., 1. c.),

Coalescence of hexagonal ring complexes eventually leads tonucleation and growth of tabular, pseudohexago”al gibbsite

6

0 ●

GA ~24, ,

6[Al(H20):+] - 12H + - 12H20

+Ale(OH):j.12H20

AI,0(OH):;.16H20

A132(oH)~y.28H20

(

@

Al=(OH);~.36H20

Refi Hsu and Bates 1964Hem and Roberson 1967Nail, White and Hem 1976

‘Each external apex rapreaants an Al(H20)& complex

Formation of Polynuclear Rings by Condensation

Figure 2.4

crystals (Figure 2.6), Somatoids* of bayerite (Figure 2.7) formby the stacking of AI(OH)3 platelets perpendicular to their basalplanes,

Under the conditions which favor the proposedpolycondensation mechanism that results in chain-like polymers

(OH/Al s 2), fibrillar and sheet-like morphologies of thegelatinous material develop, These conditions are high salt

concentration, i.e., low water activity (Hsu, 1967), low

‘See Chapter 2.22.


!

I

hydroxyl concentration or temperatures above about 350K.

Bugosh (U.S. Patent, 1959) described the preparation oftibrillar, gelatinous khmite from acidic solutions. Llppens

(1961) reported a sheet-like morphology of gels (“wrinkledtissue papr”). ~is gelatinous material contains porly

crystallized khmite. It can be peptized by diliute mineralacids. Plefre and Uhlmann (1986) prepared an X-ray indifferent(“su~r-amorphous”) gel they believe to be composed ofswollen and folded sheets of boehmite, The swelling is ascribed

to intercalation of water within the double layers of wtahedrawhich me the stmcturaf elements of boehmite. This gel has

been produced by hydrolyzing aluminum butoxide. Theprecipitate was ~ptized with 0,28 N HN03 before gelling anddrying.

Gelatinous hydroxides prepared by hydrolysis of alkoxides(Chapter 2. 12) develop via reaction paths analogous to those

. . >“,,,

+ +W82[Al(H20)~+ – H+-AI(H20)50H2+] – 2H20 -A12(OH)2(H20):+

Y-—–—––?

Y-–—–—–ii

–H20Al(OH)n(H20):+ —- AlO OH

~o o!

-*–4H20-

g8;

x

Al(OH)n(H20):+ – (OH + OH- H20 + O) - H20-IA1202(OH)21.

@

above ~laneo

OH-● Al

below

@

H20 o0--in plane

Deprotonatio”/Condensation Reactions Leading to ChainsFigure2.5

J. J. Ptasienskl 75,000X

Tabular GibbsiteFigure 2,6

which govern the precipitation and aging of hydroxides formed

from aluminum salt solutions.

Gelatinous boebmite forms in the same environments that leadto chain polymers from aluminum salt precursors, namely, lowwater-to-afuminum ratios, low pH, or tem~ratures above

350K. It is conceivable that cross-linklng or end-to-endconnection of chains via residual alkoxide groups aids thedevelopment of larger particles or of domains in a continuous,~lymeric network gel:

o 00 0

\/\/ \/\/Al Al

/A’\ / \Al

/\/\OH OR . . RO OH

However, Lippens’ (1961) model of the bonding of [AIOOH]2

double chains in pseudoboehmite would account for the sameeffect:

OH OH

I IAl-O . HOH . O-Al

; J, J, ~

I 1-OH OH

7

Alcoa Laboratories

J. J. Ptasienskl

Bayerite SomatoidFigure2.7

40,000X

This bonding mechmism may also operate between colloidal

p~icles in particulate gels, in addition to, or replacing, vander W~al’s forces.

The literature, so far, has not been ve~ explicit shout the bond

chemistry of the so-called p] ymeric net work gels produced bycontrolled hydrolysis of aluminum afkoxides. Schlenker (1956,1958) discussed the suucture of “Alukones,” which arehydrocarbon compounds such as Iong-chain fatty acidscross-linked via aluminoxane bonds. These bonds wereintroduced by reacting aluminum afkoxides with suitablefunctional groups. Evidence has not yet been presented for an

analo80us cross linking of the inorganic polymer [A1ooH]. byhydrocarbon groups. The question remains whether“inte~netrating network gels of chemically linkedhydrocarbon and alumina polymers can k produced, or if such

gels constitute a mixture of organic and inorganic colloids.

When aluminum alkoxides are hydrolyzed with a Iwge excess

of water, the rapidly precipitated, amorphous solid pafiiallyconverts to pseudoboehmite before stable, crystalline aluminumtrihydroxide forms. Gelatinous products precipitated fromalkaline aluminate solutions or from acidic salt solutions, but

aged at high pH, undergo the same sequence of structuralstages. The appexance of gelatinous boehmite as a - generallyvery short lived . intermediate stage eve” at low temperatures

aPPe~s tO contradict the assumption that low pH or lowactivity of water ue preconditions for the chain polymerization,However, this contradiction can be resolved when we consider

the steep S1OFS of the volubility curve of aluminum hydroxidethat is shown in figure 2.1. As mentioned above, precipitation

can occur very rapidly upon a slight change in PH. Whetheracidic salt solutions are neutralized by alkali or aluminate

solutions are treated with C02 or mineral acids, the initialprecipitate will include anions due to its positive surfacecharge. Topochemical reaction (polycondensation) inside the

solid satisfies the second condition, i.e., low activity of water.Gelatinous khmite can therefore form, but will be convertedto Al(OH)J by a solution reaction of the solid with the mother

liquor

AIOOH + H20 + OH-+ Al(OH); ~ Al(OH)T + OH”

and

AI(OH)3 + AI(OH)I e 2A1(OH)3 + OH-

The latter reaction prevails in the spontaneous precipitation of

aluminum trihydroxide from supersaturated aluminate solutionsor when crystallization is induced by seeding. Gelatinousintermediates do not occur under both conditions.

2.14 Gelatinous Boehmite or Pseudoboehmite

In the preceding chapters we referred to the product occurdngas an intermediate stage in the aging sequence of aluminumhydroxide gel as pseudoboemite, gelatinous boehmite or poorly

crystallized boehmite. All these terms have ken usedinterchangeably in the literature. They describe a solid, theX-ray diffraction pattern of which shows broad lines that

coincide with the major reflections of well crystallized

Y-AIOOH. Depending on the method of preparation of thematerial, the diffraction lines are shifted to varying degreestoward higher d-values; the largest increase in the lattice

constant is measured in the direction of the b-axis (OkO).Pseudoboebmite generally contains more water than the 15% byweight corresponding to the composition A1203 H20. Up to

30% has been reported.

When aged under aqueous solutions at temperatures lower than

350K, pseudoboehmite converts to crystalline AI(OH)3, thephase thermodynamically stable in this temperature range (see

phase diagram, Figure 3. I ) This observation has supported theview that “pseudoboemite” is a transitional stage in the aging

of aluminum hydroxide gels rather than Y-AIOOH of a particlesize small enough to cause X-ray line broadening. Calvet et al.

introduced the term pseudoboemite in 1953. Pa@e et al. (1958)assumed that intedayer water caused the difference between

“mineral” boehmite and the gelatinous material. Lippens (1961)concluded that in “gelatinous” boehmite tbe (AIOOH)l chainsof the boehmite structure are linked via double hydrogenbridges provided by water molecules (see Chapter 2. 13). He

I

I

I

I

8


showed an increase of the ““it ~eII dime”sio”s i“ theb-direction proportional to the amount of water exceedingI H20 per A1203. Bye a“d Robinson (1974) believed that themajor portion of the excess water is present as a SUfiaCe

monolayer on the generally small (<20 nm) crystallite; only aminor amount being responsible for lattice strain. Baker andPearson (1974) rejected Pa@e’s and Lippen’s intercalation

model, ascribing all excess water to HZO coordinated to Al oncrystal surfaces.

Tettenhorst and Hoffmann (1980), having investigated 32boehmite samples prepared under varying conditions, concludedthat “pseudoboehmite consists of similar octahedral layers but

lacks the three-dimensional order of boehmite because of arestricted number of unit cells along (b). ” “It contains morewater which is commordy intercalated ktween octahedrallayers.” Hsu ( 1967) considered psuedoboehmite to be an

incompletely dehydrated boehmite, while Pierre and Uhlmann( 1986) assume a random intercalation of water.

This incomplete review of pertinent literature would indicate asignificant divergence of interpretation if one assumed the

subject to be a stoichiometrically, stmcturally, andthennodynamirally defined phase. It was shown in precedingchapters that gelatinous boehmite can be formed in a variety of

chemical environments, but always represents an intermediatestage in the polycondensation of less ordered aluminum

aquo-hydroxo complexes.

The aging via polycondensation of a gelatinous precipitate ispredominantly a topochemicai process; i.e., it takes place eitherin the solid phase or by solid/solute interactions over very short

transport distances due to the microporous texture of the gel.

Unlike the precipitation of a crystalline compound, such as asimple ionic salt from a homogeneous solution, thetopchemical transformation of the shon-range ordered (X-rayindifferent) aquo-hydroxo complex to the long-range ordered

AIOOH is a progressive process. As a result, the gels ingeneral are structurally and compositionally inhomogenous;longer-range ordered areas may & embedded in an“amorphous” matrix.

Because of this inhomogeneity, there obviously is nomeaningful correlation between the stmctural properties of the

“X-ray active,” ordered material and the stoichiometry; i.e., theA1203:H20 ratio, of the entire solid. Difficulties arise not onlyfrom the unknown proportion of hydrates to oxide-hydroxide,but also kcause of the different forms in which water can kpresent.

If one excludes “mobile” water in meso and macro pores,molecula H20 can be associated with the gelatinous solid inseveral ways. AS a result of incomplete polycondensation, H20can occupy oxygen or hydroxyl positions in the stmct”re oflonger-range ordered chains. This induces lattice strain because

of the distortion of the AI(0,0H)6 coordination polyhedr%there is also an increase of the cell dimensions in addition to a

positive charge, resulting from the replacement of a negativeion by the changeless water molecule.

Water adsorbed on the surface of a particle is bonded byhydrogen or dipole bonds, the ty~ and strength of the bond

being determined by the stmctural configuration of theadsorbing site, as shown schematically below:

H H O-Al-OH H

HOA14 / \o/ \o/

\/\ /\H H O-Al-OH H

The contribution of the first layer, chemisorkd water, to thestructure and composition of the longer-range ordered solid is

significant, due to the small particle size of the gelatinousoxide hydroxide. Assuming an approximate density of 3 g/cm3for gelatinous A1OOH, ideal cubic panicles of 10 nm (100~)edge length would have a specific surface area of 200 m2/g, or24,000 m2/mole of A1203 H20. At a surface coverage of 10

H20 molecules per nm2 (10 ~2 per molecule), a monolayerrepresents approximate y 0.35 mole of H20 per mole ofA1203 H20, Considering the larger surface area of

nonisometric particles and the actual ranges of particle sizesand surface areas reported in the literature, values of 0,5 moleand greater are realistic,

Another consequence of the small particle size (which can beequivalent to only a few (AIOOH) layers in the short directionsof acictdar crystallite) and the concomitant large

surface-to-mass ratio is the likelihmd that panicle surfaces maybe linked by hydrogen bonding of chemisorbed water,

.[HO -Al - O]H -0- HIO - Al - OH]n

as discussed in the preceding chapter. Whether this linkagerepresents structural intercalation of water, or agglomeration of

particles, depends on one’s point of view, This type of bondingof water cmxists with the other forms of association throughoutthe gel, The temperature and chemistry of synthesis and the

degree of aging determine which form of water associationdominates. DTA and infrared analyses by various workers

(e.g., Tettenhorst and Ho ffmann, l.c.; Hem et al., l.c.) clearlyshow a broad range of bond energies for H20 and OH, theproportion of H20 increasing with lower precipitation

temperatures and shoner times of aging of the gelatinousmaterial.

This discussion shows that each of the models proposed i“ the

literature applies to cenain aspects of the stmcture andcomposition of the gelatinous material. Most authors agree thatthe oxide-hydroxide double chain is the basic structural

9

Alcoa Laboratories

element, the same configuration that makes up the lattice of

Y-AIOOH, boehmite. In the authors’ opinion it is irrelevant thatgelatinous hohmite transfoms to .AI(OH)3 when aged attemperature below 350-370K. The formation of the [AIO, OH,H20] chain plymer is a kinetically controlled process.Whether this material transforms to crystalline A1OOH orAI(OH)3 de~nds ultimately on the temperature and pressure

conditions, i.e., the phase field in the A1203-H20 system, andnot on composition or degree of crystalline order. In view of itspredominant stmctural features, the authors prefer the term

“gelatinous kehmite” for this material.

2.2 Aluminum Trihydroxides

2.21 Gibbsite

J. J. Ptaslenski mx

Al(OH)j, gibbsite, is the principal constituent of bauxites of thetropic region. ~is mineral also occurs in North American and

European deposits. Gibbsite bauxites ae mostly, although notexclusively, of tertiary or younger age.

The crystal habit of natural gibbsite is usually pseudohexagonal

tabular, while that of synthetic gibbsite is determined by theconditions of cvstallization.

Grains of gibbsite precipitated in the Bayer process are

aggregates of tabular and prismatic crystals (Fig. 2.8). Gibbsiteusually contains a few hundredths to several tenths of a percentof alkali metal ions. The highest alkali concentrations xe found

in technical trihydroxide produced in the Bayer process.Ginsberg and Koster (1952) and Wefers ( 1965) showed thatsodium is atomically dispersed in the c~stal lattice of gibbsite.Wefers (1962) also repofied on the effect of sodium and

ptassium ions on the morphology of gibbsite. Crystals grownfrom sodium aluminate solution had a tabular habit.Pseudohexagonal, elongated prisms prevailed when gibbsitewas precipitated from ~tassium aluminate solutions. (See also

Misra and White, 1971).

Elongated prisms will also form if gibbsite is grown fromsodium aluminate solutions under conditions of low

super-saturation and high temperature (>330 K), The reasonsfor the preferential growth of the prism faces or the basalplanes are not well understood.

Ginsberg et al. (1962), Torkar et al, ( 1960), Wefers (1962,1967), and Saalfeld et al, (1968) believe that alkali ions arenecessary to stabilize the gibbsite structure. Barnhisel and Rich

(1965), however, stated that gibbsite forms from gels if the pHis less thm 5.8; Schoen and Roberson (1970) confirmed thisobsewation. The relationships between the structure of

aluminum trihydroxides and the conditions of preparation willbe discussed in Chapter 2.25,

Aggregate of Technical GibbsiteFigure 2.8

Pauling ( 1930) first proposed the concept of the gibbsitestructure which was subsequently confirmed by Megaw (1934).

Double layers of OH ions, with Al ions occupying two-thirdsof the octahedral interstices within the layers, form the basicstructural element. The hydroxyls of adjacent layers are situated

directly oppsite each other, i.e., in a cubic packing. Thus, thesequence of OH ions in the direction perpendicular to theplanes is AB-BA-AB-BA (Fig. 2.9).

This su~rposition of layers and the hexagonal arrangement of

Al ions lead to channels through the lattice parallel to thec-axis. The hydroxyl ions in the gibbsite stmcture are

considerably deformed. Hydrogen bridges originating from thedipoles operate between OH groups of adjacent double layers.From proton magnetic resonance measurements Kroon and v.d.

Stol~ (1959) deduced a model of the spatial distribution ofthese H-bonds.

The monoclinic symme~ of the gibbsite lattice, i.e., itsdeviation from a hexagonal close packing of the hydroxyl ions,can he described as a displacement of the double layers relative

to each other in the direction of the a-axis. An additionaldisplacement in the same plane, but in the direction of theb-axis, was determined by Saalfeld (19@). He analyzed the

stmcture of single crystals of gibbsite from the Ural mountains.Several crystals showed the additional displacement of thedouble layers which reduced their symmetry to triclinic. The

indices of refraction were identical for monoclinic and triclinicgibbsite. The angle of the optical axes, 2V, was found to be O

10

Alcoa Laboratories

Table 2.1

Mineralogical Properties of Oxides and Hydroxides I

I

Reference

Index of Refraction nDMobs

iardness

2-1 /2 to3-1/2

... ,., ,,

3-112 to4

6-1 /2 to7

9@

:Ieavage

(001)Perfect

(010)

(010)0Perfect

None@

Brittleness

Tough

. .

Brittle

Tough wher

compact

LusterPhase

Gibbsite

Bayerite

Boehmite

Oiaspore

Corunduma

P

1.566

1.6596

1.722

l--

Averagea

1.566

1.649

1.702

PearlyVitreous

BrilliantPearly

Pearly

Adamantine

Dana1.587

1.665

1.750

T

. . . .

1.563

. .

Montoro

:rvin and Osborn

Bonshtedt-Kupletskaya

Dana

I o I Average

1760m Dana

I 1

achromatic dispersion e, – c,= o~ – WC= 0.011 (Castor).

NO along optic exis = 1.74453 + ~

1@Knoop hardness 1525-2000 (Castor

@Fracture uneven to conchoidal; parting (0001).@Value in Ewin and Osborn is misprint.

by crystal faces. These sha~s, reflecting the interplay ofcrystallization forces with the environment, resemble hourglasses, cones, or spindles (Fig. 2.7).

the name bayerite II, since structure and growth features were{

closely related to those of bayerite. In honor of the author, thistrihydroxide was later called nordstrandite.

,

Pap&e et al. ( 1958) confirmed the existence of the new form ofAl(OH)J and published a more complete X-ray diagram. /Shimizu, Mi yashige and Funaki (1958) reproduced Pa@e’sresults.

Bayerite is produced commercially—principally for themanufacture of catalysts or other applications which require analuminum hydroxide of high purity. Physical properties of

bayerite are listed in Table 2.1.

2.23 Nordstrandite Hauschild ( 1963) showed that very pure nordstrandite can beprepared by reacting aluminum, aluminum hydroxide gel or

hydrolyzable aluminum compounds with aqueous solutions ofalkylenedlamines, especially ethylene diamine.

Van Nordstrand, Hettinger and Keith ( 1956) publisbed the

X-ray diagram of an aluminum tribydroxide which differedfrom the diffraction patterns of gibbsite and bayerite. Theyobtained this trihydroxide by precipating a gel from aluminumchloride or nitrate solutions with ammonium hydroxide.

After Van Nordstrand’s and Pa@e’s X-ray diffraction diagramshad been published, nordstrandite was identified as a

component of tropical red soils (terra rossa) in West Sarawak(Wall et al., 1962) and on the island of Guam (Hathaway and

Schlanger, 1962).

Upon aging under the mother liquor at a pH of 7.5 to 9 the gelconverted to tbe crystalline phase. Van Nordstrand proposed

12


I

I

I

I

}

1,

Van Nordstrand and co-workers (1956) ~onsjdered nordsv~dite

a screw dislocation polymorph of the gibbsite lattice. Llppens(1961) proposed a structure i“ which two double layers withgibbsite sequence and two double layers with bayerite sequence

6re 6Jtematel y stacked,

Saalfeld and Mehrotra ( 1966) analyzed single c~stals fromS6rawak. They found a triclinic unit cell containing 8AI(OH)3.

Sadfeld and Jarchow (1968) refined the stmcture, ~inting outthat the sequence of layers is AB-AB as in bayerite; on tbe

other hand, OH ions of adjacent double layers ze located

OPPsite each other. This places the lattice of “ordstranditebetween those of bayerite and gibbsite. Bosmans (1970)

c~lculated a structural model based on the Oiclinic space groupPI.

From infrared spectra, Hauschild (1963) concluded that thelayers in the nordstrandite stmcture are linked by hydrogen

bonds.

SOuctural data of nordstrandite are listed in Table 2.2.

Phase

Gibbsite

Gibbsite

Bayerite

Nordstrandite

Boehmite

Diaspore

Tohdite

Formula

AI(OH)3

AI(OH)3

Al(OH)a

AI(OH)3

AIOOH

AIOOH

5A1203H20

A1203

CystalSystem —

Trictinic

Monoclinic

Trictinic

Orthohombic

OrihorhomMc

Hexagonal

Hexagonal

(Rhomkhedral)

Space

Group

C;h

Although the technical production of nordstrandite is coveredby patents (Hauschild, 1964), the material has not been used

commercially to date.

2.24 Doyleite

In 1985, Chao et al, reponed the structure determination of anAI(OH)3 form which they named doyleite. This mineral isfound at Mont St. Hilaire and on Montre6J Islmd, both in

Que&c. At the former site, it crystallizes in tbe tabular habit

tyPical fOr gibbsite; at the latter it fores irregular prismatic a“dtabular c~stals clustered in aggregates.

Triclinic symmetry was determined, space group PI or Pi. Theinfrared spectim resembles that of nordstrandite. Suucturaldata indicate that in this form of A1(OH)3, the stackingsequence of the hydroxyl double layers can be considered an

intermediate between that of triclinic gibbsite and nordstrandlte,Chemical analysis shows Si02, FeO, Na20, md CaO to bemajor contaminants. Whether or not this material can k

Table 2.2.Structural P20perti~

MoleculesPer Unit Unit MS Length, nm Densily

Cell a bc Angle _g/cm3

4

16

2

2

2

2

1

2

0.8664 0.5078 0.9136

1.733 1.006 0.973

0.5062 0,6671 0.4713

0.5114 0.5082 0.5127

0.2866 0.1223 0.3692

0.4396 0.9426 0.2644

0.5576 - 0.8766

0.4759 - 1.2992

94” 10!92. 08;9o” 0<90” 27’

70~ 1674° 0,56” 28,

2.421

2.531

3.012

3.443

3.724

3.985

Referene

Saalfeld and Wedde

Saalfeld (1960)

Rothbauer

❑osmans

Corbato et al,

Swanson and Fuyat

Yamaguchi

Phillips et al.

1- Roth2- Fricke and Severin3- Dana4- Alma Lab.5- Toropv

13

Alcoa Laboratories

Bayerite

GA202464

B

1

A--

A

B--

cA

[

B--

B

A--

B

A--

Nordatrandite

Layer Stacking in AI(OH)3Figure 2.10

considered a fourth modification of Al(OH)J should be left

open to further debate.

2.25 Interrelationships

The stability relationships of the aluminum trihydroxides havebeen the subject of much debate, Natural abundance of gibbsitehas often been used as an argument to support the assumptionthat this form of AI(OH)3 is the thermodynamically moststable. Experimental data (Wefers, 1967), however, indicate

that bayerite is more stable than gibbsite. This should beexpected as the bmcite-type bayerite structure has the highestsymmetry of all structural variants of Al(OH)q, and also is the

densest (see Table 2.2).

As demonstrated in Figure 2.10 the structures of gibbsite,bayerite, and nordstrandite differ only in the stacking order of a

common stmctural element, (he [A12(OH)6]n double layer. Inthe structure of monoclinic gibbsite, these layers are displacedin one directon (a-axis) relative to the stacking sequence in

bayerite. In triclinic gibbsite, nordstrandite and doyleite, thestrata are shifted to varying degrees in botb the dbection of the

14

a and the b axes. The reasons for these energetically lessfavorable managements of the layers are not well understood.

Schoen and Roberson (1970) assume a different degree of

polarization of the hydroxyl ions to & responsible forvariations betw=n tbe layers of gibbsite, bayerhe, and

nordstrandite. In all three forms, the individual layers are,

however, chemically and structurally identical, all mainvafencies of the A12(OH)6 “molecule” being satisfied within alayer. The symmetry of the hydrogen bonds between the

sheets, therefore, is more likely the result rather than the causeof the arrangement of layers relative to each other.

Stronger evidence points to impurities as a factor determining

the stacking sequence of the hydroxyl in double layers.Synthesis of Al(OH), from very pure aluminum and water, orvia hydroxylation of pure, active alumina, always leads to

bayerite. Gels prepared by hydrolysis of aluminum butoxiderecrystallize to bayerhe after prolonged aging under very pure,neutral water (Bye and Robinson, 19M).

G~ns&rg et d. (1961, 1962) have ~inted out that in thereaction sequence

aluminumsalt solution+ gelatinoushydroxide - .Vstalline trihydr.xide

bayerite occurs as the initial crystalline phase. Aging underconcentrated ammonium hydroxide solutions (125 and 250 gNH3 per liter) leads to nordstrandite as a final product. If the

mother liquor contains NaOH or KOH, bayerite is converted togibbsite. Nordstrandite will also transform to gibbsite in sodium

or potassium hydroxide solutions, according to these authors.Hauschild (1963) claimed that nordstrandite is not converted togibbsite by aging under 0.01 normal NaOH or 2 ~rcent KOH

solutions.

Wefers (1967), investigating the system Na20-A1203-H20,found the transformation bayerite * gibbsite to be

irreversible. Gibbsite was formed by a solution-redepositionprocess, growing epitaxially on bayerite somatoids. Epitaxialgrowth on the nealy isostructural bayerite reduces the energy

of nucleation considerably, mtilng the solution reaction therate determining step in the process of transformation. This

may explain the discrepancy between the results of Ginsberg etal. and Hauschild. Wefers concluded that gibbsite does not

have a field of stability in the binary system A1203-H20, as allsamples produced under various conditions contained sodium.The concentration of Na could be related to temperature and

supersaturation.

Hauschild (1964) prepared gibbsite essential y free of alkali by

hydrolyzing triethyl aluminum in an aqueous solution ofethanolamine. The precipitate showed broadened X-raydiffraction lines of gibbsite. It contained less than 10 ppm

Na20, but 0.2 prcent N and small amounts of organic

I

I

t


I

material. Hauschild assumed that ethanolamine occupied anionvacancies in the gibbsite lattice,

According to Bamhisel and Rich (1965), Hem and Roberson(1967), and other authors, acid conditions favor theprecipitation of gibbsite from dilute solutions. This process

involves, as shown in Chapter 2.13, the polymerization via adeprotonation and condensation reaction of a positivelycharged, aluminum-aquo complex,

Hydronium ions could conceivably be incorporated in the

structure in the same way as are alkali ions in gibbsiteprecipitated from alkali aluminate solutions, Excess positivechages may also be provided through a substitution of Al]+ by

Si4+, a possibilityy to be considered in the case of the naturallyoccurring doyleite and triclinic gibbsite.

Determiwdtion of the position and structural significance ofimpurities in Al(OH)l plymorphs remains a challenging

problem which appears solvable by today’s powerful techniquesof structure analysis.

If foreign ions should, indeed, be the cause of the stmcturalvatiet y of Al(OH), forms, the question of their relative

thermodynamic stability would k irrelevant.

2.3 Aluminum Oxide Hydroxides

2.31 Boehmite

This modification of AIOOH is a major constituent in many

bauxites of the Meditcmanean ty~. The mineral ~curstogether with gibbsite in deposits of Tertiary and UpFrCretaceus age. In many bauxites of older Mesozoic strata,

boehmite is the only aluminum hydroxide. It forms the finecrystalline matrix of these aluminum ores, usually mixed withhighly dispersed iron oxide and hydroxide.

Bmhmite can be precipitated by neutralizing aluminum salts oraluminate solutions at temperatures near and above the boilingpoint of water. Treating amalgamated (activated) aluminumwith boiling water is another method of preparation. Aluminumcovered with an oxide film reacts very slowly under these

conditions. The reaction product formed on the sutiace consistsof X-ray indifferent material and gelatin~us boehmite; i.e., thepoorly crystallized, hydrated form described in Chapter 2.1. In

the autoclave, crystalline boehmite grows at a measurable rateabove 375K. Increase in pH reduces the minimum temperaturenecess~ for the conversion of bulk aluminum to AIOOH.Formation of behmite by a solid state reaction is observedwhen gibbsite is heated in air to temperatures between =380

and 575K. Conversion of the trihydroxide to measurableamounts of AIOOH requires rapid heating and coarse particles.[t is, therefore, assumed that locally high water vapor pressuresgenerated within Imge gibbsite grains during rapid dehydration

lead to the formation of the aluminum oxide hydroxide (see

Chapter 4.2 1). AI(OH)3 heated above 375K under water ordilute alkaline solutions is quantitatively converted to boehmitc.An electron micrograph of hydrothermally grown boehmite isshown in Figure 2.11.

The stmcture of boehmite (Ftgure 2. 12) consists of doublelayers in which the oxygen ions xe in cubic packing. Theselayers are composed of chains formed by double molecules ofAIOOH which extend in the direction of the a-axis (Ewing,1935; Reichert and Yost, 1946; Milligan and McAtee, 1956).Hydroxyl ions of one double layer we located over the

depression between OH ions in the adjacent layers. The doublelayers ze linked by hydrogen bonds between hydroxyl ions inneighboring planes. Average O-O distance of the hydrogen

bridges is 0.27 nm.

_.. ..__

J. J. Ptasienski W,mxlx

Boehmite CrystalsFigure 2.11

15

L,,


ere has been considerable discussion regarding the exactposition of hydrogen in tbe boehmite structure. If an

t

orthorhombic uni~ cell is assumed to include two layers of4AIOOH and to &long to the space group Cmcm (D];),

hydrogen bonds should be symmetric. Recent refinements ofthe stmcture (Hill, 1981, and Corbat6 et al., 1985) placehydro8en asymmetrically between bonded pairs of oxygen ions.Corbato et al. concuded ibat hydrogens are located in two

positions, king closer to one oxygen than to the comspondingone in the next layer, Thus, the structure is kst described inspace group Cmcm, Russell et d, (1978) found evidence, byinfrared spectroscopy, for Cmc (C~~) rather than Cmcm (Dj~)

symmetry.

Boehmite in bauxites is an important raw material for theproduction of aluminum oxide. Fine crystalline, synthetictihmite is produced as a precursor for activated (transition)

aluminas used in tbe manufacture of catalysts and absorbents.

2.32 Diaspore

Dlaspore occurs in the high al”mi”a clays of east central

Missouri and central Pennsylvania. This mineral is also a majorconstituent of bauxites in Greece, Romania, and EuropanRussia, Metamorphic rocks, such as the alumina-rich shales ofChina, often contain considerable quantities of this aluminumoxide hydroxide.

Because diaspore is usually associated with older bauxites andmetamorphic rinks, high pressure and elevated temperaturewere believed necessary for the formation of this mineral. The

hydrothermal synthesis of diaspore at temperatures above ~°Creported by Laubengayer and Weiss (1943), seemed to confirmthis theory, Occasionally, however, diaspore grows epitaxially

on natural corundum c~stals, obvious] y formed by weatheringof tbe aluminum oxide. Koster (1955) identified diaspore inlaterite deposits in Indiq i.e., as a product of soil formation.This indicates that diaspore can form at ambient temperatures

and pressures. Wefers (1967) synthesized diasporehydrothermally below 370K, using coprecipitated iron andaluminum hydroxide gels as a starting material. Goethite,a-FeOOH, which crystallizes spontaneously below 370K at

ambient pressure, provided the substrate for epitactic growth ofthe isostmctural aluminum oxide hydroxide, thus reducing thenucleation energy for diaspore,

The structure of diaspore (Figure 2. 12) was first investigated byDeflandre (1932) and Takane (1933). As in the lattice ofboehmite, the basic elements are chains of double molecules.

Ho o

These chains xe, however, arranged in a nearly hexagonal

close packing; whereas, the arrangement in b~hmite is cubicA schematic representation of the stacking of the (2A 100H).chains was given by Llp~ns (1961); it is shown inFigure 2.12.

Refinements of the diaspore structure were published by Ewing(1935) and Hopp (1940). Busing and Levy (1958) established

more precise hydrogen positions by neutron diffraction. Thereis agreement among investigators that the A106 coordinationoctahedra are distorted, resulting in two Al-O hnd Iengtbs of0.185 and 0.198 nm, respectively. While Busing and Levy’s

results indicate the presence of hydrox yl ions, stmcturccalculations by Giese et al. (1971) favor a more ionic proton.

No conunerciaJ use or large-scale synthesis of diaspore has

been reponed so far. Use of diaspore bauxites as a raw materialfor alumina production via the Bayer or tbe sinter process goesback almost one hundred years, but has been abandoned inmany countries in favor of gibbsitic bauxites which require

considerable y lower processing temperatures. Acicular crystals

typical fOr diaspre are shown in Figure 2,13.

2.33 KI-A1203 or Tohdite

In 1951, Houben reported to have found an alumina form ofthe composition 5A1203.H20. Torkar and Krischner (1960)synthesized a similm compound by hydrothermal treatment of

aluminum or gelatinous alumina in the temperature range of

\ / \A,Al

\/\ Diaspore Crystalso OH Figure 2.13

17

Alcoa Laboratories

650 to 8013K at pressures between - 10il to 3W bar. X-ray

diffraction data, but no structural information, were given ofthis reaction product which these workers called KI.

Yamaguchi et al. ( 1964) determined the structure of a new

mineral named tohdite which they found to be identical withTorkar’s KI with regard to composition and the temperaturepressure conditions of formation. The same authors later

described in detail various methcds of synthesis (1966) and arefinement of the structure (1969) of this alumina. Althoughnamed a “hydrate” by Yamaguchi et al., infrared analysisshowing an OH-stretch frequency of 3240 cm–l and the

dehydration tem~rature ktween 1100 and 1200K (Yamaguchiet al., 1964) indicate an oxide hydroxide. The OH-stretchadso~tion band lies between two strong bands of Y-A1OOH,

boehmite.

The stmcture of tohdite-KI is closely related to those of kappaand beta alumina (see Chapters 4.41 and 3.2). Yamaguchi etal. (1964, 1969) calculated a hexagonal unit cell with thelattice parameters

a. = 0.5576 nm, co = 0.8768 nm

space group P6Cmc. The unit cell contains one formula unit;i.e., 10 Al, 160, 2 H. About one-fifth of the aluminum ionswe in tetrtiedral coordination, tbe other ones are surrounded

by six oxygen. Yamaguchi et al. did not determine hydrogenpositions. In view of the similarity of the tohdite lattice withthe structure of kappa alumina, and considering the hightem~rature of dehydration, a rather “metallic” hydrogen ion

can be assumed. One may, therefore, conclude that the layerstructure contains spinel* blinks in which charge balance fortetrahedral sites occupied by the trivalent aluminum is providedby protons:

(A13+H+O:-)lV 2 (Al;+O~-)vl

As this composition is very close to the reported formula5A1203H20, tohdile may be an ordered hydrogen spinel.

DeBoer and Houben (1952) suggested gamma alumina to hehydrogen spinel. Later work showed that this transition aluminahas a defect spinel stmcture with high degree of disorder and arandom distribution of hydroxy l-ions (see Chapters 4.43 and4.53),

Further work is needed to ascemain the exact positions ofhydrogen in the structure of tohdite. The thermodynamic

stability in the A1203-H20 system and the relationship to otheralumina phases must also b established. A comparison withthe H20-beta alumina synthesized by Saalfeld et al. (1968) by

exchanging sodium ions with water molecules (Chapter 3.2)would also yield valuable information.

‘nomal spinel: (MeO)lv(M.203)V1,exampleMgA1204

4..{i

Data on stmcture and physical properties of tohdhe-KI weincluded in Tables 2.1 and 2.2.

2.4 Aluminum Oxide

2.41 Corundum, a-A1203

Corundum, a-Al*03, is the only thermodynamically stableoxide of aluminum. Occurrence of corundum is commonlyassociated with igneous and metamorphic rocks. It is the chief

component of the abrasive mineral emery. Red and blue gemquality corundum crystals are known as ruby and sapphire,

respectively. The red color of mby is derived from the presenceof chromium and that of blue sapphire is related to the presence

of iron and titanium.

2.411 Crystal Structure

Corundum crystallize> in the hexagonal-rhombohedral system,space group D~d or R3C. This stmcture type is often referred to

as “corundum structure” in crystallography. Hexagonal tabularand prismatic crystal habits are most common (Figure 2. 14).

The c~stal structure of corundum was fully investigated by

Pauling and Hendricks (1925) following emlier work by Bragg

and Bragg (19 16). Refinements have been published later byNewnham and DeHaan (1962) and Cox et al. ( 1979).

J. J. Ptaslenski 800X

Corundum CrystalsFigure 2.14

18

Alcoa Laboratories

this reason the actual Al-O-Al hnd angles in corundum show

considerable departure from ideal vafues for regular wtahedra.

Because of their identical structures, solid solutions of A1203with Cr20g md Fe203 are possible over a wide range of

concentrations. The A1203-C1203 series of mixed crystalsprepared by reaction ~tween the oxides at 15W-16CH3K haveken investigated by Thilo et al. (1955) who have reported cellparameters for 22 different compositions. Similar studies for

A1203-Fe203 solid solution have been published by Mum andGee (1955). Corundum can lake up approximately 20% Fe203in solid solution at 1700K.

2.412 Preparation

Corundum can be synthesized by thermal and bydrothemalmetbcds, Aluminum oxide is formed by tberrnal dehydration of

aluminum hydroxides. The extent of conversion to thecomndum structure de~nds on the temperature and time ofthermal treatment. Total conversion occurs on heating abve

15WK for more than one hour, Technical grades of “calcined”aluminas, used for aluminum smelting, ceramics, abrasivesetc., represent materials with different degrees of conversion toa-A1203, varying from 5 to 100%, The tem~rature and rate of

conversion is affected by impurities and “mineralizers. ” Thesize of corundum crystals formed is a complex function ofprecumor hydroxide, temperature, time and calcination

environment. The presence of fluoride ions is known tofacilitate the growth of a-A1203 crystals at lower tem~ratures.

Fused corundum is produced by melting calcined alumina in an

electric arc furnace. “Tabular alumina” is composed of large,well developed, tablet-like a-A1203 crystals. It is produced by

heating aluminum oxide made by dehydration of gibbsite to atemperature close to the fusion temperature.

The Vcmeuil method has been used for the preparation of largesingle crystals of comndum includlng jewel quality sapphireand ruby crystafs. In this process, an oxygen-hydrogen flame is

used to fuse tine A1203 pwder. The molten oxide solidifies ona seed c~stal, The cimling rate is carefully controlled toproduce single cwstals measuring several centimeters in size.Single crystals of appreciable size can also be grown from

melts containing lead fluoride or cryolite as ineri flux(Timofejeva and Woskanjan, 1963; Klekr and Fehli”g, 1965).

Chemical vapor deposition of a-A1203 has ken studiedextensively. This process may involve pyrolytic decompositionof aluminum alkoxides (Aboaf, 1967) high temperature(> 15WK) hydrolysis of aluminum chloride (Wong andRobinson, 1970) and of aluminum ff”oride (LOcsei, 1962).

Comndum is also formed by direct oxidation of aluminum

metal with oxygen at high temperatures. The reaction is highlyexothennic; fused corundum is produced if a sufficiently hightem~ramre is maintained (Griffiths, 1981).

The hydrothemd formation of ct-A1203 is discussed inChapter 3.1. Laudise and Ballman (1958) have descrikd the

hydrothetmd synthesis of large crystals of sapphire, Highgrowth rates were obtained by the use of Na2C03 solution as a

solvent at temperatures above 670K and pressures in excess of20W bars. Doping with chromium, realized by the addition ofsodium dicbromate to the solvent, resulted in the formation ofsynthetic mby.

Alumina ceramics made by the sintering of poly-crystalline

a-alumina we of considerable technical and commercialimportance. The properties of the find product are controlled

by the different steps of the fabrication process, Dense, sinteredarticles of afumina are prepared by forming tine a-Alz03powder to the required shape (either by d~ compaction or by

slip casting) and subsequently fting to bigb temperatures(18@-21MK) to produce sintering and eliminate porosity. Thefact that A1203 can be sintered to theoretical density was first

demonswated by Cabmn and Christensen in 1956 andsubsequently by Coble (1959). These and others have shown

the imptiance of suppressing grain growth for pore eliminationand the role of smafl additions of MgO in achieving this.

Techniques for preparation, and properties of fibrous alumina

crystals, so-caf led “Whiskers,” have ~n studied intensively inrecent years. The tensile strength of single crystal whiskers can

be several orders of magnitude higher than that of theplycrystalline bulk material. Values of 5,500 MPa

(8 X 105 lbs/in2) have ken measured for corundum whiskers(Levitt, 1966). The product is currently of considerable interestfor the fabrication of fiber reinforced polymer, metal andceramic composite materials.

Several metbcds of growing alumina whiskers, generally fromthe gas phase, have been described. Campbell ( 1963) reacted

AlC13 with C02 in the presence of H2 at 1500K, Grimshaw, etal (1975), have described a process in which gaseous AIC13 ispassed over a mixture of alumina and carbon at 1700-2100K to

form aluminum monochloride and CO. This stream is thenpassed over a zone of lower temperature, 1300- 1500K where

tbe reaction is reversed producing A1203 whiskers.Decomposition of “spun” fibers made from aluminumcompounds and organic binders has ken used to produce

polycrystalline alumina fibers and filaments. Gitzen (1970) hasreviewed methods of preparation and properties of aluminawhiskers and fibers.

Alumina is an important high temperature ceramic material andits properties have been explored thoroughly.

2.413 Thermal Properties

Data and reviews of thermal pro~rties of alumina have kenpreviously compiled by Gitzen (1970), Dorre and Htibner

20


(1984), JANAF (197S), NBS (19S2), U.S. Geol. Survey Melting Point(1979), Thermophysi.al Institute (1977), CRC Handbook ofMaterial Science (1975) and many other publications, A Literature values range from 2310 to 2329K (2037” to 2056°C).summary of important and recent results is discussed in the Tbe value of 23273 6K (2054 * &C) was recommended byfollowing. Schneider (1970) from tbe results of cm~rative measurements

Table 2.3.Thermodynamic Data

298,15K (25°C) and 0.1 MPa (1 bar)

Substance

AI(OH)3gibbsite

AI(OH)3

bayerite

AIOOHboehmite

AIOOH

diaspore

a-A1203

MW

78.004

78.004

59,989

101.981

P-A1203

y-A120~

K-A1203

8-A1203

AIO

A120

A1202

42.981

89.982

85.962

State

Cryst.

Cryst.

Cryst.

Cryst.

Cryst.

Gas

Gas

Gas

Molarvol.

cm3/mol

31.956

19.55

17.76

25.575

AH; AG;

kJ/mol

–1293.2 –11 55.0

–1288.2 –11 53.0

–990,4 –915.9

–999.8 –921 .0

–1675.3 –1562.3

– 1657

– 1657

– 1662

– 1666.5

67 41

–145 –173

–395 –400

Sources: (1) NBS Tables of Chemical Thermodynamic Properties Vol. 11, 1982,

Muddlement No. 2.(2) JANAF Thermochemical Tables, 1976 Supplement, and Third Edition, 1965

(3) Geological Survey Bulletin 1452, Reprinted with Corrections,1979.

(4) Kuyunko et al. (1983)

s“ Cp—

Jlmol K

68.44

46.43

35.33

50.92

218.4

252

281

91.7

65.6

53.3

79.0

30.9

52.0

67.2

21

Alcoa Laboratories

Table 2.4Equation for Cp

CP(TK) J/mol.K = a + bT – CT-2

TemperatureRange, K a b c

Gibbsita 240-500 50.46 179.49 x 10-3 10.96 x 105

Boehmite 240-600 56.23 82,93 13,60

Diaspore 240-600 46.94 64.16 11.30

Corundum 296-1500 114.77 12.80 35.44

Sollrce: Kuyunko et al. (1963)

by nine groups in seven countries. Reported values for entbalpy

change on melting 6re in tbe range of 107.5 ? 5.4 to

118.4 t 2.5 kJ/mol.

Boiling Point

Brewer and Semcy (1951) reported the boiling point ofa- A1203 to be 3803K (3530”C). Yudin and K6rklit (1966)re~rt a value of 3990K (37 17°C) for the boiling point and theheat of vaporization to be 1976.1 k.flmol. We consider thesedata to have little practical meaning as aluminum oxidedecomposes to suboxides (Chapter 2.42) at this temperature,

Thermodynamic Data

Recent values for heats of formation, Gibbs free energy and

entropy at 298. 16K are given in Table 2,3,

A large number of heat capacity (Cp) determinations over awide temperature range have been published, A survey ofliterature by Dhmars and Douglas (1971) includes some twenty

previous studies. Correlations have hen published by Robie etal. (1978) and Kuyunko et al. (1983) in the form:

Cp=a+bT–+T

Values of the coefficients we given in Table 2.4. As shown in

figure 2.17, the specifc heat at constant volume, Cv,

approaches the tberrnodynamic value of 3R (=24.94J/g atom K) at high temperature, a-A1203 has been used as acalorimetric standad.

Thermal Expansion

Thermal expansion of U-A1203 bas ken studied by Wachtman

et al. (1962) who determined the relative linear thermalexpansion of single crystal a-A1203 parallel and perpendicular

22

to the C-axis and of a polyc~stalline material in the

temperature range of 100- 1100K. The anisotropy of thehexagonal comndum stmcture causes the expansion coefficientparallel to the C-axis to be higher than that perpendiculu to it

by about 10%. Klein (1958) showed a nearly constant ratio ofthe expansion coefficients in the two directions between

300-2000K. The values for pcdycrystalline alumina areintermediate ktween those for the two single crystalorientations. Expansion coefficient data up to 2 100K we

reprted by Nielsen and Leipld (1963) for polycrystal linealumina. Yates etal. ( 1972) reported results between 373 toI073K for single crystal and calculated the mean GrOneisen

parameter as a function of temperature. The ThennophysicalResearch Institute has publisbed a graphical presentation of

published data including a recommended curve for m-A1203single crystals.

As shown in Figure 2.18, the thermal expansion cm fticient foralumina approaches a constant value at high temperaturessimilar to heat capacity behavior. This is in accordance with

theoretical predictions and was verified by Hoch andVemardakis for Q-A1203 ( 1975),

Thermal Conductivity

Data on thermal conductivity of alumina ue well documentedand generally consistent, Charvat and Klngery (1957) publishedvalues for both single crystals and plycrystalline sintered

materials. In sintered alumina the thermal conductivity must k

GA 20

I I I I I I I I

VI I I I I I I I

0 200 400 600 800 10001200140016001800

Temperature K

IHeat Capacity of a-A120J as a Function of Temperature

Figure 2.17 I


1,

\

I\

\

~

Temperature

K “c

100

200

298

400

500

600

800

1000

1200

1400

1600

–173

–73

25

127

227

327

527

727

927

1127

1327

Table 2.5Selected Thermal Properties of a-A1203 at Various Temperatures

Linear

TharmalExpansion

Heat Capacity

* ~

0.03

0.12

0.18

0.22

0.25

0.26

0.28

0.29

0.296

0.298

0.30

0.126

0.502

0.753

0.920

1.046

1,088

1.172

1,216

1.238

1.247

1.255

Thermal Conductivitycal/cm. s.°C J/cm.s.K

1.075 4.5

0.196 0.82

0.110 0.46

0.0773 0.324

0.0579 0.242

0.0451 0.189

0.0310 0.130

0.0251 0.105

lPK~l O”s

0.6

3.3

5,5

7.1

7.5

7.9

8.5

9.1

9.6

10.1

10.5

corrected for porosity. The relation K = Kp/( 1-p), where Kp isthe experimental thermal conductivity and p the prosity, has

been generally used. Representative thermal conductivity dataof several investigators are presented in Figure 2.19. Thevariation of thermal conductivity with temperature shows aminimum. This arises from the mechanisms of heat transport.At low temperatures, the rate of heat transport is related to the

mean free path of phonons between two collisions which varieswith Iff K. At higher temperatures, electromagnetic radiation

becomes the dominating mechanism (K & T3). Data of Charvatand Kingery show that the minimum for single crystal alumina

occurs at 11OL-i 200K compared to about 1500K for thepolycrystalline material; the difference has been explained onthe basis of greater transparency of the single crystal. Nishijimaet al. ( 1965) fitted the equation: k = A/(B + T) + CT3 to their

experimental data. Results of Fltzer and Weisenburger (1974)are in good agreement with those of Charvat and Kingery. Thesignificant effect of small amounts of impurities on thermalconductivity arises out of the loss of regularity of the ionicarrangement of the crystal lattice.

2.414 Diffusion and Related Properties

Thermal

_

0.075

0.058

0.039

0.030

0.022

0.017

0.013

0.012

0.011

Diffusion, pmiculady at high temperatures, is the principalmechanism underlying several properties of alumina including

electrical conductivity, sintering, grain and crack growth atelevated temperatures, and creep. There is a considerablevolume of experimental observations in the published literature

describing these phenomena in A1203, However, because oftheir strong dependence on preparation methods, impurity

content and environmental factors, data and their interpretationlack consistency and are often confusing,

The nonstoichiometry of well-crystallized comndum is very

small, so is the defect concentration. Mohapatra and Kroger( 1978) proposed that Schottky disorder is the main atomic

disorder mechanism in alumina. However, many experimentalresults of diffusion controlled phenomena can be and have beeninterpreted in terms of Frenkel disorders. Thus, the question ofwhich disorder mechanism determines the bhavior of pure and

doped A1203 remains ambiguous.

23

Alcoa Laboratories

12

o

I I I I I I I IPolycrystalline

IIc - axis

iVI 1 I I I I I I I

o 200 400 600 60010001200140016001800

Temperature, K

Thermal Expansion ofa-A1203as a Functionof Temperature

figure 2.18

The concentration of Schottky defects in pure a-A1203 new themelting pint has ken estimated to k of the order of 10– 12.This concentration is orders of magnitude smaller than that

caused by even very small amounts of impurities. Evidently,obsewed diffusion related phenomena are linked primarily todefects caused by the presence of impurities rather thanintrinsic defects in pure A1103. Dome and Hiibner ( 1984) havetabulated (reproduced in Table 12.6) energies for the formation

of various defects in alumina.

GA a,i, ,I I I I I I I

\ Polycrystalline\

___ _I I I 1

0 200 400 600 6001000120014001600 laOO

Temperature, K

Thermal Conductivity of a-A1203 as a Functionof Temperature

1.2

1.0

0.8

0.6

0.4

0.2

0

Self diffusion of aluminum and oxygen in alumina at high

temperatures has been studied by direct measurements usingtracers (Oishi and Kingery, 19@, Paladlno and Kingeg, 1962)

and from plastic deformation investigations (Cannon et al.,1980). Data from recent publications are presented inFigure 2.20.

2.415 Electrical Conductivity

An important diffusion related property of alumina is electrical

conduction. Alumina is a ve~ good electrical insulatoq itretains high electrical resistivity to very high temperatures. The

accurate measurement of true electrical conductivity of a vevhigh resistance material at high temperatures is very dlfticult asit is of the same order of magnitude as that for surface and gas

conduction. There is strong evidence that data published before

GA2024610

K

2300 2000 1700 1500

–lo

–11

~

“g –12

o

; –13A

–14

–15

I I I

\

Aluminum ion inpolycrystslline A1203

\

Oxygen ion inpolycrystalline A1203

Oxygen ion

crystal A1203

I I I I I I

0.46 0.50 0.54 0.68 0.62 0.66

103~(K)

Self Diffusion of Al and O in AIz03Figure 2.19 Figure2.20

24


r

I

Table 2.6Possible Disorder Mecbani9ms in A1203

Type of disorder Equation of formation and Formation energy Preexponential

equilibrium constant Hi/n, in eV term, exp(S~nk)

Frenkel disorder Al~l – All+V~L 10.0 ~,o-4

of Al

[Vfi] = [Ali]=K~fi) 8.3

4.45

KFl{~l)=l .32 1018 exp (–4.45/kT) cm-3

Frenkel disorder o~e o~+v~ 7.0of o

[vO]=[O(] = K$(~} 7.1

Schoftky disorder null * 2vfi + 3V0 5.7 ~,o-4

5.2

‘~=:’vol=(:)‘5K5”54.1

3.83

K~/5 = 1,3tj 10I9 exp (–3.83/kT) cm-3

Intrinsic electronic null~e’ + h’ 5.18disorder

[h’] = [e’]= Kj’2

150

KJJ2 = 3,14. 101g ~ exp (–5.18/kT) cm-3

Source: Dorre and HUbner

i’,) 1971 are not reliable. Figure 2,2 I shows results from literature

published during 1960-1973.

The electrical conductivity is a function of temperature, oxygen

partial pressure (po2), impurity (or dopant) content and grainsize (for polycrystalline alumina). Behavior with respect to p02shows a minimum in the range of 10–6 and 10-4 bar. This hasken interpreted in terms of contributions by different

conducting species, Two types of hhavior have been obsewed.In one, found in alumina containing “acceptor” type dopantssuch as Mg, Fe, Co or V, conductivity is mainly ionic at lowand electronic at high oxygen partial pressures, The reverse isfound when “donor” dopants, TI, Y, H, Si are present. Thus

the electrical behavior of A1203 is primarily dependent on the

tY~ and amOunt of doping and the interaction of the dopa”twith oxygen.

In polycrystalline alumina, the grain boundaries have beenidentified as zones of increased conductivity -- resulting in

much higher overall conductivities than those of single crystals.Conductivity increases with decreasing grain size (Bozkan andMoulson, 1971),

2.416 Dielectric properties

The low dielectric loss and high permittivity of alumina haveled to wide applications in electronic ceramics, andconsequent] y, its dielectric properties have been thoroughlyinvestigated and documented.

Dielectric properties (dielectric constant, dielectric loss tangent)

of corundum were extensively investigated and publisbed by

25

Alcoa Laboratories

K

2000 1000 750 500

I I I I

L I I I

5 10 15 20 25

104/T(K)

Electrical Conductivity of a-A1203 as a Functionof Temperature

Figure2.21

Von H]ppel (1953, 1954, 1959) up to the temperature of

I073K (Table 2.7) Data for highter frequencies (1010 Hz) andtemperatures (23WK) have been re~rted by Lhovchenko et al(1983). Hill (1974) d,scussed experimental techniques for the

precise determination of the dielectric propefiies ofpolycrystzlline alumina ceramics ad gave data for microwavefrequencies of 1010 Hz.

Dielectric propeflies of single crystals of a-A1203 and ofa-A1203 doped with chromium and vanadium were studied by

Govinda and Rao (1975). The results were interpreted in termsof the polarization mechanisms operating at vtious frequenciesand temperatures. These authors found that among the fourpolarizations (electronic, ionic, dipolar, and space charge)contributing to the dielectric constant of A1203 only electronic

md ionic polarization are present at a temperature of 300K.The concentration of defects (contributing to. space charge

26

polarization) in pure and doped corundum is low. This wasborne out by the fact that the dielectric loss in these sampleseven at 102 Hz and 373K is very low (tan 8< 0.0005). Thesimilar ionic radii of Al (0.057 rim), Cr (0,064 rim), and V

(0.065 nm) suggest that Cr and V enter the A1203 lattice

without much distortion of the lattice. The frequency-dependentincrease in the dielectric constant (K) at temperatures above473K, the large dielectric loss at the higher temperatures and

the larger changes in K and tan 8 (with temperature) in the Crand V doped crystals are attributed to space charge polarization

and the contribuiton made by the dopants to defectconcentration. The frequency-dependent increase in dielectric

constant with temperature is apparently due to increase in ionicpolarization at higher temperatures.

The dielectric breakdown of alumina has been shown to becaused by two mechanisms: intzinsic (electronic or avalanche)and thermal. The former is impoflant in the case of thin tilms

of amorphous aluminas and is generally used to describe.lowtemperature phenomena. The thermal breakdown mechanism ismore complex and depends, in addition to intrinsic

characteristics, on such factors as geometry and frequent y andwavefom of applied electric voltage.

Miyazawa and Okada (1951) reported that the electricalstrength of a 70 pm thick layer of alumina (exact structure notspecified) was constant at 2 X I@ V/cm from roomtemperature to 1073K and then decreased to 6 x 103 V/cm at

1600K. Brht and Davis (1971) studied plycrystalline a-A120S“Lucalox” samples from rwm temperature to 1373K concludedthat an avalanche bre~down recurred at <725K, where a

maximum strength of 3.3 X 105 V/cm was observed. Thestrength decreased at higher temperatures. This was attributedto a thermal breakdown mechanism.

Yoshimura and Bowen (1981) studied the dc electrical strengthof sapphire and pcdycrystalline alumina up to the temperature

of i673K. The electrical strength was essentially the same forboth the materials. The rmm temperature value of >106 V/cmfor a 100 pm thick sample decreased gradually with

temperature to 2.6 X 105 V/cm at 1173K, then dropped rapidlyto 2 X 104 V/cm at 1673K. The electrical strength decreasedwith the saznple thickness; it was inversely proportional to the

thickness for samples thicker than 600 pm at 1473K. Thebreakdown behavior above 1173K was described by a thermal

breakdown model.

I2.417 Optical Properties

a-A1203 (corundum, sapphire) is usually uniaxial negative,

though twinning could give rise to anomalous biaxial character.Colors found in corundum crystals are related to the amount ofother ions replacing aluminum: ruby has 2 to 3% Cr203 while

sapphire contains Fe3+ and TI, yellow comndum may have NI


Table 2.7Dielmtric Properties of Corundum

Temperature

I

\

“C/K

25/298

50/323

100/373

200/473

1,

)300/573

!400/673

i

‘? 500/773

I600/873

700/973

8OOI1O73

Frequency

Cycle51s0c

1O“10“10“10.

6.5 X 10”

10-I 10.

10.10.

8.5 X 10°

10-1O*10.10“

6.5 X 10”

10-10.10.10.

8.5 X IO.

10.1O“10s10s

8,5 X 10°

10“10“10.10.

8.5 X IOC

10-10.10°10-

8,5 X 10m

10“10“10“10“

8.5 X 10°

10°10“10“

8.5 X 10°

8.5 X 10-

UelectricConstant

Sapphire@

L OptiC tiS IIoptic hs

8.349.349.349.349.34

9.369.369.369.369.36

9.419.419.419.419.41

9.539.539.539.539.53

9.659,659.659.659.65

9.829.789.769.769.76

10.029.959,929.929.92

10.5510.1710.0710.0710,07

10.2610,2610.26

10.40

11.5511.5511.5511.5511.55

11.5811.5611.5611.5811.5a

11.ea11,6611.6611.8J311.ea

11.8711.6711.8711.8711.87

12.0912,0912.0912,0912.09

12.3512.312.312.312.3

12,712.5512.5512.5512,55

13.1512.6312.6312.W12.63

13.1513.1513.1513,15

13,50

Dielectric LOSSTangent(Dissipation Factor)

Sapphirea

L Optic tis IIOptic Axis

0.00003

0.00001

0.000032

0.0002

0.000036

0.000120.0000360.000012

0.000620.000230.0000780.0000370,000062

0.00330.0010.000360.000160.00008

0.0260.0050.00120.000560.000093

0,210.0320.00540.00110.00013

0.00021

0.00043

0.000012

o.oom86

0.000016

0.00009

0.0000380,00001

0.000086

0,000170.0000240.00001

0,00011

0,000660.000150.00005

0.00013

0.0030.00090.0003

0,00014

0.0150.00350.00170.00110.00015

0.160,0210,00360.00170.00017

0,00019

0.00021

@ Oata from MIT Technical Report 126: von Hippel (1953, 1959)

w values In parenthesis were obtained from a difterent sample of sintered alumina

DielectricLOSS

Sntered Aluminaa(99.9%A120.)

10.510.510.510.59.6

10.610.610.610.69.7

10.710.710.710.79.7

12.010.610.610.89.6

21.612.611,211.1

9.9

10021.513.111.510.0

2576919.01310.1

10.2

10.3

0.000660.000310.00011

0.00049(0.00026)@0.00150.000560.000190,0001

‘0.200.0330.00670.00160.00062

(;:003)

0.330.060.0120.00076

(0.00034)1.031.170.300.0620.001

(;0.)

1.100.800.240,0016

(0,00048)

0,003(0.00062)

0.0059(0.00093)

@ All measurements of sintered alumina at 6.5 x 10” were made with electric field perpendicular to the direction of processing. At other

frequencies the field was parallel.

Alcoa Laboratories

or Fe3+; brown varieties have Fe3+ and Mn and green cO1Orhasbeen attributed to V, Mo, Ni and Fez+.

The refractive indices are fairly constant and the commonly

accepted values are El .760 and 61,768. Measurements of Eand m for synthetic sapphire were reported by Jeppesen (1958)for various wavelengths from 690.7 to 253.6 nm. leppesen also

determined the temperature cm fficients for refractive indicesand bircfringence. me refractive index of sapphire has alsobeen reported by Malitson et al. (1958). Other reports of dataon optical constants of sapphire have been published by:

Lowenstein (1961) -270 to 1000 #mRussell and Bell (1967) -50 to 400 #m

Lang and Wolfe (1983) -0.3 to 25 Wm

The temperature dependence of the far-infrared ordin~-rayoptical constants of sapphire was determined by Cook andPerkowitz (1985) for frequencies ktween 30 and 230 cm-1.

Optical constants of a-A1203 at 27W-30WK have beenreported by Klabukov et al. (1983).

Bhagavantam and Venkatarayedu (1939) predicted eighteenfrequencies for a-A1203 of which two symmetric and fivedoubly degenerate ones should be Raman active. Krishnan(1947) found the Raman frequencies and their relativeintensities to be 375 (8), 417 (10), 432 (4), 450 (2), 578 (3)

@2 (6), and 751 (7) cm-l. The Raman spsctrum of a-A120qwas confirmed by Rarnm in 1951.

Pure a-A1203 is neither phosphorescent nor fluorescent

although 0.000 I % of chromic oxide produced a distinctfluorescence (Kroger, 1948). Experimentally, ruby has beenfound to exhibit fluorescence in the famous RI and R2 regionsat 694.27 and 692.85 nm as a resonance process (MendenhaOand Wood, 1914; Wieder, 1959) and when pumpsd with

electric fields, in the absorption bands at 250, 400, and 560 nm(Maiman, 1960). Ruby has been used as a laser material. Its

behavior as a macroscopic fluorescing and laser material wasdiscussed by Mahan et al. (1969). The Faraday effect in rubyand corundum was investigated by Boiko and Soika (1984)who showed that the angle of rotation for the plane of high

polarization at 632.8 nm is a linear function of magnetic fieldstrength.

The emissivity of alumina decreases from 0.98 at roomtemperature to 0.9 at 373K, 0.8 at 673K, 0.5 at I023K, 0.3 at

1273K, 0.18 at 1873K, and 0.12 at 3073K (Heilman, 1936;Michaud, 1949; Kilham, 1949; Taylor and Edwards, 1939;

Leedy, 1954). The monochromatic emissivity at 655 nm is0.15 in the temperature range of 1273- 1873K (Michaud, 1949).

Infrarsd spctra data are given in Table 2.11,

Za

2.418 Magnetic Prowrties

Magnetic susceptibility of aluminum oxides and hydroxides wasmeasured by Selwood (1950) and m given in Table 2.8.Additional data for a-A1203 were re~rted by Rao and Leela

(1953).

2.419 Mechanical Properties

Due to its wide use as a structural ceramic material, the

mechanical pro~rties of aluminum oxide have been extensivelystudied and documented. These studies have covered both

single crystals of a-A1203 (Imge crystals of sapphire have beengenerally used) and polycrystalline sintered products. Themechanical khavior of the latter is, as would k expected,

strongly influenced by the micros~cture. A thoroughdiscussion of the subject is given in the monograph of Dorre

and Hiibner (1984).

The elastic khavior of the hexagonal comndum crystal is

described by six elastic constants of the general 6 X 6 elasticitymatrix. These constants and values for the compliance havebeen experimentally determined by Wachtman et al. (1960)who have also reviewed previous reports. Strong anisotropic

bebavior was expected and observed. The variation of elastic

constants in the temperature raoge of 77-850K was determinedby Wachtman et al. (1961) and by Tefft (1966).

In plycrystalline alumina, which can k considered isotropic

on a macroscopic scale, the Young’s modulus and the shearmodulus can k estimated from the orientation de~ndent

Table 2.8Magnetic Susceptibility

Unib of 106; Data of Selwnod

Form Susceptibility

G!bbsite –0.43

–0.46

Boehmite –0.37

Gamma Alumina –0.34

Corundum –0.23

See also Rao and Leela for more values on corundum.Pascal found partially dehydrated gibbsite nearly

identical with corundum.


Property

Plastic Constants

Plastic Compliance

Mtiulus of Elasticity

Bending Strength

I

Modulus of F6gidity

Poisson’s Ratio

Volume ChangeaImpact StrengthStrength

Table 2.9Mechanical Properties of Corundum

25° C Unless Otherwis Noted

Condtions Values, psi

SllS12

S1?

Sll

s?,

S11

Z&c500”C

1ooo~c1zoo-c

25-C500”C

1Oovc1zoo-c

Plates

1ooo~c25°C

1ooo~c

25°C25-C

25-C

25°C

25.1 OOOQC1400”C

800”C1ooo~c115rc1300”C1500”C

72.0.1023.7.1016.1.10–3.4.10W72.2.1 W21,4.10.

2.35.10cm2. dynl-0.72 .10cm2. dyn>–0.37.10”cm2. dynl0.49.1 Ocmz. dynl2.17.1 Ocm2. dynl6.94.10”cm2. dynl

52.6.10”48.1.1043.5.1041.9. IW

59,3.1057.3,1054.9.1053.6,10

44,300-1 15,600(2)24,000-36,00016,000.22,00037,000(1)26,000(2143,000(1)32,000(2)

43,000-131,00094,000 -t55,00017,000-50,000(540~c)

23.3( Reuss).24.l(Voigt)23.3

23.9

0.257@

0,[email protected]@

o.oo355@43,00027,00024,00016,00011,000

3,400

Alumina

CorundumCorundumCorundumCorundumCorundumCorundum

CorundumCorundumCorundumCorundumCorundum

SapphireSapphheSapphireSapphire

POlycWstalfinePOlycVstallinePolycrystalline

CorundumRAE S!”tered171SapphireSapptireSapphire

SaDDhireSapphireSapphire

Sapptirehot pressedzero porositycold pressedzero porosity

cold pressed

zero porositySinteredSntered

SapphireSnteredSinteredSinteredsinteredSinteredSintered

Reference

Wachtman et al (1960)Wachtman et al (1960)Wachtman et al (1960)Wachtman et al (1960)Wachtman et al (1960)Wachtman et al (1960)

TetftTeHtTefftTefftTefftTetft

Wachtman and LamWachtman and LamWachtman and LamWachtman and Lam

Crandall et alCrandall et alCrandall et alCrandall et al

Klassen-NoklyudovaRoberfa and WanRoberta and WanRoberta and WattRoberla and Watt

Wachtman and Maxwell (1954)Wachtman and Maxwell (1954)Wachtma” and Maxwell (1954)

Wachtman et al (1960)

Lang

Lang

Spriggs and BrissetteRyshkewilch (1942)Ryshkewitch (1942)

Bridgma”MooreStavrolaNs and NortonStavrolatis and NotionStavrolaks and Noflo”Stavrolalds and NotionStavrolakis and Notion

29

Alcoa Laboratories

Properly

Strain

Compressive Strength

Tensile Strength

Table 2.9Mechanical Properties of Corundum

25° C Unless Otherwis Noted

Condtions

TTension,

Temp. psi

1300°C1Ooo”c 7,200

1Ooo”c 20,000

1Ooo”c 15,100

1Ooo”c 12,000

llOO°C 17,100

IIOO”C 13,100

1200”C 12,100

1300”C 7,250

13oo”c 8,620

luration,

hr

402

769

77

431

95

431

53

3

44

25°C25°C

4oo~c1Ooo”c16oo~c

20”C800”C80°C

1300”C1300”C14oo~c1400”C1460”C15oo~c

45” to c-axis

30”C300”C800”C

11Oo”c30° to c.mis45” to c-ais60mto c-axis75- to c-axis

Values, psi

Observable10 10-2%

1 10-2%

0.49%

1.52%

,08%

2.06%

0.7%

2.34%

32%

443,000 to 495,00050,(XX)213,000128,000

7,100

21,00037,40019,50033,90011,6006,4007,3004,2001,500

0

71,00052,560052,50088,000

100,000

78,00065,00094,000

Alumina

3ntered7AE

sintered @?AE

sintered @;apphre

;apphre

;apphite

Sapphire

3appMre

3apphhe

Sapphire

SapphirePolycrystallinePOlycrystallinePOlycrystallinePolycrystalhne

PolycrystallinePolycrystallinePolycwstalfinePOlycVstal linePOlyc~stalhnePolycfystallinePolycvstallinePolycrystalfinePolycwstalinePolycrystaline

SapphireSapphireSapphireSappNreSappMreSapphireSapphireSapphire

Reference

Stuvrolukim and NortonRoberts and Watts

Roberts and Watts

Wachtman and Maxwell (1954)

Wachlman and Maxwell (1954)




Wachlman and Mwell (1954)


PavlushkinRyshkewitch (1941)Ryshkewitch (1941)Ryshkewitch (1941)Ryshkewitch (1941)

SchwarlzRyshkewitch (1941)SchwartzRyshkewitch (1941)SchwartzRyshkewitch (1941)SchwartzRyshkewitch (1941)Ryshkewitch (1941)Ryshkewitch (1941)

Wachtman and Mawell (1954)Wachtman and Maxwell (1954)Wachtman and M~well (1954)Wachtman and Maxwell (1954)CastorCastorCstorcastor

@ Optic misl length oftestpiece. Benlinplane ofoptic aXis~ Optic misl length oftestpiece. Benl Ltoplane ofoptic aXis@ Four point loading. Angle between rodaxis and slip@ 10,000 kg/cm2(142,200 psi)@ Heating 1500to1900"C followed byslowcoofing increased strength@ No.nits

I

~ ipecifiigravity 3.48-3.620 Specific gravity 3.610 Specific gravity 3.76@ Crushing

30


values of the respective single-crystal mcduli. These values aregiven in Table 2.9,

Conrad (1965) reviewed the mechanical behavior of sapphirewith respect to plastic flow involving slip, twinning, creep, andfracture. Sapphire deforms plastically above about 100K; the

stress for yielding and the amount of strain before fracture arestrongly dependent on tbe strain rate. This yield bebavior atelevated temperatures is characterized by tbe presence of up~r

and lower yield points whose values depend on the temperatureand the strain rate.

[n the aluminum oxide crystal, sliD can occur in the three

different systems; i e., ba~al, prismatic, and pyramidal. Themost_common mode of deformation is basal slip in the<1 I 20> direction; the prismatic and pyramidal slip systems

are activated at high temperatures of > 1900K. Kronberg (1957)

applied dislocation concepts to determine the significance ofcrystallographic orientation on slip behavior.

Twinning has been observed frequently in single-crystal and

polycrystalline alumina and is considered to be an importantplastic deformation mechanism (Kronberg, 1957). Onexamination of sapphire specimens tested in compression,

Stofel and Conrad ( 1~63) found two distinct types of twins,one puallel to the (O11 I ) plane and the other parallel to thebasal plane. It was established that twinning occurred only at

lower temperatures and high strain rates.

Creep is an important deformation process for aluminum oxideat high temperatures. The mechanisms involved includediffusional creep at small and intermediate stresses (Cannon

and Coble, 1975) and non-viscous Nabarro climb type creep athigh stresses (Weetiman, 1968). The effects of temperaturesand stress on the creep rate of sapphire was studied by Rogerset al. (1961) in the range of 1250- 150i)K and by Cbang (1960)

at 1750 -2200K.

The modulus of rupture of both sapphire and polyctystallinealumina as a function of temperature was reprtcd by Jacksonand Roberts (1955). For single crystals, the modulus of rupturepassed through a minimum at 873K and returned to its room

temperature value at 1273K. The same behavior was observedby Wachtman and Maxwell (1959).

According to Conrad ( 1965), the fracture of 60” orientedsapphire rods in tension, bending, and compression generallyoccurs on a plane approximately normal to the tensile stress

and the fracture surface is conchoidal. The fracture stressdecreases with increase in previous plastic strain, independentof temperature and strain rate. The mechanism of failure seemsto be the interaction of edge dislocations with pre-existingsurface cracks. When twinning occurs, the fracture surfacetends to follow the twin interface suggesting that this is weaker

than the untwinned crystal. _~e O“ oriented crystals, stressed inbending, fractured along {O111} planes. This observationsuggests cleavage fracture. Detailed discussions of mechanicalproprties of Polycrystalline aluminum oxide ceramics including

strength-grain size relationships, fracture strength, thermal andmechanical shock proprties, fatigue, hardness, etc., can befound in books by Gitzen (1970) and by Dome and Hiibner( 1984).

2.4110 Surface Properties

Surface related properties of aluminum oxide such asadsorption, catalytic activity, wetting, sintering khavior are theresults of the atomic and electronic arrangement of the surfacewhich differs from that of the bulk crystal. A recent review by

Henrich ( 1985) Wints to the lack of investigations of the

surface pro~rties of a-A1203 in spite of the technical andcommercial importance of this material. One practical reasonis, since A1203 is an excellent insulator (optical band gap9.5 eV), the electron spectroscopic techniques, appliedsuccessfully to metal and semiconductor surfaces, are difficultto apply.

The LEED (Low Energy Electron Diffraction) technique basbeen mostly used to study the surface of A1203. Notable arc tbeworks of Charig (1967), Chang (1968), French and Somorj ai

( 1970), and Wei and Smith (1972). Most of the measurementswere made on the (Oi)Ol) basal plane of a-A1203 crystals.

At moderate temperatures, the (WO 1) face of U-A1203 exhibitsthe two-dimensional hexagonal 1 X 1 lattice of aluminum atoms

expected from the bulk crystal stmcture. French and Somorjaishowed that the ( 1 x 1) bulk-like structure remains up to about1250K under vacuum. On further increase of temperature anew ordered surface structure designated at (~ x W)(rotated 30”) appears. Subsequent beating to even highertemperatures (- 1700K) produces the ordered (fix ~)

(rotated 9“) surface structure, This surface was shown to beoxygen deficient with composition corresponding to A120 (orA1O).

Two other crystal faces, (101 2) and ( 11~3), have been studied

(Cbang, 1969). They have (2 X 1) and (4 X 5) surfacestructures, respectively, at high temperatures (> 1~K).

In reality, an atomically flat surface is unlikely and surface

steps and vacancies have considerable influence on surfacebehavior. The only information that is available on defects onAllO, surfaces come from Auger spectroscopy work by Olivierand Poirier ( 198 I).

Other investigations using UPS (ultraviolet photoelectronspectroscopy) (Bianconi et al., 1979), electron energy loss(Olivier and Poirier, 198 1), and surface pho”on loss

Alcoa Laboratories

spectroscopes (Liehr et al., 1984) have not shown any unusual

electronic structure of the surface.

Experimental investigations of specific surface energy of A1203

have used two techniques - liquid metal contact angle and the~ndant-drop method. Values range from 1.0 J/mz at 1273K to0.65 a! the melting temperature (Norton et al., 1953;

Rasmussen and Nelson, 197 1).

2.42 Aluminum Suboxides

The only confirmed suboxides of aluminum are A120 and AIO,

though spectroscopic identification of other species such asA1202, A102 has been repofied at high temperatures.

Brewer and Searcy (1951) concluded from vapor pressurestudies that the two suboxides of aluminum A120 and AIOmust exist in the gas phase. Hoch and Johnston ( 1954) studiedthe reactions: (I) 4AI( 1) + A1203(s) - 3A120(s) and (2)

Al(I) + A1203(s) ~ 3AIO(S) between 1300K and 23WK bymeans of a high temperature X-ray technique. Tbe studyindicated the formation of solid A120 and AIO above certaintemperatures. At 13WK no reaction occurred. Between 1373K

and 1773K A120(s) is formed according to ~uation ( I).Between 1773K and 1873K bth reactions (1) and (2) occursimultaneous] y; whereas, above 1873K only reaction (2)occurs. On cooling or rapid quenching both compoundsdispropofiionate into Al and A1203. X-ray diffraction patternsshowed that both the compounds are cubic, the lattice constants

being 0.498 nm for A120 at 1373K and 0.567 nm for AIO at1973K.

The literature until 1967 was extensively reviewed byMackenzie ( 1968) who also summarized the properties of AIOand A120. Giltesen et al. ( 1968) and Yanagida and Kroger( 1968) reinvestigated the system Al-AlzOj and critically

discussed the results of previous investigations. Both groupsagreed that no suboxides of alumium exist in the solid state.

Yamaguchi ( 1974) concluded from electron diffraction studiesthat solid A120 and AIO are formed at the bound~ betweenthe Y-A1203 and the Al metal surface in the oxide layer on

aluminum. The lattice constant of the cubic suboxide wasreported to vary bstween 0,406 and 0,430 nm, Thealuminum-rich layer may be due to the incorporation of oxygeninto the metal lattice (Eberhardt and Kunz, 1978), or thereduction of excess A13+ ions in the interface region (Wefers

and Wallace, 1976).

Ivanov et al. (1973) made an electron diffraction study of thealuminum suboxide A120 molecules formed by heatingAIIOJ-Al mixtures at 23~K to 2400K. They showed that the

A120 molecule has C2V symmetry; the Al-O distance is 0.173tO. Wl nm and the AI-O-AI bond angle is 141 ~ 5 degrees,

Mass spectrometric studies of vapor over alumina at high

temperatures ( 1900-26WK) have been used to identify anddetermine thermodynamic parameters of the different species(Far&r et al., 1972; Ho and Bums, 1980). Both groups ofworkers claim the formation of the species A102 and A1202 in

addition to AIO and A120 and have reponed partial pressures .for the various s~cies. !

(Literature data on thermodynamic pro~rties of the differentsuboxides of aluminum we given in Table 2.3.

/2.5 IR-Spectra and X-ray Data

Table 2.10Infrared Spectra of Oxides and Hydroxides

(Wave Number vs. Intensity)

Ph.% OH Stretch OH ESnd

~btite 3616.6 (M) 1015,2 (VS)

3518.6 (S) W.3 (M)

3428.2 (VS)

2376,4 (M)

2361,3 (S)

Bayerite 3533.6 (M) 1016.2 (s)

3518.6 (M) 975.6 (S)

24w,2 (M)

NO1 ,3 (M)

~hJ),22~6(s)1142,6 (M)

3079.8 (s) 1069,5 (S)

Daswre 2924.0 (VS) 1069,5 (S)

2341,9 (M) 959.7 (s)

2114.2 (M)

1984.1 (M)Corundum

M= Mediumlntensi~ S = Slrong

.= Shoulder

a See also Adler (1950)

unassigned AI-O Stretch fleference~

909.1 (W) 741.8 (S) Fredrick,o”

Hannah

626.4 (M~

794.9 (s)

662,1 (M) 776.4 (S) Fredrickson ‘.

Hannah

613.0 (M).

719.4 (MP (

740.7 (S) Fredrickson

Hannah

A

720,0 (VS) Cabannes-

cm c

614,6 (M) 759,9 (Vs) Ha...

802,0 (s) K.les.va

559,0 (M]

490,0 (W)

450,2 (S)

VS = Vefy Str.ng W = Weak

32


Table 2.11X-Ray Diffraction Data of Aluminum Hydroxid~ and Oxide

3

hkl

m110mm2112

112

;:02<9W

311121312o=l=

312114313222023

123,222322024124124

314224,130

414315

m,mo

m216,=+~,324

415=2

3163320%

E

4166=

d,t,1(C—

VIj—

w25,81212

86442

1;

2:6

w4424

—

25

a!.

3

hkl

,.

Cu.

dh

3.4792.s22,3792.1852.W5

1,%, .74a1.ml1.W4.5!4

1,5101.4M1.3741.=,1,276

$.m31.*1.1898I,lm1.1470

I.lmz1.12551.12461.W1.W1

1,07811.M261.01750.59780.9857

0.9819Owl0.%130,9%50,9?78

o.su7e0.5Q520.85910,-

B,a.A

,.113

s

w?1IO,om

011,m

11,,02, +

121m

k?210

CQ2012

201,131+ZzaM

211211

,, 2,022+140221

Ml

E141a32

231~2,132+

212212310

Z&l=051

222,W2+m,003

311,241+142442321m

232,023+330Om2W123

;:ml,lm+061,03

251

251312

312,242+18,,m+

213

322322

344,223+4C!

a

<20

hti

Iwm210,102110

103m112mlm

mz203122105w

laX2mms220

222,131132125m

1~,107.

al

m:i7al ,W8

,1i

1

c

*

;

6466

02262

84224

246

:

2

;22

22222

22224

t Al,

ul

m7[x1>x

2!1:2!x1:

554

lE2C

27

:

28—

e

:44

309778

3

;1819

136

!074

—

33

—dA—

:23257Ea

5c04163523=192

115WI605M623

w511al4,9394—

328281265251217

197led114096

—

I.iA

4.794.324.214,163,89

3.613.433.032.W2.71[

2,%12,4M2.45!2.3%2.=

2,2712,217ZIG2.11:2.074

2.01ii .991,.97$1.M:1.m2

1,8771,8211.W4

dA

4.714.%4.143,B3.20

2.6592.53,2,5102.W2.W

2,3562.2742.2222.IM2.16n

2.1E42.1%2.0751.B31.%3

1.9691.9%71.W, .m, ,8%

4.7651.7231,6951.WT.6%

1.W1.&l?,=81.m1.572

,.5541,5a, .Sm1.4921.W

1.4781.4571.451.al1.4m

1,4161.3971.3921.%71.370

1.=1.m1.W, .3?31,3m

1.3111.3051.2771.2651.2W

dA

4.e4864.371,4,3,873.35W3.3122

3.18293.1OW2.46*2.45=2.4224

2.W12.W712.28=2,24Ed2.1924

2.16472.~52.W92.02341,m

1.96371.U21.75%71.7W,.6974

1.M54.65764.5926?,%651.5739

t .%251.WI .45n1.W51.4115

1.mzl1.W, .m1,-111.3247

1.=991.2316i .2267I,zl Ea1.2116

1.17911.14Q1.mI .mze

dA—

71592!45=4W

3963%317131077

9)48157W712678

mm570522680—

4314=w376w

3mm289279256

243218m1783I739

!408!C03N23

hkl d.4 hkl—

012Iw,40M113

202024116211122

018124Om125m

,.0.1!119220306223

314312?281.2.1,).0.1:

;:021,2,1!1,1.1:

40431,2.13182=

326),1.1410m

—010ml101lW110

11,011011111110

2?1120,402

101020,201

121

112240,021

112242,m22W,221

121

Om,10,m

L?10104a,11,2,140

131C41

;?141

221240w231

lm,15f

ZmwmWII*2

mml

37,,170251321

W,132C42WIw142

Mlm410

11

E9mm

8W770E62527453

4X412396w369

312m224m178.

17111-137!152~17

C45902849W398189%6

9310924791059023B937

BSQ7B~W7Wleue6—

m,m

140,03?131051

mm151w231

W2,800221712W

25,

H3m2W

C62274m2222.U1

w351262

371

wW,3Z

282,33%2

053lWalm.5,

021111m211

012120222

120,212022

13L,21O211,311.112,312

lC.3731,102,

,W

240,425142

118624M632

t

I

33

Alcoa Laboratories

I I I 1 I100

};

80

60

40

20

0

I I I I

100

80

60

40

20

0 I

100

80

60

40

20

0

Bayeritea - AI(OH)3

i

I I I I

Nordstrandite

Al(OH),

JIL, I

GA 202’

I I I I I I

I I I I I I 1

—

Gibbsitey - AI(OH)3

1

Boehmitey - AIOOH

I

I A> dA 1 h LIAI I I I I I

fNaaporea - AIOOH

LI I I I I I

I L

Corunduma - A1203

20.00 30.00 40.00 50.00 60.00 70.00 80.00 20.00 30.0040.00 60.00 60.00 70.00 60.00

2“ THETA, CU Km

X-Ray Diagrams of Aluminum Oxide and HydroxidesFigure2.22

34


t,

0.2,246 t3

I I I I I 1 I I

I I I I I I I I I I I I I I I I I I I I I I I I I 1.

I I I I I I I I I I I I I I I I I I I

Bayerite - a-Al(OH)S

I I I I I I I I I I I I I I I I I I I I I I I I I I

I I I I I I I I I I I I I I I I I I I

Boehmite - y -AIOOH

1 I I I I I I I I I I I I I I I I I I I I I I I I I

I I I 1 I I I I I I I I I I I I I I I I I I I 1 I

I 1 I 1 I I I I I I I I I I I I I I I I I I4000 3000 2000 1000 400

cm-l

Infrared Spectra of Aluminum HydroxidesFigure2.23

Alcoa Laboratories

3.

3.1

Binary md Ternary Alumina Systems

The A1203-H20 System

Tbermcdynarnics and phase relations of the A1203-H20 systemhave received considerable attention in the technical literature.

Results of numerous studies have found application in suchdiverse areas as geochemistry, soil chemistry, aluminaextraction from bauxite, ceramics, corrosion of aluminum, andfor the experimental verification of therm~ynamic concepts

and relations.

3.11 Phase Relations

Phase relations in the A1203-H20 system have beenexperimentally studied by Laubengayer and Weiss (1943),Ervin and Osbom (1951), Kennedy (1959), Neuhaus and Heide(1965), Fyfe and Godwin (1962), Fyfe and Hollander (19@),

Haas and Holdaway (1970), Matsushita et al. (1967), andHaas (1972). Some of these authors have comparedexperimental observations with thermodynamic calculations anddiscussed discrepancies, Studies were carried out under

bydrothennal conditions with mixtures of water and aluminumhydroxide in closed bombs. Kennedy ( 1.c. ) also used apiston-anvil apparatus for experiments at pressures higher than

2 kbar. Torkar and Krischner (1960) studied formation of solidphases in the reaction of aluminum metal with water attemperatures between 600 and 8WK and water vapor (steam)

pressures up to 300 bar.

The stability relationships of the three trihydroxides, bayerite,gibbsite, and nordstrandite were discussed in Chapter 2.25. Itwas concluded that bayerite is the stable phase in pure water at

298K. This conclusion is further supported by the higherdensity of 2,53 g/cm3 for bayerite compared to 2,42 g/cm3 forgibbsite; dehydration of boehmite to bayerite at very highpressures (Dachselt and Pitz, 1957), and on extreme grinding

(Gout et al., 1983), and the formation of bayerite duringdehydration of rapidly dehydrated gibbsite (Section 4.2 I). ,,

transformation temperatures by decreasing the activity of wateras shown by Wefers (1967) for the system Na20-A1203-H20.

I

Phase transformations proceed via an intermediate solute phase

(Kennedy, 1959; Neuhaus and Heide, 1965; Wefers (1967)

As the sum of tbe molar volumes of A1OOH + H20 is greater

than the moku volume of Al(OH)j, the conversion temperatureincreases with higher pressure. For the same reason the

transition temperature boehmite/diaspre is lowered byincreasing pressure, because tbe moku volume dlmisbes from19.87 to 17.44 cm3 in the phase reaction. The density of

diaspore (3.44 g/cm3) is higher than that of boehmite (3.ol

g/cm3).

Below 575K the spontaneously crystallizing oxide-hydroxidephase is always boehmite. Spontaneous growth of dlaspre has

been observed only above 575K and 2Ctilbar pressure.However, in experiments using diaspre seed crystals, growth

of diaspore occurs at temperatures as low as 450K at theexpense of boehmite. The reverse reaction has not beenobserved. Wefers (1967) bas reported epitactic growth of

dlaspore on isomorphous goetbite (a FeOOH) at temperaturesklow 373K.

~ese observations have led to the conclusion that boebmitc ismetastable in the A1203-H20 system. However, bmhmiteformation is kinetically favored at low temperatures andpressures due to its lower density; afso bcause common

373 473 573 673

‘“””m :

!’OO[’”““!$nBayerite is also the only ~hydmx~de which can be synthesized $

by reaction of aluminum and water.

Figure 3,1 is a representation of phase relations in theA1203-H20 system developed from the data of Kennedy (l.c.) ~

and Neuhaus and Heide (l.c.),

AI(OH)3 transforms hydrothermally to AIOOH at around 373K.No significant differences in the temperature of conversion

have so far been reported for tbe three trihydroxides. As therate of conversion is very slow near the equilibrium ,Outemperature of around 373K, the exact temperature cannot be o 100” 20W 300” 400” So””caccurately determined from direct experiments but only by

extrapolation and thermodynamic calculations, The rate isincreased significantly under higher pH (alkaline) conditions Phase Diagram A1203-H20(Gins&rg and Koster, 1952), Dissolved salts lower the F,gure 3.1

36


i

t)

structural elements in the crystal lattices of AI(OH)3 and

boehmite allow epitactic growth of boehmite on thetrihydroxide. Both factors lower the nucleation energy. A tmephase boundary does not exist between diaspore and hoehmitein the A1203-H20 phase diagram. The coexistence of boehmite

and diaspore in many clays and bauxites may be explained bythe extremely low volubility of boehmite in the pH range ofnatural waters, Though this is the generally accepted view, free

energy calculations published by Gout and Dandurand (1983)indicated that boehmite may have a stability field in thesystem.

Fyfe and Hollander (1964) determined the equilibriumtemperature for conversion of oxide-hydroxide to oxide

a-AIOOH = CK-A1203+ H20 (liquid)

to be 640 & 7K under pressure corresponding to the watervapor pressure (=200 bar), This agrees with data of Kennedy,Neubaus and Heide, and Haas. Haas determined the

diaspore-comndum equilibrium by observing expituy ofdiaspore on corundum above 660K and 1.5 kbar pressure.Ex~rimental equilibrium points were used to derive

thermodynamic data which allowed calculation of theequilibrium curve below 660K. The curve passes through 170bar at 64(3K. The equilibrium curve in P-T coordinates has a~sitive slope which means that the transformation tem~ratureincreases with pressure. Thenntiynamic considerations suggest

that reduction of water activity shifts the diaspore-comndumequilibrium towards lower temperatures. This was verifiedex~rimentally in the Na20-A1203-H20 system (Wefers, 1967).

The transformation of boehmite to comndum occurs atsomewhat higher temperatures of around 675K at <200 barpressure. Since behmite has no stability field in the

A1203-H20 system, a boehmite-comndum equilibrium line hasno significance. Itis likely that this converion occurs viadiaspore as an intermediate phase, but no direct ex~rimental

evidence is available. The rate of transformation of bothboehmite and diaspore to comndum increases when thevolubility is raised by increasing the pH of the solution.

Thenntiynamic data for compounds in the A1203-H20 systemhave been analyzed for consistency by many reviewers

(Kuyunko et al., 1983). Values given in Table 2.3 arecompiled from recent publications.

Several forms of aluminum oxides (KI, a*, and autocalve-y),reprted by Torkar and Krischner (1960) in low pressure (<300bar) hydrothermal ex~riments at 600-800K may k considered

metastable. Of these, only the KI form (also known as Tohdlte,Chapter 2.23) has a well defined X-ray diffraction pattern.Tohdite converts to a-A1203 at 675K and 2W bar pressure.

3.12 Volubility of A1203 in Water

The volubility of the compounds in the A1203-H20 system isanother area of interest. Volubility is primarily a function oftemperature and PH. which determine the nature of the ionicspecies in solution. Volubility is very small in the pH range of4.5-8.5. In the A1203-H20 system, the stable solid phases inequilibrium with the solution are:

<375K - trihydroxide (gibbsite**or bayerite) Al(OH)J

375-640K - oxide hydroxide (diaspore) a-AIOOH

>640K oxide (comndum) a-A1203

Volubility data for the hydroxides have been reported by manyworkers, particular] y at high pH values which relate to tbe

Bayer process for alumina extraction from bauxite. Volubilitydata also have been used to develop and refine thermodynamicproperties, and for comparison with calorimetricdeterminations. Discrepancies have ken frequently encountered

and attempts made to reconcile them (Parks, 1972; May et al.,1979).

Though there is much discussion on the nature of the ionics~cies in solution, no clear picture has yet emerged. There is

general agreement that at room temperature tbe mononuclearA13+ ion exists in acidic solutions &low pH 4. As the pH isincreased, a seties of bydroxy -aluminum complexes is formed,There is much controversy as to which complexes are stable.

Some authors find that their volubility data are adequatelyrepresented by mononuclear complexes of the type A1(OH) ~-”with n in the range of O = n <4, i.e., A13+ to Al(OH)~, Somefind it necess~ to assume the existence of plymeric

complexes. The mtiel propsed by Hem and Roberson (1967)may & considered a gd example of the latter. They pro~sedthat at pH values two or more units blow tbe pint ofminimum aluminum volubility (about pH 6), the aluminum ion

is cwrdinated by six water molecules. The strong positivecharge of the Al 3+ ion polarizes each water molecule and, as

pH increases, eventually drives off a proton forming themonomeric complex ion Al(OH)(OH2)~ + At abut pH 5 this

complex ion and tbe hydrated A13+ ion are equal in abundance;more hydrated A13+ ions exist at lower pH and more of thecomplex ions at higher pH values. Two of these complex ions

may join by losing two water molecules to form a dimec

[A11(OH)2(OH,),~’

*-1” Icmary systems

37

Alcoa Laboratories

Further deprotonation, dehydration, and polymerization of

monomem and dimem yields a ring structure of six octabedrallycoordinated aluminum ions with the formula

[A16(OH),2(OH),2]6+

Coalescence of rings into layers by further growth and stackingof layers results in the formation of crystalline aluminumbihydroxide (see Chapter 2. 13).

More recently, May et d. (1979) determined volubility curvesat 298K for synthetic and natural gibbsite from pH 4 to 9. Thesolubilit y curve obtained for synthetic gibbsite is shown inFigure 3.2. The slight inflection at pH 6.7 was indicative of the

possibility that a solid phase less soluble than gibbsite controlsthe solution composition above pH 6.7. However, this phasecould not be identified.

Ion activity products re~fied by May et al. (l.c.) for varioussolute species are listed in Table 3.1.

The calculated concentrations of the individual mononuclear Al

ion species in equilibrium with gibhsite are also shown asfunctions of pH in Figure 3,2.

At pH abve 8.5 the volubility of aluminum increases verysharply with the formation of Al(OH)i– ions. This part of thevolubility curve has been extensively studied kcause of its

imponance to the industrial Bayer process for aluminaextraction from bauxite (Chapter 3.3).

–3

–4

–7

–8

0.2024614

\I I I I I

\ - — Natural gibbaite

\ — Synthetic gibbsite\

\\

/

\/

L/’

/

\ /

\0

/\ /

\ /@, /-.

.

4 5 6 7 8 9 10

PH

Volubility of Gibbsite in WsterFigure 3.2

Table 3.1Ion Activity Product3

Al(OH)a + 3H+ % A13+ + H20 “Kw= 1.29 X 106

AI(OH)3 + 2H+ s AIOH2+ + 2H20 ‘K~l = 1.33 X 103

AI(OH)3 + H+ s AI(OH)J + H20 ‘K~2 = 9.49 X 10-3

AI(OH)3 + H20 s AI(OH)I + H+ ‘K% =8.94 X 10-i5

AI(OH)3 s Al(OH)~

Ionic equilibria in the temperature range of 370-570K,corresponding to equilibrium solid phases of diasporelboehmite,

also have kn studied experimentally by several investigators.Data of previous work for boehmite have ken reviewed and

additional results reported by Kuyunko et al. (1983). Diasporesolubllity was investigated by Wefem (1967), Dmzbinina(1955), and Bemshteyn et al, (1965). The high temperature

equilibrium at pH above 7 has been shown (Kuyunko et al.,1983) to involve the possible presence of the species Al(OH)~in the aqueous phase.

Several workers have determined the volubility of corundum inwater at high temperatures, Becker et al, (1983), Ragnwsdottir

and Walther (1985) have reviewed earlier work and providedadditional data. Morey (1957) reported corundum solubilities at775K and 1,04 kbar, Anderson and Bumham (1967) and

Bmnham et al. (1973) undertook experiments at 775-975K and6 kbar. Ganeyev and Rumy3ntsev (1974) reported experimentalresults from 640-780K at 0.2 to 2 kbm and the most recent

work of Ragnarsdottir and Walther (1985) gives data ktween675-975K and I-3 kbar. All these studies show wide variations

in the volubility results in the range of 1-4 ppm Al with no

app~ent effect of temperature and pressure. The sol”bilityresults have ken interpreted on the assumption of the

formation of Al(OH)~ as the dominant solute species.

Using the equilibrium constants for the aqueous phasereactions, Kbodakovskii et al. ( 1980) have calculated theprevalence of the different hydroxo-aluminum species atvtious pH and temperature conditions. As can be seen in

Figure 3.3, the stability ranges of the cationic form of

aluminum decreases sharply with increasing temperature. Theactivity of the dissolved neutral hydroxo complex Al(OH)~increases correspondingly. As a result, the volubility of

aluminum hydroxide and comndum in acidic solutionsdecreases with increasing temperature. The ratio of Al(OH)~ toAl(OH)~ solute species is practically independent of

temperature.

(

I

!

38


600

300

I I I I

1 3 5 7

PH

Stabifity Ranges of Aluminum HydroxyComplexes in Water

Figure3.3

3.2 The Na20-A1203 System

3.21 Phases Occurring in the System

Sodium oxide forms a numkr of compunds with aluminumoxide. The molar ratio of Na20 to AlzOJ of phases repofied inthe literature ranges from 1 to 12. Compounds having molarratios of 1:5 and higher age generally formed at temperatures

ahve 1300K. The composition of structurally defined phasescan vary, some of the many forms descrikd in the literaturemay be compositional variations of a structure type.

Well established phases are, besides Na20 and A120J, the 1:1

sodium aluminate NaAIOz and the so-called beta afuminas. Thelatter phases range in composition from Na20 5A1203 to-Na20 9A1203. A molu ratio of Na20 to A1203 of 1:11 wasassumed by some authors (see klow).

Several other compounds in the Na20 system have beenreported in the literature. Elliott and Huggins (1975) found aphase designated at A-Na20 XA1203 (with x varying from 3 to

- 12) in the temperature range 1075 to 1325K. The structure of

this phase is very similar to that of mullite and is likely to kthe same as that observed by Pemotta and Young (1974).Barker et al. (1984) identified three sodium rich aluminates,

Na7A1308, Na,7A15016, and Na5A104, in a study of reactions ofsodium oxide with a-A1203. These compounds are isostructuralwith corresponding sodium rich compounds in the iron and

gallium systems.

The high temperature (> 1000K) region of the Na*O-A120~system has been studied by several investigators. The hteratureup to 1967 was critically examined and reviewed by DeVries

and Roth (1969). These authors also proposed two versions ofa phase diagram of the Na20-A1203 system based on available

data. Subsequent investigations by Llebertz (1971) and hCarset al. (1972) were aimed at resolving uncenainties in the

previous reprts. The remaining uncertainties have beendiscussed more recently by Stevens and Binner (1984), whoquestioned the existence of several phases.

The alumina-rich, high temperature area of the phase diagram

that was proposed by LeCars et al. in 1972 is shown inFigure 3.4. Four solid phases occur in this region: NaA102,p-alumina, fl-alumina, and u A1203. The system has a eutecticktween 8 NaA102 and p-alumina at 1855K (not shown in thefigure) and a perhectic between p-alumina and a-A1203 atapproximate y 2275K.

3.22 Sodium Aluminate NaA102

The 1:1 sodium aluminate NaA102 can & prepared by heatingmixtures of aluminum oxide, hydroxides, or salts with sodiumhydroxide or carbonate to about 12COK. Decom~sition of

dawsonite (A1203 Na20 2C02 2H20) yields sodiumaluminate at temperatures of 901.-1 ~K.

Stiium aluminate (1:1) has ken found to exist in threeallotropic forms (Thery, Briancon, and Collogues, 1961). Thep-NaA102 modification is stable to 743K; above thistem~rature it transforms to the y-form. The p-form is

ortborbomblc; its unit cell contains four molecules of NaA102.The structure is similw to that of ~-NaFe02, with which itforms a series of solid solutions. The transformation to

Y-NA1O* at 743K is reversible. The bigher temperature‘y-modification is tetragonal. A third modification, ij.NaA102,has ken shown to occur akve 1683K (Tbery, Lejus,Briancon, and Collogues, 1961).

In the presence of moisture, NaA102 rapidly hydrolyzes toaluminum trih ydroxide at room temperature. The melting pointof NaA102 has been reported to be 1923K. However, whenheated in air, it loses sodium oxide above 1273K and

decomposes to aluminates richer in A1203, the so-called betaaluminas.

39

Alcoa Laboratories

2300

1300

GA2“,4

Liquid“q=

/’

13+A1203

80 90 100

Mole “IoA1203

Na20-A1203 Phase DiagramFigure 3.4

3.23 Sodium Beta Aluminas

Sodium beta alumina was first reported by Rankin and Merwin(1916) who considered il to be a new modification of A1203,

Later investigations by Stillwell ( 1926), Bragg et al. (1931),Beevers and Brobult (1936), Beevers and Ross (1937), andAdelskold (1938) showed that sodium oxide is necessary forthe formation of beta alumina, The unfortunate designation has

been continued in the literature nevertheless.

Beta-alumina is prepared by vtious methcds. Rankin andMerwin (1916) observed small hexagonal platelets of ~-alumina

on cwling of molten aluminum oxide, Sodium ~-alumina

crystallizes from melts containing aluminum oxide and sodiumoxide or other sodium compounds such as sodium fluoride,

sodium hexafluoroaluminate (cryolite), or sodium nitrate.Foster (1964) determined the phase field of p-alumina in theNti-Na3AlF6-A1203 system, Saalfeld (1956) grew p-aluminaepitactically on corundum crystals which he had exposed to

cryolite vapr or fused cryolite at 13WK. The literaturedescribes a number of other reactions ktween aluminum oxideand hydroxides and alkali metal compounds that lead to theformation of ~-alumina,

During preparation of ~-alumina, the metastable form~-alumina is often encountered. Beta” is considered a

nonstoichiometic form with sodium oxide contents betweenNa20 5A1203 and Na20 8.5 A1203. There has been some

disagreement on the stabOity region of this form in theNa20-A1203 phase diagram. CoOongues, Thery, and Boilot(1978) suggest that ~-alumina is metastable in the pure

Na20-A1203 binary system.

Beta” always occum together with 13, regardless of the methodof preparation. According to LeCars (1972), 13° is imeversiblytransformed to13at 1823K. The region of coexistence of the

two phases is shown in Figure 3.4. However, it has ken foundthat in the presence of additions such as MgO and U20, the ~phase can be stabilized even at temperatures as high as 2~K

(Imai and Harata, 1972). Also, the replacement of A13+ by

Ga3+ shifts the existence range and stability of the phases(Boilot, Tbery, and Collogues, 1973). The ~ phase kcomes

more stable; pure ~-gallate can be obtained without thepresence of the ~-form for a small composition range.

Bragg et af. (1931) and Beevers and Ross (1937) determinedthe main structural features of p-alumina which they assumed

to have a composition of Na10 .11 A1203, They foundhexagonal symme~, space group D~h-P63/MMC. Latticeconstants are ~ = 0.558 nm, co = 2.245 nrn, The unit cell iscomposed of two spinel-like blocks separated at a mirror plane

by a loosely packed layer of oxygen and sodium ions, Thislayer is linked to the blinks by bridging oxygen ions(Figure 3.5a). Vacant sites in these planes give rise to mobility

of the sodium ions resulting in high values for two-dimensionalionic conduction (u E 10-20 cm– 1 at room temperature).

Beta’’-glumina has rhombohedral symmetry, the space group isD3d-R3m;

~= bO=0.56nm, CO=3.3 nm

The unit cell contains three spine] blocks of the same ty~ as

those in ~-alumina (Figure 3 ,5b). The ideal composition ofNa20 .5.33 AlzOq is obtained by the presence of two sodiumions in an interlayer which is not a mirror plane.

Both ~ and ~ are nonstoicbiometric. According to Collogues

et al. (1984), electric charge compensation in nonstoichiometric~-alumina is obtained by an excess of oxygen ions ininterstitial positions; in ~ excess positive charges are balanced

through replacement of Al3+ by ,Ower.vale”t cation. Besides

Collogues et al, (l.c.), Fanington and Briant (1979), Roth etal. (1976) and Stevens and Binner (1984) also reviewed the

structure, composition, and conductive properties of~-aluminas.

40

Alcoa Laboratories

Two-dimensional disorder of the sodium ions located in theintedayers between spinel blocks allows high mobility of theseions. The result is not only high ionic conductance, but alsosignificant ion exchange properties. Monovalent ions such anLit, K+, Rb+, Ag+, Cu+, H30+, NH4+ can be completelyexchanged. Only partial exchange is possible with divalent ionssuch at Mg2+, Ca2+, or S?+ (Kummer, 1972).

3.24 Beta Aluminas as Sofid Electrolytes

Since the discovery of the fast ionic conductivity of sodiumbeta alumina (Weber and Kummcr, 1967), research into thepropefiies of this material has grown rapidly, particularly workin connection with the development of the sodium-sulfurbattery. This battery uses as an anode molten sodium which isseparated from the molten sodium polysul fide cathode by asolid electrolyte, p-alumina. During operation, sodium ionsmigrate through the p-alumina, producing sodium polysulfidesat the cathode. Electrons flowing in tbe external circuitgenerate an open circuit voltage of 2.08 V.

Tubes made of polycrystalline p-alumina are generally used inthese batteries. Both sintering and fusion processes areemployed in the fabrication of these tubes. A crystal growthtechnique called “edge defined film fed growth” (EFG) hasbeen developed to produce large single crystals having aconstant cross section (Cocks and Stormont, 1974). Accordingto the methods of synthesis described in the literature, the ~“phase is the predominant component of the solid electrolytes(Collogues et al., 1978).

Production techniques are constantly being improved to meetmechanical, themal, corrosion, conductivity, and other

requirements critical to commercial development of thesodium-sulfur battery.

The formation of an MgO I IA1203 magnesium aluminatewhich has ~-alumina strocture was reported by Bragg et al.(193 I). Kordes (1935) synthesized the Li20 5A1203 lithiumaluminate and showed that this compound crystallizes in thespine! (more correctly, inverse spinel) structure. The lithiumions occupy octahedral and tetrahedral sites in the cubic,close-packed oxygen lattice; no lithium-oxygen layers arepresent between spinel blocks. This compound has been named“lithium zeta.” Lejus and CoOongues (1961) and kjus (1964)showed a variability of zeta-alumina composition betweenLi20 5A1203 and L120 8A1203.

Scholder and Mansmann (1963) showed that ~-alumina typestructures are also formed with K20, Rb20, and CS20. They,however, concluded that botb these and sodium ~-alumina areI:6 compounds of the general type A20 6M203. Thisassumption is not acceptable in view of the variability ofcomposition of ~-alumina discussed earlier.

Hexa aluminates of alkaline earth oxides, AEO 6A1203(AE = Ca, Sr, Ba), are also reported in tbe literature(Adelskold, 1938; Ligerquist, 1937). The structures of thesecompounds are of the magnetoplumbite type of theconesponding femite system. This type is very similar to thep-alumina structure.

Structural data on p-alumina and related compounds aresummarized in Table 3.2.

Table 3.2Strucural Properties of Beta Aluminas

Phase

Beta SodiumAluminate

Gamma SodiumAluminate

Sodium Beta

Sodium Beta

PotassiumBeta

MagnesiumBeta

Calcium Beta

StrontiumBeta

Barium Beta

Lithium Zeta

~ToroPov; Toropov al

42

Formula

NaA102

N*102

Na20 11A1203

Na20 5AIz03

K20 11A1203

MgO 11A1203

Cao 6A1202

SrO 6A1203

BaO 6A1203

Li20 5A1203

Galakhov

CrystalSystem

Orthorhombic

Tetragonal

Hexagonal

Hexagonal

Hexagonal

Hexagonal

Hexagonal

Hexagonal

Hexagonal

cubic

SpaceGroup

c%

D;h

Moteculesper

Unit Cell

4

1

i

1

2

2

2

2

Unit MS Length, nm

a

5376

5325

,558

,561

,556

.556

.554

.556

,558

.790

b

5216

c

.7075

.7056

2.245

3.395

2.267

2.255

2.1B3

2.195

2,267

Densityg/cm3

2.693

3.24

3.30

3.6$

3,61

Reference

Th6ry andBriancon

Th6ry andBtiancon

Beevers andBrohult

Th6ry andBriancon

Beevers andBrohult

Bragg et al.

Lagerquistet al

L~gerquist

et al

Adelsk61d

Kordes; Braun

I


f

I

3.3 The Na20.A1103-H20 System

This system is the basis of the hydrothermal Bayer process

which is the most widely used process for producing aluminum

oxide from bauxite, Worldwide, the annual production exceeds

30 million tons.

In the Bayer process, bauxite is treated at temperatures between

415 and 560K with caustic solutions containing 140 to 350 g

Na20 per liter. Temperatures and sodium oxide concentrations

are selected according to the aluminum hydroxide mineral

prevailing in tbe bauxite and the type of pressure vessels

(autoclaves) available.

The digest reactions are

Al(OH)t + NaOH % Na+ + A1(OH)I

AIOOH + H20 + NaOH s Na+ + Al(OH)~

In the first step of the process sodium aluminate solutions are

obtained having a molar ratio Na20:A1203 between 1.5 and

1.7. To recover the dissolved aluminum hydroxide tbe liquor is

diluted, if necessary, to 130 to 150 g Na20 per liter and cooled

to 315-335K. Large quantities of recirculated seed gibbsite (up

to 400’?6 of the amount dissolved) are added and the suspension

is stirred. Gibbsite crystallizes out until a molar ratio of about 4

is reached. Above 375K, boehmite is the crystallization

product,

Tbe khavior of aluminate solutions is remarkable in two ways:

For one, they must be diluted for the precipitation of hydroxide

to be initiated. Moreover, gibbsite crystallizes only after

intensive seeding, even though tbe solutions are highly

supersaturated at 3 15-335K.

At low concentrations (less than 1 molar) the monomeric

hydroxy-aluminate prevails. This was found by Jabr and

Pemoll ( 1965) who applied EMF and diffusion measurements

to study the ionic stmcture of tbe solute species, Moolenaa et

al. ( 1970) confirmed this result. These workers investigated

sodium aluminate solutions using infrared and Ramanspectroscopy as well as Nuclear Magnetic Resonance, Their

data were consistent with the existence of the tetrahedral

Al(OH)~ ion. At concentrations above 1.5 molar, condensation

of tbe tetrahydroxy. aluminate occurs, leading to the species

[(OH), A1O Al(OH),]’-. According to Moolenaar, tbemonomeric and the dimeric ions cxxist in solutions of

concentrations up to 6 moku.

Concentrated sodium aluminate solutions me extremely viscous.Wefers (1967) suggested that, besides polymerization of tbe

Al(OH)~ ion, association of the hydrated complexes throughhydrogen bonds can be tbe cause of the high viscosity, Thus,several factors may account for the unusual crystallizationbehavior of aluminate solutions. The high viscosity reduces the

rate of diffusion. Furthermore, tbe transition solute - solidrequires activation energies for the following steps: breaking ofthe hydrogen fwnds; breaking up of tbe dimeric (or higherpolymeric) complexes that do not tit into the AI(OH)3 lattice;

finally, change of the A1-cwrdination number from 4 to 6.

Glastonbury ( 1969) in a review on the nature of sodiumaluminate solutions concluded that considerable hydrogen

bonding ktween aluminate ions xcurs at moderateconcentrations. Dehydration of AI(OH); to AIO~ takes placewhen the concentration exceeds 25% Na20. The solid-liquidequilibria in the ternary system indicate that this dehydration isnot merely a function of concentration, but also depends

strongly on the temperature, Solution isotherms for 303 and333K have been given by Fricke and Jucaitis (1930). Sprauerand Pierce (1940) have examined the solution equilibria at298K, Cistjakova (1964) at 293K, Solution ~lythenns, based

on their own work and literature data, were compiled by Volfand Kuznetsov (195 I ) and Russell, Mwards, and Taylor(1955). Berecz and Szita (1970) determined solubilities in tbesystem AI(OH)3 - NaOH-H20 by an electrochemical method.

Isothermal sections of the phase diagram Na20-A1203-H20 at333, 368, 423, and 623K, published by Wefers in 1967, areshown in Figure 3,6. Pol ytherms replotted from We fers’ data

are represented in Figures 3.7 and 3.8,

Several solid phases are in equilibrium with the solution. At tbelowest temperature, 333K, AI(OH)3, NaOH H20, and a

sodium aluminum bydroxo hydrate, -2 NaA101 2.5H20,cmxist with the liquid phase. The point of maximal volubility;i.e., the nonvariant point at which trihydroxide and aluminate

we in equilibrium with the solution, lies near the concentration21 wt. % Na20 and 20 wt. % A1203. Between 343 and 348K, at22% Na20 and 237. A1203, AIOOH forms as a fourth solidphase in equilibrium with the liquid. This may indicate the

occurrence of a dehydrated aluminate ion, AIO~, in tbesolution.

Assuming that gibbsite and diaspore are the thermodynamicallystable phases in the ternary system, Wefers (1967) determinedsolution polythems using equal amounts of y-Al(OH)3 and

a-AIOOH as solids. The inflections in the polythenns shown inFigure 3.7 coincide with an increase of a-AIOOH at higher,and a growing proprtion of y-Al(OH)3 at lower temperatures.

The inflections can & extra~lated to 373K at 070 Na20, andto 348K at 22% Na20, 23% A1203, The temperature of the

43

..-. . .Alcoa LaDoratorles

transition AI(OH)3 + AIOOH + H20 is obviously a function of

the activity of water.

At 368K the solution field has widened, NaOH has replaced

NaOH H20; and Al(OH)J, AIOOH and 2NaA102 5H20 are

coexisting solid phases. The hydroxo aluminate hydrate

decompses to ~-NaA102 above 41 OK; AIOOH is the only

solid phase in the A1203-H20 boundary system.

H20

t.laOHI(OH)3

70 70

90

4

2 Na A102-2.5 H20

With increasing temperature, the solution field widens and the

nonvariant pint moves toward tbe Na20-A1203 axis. The

solubllity curve straightens out, approaching a constant molarratio Na20:A120q. Slightly above 595K (the melting point ofNaOH in the Na20-H20 bundary system) AIOOH becomes

unstable at the point of maximum volubility. a-A1203 forms,widening its phase field with increasing temperature

(Figure 3.6D). At 633K, diaspre is replaced by corundum atO% Na20.

GA 20246 .,8

H20

N OH

90

~A,203Na20~A12Q3 ..2Q

H,O H,O

N

Phase Diagram Na20-A1203-H20Figure 3.6

44

—


I I I

Wt. “h NasO

m - AIOOH 17.5

&

y - AI(OH)3 15

12.5

16

7.5

5\

323 373 423

Temperature, K

Volubility of A1203 as a Function of Temperature andConcentration of Na20

Figure3,7

a-AiOOH as solids. The inflections in the polytbenns shown i“Figure 3.7 coincide with an increase of a-AIOOH at higher,

and a growing proportion of Y-AI(OH)3 at lower tem~ratures.

The inflections can be extrapolated to 373K at O% Na20, andto 348K at 22% Na20, 23% A1203, The temperature of the

transition Al(OH)q ~ AIOOH + H20 is obviously a function ofthe activity of water.

At 368K the solution field has widened, NaOH has replacedNaOH HZO; and Al(OH)~, A1OOH and 2NaA102 5H20 arecmxisting solid phases. The hydroxo aluminate hydrate

decomposes to 13-NaA102 above 41 OK; AIOOH is the onlysolid phase in the A1203-H20 boundary system.

With increasing temperature, the solution field widens and the

nonvariant point moves toward the Na20-A1203 axis. Thevolubility curve straightens out, approaching a constant molxratio Na20A1203. Slightly above 595K (the melting pint of

NaOH in the Na20-H20 boundary system) AIOOH &comesunstable at the point of maximum volubility. a-A1203 forms,widening its phase field with increasing temperature(Figure 3.6D). At 633K, diaspore is replaced by corundum at

O% Na20.

The transition temperatures Al(OH)j ~ AIOOH - A1203

determined in the ternary system at 0% Na20 agree with thosepublished for the AIZ03-H20 boundary system (Chapter 3.2),Higher Na20 conce”tratio” lowers the activity of water in the

system, causing a concomitant decrease of the transformationtemperatures.

From systematic seeding experiments with gibbsite and

bayerite, Wefers ( 1967) concluded that bayerite dms not have astability field in the system Na20-A1203-HI0. Gibbsiteprecipitated from sodium aluminate solutions contains small

amounts of Na20. The sodium concentration is lowest whengibbsite crystallizes at low supersaturation; i.e., undernear-equilibrium conditions.

The solution-p~cipitation cycle of the Bayer process,particularly the fact that solutions must be diluted to induce

crystallization, is explained by the diagram in Figure 3.8: Asolution having the concentration and molar ratio representedby Point A in the plot is used to dissolve aluminum hydroxideat temperature B, until saturation is reached. Datution toconcentration C and cooling to temperature D supersaturates the

solution by the amount C-D. Crystallization accelerated byseeding brings the alumina concentration to point D. Byvacuum evaporation, the spent solution is brought back to the

initial concentration A. Formation of gibbsite particles involvesnucleation, growth, and agglomeration mechanisms (Misra andWhite, 197 1), Rates of crystal growth are very slow only afew micrometers PI hour. For this reason, the large quantitiesof gibbsite “seed” crystals are needed to obtain an economically

acceptable yield of gibbsite in a reasonable time period.

G. m246.20

I I I1.0 1

0 10 20 30 40

Wt. “k Na20

Solution Isotherms h the System Na20-A1203-H20Figure3.8

45

Alcoa Laboratories

4. Phase Relations of Aluminum Hydroxides and OxidesUnder Nonequilibrium Conditions

4.1 Properties and Applications of Calcined Hydroxides

Large quantities of metallurgical alumina (30 X 106 t each year)are produced by thermal decomposition of technical gibbsite atambient pressure. In the early years of the now century old

Bayer and Hall-Heroult processes, it was recognized thatcalci nation temperatures of 1400- 1500K were needed to obtainan anhydrous A1203. This is a more than 800K higher

temperature than that required for the conversion of hydroxideto corundum under the equilibrium pressure of water in thesystem A1203-H20 (Figure 3. I ).

While the fully calcined oxide was found to be identicd with

the mineral corundum, it was discovered that heating afuminumhydroxide to temperatures lower than that required for complete

dehydration, produced a material with unique pro~rties. Crossreprted as early as 1879 that gibbsite dehydrated attem~ratures klow 1lWK is a desiccant. Johnson (1912)

determined the adso~tion limit for water to be 18% by weight.

Later investigations revealed that partially dehydrated aluminumhydroxides not only reversibly adsorb water, but also anions,hydrated cations, and a num~r of organic compounds. Highspecific surface area and prosity, thermal stability andchemical purity, as well as easy agglomeration and shaping,

made such aluminas attractive materials for chemicalprocessing.

Partially and fully calcined aluminas are employed as supportsfor metal catalysts (Oberlander, 1984). In a number of catalyst

systems the alumina support is not inert, but contributes to thereaction. By itself, “active” alumina is an effective catalyst forisomerization, dehydration, and dehydrogenation reactions. The

commercially most significant application is the catalyticremoval of hydrogen sulfide from gas streams, according to theClaus reaction:

2H2S + S02 -i 2H20 + 3S

The preparation of aluminas for catalytic processes and forseparation technology has been described in countless patents

and publications since the 1940s. It was soon recognized thatspecific properties such as surface area, pre size distribution,or catalytic activity could be controlled by varying heating rate,calcination temperature and furnace atmosphere. Tbe stronginfluence of composition and structure of tbe starting material

on final properties also became apparent,

Scientific understanding of the relationships ktween structuraltransformations that occur in tbe course of thermal treatmentsof hydroxides and the chemical and physical properties of theproducts developed concurrently with an increasing commercial

46

application. With the emergence of improved tecbniq”es forchemical and physical characterization of solids and reactionsin heterogeneous systems, progressive y dee~r insight wasgained into the kinetics of dehydration, micro-morphology,c~stal stmcture of calcination products, the development of

surface area and porosity, and lately, the configuration and

reactions of surface sites at tbe atomic level. Besides manyindividual investigators, institutes and research groups directedby the following scientists have made valuable contributions to

this field of physical chemistry: J. H. DeBoeq R. Fticke;H. Ginsberg; G. F. Hiittig; J. W. Newsome; A. S. Russell;R. Tertian and D. Pap&e; and K. Torkw.

Stumpf et al. in 1950 published the results of extensive X-ray

diffraction studies showing that, in the course of thermaltransformation of aluminum hydroxides, a series of stmcturalforms develops the properties of which are determined by the

stmcture of the starting material and the temperature ofcalcination. Later investigations by numerous workers

confined the principal features of the structures determined byStumpf as well as the reaction paths.

A general scheme of the decomposition sequence is illustratedin Figure 4.1.

4.2 Thermal Decomposition of Aluminum Trihydroxides

4.21 Gibbsite

Achenbach (1931) and Damerell et al. (1932) reported thatgibbsite crystals after complete conversion to comndum retain

their original habits. A comparison of the scanning electronmicrograpbs in Figure 4.2 and Figure 4.3 demonstrates thispseudomorpbic transformation. A large internal porosity

develops in the process of conversion, as the density increasesfrom 2.42 to 3.98 g/cm3, Loss of water, development ofporosity and internal surface area, and change of density with

increasing temperature are presented graphically in Figure 4.4.More than 20% of the mass is lost in the temperature rangebetween 450K and 650K; another 10-1270 within tbe following

2CX3K.The remaining bydroxyl ions are driven offprogressively; anhydrous a-A1203 is obtained at a calcinationtemperature of 1375- 1450K.

The reaction path and kinetics of tbe dehyroxylation process areinfluenced by pmicle size of tbe gibbsite, rate of heating, and

the water vapor pressure in the atmosphere surrounding tbehydroxide, This is illustrated by the DTA curves in Figure 4.5.A coarse, technical gibbsite undergoes a sharp endothermic

reaction above 450K, which is followed by an exotbernticenthalpy change around 5WK and a second strong endotbertnicmaximum near 625K, A weaker endothermic effect occurs at

800-825K, and an exotherrnic peak at approximately 1450K.The sample having a particle size smaller than one micron only

—

.- -- . .oxides and Hydroxides Ot-Aluminum

“c

100 200 300 400 500 600 700 800 900 1000 1100 1200

I I I I I I I I [ I I I

=~CHI 9 Kappa I Alpha

*“ Gamma I Delta 1 Theta I Alpha

Diaspore — Alpha

I I I I I I 1 I I I I I

400 500 600 700 800 900 1000 1100 1200 1300 1400 1500

K

Transformation Sequence A1(OH)3 + A1203Figure 4.1

shows one strong endothermic reaction between 475K and

620K.

DeBoer et al. (1954) recognized that the exothennic reactionnear 500K is caused by the formation of boehmite, Y-A1OOHas a result of hydrothermal conditions occurring inside coarse

gibbsite particles from which water cannot rapidly evaporate.Boehmite forms concurrently with chi-alumina, a disordered,nonstoichiometric oxide containing several percent residualhydroxide (Glemser and Rieck, 1956).

The Y-AIOOH does not develop in particles of gibbsite smallenough to let water escape without significant increase inpressure.

The chl-form is replaced by kappa-alumina above IOCOK. Thetemperature range in which this transition alumina occurs

depends on the thermal and chemical history of a particularsample; particular y the content of alkali metal ions affectstransformation (Ginsberg et al,, 1957). Kappa alumina has aloss on ignition of 1-2%. Only after prolonged heating above1400K is this value diminished to 0.1 -0.2%, as completeconversion to corundum is obtained.

An X-ray indifferent (“amorphous”) transition form,rho-alumina. is obtained bv “flash calcination. ” i.e.. heati”~... ...—..-the hydroxide for a few seconds in a fluidized bed employinggas temperatures higher than 1100K, Rho-alumina is alsoformed as a result of rapid heating of gibbsite to 575-700K in

high vacuum. Upon heating in air at higher temperatures, rhotransforms to eu-alumina, above 11WK to theta whichconverts to alpha-aluminum oxide at 1400K.

Rho alundna is very reactive because of its high free energyresulting from lattice disorder and kuge surface area. In contact

with water, rho afumina reacts to form crystalline a-Al(OH)3 or

Y-A1OOH, depending On temperature and PH20 (Tertian andPap6e, 1958). This reaction proceeds at a significant rate even

at temperatures below 373K; itis utilized commercially for

agglomerating ~d shaping flash calcined alumina throughinterlocking bayerite crystals grown on the surfaces of ptiicles.

The partially hydroxylated agglomerates retain their shapesafter a subsequent dehydroxylation treatment (“re-activation”) at750-SOOK (Osmont and Emerson, 1965).

Smafl amounts of fluoride in the furnace atmosphere drasticallychange the sequence of dehydrox ylation. Ch]-alumina dircctl y

47

Alcoa Laboratories

D. R. Micholas 4,0mx

Acicular GibbsiteFigure4.2

converts to corundum near 1200K; the tsansfonnation isindicated by a sharp exothennic effect (we Figure 4.5) whichoccurs as the surface area decreases by a factor of akut lCOfrom >30 m2/g, and the loss on ignition falls below 0.2%

(Wefers, 1964). Tabular crystals of corundum are formed(Figure 4.6).

Berates, chlorides and fluoroborates also have been used as“mineralizers,” i ,e., agents which affect transformation

temperature, rate of transformation and size and habit of thecrystalline product. A number of procedures were described inthe patent literature and in technical publications (e. g.,Gopienko et al., 1970; Cherdron et d., 1981; Kruse andHausner, 1979), The chemical reactions involved in themineralization process, however, have not been citified so far.

4.22 Bayerite and Nordstrandite

The crysti lattices of bayerite, nordstrandite, and gibbsite mmade up of identical stmctural elements that are stacked in a

slightly different pattern, as was shown in Chapters 2.22-2.23.It is, therefore, not surprising that tbe debydroxylationsequences of the three trihydroxides we very similar.

The DTA and TGA curves of coarse or tine grained bayeriteand gibbsite me nearly identical. Some boehmite forms due to

intergranular hydrothermal conditions in lmge particles ofbayerite. Finely divided material transfoms between 500K and620K to the highly disordered eta-alumina. A. ktter ordered

transition form, theti-alumina, occurs above 1070-1120K. Thisalumina is converted to corundum above 140iIK in air.

Calcination under high vacuum leads to X-ray indifferentrho-alumina, which subsequently transforms to eta, theta, mdalpha. Among others, Tertian and Pa@e (1958), Torkar,

Egghart et d. (1961), Sato (1962), and Torkar and Bertsch(1961 ) investigated the sequence of dehydroxylation ofbayerite. The latter workers reported that bayerite containing

alkali metal oxides decomposes at lower temperatures, sodiumoxide having the greatest effect. Russell and Cochran (1950)measured the surface area of calcined bayerite. They found a

slightly greater specific surface area at optimum calcination

temperature than for gibbsite.

According to Hauschild (1963) the dehyroxylation sequence of

nordstrandite is identical with that of bayerite.

4.3 Decomposition of Aluminum Oxide Hydroxides

4.31 Boehmite

Goton (1955) re~tid heating of well crystallized boehmite ata rate of 4. 5“ pr minute to produce a strong endothermic effect

at 780K. The total heat required for debydroxylation was 145k.f/mole. Debydmxylation to a residual 0.1 mole of HIO ~rmole of A1203 could & achieved also by prolonged heating at725K. The specific surface aea increases to <100 m2/g, and a

transition alumina, gamma (Ubich, 1925) forms.

D. R. Mlchdas 4,MDX

Acicular CorundumFignm 4.3

I

4s


., !0246 ,,

~c200 400 000 800 ?Ow 1200

I I I I I k/

, 3.9

/“

LI//

‘\ /.- ~ien.ily

/

/

i sP. s.fiace .,.,

I

\ L.JSSon ignition

/

\

~-

30

35

0 500 700 9W 1100 ~300 15~

K

-c200 WO 600 800 1000 1200

I I I I I I

sPec. .. fiac. are.

I I I I I

Surface Area vs. Temperature of Calcined GibbsiteF,gure4.4

GA 20246.24

200 400 600 800 100012001400

+

~~~~~~~~Gibbsite + 0.1 wt.”A AIF3

I I I I I I I I

400 600 80010001200140016001800

K (C. Bates)

“c

200 400 600 800 1000 1200 14000, m,,. 2,

I I I I I I I 1 I— Boehmite----- Gelatinous Boehmite

. . . . . . . . . . . Diaspore

‘r ti I I I I I I I I

400 600 600 1000 120014001600 1800

K (C. Bates )

DTA - Curves of Heated HydroxidesFigure4.5

49

Alcoa Laboratories

J. J. Ptaslenskt 2,fmox

Tabular Alumina from Gibbsite Calcined with FluorideFigure 4.6

Goton’s data agree well with the tindings of Lippens (1961)who investigated the dehydroxylation of well c~stallized and

of gelatinous kehmite. The latter material develops surfacearea and transition aluminas similar to those forming in thedecomposition sequence of bayerite. This was confirmed byAbrams and bw (1969). These workers investigated the

thermal decom~sition of the fibrous hoehmite described byBugosh (1959). They found the fibrous AIOOH to consist ofwell crystallized and gelatinous boehmite. The well crystallized

part decom~sed to gamma, then formed theti-alumina whichwas followed by alpha, The gelatinous form converted to eta,

then via theta to alpha. (See also Clark and Lanutti, 1983).

X-ray indifferent (“amorphous”) gelatinous hydroxide

undergoes a succession of transition forms similar to that ofgelatinous boehmite. Up to about 900K, the product ofdehydroxylation remains X-ray indifferent, Above this

temperature a poorly ordered, spinel-type transition aluminadevelops, which transforms to theta near 1300K and to a-A1203above 1450K (Sate, 1972). The spinel-t ype transition aluminahas ken referred to as eta, gamma, or “gamma prime.” This

inconsistency reflects the difficulty of differentiating thesepoorly crystallized forms solely on the basis of a few, broaddiffraction peaks.

Teichner’s group claimed that an oxygen-deficient, black,n-type semiconducting transition alumina can be obtained byheating gelatinous hydroxide to 770K at a pressure of 10-6 bar

(1959). Torkar and Mostad (1962) disputed this claim. They

attributed the black color to carhn originating from thereduction of COZ adsorbd by the gel.

Saalfeld (19@) found the sequence gamma-theta-alpha aluminawhen heating oriented aggregates of boehmite formed byhydrothermal treatment of large gibbsite crystals. According to

Tertian and Pa@e (1958), L1ppens (1961), and Sato (1962),delta-alumina occurs as an intermediate between gamma and

theta.

Lippens and DeBoer (19@), Saalfeld and Mehrotra (1965), andWilson (1979) showed by electron diffraction that the

transformation from boehmite to theta-alumina proceedstopotactically. * Although differing in details, the workers

agreed on tbe close relationship between the stmcture ofttoehmite and those of the transition forms, As a result of akinetic study of the transformation of well crystallized

boehmite, Wilson and McConnell (1980) concluded that thesequence gamma-delta-theta is kinetically favored because ofthe structural similarities among the three forms. They did,

bowevcr, point out that the reaction sequence and kineticsdepend on the propetiies of tbe original boehmite. This doesnot only apply to extremes such as gelatinous and well

crystallized boehmite, but also to highly crystalline y-AIOOHprcpaed hydrothermally at various temperatures ranging from420K to 570K, as shown by Tsuchida et al. (1980). Accordingto these workers, the activation energy for the

Y-AIOOH-?-AII03 transition can range from 220 to 325KJlmole depending on the tem~rature of preparation of the

boehmite sample and the water vapor pressure prevailing duringdebydrox ylation. Wilson and McConnell (1.c. ) reponed valuesof 185-205 KJ/mole. These examples demonstrate the difficultyof assigning definite pro~rties to the metastable phase;

Y-AIOOH, and the transition forms.

4.32 Dla.spttre

The DTA cume of diaspore is similw to that of boehmite; an

endothermic peak lies between 770K and 820K. Prolongedbeating leads to dehydroxylation at lower temperatures. Tbeequilibrium temperature for the transformation

a-AIOOH s a-A1203 was calculated to be 630K by Fyfe andHollander ( 19@). These workers determined tbe free energy of

formation of a-A1203 at 298K as 1835 kJ/mole.

X-ray diffraction patterns of decomposed diaspore show

comndum as the only crystalline reaction product. Because of

tbe stmctural similtit y ktween the hexagonal close-packedlattices of diaspore and comndum, the phase transformation

requires relatively small rearrangements of the oxygen and thealuminum psitions. From single-ctystal diffraction patterns

(“shofl-mge, in-situsmcture rewgeme.t)

I

I1

I

50

-.. -- ..- . . .Oxides and Hydroxides 01”Aluminum

De flandre ( 1932) concluded that the conversion of diaspore to

corundum takes place in situ (topotactic), the a~ b, c axes ofthe oxide hydroxide becoming the c, a, and [ 1I .0] directions ofa-A1203. Schwiersch (1933) confirmed Deflandre’s observationby an optical study of the orientation of newly formed

corundum with respect to the decompscd diaspore crystal.Ervin (1952) reported identical results.

Upon complete dehydroxylation diaspore loses approximately1590 of its original weight. Contraction of the remaining solidcauses a system of regularly spaced cracks to develop. These

are oriented parallel to (001), and ~rpendiculm to (O10), tbemain plane of cleavage of diaspore, At calcination temperatures

~low about 107OK, corundum nucleates within the lath-shapedsegments of the former diaspore in the form of crystallite of afew tens nm diameter, Electron diffraction patterns of thenonrecrystallized material surrounding the comndum particlesindicate a linear texture. The refractive index of the partially

convefied oxide, n~ = 1.60, is considerably lower than that ofdiaspore or comndum (We fers, 1962).

Lima-de-Faria ( 1963) found satellite reflections in single c~stalX-ray patterns of partially converted diaspore. This workerconcluded that an intermediate state of structure occurs during

transformation, caused by an alternating sequence of areas withordered and disordered cation distribution. Lima-de-Faria alsodescribed a pore system which developed during calcination.

He determined the activation energy for tbe dehydroxylation tobe -100 kJ/mole.

The transition aluminas cannot be considered tme polymorphs

of A1203. The low temperature forms in particulw containhydroxyl ion. Moreover, the sequence of transformation is notreversible; i.e., neither a- A1203 nor any of the high

temperature forms can k converted to one of the transitionaluminas that occur at lower temperatures. They may,therefore, be classified as thermodynamically unstable,although reasonably reproducible, states of structural

reordering. This point of view is supported by the fact that thestmcture of the stwing material determines the type andsequence of transition forms in the course of thermal

decomposition.

The schematic diagram of Figure 4. I only indicates general

ranges in which the transition forms occur, but should not beconsidered a “phase diagram. ” More than one form may bepresent simultaneously within a particle of decomposed

hydroxide; the low temperature transition aluminas initiallydevelop as domains in an X-ray indifferent matrix.

4.33 Other Methods for Preparing Transition Aluminas

Some of tbe transition aluminas can be prepared by methodsother than thermal decomposition of hydroxides, Stumpf andRussell (unpublished) identified transition aluminas on heating

ammonium alum. Rooksby (1950) noted kappa-alumina in tbethermal decomposition of hydrated aluminum chloride. Frickeand Jockers (1950) found gamma type alumina upon heatingAIC13 6H20. Caglioti and D’Agostino (1936) produced atransition alumina by blowing a stream of air rapidly through

molten aluminum.

Eta-alumina is formed from tbe oxidation of aluminum attemperatures near its melting pint (Hunter and Fowle, 1956)as well as in anodic oxidation of aluminum under certainconditions.

A fused alumina-aluminum nitride mixture, containing 0.347.N, yielded a pattern similm to that of delta (brig and Foster,1961 ). Delta-alumina has also been obtained by combustion of

aluminum in air or oxygen, and of aluminum carbide in oxygen(Foster et al., 1956; Wmenberg, 1952; Schneider and Gattow,1954).

Rooksby and Rmymans (1961) prepared delta-alumina byigniting ammonium alum impregnated with ammoniummolybdate, by quenching molten alumina, by rapid

condensation after passing an oxygen-hydrogen flame throughvapor from an aluminum chloride bath.

Plummer (1958) found that gamma- or alpha-alumina passedthrough an oxygen-hydrogen flae formed well-develo~d

crystals of delta- and theta-alumina, provided the final ptiicleswere spheres of less than 15 microns diameter. Larger pmicleswere alpha-alumina, Thus, extremely rapid solidification of

molten alumina, and ~ssibly contamination by nitrogen orcarbon, stabilize delta-alumina.

Foster (1959) reported a new alumina having an X-ray patternsimiku to that of the aluminum silicate, mullite, This materialwas designated iota-alumina, itformed during rapid quenchingof c~olite-alumina melts. Slow cooling gave comndum as the

only equilibrium phase. This mullite-like form was metastableand transformed through eta-alumina to alpha-alumina onheating above 1400K. Saalfeld (1962) described a similar formhaving slightly smaller lattice constants. He called it a

modification of A1203 with sillimanite structure, It is doubtfulwhether this material is a transition alumina, or rather a

compound in tbe system NaF-A1203.

4.4 Structures of Transition Aluminm

A comparison of tbe X-ray diffraction patterns of the transition

aluminas is given in Figure 4.7; Table 4.1 lists the most recentd-values published for these forms, structure data are compiledin Table 4.2.

To emphasize tbe relationships between tbe structures of theprecursors a“d the transition aluminas, in the following sections

these forms will be grouped according to the debydroxylationsequences discussed above.

51

Alcoa Laboratories

4.41 Chi and Kappa Alumina

Stumpf et al. indexed X-ray diffraction patterns of chi aluminaassuming a cub]c (not spinel-type) unit cell having a latticeconstant of 0.795 nm. Saalfeld (1960) examined aggregates ofchi formed in heated single crystals of gibbsite. He calculated ahexagonal cell with the parameters.

+ = 0,556 nm co= 1.344 nm

space group P61m or P631mcm

Certain stictural elements of the pseudohexagonal gihbsitelattice remain, es~cially the arrangement of anions in thedirection of the a-axis. Chi-alumina has a layer structure, thealuminum ions occupying octahedral sites within the hexagonaloxygen layers. The stacking sequence of the layers is stronglydisordered, perpendiculw to the c-mis.

Brindley and Chm (1961) also determined a hexagonal

structure with

~ = 0.557 nm, co= 0.864 nm

Thew workers pointed out that the structure of chi-aluminacontains appreciable amounts of hydroxyl iuns.

SaaJfeld calculated a hexagonal layer structure for kappaalumina. The kuge cell contains 28 A1203. Lattice constants are

G = 0.971 nm, co= 1.786 nm.

A layer sequence ABAC-CABA. with aluminum in bthoctabedrd and tetrahedral sites show some similarity with thestructure of ~-aluminas. According 10 Brindley (1961), two

types of hexagonal kappa exist, one with N = 0.970 nm, theother with % = 1.678 nm.

With the loss of almost all OH ions, the anion lattice of kappa

is htter ordered than that of the low temperature chi-form. Inaddition, a reordering of cations takes place in the transition chi+ kappa,

Space groups have not been reported for the two kappa forms,kcause of the unceflainties of interpreting powder diffractiondata of marginally crystallized specimens (see Figure 4.7),

4.42 Eta and Theta Alumina

Eta alumina has a spinel structure, Stumpf et al. (1950)detemined the cubic lattice constant to be

do = 0.79 nm

52

I

a:

eta

Pchi

80° 60” 40” 20”

20 CUK,,

X-RaY Oiflraction Patterns of Transition AluminasFigure4.7

While in the normal Me3+Me2+ spine] 32 oxygen ions and 24metal ions make up the unit cell, only 21 1/3 A13+ areavailable for the cation psitions in eta as well as in theisostructural gamma alumina, Lippens (1961) assumed eta to betetragonally deformed. According to Yamaguchi and Yanagida

(1964) the degree of tetragonal deformation is a function of theresidual OH.content, The cub]c close-packed stacking of

oxygen ions in the stmcture of eta is one-dimensionallydisordered in the direction perpendiculm to the c-axis of theprecursor bayerite (Lippens and OeBoer, 1964). John, Alma,


and Hays (1983) used solid-state nuclear magnetic resonance 2(NMR) to determine the coordination of Al in transitionaluminas. They found 35 * 4% of the AI ions in eta”on

tetrahedral sites. Cation vacancies, therefore, appea to beequally distributed over tetrahedral and octahedral psitions.

The same workers showed that in theta alumina prepared fromboehmite, the aluminum ions were almost exclusively inoctahedral coordination. This is in contrast to Sadfeld’s (1960)

stmcture analysis who assumed Al ions to preferentially occupytetrahedral sites. Theta alumina is isostructural with ~-Ga203,which can be grown in large single crystals (Roy, Hill, and

Osbom, 1953; Kohn, Katz, and Brcder, 1957). Geller (1960)and Saalfeld (1c.) used single crystals to determine thestructure of @-Ga203. Saalfeld inte~reted powder data of thetaalumina on the basis of the ~-Ga203 structure. He calculated amonoclinic cell, C~h C2/m having the constants:

a. = 1,124 “m, b. = 0,572 nm, co = 1.174 nm, P = 103”20’

The stmcture is a deformed spinel lattice and may beconsidered an intermediate between the cubic packing of thelow-temperature transition aluminas and the hexagonally close

packed comndum.

4.43 Gamma and Delta Alumina

The stmcture ofgamma alumina is also of thespinel type, butit is significantly more tetragonally deformed than the etalattice.

Saalfeld and Mehrotra (1965) determined the lattice constantsto be:

G = 0.562 nm, co= 0.780 nm

W1lsOn (I 979) gave:

aO=0.796nm, CO=0.781 nm

These authors, as well as Llppens and DeBoer (1964),investigated the structure of 8amma by selected area electron

diffraction. Allthree groups conclude tbattbe oxygen sublatticeof gamma alumina is fairly well ordered, much more so thanthat of eta alumina. Tbe reason forthc difference in order ofthe two spinel-type stmctures lies in the stmcturesof theprecursors: the bayerite lattice ismade up of single layers of

Al(OH)6mtahedra, and two water molecules me lost duringdehydroxylation. Intbeboehmite structure, theoctabedra form

cubic packed double layers; only one H20 is driven off in thethermal transformation.

As is the case with the eta structure, tbe AIIvAIVI* spinellattice ofgamma contains cation vacancies. Saalfeld and

*IV.V1 = Cmrdinati.n .umkm

Mehrotra (1965) assumed all octahedral sites to be occupied,

the cation vacancies being confined to the tetrahedralinterstices. This was confirmed by John, Alma, and Hays

(1983) who found, by NMR, only 25 t 4% of the tetrahedral~sitions wcupied, rather than the one-third (8 of 24) available

in the spinel stmcture.

Because gamma alumina contains several Wrcent hydroxylions, DeBoer and Houben (1952) klieved it to k a hydrogenspinel, analogous to the lithium spinel described by Kordes(1935). Soled (1983) postulated tha the hydroxyl ions are a

necessary component of the defect structure of gamma (andeta), their number tilng equal to the number of cationvacancies. The composition of this transition form,. therefore,can be expressed ax

A12D0,402.8(0H)0.4**

The structure of delta alumina is related to that of gammw the

long-range order having increased so that the unit cell can &described as a triple block of spinels (Lippens, 1961; Saalfeldand Mehrotra, l.c.). Tertian and Pa@e (1958) gave latticeparameters for a tetragonal structure as

~ = 0.797 nm, co= 2.347 nm

According to Lippens (1961) the dimensions ue

a. = 0.794 nm, co = 2.350 nm

Wilson (1979) reprted

an = 0.796 nm, co = 2.342 nm

While Rmksby and Rmymans (1961) calculated

~ = 0.796 nm, CO= 1.170 nm

i.e., a tetragonal cell half the size of that described by the otherworkers. Their material, however, had been prepared by

different routes (see Section 4.4). This was also pointed out hyLippens and Steggerda (1.c. ) These authors discussed the reasonwhy delta alumina does not appew as a transition in the

decomposition sequence of bayerite or gelatinous boehmite.They conclude that the formation of the triple spinel blockrequires a shucturd rearrangement in a direction that crosses

the major pore system in dehydrated bayerhe, and is~rpendlcul= to tbe shofi dimension of gelatinous boehmiteparticles. Reorderin8,

**❑represent vacancy

53

Alcoa Laboratories

dA—

2,402,272.111.981.531.39

Cti

rIll, hkl

40 31120 22230 32120 40010 311, 333

100 440

dA

4.62.62,402,271,971.521.401.21

L1.141.03

Eta

~

402060306020

100102010

Table 4.1X-Ray Diffraction Data of Transition Alumin8s

hkl

111220311222400333440533444563

G

rdA

4.562.802.392.281.9771.5201.3951.1401.0270.989

.884

.806

mma

T1/11 hkl

40 11120 22060 31150 222

100 40030 511

100 44020 44410 73110 80010 84020 844

I 1

KaI

dA

6.24.54.23.042,792.7o2.572.412.322.262.162.112.061.991.951.871.821.741.641,541.491.451.41.391.341.121,061.041.021.01).994

3

F—3020104060206030401010603040206030206010

303080I00301010

55

4020

—

dA

5.705.454.542.8372,7302.5662.4442.3152.2572.0191.95441,90941.79981.77651.73761.66071.62161.57151.54261,51201.~631.45261.42641.3883

1111 hkl

2 20010 00116 20180 400,40165 202,00214 11160 11145 401,310

35 402,20245 311,112

8 60130 600,31214 5106 602,4024 4032 6016 203,5112 113

25 3136 603

25 113,80125 02010 600,710

I 00 712,512

dA

7.66.45,535.104,574.073.613.233.052.8612,7282,6012,4602.4022.3152.2792.1601,9861,9531.9141.8271.8101.628i .6041.5381,5171.4561.4071.396

c

<—

444

e12124448

302560168

404

7540124

8848

16

18

50100

3

hkl

101102, 004ill112113114, 105115116107, 214117222302, 116312313314, 305226. ,1.1.104001.0.12318319, 2,2.1$26, 3.1.1!,1.145131.1.15523, 5164404.0,12

proceed at a sufficient rate ii the temperature ~~gion in which

delta alumina occurs. At higher temperatures the fomation ofthe better ordered theta alumina is favored.

4.5 Texture and Surface Properties of TransitionAluminas

4.51 Development of Porosity

The evolution of specific surface area and i“temal porosity in

the course of the thermal decomposition of aluminumhydroxides has received as much attention as have the

structural changes described in the preceedi”g sections, J. H.

54

therefore, requires such a high activation enerEv that it cannot DeBoer, R. Fricke, B, Imelik, A. S. Russell, K. S. W, Sing,S. Teichner and B. Tertian are among the resewchers most

prominent in this field. A summa3y of the literature up to 1970was given by Lippens md Steggerda (1970).

It was shown in Chapter 4.21 that a large sufiace area develops

as a result of the pseudomo~hic transformation of gibbsite(Figure 4.4). The dehydroxylation of bayerhe, nordstrandite or

gelatinous hydroxide follows the same general pattern,indicating that the common chemistry of the hydroxidesdetermines the reaction sequence, rather than the small

souctural differences.

Feitknechtet al. (1961 )considered diffusion ofprotons to be

theinitial step inthe dehydration reaction. Freund (1967)


Form

Gamma

Delta

Eta

Theta

Cti

Kappa

Iota

a ~~~ ~~,e~

Cwstal System

Tetragonal

OrthorhomticTetragonal

Cubic (Spinel)

Monoclinic

cubicHexagonalHexagonal

HexagonalHexagonalHexagonal

OrthorhomticOtihorhomhc

SpaceGroup

0;

C;h

D~h or

CL

Table 4.2Strucural Properties of Trasition Aluminas

Moleculesper

Unit Cell

12

10

4

10

28

., ...,..

43

Unit

a

.562

.425

.790

.790

1.124

.795

.556

.557

9.719.70

1.678

.773

.759

pointed OU1 that thermal dissociation of the hydroxyl mostlikely is not the source of protons for the reaction:

O--- H+ OH* O+HOH

The OH-stretch vibration occurs neu 370Q cm-1; the thermal

equivalent of which is an excitation temperature of -55WK forthis vibrational state. This is about an order of magnitudehigher than the temperature of the dehydration maximum for

AI(OH)3. Freund assumes that in the temperature range inwhich dehydroxylation occurs, the energy levels of single and

double protonated states of the OH ion are broad enough, andoverlap sufficiently, to permit proton exchange between

adjacent OH ions via an inelastic tunneling process. H20molecules formed through this mechanism diffuse out of theIatticc via anion vacancies (Frettnd, 1965). Assuming randomremo~al of hydroxyls, Freund argues that complete

IS Leng!

b

,780

1,275

.572

.778

.767

m

c

1.021

2.34

1.174

1.344

8.64

,1786.1766.1786

.292

.287

Angle

103~20<

... ,..

Density

a cm

3.2a

3.2~

2.5–3.6m3.56@

3.o@

3.1.3.3@

3.71 ~3.0

Saalfeld andMehrotra

Stumpf et alRooymans

Stumpf et al

Saalfeld (1960)

Stumpl et alSaalfeld (1960)Brindley and Choe

Saalfeld (1960)Brindley and ChoeBrindley and Choe

Stumpt et alSaalfeld (1960)

dehydroxylation of aluminum hydroxide is not possible at lowand moderate temperatures, ad that the rate of decompositionshould slow down after about two-thirds of the OH-ions have

been removed (equivalent to a 10% loss on ignition fortrihydroxides). The reason in both cases is the spatialseparation of hydroxyls, i.e., remaining OH-ions have nohydroxyls for nearest neighbors. Freund’s arguments are

supported by the experimental evidence.

Rocquerol et al. ( 1975, 1978) also imply diffusion of water via

a “step process,” i.e., vacancy hopping. They consider the“structural channels” in the gibbsite lattice, which originatefrom superposition of the “empty” hydroxyl octahedra in the

direction of the c-axis (see Figure 2.9), to be favored diffusionpaths. At high decomposition rates the (CKII) planes betweentwo layers of mtahedra are also prefemed sites for the removalof water.

55

Alcoa Laboratories

The porosity and concomitant large specific surface area of

transition aluminas are the result of the rapid loss of mass thatis not accompanied by a decrease of the external dimensions ofhydroxide particles. Much work has been done to relates~cific surface area, size and shape of pres, as well as pore

volume to tbe properties of the starting material and theconditions of the dehydration process. In earlier investigations,gas adsorption and resorption techniques, and mercuryporosimetry were applied to gain insight into the texNre of

porous transition afuminas. High-resolution electron microscopyhas been used since the 1960’s to directly observe and measure

morphological features inferred from the earlier, indirectexperimental approaches.

I

J. J. Ptaslenskl 2S0,000X

Chi Alumina lCFigure 4.8

The large surface area developing in the initial, rapid stage ofdehydration of trihydroxides (figure 4.4) is attributed to a

system of slit-shaped pores which are less than 2 nm wide.This pore system was earlier descrikd as consisting of <I nmwide, rod-shaped pores oriented perpendicular to the basal

plane of the trihydroxides, and slit-shaped, 2-3 nm poresparallel to these planes. “Ink bottle” sbaps were deduced from

BET adsorptiotidesorption isotbenns, obviously based on acombination of the two morphological features (de Boer et al,1954, 1956). Lippens (1961), Wefers (1964) and others showed

by transmission electron microscopy that a sponge-like texturedevelops within a coherent solid. Paraflel pores located in the

cleavage plane (001 ) separate the solid into Iamellae that are on

K. Welem 220,41xlx

Chi Alumina IlcFigure 4.9

56

—


the order of 2-3 nm thick, about twice the average pore width.A network of imegukuly shaped slits, about 1 MI wide,extends Perpendic”laly to, and is connected with, the pores

parallel to (001). Figures 4,8, 4.9, and 4.10 show these presystems in particles of chi and eta alumina, The intersecting

pore systems divide tbe Iamellae into interconnected, irregularlyshaped domains of solid. These measure on the average lessthan 10 nm in their longest dimension, equivalent to the lengthof about 10-15 A 106 octahedra. It is obvious from the electron

micrographs that the surfaces of the solid domains are “roughon an atomic scale and do not represent refined crystal faces.

Tbe longer-range periodicity necessary for the generation of anX-ray diffraction pattern is provided by an approximatelyparallel alignment of the irregukir domains in the layers of the

Al(OH)3-precursor.

J. J. Ptasienski 400,000X

Eta Alumina IlcRg”re 4.10

Surface areas resulting from this porosity can exceed 400 m2/g.Lippens (1961), Lippens and Steggerda (1970) and Wilson andStacey (1981) pointed out that the surface areas measured bygas adsorption methods, particularly when nitrogen is used,generally are lower than those calculated from microscopicallymeasured values for pore size and spacing. The obvious reason

is the limited access to, or incomplete tilling with gasmolecules of pores smaller than about 1 nm. Limitations of themicroscopic methcd lie in the dlfticulty of precisely

determining and measuring outlines and dimensions of highlyimegulm Pres, and differentiating micropores from clusters ofvacancies. The dimensions of both these morphological featuresoverlap; the definition, therefore, is arbitr~.

Total pore volume of transition aluminas having maximal

specific surface areas is in the range of 0.2 cm3/g A1203. If thecalcination temperature is raised above 650-700K, the surfacearea diminishes (Figure 4.4), but the pore volume increasesslightly. As shown in Figure 4.4, tbe adsorption of water at

high PH20 reaches its maximum nez the first inflection of thedown slope of the surface area-versus-temperature curve,staying nearly constant up to about 980K (Wefers 1964).

These data indicate that the smallest pores contribute the majorpart of the kirge specific sutiaces areas formed below 650K,but are not all accessible to water. Bridging of nmow (1-2 nm)pre openings as a result of dipole interaction of water is themost probable cause. Whh increasing temperature, the smalldomains of solid coalesce into lager, more stable weas of

oxide. As a result, the spacing between solid domainsincreases; so does the volume accessible to water molecules forsurface adsorption and capilkuy condensation.

At temperatures exceeding approximately 10WK, reordering isno longer confined to short-range consolidation witbin the layerskeleton of the decomposed trihydroxide, Activation energies

ae sufficient for three-dimensional sintering to take place, thusremoving the steric hindrance for the formation of long-rangeordered, crystalline oxide. Figures 4.11 and 4.12 show aciculasgibbsite crystals heated to 1100K. Although tbe.layer texture of

the precursor is still obvious, Imellae have fomed that are20-40 nm thick, about 10 times the thickness of the layers inthe particles of chi and eta shown in Figures 4,9 and 4,10, Thelamellae occur simultaneously with the formation of kappa

alumina. They aIe embedded in a network consisting ofinterconnected crystallite and pres, each on the order of

10-30 nm in size, equivalent to a surface area of approximately30-50 m2/g. This estimate is in good agreement with BETsurface areas determined for this transition alumina,

The orientation and morphology of the crystallite seen in thehigh magnification microgmph of Rgure 4.12 suggest that thetransformation ch] + kappa is a reconstructive and not atopotactic process.

Alcoa LaboratoriesI}

(

The dehydration of well-crystallized hoehmite causes a system

of Iarnellar pores to develop (Figure 4, 15) which are oriented

1

parallel to the (100) direction of the boehmite crystal. Wilson

(1979) and Wilson and Stacy (198 1) measured average porewidths of 0.8-1 nm; the spacing between the pores was

determined to be between 3.5 and 4 nm.(

;?

K. Wefers

Kappa Alumina from Acicular GibbsiteFigure 4.11

80.000X

The morphological pattern established in the temperature rangefrom 1000 to 1100K persists beyond the formation of a-A1203

at temperatures above 1450K. Lamellae continue to increase inthickness by fusion, as demonstrated in Figure 4.13. The gapsbetween Iamellae also grow with increasing densification,because the hahit and parallel texture of the original gibbsite

crystal are retained (Figure 4.3). In fragments of oxide havingnetwork texture, a comsening of crystallite and pores is

aPP~nt, aS shOwn in Figure 4.14. These morphologicalchanges shift the average pore size into the range of

100-200 nm. While the surface area diminishes to 1-3 m2/g, tbe

Surface areas obtained by dehydmxylation of boehmite we onthe order of less than 100 m2/g, pore volumes generally <0.08cm3/g. The smaller values, compared to those measured with

decomposed trihydroxides, reflect the loss of only 10-12% ofmass during the initial phase of dehydrox ylation. Wilson,

~re volume remains near 0.1 cm3/g A1203. K. Wefers 250,WOX

Complete recrystallization which obliterates tbe habit of the Kappa Alumina Ilc of Gibbsiteprecursor and the porosity, occurs at calcination temperatures Reordering by Bridging Lamellaeexceeding 1800K. Fi8ure4.12

58


>

I

K. Wefem

Alpha Alumina from Acicular GihhsiteFigure4.13

20,MXIX

McConnell and Stacy (1980) found that surface area, porevolume and spacing of the lamellw pores are controlled by the

water vapor pressure prevailing during dehydration at

temperatures below 700K. Llppens and Steggerda (l.c. ) showedthat the microporosity of dehydrated boehmite decreasedsharply above 850K, leaving a surface area of less than20 mzlg, but not affecting the lamellw shape of the pores.

The dehydroxylation of gelatinous boehmite follows a differentpattern. High surface areas exceeding 600 m2/g can by

-1

II

,)

..‘~

K. Wefers 2m,ooox

Alpha Alumina, Primary Crystals and PoresFigure4.14

obtained, as discussed in Chapter 2.12, by extracting the liquidphase from gels. The surface area of the resulting,open-network aerogels is the surface of the c~stallites formingtbe network, and thus a function of the crystallite size, which

may be as small as 2-3 nm. Dehydroxylation at temperaturesbetween 450 and 750K does not lead to significant additionalporosity, due to the small size of the solid particles, as Llppens

and Steggerda (1.c.) pointed out. Loss of water causes thecrystallite to shrink and the network to progressively collapse,leaving surface areas comparable to those obtained bydecomposition of crystalline trihydroxides.

59

—

Alcoa Laboratories

J. J. Ptasienski

Gamma Alumina from BoehmiteFigure4.15

300,DOOX

4.52 Relationships Bet ween Texture and Structure

Crystal structure and texture of the transition aluminas areclosely related. Although the stmctures of individual transitionforms differ in some details, they can & grouped into low and

high temperature types that are characterized by commonstmctural and micro-mo~hological features.

The eta, chi, and gamma forms occur in a temperature region(<1 lWK) in which diffusion rates are low, and in whichprogressive loss of mass in form of hydroxyl and hydrogen ionsleads to lower cmrdination numbers for a sig”ifica”t fraction of

the aluminum ions, and the breaking “p of the solid into small

domains separated by gaps (pores) that are as wide as 5 to 20

oxygen ion diameters,

For these reasons, structural consolidation is confined toshort-range reordering witbin the domains. A spinel. type

arrangement is energetically y favored as it can incorporateoctabedrdly and lower - tetrahedrally coordinated aluminum.

Hydrogen ions may k retained in areas of cation deficit tobafance excess negative charges.

Significant sintering and three-dimensional fusion of layersoccurs above 1lWK, as was demonstrated in the precedingsection. brig-range reordering, thus, kcomes possible, The

resulting delta, kappa, and theta forms represent, to varyingdegrees, structure transitions from the spinel-type to the

hexagonally close-packed corundum lattice.

Reordering of the oxygen sublattice precedes that of the cation

sublattice of the high temperature transition forms. Hydrogenions are retained in these structures until the conversion tocomndum is completed (see L,O. I. values in Figure 4.4A). Tbe

cause/effect relationship is not clem are protons retainedbecause of locaf charge imbalance due to the cation disorder, or

does the presence of protons in A13+ “positions” lower thedriving force for the diffusion of A13+ ions into tbe respectiveinterstitial sites? The observed rapid crystallization of corundum

at temperatures around 11OOK in furnace atmospherescontaining fluoride ions appears to suppofl tbe latterexplanation: fluorine is likely to react with protons to form HF,a stable gaseous species in this temperature range, Removal of

protons creates negatively charged sites which accelerate the

diffusion of cations. As this process takes place in the earlierstages of the Iamellar texture, distinctly tabular corundum

crystals should be ex~cted. These have been observed(Wefers, 19@),

The interde~ndence of micro-morphology and stmctureexplains the strong effect of the precursor on the path of

reordering of tic thermally decomposed hydroxides andoxide-hydroxides. In the writers’ opinion, tbe structural details

and distinctions of the individual forms have been somewhatover-emphasized in the literature, leaving the impression thatthey are thermodynamically defined polymorpbs of A1203,

rather than transitional stages in a continous solid-statereordering process,

4.53 Properties of Active Alumina Surfaces

The term “active” or “activated aluminas describes transition

forms which have the capability of catalyzing certain chemicalreactions and chemisorbing a variety of molecula and ionic

species. This reactivity develops concurrently with tbe changesin stoichiometry, texture, and structure that occur in the prmessof deh ydrox ylation.

60


Fully dehydroxylated, crystalline a-A1203 is not active, neitherare the well-ordered trihydroxides and oxide hydroxides ofaluminum. Activity is directly related to the defect stmcture ofthe transition forms. ~Is relationship has been the subject of

numerous investigations.

Tamele (1950) ascribed the obsewed hwis acidity (electron

acceptor) to exposed, coordinatively unsaturated (CUS)aluminum ions. Brbnsted-acid (proton donor) properties ofafumina surfaces were discussed by Trambouie and Pernn(1953) and Mapes and E1schens (1954), Cornelius et al. (1955)

assumed “strained” oxygen bridges, resulting from thecondensation of two adjacent hydrox yl ions, to be thecatalytically active centers. Flockbari et al. (1966, 1969)pointed out that active alumina surfaces exhibit both electron

acceptor and electron donor properties, and can catalyzeoxidation as well as reduction reactions.

Peri in 1965 developed a model for the progressivedehydroxylation of a cubic (100) surface plane. He used astatistical treatment (Monte Carlo) of m assumed, random

interaction of hydroxyl ions in a surface monolayer, Tbecomputer model demonstrated that an ordered, planaroxygetihydroxyl lattice is maintained until two-thirds of the

hydroxyl ions have been removed. Further condensation ofadjacent OH-ions creates local dlsordeL at 90.4% removal ofbydroxyl ions, no pair of neighboring OH-ions is left.

Surface migration of protons or hydroxyl ions is required for

any additional dehydration (see also Freund’s arguments,referenced in Chapter 4.51).

The sutiace states resulting from the removal of all OH-pairs

are shown in Figure 4.16. According to Peri’s model, fivecombinations exist of bydroxyl ions with oxygen and CUSaluminum ions (the latter located in the next lower layer). Five

OH-stretch vibrations are registered in the infrared spectrum ofthe transition aluminas chi, eta, and gamma. Peri tentativelyrelated the frequencies of these OH-stretch vibrations to tbenumber of oxygen ions adjacent to the respective hydroxyls.

Knozinger and Ratnasamy (1978) refined Peri’s mcdel,extending it to include (111) and (110) planes. These authors

calculated net chages for the five types of surface hydroxyls,taking into account that aluminum ions to which they ue linkedcan be either tetrahedral y mtahedrall y coordinated. A graphicrepresentation of the five hydroxyl configurations is given in

Figure 4.17. In the top row the most basic hydroxyls xeshown. These c6ny a net chmge of –0.5 and are coordinatedto one Alrv or Alvl ion, respectively. The OH-group linked tothree octahedrally coordinated aluminum ions (bottom) has themost acidic character.

(Peri, 1965)

Configurations of Surface Hydroxyl GroupsFigure4.16

The geometric models developed by Peri and Knozinger and

Ratnasamy are somewhat idealized. Es~cially the assignmentof particular configurations to defined c~stal planes appearsquestionable in light of the microscopic evidence whichindicates that tbe internal surfaces of transition aluminas are

“rough” on the scale of the size of Al-O coordination polyhedra(see Figures 4.10 and 4. 11). The concept, nevenheless, is ingood agreement with experimental observations.

Considerable research effort has been directed at understandingthe contribution of specific sites to adsorption and catalyticsurface reactions. Rosynek and Hightower (1972) classified

sites according to their specific reactivity; i, e,, isomerization ofbutene, chemiso~tion of butene, hydrolysis of ketones andnitrites, and D ~ H exchange, respectively. However, even an

apparently simple reaction such as the adsorption of C02 IIIaYinvolve more than one site. Parkyns (1969) and Monema et al.(1977) showed that, depending on the degree ofdehydroxylation of the alumina surface and the C02-pressure,

61

Alcoa Laboratories

H

o

IAl

/l\

H

\l/ 0 \Al Al

/l\ /l\

GA202,628

H

\i’//;\

H

\l/ 0 \l/

/y\ A, /y\

/l\

Five Coordination of Surface Hydroxyl GroupsFigure 4.17

four configurations of the adsorbed species occur, including abicarbonate group formed via transfer of a hydroxyl ion:

o/ /O\

Al—O—C Al C=o\ \/

o 0

Al—O\ /

o

C=o Al—O—C/

Al—O\

OH

More than one site also is involved in the cbemisorption ofwater. As shown schematically y in Figure 4.18, a Lewis acid

site is hydroxylated and an adjacent Br6nsted base (0-)protonated in the fust-layer reaction. Tbe second layer isadsorbed via dipole bonds with surface hydrox yls. While this

second layer, as well as liquid water tilling pores as a result ofcapill~ condensation, can be desorbed at temperatures lowerthan 400K, removal of the fust layer requires “reactivation” at

higher temperatures.

OH OH “ 2024’2’

—O—Al—O—Al—O —

/l\ /l\I

I– HOH

o–

—O— AI+ —O— AI— O —/l\ , /l\

\

+ HOH

OH OH

—O—AI—O—AI—O —

/l\ /\\

H+ 2HOH

H H H\o/ \/

o

H H\ /

i’ ?—o—Al—o—Al—o —

/l\ /l\

Adsorption of First and Second Layer of Water on A1203

Figure 4.1!3

62


?

,

1

[’

Table 4.3IR Frequencies of Surface Hydroxyl Groups

Wave No. No. of Adjacent Net Charge~m-l Oxygens on OH “Peri-Type”

3700-3710 0 +0.5 c

3730-3735 1 +0,25 E

3740-3745 2 0 B

3760-3780 3 –0.25 D

3785-3800 4 –0.5 A

Flockhart et al. (1965) found that the catalytic reduction oftetracyanmthylene on active alumina formed at temperaturesbelow 625K is associated with a surface hydroxyl (Br6nsted

base). On alumina prepared at 775K and above, thecombination of an electron-donor oxygen (f.ewis base) with a

CUS Al ion is believed to catalyze this reaction. Ghorbel et al.(1973, 1974) pointed out that electron acceptor and electrondonor sites are involved in tbe catalytic isomerization ofI-butene, and that these sites can ~ deactivated by selective

poisoning with bases such as ammonia and acids such as aceticacid, respectively. Knozinger (1976) gave a comprehensivereview of !he literature concerning selective poisoning of active

sites to probe their catalytic reactivity.

Basic and acidic propemies of alumina surfaces have beenenhanced by treatment with alkali (Scokart et al., 1981) orhalogen ions (Peri, 1966, 1968).

Much work remains to be done before a comprehensive

understanding of tbe relationships between composition,electronic properties, geometric configuration, and chemicalreactivity of the shon-range ordered surfaces of transitionalumina can be developed. The advent of surface analyticaltools capable of probing structures and interactions at the

atomic level has greatly accelerated progress towards this goal.

?

63

Alcoa Laboratories

5. Surface Oxides on Aluminum Metal

5.1 Oxides Formed by Sofid-Gas Reactions

Aluminum owes its usefulness as a commercial metal to theproperties of the surface films which it forms in contact withthe atmosphere. In an ambient environment, elemental

aluminum is thermodynamically unstable with respect to itsoxides and hydroxides. The reaction

2AI + 1.502- A1203 – 1675,7 K]

is highly exothemic. Aluminum also readily reduces water, asthe standard potential of the electrode reaction

~ A13+ + e- ~~ Al”, EO= 1.706 V

is considerably negative,

Surface films on aluminum can form by one or more of thefollowing reactions:

(1) 2 AI+ 1.502 -A1203

(2) 2 Al + 3 H20 + AI,03 + 3 Hj

(3) Al + 2 H20~ AIOOH + 1,5 H2

(4) Al + 3 H20- AI(OH), + 1.5 Hz

Secondary reactions of the surface film with water arecontrolled by pH20 and temperature:

(4) A1203 + H20 ~ 2 AIOOH

(5) AlzO, + 3 H20 -2 Al(OH)j

(6) AIOOH + 2 H20 ~ Al(OH)J

Oxide layers developing near room temperature in oxygen ordry air reach a limiting thickness of 2-4 nm after several hours,

They are generally considered to be non-crystalline, butshon-range, cubic ordered stmcture has ken observed by Pryor

( 1971). At an estimated density of -3 g/cm3 of the short-rangeordered oxide, the molar volume increases by a factor of -1,7

in reaction ( I). Because of this strongly positive volumechange, the surface films are continuous even at oxidethicknesses only 2.3 “m, thus Iimiti”g mass transfer to

solid-state (lattice) diffusion, The initial stages of oxidationhave received considerable attention in recent years; in part dueto the availability of ultra-high vacuum instrumentation

providing tbe means for creating tmly clean aluminum sutiacesand precise dosing of gases, Much of tbe renewed interest is

64

also related to the use of aluminum and aluminum oxide in

microelectronic devices.

There appears to be general agreement that oxygen adsorbs

dissociatively on aluminum. Batra and Klein man ( 1984), in areview of the literature on oxygen interactions with aluminumsurfaces, pointed out that the breaking of the O-O bond in 02

to form two Al-O bonds results in an energy gain of about5 eV or 482 k.flmole. Simultaneous occupation of surface and

subsurface sites by oxygen was shown to be the first step of theinteraction of 01 with Al( I I I ) planes (Flodstrom et al. 1978;Erskine and Strong, 1982; Crowell, Chen, and Yates, 1986).

According to the latter authors, A1203 begins to form when acritical number of oxygen atoms cluster together after

prolonged exposure to 01 or when temperature is increased.

Similw mechanisms have hen reported for Al( 100) andAI(1 10) crystal faces (Batra and Kleinman, l.c., Cocke,

Johnson and Merrill, 1984),

After a continuous layer of A1203 has been established, further

growth follows exponential rate laws. According to the theoryof Cabrera and Mott (1948/49), thin films grow at low

temperature predominantly by cation migration, therefore,inverse logarithmic rate laws are observed. Electron transporttakes place by tunneling; this limits the film thickness to lessthan 5 nm in the temperature range in which thenno-ionic

emission of electrons dms not contribute significantly to chargetransport. Kirk and Hu&r (1968) reported that film grow[h and

change of work function are pressure dependent and follow alogarithmic rate law. Fehln and Mott ( 197 I) modified the

original Cabrera-Mott theory by assuming cation and anionmigration, the driving force being a space charge resulting froma cation excess and anion deficit near the metalloxide boundary

and the reverse nem the oxidelair interface, This growthmechanism is consistent with a logarithmic time function. Adefect stmcture near the metalloxide interface consisting of

excess metal combined with oxygen vacancies which trap twoelectrons each was propsed by Greenberg and Wright i“ 1955.Heine and Sperry ( 1965) found a metal-rich region extending

more than IO- 15 nm into the oxide (see also Pryor, 1971 ).

Lawless in 1974 critical] y reviewed much of the work on thin

oxide films. He pointed out that discrepancies regarding resultsand interpretation of mechanisms are mainly attributable to the

experimental difficulties of precisely determining mass andthickness changes of very thin oxide layers.

With increasing temperature up to about 650K the amorphous

nature of the A1203 films remains unchanged. Rate laws werereported to be paralinear; i.e., changing to linear after an initial

parabolic function at temperatures exceeding 700K, as shownschematically in Figure 5.1. The rate of mass gain changesback to an exponential function after several hours at these

—


.“ w

i

Non - linear (expon.)

650- 700K

DAl metal

~ ~::~ho.a

~, - A1203

Growth and Transport Mechanisms of Thermal Oxides, SchematicFignre S.1

temperatures (Cochran and Sleppy, 1961; Gulbransen andWysong, 1947).

The transition from a parabolic (logarithmic?) to a linear time

function coincides with the occurrence of crystalline transitionalumina of the gamma or eta form. This was first observed bydeBrouck&re (1945). Doherty and Davis (1963) showed that

cVstaIline transition alumina nucleates at the metalloxideinterface. Nucleation is very likely epitactic, as the difference

between the lattice parameters of Al (2 x 0,404nm = 0.808 nm) and spinel-type transition forms (~ ofeta = 0.79 nm) is VeIY small, Recent high-resolution electronmicroscopy bas confirmed such oriented growth (Ackland et

al., 1986).

Dignam et al., ( 1966) assumed that the crystalline aluminagrows laterally and vertically by continuous oxidation of metal,

and not by recrystallization of tbe amo~hous film. Oxygen,therefore, must be supplied from the atmosphere. Consideringthe slow rate of lattice diffusion in a refractoy oxide such asA1203 at temperatures below 900K, the linear growth rate isconsistent with a phase boundary (zero-order) reaction betweenoxygen and metal. This reaction mechanism requires open

pathways ktween atmosphere and metal. Flaws in theamorphous layer, wbicb occur simultaneously with the

crystallization of transition alumina, most likely provide suchchannels. Whether these flaws ind~cate a partialrecrystallization and formation of interfaces within tbe

amorphous oxide (Wefers, 1981), or whether they are causedby a different mechanism, is still being debated, Observationsby Bianconi et al. ( 1979) support tbe recrystallization

mechanism, This is also clemly seen in tbe micrographs ofFigure 5.2 A and B.

Continued lateral growth of transition alumina crystalliteeventually results in a continuous layer of thick crystallineoxide which limits oxygen transport to grain boundary diffusion

or lattice diffusion. Both processes follow no”.linear rate laws.

me interface ktween the afumi””m a“d the crystalline as WeIIas the amorphous oxide layers is sharp. Tlmsit et al. ( 1985)

showed that Al( 111 ) and (1 10) oriented i“tefiace~ bet~ee”metal and non-crystalline oxide are “rough” on an atomic scaleonly, with step heights not exceeding two interplan~ spacings.Neither Timsit et al. (l.c.) nor Ackland et al. (I,c. ) found

evidence of a transition zone of cubic suboxide

65

Alcoa Laboratories

Nucleation of Gamma Transforming Amorphous Oxide

J.J. Ptaslenskl/K. Wefem 1Oojmx

Progressive R&rystalliz.ation of Amorphous Surface Oxideon Afuminum

Fiwre 5.2

(a. = 0.406 – 0.430 nm) which has bee. reported hy

Yamaguchi in 1975.

The effect of water vapor on the oxidation of pure aluminum

has not been clearly established. At temperature exceeding650K, rates and mechanisms ap~ar to he very simila for dry

and “moist” (PH2 <100 mbar) atmospheres, as shown by

Cocbran and Sleppy (l.c.), Smeltzer (1956), Blackbum andGulbransen (1960), and Breaks~are (1970), ‘l’be reaction

products w AI*03 and H2, according to reaction (2). Evidencefor the formation of hydroxides has not been reported.

Little information is available regarding tbe composition andgrowth rates of surface films forming in moist atmospheres at

temperatures below 450”K. One may speculate that theinjection of (OH)– into the near-surface region of negative

space charge requires a lower activation energy than thetransfer of the 02- ion. Reduction of H+ may drive thecathmiic reactioq i.e., the extraction of electrons at grain

boundaries or defect sites. In careful studies of the compositionof air-formed films (<10 nm thick), Miller (1986) found a ratio

of oxygen to alutinum of 1.5 near the metalloxide interface,but a 2 to 1 ratio near the oxide/air bound~, suggesting acomposition approaching A1OOH in this region. Much morework is needed to close the gap kt ween studies of first-layer

interactions of clem Al surfaces under ultra-high vacuumcondhions, and the reasonably well known properties of thickfilms grown at high temperatures.

5.2 The Reaction of Aluminum Surfaces with Water

The reaction with water of aluminum and its alloys has been of

concern to corrosion chemists almost since the beginning ofcommercial use of the metal. It was shown in Chapter 2.12 thataluminum which has ken depassivated by slight amalgamationof its surface vigorously reacts with cold water or moist air

until all metal is consumed. Passive aluminum; i.e., metalprotected by a continuous oxide film, is on the other band

practically stable even in boiling water. Reaction with water orsteam at temperatures below 400K under certain conditionsreinforces the protective properties of the surface oxide. Thisphenomenon has been known and commercially utilized for

decades. The literature on this subject was summarized byAltenpohl in 1962.

A number of researchers have studied growth mechanisms mdproperties of hydroxide films formed by reaction of aluminumsurfaces with water. An extensive review of the

aluminum-water system was published by Alwitt in 1976.

Several workers, Hart (1957), Bernard and Randall (1961 ), andVedder and Verrnilyea (1969) observed that, in the temperature

range between 293K and 373K, increase in mass of the surfacefilms follows exponential time laws after an induction period,

,f

,

66


?

the duration of which is inversely proportional to temperature.While Vedder and Vermilyea proposed a pwabolic rate,

indicating a diffusion controlled reaction, Alwitt (l. c.) pointedout that the exponent of the rate equation varies withtem~rature.

Mass change data are not unambiguous indicators of growthrates, because neither composition nor mo~hology of sutiace

films on aluminum ~e uniform. Hart (1957) was one of thefirst researchers to find that the films formed between 293Kand 353K are layered an amorphous layer near the metal

intetiace is followed by a zone consisting of gelatinousboehmite; bayerhe or well crystallized boehmite grow on theouter surface (Figure 5.3). The ratio of the three strata changeswith temperature, pH, and time of exposure. when formed at

room temperature, films may be made up of only two opticallydistinguishable layers, both of them consisting of X-rayindifferent f’amorphous”) material (We fers, 1961),

Bayerite occurs during exposures of up to 350-360K at ambientpressure. Well crystallized boehmite develops in the outer

region of the films at higher temperatures. Dachille et al.( 1971) showed that bayerite may persist up to 425K if thepressure is raised above 675 bars. Although A1OOH is the

stable phase under these temperaturelpressure conditions (seefigure 3. I), the phase reaction

Al(OH)g - AIOOH + H20

which results in an increase in moleculz volume, may be

kinetically hindered at higher pressures, especially if nuclei of

Y-AIOOH are not present. At temperatures exceeding425-450K, the surface hydroxides recrystallize and form adiscontinuous layer of individual boehmite crystals which no

longer protects the metal (Altenpohl, 1962; Wefers, 1961).Diaspore, a-AIOOH, corundum a-A120q, or transition aluminasare tbe reaction products at higher temperatures and pressures

(Alwitt, 1976).

It is obvious from these observations that the interaction ofaluminum surfaces with water produces the same phases andfollows reaction paths similar to those observed withhydroxides precipitated from homogeneous solution.

According to the phase diagram A1203-H20 (Figure 3.1), thethin, air-formed surface oxide on aluminum is unstable attemperatures lower than 635K in the presence of water. Below-360K, a porous layer of gelatinous hydroxide is fomed by

the reaction of tbe oxide according to Quation 4 inChapter 5.1. The concomitant thinning of the continuous oxide

barrier initiates diffusion of aluminum ions which becomehydrated as they enter the porous gel.

Due to the ve~ low volubility of aluminum in pure water, thesolvated Al-ions precipitate, most likely via the

D. R. Mlcholas Io,ooox A

Cryst. AIOOHI AI (OH,,\ , Al (OH); SOI.

A Al Immersed in Watar, 1 day st 335KE Film Texture, Schematic

Multilayered Hydroxide Film on AluminumFigure 5.3

deprotonationlcondensatoin process depicted in Figure 2.5, thus

increasing the thickness of the gelatinous layer, At the

gel/liquid interface, crystalline hydroxide is precipitated by asolutionlredeposition reaction. Temperature and pressure

determine which phase is formed. Crystals of boehmite areshown in Figure 5.4 which represents aluminum treated inboiling water for 8 hours.

The layered texture of the surface films reflects thepredominant transport mechanisms and relative activities of

67

Alcoa Laboratories

J. J. Ptssiensld 8,000X

Duplex Film on Al Formed in Boiling WaterF,g”re 5.4

aluminum and water both the oxide adjacent to the metal andthe topochemically formed gelatinous hydroxide develop as aresult of the diffusion of aluminum ions into and through a

continuous barrier. The development of the outer, crystallinelayer is a function of volubility; i.e., pH and temperature. Theratio of aluminum to oxygen changes accordingly: it is tbehighest, 0.66, in the A1203 region near the metal surface;decreasing to around 0.5 in gelatinous hoehmite(AIOOH + aq, ) and 0,33 in AI(OH)3, bayerite.

To obtain an optimally protective coating, it is obviouslyadvantageous to promote the growth of tbe two inner, diffusioncontrolled barriers and to suppress the recrystallization reactionwhich transforms tbe continuous, gelatinous hydroxide,

Treatment of aluminum surfaces for shon times (<1 h) withunsaturated steam at 380-400K meets these requirements.

5.3 Oxidss Formed by Anodic Polarization of Aluminum

5.31 Non-Porous Oxide Films

A widely used method of forming surface oxides on aluminumis the ancdic polarization of tbe metal in liquid, aqueous andnon-aqueous electrolytes, or in plasma environments containing

oxygen-beting sWcies. Surface films of varying thickness,composition, and texture can k produced by selecting

appropriate electrical and chemical conditions. Detailed reviewsof tbe electrochemistry of anodic oxidation as well as of thestmcture and composition of anodic films were given byTajima (1970) and Diggle, Downie, a“d Gouldi”g (1969). The

practice of commercial “anodizing” has been summarized in a

number of biwks; e.g., Wemick and Pinner (1972), Henley(1982).

Countless andlzing systems and procedures; i.e., combinationsof electrolyte chemistry, temperature, voltage, and current,

have kn descrikd during more than 90 years of research anddevelopment in this field. The oxide films produced,

nevertheless, can & grou~d - with some simplification - intoonly two types: non-porous “barrier” oxides and porous anodlcfilms. Barrier oxides form in electrolytes in which aluminum

oxide is not significantly soluble, while prous coatingsdevelop in electrolytes which dissolve A1203. This

differentiation, again, is somewhat simplistic but will behelpful in the context of this short summary of the complex

subject.

Figure 5.5A shows the voltage/cument relation of a cell inwhich an aluminum electrcde is anodically polarized; e.g., in a

neutral solution of scdium borate and boric acid. After thevoltage has been applied, the current surges briefly before

decaying exponentially to a ve~ low value. This processrepeats itself each time the voltage is raised to a higher level.A current-limiting passivation layer is formed, the thickness of

which is a linear function of the anodic electrode potential. Inthe absence of chemical dissolution of the oxide, the filmthickness corresponds to 1.4 nm ~r volt of anodic potential, or

a minimum field strength of 7.1 X 106 V/cm.

GA 202,632

I I I I

I

——

Time 101 1Oz 103 104 see

A: highly, E moderately, C: non-solvent electrolyte

Voltage-Current Relationships in Anodizing AlFiwre S.5

6s

Oxides and Hydroxides of Aluminum1’

;

‘,

1,

Current efficiency is close to 10Q% of the theoretical Fxadayequivalent under these conditions. Ultimate thickness of thebarrier oxides is limited by their dielectric strength. Breakdownby sparking occurs at cell potentials above 600.7W V,corresponding to a film thickness of -0,85-1 pm.

The current flowing across the harrier is ionic at tield strengthshigher than the minimum value. Below this limit, tbe small,residual current is electronic and probably related to local

impurities. Both aluminum and oxygen ions are mobile undertbe influence of tbe high field, contributing equally to film

growth at the metalloxide and the metallelectrolyte interface.This was found by Davies et al, (1965) and Xu et al. ( 1985)who used radioactive (Xe12s) marker techniques to study tbe

relative movement of cations and anions during film formation.A transport numkr of -0,5 for A13+ was detemined in bothinvestigations.

Major contributions to the understanding of profxrties andgrowth mechanisms of anodic oxides were made by

A. Giintberscbultze, D, A, Vermilyea, M, J, Dignam, W. J.Bernard, N. F. Mott, G. C. Wocd, and their work groups.

me barrier films consist of almost pure A1203; they containgenerally less than 1% electrolyte anion and water. Oxidesformed at less than 2Ctl-3W V are X-ray indifferent; eta andgamma transition aluminas have been found in film,. p,uduced. .

at higher potentials, Being optically featureless even at

dimensions of a few nanometers, barrier films can be used toreplicate aluminum surfaces for transmission electronmicroscopic examination.

The main application for barrier oxides is in electrolyticcapacitors. Although the dielectric constant of the oxide filmsis only moderate (K = 8- 10), high capacitance Fr unit area is

obtained by etching the aluminum substrate before anodicoxidation. A special electrolytic etching procedure leaves deep,

crystallographically oriented “tunnels” which are tilled with theanodic oxide, as shown in Figure 5.6.

Plasma anodization of aluminum is employed in themanufacture of microelectronic devices to form insulating filmsin-situ. tittle is known about growth mechanisms andstructures of these films. A review of this technique was given

by Hyde and Yep in 1976.

5.32 Porous Anodic Oxide Coatings

5.321 Properties

When aluminum is anodically ~l&zed in electrolytes which

dissolve aluminum oxide, the current-limiting barrier thicknessof 1.4 nmlV generally cannot be obtained. Ionic current,therefore, continues to flow. In the extreme case, the rates at

W. T. EvanS 3,000X

“Tunnel” Etched Al with Barrier OxideFigure 5.6

which oxide is formed and dissolved are equal, Suchconditions, under which the anodic current efficiency is zero,

Xe employed in “electropolishing”, a form of controlledelectrolytic etching of aluminum and its alloys. Numerouscombinations of electrolyte chemistry, cument, voltage, and

temperature have been reprted in the literature which produceandlc films having properties that place them between thenon-porous barrier layer and the thin (<5 nm) oxide remainingafter electropolishing.

Oxide films having a very distinctive texture develop inelectrolytes containing sulfuric, oxalic, and phosphoric acid,Two examples of the anodic polarization behavior of aluminum

in these electrolytes are given in Figure 5.5: curve B shows thecurrent vs. time transient of an aluminum anode polarized at

constant potential in 370 oxalic acid, a moderately solventelectrolyte. Curve C represents the current vs. time relationobserved when 1590 sulfuric acid, a strong solvent, it used,

Both curves digress from the exponential decay of currentwhich occurs after the formation of a continuous barrier layerof 1,4 rim/V (curve A). Having passed through a minimum,current recovers to a constant level (C) or again slowly declineswith time (B). (If the anodlc current is held constant, voltage

rises in case B).

69

Alcoa Laboratories

In the first few seconds of anodic polarization, a barrier layergrows as it does in a non-solvent electrolyte (A). Recovery ofionic current is the result of a stmctural and morphologicalchange of the initial btier. bcalized reaction sites develop in

which a thinner (<1.4 rim/V) oxide advances into the metalsubstrate (Figure 5.7). First shaped almost hemispherically, thepenetrating oxide “cells” elongate into hollow cylinders closed

by a rounded bottom, Being thinner and, therefore, absorbingfewer electrons, the cell bottoms appear light in the electronmicrograph. The annular cell walls, which are oriented parallel

to the electron beam, appe~ dark due to the longer path lengthand higher absorption of the electrons. Ap~aring first in areasof higher free energy such as subgrain boundaries, the cellsrapidly cover the electrode surface, attaining a polygonal,ideally hexagonal, cross section as they fill the available space.

Once the cell pattern has ken established, it remainsunchanged while the oxide continues to grow (at constant

electrcde ptential). The length of the cells; i.e., total filmthickness, may reach more than 100 pm. Their diameter is afunction of voltage and electrolyte composition. It ranges from2.6 to 3.0 nmlV. Pores formed in moderately solvent

electrolytes measure less than one-third of the cell diameteqlarger pores of more than 1 rim/V develop in aggressivesolutions.

The mo~hology of a thick, cellular oxide film is schematically

represented in Figure 5.8. Setoh and Miyata (193 I ) andRummel (1936) fust recognized the porous nature of thickanodic oxides. These workers ~stulated a duplex structure ofanodic coatings on the basis of measured electrical proprties: ahigh-im~dance layer of less than 100 nm thickness adjacent to

the metal surface, covered by a second layer the (low)impedance of which is nearly independent of thickness. Directevidence of the porous texture was provided by electronmicroscopic investigations of Fischer and Kurz in 1942. Keller,Hunter, and Robinson published the now classical Alcoa modelin 1953. Although they microscopically analyzed mostly

replicas of the metalloxide and oxidclair interfaces, and verythin oxide films sepaated from the metal (as did Fischer and

Kurz), their concept agrees in its essential features with themodel of Figure 5.8, Electron micrographs of cross sectionslater published by Booker, Wood, and Walsh (1955), Akahori

(1961), and Ginskrg and Wefers (1962, 1963) contimed andadded minor details to the Alcoa model (Figure 5.9).

The concept of the duplex structure only describes the electricalpro~rties of the celhdar oxides. There is no structural orcompositional separation between the barrier and the porous

pan of the cellular oxides. Any distinction is a functional one.The rounded bottom of the cells constitutes the

G. A. Nitowskl 25.000X

Initial Cell DevelopmentFigure5.7

electrochemicafly active “barrier” while anodic oxidation isproceeding. Aluminum and anions are transported through thisbarrier along the lines of force of a cone-shaped field. As thecell advances into the substrate by oxidizing metal, the apex of

the field also moves downward and the upper portions of thecell bttom become cell wall. Consequently, each cell is anindividual entity which grows independently of others.

Advancing single cells and their circula cross sections areshown in Figure 5.10. The non-uniform progression of cellshad been induced by a sudden increase in current.

Oxid’es and Hydroxides of Aluminum

Pore

i

Cell wall

(

\Barrier layer(cell bottom)

t

Sealed cells

rexture of Porous AndIc OfideFigure 5.8

Whereas the diameter of the cells is a function of the anodic

voltage, both pore volume and chemical composition of theanodic films are determined by tbe electrolyte. As early as1899, Norden reported that anodic films formed on aluminum

in dilute sulfuric acid contained 13% S03 and almost 15%H20. Infrued analysis indicates sulfate ion to be thepredominant species. The concentration varies; it is determined

by current density, concentration, and temperature of theelectrolyte. Anodic coatings produced in oxalic acid containbetween two and five percent oxalate iow similar amounts ofphosphate are found if phosphoric acid is used as the

electrolyte. (Films formed in chromic acid have a poorlydeveloped cell texture; they are repotied to be practically freeof chromate ion; Tajima, 1970. )

The Pre volume of anodic films is controlled by the solventpwer of the electrolyte. Values near 209. were measured withcoatings produced in 15% H2S04 at 300K (Mason, 1957).

Porosity may be as low as 5% in films formed in dilute (<2%)sulfuric acid at 280K. Specific surface area is a function of the~rosity. For an ideal oxide of 10 pm thickness, having celldiameter of 30 nm and a ~re diameter of 10 nm, a surface

area of abut 45 m2 per mz of geometric surface (1 mz X 10pm = 1 cm3) can be calculated. At a density of nearly 3 glcm],this value comespnds to approximately 15 m2/g.

Porosity is an indirect measure of the current efficiency of the

anodic process. The larger the pre size at a given celldiameter, the lower the current efficiency. It is obvious from

Alcoa Laboratories

10,000 x

Cross Section and Surfaceof Porous Anodlc Oxide

Rpre 5.9A

the mcdel of Figure 5.8 that porosity or current efficiency can

be expressed as cell wall thickness in nanometers pr volt (atthe bottom of the cell; in the upper portions, chemical attackduring immersion in the electrolyte causes some additional

thinning). Values ranging from 0,85 to to 1.25 rim/V have kenreported (Diggle et al., 1969; Tajima, 1970). As 1.4 rim/Vcorresponds to a current efficiency of 100Yo, only -6090 of thealuminum ions contribute to film formation at 0.85 nmlV.

A comprehensive understanding is still lacking of theelectrochemical and chemical reactions involved in thefomation of porous anodic oxides, especially the development

of the unique regular, cell texture, There appears to be general

agreement that a purely chemical dissolution of anodically

F. A. Mozelewsld 4,000 x

Surface of Anodic OxideFigure5.9B

formed oxide is not likely to be the cause of the uniform~rosity and the defined, voltage-dependent barrier layer

thickness.

Both anions and cations are moved into and across the ceil

bottom (banier) by the electrical field. The chemical

composition of the electrolyte, current density, and tem~rature

are the factors which determine the fraction of aluminum ions

moving into the electrolyte without forming oxide (“field

assisted dissolution”) and the amount of acid anion incorporated

into the film, It remains an own question if there is a causal

relationship ktween the loss of aluminum ions and the

incorporation of multivalent cations such as sulfur, phosphoms,

or carhn.

In the cell wall, the concentration of electrolyte anions is

highest near the pore and decreases towards the periphery

(Ginsberg and Wefers, 1963; Thompson et al., 198 I ). It has

ken suggested that a thin (<50-60 nm) “shell” of aluminum

oxide is formed by a solid-state reaction at the metal interface,


W.T. Evans

5.322 Application of Porous Anodic Films

Porous coatings are widely used to enhance the appearance andimprove the corrosion resistance of fabricated alumium parts.Building, automotive, and aircraft industries provide the largest

markets. Coatings can be colored by several methods. One isthe impregnation of the pores with dyes such as azo dyes oranthraqu inone derivatives (Tajima, 1970). Precipitation ofinorganic pigments in the ~res; e.g., iron hydroxides, is

another commercial process. Widely used for the coloration ofarchitectural panels is electrolytic deposition of metal. Bycathodic polarization of freshly anodized aluminum workplaces

in appropriate plating baths, Pres are partially filled with

. .

1lul,ooox

Fragments of Anodic OxideFigure 5.9C

and that aluminum ions moving across this layer precipitate

anion-rich, microcrystalline cell wall material by reaction withthe electrolyte (Thompson et al., 1978). Electron micrographsof cells produced at high voltage show, indeed, a distinctly

different morphology of the outer “shell” and the inner porewall material-(Figu~e 5 .9C, from Wefers and Wallace, i97@see also Thompson et al., 1978). This interesting conceptwould give the term “duplex film” a new definition. The

similarity to the dual-layer texture of hydrothermally formedsurface films may k more than coincidental. These, too,

consist of a zone of pure oxide produced by a solid-statereaction and an outer layer resulting from a solution-depositionprocess, as was shown in Chapter 5.2.

Many questions remain unanswered. These authors consider therelationship &tween the electrolyte composition and theunique, regular cell texture the most impoflant puzzle to be

solved.

.

K. Wefers 15,000X

Cells of Anodic Oxide Advancing into MetalFigure 5.10

73

Alcoa Laboratories

~,- ‘ -= -’-- -“””- JF. A. Mozelewskl 15,000X

Cross Section of Sealed Anodic OxideFigure 5,11

bismuth, copper, nickel, and other metafs. So-called integral

color prmesses use specific electrolytes and voltagelcumentcontrol to produce color centers (Frenkel-type defects) in theanodic oxide (Tajima, 1970; Wefers and Wallace, 1976).

Pigmentation of the anodic films can be achieved by the use ofafloys containing constituents which do not oxidize in thecoatings process.

The protection given to the metal by thick porous oxides isgreatly improved hy a hydrothermal treatment of the coatings in

boiling water or steam. In the “sealing” process, the pore wallmaterial reacts with water to form gelatinous boehmite and

amorphous oxide hydroxide:

Alx(anion)Y + H20 ~ AIOOH + acid

A1203 + H20 ~ 2 AIOOH

The reaction proceeds Wrpendiculm to the pre walls,

eventually tilling the pore with aluminum oxide hydroxide(Hoar and Wood, 1962; Wefers, 1973). The morphology of

sealed oxides is shown schematically y in Figure 5.8 and in theelectron micrographs of Figure 5.1 I.

Interesting applications for anodic oxides were proposed byGrubitsch et al. (1961), Honicke ( 1983), and Rai andRuckenstein (1975). Gmbitsch et al. suggested the use ofporous oxides as microporous filters for the separation of

isotopes, vims, or proteins. Honicke, and Rai and Ruckensteinused anodic oxides as catalysts and as substrates forcatalytically active metals. Dehydration of alcohols and

isomerization of alkenes are examples of reactions which xecatalyzed byanodic oxides (see also Cockeet al. 1984).

There are many more commercial applications for prousanodic coatings. Their discussion would exceed the scope ofthis monograph.

74

—


Bibliography

Aboaf, J. A., J. Electrochem. Soc., 114(9) (1967) 948-952

Abrams, L, and bw, M. J. D,, Ind. Eng. Chem. Prod. R&D, ~ (1969) 38-39

Acbenbach, H., Cbem. Erde, 6 (193 I) 307-356

Ackland, D. A., Dabmen, V,, Ether, C. J. et al., J. of Metals (1986) 19-24

Adelskold, V., Ark, Kern. Mineralog. Geol. Ser. A., ~ (1938)

Adkins, A., J. Amer. Chem. Sot., Q (1922) 2175

Adler, H. H., Infrared S~ctra of Reference Clay Minerals, Section I, Amer. Petrol. Inst. Proj. 49, Columbia Univ. ( 1950)

Akahori, H., J. Electron Microscopy, Q (1961) 175-179

Aldebert, P, and Traverse, J. P., J. Amer. Ceram. Sot., 65(9) (1982) 460-463

Altenpohl, D. G,, Corrosion, ~ (1962) 143-153

Alwitt, R. S., Oxides and Oxide Films, Vol. 4 (M. Dekker, New York/Basel) (1976) 169-254

Anderson, G. M. and Bumham, C. W., Amer. J. Sci., ~ (1967) 12-27

Baker, B. R. and Pemson, R. M., J. Catal., ~ (1974) 265-278

Bale, H. D. and Schmidt. P. W., J. Cbem. Phys., U (1959) 1612-1618

Barker, M. G., Gadel, P. G., and Begley, M. J., J. Chem. Sot. Dalton Trans. (1984) 1139-1146

Barlett, H. B., J. Amer. Ceram. Sot., ~ (1932) 361-3@

Bamhisel, R. 1. and Rich, C. I., Proc. Soil Sci. Sot. Amer., ~ (1965) 531-534

Batra, I. P. and Kleinman, L., J. Electr. Spectr., ~ (1984) 175-241

Bauer, A. A. and Bates, J. L., Battelle Mere. Inst. Rept. 1930 (1974)

Beck, A. F., Heine, M. A., Caule, E. J., and Pryor, M. J., Corr. Sci., ~ (1967) 1-22

Becker, K. H., Cemic, L., and Langer, K. E. O. E., Geochim. Cosm~him. Acts, fl (1983) 1573-1578

Beckmann, J., A History of Inventions, Discoveries, and Origins, translated from the German by W. Johnston, H. G. Bobm,

publ., London (1846)

Beevers, C. A. and Brobult, S., Z. Kristallogr., ~ (1936) 472-475

Beevers, C. A. and Ross, M. A. S., Z. Kristallogr., ~1937) 59-66

Bentor, Y. K., Gross, H., and Heller, L., Amer. Mineral., ~ (1963) 924-930

Berecz, E. and Szita, L., Electrocbim. Acts, 15 (1970) 1402-1407

75

Alcoa Laboratories

Bibliography (Cent)

Bernard, W. J. and Randall, J. J., J. Electrochem. SW., 108 (1961) 822-825—

Berthier, P., Ann. Mines, ~ (1821) 531-534

Bhagavantam, S. and Venkatarayedu, T., Proc. Indian Acad. Sci., ~ (1939) 224-258

Bianconi, A., Bachrach, R. Z., Hagstrom, S. B. M., and Flodstrom, S. A., Phys. Rev. B, Q (1979) 2837-2843

Bianconi, A., Bachrach, R. Z., Hagstrom, S. B. M., and Nodstrom, S. A., Phys. Rev. B ~ (1979) 2847-2850

Bohm, J., and Niclassen, H., Z. Anorg. Allg. Chem., ~ (1924) 1-9

Bohm, J., Z. Anorg. Allg. Chem., ~ (1925) 203-216

Boiko, B. B. and Soika, A. E., Dokl. Akad. Nauk BSSR, 28(11) (1984) 984-987

Boilot, J. P., Thery, J,, and Collogues, R., Mater. Res. Bull,, ~ (1973) 1143-I 145

Bonshtedt-Kuplelskaya, E. M., C. R. Acad. Sci., URSS, Q (1945) 587-589

Booker, C. J. L., Wwd, T. L., and Walsh, A., Nature, 176 (1955) 222-223—

Bosmans, H. J., Acts C~stallogr, Sect, B, 26 (Pt. 5) ( 1970) 649-652

Bragg, W, H, and Bragg, W. L., X-Rays and Crystal Stmcture, Bell & Sons, bndon (1916) 169

Bragg. W. L., Gottfried, C., and West, J,, Z, Kristallo8r., 77(3-4) (1931) 255-271

Braun, P, B., Nature (hndon), 170 (1952) 1123—

Breakspeare, R, J., J, Appt. Chem., g (1970) 208.212

Brewer, L. and Searcy, A. W,, J, Amer. Chem. Sot., 73(11) (1951) 5408-5414

Bridgman, P, W,, Proc. Amer. Acad. AI-IS Sci., ~ (1949) 187.234

Brindley, G. W, and Chin, J. O., Amer. Mineral., ~ (1961) 771-785

Brindley, G. W., Amer. Mineral., ~ (1961) 1187-1190

Britt, E. and Davis, M. V., IEEE Thermoio”ic Conversion Specialists Conf., San Diego, Calif., October W-07 ( 1971) 137-146

Bmnham, C. W., Ryzhenko, B. N., and Schitl, D., Geochem. [ntem., Q (1973) 1374-1375

Bugosh, J., U.S. Patent 2,915,475 (1959)

Bugosh, R. L., Brown, R. L, McWorther, J. R., Sem, G, W., and Sippel, R, J., Ind. Eng. Chem., ~ (1962) 157-161

Busing, W, R. and Levy, H. A., Acts Crystl., I I (1958) 798-803

Bye, G. C. and Robinson, J. G., J. Appl. Chem, Biotechnol., ~ (1974) 633-637

76


I

Bibliography (Cent)

Cabannes-Ott, C., C. R. Acad. Sci., 244 ( 1957) 2491-2495—

Cabrera, N. and Mott, N. F., Rep, Progr. Phys., ~ (1948) 163-184

Caglioti, V. and D’Agostino, O., CU.. Cbim,, Ital., @ (1936) 543-548

Cahoon, H. P. and Christensen, C, J., J. Amer. Ceram, Soc,, ~ (1956) 337-344

Calvet, E., Boivinet, P., N&l, M., Thibon, H,, Maillard, A., and Tertian, R., Bull. Sot, Chim, Fr. (1953) 99-108

Campbell, W. B., 4th Qumerly Rep. AMRA CR 63-0314 (Lexington Laboratories) (1963)

Cannon, R. M,, Rodes, W. H,, and Heuer, A. H., J. Amer. Ceram. SIX., ~ (1980) 46-53

Cannon, W. R. and Coble, R. L., Deformation of Ceramic MateriaJs, Edited by R. C, Brat and R, E. Tressler, Plenum Press,New York (1975) 61-100

Castor, C. R., Gems, Synthetic, Kirk-Othmer Encyclo~dia of Chemical Technology, 2nd Ed,. Vol. 10, Interscience PublishersInc., New York (1966)

Chang, C. C., J. Appl. Phys., ~ (1968) 5570

Chang, C. C., “Proc. Conf. Structure and Chemistry of Solid Surfaces,” ~ited by G. A. Somorjai, John Wiley and Sons, NewYork ( 1969) 77

Chang, R., J. Appl. Phys., 31(3) (1960) 484-487

Chao, G. Y., Baker, J., Sabina, P, S,, and Roberts, A. C., Canada. Mineralogy,, ~ (1985) 21-28

Cbarig, J. M., Appl. Phys. htt., ~ (1967) 139

Charvat, F. R. and Kingery, W. D., J. Amer. Ceram. SW., @ (1957) 306-315

Cherdon, E., Hellinghausen, H., Kramer, A., Pohland, H., Metzer, A., U.S. Patent 4,308,088 ( 1981)

Christensen, A. N., Lehmann, M. S., and Wright, A., Acts Chem. Stand., A36 (1982) 779-781

Cbristoph, G. G., Corbato, C. E., Hofmann, D. A., and Tettenhorst, R. T., Clays and Clay Minerals, 27(2) (1979) 81-86

Cistjakova, A. D., Tsvet. Metally, ~ (1964) 54-60

Clark, D. E. and Lanutti, J. L., Ultrasuuct. ProcessUbicb, W]ley, New York (1983) 126-141

Coble, R. L. and Kingery, W. D., J. Amer. Ceram.

Coble, R. L., J. Appl. Phys., ~ (1961) 793-797

Cochmn, C. N. and Sleppy, W. C., J. Electrochem,

Ceram., Glasses Compos, (Proc, Int. Conf .), Edited by L. L. Hench, D. R

Sot. , ~ (1956) 377-385

SW., 108 (1961) 322-327—

Cocke, D. L., Johnson, D. E., and Merrill, R. P., Catal. Rev. - Sci. Eng., ~ (1984) 163-231

77

Alcoa Laboratories

Bibliography (Cent)

Cocks, F. H. and Stormont, R. W., J. Electrochem Sot., 121 (1974) 596-602—

Collongu~s, R., Th~ry, J., and Boilot, J. P., Solid Electrolytes, Edited by P. Hagenmuller and W. VanCool, Academic, NewYork (1978) 253-276

Collongu~s, R., Courier, D., Kahn, A., Boilot, J. P., Cojomban, Ph., and Wicker, A., J. Phys. Chem. Solids, 45(10) (1984)98 I-1OI3

Conrad, H., J. Amer. Ceram. Sot., 48(4) (1965) 195-201

Cook, W. B. and Perkowitz, S., Appi. Opt., 24(12) (1985) 1773-1775

Corbato, C. E., Tettenhorst, R. T., and Christoph, G. G., Clays and Clay Minerals, ~ (1985) 7 I-75

Cornelius, E. B., Milliken, T. H., Mills, G. A., and Oblad, A. G., J. Chem. Phys., ~ (1955) 1O-I4; 809-813

Cox, D. E., Moodenbaugh, A. R., Sleight, A. W., and Chen, H. Y., Brookhaven National Lab. Tech. Rep. No. 26502 (1979)

Crandall, W. G., Chung, D. H., and Gray, T. J., Mechanical Properties of Engineering Ceramics, Edited by W. W. Kriegal andH. Palmour, Interscience Publisher Inc., New York (1961) 349-379

CRC Handbwk of Material Science, Volume 11, CRC Press Inc., Cleveland (1975) 355-363

Cross, C. F., J. Chem. Sot., ~ (1879) 796

Crowell, J. E.. Chen, J. G,, and Yates, J. T., Surf. Science, 165 (1986)

Czandema, A., Methods of Surface Analysis, Elsevier, New York ( 1975)

Dachille, F., White, E, W, and Gigl, P, D. Tech. Rep. No. 7, Office of Naval Research, ONR 032-502 (1971)

Dachselt, E, and Patz, K., Z. Anorg. Allg. Chem., 292 ( 1957) 262-274

Damerell, V. R., Hovorka, F., and White, W. E,, J, Phys. Chem,, ~ (1932) 1255-1267

Dana, J. D., Systems of Mineralogy 1, 7th Ed. Revised and Edited by Palache, C., Berman, H., and Frondel, C., New York,Wiley ( 19ti)

Davies, J. A., Domeij, B., Pringle, P, S,, and Browa, F., J. Electrochem. Sot., 112(1965)675-679

Davy, H,, Phil. Trans., ~(1805) 155; 1621

DeBoer, J. H, and Houben, G. M. M., Proc. [“t. Symp. Reactiv, Solids (Gothe”burg) Part I (1952)237-244

DeBoer, J. H., Fortuin, J. M. H., a“d Steggerda, J, J., Proc. Ko”. Ned. Akad. Wete”sch., 57(1954)434-443—

DeBoer, J. H., Steggerda, J, J., a“d Zwietering, P., PrcIc. Ko”. Ned. Akad. Wetensch,, 59(1956)435-444—

DeFlandre, M., Bull. Sot. Fr. Mineral., 55(1932) 140.165

DeLapparent, J., Compt, rend., 184(1927) 1661-1662

i,,

1

i.

‘i((

/

1!.

i

78


Bibliography (Cent)

DeMorveau, L. B., in Encyclopedia Methodique Paris; D:ctionaire de Physique, Atiicle “Alun” (1786)

DeVries, R. C. and Roth, W. L., J. Amer. Ceram. Sot., ~ (1969) 364-369

Dewey, C., Amer. J. Sci., ~ (1820) 249

Diggle, J. W., Downie, T. C., and Goulding, C, W., Chem, Review (1969) 365-405

Dignam, M. J., Fawcett, W. R., and Bohni, H., J. Electrochem. Sot., 113 (1966) 656-662

Dignam, M. J. and Fawcett, W. R., J. Electrochem. SW., ~ (1966) 663-671

Ditmars, D. A. and Douglas, T. B., J. Res. Natl. Bur, Std., 75A (1971) 401-411

Doherty, P. E. and Davis, R. S., J. Appl. Phys., ~ (1963) 619-623

Dome, E. and Hiihner, H,, Alumina Processing, Pro~rties and Applications, Springer Verlag, New York ( 1984)

Dmzhinina, N. K., Tsvet. Metal] y, ~ (1955) 54-57

Eberhardt, W. and Kunz, C., Surf. Sci., ~ (1978) 709-714

Edwards, J. D., Frary, F. C., and Jeffries, Z., in “The Aluminum Industry,” McGraw Hill, New York (1930) l@-169

Elliott, A. G. and Huggins, R. A., J. Amer. Ceram. SW., ~ (1975) 497-500

Engel, T. K,, Nucl. Mater., ~ (1969) 211-214

Erskine, T. L. and Strong, R. L., Pbys. Rev. B, ~ (1982) 5247-5249

Ervin, G. and Osbom, E. F., J. Geol., ~ (1951) 381-394

Ervin, G., Acts Crystallogr., : (1952) 103-108

Ewing, F. J., J. Chem. Phys., ~ (1935) 203-207

Farbr, M, Srivastava, R. D., and Uy, O. M., J. Chem. Sac. Faraday Trans., 68(2) (1972) 249-258

Fanington, G. C. and Briant, J. L., Science, 204 (1979) 1371-1379

Fehler, F. P. and Mott, N. F., Oxid. of Metals, ~ (1970) 59-99

Feitknecht, W., Wittenbach, A., and Buser, W., 4tb Symp. Reactiv. Solids, Elsevier (1961) 243-249

Fiescher, H. and Kurz, F., Korrosion und Metallschutz, ~ (1942) 42-47

F,tzer, F. and Weisenburger, A., Advances in Thermal Conductivity, Edited by R. L. Reisberg, Roll. (U. S. A.) (1974) 42-46

Flockhti, B. D., Leith, 1. R., and Pink, R. C., Trans. Faraday SIX., @ (1966) 730-730; ibid, Q (1969) 542-551

Flodstrom, S. A., Martinson, C. W. B., Bachrach, R. Z., Hagstrom, S. B. M., and Bauer, R. S, Pbys. Rev. B, ~ (1978)6495-6498

79

Alcoa Laboratories j)

{

Bibliography (Cent)

Foster, L. M,, Long, G., and Hunter, M. S., J. Amer. Ceram. Sot., ~ (1956) 1-11

Foster, P. A., J. Electrochem. SIX., 106 (1959) 971-975— /(

Foster, P. A., Chem. Eng. Data, ~ ( 19@) 200-203

Fredrickson, L. D., Anal, Chem., 26 (1954) 1883-1885

French, T. M. and Somorjai, G. A., J. Phys. Chem., 74(12) (1970) 2489-2495

Freund, F., Ber. Deut. Keram. Ges., ~ (1965) 23-25

Freund, F., Ber,, Deut. Keram. Ges., 44 (1967) 51-58; 151-156.

Fricke, R., Z. Anorg. Allg, Chem., 175 ( 1928) 249-256—

Fricke, R,, Z, Anorg. Allg. Chem,, 179 (1929) 287-292—

Fricke, R. and Jucaitis, P., Z. Anorg. Chem., 191 (1930) 129-149—

Fricke., R, and Wiillhorst, B., Z. Anorg. Allg. Chem., ~ (1932) 127-144

Fricke, R. md Severin, H., Z. Anorg. Allg. Chem., ~ (1932) 287-308

Fricke, R. and Jockers, K., Z. Naturforsch,, 26 (1947) 244

Fricke, R. and Schmah, H., Z, Anorg. Chem., 255 (1948) 253-268—

Fricke, R. and Jockers, K., Z. Anorg. Chem., ~ (1950) 3

Fripiat, J, J., Trans. Faraday Sot., ~ (1969) 1236-1238

Fyfe, W. S, and Godwin, L. H,, Amer. J. Sci., 260 (1962) 289-293—

Fyfe, W, S. and Hollander, M. A,, Amer. J. Sci., 262 (1964) 709-712—

Geiling, S. and Glocker, R., Z. Elektrochem., ~ (1943) 269-273

GeOer, S. J., Chem, Phys., ~ (1960) 676-677

Ghorbel, A., Hoang-Van C., and Teichner, S. J., J. Catlaysis, ~ (1973) 298-308; ibid, ~ (1974) 123-132

Giese, R. F., Wellers, S., and Datta, P,, Z, Kristallogr., 134 (1971) 275-277—

Ginsberg, H. and Koster, M., Z. Anorg. Chem., 27 I (1952) 41-48—

Ginsberg, H., Hiittig, W,, and Strunk-Lichtenberg, G., Z, Anorg. Allg, Chem,, 293 (1957) 33-46; 2W-213—

Ginsberg, H., Hiittig, W., and Stiehl, Z, Anorg, Allg. Chem., ~ (1961) 233-244

Ginsberg, H., Hiittig, W., and Stiehl, H., Z. Anorg. Allg. Chem., 318 (1962) 238-256—

80

—

Oxide and Hydroxides of Aluminum

Bibliography (Cent)

Ginsberg, H. and Wefers, K., Metall., ~ (1962) 173-175; ibid, ~ (1963) 202-209

Gitlesen, G., Herstad, 0,, and Motzfeld, V., Selected Topics in High Temp. Chem. (Oslo) ( 1966) 180-196

Gitzen, W. H., Alumina as a Ceramic Material, The American Ceramic Society (1970)

Glastonbury, J. R., Chem. and Ind. (London) (1969) 121-125

Glemser, O. and Rieck, G., Angew. Chem., ~ (1956) 182

Ganeyev, 1. G. and Rumyantsev, V, N., GeoW]miya, ~ (1974) 1402-1403

Gopienko, G. N., Zavaritskaya, T. A., and Pashkevich, L. A., Zvei. Metall., Q ( 1970) 53-56

Goton, R., Thesis No. 146, Univ. Lyon (1955)

Gout, R. and Dandurand, J. L. Tran. ICSOBA, 13(18) (1983) 117-126

Govinda, S. and Rae, K. V., Phys. Status Solidi A, 27(2) (1975) 639-644

Green, R. H. and Hem, S. L., J. Pharm. Sci., @ (1974) 635-637

Greenberg, L. and Wright, K. R. N., Proc. Royal Sot. A., 232 (1955) 403-406

GreviOe, C., ref. in Encycl. Britannica 1Itb ~., “Corundum” (1910) 207

Griftiths, K. F., United States Patent 4,291,011 (1981)

Grimshaw, R. W. and Peacey, J. G., U.S. Patent 3,947,562 (1976)

Gmbitsch, H., Geymer, W., and Burik, E., Aluminum, ~ (1961) 569-573

Gulbransen, E. A. and Wysong, W. S., Pbys. COO. Chem., U (1947) 1087-1091

Haas, H. and Holdaway, M. J., EOS Trans. Amer. Geophys. Union, ~ (1970) 437-441

t

Haber, F., Naturwissen., ~ (1925) 1W7-1012

Hannah, R. W., Correspondence (1960)

Harris, M. R. and Sing, K. S. W., J. Appl. Chem., ~ (1958) 586-589

Hart, R. K., Trans. Faraday SW,, ~ (1957) 1020-1025

Hass, H., Amer. Mineralogist, ~ (1972) 1375-1385

Hathaway, J. C. and Schlanger, S. O., Nature (London), ~ (1962) 265-266

Hauschild, U., Z. Anorg. Allg, Chem., 324 (1963) 15-30—

Hauschild, U., Brit. Patent 950,165 (19M) and Ger. Patent 1,182,643 (1964)

81

Alcoa Laboratories

Bibliography (Cent)

Haiiy, R. J., Traite de Mineralogies, Ptis, ~ (1801) 358

Heilman, R, H., Mech. Eng., ~ (1936) 291-292

Heine, M. A. and SFITY, P. R., J. Electrochem. Sm., 112 (1965) 359-363

Hem, J, D. and Roberson, C. E., U.S. Geol. Suwey Water Supply Paper 1827-A (1967)

Hem, S. L., Russo, E. J., Bahal, S. M., and Levi, R, S,, J, Pharm, Sci., 59 (1970) 317-321

Henley, V. F., Ancdic Oxidation of Aluminum and Its Alloys, New York, Pergamon Press (1982)

Henrich, V. E., Rep. Prog. Phys., ~ (1985) 1481-1541

Hill, G, J,, IEEE Trans. Instrum. Meas., IM-23(4) (1974) 443-446

Hill, R. J., Clays and Clay Minerals, 29(6) (1981) 435-445

Ho, P. and Bums, R. P., High Temp. Sci,, ~ (1980) 31-39

Hoar, T. P. and Wood, G, C,, Electrochim. Acts, 7 (1962) 333-336

Hoch, M. and Johnston, H. L., J. Amer. Chem, Sot., ~ (1954) 2560-2561

Hoffmann, F., Observationum Physico-Chimicarum Selectroium, Hallae (1722)

Honiche, D., Appl. Catalysis, ~ (1983) 179-198

Hoppe, W,, Z. Krist,, 103 (1940) II-15

Houben, G. M. M., Thesis, Delft (1951)

Hsu, P, H, and Bates. T, F., Mineral, Msg., ~ (1964) 749-768

Hsu, P. H., Soil Sci., 103 (1967) 101- I 10—

Hsu, P. H,, Alumin. Hydrox, and Oxyhydroxides, in Minerals in Soil Environments, Soil Sci, Amer. Publ. ( 1977) 99-143

Hunter, M. S. and Fowle, P., J. Electrochem. SW., 103 (1956) 482-485—

Hyde, S. B. and Yep, T. O., J. Electrochem. Sot., 123 (1976) 1721-1725—

lmari, I. and Harata, M,, Japan J, Appl, Phys., ~ (1972) 180-186

Ingebrethsen, G. J. and Matijevic, E. J., J. CoOoid Interface Sci., ~ (1984) 1-7

Ivanov, A. A,, Tolmachev, S. M., Ezhov, Yu. S,, Spiridonov, V. P., and Rambidi, N, G., Zh. Strukt. Khim,, 14(5) (1973)917-919

1

/

Jackson, E. A. and Roberts, J. B, Trans. Brh. Ceram. Sot., ~ (1955) 389-398

82


Bibliography (Cent)

lahr, K. F., and Pemoll, 1,, Z. Elektrochem., Q (1965) 221-231

Jan, J. P., Steineman, S., and Dinichert, P., J. Phys. Chem. Sol., ~ (1960) 349.350

JANAF Thermochemical Tables, J. Phys. Chem, Ref. Data, 7(3) (1978) 793-940; ~ (1985)—

Jeppesen, M. A,, J. Opt. Sot. Amer., g (1958) 629-635

John, C. S., Alma, V. C, M., and Hays, G. R., Appl. Catal., Q (1983) 341-346

Johnson, F. M. G., J. Amer. Chem. Sot,, ~ (1912) 911-912

JuiOet, F., Prettre, M., and Teichner, S. J., Comptes Rendus, 248 (1959) 555-7; ~ (1959) 1356-8—

Keller, F., Hunter, M. S., and Robinson, D. L., J. Electrochem. Sot., 100 (1953) 411-415—

Kennedy, G. C., Amer. J, Sci., 257 (1959) 563-573—

Khodakovoskii, I. L., Katorcha, L. V., and Kuyunko, N. S,, Geokhimiya, ~ ( 1980) 1606-1624

Kilham, J, K., Third Symp. on Combust. Flame and Explo. Phenomena, The Williams and Wilkins Co., Baltimore, M~land

(1949) 733-740

Kingery, W. D., Bowen, H. K., and Uhlmann, D. R., Introduction to Ceramics, John Wiley, New York (1976)

Kirk, C. T. and Hu&r, E. E., Jr., Surf, Science, ~ (1968) 217-245

Klabukov, V, Ya., Grebenshchikov, L. T., Kudryavtsev, V. A., and Iva.shevskii, M. A,, Teplo. Mass, Dugiat. Letat, Appar.,~ited by V. K. Shuckin, Kazan Aviats. Inst., Kazan, USSR (1983) 34-39

Klarsen Neklyudova, M. V., J. Tech. Phys. (USSR), ~ (1942) 519-551

Kleber, W. and Fehling, W., Z. Anorg. Allg. Chem., 338 (1965) 134-140—

Klein, D. J., U.S. Atomic Energy Corn. NAA-SR-25-42 (1958) I-24

Knozinger, H., Adv. in Catal. (1976) 184-271

Knozinger, H. and Ratnasamy, P., Catal. Rev. Sci. Eng., ~ (1978) 31-70

Kohn, J. A., Katz, G., and Broder, J. D, Amer. Mineralog., ~ (1957) 398-407

Kolesova, V. A., Opt. Spktrosk., VI (1959) 20

Kordes, E., Z. Kristal18r., U (1935) 193-228

Koster, H. M., Heidelberg Beitr, Mineral. Petro8r., ~ (1955) 23-@

Kraut, H., FIAe, E,, Schmidt, W., and Volmer, H., Ber., @ (1942) 1357-1373

Krishnan, R. S., Nature (bndon), 160 (1947) 26—

Kroger, F. A., Some Aspects of the Luminescence of Solids, Elsevier Publishing Co., New York (1948)

83

Alcoa Laboratories

Bibliography (Cent)

Kronberg, M. L., Acts Met,, 5(9) (1957) 507-524

Kroon, D. J. and Stolpe, C.v,d., Nature (bndon), 183 (1959) 944—

Kruse, R, D. and Hausner, H., Ber. Deut. Keram. Ges., ~ (1979) 192-196

Kummer, J. T., Progr. Solid State Cbem., ~ ( 1972) 141-175

Kuyunko, N. S., Malinin, S. D., and Khodakovskii, I. L., Geokhimiya, ~ (1983) 419-428

Ligerquist, K., Wallmark, S,, and Westgren, A., Z. Anorg. Chim., ~ (1937) 1-16

Lang, M. L, and Wolfe, W. L., Appl. Opt., 22(9) (1983) 1267-1268

Lang, S. M., Communication (J. B. Wacbtman) (1960)

Laubengayer, A. W. and Weiss, R. S., J. Amer. Chem. Sot,, ~ (1943) 247-250

Laudise, R. A. and BaOman, A. A., J, Amer. Chem, Sm., 80 (1958) 2655-2657—

Lawless, K. R., Rep. Progr. Phys., ~ (1974) 231-316

LeCars, Y,, ‘f’hery, J,, and Collogues, R., Rev. Hautes Temper. Refract,, ~ (1 972) 153

Lejus, A. M. and Collogues, R., C. R. Acad. Sci., ~ (1961) 1475-1476

hjus, A, M,, Thise, Parais (March, 1964); Rev, Hautes Temper. Refract., ~ (1964) 53.95

Levitt, A. P., Mater. Res. Stand,, 6(2) (1966) @-71

Liebertz, J., Ber. Deut. Keram, Ges., 49(9) (1972) 288-29o

Liehr, M., Thiry, P. A., Pireaux, J. J., and Candano, R,, J. Vat. Sci. Technol., ~ (1984) 1079-1082

Lima-de-Faria, J,, Z, Kristallogr., 119 (1963) 176-203—

Lippens, B. C,, Thesis, Tech. Univ. Delft (1961)

Lippens, B. C. and DeBoer, J, H., Acts Crystallogr,, ~ (1964) 1312-1321

Llppens, B. C. and Steggerda, J. J., Physical and Chemical Aspects of Absorbents and Catalysts, Mited by B, G. Linsen,Academic Press, bndon (1970) 171-209

Litovchenko, A. V., Brekhovskikh, S, M., and D~onyanov, V. V., Izv. Akad. Nauk SSSR, Neorg. Mater., ~ (1983)1488-1491

Locsei, B. P,, Nature (bndon), M (1962) 177-178

Long, G. and Foster, L, M., J. Amer. Ceram. Sot,, 44 (1961) 255-258—

bwenstein, E. V., J, Opt. Sot. Amer., ~ (1961) 108-112

Mackenzie, K. J. D,, J. Brit. Ceram. Sac., 5(2) (1968) 183-193

MacKenzie, K. J. D., Thermal Anal., 5 (1973) 363-367—

Mahan, A. I., BitterIi, C,, Cannon, S. M., and Grant, D. G., J. Opt. Sot. Amer., 59(1) (1969) 49-59

Maiman, T, H., Nature (~”do”), 187 (19613) 693—

84


Malit.son, 1. H., Murphy, F. V., and Rodney, W

Bibliography (Cent)

S., J. Opt. Sot. Amer., ~ (1958) 72-77

Mapes, J. E. and Eischens, R. P., J, Phys, Chem., 58 (1954) 1059-1064—

Marhoe, E. C. and Bentur, S., Silicates Jnd., 26 (1961) 389-399—

Margraff, A. S., Mere. Acad. Berlin, U (1754)

Mason, R, B., Metal Finishing J., ~ (1957) 67-69

Matijevid, E. and Stryker, J. L., J. CoOoid Interface Sci., ~ (1966) 68-77

Matsushita, S., Kennedy, G, C., Akella, J., a“d Haygarth, J., Amer. J. Sci., 265 (1967) 28-44—

May, H. M,, Helmke, P. A., and Jackson, M. L., Geochim. Cosmochim, Acts, ~ (1979) 861-868

Megaw, H. D., Z. KristaOogr,, A87 (1934) 185-204

Mendenhall, C, E. and Wood, R. W., Phil. Msg., ~ (1914) 316-320

Michaud, M., C. R. Acad, Sci,, 228 (1949) 1115-1116—

Miller, A, C., Private Communication (1986)

Milligan, W. 0. and McAtee, J. L., J. Phys. Chem., @ (1956) 237-277

Misra C. and White E. T., J. Cryst. Growth, ~ (1971) 172-178

Miyazawa, H, and Okada, J., J. Phys. Sot. Japan, ~ (1951) 55-59

Mohapatra, S. K. and Kroger, F. A., J. Amer. Chem. SM., Q (1978) IO6-109

Montoro, V., Ricerca Sci. Prog. Teen. Roma, ~ (1942) 565-571

Mmlenar, R. J., Evans, J. C., and McKeever, L. D., J. Phys. Chem., ~ ( 1970) 3629-3636

Moore, B. J., J. Sot. Chem, [rid. (London), ~ (1940) 119-120

Morey, G. W., Econ, Geol., ~ (1957) 225-251

Morterra, C., Zecchina, A., Coluccia, S., and Chiorino, A., J. Chem. Sot., Faraday Trans., ~ (1977) 15w-1560

Muan, A. and Gee, C. L., J. Amer. Ceram. Sot., ~ (1955) 207-212

Nail, S. L., White, J. L., and Hem, S. L., J. Pharm. Sci., ~ (1976) 1188-1191; ibid, ~ (1976) 1195-1198

NBS: Tables of Chemical Thermodynamic Properties, J. Phys. Chem. Ref. Data, ~ (1982) Sup. 2, 2-127

Neubaus, A. and Heide, H., Ber. Deut. Keram. Ges., Q (1965) 167-184

Newnham, R. E. and DeHaan, Y. M., Z. Kristallogr., ~ (1962) 235-237

Alcoa Laboratories

Bibliography (Cent)

Nielsen, T. H. and Leipld, M. H. J. Amer. Ceram. Sot., 46 (1963) 381-387

Nishhijima, T., Kawada, T., and Ishihata, A., J. Amer. Ceram. SW., ~ (1965) 31-34

Norton, F. H., Kinge~, W. D. Economos, G., and Humenik, M., USAEC Report NYO-3144 (1953)

Oberlander, R. K., Appl, Ind. Catal., Vol. 3, Acad. Press (1984)

O1sbi, Y. and Kingery, W. D., J. Chem. Phys., ~ (1960) 480-486

Oliver, J. and Poirier, R., Surf. Sci., 105 (1981) 347-356—

Osmont, H. E. and Emerson, R. B., U.S. Patent 3,222,129, Kaiser Alum. (1965)

Ozkan, O. T. and Moulson, A. J., J. Appl. Phys., ~ (1971) 983-986

Paladino, A. E. and Kingery, W. D., J. Chem. Phys., ~ (1962) 957-962

Pa@e, D., Tertian, R., and Biais, R., Bull. Sm. Chim. Fr. (1958) 1301-1310

Parks, G. A., Amer. Mineralogist, 57 (1972) 1163-1189

Parkyns, N. D., J. Chem. Sm. (A) (1969) 410-417

Pauling, L. and Hendricks, S. B., J. Amer. Chem. Sm., Q (1925) 781-790

Pauling, L., Proc. Nat. Acad. Sci., ~ (1930) 123-128

Pavlushkin, N. M., Steklo Keram., 14 (1957) 14-15

Peri, J. B., J. Phys, Chem., & (1965) 220-230

Peri, J. B., J. Phys. Chem., ~ (1966) 1482-1491

Peri, J. B. J. Phys, Chem,, ~ (1968) 2917-2925

Perrotta, A. J. and Young, J. E., J. Amer. Ceram

Petz, J. I., J. Chem. Phys,, ~ (1968) 909-911

Sot. , ~ (1974) 405-407

Phillips, D. S., Mitchell, T. E., and Heuer, A. H. Phil. Msg. A42 (1980) 417-432

Pierre, A. C. and Uhlmann, D. R., MIT Ind. LLa, Rep. 3-3-86, Dir. Cum. Res. 5.14,029

Plummer, M., J. Appl. Chem., ~ (1958) 35-44

Pott, J. H., Lithogeognosia, Potsdam (1746); Berlin (1754)

Pryor, M. J., Oxidation of Metals, ~ (1971)

Ragnwsdottir, K. V. and Walther, J. V., Geochim, Cosmmhim. Acts, Q (1985) 2109-2115

86

)1.

I


Bibliography (Cent)

Rai, K. N. and Ruckenstein, E., Joum. Catal., 40 (1975) 117.123

Raman, C. V, Proc. Indian Acad. Sci., 34A (1951) 61-71

Rankin, G. A. and Merwin, H. E., J. Amer. Chem. Sot., 38(3) (1916) 568-588

Rae, S. R. and hela, M., Cum. Sci. (India), ~ (1953) 72-73

Rasmussen, J. J. and Nelson, R. P., Surface Tension and Density of Molten Alumina, Rep. Battelle-Northwest, Richland, Wash.(1970)

Reichert, P. P. and Yost, W. J., J. Chem. Phys,, 14 (1946) 495

Ridgway, R. R., Klein, A. A., and O’Leary, W. J., Trans. Electrochem. Sm., ~ (1036) 71-86

Roberts, J. P. and Watt, W., Ceramics and Glass, Selected Govt. Res. Reps., Minist~ of Supply (London), ~ (1952); Trans.Brit. Ceram. Sot., 48 (1949) 343-351; 50 (1951) 122-143

Robie, R. A., Hemingway, B. S., and Fisher, R. J., Geological Survey Bull. 1452, U.S. Govt. Print Press (1978)

Rogers, W. G., Baker, G. S, and Gibbs, P., Mechanical Properties of Engineering Ceramics, Edited by W. W. Kriegel and H..Palmour, Interscience Publishers Inc., New York (1961) 303-311

Rooksby, H. P., Correspondence (1950)

Rooksby, H. P, and Rooymans, C. J. M., Clay Miner. Bull., ~ (1961) 234-238

Rosynek, M. P. and Hightower, J. W., Proc. 5th Int. Congr. Cat., 2 ( 1972) 851-859

Roth, W. A., Wirths, G., and Berendt, H., Z. Elektrochem., ~ (1942) 2M-267

Roth, W. L., Reidinger, F., and LaPlace, S., Super Ionic Conductor, Edited by G. D. Mahan and W. L, Roth, Plenum Press,New York (1976) 223-230

Rothbauer, R., Zigan, F., and O’Daniel, H., Z. Kristallogr. ~ ( 1967) 317-33 I

Rouquerol, J., Rouquerol, F., and Ganteaume, M., J. Catal., ~ (1975) 99-11 O; ~ (1979) 222-230

Roy, R., Hill, V. G., and Osbom, E. F., Ind. Eng, Chem., Q (1953) 819-820

Rummel, T., Z. Physik, ~ (1936) 518-522

Russell, A. S. and Cwbran, C. N., Ind. Eng. Chem., Q (1950) 1336-1340

Russell, A. S., Edwards, J. D., and Taylor C, S., J. Metals, ~ (1955) 1123-1128

Russell, E. E. and Bell, E. E., J. Opt, Sot. Amer., ~ (1967) 54-546

Russell, J. D., Farmer, V, C., and Lewis, D. G,. Spectrochim. Acts, Part A, 34A(12) (1978) 1151-1153

Rysbkewitz, E., Ber. Deut, Keram. Ges., ~ (1941) 54-65; 363-371

Alcoa Laboratories

Bibliography (Cent)

Ryshkewitz, E., Ber. Deut. Keram. Ges., ~ (1942) 243-260

Saalfeld, H., Z. Anorg. Allg. Chem., 286 (1956) 174-179; 291 (1957) 117-21—

Sdfeld, H., Neues. Jahrb. Mineral. Abb. ~ (1960) 1-87

Saalfeld, H., Proc. 8th Int. Cemm. Cong., Copenhagen (1962) 71-74

Saalfeld, H. and Mehrotra, B., Ber. Deut. Keram. Ges. Q (1965) 161-166

Saalfeld, H. Mehrotra, B. B. Naturwiss., ~ (1966) 128-129

Saalfeld, H. and Jarchow, 0., Neues. Jahrb. Mineral. Abh., ~ (1968) 185-191

SaaIfeld, H., Mathies, H., and Datta, S. K., Ber. Dtsch. Keram. Ges., ~ (1968) 212-215

Saalfeld, H., Mathies, H., and Karmakar, M., Neues Jahrb. Mineral. Abn., 109 (1968) 232-237—

Saalfeld, H. and Wedde, M,, Z. Kdstallogr., 139(1-2), (1974) 129-J35

Sate, T., J. Appl. Chem., ~ (1962) 9-12; 553

Sate, T., Z. Anorg. Allg. Chem., 381 (1972) 69; 167

Schlenker, F., Fark und Lack, ~ (1956) 9-14; ibid, ~ (1958) 174-183

Schmti, H., Z. Naturforscher., ~ (1946) 322-323

Schmfi, H., Z. Naturforscher., ~ (1946) 323-324

Schneider, A. and Gattow, G., Z. Anorg, Chem., 277 (1954) 41-48

Schneider, S, J,, Pure Appl, Chem., ~ (1970) 115-119

Schwn, R. and Roherson, C. E., Amer. Mineral., ~ ( 1970) 43-77

Scholder, R, and Mansmann, M,, Z, Anorg. Allg. Chem., 321 (1963) 246-261—

Schwartz, B., J, Amer. Ceram, Sm., ~ (1952) 325-333

Schwierscb, H., Chem. Erde, ~ (1933) 252-315

Scokart, P. 0,, Amin, A., Defosse, C., Rouxhet, P. G., J. Phys. Chem., ~ (1981) 140f-1412

Selwmd, P. W., Coi-respndence (1950)

Sema, C. J,, White, J. L., and Hem, S. L., Soil Sci., ~ (1977) 1009-1013

Setoh, S. and Miyata, A., Sci. Pap. Inst. Phys. Chem. R.s,, Tokyo, ~ (1932) 237-241

Shimizu, Y,, Miyasbige, S,, and Funaki, K., J. Chem, Sot. Jap,, Ind, Chem. Sect., g (1958) 475

88


Bibliography (Cent)

Smeltzer, W. W,, J. Electrochem. SO.., 103 (1956) 209-214

Smith, R. W. and Hem, J. D, U.S. Geol. Survey Water Supply Paper, 1827-D (1972)

Soled, S., J. Catal., ~ (1983) 252-257

Souza-Santas, P., VaOejo-Freire, A., and Souza.Santos, H. L., Kolloid-Z., 133 (1953) 10 I-107—

Sprauer, I. W. and Pierce, D. W., J. Phys. Chem., M (1940) 909-916

Spriggs, R. M. and Brissette, L, A., J. Amer. Ceram. SW., Q (1962) 198-199

Stavroltils, J. A. and Norton, F. H., J. Amer. Ceram. Sot., ~ (1950) 263-268

Stevens, R. and Binner, J. G, P,, J. Mater. Sci., ~ (1984) 695-715

Stillwell, C. W., J. Phys. Chem., ~ (1926) 144-1466

Stofel, E. and Conrad, H., Trans. AIME, ~ (1963) 1053-1060

Stumpf, H. C., Russell, A. S., Newsome, J. W., and Tucker, C. M., Ind. Eng. Chem., Q (1950) 1398-1403

Swanson, H. E. and Fuyat, R. K., Nat. Bur. Stand. (U. S.) Cir., ~, ~ (1953), ~ (1954)

Swanson, H. E., Cook, M. L., Isaacs, T., and Evans, E. H., Standard X-Ray Powder Diffraction Patterns, Nat. Bur. StandCirc., ~ (1960) 539

Tajima, S., Adv. Con. Sci. Technol., ~ (1970)227-364

Tak~e, K., Prm. Imp. Acad, Japan, ~(1933) 113-117

Tamele, M. W., D,s. Faraday Soc., No. 8(1950)270-279

Taylor, C. S. and~wards, J. D., Heat. Piping Air Cond., l_l (1939) 59-63

Teichner, S. J., Nicolaon, G. A., Vicarini, M. A., and Gardes, G. E. E., Adv. CoO, Interf. Sci., ~(1976) 245-273

Tefft, W. E., J. Res. Nat. Bur. Stand. (U. S.), Sect. A,~(1966) 277-280

Tertian, R. and Pa@e, D., J. Chim. Phys., ~ (1958) 341-353

Tettenhorst, R. and Hoffmann, D. A., Clays and Clay Miner., ~ (1980) 373-380

Thermophysical Properties of Matter, Thermophysical Propetiies Research Center, Purdue University, IFf/PLENUM, New York(1970)

Thery, J., Briancon, D., and Collongues, R., Compt. Rend., @(1961) 1475-1477

Thery, J., Lejus, A. M,, Briancon, D., and Collongues, R., Bull. Soc. Chim. (1961)973-975

Thery, J.and Briancon, D., Rev. Hautes Tem~r. Refract., ~ (1964)221-227

Alcoa Laboratories

Bibliography (Cent)

Thilo, E., Jander, J., and Seeman, H., Z. Anorg. Allg. Chem., 279 (1955) 2-17—

~ompson, G. E., Fumeaux, R. C., Wood, G. C., Richardson, J. A., and Goode, J. S., Nature, ~ (1978) 433-435

Thompson, G. E., Fumeaux, R. C., Wood, G. C., and Hutchings, R., Pratt. Metallogr., ~ (1981) 594-599

Thompson, G. E., Wood, G. C., and Shimuzu, K., Electrochimica Acts, ~ (198 I) 951-954

Tlmofejeva, W. A. and Woskanjan, R. A., Kristallografiya, ~ (1963) 293

Tlmsit, R. S., Waddington, W. G., Humphreys, C. J., and Hutchinson, J. L., Appl. Phys. Lett., ~ (1985) 830-832

Torkar, K. and Bergmann, O., Monatsh. Chem., U (1960) 400-405

Torkar, K. and Krishner, H., Monatsh. Chem., ~ (1960) 757

Torkar, K. and Krischner, H., Monatsh Chem., ~ (1960) 76’-773

Torkar, K., Egghart, H., Krischner, H., and Worel, H., Monatsh. Chem., ~ (1961) 512-524

Torkar, K. and Bertsch, L., Monatsh. Chem., ~ (1961) 525-532

Torkar, K. and Egghart, H., Monatsh. Chem., ~ (1961) 755-767

Torkar, K. and Mostad, F., Monatsh. Chem., ~ (1962) 657-660

Torkar, K., Giegler, H., Egghart, H., Faye, G., and Worel, H., Monatsh. Chem., U (1967) 450-455

Toropov, N, A., C. R. Acad. Sci, URSS, I (1935) 147-149

Tomey, J., New York Med. Phys. Joum., ~ (1822) 68

Tramhouze, H. and Pernn, M., Compt. rend., 236 (1953) 1261—

Tsuchida, T., Fumichi, R., and Ishii, T,, Thermochimica Acts, ~ (1980) 103-115

U.S. Geological Survey: Bulletin 1452, U.S. Govt. Printing Office, Washington, Reprinted with Corrections (1979)

Ubich, F, Norsk. Geol. Tidsskrift, ~ ( 1925) 115-122

Unmack, A., 2nd Int, Congr, Crystallogr. Abstr. of Papers, Stockholm, (Inorganic Structures No. 8, Section G) ( 1951)

Van Nordstrand, R. A., Hettinger, W. P., and Keith, C. D., Nature (bndon), 177 (1956) 713-714—

Vauqelin, L. N,, Ann, Chim. Phys, (l), g (1802) 113

Vedder, W. and Vermilyea, D, A,, Trans. Faraday SW., Q (1969) 561-584

Vicarini, M. A., Nicolaon, G. A., and Teichner, S. J., Bull. Sot. Chim, France, : (1969) 1466-76; ~ (1970) 431-46; ~ (1970)3384-7

Volf, F. F. and Kug”etsov, S. 1., J. App. Chem, (USSR), Eng. Trans., ~ (1953) 255-264

90


I

I

Bibliography (Cent)

Von Hippel, A, R., MIT Re~rt 126 ( 1953, 1959); Dielectric Materials and Applications, John Wiley, New York ( 1954)

Wachtman, J. B. and Maxwell, L. H., Amer. Cer. Sot. Bull., 32 ( 1953) 145; J, Amer. Ceram, SW,, 37 ( 1954) 29 1-299; WrightAir Dev. Cent. Tech, Rep. 53-265 (1954), 57.526 (1957); J, A~er. Ceram. Sot., 42(9) (1959) 432-437

Wachtman, J. B. and Lam, D. G., J. Amer. Ceram. SW,, 42 (1959) 254-260

Wachtman, J. B., Tefft, W. E,, Lam, D. G., and Stinchtield, R. P., J. Research, National Bureau of Standards, @A(3) ( 1960)

213-228

Wachtman, J. B., Tefft, W. E., Lam, D. G., and Apstein, G. S., Phys, Rev., ~ (1961) 1754-1759

Wachtman, J. B,, Scuderi, T. G., and Cleek, T. W., J, Amer. Ceram, Sot., ~ (1962) 319-323

Wall, J, R, D,, Wolfenden, E. B,, Beard, E. H., and Deans, T., Nature (London), 196 (1962) 2@-265—

Wartenherg, H., Z, Anorg. Chem., 269 (1952) 76-85; 270 (1952) 328—

Watson, J, H. L., Pmsons, J,, Vallejo-Freire, A., and Souza-Santos, P., Kolloid-Z., 140 (1955) 102-II 2; ibid, 154 (1957) 4-15— —

Weber, N. and Kummer, J. T., ASME, Advan. Energy Conv. Eng, (1967) 913

Weetiman, J., Trans. ASM, Q ( 1968) 681-694

Wefers, K,, Aluminium, ~ (1961) 19-28

Wefers, K., Erzmetall., 15 (1962) 339-343

Wefers, K., Naturwissenschaften, 49 (1962) 204-205


Wefers, K., Erzmetall., ~ (1965) 459-463

Wefers, K,, Metall., ~ (1967) 423-431


Wefers, K., Aluminium, ~ (1973) 553-561; 622-625

Wefem, K. and Wallace, P. F., Aluminium, 52 (1976) 485-490

Wcfers, K., Aluminium, 57 (1981) 722-726

Wei, P. S, P. and Smith, A, W., J. Vat. Sci. Technol., ~ (1972) 1209-1213

Weis, J,, Z. Anorg, Allg, Chem., 354 (1967) 149-162

Weiser, H. B. and Milligan, W. 0,, J. Phys. Chem., ~ (1934) 1175- I 182

Weiser, H. B,, I“org. Colloid Chemistry, Vol. II, The Hydrous Oxides and Hydroxides, J. Wiley, New York publ, (1935)90-120

91

Alcoa Laboratories

Bibliography (Cent)

W,ells, A. F., Structural Inorganic Chemistry, 5th Edition, Clarendon Press, Oxford (1984) 190-191

Wemick, S. and R. Pinner, The Surface Treatment of Aluminium and Its AI1oYs, R. Dra~r, publ., Teddington UK (1972)

Wleder, 1,, Rev, Sci. Inst., ~ (1959) 995-997

Willstatter, R., Kraut, H., and Erbacher, 0., Ber., 588 (1925) 2448-2458—

Wilson, S. J. and McConnell, J. D. C., J. Sol. State Chem., ~ (1980) 315-322

Wilson, S. J., Mineral Msg., Q ( 1979) 301-306

Wilson, S. J., F’roc. Brit. Ceram. Sot., ~ (1979) 281-294

Wilson, S. J., McConnell, J. D. C., and Stacey, M. H., Joum. Mat. Sci., ~ (1980) 3081-3090

Wilson, S. J. and Stacey, M. H., J. Colloid Intefi. Sci., ~ (1981) 507-517

Wong, P. and Robinson, M., J. Amer. Ceram. Sm., 53(11) (1970) 617-621

Xu, Y., Thompson, G. E., and Wocd, G. C., Trans. Inst. Met. Finish., ~ (1985) 98-103

Yamaguchi, G., Yanagida, H,, and One, S., Bull. Chem. Sot. Japan, ~ (1964) 752-754; 1555-1557

Yamaguchi, G. and Yanagida, H., Bull. Chem. Sot. Japan, ~ (1964) 1229-1231

Yamaguchi, G., Yanagida, H,, and One, S., J. Ceram. Assoc. Japan, 74(3) (1966) 26-30

Yamaguchi, G. and Okumiya, M., Bull. Chem. Sm. Japan, ~ (1969) 2247-2249

Yamaguchi, S., J. Appl. Phys., 45(7) (1974) 3207-3208

Yamagucbi, S., Z. Phys. Chemie, 256 ( 1975)—

Yanagida, H, and Kroger, F. A., J. Amer. Ceram. Sot., ~ (1968) 70C-706

Yates, B., CmPer, R. F., Pojur, A. F, J. Phys.: Solid State Physics, ~ (1972) 1W6-1058

Yoldas, B. E,, J. Mat, Sci,, ~ (1975) 1856-1860

Yosimura, M. and Bowen, H, K,, J. Amer. Ceram. Sot., ~ (1981) 404-410

Yudin, B. F. and Karklit, A. K., Zh. Prikl. Khim., ~ (1966) 537-44

=Iinski, B. J. J. and Uhlmann, D. R., J. Phys. Chem. Solids, Q (1984) 1069-1090

Zigan, F., Joswig, W., and Burger, N., Z, KristaOog,, 148 (1978) 255-273—

/

92

Oxides and Hydroxides of Aluminumepsc511.wustl.edu/Aluminum_Oxides_Alcoa1987.pdf · Preface One hundred years after the invention of the Bayer Process for the production of aluminum

Documents