MATERIALS FOR ADAPTIVE STRUCTURALACOUSTIC CONTROL

Cv

Period February 1, 1992 to January 31, 1993

__ •Annual ReportA _DTICVOLUME IV IVMY121993*

OFFICE OF NAVAL RESEARCHContract No. N00014-92-J-1510

APPROVED FOR PUBLIC RELEASE - DISTRIBUTION UNLIMITED

Reproduction in whole or in part is permitted for any purposeof the United States Government

L. Eric Cross

PENNSTATEV

THE MATERIALS RESEARCH LABORATORYUNIVERSITY PARK, PA

93-11219

REPORT DOCUMENTATION PAGE 0 No ....OMS No 0704G0I86

1. AGENCY USE ONLY (tea,, bla,l 2. REFORT DATE -• R(PORt TYPE AND DAtES COVERED.,

104/06/93 ANNUAL REPORT 02/01/92 TO 01/31/934. TITLE AND SUBTITLE S. FUNDING NUMBERS

MATERIALS FOR ADAPTIVE STRUCTURAL ACOUSTIC CONTROL

6. AUTHOR(S)

L. ERIC CROSS

7. PERFORMING ORGANIZATION NAME(S) AND AOURESS(ES) 8. PERFORMING ORGANIZATION

MATERIALS RESEARCH LABORATORY REPORT NUMBER

TilE PENNSYLVANIA STATE UNIVERSITYUNIVERSITY PARK, PA 16802 N00014-92-J-1510

9. SPONSORING/MONIT'ORING AGENCY NAME(S) AND AOURESS(ES) 10, SPONSORING (MONITORING

OFFICE OF NAVAL RESEARCH DOUGLAS E. IIEATON AGENCY REPORT NUMBER

CODE 1513:NRJ DEPT. NAVY/ONR, RES. REP.800 NORTH QUINCY STREET THE 01110 STATE UNIV. RES. CTR.ARLINGTON, VA 22217 1960 KENNY ROAD

COLUMBUS,,O1 _43210-106311. SUPPLEMENTARY NOTES

12a. DISTRIBUIlON/ AVAILABILITY STATEMENT I2b. DISTRIBUTION CODE

13, ABSTRACT (Mfaxinluri 200 woods)

SEE FOLLOWING PAGE

14. SUBJECT TERMS .5. NUMBER OF PAGES

16, PRICE CODE

Ti. SECURITY CLASSIFICATION "'. SECURITY CLASSIFICATION 19, SECURITY CLASSIFICATION 20. LIMITATION OF AB '

Of REPORT OF THIS PAGE OF ABSTRACT

,tndsed'rO'm 1 98 IqPv 2 89)

lii 3

ABSTRACT

"This report documents work carried out in the Materials Research Laboratory of thePennsylvania State University over the first year of a new ONR sponsored UniversityResearch Initiative (URI) entitled "Materials for Adaptive Structural Acoustic Control." Forthis report tie activities have been grouped under the following topic headings:

1. General Summary Papers.2. Materials Studies.

3. Composite Sensors.

4. Actuator Studies.

5. Integratio;n Issues.

6. Processing Studies.

7. Thin Film Ferroelectrics.In material studies important advances have been made in the understanding of the

evaluation of relaxor behavior in the PLZT's and of the order disorder behavior in leadscandium tantalate:lead titanate solid solutions and of the Morphotropic Phase Boundary in thissystem. For both composite sensors and actuators we have continued to explore and exploitthe remarkable versatility of the flextensional moonie type structure. Finite element (FEA)calculations have given a cOear picture of the lower order resonant modes and permitted theevaluation of various end cap metals, cap geometries and load conditions. In actuator studiesmultilayer structures have been combined with flextensional moonie endcaps to yield highdisplacement (50 Vi meter) compact structures. Electrically controlled shape memory has beendemonstrated in lead zirconate stannate titanate compositions, and used for controlling a simplelatching relay. Detailed study of fatigue in polarization switching compositions hashighlighted the important roles of electrodes, grain size, pore structures and mnicrocracking anddemonstrated approaches to controlling these problems. For practical multilayer actuators auseful lifetime prediction cmi be made from acoustic emission analysis.

New modelling of 2:2 and 1:3 type piezoceramic:polymer composites has given moreexact solutions for the stress distribution and good agreement with ultradilatometermeasurements of local defornations. Composites with 1:3 connectivity using thin wall ceramictubes appear to offer excellent hydrostatic sensitivity, unusual versatility for property controland the possiblity to use field biased electrostrictors in high sensitivity configurations.

Processing approaches have continued to use reactive calcining and have supplied the groupwith the wide range of ccramics used in these studies. For lead magnesium niobate:lead

ASISTRACT (continued)

titanate solid solutions grain size effects in samples of commerical purity have been traced to a

thin (-20 n meter) glassy layer at the grain boundary. In parallel with the ONR URI the

laboratory has extensive DARPA and Industry sponsored research on ferroelectric thin films, a

very short selection of most relevant papers has been included for the convenience of users.

.S . 1% -7 - /,

4'• # . .

\'7 - .";", /

MATERIALS FOR ADAPTIVE STRUCTURALACOUSTIC CONTROL

Period February 1, 1992 to January 31, 1993

Annual Report

VOLUME IV

OFFICE OF NAVAL RESEARCHContract No. N00014-92-J-1510

APPROVED FOR PUBLIC RELEASE - DISTRIBUTION UNLIMITED

Reproduction in whole or in part is permitted for any purposeof the United States Government

L. Eric Cross

PENNSTATEVTHE MATERIALS RESEARCH LABORATORY

UNIVERSITY PARK, PA

TABLE OF CONTENTS

ABSTRACT ................................................ 6

INTRODUCTION ........................................... . 7

1.0 GENERAL SUMMARY PAPERS .......................... 10

2.0 MATERIALS STUDIES ................................ 10

3.0 SENSOR STUDIES ................................... 12

4.0 ACTUATOR STUDIES ................................. 13

5.0 INTEGRATION ISSUES ............................... 14

6.0 PROCESSING STUDIES ............................... 14

7.0 THIN FILM FERROELECTRICS ......................... 15

8.0 HONORS AND AWARDS .............................. 15

9.0 APPRENTICE PROGRAM .............................. 17

10.0 PAPERS PUBLISHED IN REFEREED JOURNALS ............. 18

11.0 INVITED PAPERS PRESENTED AT NATIONAL ANDINTERNATIONAL MEETINGS ........................... 21

12.0 INVITED PRESENTATIONS AT UNIVERSITY, INDUSTRYAND GOVERNMENT LABORATORIES ..................... 23

13.0 CONTRIBUTED PAPERS AT NATIONAL ANDINTERNATIONAL MEETINGS .......................... 24

APPENDICES

General Summary of Papers

I. R. E. Newnham, "Memories of Arthur Von Hippel," Ferroelectrics 1M2, 17 (1992).

2. C. Rosen, B. V. Hiremath and R. E. Newnham, "Piezoelectricity," American Inst. ofPhysics, New York (1991).

3. R. E. Newnham, "Ferroelectric Sensors and Actuators Smart Electroceramics,"Ferroelectric Ceramics, Editor N. Setter, Proc. of Summer School on Ferroelectrics,Ascona (1991).

4. R. E. Newnham, "Smart Electroceramics in the 1990's and Beyond," J. EuropeanCeramic Soc. (1992).

5. V. Sundar and R. E. Newnham, "Electrostriction and Polarization," Ferroelectrics 135,431 (1992).

Materials Studies

6. J. R. Giniewicz, A. S. Bhalla and L. E. Cross, "Identification of the MorphotropicPhase Boundary in Lead Scandium Tantalate-Lead Titanate Solid Solution System."

7. J. R. Giniewicz, A. S. Bhalla and L. E. Cross, "Variable Structure Ordering in LeadScandium Tantalate-Lead Titanate Materials."

8. J. R. Giniewicz, A. S. Bhalla and L. E. Cross, "Lead Scandium Tantalate:LeadTitanate Materials for Field Stabilized Pyroelectric Device Applications," FerroelectricsLetters J4, 21 (1992).

9. A. S. Bhalla, R. Guo, L. E. Cross, G. Burns, F. H. Dacol and R. R. Neurgaonkar,"Glassy Polarization in the Ferroelectric Tungsten Bronze (BaSr) Nb 20 6 ," J. Appl.Phys. 71 (11), 5591 (1992):

10. J. S. Yoon, V. S. Srikanth and A. S. Bhalla, "The Electrical Properties ofAntiferroelectric Lead Zirconate-Ferroelectric Lead Zinc Niobate Ceramics withLanthanum," Proc. ISAF 1992, Greenville, South Carolina, pp. 556.

11. E. F. Alberta, D. J. Taylor, A. S. Bhalla and T. Takenaka, "The DC Field Dependenceof the Piezoelectric, Elastic and Dielectric Constants for a Lead Zirconate BasedCeramic," Proc. ISAF 1992, Greenville, South Carolina, pp. 560.

12. W. Cao and L. Eric Cross, "The Ratio of Rhombohedral and Tetragonal Phases at theMorphotropic Phase Boundary in Lead Zirconate Titanate," Japan Journal of AppliedPhysics 31 (Pt. 1, No. 5A), 1399 (1992).

13. C. A. Randall, M. G. Matsko, W. Cao and A. S. Bhalla, "A Transmission ElectronMicroscope Investigation of the R3m - R3c Phase Transition in Pb(ZrTi)0 3 Ceramics,"Solid State Comm. K (3), 193 (1993).

14. Shaoping Li, Chi Yeun Huang, A. S. Bhalla and L. E. Cross, "90' Domain Reversal inPb(ZrxTil.x)Q 3 Ceramics," Proc. ISAF 1992, Greenville, South Carolina.

15. Shaoping Li, Jyh Sheen, Q. M. Zhang, Sei-Joo Jang, A. S. Bhalla and L. E. Cross,"Quasi Lumped Parameter Method for Microwave Measurements of DielectricDispersion in Ferroelectric Ceramics," Proc. ISAF 1992, Greenville, South Carolina.

16. H. Wang, Q. M. Zhang, L. E. Cross and A. 0. Sykes, "Piezoelectric Dielectric andElastic Properties of Poly (Vinylidene FluoridefTrifluorethylene)."

17. H. Wang, Q. M. Zhang, L. E. Cross and A. 0. Sykes, "Clamping Effect on

Piezoelectric Properties of Poly (Vinylidene Fluoride/Trifluoroethylene) Copolymers."

Composite Sensors

18. Ki-Young Oh, Yutaka Saito, Atsushi Furuta and Kenji Uchino, "Piezoelectricity in theField Induced Ferroelectric Phase of Lead Zirconate Based Antiferroelectrics," J. Amer.Ceram. Soc. 75 (4), 795 (1992).

2

Composite Sensors (continued)

19. R. E. Newnham, Q. C. Xu, K. Onitsuka and S. Yoshikawa, "A New Type ofFlextensional Transducer," Proc. 3rd Int. Mtg. on Transducers for Sonics andUltrasonics (May 1992).

20. C. A. Randall, D. V. Miller, J. H. Adair and A. S. Bhalla, "Processing ofElectroceramic-Polymer Composites Using the Electrorheological Effect," J. Mat. Res.8 (4), 1 (1993).

21. C. A. Randall, S. Miyazaki, K. L. More, A. S. Bhalla and R. E. Newnham,"Structural-Property Relations in Dielectrophoretically Assembled BaTiO 3Nanocomposites," Materials Letters 15 26 (1992).

22. C. A. Randall, S. F. Wang, D. Laubscher, J. P. Dougherty and W. Huebner,"Structure Property Relations in Core-Shell BaTiQ3:LiF Ceramics," J. Mat. Res.fin press).

23. C. A. Randall, G. A. Rossetti and W. Cao, "Spacial Variations of Polarization inFerroelectrics and Related Materials."

24. Jayu Chen, Qi Chang Xu, M. Blaszkiewicz, R. Meyer, Jr. and R. E. Newnham, "LeadZirconate Titanate Films on Nickel-Titanium Shape Memory Alloys: SMARTIES,"J. Amer. Ceram. Soc. 75 (10), 2891 (1992).

Actuator Studies

25. D. Darnjanovic and R. E. Newnham, "Electrostrictive and Piezoelectric Materials forActuator Applications," J. Intell. Mat. Syst. and Struct. 3, 190 (1992).

26. Y. Sugawara, K. Onitsuka, S. Yoshikawa, Q. C. Xu, R. E. Newnham and K.Uchino, "Metal-Ceramic Composite Actuators," J. Amer. Ceram. Soc. 75 (4), 996(1992).

27. Q. C. Xu, A. Dogan, J. Tressler, S. Yoshikawa and R. E. Newnham, "Ceramic-MetalComposite Actuators," Ferroelectrics Special Issue.

28. K. Uchino, "Piezoelectric Ceramics in Smart Actuators and Systems," Proc. 1stEuropean Conference on Smart Structures and Materials.

29. A. Furuta, Ki-Young Oh and K. Uchino, "Shape Memory Ceramics and TheirApplication to Latching Relays," Sensors and Materials 3 (4), 205 (1992).

30. K. Uchino and A. Furuta, "Destruction Mechanisms in Multilayer Ceramic Actuators,"Proc. ISAF 1992, Greenville, South Carolina, pp. 195.

31. Q. Jiang, Wenwu Cao and L. E. Cross, "Electric Fatigue Initiated by SurfaceContamination in High Polarization Ceramics," Proc. ISAF 1992, Greenville, SouthCarolina, pp. 107.

32. Q. Jiang, We-"-'u Cao and L. E. Cross, "Electric Fatigue in PLZT Ceramics."

3 ONR 1992 Annual Report

Integration Issues

33. Wenwu Cao, Q. M. Zhang and L. E. Cross, "Theoretical Study on the StaticPerformance of Piezoelectric Ceramic-Polymer Composite with 1-3 Connectivity," J.Appl. Phys. 72 (12), 5814 (1992).

34. Q. M. Zhang, Wenwu Cao, H. Wang and L. E. Cross, "Characterization of thePerformance of 1-3 Type Piezocomposites for Low Frequency Applications," J. Appl.Phys. 73 (3), 1403 (1993).

35. Q. M. Zhang, Wenwu Cao, H. Wang and L. E. Cross, "Strain Profile andPiezoelectric Performance of Piezocomposites with 2-2 and 1-3 Connectivities," Proc.ISAF 1992, Greenville, South Carolina, pp. 252.

36. Q. M. Zhang, H. Wang and L. E. Cross, "Piezoelectric Tubes and 1-3 Type TubularComposites as Tunable Actuators and Sensors."

37. Q. M. Zhang, H. Wang and L. E. Cross, "Piezoelectric Tubes and Tubular Compositesfor Actuator and Sensor Applications."

Processing Studies

38. Thomas R. Shrout and Scott L. Swartz, "Processing of Ferroelectric and RelatedMaterials: A Review."

39. G. A. Rossetti, D. J. Watson, R. E. Newnham and J. H. Adair, "Kinetics of theHydrothermal Crystallization of the Perovskite Lead Titanate," J. Crystal Growth 1251 (1992).

40. A. V. Prasadarao, U. Selvaraj, S. Komarneni and A. S. Bhalla, "Sol-Gel Synthesis ofLn2(Ln = La, Nd) Ti20 7," J. Mat. Res. .7 (10), 2859 (1992).

41. A. V. Prasadarao, U. Selvaraj, S. Komarneni and A. S. Bhalla, "Sol Gel Synthesis ofStrontium Pyroniobate and Calcium Pyroniobate," J. Amer. Ceram. Soc. 75 (10),2697 (1992).

42. A. V. Prasadarao, U. Selvaraj, S. Komarneni and A. S. Bhalla, "Fabrication ofLa2Ti20 7 Thin Films by A Sol-Gel Technique," Ferroelectrics Letters 14, 65 (1992).

43. S. F. Wang, U. Kumar, W. Huebner, P. Marsh, H. Kankel and C. G. Oakley, "GrainSize Effects on the Induced Piezoelectric Properties of 0.9 PMN-0.IPT Ceramic,"Proc. ISAF 1992, Greenville, South Carolina, pp. 148.

44. C. A. Randall, A. D. Hilton, D. J. Barber and T. R. Shrout, "Extrinsic Contrbutionsto the Grain Size Dependence of Relaxor Ferroelectric Pb(Mgl/3Nb2/3)O3:PbTiO3Ceramics," J. Mat. Res. 1 (4) (1993).

45. B. V. Hiremath, R. E. Newnham and L. E. Cross, "Barrier Layer Capacitor UsingBarium Bismuth Plumbate and Barium Plumbate," J. Amer. Ceram. Soc. 75 (11),2953 (1992).

46. U. Kumar, S. F. Wang, S. Varanasi and J. P. Dougherty, "Grain Size Effects on theDielectric Properties of Strontium Barium Titanate," Proc. ISAF 1992, Greenville,South Carolina, pp. 55.

4

Processing Studies (continued)

47. U. Kumar, S ' Wang and J. P. Dougherty, "Preparation of Dense Ultra-Fine GrainBarium Titaiiate-Based Ceramics," Proc. ISAF 1992, Greenville, South Carolina,pp. 70

Thin Film Ferroelectrics

48. J. Chen, K. R. Udayakumar, K. G. Brooks and L. E. Cross, "Dielectric Behavior ofFerroelectric Thin Films at High Frequencies," Proc. ISAF 1992, Greenville, SouthCarolina, pp. 182.

49. K. Uchino, N-Y. Lee, T. Toba, N. Usuki, H. Aburatani and Y. Ito, "Changes in theCrystal Structure of RF-Magnetron Sputtered BaTiO3 Thin Films," J. Chem. Soc.Japan 10Q (9), 1091 (1992).

50. R. E. Newnham, K. R. Udayakumar and S. Trolier-McKinstry, "Size Effects inFerroelectric Thin Films," Chemical Processing of Advanced Materials, Edited byLarry L. Hench and Jon K. West, John Wiley and Sons, Inc. (1992).

51. S. Trolier-McKinstry, H. Hu, S. B. Krupanidhi, P. Chindaudom, K. Vedam andR. E. Newnham, "Spectroscopic Ellipsometry Studies on Ion Beam Sputter DepositedPb(Zr, Ti)03 Films on Sapphire and Pt-Coated Silicon Substrates."

5 ONR 1992 Annual Report

PROCESSING- STUDIES

APPENDIX 38-

Processing of Ferroelectric and Related Materials:A Review

Thomas R. Shroutt and Scott L. Swarxz*tMaterials Research Laboratory

The Pennsylvania State UniversityUniversity Park. PA 16802*Battelle Memonal Institute

Columbus, OH 43201

ABSTRACT Overall, however, in the field of ferroelectnc polvcrystallincerami;co, No new materials are foreseen that will provide

The objective of this article is to present a synopsis of revolutionary impact to new applications. Instead.recent, ongoing, and future evolutionary developments pertaining evolutionary advances in the processing of ferroelectnc ceramicsto the fabricatior.Iprocessing of ferroelectric materials, with an will lead to improvements in existing commercial applications andemphasis on polycrybtalline ceramics. The basis for the review to a gradual implementation of existing materials in newwas derived from relevant literature over the last dozen years, applications. For example, although ceramic materials such asincluding the responses from a worldwide questionnaire. This relaxor ferroelectrics have been known for some forty years,survey addressed issues such as chemical synthesis methods vs. progress in their commercialization for MLC capacitors andconventional processes and anticipated future developments in actuators has resulted from recent enhancements in theirprocessing and their impact on new applications, if any. The processing and fabrication.{ 13-16]survey participants were scientists and engineers, both academicand industrial, involved in research and cevelopment offerroelectrics and related rmaterials. The primary conclusion of this III. FABRICATION/PROCESSING OFsurvey revealed that evolutionary advancements in the processing FERROELECTRIC CERAMICSof ferroelectric ceramics will continue to impact commercialproduction over the next ten years. Fabrication technologies for all electronic ceramic materials

have the same basic process steps as presented in Figure 1,regardless of the application: powder preparation, powder

I. INTRODUCTION processing, green forming, and densification.

Ferroelectric and related materials continue to be exploited A. Powder Preparfor numerous applications, including recent concepts of "smart" or The goal in the preparation of ferroelectric powders is tointelligent sy.!tems, whereby multifunctional components are achieve a ceramic powder which, after consolidation, yields a

required.(0-4) In recent years. substantial research and product satisfyingspecifiedperformance standards. A secondary

development has been devoted to ferroelectric materials in the form goal is to produce a powder that can be consolidated/densified at

of single crystals, polymers, composites, and especially, thin lower temperatures. The most important commercialized powder

films. References are provided for these important topics, but the prep eratures T or fmprtant ceraizs powder

primary focus of this paper pertains to polycrystalline materials. preparation methods fonr feoelectric ceramics include:

The most widely used and researched ferroelectric structural types, mixinlcalcination, dicdreiheralion from solvents. meial oranic

compositions, applications, and their state of development am dpepmaration and htwrdthsrmal powe ssinga The trend in powder

presented in Table 1. Although the materials and ceramic preparation is towards powders having particle sizes less than I

fabrication methods for the ferroelectric materials described had gm and little or no agglomerates for enhanced reactivity and

their origin decades ago, this paper attempts to review uniformity. Such powder qualities allow for the development of

evolutionary advances in the processing and fabrication of fine-grained microstructures with enhanced mechanical and

ferroelectric ceramics. The data presented is based on the electrical properties. Most importantly, fine particulates are critical

compilation of responses to a questionnaire from more than 100 for the continuing miniaturization of electro-ceramic devices.

international scientists and engineers, of both academic and Examples of the four basic methodF are presented in Table 2 for

industrial backgrounds. In addition, recent processing the preparation of BaTiO3 powder. References relevant to

devwiopments and novel fabrication schemes, their future impact processing advancements, particularly in regard to multilayer

on new applications, if any, are discussed. capacitor applications. including review articles0t8,43), are given atthe end of this paper.

Specific examples of significant developments inII. MATERIALS AND APPLICATIONS mixinugcakination processes are given for the PbO-based relaxor-

As presented in Table I, the most widely used ferroelectrics based materials Pb(Mgl/3Nb2/3)03 [PMNJ in Table 3. Theare found in the perovskite family general formula ABO3, which primary limitation for the development of ferroelectric relaxors hasincludes hundreds of compositional modifications (e.g., solid been in the consistent synthesis of phase-pure perovskite powderssolution substitutions and/or dopants). In addition to ceramic and ceramics. However, thi- problem has been successfullymaterials, ferroelectric polymers. including PVF 2 and other co- addressed through an improved undrrstanding of the surfacepolymers, are well established in the market place. Perhaps the properties of the constituent oxides, crystal chemistry of PMN,most significant development comes not from monolithic materials and overall kinetics of the synthesis reaction. Specifically, the "B-but in the tailoring of physical properties using the nature of site precursor method, whereby the B-site oxides are prereacted to

composites. Through the concept of phase connectivity(5 "7), form a columbite phase, allows enhanced dispersion and favors

properties can be engineered to give orders of magnitude formation of the perovskite phase in contrast to lead-niobate

performance enhancement. Specifically, piezoelectric composites pyrochlores.are finding increasing use in applications such as ultrasonictransducers for bio-medical imaging(B"9) and towed-array Note, many of the advances in mixing'calcination havehydrophones.0) imancome through the better understanding of process-structure-

property relationships. For example, knowledge of the underlyingThe ability 4o fabricate known ferroelectric materials in thin- physiocheinical behavior of Pb-based perovskites. also including

film form offers applications such as non-volatile memories and pyrochlores and Bi-layer structures, allows for controlledDRAMs. both soon to be commercial realities. Furthermore, thin- morphological developments during calcination or "softfilm offers the potential of engineered crystallographic technology, agglomeration." whereupon high reactive materials can be readilythus allowing the exploitation of non-centrosymmetric materials prepared.such as ZnO and AIN, which in polvcrystalline form cannot bemade piezoelectrically active.(0 1.12)

Cogreciitation is a chemical method whereby insoluble Hydrothermal synthesize uses hot (above lX00C) water

compounds (e.g., hydroxides or oxalates, etc.) are precipitated under pressure to produce crystalline oxides' 62- and is widely used

from a solution of metal salts (e.g., chlorides), and the precipitated in the formation of A120 3 (Bayer Process). The major advantage

product is calcined to form the desired oxide powder. The of the hydrothermal technique is that crystalline powders of the

advantage of this technique over calcination of mixed oxides is that desired stoichiometry and phases can be prepared at temperatures

intimate mixing of the precursors (in the solution phase) leads to significantly below those required for calcination. Another

lower calcination temperatures and the preparation of high-purity advantage is that the solution phase can be used to keep the

and fine-particle-size powders, see Table 3. Limitations are that particles separated and thus minimize agglomeration. The major

the calcination step may once again result in agglomeration of fine limitation of hydrothermal processing is the need for the

particulates and the need for milling. An additional problem is that feedstocks to react in a closed system to maintain pressure and

contaminants from the coprecipitation process (e.g., chlorides, prevent boiling of the solution. currently, Sakai Chemical and

carbonates, etc.) may linger in the powder after calcination. For Cabot Corporation offer commercially available BaTiOi-basedexample. commercially available, high-purity BaTiO3 powder powders, with PZT matenals from Morgan Matroc in the late

prepared by the well-known oxalate process. whereby a Ba-Ti stages of development.chloride solution is precipitated by oxalic acid and the resultingprecipitate is calcined. However, the sintering performance of Laboratory research has shown the considerable benefits of

these high-purity BaTiO3 powders is hindered by the presence of hydrothermalty-synthesized powders, e.g. Ba:Ti homogeneity.

BaCO3. which is formed during calcination. lower sintering temperatures, etc. However, these powders

Metal organic decomlosition (MOD), often referred to as perform dramatically differently in post processing and sintering

sol-gel processing, in which metal-containing organic chemicals behavior requiring further developments for successful

(e.g. alkoxides) react with water in a non-aqueous solvent to commercialization. As an example of their unique I-havior. the

produce a metal hydroxide or hydrous oxide, or in special cases, intrinsic nature of OH-bonding in hydrothemially derved BaTiOi

an anhydrous metal oxide. Powders typically require calcination powders is thought to greatly influence densification( 6 3).

to obtain the desired phase. A major advantage of the MOD particularly when accompanied by a Bi 203-based flux, whereby

method is the control over purity and stoichiometry that can be densification could be achieved at temperatures less than

achieved with powder crystallite size on the order of 5-50 nm. 800"C.(64)

However, powder processing methods for nano-sized particulateshave not been developed to take full advantage of such fine Hybrid S riLI

powders.powders a tA wide spectrum of ferroelectric ceramic powders may also

The advantage of this technique over mixing/calcination is be produced by the hybrid process involving both chemical

exemplified by multilayer capacitors fabricated using alkoxide- synthesis and conventional powder processing steps. For

derived relaxor based dielectrics. Capacitors produced with this example, promising results have been reported for PZsT cerapFcs

powder have lower sintering temperatures and submicron grain derived from powders prepared from the conventional processing

sizes, thus allowing thinner layers and enhanced dielectric of a mixture of PbO and a hydrothermal zirconium titanate

breakdown strength. Such benefits, however, are contrasted by

expensive chemicals and processing methods. precursor.(60.65) The use of chemical synthesis methods to

Table 1. Ferroelectric Materials and Applications(172-3)

Structural Family Composition Application Development Stage

Perovskitt BaTIO3 Cajmcitos Commercialized

(BaLSr)TiO3 IR Detectors Develolnent(24)

(BaSr)TiO3 (dope' PTCR Thermistors Commenrtiabzatmi

Pb(Zr.Ti)Oi (PZT) Trnsdues CommerahzatonActuators Development(25)

Pb,La(rTi)03 (PL.ZT) E.lecius-ol: Commercauiamion

Ca-doped PbTiO3 Transducrs Development/Cmrmrclahzauon(26"2"7)

Sm-doped PbTiO3 (0yftphones)

Pb(SCTa)03 IR Detectors Development(23)

(Na.Bi)T'i 3 Trnsducers (P'- free) Resewth29)

Pb(MgKq.Nb)O3 (Relaxors) Capaitors Commerclizabon(30-313 )Actuators Deveopmei3433)Electro-optics Resarch(34.

3 5)

Ba(ZnTaI)3 Microwave Rescturs Commercializatuon

Tungsten.Bronze PbNb2O6 Tninsducr (hydrophaurs) CommercahitAuon

(SrBa)Nb206 Eiecoo-opf" Resewrlc/Devclopmeat(36 )

Bismuth-Laytr Structure BfLTi3Ot2. Bi2WO6 Tmnswduc (Wocromewc ) Commmacializaoion

Perovsklte-Layer Structure Sr 2 Nb2O7 Tr-sdUeCs R¢seruh(37)

L-Ti207 gh-te'nperfnlie)Bi2(Zn.Ni.Nb)02 Capacitors Develpn't38)/Commrcialzaoorii

39)

Comn posi tes PZT/Polymmr Transdiwl Development(9,10 )

Polymers PVF2.Co-polymeis Traisduirs Develoment/Comm¢tdization(40"42)

Miscellaneous Li2B40"&M AIP04 (Crystals) Trzsdauhr Development

ZnO films (high-frequency) Cmmsermlt zd

AIN films R•u•iahr i)

Note, Developm~ent stage may refer to limited commercialI inmroucuon stpecffc to cerwn geograph"re.l gions.

unmiormly distribute dopants to conventionally prepared powders Information on tap and pour density, partcle size distribution.of BaTiO3 for capacitors is another way to combine the specific surface areas, and chermcal and phase analysis are crncalperformance advantages of chemical synthesis and cost Uniaxial compaction behavior, in particular, is easily measuredeffectiveness of conventional processing.(6 1) and provides data on the nature of the agglomerates in a powder.

B. Powder Processing, Milling is required for most powders, either to reduce

A basic guideline of powder manufacturing is to do as little particle size or to aid in the mixing of component powders.

processing as possible to achieve the targeted performance Commonly employed types of comminution include ball milling.

standards. Ceramic powder fabrication is an iterative process vibratory, attrition, and jet milling, each possessing its own

during which undesirable contaminants and defects can enter into advantages and disadvantages. For example, ball milling is well

the materials at any stage. Therefore, it is best to keep the powder suited for mixing, but not for comminution, unless varying media

processing scheme as simple as possible to maintain flexibility, sizes and long milling times are used. Vibratory milling is well

Uncontrollable factors such as changes in characteristics of as- suited for comminution, but extreme care in dispersionlrheological

received powders must be accommodated to achieve behavior must be considered, whereas attrition milling, though

reproducibility in the processing from batch to batch of material, very effective, is generally more expensive. Example of advances

Keeping the processing simple is not always possible for in terms of powder processing are given in Table 4. As presented,

ferroelectric ceramics, based simply on their complex nature, e.g. attrition milling allows for the preparation of extremely fine

the need for precise stoichiometry control and dopants. particulates, generally are achievable by chemical synthesismethods. Along with the introduction of high performance milling

A fundamental requirement in powder processing and media, e.g. partially stabilized zirconia (PSZ), minimizingperhaps key to the continued performance enhancement is contamination, attrition milling is finding growing usage for thecharacterization of the as-received or synthesized powders.(66) processing of electronic ceramics.

Figure 1. Schematic of iterative processing of ferroelectric ceramics and key characterization methods. "Keyto performance and reliability lies in the understanding of pre-ursor-process-structure propertyrelationships."

"* mixing/calcination Powder Processing Uniaxial pressing

"* coprecipitation - comminution Isostmti pressing* metal organic __ * powder flowability • E injection molding

decomposition * dopant distribution Multilayering

Characterization - dispersability * rheological behavior"* chemical and phase purity - impurities * organic interactions

"* particulate morphology * segregation - binder burnout"* size/distribution - tap density * green density"* agglomeration * compaction behavior * green strength"* surface area

"* microstructure Densification"* density, % theo - pressureless sintering"• flaw distribution • HIP, HUP• interface reactions 0- 0, Air, P02,"* electrical and mechanical reducing

properties

Table 2. Methods Used to Prepare BaTiO 3 Electronic Ctramic Powders

Method Reaction Particle Size

Mixing/calcination BaCO3 + TiO2_T BaTiO3 +CO 2 T I Pim to 100 pm

Copreciptaon Ba2+ + TiO2+ +C 202 BaTi(C204)24H2) AT 0.5 pim (after calcining

o4 + - and milling)

H20 pH

Hydrothermal Ba(OH)2 + Ti(O-PH + -4 BaTiO 3 + 2H 20 Nanosize to 5 pmAT, P

H2o AT 5.0-50.0 nm.

Metal organic Ba(OR)2 + Ti(OR), + sli- t BaTi(OR) A BaTiO3 + 6R (crystallite size)(Alkoxide) [aggomerte size

larger)

Progress in understanding the surface chemistry of magnetic tape fabrication technology. whereupon thin sheets can becomponent and reacted materials has allowed the wide spread formed >600 ftJmin., while being simultaneously electroded.T)iusage of aqueous processing as well as the ability to disperse fine Recent developments in the fabrication of piezoelectric-particulates through surface passivation.( 67 ) With growing polymer composites. primarily for bio-medical ultrasound andenvironmental concerns related to the use and/or disposal of towed array transducers, are given in Table 7. In addition tonazardous solvents, aqueous processing will continue to become achieving fine-scale composites, current emphasis lies in the abilityimportant in the future, to economically fabricate large areas (>meter-square).

"With the advances presented above and the overall controlof incoming raw materials, e.g morphology, purity, etc., high D. Densificationperformance ceramics are generally achievable without theimplementation of expensive chemical methodologies, making the Densification of ferroelect-ic materials generally requiresmixing/calcination process the method of choice for most high temperature and atmosphere control (02, PbO, vacuum, etc.) toferroelectric materials well into the future," However, continued minimize the porosity in consolidated ceramics. Heat-controlleddevelopment of powder processing methods for fine-particle size cycles are critical to microstructural development and grain sizepowders will allow future commercialization of chemical methods control. Techniques such as fast firing and rate controlled sintenneif the cost of their advanced powders can be reduced, have been utilized to inhibit or eliminate undesirable sinterng

mechanisms. Hot isostatic pressing (HIP). which employ, aC. GreenFori, gaseous pressure at high temperature, has found morecommercialization in contrast to hot uniaxial pressing, which is

Green forming is one of the most critical steps in the limited to relatively simple shapes. The HIP process has beenfabrication of ferroelectric ceramics. The choice of green forming shown to greatly enhance the dielectric breakdown strength (DBS)technique depends on the ultimate geometry required for a specific of multilayer ceramics and actuators and can be used to prepareapplication. There are many different ways to form green ceramics, transparent materials, including PLZTs and PMN. Advancements inseveral of which are summarized in Table 5. Perhaps, of all the HIP processing of PbO-based ferroelectrics have also been mademethods, the most significant advances in processing have been in with the introduction of oxygen atmosphere compatibie s' stems (seethe realm of multilayer fabrication, which includes: Table 8).piezoelectric/electrostrictive capacitors (>50 billion units/year),piezoelectric/electrostrictive actuators, and varistors, as well as Work continues to find economical densification processesseveral in non-ferroelectric applications (e.g. ceramic by which macro-defect free ceramics with near theoretical densitiespackaging).(69 30) can be achieved, allowing one to approach the maximum properties

allowable in ferroelectric ceramics. Much of the progress in thisTable 6 summarizes recent developments in MLC direction, however, will be made through advances in the powder

fabrication. Naturally, the enhanced performance arises from the synthesis, processing, and forming methods.corresponding development in binders, dispersants, and overallorganics and their role in sheet formation. In addition, ultra-thinMLCs have led to the need for correspondingly thin metallization.Of particular significance in the fabrication of MLCs is that based on

Table 3. Advances in Synthesis of Ferroelectric Materials

Advancement Material Demonstrated Benefits

Conventional (Mixing/Calcination) Powder Synthesis

Pre-reaction of B-site precursors Pb(MgNb)03 Improved perovskite phase purity(")(Columbite method)

Modification of surface chemistry Pb(Mg.Nb)03 Improved perovskite phase purity(14 )

(pH control) to optimize dispersion

Crystal chemical engineering Pb(Zn,Nb)03 Improved perovskite phase purity(45,46)(doping with BaTiO 3, SrTiO3)Reactive calcination (optimization PMN, PZT Improved sintering reactivity(47)of calcination conditions

Advanced Powder Synfhesis Methods

Alkoxide Synthesis Pb-based relaxors Fine-particle-size powders( 31,49)Thin-layer MLCs

Pechiney Method BaTiO3 Fine-particle-size powders(SO5-1. 52)(Citrate decomposition) Ba,SrTiO3,PZT High purity

Oxalate Coprecipitation BaTiO 3 Fine-particle-size powders(53-54)High purity, stoichiometry

Hydrothermal Synthesis BaTiO 3, PZT Fine-particle-size powders(55.5 6.57.58.59)Lower sintering temperatures

Hybrid MethodsPbO + hydrothermal (ZrTt)04 PZT Improved compositional uniformty(60)

Fine grain size

Chemical methods for dopant BaTiO 3 Uniform dopant incorporation(6t)addition "nanoheterogeneity" MX"R dielectrics)

Table 4. Advances in Powder Processing IV. SUMMARY

Concept BThe primary conclusion of this review is that the fabricationC tBenefit of ferroclectric ceramic materials will continue to see evolutionary

" H e gadvances, primarily in the areas of synthesis and processing. TheHigh energyrnil. • * Submicron powders(47 ,68 ) implementation of recent improvements in processing methods fore.g. attrition * Dispersion(t 4) conventional powders (e.g. attrition milling, dispersion, etc.) willPbO+MgO+Nb2O5-iPMN extend the pix.rformance of ferroelectric ceramics. However. there

will be a need for advanced powder synthesis methods and"* PSZ media * Minimal contamination associated handling and consolidation procedures for the high-

performance end of certain applications, e.g. cryogenic actuators"* Aqueous processing * Non-toxic solvent for space. Without revolutionary advances, the primary focus of

- Low cost development will be on reducing cost of advanced powdersynthesis methods, such as hydrothermal synthesis and"* Surface powder chemistry - Rheological control coprecipitation. If these cost issues can be addressed, advanced

understanding (dispersability) synthesis methods will significantly enhance existing materialse.g. "Passivation - BaTiO 3" (e.g., BaTiQ3-based dielectric ceramics and PZT-based

piezoelectric ceramics). Although the discovery of newferroelectric materials is not anticipated, continued improvementsin the processing of emerging ferroelecu'ic ceramics, such as PbO-based relaxors, will lead to their increased use in existingapplications (capacitors and actuators), and enable theirdevelopment for emerging applications (e.g., E-field tunabletransducers for sonar and bio-medical ultrasound).( 97)

Table 5. Green Forming Procedures for Ferroelectric Ceramics

Green Forming Geometries ApplicationsMet hod

Uniaxial Pressing Disks, toroids, plates Disk capacitors, piezo transducers,igniters, inch-worm actuators

Cold Isostatic Pressing Complex and simple High-frequency ultrasonictransducers

Colloidal Casting Complex shapes Transducers

Extrusion Thin sheets (>80 pzm) rods, Igniterstubes, honeycombs, substrates PTC thermister heaters

Injection Molding Complex shapes PZT-composites

Multilayer Fabrication Thin sheets/multicomponent Capacitors, actuatorsVaristors

Table 6. Advances in MLC Manufacturing

Advances Benefits Development Stage

Ultra-thin MLCs • Cap. Vol. Eff. Comncialized(3 1 ,72), %10La • High energy storage(73)

(replace tantalum &electrolytics)

S 6;s Research/Development( 74)

* Fine Metallizaion DevelopmentC75)

* High Speed Fabrication • Low cost Development(71)(>600 ft/min.)

* Low-Fire - Low cost Com iali,-,d- Ag:Pd electrodes

* Material/Dielectrics • Cap. Vol. Eff. Development(76)/Commercializcd

--High K >5000 X7R

-Relaxors, e.g. PMN-based * Cap. Vol. Eff. Commercialized

-Varistors (ZnO) • Surface Mount Chip CommercializedIntegration

-PZTs, PMN * Actuators Comnmnerialized

V. SURVEY RESULTS Areas identified as seeing significant recent developments

Prior to conducting this review, a worldwide survey was included hot-pressing and HIP techniques and piezoelecmc

conducted of scientists and engineers involved in ferroelectric composites. wherasresearch topics identified as important

materials research, from both academic and industrial to future applications include thick film processing fororganizations. This survey addressed significant recent and mulocomponent packaging, assembLk:e of nano-composites

anticipated developments of ferroelectric materials, and the for electro-optlcs and ferro-fluids. relaxor ferroelecmcs.question of whether conventional fabrication methods for smart materials, and optoelectronic materials.ferroelectric materials will be sufficient for future applicationsrequirements. Responses were categorized by geographical region ACKNOWLEDGEMENTS(North America, Europe, and the Far East). Results of this survey The authors wish to thank all those who took time out fromare summarized below: their busy schedules to answer our questionnaire. The following is

a partial list of industries, universities, and government laboratoriesThe field of ferroelect~c films, although outside the scope of fr'om which the responses came:this review, was recognized as an area where significantrecent developments were achieved, and where an evenlarger number of future developments and applications wereanticipated. The importance of ferroelectric films was Toshiba, Sakai Chemical. Denka, NEC, Sumitomo Metals.recognized by the largest percentage of respondents, Matsushita Electric, Murata. Nippon Soda. Mitsubishi Mining and

regardless of geographical region. Cement (Materials), Kyocera, Hitachi. TDK. AT&T, DuPont.

SMultilayer ceramics, both actuators and capacttors. were Battelle. Martin Marietta, BM Hi-Tech. Piezo Systems, Alcoa. MRA

identiayed easmanppicatibonh areatweresigniicanitos wreen Laboratories, Centre Engineering, Acoustic Imaging. Krautkrameridentified as an application area where significant recent Branson. Hewlett Packard, Alliant Cerarmcs, Cleveland Crystals,developments, particularly in terms of ultra-thin layers and ITEK Optical, IBM, Vitramon. Army (Fort Monmouth). Siemens.the incorporation of relaxor ferromlectrics,.havc been made, Philips, GEC Marconi, ferroperm, CNET/France Telecom. ISMRAespecially by respondents from the Far East Lab CRTSMAT, Toshiba Tungalloy. Morgan Matroc

Advances in conventional processing was cited as an area of Shonan Institute of Tech. National Defense Academy.significant recent achievement, although there was little Science University at Tokyo, Nagoya, Kyushu Institute of Tokyo.confidence that conventional processing would see additional Penn State. Texas A&M. Office of Naval Research. Floridaadvances in the future. Chemical synthesis of ceramic University. Montana State. Naval Research Lab, Institute ofpowders, by sol-gel and coprecipitation methods, was Physics-Rostov, Ben-Curion University, University desrecognized as important to future development of existing Saarlandes. Ecole Polytechnique de Lausanne, National Institute inand emerging applications, but strives must be made to Inorganic Materials, Hachinohe Institute of Technology. Institute ofachieve economical feasibility, perhaps being achievable Crystallography (Moscow), Indian Institute of Science, Osakathrough combined methodologies, i.e. hybrid processes. University

A special thanks to JoAnn Mantz for her help in puttingtogether the questionnaire and this manuscript.

Table 7. Advances in the Fabrication of PiezoPolymer Composites

Advances Benefits Development Stage

Extrusion <100 micron PZT rods Development(T":.10 micron Research(e8)

Pick and Place (Weaving) Large Area (> meter-squared) Deveiopment< 9 )

Lost Mould Method 10 g. -100 scale Developmente 80.82 )

Injection Molding Low cost Developmene83)

Fill and Dice Simplistic Conmercialized

Table 8. Advances in the Densification of Ferroelectric Materials

Advances Benefit Development Stage

"* Pressureless sintering - Transparent PLZT Developtent(S.484)-Vacuurriatmosphere control - Lower Firing Temp PZTs Research(86)-Rate-controlled densification

"• Hot uniaxial pressing (HUP) - PZTs (pymelectric -10 g wafers Commercialized* Transparent PMN, SBN, Research(87s)

"* Hot isostatic pressing (HIP) • Complex shapes Research/Development( 89.9 1)• Multilayer capacitors- fatigue reduction

• Multilayer actuators Comu(8ercialized08 9 1 )* Transparent PMN, PLZT

"* Hot forging Grain orientation Research(293 7.9 2)

e.g. B4 T13012

"• Mixed sintering - Flat tT.C.C. MLCs Research(9 314)

"* Fluxes (liquid phase) • Low firing temps Research!D)evelopment0 5,96)e.g. lithium oxide & fluorides

tTemperamture coefficient of capacitors.

REFERENCES 22. R.C. Pohanka.P.L. Smith, and G.H. Haerling. "Recent Advances inPiezoelectrc Ceramics." in Electronic Ceramics: Pioeprues- [kvjcecs

1. R.E. Newnham and G. R. Ruschau, "Smart Electrocerarnics." 3A . and Aplication, ed. by L.M. Levinson, Marcel Dekker, Inc.. NewCer. Soc.. 74, pp. 463-479 (1991). York, pp. 45-145 (1988).

2. R.E. Newnham, "The Golden Age of Electroceramics," Adv. Ceram 23. Ilelmut Thomann. "Piezoelectrc Ceramics.' Advanced Materials, 2,Mater., 3. 12-16 (1988). pp. 458-463 (1990).

3. H. Yanagida, "Intelligent Materials--A New Fronuer," Anfgew. Chem.. 24. B.M. Kulwicki, A. Amin, H.R. Beratan and C.M. Hanscon.100, 1443-46 (1988). "Pyroelectric Imaging." (to be published, Proc. 8th Int'l Symp. on

Application of Ferroelectrics (1992).4, C.A. Rogers, "From the Editor," Journal or Intellieent Material

Systems and Structres, 1,3 (1990). 25. S. Takahashi, "Recent Developments in Multilayer PiezoelectricCeramic Actuators and Their Applications," Ferroe-lectr r. 91, 293

5. R.E. Newnham, "Composite Electroceramio," 1. Mater. Educ., 7, 605- (1989).51(1985).

26. Y. Yamashita. et al.. "(PbCa)((CoI/2W 1/2),Ti)O3 Piezoelectric6. R.E. Newnham. et at., "Connectivity and Piezoelectric-Pyroelectric Ceramics and Their Applications," Jnn J. AnnIl Phys., Vol. 20.

Composites," Mater. Res, Bull., Vol. 13, No. 5, pp. 525-536, 1978. Suppl. 20-4, pp. 183-187 (1981).

7. H. Banno and S. Saito, "Piezoelectric an" Dielectric Properties of 27. H. Takeuchi, et al_ "Electromrchanical Properties ofComposites of Synthetic Rubber and PbTiO 3 or PZT," Jpn. 1. Anlo. (PbLn)(Ti.Mn)03 Ceramics." 1 Acoustic Soc. A•e., Vol. 72, No.PF'yj., Vol. 22, Suppl. 22-2, pp. 67-69 (1983). 4, pp. 1114-1120 (1982).

8. H. Masuzawa, Y. Ito.. Nakaya. et al., "Electrostrictive Materials for 28 R.W. Whatmore; P.C. Osbond. N.M. Shorrocks, "FerroelectricUltrasonic Probes in the Pb(Mgi/ 3Nb213)03-PbTiO3 System." Materials for Thermal IR Detectors." Ferroelectrics, Vol. 76, pp. 351Jamanese Journal of Annlied Physics, Vol. 28. Supplement 28-2, pp. (1987).101-104 (1989). 29. T. Takenaka and K. Sakata. "Grain Oriented and Mn-Doped

9. Wallace A. Smith, "New Opportunities in Ultrasonic Transducers (NaBi)(1.x)/2CaxBi4Ti405 Ceramics for Piezo- and PyrosensorEmerging From Innovations in Piezoelectric Materials." 1 Materials," Sensors and Materials 1 pp. 35-46 (1988).SPIE Int'l Svmno New Developments in Ultrasonic Transducers andTransducer Systems, July 1992. 30. T.R. Shrout and J.P. Dougherty, "A World Review on Lead Based

(Pb(BIB2)0 3 ) Relaxors Versus BaTiO3 Dielecmcs for Multilayer10. R.Y. Ting, "Evaluation of New Piezoelectric composite Materials for Ceramic Capacitors," Proc. Symp. on Ceramic Dielectrics, H.C. Ling

Hydrophone Applications," Ferroelectrics, Vol. 67, pp. 143-157 (ed.). Am. Ceram. Soc. (1989).(1986). 31. Masatomo Yonezawa, Kazuaki Utsurni A sushi Cichi. and Toru Mor.

11. N.D. Patel and P.S. Nicholson. "High Frequency, High Temperature "Research and Development of Relaxor Ceramics at NEC, 7Ultrasonic Transducers,- NTT International 23 (5) pp. 262-266 (1990). IEEE Int'l Sv'ns on Apnlications of Ferroelectrics. pp. 159-164

(1990).12. Y. Ito, K. Kushida, H. Kanda, H. Takeuchi, and K. Sugawara, "Thin-

Film AIN Ultrasonic Transducer Arrays for Operation at 100 MHi," 32. R. Aldrich, "Requirements for Piezoelectric Materials for DeformableFerroelectrics in press (1992). Presented at the Second European Mirrors." Ferroclactrics 27, 19-25 (1986).Conference on the Application of Polar Dielectrics. London. UnitedKingdom, 12-15 April 1992. 33. K. Uchino, "Electrostrictive Actuators: Materials and Applications,"

Am. Cenm Soc Bull- Vol. 65, No. 4. pp. 647-652 (1986).13. T.R. Shrout and A. Halliyal, "Preparation of Lead-Based Fenroelectric

Relaxors for Capacitors,: Am Ceram Soc Bull.., Vol. 66, No. 4. pp. 34. N.C. Kim, D.A. McHenry, SJ. Jang. and T.R. Shrouu. "Fabrication of704-711 (1987). Optically Transparent Lanthanum Modified Pb(Mgj/ 3 Nb2/)O3 Using

Hot lanstatic Pressing." J Amer Ceram. Sc., 73. 923-928 (1990).14. T.R. Shrout, et al., "Enhanced Processing of Perovskite

Pb(Mgl/3Nb2/3)03 Relaxors through Understanding of the Surface 35. D.A. McHenry, J.R. Giniewicz. SJ. Jang, T.R. Shrout. and A.S.Chemistry of the Component Powders," in Ceramic Powder Science- Bhalla. "Optical and Electro-Optical Properties of Lead MagnesiumCeramic Transactions, Vol. 1, ed. by G.L. Mesing, et al., American Niobate-Lead Titanate," e 107.45-46 (1990).Ceramic Society. Weserilt, OH, pp. 5 19-527 (1988). 36. S. Musikant, Optical Materials An Introduction to Selection and

15. M. Lejeune and J.P. Boilot. "Low-Firing Dielectrics Based on Lead AD0hcadons MarceJ-Dekker. Inc., New York. 1985.Magnesium Niobate, Res-B , Vol. 20. pp. 493-499 (1985). 37. P.A. Fuierer and R.E. Newnham. "La2Ti207 Ceramics." LAm-

16. Joseph P. Dougherty, "Fifth U.S.: Japan Workshop on Dielectric and Cer-. Soc 75 1111 pp. 2876-41 (1991).Piezoelectric Materials." Office of Naval Research RenorL N66005(1990) 38. MF. Yan, H.C. Ling. and W.W. Rhodes," Low-Firing, Temperature-

stable Dielectric Compositions Based on Bismuth Nickel Zinc17. T. Shiosaki. "Recent Developments in Piezoelectric Materials," Niobates," I- Am Ceram Sc.. 73, 1106-117 (1990).

Femlegtrics, 91. pp. 39-51 (1989). 39. Xao Yi: private communication.

18. S.L. Swanz, 'iTopics in Electronic Ceramics," IE rn lelns.l. Swart opDielecics 25, 935-987 (Oct. 1990) 40. T.T. Wang, J.M. Herbert, and A.M. Glass, editors, The Applications ofInsulF Digest on Polmers Blackie. Glasgow, United Kingdom (198n).

19. L.L. Hench and J.K, West (Eds.), Princinles of Electronic Ceramics 4John Wiley & Sons. Inc.. New York. 1990. 41. H. Ohigashsi. K. Koga, M. Suzuki. 1'. Nakanishi. K. Kimura, and N.Hashimoto, "Piezoelectric and Ferroelectric Properties of P(VDF-TrFE)

20. J.M. Herbert. ert, olectric Trnsducers andSensa.Gondon and Brech Copolymers and Their Application to Ultrasonic Transducers."

Science Publishers, Inc., New York (1982). Errvieletrics 60,263-276 (1984).

21. G.H. Haerding, "Piezoelectric and Electrooptic Ceramics." in Ceramic 42. T. FurukawaL "Piezoelectricity and Pyroelectricity in Polymers," MEEMaterials for Electronics' Processing- Proertie&s and Aiicatiot, ed. Transactions on Electrical Insulation 24. 375-394 (1989).

by R.C. Buchanan. Marcel-Dekker. Inc., New York. pp. 139-225(1986). 43. J.H. Adair, et al.. "A Review of the Processing of Electronic Ceramics

with an Emphasis on Multilayer Capacito Fabrication," L. Matr. E.Vol. 9. No. 1-2, pp. 71-118 (1987).

44. L. Swartz and T.R Shrowt. "Fabrication of Perovskite Lead 63. D. Henningi and S. Schrememacher. "Charactenzation of Hydrothermal'lagnesium NjobatL Res Bul. 17, 1245.50 (1982). Barium Titanatc."i ]_m CerM Six., 9.41-46 (1992).45. A. Hailiyal, et al., "Stabilization of Perovikite Phase and Dielectric 64. U. Kumar, S.F. Wang, and J.P Doughery. "Preparauon of Dens

Properties of Ceramics in the Pb(Znl/3Nb2/3)03.BaTiO3 System." U.uarF . Wan and J.P. Dougerti. P epar an enseAm. Ceram Soc Bull., Vol. 66. pp. 571-676 (1987). Ultra-Fie Gran Barium Titanale-Based Cemics," (to be publihed).65. T. Takagi. K. Anetami. and K. Shimizo. "Lead-containing Oxide

46. 0. Furukawa. Y. Yamashita. NI. Harata. T. Takahashi. and K. lnagaki, Powder." U.S. Patent 44,812,426 (1989)."Dielectric Properties of Modified Lead Zinc Niobate Ceramic," IU a . (Appl hs., 24. 96-99 (1985). 66. K.K. Vema and A. Roberts in GY. Onoda, Jr., and L.L. Hench. eds.,

Ceramic Processing Before Finne, John Wiley & Sons. Inc.. New47. T.R. Shrout. P. Papet. S. Kim, and G.S. Lee. "Conventionally York, pp. 391407(1978).Prepared Submicron Lead-Based Perovskite by Reactive Calcinauon,"Journal of the American Ceramic Society 73, 1862-1867 (1990). 67. James H. Adair and Thomas R. Shrout. "Surface Passi"=tion ofPerovskite Compounds in AqueOus Suspension," (to be patented).

48. S. Kim. G.S. Lee, S. Venkatarimani. and TAR. Shrout, "Fabrication ofFine Grain Piezoelectric Cer.tmics Using Reactive Calcination and 68. P. Papet. J.P. Dougheny. and T.R. Shrout, "Particle and Grain SizeMilling," Journal of Materials Science 26. 1141 (1991). Effects on the Dielectric Behavior of the Relaxor Ferroelectric

49. F. Chaput and J.P. Boilot. "Alkoxide-Hydroxide Route to Synthesize Pb(MglsNbzl)03." Lhlat Re., 5. 1-8 (1990).Barium Titanate Based Powders," J. Am. Ceram- Soc., 73, p. 942 69. K. Utsumi. Y. Shimada. and H. Takamizawa, "Monolithic(1990). Multicomponent Ceramic (MMC) Substrate." in K.A. Jackson, R.C.

Pohanka. D.R. Uhlmann, and D.R. Ulrich, eds., Electronic Packaging50. . Mulder, "Preparation of BaTiO and Other Ceramic Powder by Science Materials Rearch Society, Pittsburgh, PA, pp. 15-Coprecipitation of Citrates from Alcohol," Am. Cer, Soc. Bull., 49 26(1986).[111.990-993 (1970).

70. H.T. Sawhill and co-workers, "Low Temperature Co-Firable Ceramics51. P.C. Osbond, N.1 Payne. N.M. Shorrocks. R.W. Whatmore. and R. W. with Co-Fired Resistors." International Solciety of HybridAinger, "Dielectric and Microstructural Properties of Barium Strontium Microelectronics Proceedingr', p. 4 73480 (1986).Titanate Ceramics Prepared form Citrate Precursors," IntjSvmn. on Apnliations of Ferroelectrics .pp. 748-51 (1986). 71. Joe Rainwater, Vistatech Corporation, Dallas, Texas (private

communication).52. Maria A. Zaghete. Carlos 0. Poira Santos. Josd A. Varela. Elson 72. Detler Hennings, Mareike Klee. and Rainer Waser, "AdvancedLongo. and Yvonne P. Mascarenhas. "Phase Characterization of Lead Dielectcs: Bulk Ceramics and Thin Films, Advanced Materials, 3Zirconate Titanate Obtained from Organic Solutions of Citrates." J p. 334-340 (1990).Ame- Cra, Soc., 7S [8) 2088-93 (1992).

73. G.R. Love. "Energy Storage in Ceramic Dielectrics." Lhajaxa..•tI53. K. Osseo-Asare, FPJ. Arriagada. and I.H Adair, "Solubility 73 . 73 3Lv -e2 r (1989).

Relationships in the Coprecipitation Synthesis of Barium Titanate:Heterogeneous Equilibria in the Ba-Ti-C2 0 4 -H20 System," in G. L. 74. G.H. Maher, "MLC Capacitors with 6 Microns Active Dielectric andMessing, E.R. Fuller. Jr., and Hans Hausin, eds.. Ceramic Powder X7R Electrical Characteristics," presented at the 91st Am. Cer. Soc.Science, Vol. 2, 1987, pp. 47-53. Meeting. Indianapolis. IN (1988).

54. 3.M. Criado. F.J. Gotor. C. Real. F. Jimenez, S. Ramos. and 3. Del. '75. A. Buret ,"MC Tech. Co. (private communication).Cerro. "Application of the Constant Rate Thermal Analysis Techniqueto the Microstructure Control of BaTiO3 Yielded From Coprecipitaxed 76. T.T. Srinivasan. Ferro Corp. (private communication).Oxalate," Ferroelectrics. 115. pp. 4348 (1991).

77. H. Park, CPS, Inc. (private communication).55. WJ. Dawson and S.L. Swartz, "Process for Producing Sub-Micron

Ceramic Powders of Perovskite Compounds with Controlled 78. Shoko Yoshikawa, The Pennsylvania State University (privateStoichiomeury and Particle Size," U.S. Patent 5,112,433, 12 May communication).1992.

79. Dean Bathol. Fiber Materials, Inc. (private communication).56. W.I. Dawson, 'Hydrothermal Synthesis of Advanced CeramicPowders." Am. Ceram. Soc. Bull., Vol. 67, No. 10, pp. 1673-1678 80. K. Lubitz. A. Wolff, G. Preu. R. Stoll, and B. Schulmeyer. "New(1988). Piezoelectric Composites for Ultrasonic Transducers," Fttoelectrics in

press (1992). Presented at the Second European Conference of the57. K. Abe. et al.. "Process for Producing a Composition which Includes Application of Polar Dielecmcs, London, United Kingdom. 12-i5 AprilPerovskite Compounds." U.S. Patent #4.643.984 (1987). 1992.

81. Ulrich Bast. Dieter Cramer, and Andrew Wolff, "A New Technique forthe Production of Piezoelecuic Composites with 1-3 Connectivity,"U.S. Patent #4.832,939 (1989). Ceramics Today-Tomerrow's Ceramics; Materials Science Monographs,66C. Ed.. P. Vincenzini, Elsevier Publishing, Proc. of the 7th tr'l59. K. Fukai, K. Hidalka. M. Aoko, and K. Abe. "Preparation and Meeting on Modern Ceramicit Tech., Montecatini Fermi, Italy, 2 JuneProperties of Uniform Fine Perovskite Powders by Hydrothermal (1990)

Synthesis," Ceramics InteCMati•nWl, 16 pp. 285-290 (1990).

82, G. Preu. A. Wolff. D.Creamer, and U. Bast. "Microstructuring of60. Takashi Yamamoto. "Optimum Preparation Methods for Piezoelectric Piezoelectric Ceramic," Z (European Ceramic Soc.). Augsburg.Ceramics and Their Evaluation." CeramicRO., 71. pp. 978-985 Germany (Sept. 1991).(1992).

83. L. Bowen, Materials Systems, Inc. (private communication).61. D.Swanson, S.A. Bruno. I. Burn, K. Sasaki, arnd H.E. Rergna,"Advanced Dielectric Powders for Improved Capacitor Reliability," 84. K. Nagata. et ial.. "Vacuum Sintering of Transparent Piezo-Ceramics."Proc. Multilaver Ceramic Reliability, 68-86, The Pennsylvania State C,,nL-JI., VoL 3, No. 2. pp. 53-56 (1977).University (1991).

85. 0.•. Snow, "Fabrication of Transparent Electrooptic PLZT Ceramics62. E.P. Stambaugh and J.F. Miller, "Hydrothermal Precipitation of High by Atmosphere Sintering." 1.A m ., Vol. 56. No. 2, pp.Quality Inorganic Oxides," in S. Somiya. ed.. Proceedingt of First 91-96 (1973).International Svmnosium on Hydrothermal Reactions- GakujutsuBunken Fukyu.kai (c/o Tokyo Institte of Technology), Tokyu, ;apan, 86. K. Lubitz, H. Hellebrand. D. Cramer. and I. Probst, "Low Sinte7ingpp. 859-872 (1983). PZT for Multilayer Actuators." r Augsburg, Germany

(Sept. 1991).

87. Nobuko. S. Van Damme, Audrey E. Sutherland. Lon Jones, KeithBridges. and Stephcn R. Winzer. "Fabrication of Optically Transparentand Electrooptic Strontium Barium Niobate Ceramics." L Ame.. - ., 74. pp. 1785-1792 (1991).

88. J.R. Giniewicz, D.A. McHenry, SJ. Jang, T.R. Shrout. A.Bhala. andF.Ainger. "Characterization of (I-x) Pb(Mg 1 3Nb 2/3)O3 4(x)PbTiO 3and Pb(Scl/ 2 Ta 1/2 )0 3 Tra-"narent Ceramics by Uniaxial Hot-Pressing," Ferroelectrics, 109, to '-172 (1990).

89. LJ. Bowen. W.A. Schulze, and J.V. Biggers. "Hot Isostatic Pressingof PZT," Powder Metallurgy In., 12 (2], 42-96 (1980).

90. M. Takata, S. Kawahara, K. Kayeyama. and S.Toyota. "The HIPTreatment of Magnetic and Piezoelectric Ceramics," Prec. of Int'l Conf.Hot hsostatic PressineALulea. 11. 399-401, 15-17 June (1987).

91. Mituhiro Takata and Keisuki Koyeyama, "The High FrequencyTransducer of HIP-Densified Piezoelectrc Ceramics." jap. J. of AI!.Phy.c, 22 Supplement 22-2. pp. 148.149 (1983).

92. H. Watanabe, T. Kimura, and T. Yamaguchi, "Particle OrientationDuring Tape Casting in the Fabrication of Grain-Oriented BismuthTitanate," I Am. Ceram- Soc., 72 [2) pp. 289-93 (1989).

93. 1. BeLsick, Y. Yamashita. and M. Harata. "Ceramic/Ceramic Compositefor Multilayer Capacitor Application," Proc IEEE 7ih Int'SvmAoljication of Ferroelecwtcs. 44-47 (1990)

94. Brahim Boufrou, Filbert Desgardin, and Bernard Raveau. "TetragonalTungsten Bronze Niobate, K0.2Sr0.4NbO3: A New Material forCapacitors with Flat Dielectric Curves," J. Am Ceram Soc.. 74, pp.2804-2814 (1991).

95. J.M. Haussanne, 0. Regreng. 1. Lostec, G. Desgardin, M. Halmi, andB Raveau, "Sintering of Various Perovskites with Lithium Sails,"Proc. 6th CIMTEC, Milan. Italy (1986).

96. G. Desgardin, I. Meg, and B. Raveau. "BaLiF3-A New Sintering Agentfor BaTiO3-Bascd Capacitors," Ceramic Bulletin. 64, pp. 564-570(1985).

97. T.R. Shrout and J. Fielding. Jr., "Relaxor Ferroelectric Materials,"Proceedings of the 1990 IEEE Ultrasonict SvmRtXium 711-720 (1990).

APPENDIX 39

Journal of Cnstal Growth 116 1(1992) 251-259 o ...... CRYSTALNorth-Holland GROWTH

Kinetics of the hydrcthermal crystallization of the perovskitelead titanate

G.A. Rossetti, Jr., D.J. Watson ', R.E. Newnham and J.H. Adair 2

Materials Research Laboratory, Ahc Penns.Yvhani State Unu ersity. Universito Park, Pennsyhania 16802. USA

Received 21 March 1991: manuscript received in final form 16 October 1991

The hydrothermal crystallization kinetics for the perovskite PbTiO5 have been investigated under autogenous conditions attemperatures in the range of 225-250"C and feedstock concentrations of (0.1-1.0 M. At these temperatures, crystalline perovskiteparticles were obtained in approximately 4-7 h. Transmission electron microscopy (TEM) of the product oxides showed nanometersized crystallites with an elongated morphology. The crystallization kinetics were monitored using X-ray powder diffraction onsamples extracted from the reaction mixture at variou, times. The crystallization rate data were analyzed according to a generalizedsolid-state kinetic treatment which, along with microstructural evidence, suggest that the PbTiO, formation reaction proceeds via adissolulion-recrystallization mechanism. It is proposed that the precursor amorphous hydrous oxides of lead and titanium dissolveand recrystallize to form the perovskite phase. The relative rates of dissolution and recrystallization were found to be stronglytemiperature dependent within the range examined. At all temperatures, the recrystallization kinetics appeared to obey azero-order rate law. An apparent activation energy of 7.2 kcal/mol was estimated for the hydrothermal PbTiO, formation reaction

1. Introduction reactions, but with correspondingly enhanced dif-fusion rates 19). The reaction mechanisms and

Hydrothermal particle synthesis involves the sequences that can lead to crystalline particletreatment of aqueous solutions or suspensions of formation therefore include dissolution or trans-precursor particles at elevated temperatures and formation of any solid precursor phase(s), diffu-pressures. The reactions occurring in hydrother- sion in solution, adsorption at the solid-liquidmall), treated solutions of inorganic compounds interface, surface diffusion, incorporation of so-can produce fine, high purity, and homogeneous lute material into the lattice, and crystal growthparticles of single and multicomponent metal ox- steps (8-12]. Unfortunately, the relatively highides under the appropriate conditions [1-81. Fur- temperatures and pressures for hydrothermalthermore, particle sizes from nanometer to cen- syntheses (e.g., 100-500°C and 0.1-14 MPa, re-timeter ranges can be synthesized depending on spectively) prohibit in most cases the use of in-situthe configuration of the hydrothermal equipment. systems to monitor the course of the reactionsHowever, the reaction sequences in hydrothermal leading to product particie formation. Conse-systems are complex, and at the present time quently, meaningful data relating to particle for-there is scant information regarding the reaction mation can at present often be obtained only bykinetics and underlying mechanisms [9]. Hy- studying the solid-state nature of the reaction.drothermal reactions are analogous to solid-state In the present work, we have evaluated the

hydrothermal formation of the binary lead tita-

1Currently with nium oxide, PbTiO. . The solution speciation and

12533, ur Aw IBM Fishkill Hopewell Junction. New York phase equilibria for this relatively complex, but2 Currently with Material Science and Engineering. Univer- technologically important, system have not been

sity of Florida. Gainesville. Florida 32611. USA. studied. However, there have been several studies

(Xt22.t124X/92/$t)5.0X) V 1992 - Elsevier Science Publishers B.V. All rights reserved

252 (;.A. Rmo.we. Jr c ti Kinc, ht of hIdrotdrmal 'rmsiaffizaiion of pwrenskize PbTiO,

verifying that lead titanate is the stable com- 2. Materials and methodspound under a wide range of hydrothermal reac-tion temperatures and pressures [2-81. Further- The hydrothermal synthesis of particulate leadmore, studies that preceded the work currently titanate employed a solution crystallization proc-being reported have demonstrated that a range ess carried out at relatively modest temperaturesfrom pH ") to pH 10 is suitable to produce stoi- and pressures (225-250°C and 1.38-5.17 MPa,

chiometric PbTiO, with the perovskite crystal respectively) [6]. The preparation of the feedstockstructure 161. The objective of this study was to materials was conducted according to theobtain kinetic data on this system to better un- flowchart shown in fig. 1. Experiments were per-derstand the particle, formation mechanisms for formed in a I liter 316 stainless steel autoclavethe complex binary oxides and, in particular, the equipped with a magnetically driven stirring unitperovskile family of materials. (Model 4521, Parr Instruments Co., Moline, IL).

LeadAetaae aTntancum lsoprot

DiWashe Wattr A1mn0te WatEth,

Miin PbTi I

NH40OH to pH 95-9.6

SFeedstock Suspension

Hydrothermally Treat(225. 235. 250 -C)

5ml Samples ExtracteIfasle Function of Time

4I

SCollect Particles by Filtration.

Wash wiith Ammoniatod Water,and Dry at 25 *C in Air

X-Ray Ditfraction

end TEM Analysis

1i- .I Fhn• chart for ili hydrothernmal synthesis of PbTiO, designed to collect crystallization data as a function of time.

G-. Rovric i. Jr. et at / KAnetcs of f•ihdron/iermal cn'staltl:zatioif (Ofpcroi S& PhilO) 253

Samples of 3 nil size were extracted underisothermal conditions at various times during thecrystallization process. The extraction of theserelatively small samples was accompanied by min-imal reactor pressure losses ( - 0.014-0.034 MPa).The solid portion was immediately separated fromthe extracted suspensions :y filtration and/orcentrifugation depending on particle size. Thesolids were then washed with dcionized waterwhose pH had been previously adjusted to pH 9.5with ammonium hydroxide, filtered again, and air 100.

dried at room temperature. It was observed inpreliminary studies that washing the powders witha solution near the pH for the minimum solubility En

of lead oxide and lead titanate was necessary to ,..

limit incongruent dissolution of the lead from the z 80.4%uJ 8.

hydrothermally treated powders. X-ray powder 2>diffraction patterns for the extracted samples iwere obtained using an automated diffractometer •employing Cu Ka radiation. The degree of crys- 37..%

tallinity of the solids was assessed by integratedintensity analysis of the (101) reflection [131. 0%Bright field transmission electron micrographs(TEM) were obtained on selected samples andused to estimate the mean particle size. Thesurface areas of the powders were determined byan automated nitrogen adsorption technique(Monosorb, Quantachrome Corp., Syosset, NY). DEGREES 20

Fig. 2. Examples of X-ray diffraction patterns showing thetypical change in crystallinity as a function of reaction time forhydrothermally treated PbTiO,. Data are for 033M feedstock

3. Results and discussion hydrothermally treated at 225"C for 0t time (tOr crystallinity),and 37.6%, 80.4%. and l(X1% crystallinity.

3. 1. Alaterials characterization

Typical X-ray powder diffraction patterns for with a relatively uniform, acicular morphology.the PbTiO, crystallization sequence as a function The influence of feedstock concentration on par-of hydrothermal reaction time are shown in fig. 2 ticle size and surface area for the PbTiO. crystal-for samples f-, m a 0.33 molar feedstock solution lized at 225°C is shown in table 1. Particle sizeat 225'C. As shown in the figure, the starting increases with feedstock cttncentration as judgedfeedstocks were amorphous and became increas- by direct observation from the TEM micrographsingly crystalline with time. Under these condi- and specific surface area measurements.tions, no change in crystallinity was detected byX-ray diffraction after hydrothermal treatment 3.2. Crystallization kineticsfor - 7 h. TEM micrographs corresponding tothese samples are shown in fig. 3. The amorphous The kinetics of PbTiO, crystallization from afeedstock particles were 20 nm in diameter, 0.33 molar feedstock suspension are shown atequiaxed, and could be clearly distinguished frore three temperatures in fig. 4. Qualitatively, thethe product PbTiO1 particles, which crystallized crystallization process may be divided into three

254 G.A Roston, Jr t af / Kiiweit v of fh.drothIcrm*i* tn'sJiah:atioI of pxron oLts" !hTiO,

(j i' lO_.m M O0_nm

Mct 4 100_nm 100 nm

Fig- 3. Examples of tinmimission electron micrographs of some typical particle samples extracted trom a 0.33M feedstoxk al 25.(as a function of reaction time: (a) 0'%. (b) 37.fil, (c) 80.41ý-. and (d) I(1)K c'mtathni1s

distinct kinetic regimes. At relatively short reac- last, a second period of cryStalhization at a Iocrtion times there is a temperature-depcndent in- rate than during the intermediate regime. Theduction time with no measurable crystallization transition between the two periods of crvstalliza-tating place, followed by an initial period ofrapid crystallization at intermediate times, and,..00

Table I

Crystallite sires and specific surface areas for the PbTiOjparticles hydrothermally synthesi'ed at 225'C to I(M)% crvs-* 225talhnity a% a function of feedstock concentration "0400 2 C

CJ DFeedstock Crystallite size Specific 2 0.200

concentration (length/diameter) surface area _

(mol/l) (nm) (m,/g) 00

LOU0 750/220 7.8 t~ .00 10000 200000,.50 560/16) 13 TIME (seconds)0.33 280/2% 22 Fig. 4 Fractional PbTiO, cr•-stallint, as a function of time for""'.0 70)/20 3.33M feedstocks hydrothermall) treated at 225. "15. and

* Estmatec from I ENI photomicrographs. 250'C

G.A. Rmsettt. Jr. et (l / Ki'tics of hVdrothernial rrystalh:.atton of peroisktie Ph Ti. 255

tion occurred at approximately 30-40% crys- where f is the fraction crystallized isothermallytallinitý for all temperatures. at time 1, r is a constant that partially depends

in order to gain further insight into the factors on nucleation frequenc:'I and rate of grain growth,controlling the hydrothermal formation of crys- and m is a constant that varies with the systemtalline PbTiO, the rate data in fig. 4 were ana- geometry. Hancock and Sharp have shown thatlyzed according to the generalized solid-state ki- for reactions obeying a single theoretical ratenetic treatment of Hancock and Sharp [141. This expression, plots of -In In(! -f) against In(t)method was originally applied to isothermal over f=-0.15-0.50 yield approximately straightsolid-state transformations such as the dehydroN- lines with slopes in having a value falling within aylation of brucitel[14] and has also been success- range characteristic of three distinct reactionfully applied to more complex heterogeneous re- mechanisms. When rn = 0.54-0.62, a diffusionaction sequences in both oxide [15] and non-oxide controlled mechanism is indicated, while a zero-[161 systcms. Care must he exercised, however, to order, first-order, or phase boundary controlledensure that a literal interpretation is not assigned mechanism is indicated for m = 1.0-1.24. Ato the rate constants or rate laws determined in mechanism involving nucleation and growth con-this way from the simple regression analyses. trol is indicated when in = 2.0-3.0. Values of inEven when precise statistical data sets are avail- lying outside the specified ranges have no obviousable, best-fit rate constants obtained f:om regres- mechanistic interpretation, but can sometimes besion analyses can be substantially in error [171. indicative of competing processes [141. The vari-Despite these reservations, careful application of ous standard solid-state reaction rate equations

simple kinetic treatments is often helpful in de- and associated values of ni are summarized inveloping a qualitative understanding of the domi- table 2. It is not possible to distinguish the mostnant processes in complex solid-state reaction appropriate rate jaw within a given group solelysystems, particularly when corroborated by mi- on the basis of the value of in. Instead, theCrostructural evidence and other data. indiidual rate laws must be tested and compared

Recognizing these limitations, and considering over the complete conversion range [201.only the solid-state nature of the transformation, In fig. 5, plots of -In In(l -f) against ln(O)a kinetic analysis was applied based on the John- over fr= 0.15-0.50 for the data in fig. 4 are pre-son-Mchl-Avrami equation [18.191: sented. For PbTiO, crystallization at 225 and

235TC, it is shown in figs. 5a and 5b that thef= I - exp( -rit), (.) kinetics are described by a two-stage rate law. In

or. in linear form, each case, the kinetics of the first stage are char-acterized by a large rn exponent (n > 5) followed

-In In(1 -f) = In(r) + n ln(i), (2) by a sharp transition at f= 0.3-0.4 to a second

Table 2Solid-siate reaction rate equations (from ref. 1141)

Function Implied mechanism Equation m

,' (f ) Dillusion controlled f 2 - k, 0.62D f(f) Diffrusion controlled (I - f) In( ] - f ) + f - kt 0.57!) •( f I)lltstI nn controlled I1 -(I - f)'1'12

- kt 0.54D, fD Diffusion controlled I - 2f/3 - (I - f)2/1 = ki(1.57

1'(f) First order - In( I - f)-= t 1.00'R,( f') Phase •oundary I -(I - f) 2

, - ki L.11R 11f) Phase boundary 1 -(I - f)'/: - kA 1.07Z'I f) Zero order f - Ot 1.24A, f(t Nucleation and growth I - In( I - f) 1t2 At 2.00Af,() Nucleation and growth l-In(I - f)]''- ki 3.00

256 (G.A Rorsent, Jr. et al. / Kinetcrs of hydrothermal cryosallizaioun of perovsikre Ph7tO

0.0 cannot be simply described by any of the ten0.5 1.22 standard solid-state reaction rate equations.

On the other hand, the in exponent of thesecond-stage crystallization in figs. 5a and 5b,

•, -1.0 along with the single in value of fig. 5c (ti 1.22,

"-.1.5 m8.83 0.80, and 1.00, respectively). suggest a reaction15 8.8mechanism best described in terms of zero-ordcr,

(..a first-order, or phase boundary controlled rate ex-7.0 7.5 8.0 8.5 9.0 9.5 pressions. Consequently, the rate expressions for

In (t) zcro-order, first-order and phase boundary con-

trolled mechanisms were tested over the second-0.0 stage crystallization tanges indicated by the ti

0.5 values of fig. 5. The first-order and phase bound-M 0.80 ary controlled rate equations gave poor fits of the

raw kinetic data when continued to completecrystallization (if= 1.0). In accordance with a

.1.5 5n=.11 zero-order rate law, however, fig. 4 shows thatthe second-stage (f> 0.3-0.4) plots of f against I

-2.0 .) lor PbTiO, crystallization at 225 and 235°C arc in7.0 7.5 8.0 8.5 9.0 9.5 fact lincar. Uig. 4 indicates that the lPbhFiO, crys-

1I1ni) I:lliz:ation kiniclics ;al 25(Ic a re allo lin;er fiur/ it t1 II 1 ($% l li,-1 11 " '" ,' t • , t h l lt

Oil lajiges. Ihicielotn. tile Zelo-olde1 liate CXilnesslIoi

-0.5 is most appropriate to describe the apparent crys-talli'alion kinetics. In fig. 4. the slight nonlinear-ity in the crystallization kinetics at 250"C lor"i 1..0

•" -1< 0.3-0.4 is contrary to the expectation from fig.

41 5c that only a single rate expression should be

Mc) obeyed. However, it is likely that the crystalliza-_........................ lion kinetics at 250"C also conform to a two-stage

"7.0 7.5 8.0 8.5 9.0 9.5 rate law, but that first-stage crystallization wasIn (1) not detected due to the rapid initial rates at the

Fig. 5. Plots based on the Johnist-MehI-Avraini analyses ofthe kinetic data From fig. 4. higher temperature.An Arrhenius plot for the hydrothermal

PbTiO3 formation reaction is given in fig, 6.stage characterized by an in value close to unity. Using a method similar to that of Culfaz and

The m exponent for the first stage dccreases Sand [21), the values plotted along the ordinate inrapidly with increasing temperature, so that for fig. 6 represent the instantaneous rate deter-crystallization at 250-C, the first stage is not obvi- mined at 50% crystallinity. This method was cho-ous and the kinetics may apparently be described sen because it makes no assumption regardingin terms of a single rate law, as shown in fig. 5c. the underlying reaction mechanisms and associ-

In figs. 5a and 5b. the in exponent of the first ated rate laws. Even in the crystallization regimestage (m = 8.83 and 5.11, respectively) reflects where the reaction is apparently isokinetic (i.e.,

the initial period of rapid crystallization in the f'> 0.4), the rate data were used in preference toearly portions of the corresponding curves of fig. rate constants in constructing the plot because

4. Comparison of these in exponents with the the error in the zero-order fits, as well as the

theoretical values presented in table 2 shows that extent to which the initial rapid crystallization at

the kinetics of the initial stage of crystallization various temperatures affects the subsequent

G6A. Ro.s.wtn, Jr- et al / Kmne:,cs of tiedrotherpaI ,rysialhzaoo of perob-kile P6Ti0, 257

-5.4 ing feedstock concentration. Such a result is notsupported by a classical nucleation and growthmodel, which would predict higher supersatura-

. Es 7.2kcal/mol tion conditions and smaller particles at higherS -S.6 - -2feedstock concentrations [ I ]

-C Alternatively, the generation of reacting-s.7 species by the process generally known as precipi-

tation from homogeneous solution (PFHS) [26--5.8 281 is often observed in systems where tempera-

ture is used to thermally decompose precursor.5.9 reactant-. 124-271. It is generally acknowledged

1.90 1.95 2.00 2.05 that a major limitation in the PFHS reactionscheme is that relatively low concentrations of

l/T x 1000 (1/K) precursor species must be used to avoid continu-Fig- 6. Arrhenius plot for the hydrothermalk crystallized ous nucleation throughout the particle formationPbTiO,. The aIpparent activation encrgp for hydrothermial process. In the current work, the microstructural

crstalfization of PbTiO, from the plot is 7 2 kcal/molt data strongly support the contention that PFHS is

taking place in the Pb-Ti-H.0 system underzero-order rate, is uncertain. The apparent acti- hydrothermal conditions. However, in this sys-vation energy obtained in this way is 7.2 kcal/mol. tern, a sparingly soluble precursor hydrous oxide

was used to generate the reacting species. Under3.3. Mechanistic interprclation these conditions, a high yield of product powder

is potentially attainable using relatively concen-The microstructural and kinetic data pre- trated precursor suspensions. Furthermore, the

sented above provide sonic insight into the mech- high concentration of feedstock is not expected toanism of the hydrothermal formation of crys- compromise the generation of nuclei as it does intalline PbTiO* . Based on the microstructural data classical PFHS because the reservoir of nutrientof fig. 3. a mechanism involving a liquid assisted stored in the solid precursor does not influencesolid-state transformation [22-241 is deemed un- solution factors such as supersaturation and ioniclikely. The crystalline material in the micrographs strength.of Fig. 3 does not appear to have grown out of Consequently, with reference to the schematicthe amorphous precursor. Furthermore, at no solubility curves shown in fig. 7. it is proposedstage in the crystallization is there any evidence that as the hydrothermal temperature is in-of partially or poorly crystalline material, as would creased, the dissolution of the precursor hydrousbe expected during thc progress of a solid-state oxides dictates the supersaturation (S,2) at whichtransformation. Similarly. the morphologies of the PbTiO1 crystal!izes. Assuming normal solubilityparticles in the micrographs of fig. 3 would not be behavior, this is consistent with the temperature-expected for a material precipitated via a classical dependent induction time observed in the kineticnucleation and growth mechanism [11,121. The data of fig. 4. When the hydrothermal tempera-particles are seen to be nearly the same size with ture is increased to the range where the solubilitysimilar acicular morphologies. A larger size distri- of the precursor hydrous oxide (S,) is greaterbution is expected if particles are precipitated than that of the anhydrous oxide (Sp,.), crystal-from a heterogeneous, locally supersaturated so- lization of the latter will take p~ace with thelutior' Ill]. In contrast, a narrow size distribution nutrient precursor material acting as a reservoiris more typical of particles precipitated from ho- for the precipitating species. If the particle growthmogeneous solution 125-281. Moreover, the parti- is not topotactic with the precursor particles, thencle size and surface area data of table I show that nucleation of the PbTiO, is required. It is be-the average particle size decreases with decreas- lieved that this corresponds to the rapid. first-

258 G.A. Rossetti, Jr. et aL / Pieties of hydrothermal crystlhzation ofpero'skare PbT1O

V ! When the nutrient is depleted to the point wherethis is no longer possible, the crystallization might

nRDC hAEtA e expected to end abruptly, with little premoni->. -- PRODUCT MATERIAL

tory diminution of the crystallization rate as 100%crystallinity is approached.

fPRECURSOR-SP

Zi 4. Summary

0 Crystalline, nanometer sized PbTiO, particles

T1 T2 were synthesized under autogenous hydrothcrrnal

TEMPERATURE conditions at temperatures in the range of 225-

Fig. 7. Reaction scheme proposed for the hydrothermal crys- 250'C and feedstock concentrations of 0.1-1.0lallifl'tion of PbTiO1 . It is proposed that the difference in molar. Under these conditions, the product parti-solubility of the precursor hydrous oxide and the product cles crystallized with a relatively uniform acicularmaterial, PbTiO. at the hydrothermal reaction temperature morphology. In contrast to expectations based onprovides the driving force or supersaturation,. S12, necessary classical nucleation and growth models, the parti-to nucleate and grow PbTiO, particles via precipitation from

homogeneous solution. cle size was found to increase at higher feedstockconcentrations. A simple solid-state analysis ofthe crystallization rate data showed that the ki-

stage crystallization. As might be expected for a netics could be characterized by three regimessuch a complex dissolution-recrystallization proc- corresponding to a temperature-dependent in-ess, the m exponents for this stage of crystalliza- duction period, an initial period of rapid crystal-tion (fig. 5) did not correspond to any of the lization, and a second period of crystallizationtheoretical values for the standard solid-state re- obeying a zero-order rate law. To account foraction rate equations (table 2). these observations, a particle formation mecha-

Once sufficient nuclei are formed, as dictated nism was proposed wherein an anhydrous oxideby the relative supersaturation at a particular leading to the perovskite phase is precipitatedtemperature, growth will commence. It is be- from a homogeneous solution, the supersatura-lieved that this corresponds to the sharp transi- tion condition of which is dictated by the solubil-tion to a second-stage crystallization at f'= 0.3- ity of a sparingly soluble amorphous hydrous ox-0.4 as observed in fig. 5. With the dissolution of a ide precursor. It is suggested that this reactionprecursor solid providing the nutrient for the scheme may be useful in preparing uniform,tilimalc cl-vallinc Ih)p:tc. tiniftitmly shaped., mono-ized particles of complex oxides from high

nearly nitotsiLcd particles aic pr-oduccd, Ilti- ciliiceilliaa lins l flutrietl .11I at high yields.

vided that the dissolution or decomposition of theprecursor material is the rate-limiting step. Thezero-order kinetic dependence of the second- Acknowledgmentstage crystallization is consistent with this re-quirement. Zero-order kinetics imply that the re- The authors would like to thank Dr. C.A.acton ateis ndeendnt f te cncetraionof Randall of the Materials Research Laboratoryaction rate is independent of the concentration of for performing the transmission electron mi-the reactants, and arc observed in systems where forcpy.the rate is controlled by a large excess of one croscopy.reactant, or is dictated by an external variable,such as the intensity of light in a photocatalyzed Referencesreaction 1291. In the mechanism proposed, crystal-lization can continue only so long as there is 11] E.P. Stambaugh and R.A. F<.)s, US Patent #3,,).771),sufficient nutrient to maintain supersaturation. 1963.

G.A. Ro.swir. Jr. et at. / Kineincs of hvdrotherinal crnstatfhiatlon of erocsAlte PbTiO, 259

121 T.R.N. Kutty and R. Balachandran. Mater. Res. Bull. 19 15] R.A. Gardner, $. Solid State Chem. 9 (1974) 336.

(1984) 1479. [16) G.A. Rosselti. Jr. and R.P. Denkewicz, Jr., J. Mater. Sci.13] M. Suzuki. S. Uedaira. HI. Masuva and H. Tamura. in: 24 (1989) 3081.

Ceramic Transactions, Ceramic Powder Science II (Ani. [171 A.-1. Weiss. Catal. Rev. 5 (1971) 283.Ceram. Soc.. 1987) p 1013. [181 M.J. Avrami. J. Chem. Phys. 7 (1939) 1103: 8(1946) 212.

[41 K. Takai. S. Shoji. H. Naito and A. Sa~aoka. in: Proc. Ist [191 B.V. Erofe'ev, Compt. Rend. Acad. Sci. URSS 52 (1946)

Intern, Symp. on Hlydrothermal Reactions. Ed. S. Somiya 511.(Gakujutsu Bunken Fukyo•kat. c/n Tokyo Inst. Technol., [201 J.t. Sharp, G.W. Brindley and B.N.N. Achar, J. Am.1982) p. 877. Ceram. Soc. 49 (1966) 379.

[5] S. Kaneko and F. Imoto. Bull. ('hem. Soc. Japan 51 121] A. Culfaz and L.B. Sand, Advan. Chem. Ser. 121 (1987)( '179) 173'). 41.

If] D).- Wtson, C.A. Randall."R.I. Newnhanm, and JJ+ 1221 -W.J. Dawson. Am. Ceram. Soc. Bull. 67 (1988) 1673.Adair. in: Ceramic "Iransactions. Ceramic Powder Sci- 123] J.I. Adair, R.P. Denkewicz. F.J. Arriagada and K. Os-ence II (Am. (cram. Soc.. 1987) p. 154. seo-Asare. in: Ceramic Trans.. Ceramic Powder Science,

[71 I. Abe. M. Aoki. I1. Rikim:oru. 1. Iho. K. Ilidaka and K. Vol. I (Am. Ceram. Soc., West,.ville. OIL 1988) p. 135.Segawa, US Patent #4'643.984. 1987. [24] E. Tani, M. Yoshimura and S. Somiya. J. Am. Ceram.

181 N.A. Ovramenko. L.I. Shvets, F.D. O'.charenko and B.-)u- Soc. 66 (1983) 11.Kornilovich. inorgan. Mater. 15 (1979) 15N)1. 1251 L. Gordan, M.L. Salutsky and H1.. Willard, Precipita-

[91 V.A. Kkiznetmiw. Kristallografiva 9 (1'J3) 123. lion from Ilomogeneous Solution (Wiley. New York,[10] A Norlund Christensen, AL . Chem. Scand. 24 (19701) 1959).

2447. [26] 11.11. Willard and N.K. Tang. J. Am. Chem. Soc. 5911)] A, F. Nielsen. Kisiclics o[ Prccipilalit•n (Perg:mnon. Ox- (1937) 1 MI.

ford. 1909'). 1271 J.T.G. Overbeek. Advan. Colloid Interface Sci, 15 (1982)[121 A.(G. Walhon. The Foi niation and I'moprlices ol Pliecipi- 25 1.

liles (Krieger, 1)979l. 1281 V.K. LaMer and R.H. Dinegar. J. Am. Chem. Soc. 72[1.31 (6R. Fox. F. l eva) anid R.L. Nc nham. J. Mater. Set. 261 (195)) 4847.

i 1991) 2566. 129] 0. Levenspiel. Chemical Reaction Engineering (Wiley,1141 J.D. hancock and 1ll. Sharp. J. Am. Ceram. Soc. 55 New York. 1972) pp. 51-53.

(1972, 74.

APPENDIX 40.

Sol-gel synthesis of Ln,(Ln = La, Nd)Ti207A. •, Prasadarao.- ulagaraj Selvaraj. Sridhar Komarnent." and Amar S. BhallatVatertaii Research Laboraiors. The Pennsýlania Stare triersirn. Lntwersir% Park. Penns•hania 16802

iRccci,,ed 4 October 1991. accepted 16 June 1992)

Lanthanum and neodymium titanates were prepared by a sol-gel route Synthesis ofLaTiO- from pure alkoxide precursors yielded an intermediate perovskite type phase.Lail, TiO. which partially transformed to LaTiO, on heating to 1500 'C. Substitutionof the lanthanum acetylacetonate for alkoxide precursor yielded La:Ti-O- without anyintermediate phase at a very low temperature of 700 *C. Sintering of the La2Ti,O- gelpowder yielded a highly dense ceramic with -97% theoretical density. Similar sintertngtreatment resulted in -92% theoretical density for Nd2TiO-.

I. INTRODUCTION particles are-desirable -for good sinterability and fine-

Interest in lanthanide titanates with the general for- grained microstructures.mula. LnTiO,. arose as a result of the observation that In recent years, the sol-gel process has become athey might be ferroelectric by analogy with Cd 2Nb,O-. method of interest for the synthesis of nanocompitteswhich is a unique and unusual ferroelectric material and electroceramics.' 6 The advantages of the sol-gelat low temperatures.' Cd2NbO 7 has a cubic structure process compared to conventional methods for pol,,crys-of the mineral pyrochlore (Na.Ca)(Nb,Ta)O 6F. Roth talline ceramics are better control of stoichiometrv andwas the first to synthesize and investigate a series of homogeneity, lower reaction temperatures. less contain-lanthanide titanates by solid-state reactions.-2 Among ination, and ease of preparation of ultrafine povders.the rare earth titanates, those with Ln = Sm to Lu thin films, and fibers for device applications. Sol-.,elare isostructural to pyrochlore whereas lanthanum and synthesis of polycrystalline or thin film electroceram-neodymium titanates are monoclinic with a space group ics is mostly based on the hydrolysis and subsequentP2 1. 34 Further investigations revealed that both lan- polycondensation reactions of component alkoxides. At-thanrum and neodymium titanates are effective ferro- tempts on a similar basis with lanthanum and titaniumelectric substances with both high Curie temperatures isopropoxides as starting materials yielded a phase ofand coercive fields.56 Their temperature stability and La;-,TiO3 instead of the required phase La:TiO-.low dielectric loss at microwave frequencies make them However, by substituting one of the precursors (seegood candidate materials for high frequency applica- below), phase pure La.Ti_, is obtained at a relativelytions. Indeed, La2Ti20-r and Nd2Ti2O7 are being used low temperature. Also, the La2T12O, precursor solutionas major components in high K microwave dielectric prepared by this method is useful for the formation offormulations.' 5 Also, La 2Ti20, has been found to have highly oriented thin films on different substrate- " Thisgood piezoelectric properties with possible use as a paper describes the first report on the sol-gel synthesishigh temperature transducer material.9'10 Nevertheless, and sintering behavior of lanthanum and neodymiumstudies on single crystals of these materials are limited titanates.because of their high melting points which make crystalgrowth difficult. The synthesis of single crystal fibers II. EXPERIMENTALof these compounds by laser heated pedestal growthhas recently been reported." Though some methods for The scheme for the synthesis of Ln.Ti;O, is out-the synthesis of LaTi2 O,7 and Nd 2Ti2.O, in the poly- lined in Fig. 1. Lanthanum/neodymium acetylacetonate

crystalline form were reported based on coprecipitation [Ln(acac)3 or LnA) and titanium isopropoxide. Ti(OPr' ,

of hydroxides.-'2' 3 thermal decomposition of nitrates,i2 (Aldrich Chemical Company, WI) were used as pre-liquid mix techniques," and hydrothermal methods,' 5 cursors and 2-methoxyethanol (MOE) was used asthese ceramics are by far fabricated only by conventional the solvent. 0.05 M of La(acac)3 was added to 2 M

solid state reactions of the oxides. However, for high MOE and dehydrated by distilling off the solvent at

quality products, pure powders consisting of uniform 120 *C. The resulting slurry was refluxed with MOEfor 24 h in argon atmosphere keeping the molar ratioof MOE/Ln(acac), around 80. The required amount ofTi (OPt")4 was then added to the above solution after

"aOn leave from Andhra University. Visakhapltnam. India. cooling. The resultant mixture was again refluxed forbIAlso with the Department of Agronomy, 12 h, cooled, and the pH was adjusted from -12.0 to

J. Mz.,. Res., Vol. 7, No. t0, Oct 1992 C 1992 Matenals Research Soaoity 2859

A V Prasadarao er ai Sol.gel synltesis of Ln2(Ln - La NdTM20o7

u1 Mm-alcohol (2 wt. 1) as binder. The pellets were subjected0.05 M Lanthanum/Neodynium to heat treatment at 1300. 1350. 1400. and 15UO "C forAcetylace.anate (LnA) in 2 M 5 h to study their sintering behavior Bulk densities were

measured by the Archimedes method. A scanning elec-2-Methoxyethanol (MOE) tron microscope (ISI-DS 130. Akashi Beam Technolo2v.

f_ Tokyo) was used for obtaining microstructures of theDistill off Solvent sintered ceramics.

Ill. RESULTS AND DISCUSSION

As stated in the introduction, the synthesis basedAdd MOE (NOEII..nA. = 80) on pure alkoxide precursors resulted in the formationLand Reflux at 125C for 12 h of a defect perovskite phase of composition close to

Laf,_,,TiO_1. The thermal behavior of the gel powderobtained is shown in Fig. 2. X-ray diffraction (XRD) pat-terns of this sample heat treated at various temperaturesare depicted in Fig. 3. The formation of such A-site defi-cient perovskite Lat_,_TiO1. where 0 < x < 0.3. fromAdd 0. num ceramic oxides was reported earlier by Kestigan and

isopropoxide and Reflux at Ward."8 The phase of LaUTiOA (corresponding to x =0.33) was also synthesized from a stoichiometric mixture125'C for 12 h of La,O, and TiO, in the presence of a small amount

of alkaline earth ion.18 Abe and Uchino '• reported theCool to 25"C synthesis of La2/ 3TiO 3-A from pure oxides under CO-H:

mixed gas atmosphere. Based on XRD results, theseauthors concluded that the structure of La2,JTiO3-, %as

Adjust pH from 12.0 to 10.0' dependent on A: when A is small the perovskite cellwith Nitric Acid and Add is distorted to orthorhombic symmetry, while a cubic

perovskite phase is facilitated with the increase of A.Water In the present case. the XRD pattern (Fig. 3) shows a

cubic cell with lattice constant. a - 3.91 A. Only partialtransformation of La,,_,,TiO. into La.TiO, occurred on

Stir at 25"C heating to 1500 'C for 6 h.In order to modify the reactivity of molecular pre-

cursors, chemical additives, such as acetylacetone andLanthnum/ eumIalkanolamine, are sometimes used.-'0°'' These additives.Lantheanum/Neodymiumaine ares times, modify Thes ades.

Titant e primarily being chelating agents, modify the molecularTitanate G&lI

FIG I Schematic outline of the preparation of tanthamineo.m -d. mium tlanate

10.0 with concentrated HNO.,. Four times the theoreticalamount of water necessary for hydrolysis was added oslowly under continuous stirring. The addition of watercleared the slight turbidity of the solution. The solutionand the gel were air dried at 60 "C.

Powdered gels were characterized by thermogravi-metric (Delta Series TGA7. Perkin-Elmer, CT) and dif-ferential thermal (Model DTA 1700, Perkin-Elmer, CT)analyzers interfaced with computerized data acquisitionand manipulation systems. Phase identification of thevarious heat treated samples was performed using a ' 2 .6'0 - -L3

diffractometer (Model DMC 105, Scintag, CA) with Ni TCMP(Rt, ATuRt MC1filtered Cu K. radiation. The La:Ti:O- gel powder was FIG 2. DTA curve for the eel obtained from lanthanum and titanium

calcined at 800 "C for I h and pelletized using polyvinyl tsopropo'ude precursor• (heating rate 10 *T'mini

2860 J Mater Res. Vot 7. No 10 Oct 1992

A. V Prasadarla or of. Sol-gel synthesis of Lfl2(Lfl La. Nd Ti20

0z

z Z

t0 20 30 40 50 60DEGREES TWO THETA (CuKe) 06

FIG. 3. XRD patterns of the gel powder obtained from lanthanumn

and titanium isopropoxide precursors and heat-treated at 800 and lp

0500 1C. 20 30 .40 so 60DEGREES TWO THETA (CijXat

_T FI0. 6 XRD patterns for the gel powder obtained from lanthanumn

too- aceihl.acetonate and titanium tsopropoxide precursors and heat-trc~itedat different temperatures.

S the lanthanum acetylacetonate itself was chosen as oneS of the precursors in this study. The change in lanthanumn

precursor facilitated the formation of La2Ti2Oi directly0 without any intermediate phase. The thermal behavior of

w gair dried gel obtained from lanthanum acetylacetonateS and titanium isopropoxide is depicted in Fig. 4. TheIthermogravimetric analysis (TGA) shows a continuous

loss up to 350 6C and another loss in the range of 580

soil to 780 'C. The earlier loss is due to dehydration and

60 2,0 i0 00* 40 ;! decomposition of organics. Differential thermal analysisTIMPIRATUOE rVof the gel powder previously heat-treated at 500 C

FIG. 4 TGA and DTA curves for the gel powder obtained from Ian- for 24 h gave a sharp exotherm at 750 *C with nothanumn aceixlacetonate and titanium isopropoxide precursors (heating corresponding weight loss observed in TGA. as shown inrate tO *C/mmn. Fig. 5. Ishizawa et al. : reported a phase transformation

of monocirnic La2TiO, to an orthorhombic form with

rG 98-

94-

tI0 2507 5 3 o 9

TEMPERATURE (*C) a.

FIGý 5. TGA and DTA cum-es for the gel powder obtained from Ian- 92thanumn acetl~lacetonate and titanium isopropoxide and heait-treated at91500 *C for 124 h theating rate 10 *C/min). 91 ý

1200 1300 1400 1500 1600

precursor in such a way that its rate of hydrolysis is al- TmeaueCCtered. The controlled rate of hydrolysis promotes homo- Tmeaue(C

geneous mixing of the resultant gel network. Instead of FIG 7 'variation of densitr as a function of sintering9 temPetatute fOr

modifyin g lanthanum isopropoxide with acetylacetone. La:Ti.-O-

j Mater ReS, Vol 7 NO 10 Oct 1992 286,

A V. Prasadarao ef at, Sol-ges synthesis of Ln 2 (Ln - La. Nd)TtrO,

space group C,,,, 2 at 780 'C. In order to verify whether Figure 7 shows the densities of the sintered LaTi2O,the exotherm at 750 'C is due to crystallization or samples as a function of sintering temperature. Maxi-phase transformation of La-Ti.07. DTA cooling run mum density was observed for the saii-pie sintered atwas performed from 800 to 500 *C. which showed no 1400 *C. For microstructuial studies, sintered cross sec.peak in this region. This indicates that under the DTA tions were prepared by polishing the surface followed bysetup conditions, it may not be possible to identify thermal etching at 1400 *C for 10 min. Figure 8 show'sthe monoclinic to orthorhombic transition. Hence, the scanning electron micrographs (SEM) of La:Ti,O, sin-above exotherm can be attributed to the crystallization of tered at 1400 °C for 1. 3. and 5 h. The grain sizeLaTiO,. The phase identification was also followed by increased with increasing heat treatment time (Fig. 8).XRD for samples heat-treated at different temperatures. Thermal behavior of neodymium titanate gel shownThe sample was x-ray amorphous up to 600 *C. The in Fig. 9 is similar to that of lanthanum titanate gelXRD patterns (Fig. 6) clearly indicate the formation of powder. The phase formation followed by XRD (Fig. 10)LaTi_,7 at 700 'C. As TiO, crystallizes at temperatures indicated the formation of Nd 2Ti.,O7 at 800 *C. Theless than 600 "C, the absence of TiO2 peaks suggests that phase formation was complete within 1000 'C. TheLaTiO 7 is formed directly. The phase formation was optimum sintering conditions obtained for lanthanumcomplete at temperatures >800 0C. titanate samples were adopted to the neodymium titanate

(a) (b)

(C)

FIG. 8 SEM micrographs of LaTi2O, sintered at IM1)0 T for (a) 1 h. (b) 3 h. and 1c) 5 h

2862 J Mater Res.. Vol. 7. No 10. Oct 1992

A V Prfasatiarao of al. Sol-gel synthesis of Ln2(Ln - LaNdITQ,20

__T the Loigering orientation factor calculated from XRDtoo ~intensities of the sintered pellets. - Grain orientation has

also been observed for L~aTiO,0 sol-gel thin films coatedton Si( 100) and fused silica. 17

IV. CONCLUSIONS

Lanthanum and neodymium titanates we're prepared* by a sol--el method for the first time at a relatively lower

temperature of 700 to 800 'C. Phase pure La'TiZO-without any intermediate phase was synthesized usingtitanium isopropoxide and lanthanum acetylacetonate.Sintering of LaTi2O7 gel powder at 1400 'C for 5 h

5c0 * 'Iyielded a highly dense ceramic with -97 theoreticalTEMPERATURE 1`0 density.

FIG 9 TGA and DTA curves lot the gel powder obtained from REFERENCESneodymium aceivacetonrat and titanium isopropoxide precursors(heating rate 10 *C/min)r I. W R. Cook. Jr. and Ht Jaffe, Phys. Rev. 38. 1426 ( 1952).

2. R. S. Roth. J. Res. Nait, But. Stand. 56. 17 (19561

gel powder. Approximately 927o theoretical density was 1 0. Knoipr. F. Brisse. and L. Castetliz. Can. 1_ Chem. 417. 971achivedfortheneoymim saplesinere at140 'C(1969).achivedfor he eodmiumsamle snteed t 140 ' 4.P. M. Gasperin. Act. Crystailogt. 8 .31. 2129 (1975).

for 5 h. La2TiO 7 and NdŽTi2O7 pellets sintered at 5. S. Nanamatsu. M. Kimura. K. Doi. S. Matsushita. and N. Yamada.1300, 1400. and 1500 'C exhibited grain orientation. The Ferroelectrics. S. 513 (1974).extent of grain orientation v as quantified in terms of 6. M. Kimura, S. Nanamatsu. T. Kawamnura. and S. Matsushita. Jpn.

1. Appi. Phys. 13. 147 (1974).7. K. Kagayamat and T. Mitsuhiro. Jpn. J. AppI. Phys. 24. 10)45

- (1985).7 8. K. Wakino. K. Minai, and H. Tamura. J. Am. Ceram, Soc. 67.

-~10'0C 278 (1984).9. M. Kimura. S. Nanamatsu. K. Doi. S. Matsushita. and M Taka-

a Z hashi. Jpn. J. Appi. Phys. 11, 904 (1972).-A 1 . 40. J. K. Yamamoto and A. S. Bhalla. Appi. Phys. ei sbitd

It. J. K. Yamamoto and A. S. Bhiatta. Mater. Lett. t0. 497 (1991).12. L. G. Shcherbakova. L. G. Mamsurova. and G. E. Sulthanrova.

33. J. Takahashi and T, Ohisulta. J. Am. Ceram. Soc. 72. 426 (1989).14, U. Balachandran and N.G. Eror. J. Mater. Res. 4. 152-5 (39893

010 r 1S. E. B. Panasenko and R. . Begunova. Russ. J . lnorg. Chem 29.1430 (1984).

sooc 36. C.1 nrinker orA. G H. 5..-heri'er The Phvsics and Chemist- ofSol-Gel Processing (Academic Press. New York. 3990).

17. A. V. Prasadaraot. U. Selvara;. S. Komarneni. and A. S. Bhatta.Ferroetectfics Lett. (1992).

18. M. Kestigan and R. Ward. J. Am. Chem. Soc. 77. 6399 (1955).19. M. Abe and K. Uchino. Miater. Res. Bull. IX. 147 (1974).

TOOGC 20. C. Sanchez. F. Babonneau., S. Ooeuiff. and A. Leanstic. in Ultra-stutre Processing of Advanced Ceramics. edited b) J D

Ma'&c~kenzie and D. R. Ulrich (John Wiley. New York. 1988). p 77.2. so 0 50 I0 21. Y. Takahashi and Y. Maisuoka. J. Mater. Sci. 23,3 2259 (1988)

0EGREES TWO THETA Kyiau) 22. N. Ishizawa. F. Marumo. S. lwai. M. Kimura. and T. Kawamura.Act. Crvstaltogr. B .38. 368 (3982).

FIG it). XRD patterns for the gel powder obtained from neodymium 23. A. V. Praosadlarao. U/. Selv ataj. S. Komarneni. and A S. Bhaiia.acetylacetonate and titanium tsopropoxide precursors and heat-treated Mater. Lett. 12. 306 (1991).at different temperatures

j. Moter Res.. Vol 7. No 10. Oct 1992 2863

APPENDIX 41

journal -

Sol-Gel Synthesis of Strontium Pyroniobate and Calcium Pyroniobate

AI.inunda V l'ratsadira(" t'ligarall Sd'. jr;Jl2 Sirtdlar Komrnrncnj.* ' and Amar S Bhiluli*

%UtacriiaI' Hc~i..iii It libi rati r% 1fic Pcfnii' 1% anti 1,t~it I fntcrts IitN1 fi. ur'it S Park Penn,% ki.nti I (0,i1

Strontium and calcium p~robnhihate% vie,-e prepared h, a the otixine of the presurso)r ol'ution it the rnolecular teesci the

siti-gel process. using striintiu micalIcium metal and nio- soIi-Lel method offers httmogenemi., reduc~ed rei~ttion ienipera

hium ethoxide as precursors. The fotrmationt of SrNh.O. lures. and ease of formation ot thin filmsn \%.e hat~e te~tnilsoccurred at 750"~U %ia in intermediate perosskite phase reported the scil-gel st7nthesi, (it La.Ti 0- and Nd Ti 0 ) Inof composition close to Sr,.b, The cr~stallization siev% of the abose interestin-, properzies outlined for !sr \h 0-

oif CaNh.0- occurred at 600*C directl% %without anv inter- Lind Ci.NhO.. ssnrthessi~ot these phise-pure niaterijii.i under-mediate phases. Sintered Sr,.Nhb.O and Ca."ib:U. pellets taken b\, the stelprocess In tact, the siti-eci

shovied a preferred grain orientation. Mdicrostructural produces single- phase materials. is opposed to the imnsen-studies revealed an increase in grain growth and associated tional pro~cessing methods. leidin, ito the torination tit c,,.>orientation with sintering temperature. ond phases,

1. Introduction 11. Experimental Procedure

',(N;teternirt, metallic oxides viith the genri formula The nlo" chart for the si~nthesis. of Sr.Nb.O- and ('a li UAr(A.B,7t compo'und'. \Aith lalered perox'skite-iype struc- goel is outlined in Fig, I For the preparatInon tit Sr Nh-U -. I.

lue ehbt neesigfeml~crcproperties. th o- srnimmtland niobium ethoxide tAidrich Chernticl Copoiunds where A =Sr. Ca and B = Nb. Ta have become the Mfilwaukee. WOt were used as starting materials stit)hsubject of1 some recent investigations " Crystals, of Sr.Nh.O. 2-methoxiiethanoil i Aldrich) as a soltent, The required ink'urn

are orthorhombic with space group Oyu-, at Iroom temperature (if strontium metal was slow[\ reacted v ith 2-meth,'sscthaniiand the crystal structure is built up of slabs of distorted NbO, in a molar ratio of I1i5 under argon atmosphere Atter the out1-

octahedra and strontium atoms along the [010] axis.' Dielectric plete dissolution of th,! metal, the solution %4as retiuxed undermeasurements on Sr:Nb:.O- single crystals revealed twAo ferro- arigon at I125'C for 4 hi. Niobium ethoxtde wxas retluxed cepa-electric transition temperatures, one at - I156'C and a second at rately with 2-methoxyethanol in a molar ratio oif 1 27 tinder

I1342'C.' Also, the electric anomalies showed a normal to similar conditions.incommensurate phase transition at 215'C.' Refractive index. A mixture of 0. IM strontium alkox ide solution. Ii V1 j,.birefringence and lattice modulation properties of the incom- tylacetone tAldrich). and -4 Of 2-meihoxu.etihnol \,,i,mensurate phase have also been investigated.- Besides being refluxed under argon at 1.15'C for 4 h. Alter the solution %iferroelectric with a high Curie temperature. Sr.Nb.O. exhibits cooled. 0. IM niobium ethoxide solution was added and turther

excellent piezoelectric and electrooptic properties comparable refluxed under the same conditions, The resulting soiution 1,% a'

to those ot LiTaO, and Ba~NaWb,O. single crystals.' cooled and the pH was adjusted from -12 (1 to 1 Q iti h H1NoJCa.Nb.O-. (in the other hand, belongs to a monoclinic system

with a space group P., and is isostructural with La.Ti.O,. "'Early attempts to detect ferotelectricity in CaNb.,O, wereunsuccessfl-"'': Nanamatsu and Kimura' reinvestigated the Strontium/ac~flMu Niobium Ethotide in

electrical properties of- Ca.NbO,. They reported a polarization in 2-Methoxyethanot 2.-Mkethoxsethianol

reversal at very high applied fields and observed no dielectricanomaly from room temperature to 1500'C. The Cune temper- Reflix st 15C fr~ature of CaNb.O- is therefore assumed to be higher than its 4 h ini Jww

melting point of 1581i'C. Ca.Nb.O, also exhibits remarkable Mix 0.1 M Strontium/Caltcurn,piezoelectric and electrooptic properties %hen an electnic field 012 M Acetylacetone and to',- 4 'oafl

is applied along~ the polar axis - All of the above studies have 4 M 2.Methoxyethanol ~ ~been performed with single crystals grown by float zone tech-niques Single-cr~ stal fibers of C-a.Nb.O, and related materials tniof15fc

%kere also prepared using laser-heated pedestal growth tech-niques. Attempts to grow Sr.NbO- and Sr-Ta.O- fibers were Ad 0. 1 M Niobium Ethoxideunsuccessful Since single crystals of these materials are in 2-MethoxyethanolJimpractical for muost device applications because of excessiveRnx21.5 ocost, lowt-temperature s,.nthesis of' polycrystalline or grain-Rflxt5fooriented ptoscrystalline ceramics is desirable.Adutpto0anHyrye

In recent sears. sol-gi~e synthesis is found to have many Adusin Water Dilutnd iHydoyeadvantages over conventional ceramic processing." Becaus o 2-Methoxyethanol

FStrontiumVCalcium1Pnyroniobate GelIVlisuswnpi No luh2ftill Rct:cised 0ooibcr IA199. ti~J pproved June I I. 1'J92.

*ilmhfl a..ir AndhraL'nCOit, SRsiiY Fig. 1. Schematic oudint: of the preparation of SeNti ) ~iidAls- - ith the tkpurimcnitit A cr.-nimy CjiNb:O. gets

2697

-100 0

I 4825

50 930 gO 450 7,00 7Q 60 240 420 600 70 9TEMPERATURE (M) TEMPERATURE MC)

Fig. 2. TG and DTA curie, for Sr Nh 0- jee ihteatng rate Fig. 4. TG and DIA curve-, for CaNb.O. gel poder theitine taleIi'Ct mm tn I Cmini

Slow bydrol.sis %as initiated with the theoretical amount of and 92% of theoretical densities. respectihel. The pellets sin-water ldiluted with 2-methoxvethanol) necessar% for complete tered at different temperatures were polished and thermall\hydrolysis and the resulting gel was dried at 6&)C A similar etched for microstructural studies with a scanning electronprocedure was followed for the synthesis of CaNb:O- gel with microscope (lSI-DS 130. Akashi Beam Technology Corpcalcium metal (Aldrich) and niobium ethoxide as precursors Tokyo. Japan).The gel powders of SrNb.O- and Ca.NbO- were characterizedby thermogravimetric iDelta Series TGA7. Perkin-Elmer. Nor- Ifl. Results and Discussionw alk. C"T) and differential thermal (Model DTA 1700. Perkin-Elmeri anal'zers interfaced with computerized data acquisition Figure 2 shows the DTA and TGA curves for Sr.Nb:O. gel inand manipulation systems. Phase identification of the various the temperature range of 50' to 850°C. As indicated b, the TGheat-treated samples was performed usinp a diffractometer curve, there is a sudden weight loss at about 310"C followed hb(Model DMC 105. Scintag. Santa Clara. (At with Ni-lilhred a cotitittuous weight loss ill the tatige of 580' to (80'"C I lihcCuKu radiation The gel powders, after the removal of carbon two weight losses are separated by a plateau from 4500 toh%, heating at 50tM)C. were pressed into pellets with 2 wt% pilý- 580(C- The DTA curve shows a series of exothermic peaks.m, in, I alcohol) a, a binder and sintered at different tempera- The peaks below 400'C are due to the loss of water and organ-tures The densites were measured hw the Archimedes method. ics. while the peaks above this temperature may. be due to theSrNbhO. and CaNb.O- sintered at 1450'C for I h showed 88% crystallization and the oxidation of residual carbon As there i,-

no sharp exother'mic peak indicative of crystallization, thephase formation is followed by X-ray diffraction of the gel po. -

S,Z Nb.0, ders heat-treated at different temperatures (Fig 3) The gelsoo*c powder heat-treated at 500'C was X-ray amorphous However.

it transformed to a perovskite phase at 60O°C. The XRD patternIof this perovskite phase matched with Sr,t,.NbO, lJoent Com-

of tmission on Powder Diffraction Standards. File No 9o79)

Ceramics with the composition Sr,NbO, have been known since

I I Cot Nbt 02 ,

700C00

600*C

600"C

500'c soo-c

iO 20 30 40 5o 60 10 20 30 40 50 60DEGREES 29 tCuKt) DEGREES 29 tC-KaI

Fig. 3. \RI) pljcrn., Wi Sr Nh 0) .cl pnkdJcr hcat-irt:J,+id .it difler- Fig. 5. \Rt) paitirn, to Cj " ()j N 0 co ,',dcr h,.i-ir.j d ,,it dm rernt wtrllp .f-aiurc'- tn he n 'rf ,frJt uic -

- 800"CIh 1450*C

,2 Z Sr2Nb 2 07 133 0 *C

S9000C I h

10 20 30 40 50 60DEGREES 28 (CuK|) v 125COC

Fig. 6. XRD patterns ofCa:Nb-O. and Sr.Nb:O. gel powder calcinedat, M00° and 900'C for I h. respectively- N

1950. It has been well established that for 0.70 < j < 0.95 the V J 6 ,formation of a single-phase perovskite is favored." " Heat ,treatment of the gel powder at 750'C yielded phase-pureSrNb,O,. Samples of SrNbO, prepared from SrCO, andNbO, by the conventional pGwder mixing method'" were

10 20 30 40 50 60DEGREES 28 (CuKq)

Sr2 Nb2 O7 Fig. 8. XRD patterns x)1ished surfaces of Ca.Nb.O- pellte, ,in

14 500C tered at 1250P. 1350*. ant i 450*C.

-. reported to have a small amount of Sr,NbO,, as impurit, %. hichT1 7was not observed in the present synthesis.

t: ! -Figure 4 shows the DTA and TGA curves for Ca.Nb.O- i2el n10- 8-Q, -o the temperature range of 500 to 1000*C. There is a continuu.,0 weight loss shown in the TG curve up to about 500°C tollowed

-t& Lby a minute weight loss up to 650TC. As in the previous case.DTA peaks below 500TC are due to loss of water and decompo-

1350*C sition of organics. Above 500TC DTA showed a sharp exothermat 6800C. due to the crystallization of CaNb,O,. This was lur-ther confirmed by XRD of ge. powders heat-treated at differenttemperatures (Fig. 5).

The XRD patterns for the calcined powders of CaNb.O. andSrNbO, are shown in Fig. 6. The (212) and (131) retfecion,are the highest intensity peaks. respectively. However., whenthe pellets were sintered at different temperatures. the corte-sponding XRD patterns of polished surfaces showed (080) j,the highest intensity peak for SrNbO, and (400) as the highetintensity peak for Ca.Nb.O, as shown in Figs. 7 and 8. This

12509C shift in intensities is attributed to a pref-rred grain orientationduring sintering. A similar grain orientation was also oberedfor La.Ti.O, and NdTiO,." CaNbO,. being isostructuralwith La,:TiO,. showed a similar trend. Sr.Nb.O,. on the otherhand. is orthorhombic and the observed XRD intensities ot the

TableLI Orientation Factors, f. for SrNb.O, and Ca.NbQ0-Sinientig Oneniaitan -Im.o II ~ ~ ~ ~ ~ ~ ~ ~ tmeau IT I If Nb0 CAN I I ~ ~ lll l

10 20 30 40 50 60 ____ __O.___,___ ___

DEGREES 28 (CuK,) 1250 0.14 00O61350 0.23 0 13

Fil. 7. XRD patterns of polished surfaces of Sr:Nb:O, pellets sin- 1450 0 21 0 20tered at 1250". 1350*. and 1450°C.

27(Mt Journal ,• Mhe American Ceramiu So ,et \-Prasadaruo el ul Vol 75. No 1O

S,.

W a N Il

Oki .3 - -3.

.,Fh

FIlt~r. 9. hLMh nIieirpso plse n dhelinacd ichdpel ,,urig 10. ShE mertn tfo~rienttpione And the,,nat , e kh, ,ape', phei

surj,.s ii S~hO. inireditiia 12(P.ibi135 .andId 45tC. urauce, ot Ca.NbO- M nerdaii Sr~(.Nb arh ,,h51 n ,nd i abl 4(

in-iered pel- sh,)sed -X~ refecio) Theeiges iteeiyriened oamplenlPab|cms P,. reare,,nta|\ tohI bul With a

peahcr-c XP whc istl<l in/tl hccrdane Srit.O thenteramic pre-It ,nrampng d,.¢aeree ot ofluen, to,,,uredam diinorintd sample t depn-

parc h= thelll p.lhs l lr mii~mthod Theb. rdexetfpreferred ot•riete samlelli.t Pdntcrase ri h ju . n orsod

•hC Irc I h ' = 5 hi 41ii •.I hlh~ ta r thre C-Nb .O- ,,intered pelehsape ts.'.,ausetheIlure, m•te asured atitrni phermllet deph.s

p,.llci. and I' = "h 4i(i jtU L I~tI hr the Ca-Nb () paxsder cal. ,,urtJ,.- at! the ,,intered pellet, ii Sr Nb.O and C., Nh-,() arc

•.ne.d at.lX (" h C ta I h a,'er III to f•5 at 2" \alue' Far a llr'n- '.ha"' ni in F-ies' a and Ill The ,'rieniailn a~l Cra 'in', k-al•-rl'

• • n | ii

'ienlncint increase in grain .ize I, notic:ed ii ith increj~inf2 in- llIri..b *In )f \t' i ') k", . N3 JR. _Mt- i.0.Ni K~hic ( ~~ ,i.)hj,J NjjF Uh 'cor~ i,!" Inj~, ^jtering temperature Brrix ) I 'iNho A i rult IN,: In-lxintenflur

4 s, Ph~i- ki.n-,I ~ J's 5M. IAJK--4111 14)(14

's \jirmmoti .nd I. fhiiuk~i An-11'o "I thc In-nm'.n,uv-w, Nlru~tu%~I V Conclusions in Sr St, I h% Elcwtin \,Ltircrswpý anid Gtrivriccrtcn (kejn, tic-.trn (J~iir.j.

Using stron~tum calcium metal and niotiury etho'ude as pre- li~ ABijin Gr(r B.49 .t. 1jnjli FcniI 1 , XIiru It.ih

cursors. phise-pure Sr:Nh:O. Ind Ca.Nb.O. \ere prepared b\, a C/,...nrjI.k T,;,hn.,quc I Am' (i.' S-. 48. 1i 1 -t .j#jsl-tgel pro~es, U'nlike the simple, prepared by consentional N khijjj.i. F %Ijrum,, S 1ý \1 Kimura jfnd P kj~murj

piw~der mil.ing Methods.. these materials. did not showk an%~ p'lun% ý.iih IPcr's'Lic Still, III The Sirutturc it Mimtin.c 5 it dr~iCNh () 4,1, (ot 1,icI 8.16. Ith f-his I4AIYIiinipurmies a\ second phas.e Sr Nh-O cr\ stallized %.I ii intnter F Joj, 6 Shirine. Icnd R PepinAk. Dic~c~trik \r. crid an Opiýjl tmediate peroiskite phase. %I-hile Ca-Nh-C) cr\'.tailized directI\ ,I Fvrrovckirn Cd Nh 0. and Rcijted Compound, Pit, , R, 98. toI, J

it (IMT~ Sintered pe~les' ot Sr.Nh.O. and taNb-.O 5ho~ed li Iit j' ih had(hrtlriiIni ctlI, mk(f aconsiderable Izriin orientation. %k hi~h A~a,. turther confirmed bi rhjd(ja~~..i.nc crcI-r~Sc~(oSE~trniro'.rucure tude.Fitter PrctduuccJ hc the Late~r Wciled Pede'iji Grc,-ih I,:hnc.,., Ph D I [hý,cSEMmicosiuctre udis.Penn'sI.cic4 State I. nioersits tjnicrscis Pajrk PA 1ýI4''C J Brinker .ind G H S, hcrcr rb, Pt .c .. cc . (.. icccco. j. -ip.It -1v' Acjdcrcisu Prc ess \s Ncrk 1W I

References .A Prjs~jtjrscc. L' Seiviric. S Isomicrrien, and A S Bhcibil s-.j..~t'S Nanamaitsu. %I Kimura. K Dosii nd 1. TjkjIIjSi. Fern'cdcciric Prop- S.snihe'cs .1 Ln~t n =Li. Ndirc o.. / Rcc' ,,~ in PIC-

crn. o SrNt'O. cn~e C~siI.JpsisocJp~ JO ~ -"Mc Fukuhicra. C V Hujcnv A S BhsIUj. and R E \v,,nh,,m (cajn (it'S Nanam~ii~u and MI Kimura. -ferriviciric Properties of CiNb.0. Single hlihittiil ad iecr~lPipriso rN rm~ Ic.

Crysitts." I Phic Sc Inn . 36. 14,4,; 197-1 '_ Rsvlr in R 1r.'h PeaicnII totu o-imb1n,; S Nanamaisu. ki Kimura, ind T Ka.wamiIra. Cr~iatjioraphtc and Dielec- DRde ndRWr.TePpaicnofaSrcnumNuuHrntinc ProlpernieotFemelecciric A.BO. lis = Sr. = Ta. NhiCr%.siats and Their wcith the Percvsstoe Structure.- I Amn Chiem 5,,1.77. 6112 1, i5. kSolid Sotuisons. J Ph~s So)c J1P, . 31.817_24147 -8 Hcensc. S A Surnshine. T Stevrist, and R Jimnterw ( rs-alfi/ii-i ,n

'A shailntiand hi Kimura. -Singte Crystal SriNbi.O Film Optical Wive- Reduced Strocntium Ind Banum Ntiibaie Pero% Aite from 8caise Flu ýe% %lim,,guide Depn~iiedbv' RF Sruitering. -ipplfPht. Lett .29.2 i4-9f t47isi Resi Bull .26. 8S5irn) 14,4 I

'N Ishilawa and F Mcaruric. "The Crystal Structure ot Sr.Nb.O.. a Co~m- 'A \, Prasadarail L: Sel%3raj. 5 Kosmameni. and A S BhjdI., (trim 1kr..ponnd with Perovkite Type Stabs." Asia tC rxircji/i . 831. 1412-15 41i 5 ii7fcflitiol inl Soil-GeI tensed Ln.iLn =La. .NdiTc 0 Ceranii,' ifa,. t

'Y Akishige, N1 Ko~icbsjihi. K ONi and E Sas-apukih. -Dielcctric and Fcr- 12. 106-l1i)t 1441)rc-ceectnc Properties in ihe Low-Temperaiture Phase of Strontium Ncobaie.' F K Loigerin g, Tcpiaii~oci Rcaction% w'ith Ferrimsrnj--ct it 'cdc' HI nI Phis, Soc Jpm .55. 27G1-77tI9NMI Hexagoinal Cr~ial Structures-$. I I/noriq Nwic Chem,, 9. 11 1-' iyt

"APENDIX 42

Feroelrtcrrcs L-eers. 1992. Vol 14. pp. 65-72 0 1992 Codon aind Breach Scence Publishers S AReprints available directly from publisher Pirined in the United States of AmericaPhotocopying permttied by license only

FABRICATION OF La2Ti2O0 THIN FILMS BY A SOL-GEL

TECHNIQUE

A. V. Prasadarao,* U. Selvaraj, S. Komameni÷ and A.S. Bhalla

Materials Research Laboratory, The Pennsylvania State University,

University Park, Pennsylvania 16802, USA.

*Permanent Address: Andhra University, Visakhapatnam, India.

*Also with the Department of Agronomy.

tRccetvird for Publication March 9. 1992

Sol-gel thin films of La2Ti 207 were deposited on fused silica and

Si(100) substrates by a spin-coating process. The La2Ti207 precursor

solution for the spin-coating was prepared from lanthanum acetylacetonateand titanium iso-propoxide dissolved in 2-methoxyethanol. Crystalline andcrack-free films of - 0.3 prm thickness were deposited on the above

substrates using a single coating and followed by annealing at a temperatureof 800 *C. Microstructural studies revealed that these films containedextremely fine grains of - 0.1 pm. Thin film X-ray diffraction patterns

indicated the formation of grain oriented films along 1100) direction on thesesubstrates.

INTRODUCTION

Much interest has been focused on the ferroelectric properties of layered

perovskite compounds of the general formula A2B1207, because of their high Curietemperatures and thermal stability. In this category of compounds, A is a trivalent

ion such as La3÷, Nd 3' with B being Ti4÷, or, A is a divalent ion such as Ca2",

Sr 2÷ with B being Nb5÷, Tas*. Among the layered perovskite compounds,

lanthanum and neodymium titanates (La2Ti20" and Nd2Ti20?) have high Curie

Communicated by Dr 0 W Taylor65

66 A V P1kA.ADARAO et at

temperatures (- 1500C) and high coercive fields.1' 2 Their high temperature

stability combined with low dielectric loss at microwave frequencies make them

good candidate materials for high frequency applications. In addition to being

ferroelectric, these materials also exhibit excellent piezoelectric and electro-optic

properties for possible use as high temperature transducers. 2.3

The room temperature modification of La 2Ti 2O7 belongs to the monoclinic

system with the space group, P2 1. The crystal structure is built up of layers of

perovskite slabs running parallel to (100) plane, with the TiO 6 octahedra bonded to

each other by the interlayer La3 * ions. 4 A high temperature modification of

La2Ti207 at - 780 °C with the orthorhombic space group, Cmc2j has also been

reported. 5 La 2Ti20 7 has also been described as a - 3 member of a homologous

series of layered structures with the general formula An.5 Bn÷ 10 3n*5 [0 s rn: ci.6

Single crystal fibers of La2Ti207 and related materials have been synthesized by

laser-heated pedestal growth technique.7 The microwave dielectric properties and

the piezoelectric properties of these single crystal fibers have also been

investigated.. 9

The sol-gel method has been widely used to deposit many ferroelectric and

dielectric thin films. When compared to vacuum based techniques such as

chemical vapor deposition and sputtering, the sol-gel processing is a solution based

technique of depositing thin films without any vacuum. The advantages of sol-gel

method over other methods are precise control of composition, low processing

temperature, better homogeneity, easier fabrication of thin films over large area on

either one or both faces of the substrate and low cost. We have recently reported

the preparation of some layered perovskite ceramics using sol-gel derived fine

crystalline powders. 0,1 In this letter we present the first report on the fabrication

of thin sol-gel films of La2Ti207.

The precursor solution for La2 T1207 was prepared from lanthanum

acetylacetonate hydrate (La(acac)3.xH201 and titanium iso-propoxide [Ti(OPrO)4I

SOL-GEL THIN FILMS OF LajTiO, 67

dissolved in the solvent 2-methoxyethanol (2-MOE). A mixture or 0.05 M ofLa(acac) 3 .x.H20 and 4 M or 2-MOE was distilled in argon to dehydrate water. To

the resulting slurry, 2-MOE was added to maintain the above lanthanum

acetylacetonate : 2-MOE ratio and retluxed in argon for 24 h. Stoichiometric

quantity or Ti(OPi) 4 was then added and the mixture was ref.uved for 12 h to

achieve homogeneous mixing. The pH of the solution after cooling was adjusted

from - 12.0 to 10 with concentrated J'N0 3 . Hydrolysis was Initiated by adding

water which resulted in a very clear solution. This precursor solution was used for

subsequent spin-coating. A portion of the solution was gelled separately and the

gel powder heat treated at 500 "C for 24 h was characterized by themogravimetric

(Delta Series TGA7, Perkin-Elmer, CT) and differential thermal (Model DTA 1700,

Perkin-Elmer, CT) analysers interfaced with computerized data acquisition and

manipulation systems. Phase identification of the gel powder heat treated at 700

and 1000° C was performed using a ditTractometer (Model DMC 105, Scintag,

CA) with Ni filtered Cu Kat radiation..

Fused silica and Si(100) were used as substrates. Prior to coating, the

substrates were cleaned by standard semiconductor processing technique.1 2

La 2Ti 2O7 precursor solution was deposited on these cleaned substrates using a

spin-coaier (Integrated Technologies P-600) operated at 2000 rotations per minute

for 20 s. The resulting films were dried at room temperature and slowly heated to

800°C at a heating rate of 2°C(min. A thin film XRD (Scintag, Model DMICRO8)

equipped with a set of angular divergence soller slits in front of the detector element

for parallel beam geometry operation was used for the phase identification of the

thin films. The fine structure and the film thickness were obtained by a scanning

electron microscope (SEM, ISt- DS.130, Akashi Beam Technology, Tokyo,

Japan). The thickness of a single layer film deposited on either fused silica orSi( 100), calculated from SEM micrograph, cofresponded to - 0.3 pm.

RESULTS AND DISCUSSION

The gel powder when heated to 5000C for 24 h was X-ray amorphous. DTA

of this powder gave an exothermic peak at 750 *C with no corresponding weight

68 A V PRASADARAO et Wi.

IT0 J G____________ _

75- xI W00

50

110 290 470 650 8,30 1010TEMPERATURE (OC)

FIGURE 1. DTA and TG curves ror La 2Ti2O7 gel heat treated at SOO*C for 24 h.

0 loooec "h,

a V I I

to 20 30 40 50 60DEGREES 29 (CU KI

29-517 LaaTij0?

FIGURE 2. XRD pattern of La2Ti2O7 gel powder heat treated at 700*C and

10O0* for 1kh.

SOL-GEL THIN FILMS OF LiaTi7O, 69

loss observed in TG as shown in Figure 1. This indicated (hat most of the

carbon species present in the films would have been removed at temperatures

below 500 *C. The exothermic peak in DTA can therefore be attributed to the

crystallization of La 2Ti 2O7. The XRP) patterns of the gel powder heat treated at 700

and 1000 °C for I h are shown in Figure 2. As can be seen from the figure, phase

formation of La 2Ti20 7 occurs at temperature as low as 700 *C. Further increase in

temperature altered the crystallite size only resulting in rather sharp peaks in the

XRD pattern.

Z

Lo2Ti2O? on Si (I00)

N

N Lo 2 Ti2 O? on Fused Silico

10 20 30 40 50 60DEGREES 29 (CuKg)

FIGURE 3. XRD patterns of La 2Ti 207 thin films on (a) fused silica and (b)

Si( 100) heat treated at 800-C for I h.

70 A V PRASADARAOe al.

- ~N,

* *b(r.;i A,1)

r~

(A "Y

ýAAýP

FIGURE~~~~~~~~~~~~~~ C.SMmcozpsofL2 iO hnflSon()fsdslcad

()Sl0)hAttaeda80Cfr1h

FIGURE 3. show mhecrogrphtes of La2Ti0h thin films spn-coate)o fusedslcan

silica and Si(lO0) and heat treated al 800 'C for I h. The gel powder calcined at

800 *C showed (112) reflection as the highest intense peak (Figure 2) which is in

SOL-GEL THIN FILMS OF LaUTiO, "1

agreement with the Joint commission on Powder Diffraction Standards (JCPDS)

file no., 28-517. La 2Ti2O7 sintered at 1400°C, however, displayed (400) as the

highest intense peak, which indicated the existence of a considerable grain

c-ientatin in the ceramic. 13 For the La2Ti2O7 films heat treated at 8000C, the (400)

peak appears to be more prominent as was observed for the sintered ceramic. As

the structure of La 2Ti2 0 7 is composed of perovskite LaTiO 3 slabs of four TiO 6

octahedra thick, bounded by shear in the perovskite 11001 planes,6 the observed

grain orientation along [1001 direction appears to be the preferred one for the

sample to grow.

The microstructures of La2Ti2 07 films are shown in Figure 4, which reveal avery fine grain size of - 0. 1 pri. The formation of needle shaped grains with

visible orientation is evident in the film coated on Si(100). Since poling of

La 2Ti2O7 single crystals is considerably difficult due to high coercive field, the

formation of thin films may be important for device applications. In this series of

compounds, Sr 2Nb 2O7 single crystal films were reported on a b-plate Sr2Ta207 by

RF sputtering for possible use as optical waveguide. 14

CONCLUSIONS

Thin films of La2Ti2O7 were deposited for the first time on fused silica and

Si(100) at temperature as low as 800°C using a sol-gel technique. Thin film XRD

and SEM results showed the formation of grain oriented films along [ 1001 directionon these substrates. The present synthesis is simple and thin film of uniform

thickness and good quality can be obtained on various substrates by this process.

Ferroelectric and dielectric properties of these thin films are presently being

investigated.

REEERENCES

I. S. Nanamatsu, M. Kimura, K. Doi, S. Matsushita and N. Yamada,

Ferroelectrics, 1, 511 (1974).

72 A V PRASADARAO ct a&

2. M. Kimura, S. Nanamatsu, T. Kawamura and S. Matsushita, Jap. J. Appl.Phys., 13. 1473 (1974).

3. M. Kimura, S. Nanamatsu, K. Doi, S. Matsushita and M. Takahashi, Jap.

:J .A221. Phys. I, 904 (1972).4. P.M. Gasperin, Acta Cryst. Bf31, 2129 (1975).

5. N. Ishizawa, F. Marumo, S. Iwai, M. Kimura and T. Kawamura, Acta

Cryst. B 368 (1982).6. T. Williams, H. Schmalle, A. Relier, F. Lightenberg, D. Widmerand G.

Bednorz, J. Solid State Chem., 93, 534 (1991).7. i.K. Yamamoto, Growth and Characterization of Ferroelectric Single

Crystal Fiber Produced by the Uaser Heated Pedestal GrowthTechniQue (Ph. D Thesis, The Pennsylvania State University, University

Park, 1990).8. J.K. Yamamoto and A.S. Bhalla, Mater. Lett., 1-0, 497 (1991).9. i.K. Yamamoto and A.S. Bhalla, J. Appl. Phys., 70, 4469 (1991).

10. A.V. Prasadarao, U. Selvaraj, S. Komameni and A.S. Bhalla, J. Mater.

Res. (1992).11. A.V. Prasadarao, U. Selvaraj, S. Komameni and A.S. Bhalla, J..Am

Ceram. Soc., (1992).12. U. Selvaraj, A.V. Prasadarao, S. Komarneni and R. Roy, ,L Am. Cneram.

Soc., (1992).

13. A.V. Prasadarao, U. Selvaraj, S. Komarneni and A.S. Bhalla, Mater. Lett,i2, 306 (1991).

14. A. ishitani and M. Kimura,AM. Phy.. Let, 29, 289 (1976).

APPENDIX.- 43°

(Ffl ) II9'N1N It tt I4NI

It Katnkid'I' mid Cf. 0ady

Mlaterial~s Reseatrch I illoratoy. 1-1 e Pcmvtt'Ilvaittiat S;Iit I If I jiii'iif m% (.$,It cw P'.1Iik. PA lI Ohl12I klietimtwteiit of Ceramic L Umtmeer itig, Lttveis it) III KMoss'iia.tRoIIh N~ If 01SIMl

$$ Echo Ulltrasounod, R. 0ý 2. Box It 1, Reetisvdlle, PA I 71)94$$$ Tetrad Corporationt 12741 E. Catey Ave., Entgetwootl. Ct) ROI I II

Abstrac Lend magnesiumt niiohcne (PhI M~g tp b-ijh,,O0I.'NiN ) cihecked )IV vcimiitaluito the XRIM pattetil %'611 filh e JCPI S 'I:Iulstiiitlead titanitte tohrio.1 -PT) sot hi sollt itolts are widely resca rhealte to lcI ie MN -, I, Ikii te.ý %s ie pt' w elle I 'N t~alcmI! to t in alpf (lilt 1tieprodiuce devices that call lie used inl low avid high electric fnied mtixtutre (it fit(I Ii taiittooi lttit Iilols it, PIA) h. gNll,( t. midt *1 i l,aipptications. For somie applicationis, such as mtedIical tiiitt;lsitti (Wlimiake. ( laik. ;flit) I ).iiiie Smaiull 'l'miiiildCII NJ t poiduf% -ittranusdutcers, it is rnecessary to pmepaic lthe cermntic Withi high Atictisi v 400tW{.75h ill a clowc iliiti ctiima (ilk(11 Ito vadiy tile Illli;l axmieatatid smalil average gr :nll size. This paper dlest ibc thle etlic 1f 10, IltIle 67.IW (;I il~ht-It IIp'C dci ý - .iil'tICe iv ) Ili' I(h itlt Ii

graitsize miltile tow mid high iel 0%pe)esof PPOtlMN-i 11 Ill1 glillilinbg .5..Ilvi-ut Ic Iird t. p1C lumlir %%(' K s il'il voilo dItIIi at i i 'I

ceramicts To prepiate highly denise Ceramioc, villrattily alilt ;tttmifoo~ fii a ii.itgclitv tiottic i%,A lit / .I ~I, iit ilil-IiA Iii 2Ill, Altc Iti 1

mitlledl poiwder-s wvere smttered bietweetn 111001-I250'C'I Ilie tvreiare a pirlilirto Ivk~i flilll t IIC %%it vlllitiilv Viilil' I mts 1 2 Iti I i

ginuin sizes of lite sittietrd camitts vatted truth 1) 7 Iii 1 5 )iii Ii o i11tliiiiy ilIlig h illop iiii l~f %11" s111 1 t~~lpclllcith waitiWiCl mitunderstitid tile graint sizec effect, dielectric. pytoelectric. few (l1411" of I .1IIoiI 1)(II \%it lim-d;I, ;i a i~l~l5ivisil loIlllII lJCII11

electiostricii ye, wni itiirhced p iezoelectric propettfies wei c iiuliiil ciii oltI l .Siall Iyfilt .Id SiIt Ii l f cI 4)I sjici e ewC iC iS IIvt .1%

tttilli~ip tiulleia lIit tocreasr lit:n Sill 1.1cC ali~l lititlier, a nolall potiiillli

11n1r0duclioit of tilt! ibratoty m il led shilly wa. alltritii't tilill(ed lmicei filel mt

aitddiiiimal 2411 'i te ball mtilled il. tiatotir v milled, atilt aliriItimReciaxor ferroeteetrrc mlateritatls atre widtely stuied it for t tie, r mi illedt pmi o lerN %iC td mimctd ;Is kit cl A, II atiut (- respecirictl

dielectric. electiostrictive aod itiduced pietrielectric tirop~eriilit thIfie I lie 7.r( I, ciititaitiiiiitol miii cfi. ad[llite It FIi Nill la(e aica Ill leipast decade, lltese miaterials we, e itseif Iin a variret y fit ;ippl iv a imi iti po% tIc us are list C4l it) I al 'le Ir.tg itig fromt miulttr.layer caipacitors to uin t~soiic tr awsuivevr 1. I21 1 Ile irtd u iv ul %%;I,; piia l il : idt\Ililt h5 wt% (I-of I )JI' p i

*flitetichi et a.1.,121 itivestigated lthe effec o 01f.91)IININ-i IIIPT 52M11 hiiirlc tii atetottie tinciliai Alter pad-siig litte mitimti ttioiiigli ac'atillic for flie toed icail iiog iig itt lite 1 -3 votiligitatimu. llecattIve I00 ivi e s i ieve. 1/2 ul dia di k \%etc ciilt fI c ScAit tii a IC NIe lite wiviOfl ite hiphl opera~ting I'rlertIety, tife dimnseioniis of filte cylitiittk.al, 310.0110- 13,000I' 1111ps iiiail Iliessitie I lie Istdi-uk Ilitil th1 itisk. W;Iceramtic I. sisi mte resricterd ito titily aI few iells tit ntivtoii til all fetioveil by a to il st.1re heat iiaitCt1oroit at 311t('lIIaltdit Itit'ii mdutCL iMUS 'I o t taiiilIacimte such Ii psts rel iably atid re 1;MuucCP ~ Sh47~h Ilue s p1sch \i, wei C '11"iIi ciilli ;I clo isti ahloiii ii (i it ii 'lik iiecessary to use high itettsitv cetutuic Wiit smlall praiti size ati 10011-I250tCII ',it A otjtl attilitli of Illii) Zc( 1, itimc tir ad

The graini size effect oi i the low vidt hiighi field poiperties 01t heated along \isdlfilte pelkcit, to ilttol tile lead oxidle atitiipliereI'MN'-tased ceramics has been itivestigated by sevetal ittsde tile citicibte All file Nimitetei Samolilv% %%ere ;mitiedll itreseaicheisl 3- ligie'l iepp lie erd hegr t , IlSIt 1'211 II iiisaplturate xte.. I'l ') [ltie we ighlt hiws 'S 111ii ig ii n-eis rediited- For cxaim ple. inl I .93PM N-I .071t'l ce ranmic, ltet peak, plepar at tll slages \\i ee toilt ttii ci attilt it cldielectric constant drops fromti 25,011M1 Ili 500014 when thle meageu~p raiii size is rethuced fromi 5 pil iiit) 0.3 pt ltd S. Thle size et Iccis Iiia etia ii

oibserveid in iliq In-e rittic were miti iihted to sIite e t insitdillc aIititritiSic effects stch as pote volumite. low-K graiti homitirary phidc. '111w 111lk1 itetIslit- $If lilit'tiMIicirl , 'Illti IM II ICIV t iltiih.iitrmtid mittrnopila dotiiil rhetisnes. Fromfit:itn literattite sorvey. it 11y rte;uilmitp rthe %weight tiatiipc iii \vtvcuu 'It lie g to slC.1ajieasittat tile lirolicities of lthe title grainreralilic ettli lie Itiiprrvt-il ir ifiiiiiulitiol a rs;141.ii1N( ;crl Ilom Oit -the ititiIlq osi i titesti rfictitivt millby miiiitioiz it ig tlite effecti of extritisic variables atit tulby mitttrtv ii il t:i poi it d is Istll di s IIi.liii %i111tcii t' et! (cv iciile IIusii i IS Ihortuogeiteity. lIt 3`11) Sevooudaiiy Oct troll sltittr'istiil

hl lthis paper. [i e effect of averalge pratmi sizSe lilt the Ii w midii I Mta iled rX~lI tItItI) dll 'c III-% 111114miif diiiitlettiti. (11oc le( itthigh electric fielid propenies of 0.9OPMNN-f I 111 ceramlic is l'-L hysteerlNi, antil ec(ler o tistil y f tcsiritive caill le Ioiiii fiiireported. Ceramics with average grinii sizes ratipitig from 01.7 tol 1 5 Ref.g. Dieclectric ttaiedtielilit% wetc matte itit goiltl ,likililevid ilistPitt were prepared froim vibratory muid attritiot mtilleid podis'tILow field propenies suich as diielectric cotistaiti antI hipgh fieldproperties- such as P-li hysteresis, fsyroclectric. elecrtiostirctive a:tti afew (if thle electric field imtidoed resuitltice liuptifiriics, weictimemigatert as a fuctoin-mt of grain qize. I PeItroitelit ns u Powdvr

Li~urtiitZi, ~ A Ci (

IIIIp~t)ritv (ssr'A I 1167 1t.71IIh

Ton prepare tile ceramiic powder. a coititimliie preetiusir It 1:. '1fl'/illediti, iiie was arloi ptd( 71, Tle ii m-uieshimt i tjhaie p1Cctiirmiir %-as% Sitl late Airca I 2() I 7 I .tplepated by calvittin? lthe auppropirtat tistuiilre of N~gli :. liaket,Phlliptitsburg, NJ) ;toit( NbO5('rmuselco. NJ) ato N4T atlid fielteat 1 tt0f)C/5Sh. 'the colitiletior of ileactiots antd lthe phase pirity were

Ci 130 8(10-7803(4A65 -9/¶2S3.0( ©IEhE 148

pyriiele irK'ic ii~l~is tile CClictioded l SiiiiI" "ere Cooiledf li%, I, :## V.

(toot it)'(' it, -75*'C Wilit ail cccivi field tit Ill KVkcii AeIr 1*. ~5vieutral izlg tlie -irac Iie Climpe(!. itile p>tce ti Cluffelf W ll'6Is ..iileasulledl by heating [lie s;pecimoliti 50i SC at4C/mioi. Uor V- , ".,i-q

li> Steresis ;nd ltrals' cis elemosfi'iflv K ilea ii~surceoiliiis. all ai I. A . *

I IimI rgular fiel ItI(f ± It0 KV/ciu I wIs applIeI d ;11 af tit lucn-v (if II I 11 Z L / l

technoitue. By differeittuatingp E. Vs. -,, 1111is. 't~lt Values as a ,

f! liol (in f electriic- field wete Caiclihii td4AC itnipedamie oleastwineiviits " ete perbittimed til irmular -P'j I i ' I*v!

disk -saitipte (if i< t uiimin %il kiti ptlere~t silver elecrioles j. aL~*~Pesoiiaiice enilv (esif Coii d i ci aice ( i. ll I esi ialitee R of tIile rjt*- -qaitiplles at vatimis iiS iases %%ec teI ordeiidl~ Willi ml itII 4 11) 9J "Isimpu1edanice :iiidlV'?et 1 25('C Sceties anid parallel Neu~wc __

frequlencies f, nod I (if thiickiiess fresuiioice modi(e were userd toi W&calculaite k, Willi tile equationi Vi'eii oin Ref 9

Results ;uid Discussiol C1 C.7

fthysical Npitrvaie

1hle desisnv atidl weight lrvs, data of batch 13 antI hatch C Fig. I . Fractured surfaces of C I and C7.samiples are lislt]d ii Tiblle 2. lit geticetal. samiiples siiteried at lowertemtpeiafures showu sliglifly fiigher detiIsify dfie lit lititted Frainl

growthi. Amon~g the two hifelics. aý tire initial aiverage parlicle sizeof C wais lower. it sititered fol compi~aratively hipgher denities. Al 1I abhe3. Dielectric I'ropcriievs of tile cetaivliChig her qint er inp letoieratures. as file g ra ii growth i tie clini-mi-iIs---___________________ _______ - ______

do:nitiate. brith bat~ch -iiuterl ill siinilat demsnies Whille batchingp g . s Kiuax 1 Illax Tan 8 5tile Pow~ders for C lic matnioll 1.) ftio Cire filie C~ioi p let iol (if renat iol li. ull T at 25*C ICC0.5 vt'7- exces P110 wvas aidded Bll aIffer milteloiig mnid 'anneal illg.higher weighit losses wete obs;erved lTible 2) As replorted by Walig B32 I 1 t7901tH470t 48 004582 41)

fli~ Si~tI~. u ~iipt uracs lo hoedasii~Iituilber of 11.3 l .2 222511H~)S() 46 0.0686 38pyiochlore grainis Sinice the pyrochliire is lead deficieti ill iiature. B4 1 3 2175t)±72h) 46 0 M19)1 34excess Pbo) is evaplorated 'shle ailticalitig 115 1.7 253)()±6(l6ft 45 0 0760 31

'Ilie polished and fractmted ilicrristruclures of a few oh' the 116 2.2 28854+±70(1 41 0 O 1)1)9 27sintieled saniples ire ieprothited inl Fig I 1 lie twcait average gtritl B7 3.3 331)011±550 42 R0(922size-s Inr all ivte samlples are I isted %% wit -Iids jard de viat ion in 'I ;di 2. -l D 1.7 13900+11300 49 001(474

(U2 1 .0 20900(±3601 41 11 !!11 : 33(11 1.4 21.150±5 10 42 (1.072 32C4 1 5 247(H11±420) 42 V110813 32C5 1,9 2545R1±8 I1) 41 4) 07945 i(1

[aible 2- Phbysical Piopetities of thle cenunic. C6 2.4 251810072111 41 0) 1185) 29_______ -__ __ _ __~--C7 3.6 27450±3(X) 40 0.0991

Si. Tell wl loss Density % 1 lico Ave.*C/11 % gui/cc g.5. (pinl) -___ ___ ___ __

112 10)5tM 1 .10 8.001±11(2 98.0)6 1.1±0.2Ili I IM1)/ 1.10) 799±00? 97 94 1-.2±013 _____________________

04 115011/ 1.11 7.99±11.02 97.84 1,3±013 40000)B5 1201)I 1 1.16 7 96±W.03 97.57 1.7±01.1136 1250/1/ 1.20 7.90±11.02 96.R4 2.2±0,8B7 1250/S5 1.32 7.83±0.03 95.98 1.3±1.3 fCl 1001)/I 0.8R5- 7.92±0.0.3 97.0)8 0.7±01.2 30000C2 10501)/ 0.99 8.00±11.0 98018 1.01±0.3C3 1100111 1.0)3 7.99±0.1)2 97.94 1.4±01.4C:4 11501h 1.4)6 7.99±4 (12 97.82 .±5C5 1210/Il 1.0)2 7.94±01(02 97.33 1.910.6200C6 12501)/ 1.13 7.88±11.113 96.59 2.4±0.8 OC7 1250/5 1.29 7.80±0.03 96.51 3.6U1.8

101000 -- -'

0 1 2 3 4(;raiii Size P~iti

Ftg.2. Effect (if aveirage grain sire not K .. U. he dietilectric colstamwas comlpenisated for pmio~sity witlh Weiner'q rule,

149

Low Electric Field Properties 0.3 .. . I" C2

The relevant dielectric properlies rue compared in Table 3. -i 2 C3Al least four samiples were used Io calculale the pmerceihage changve E

in [fie dielectric coonslant. 'use maximunm dielectric coosrtaitt KI 1 ~ I, L), 0.2 2 3- >1 3 C4

after compensating for porosity with Weiner's equthimill01, is -%-. 6 4 C5ploted in Fig.2. as a functiort of average grain size. Conmpmalively. 5 C6the Kmna of the samples fronts series 13, changes as a stfong funtion .o-

of grain size. !it botlh batches. K,,, reduces drastically whe) ithe 6 C7

graits size is < Itin. 0.1

lHigh Electric Field Proenrties o

The effect of grain size os the P vs. T hehavior is compared (.) ..in Fig. 3 amd 4. To plot these figures. lite polarizatiosi was calchnltied --5o -25 1 25 50

front the pyroelectric current. It relaxor lerroelectric maiterials. P Temperature (IC)vs h behavior below the pyroelectric depohauizatio=l temperature. 1;i.

reflects the degree of cohesiveness amiosng sacrmdomahsis. Below

Td, if the macrodomains are lhighly slatle. Al.willh respect to A Fv"

will be s11ai1. When the Inalte'ianl is he)sted, tlhe st;Ale

macrodomains will convert into microdntmaits, over a nari ow

temperature interval around T&. In our investigation. the samples

sintered at higher temperatures show these behaviors very clearly. 60

When the sintering temperature is reduced, increasingly large

chaunges in P with respect to temuperanture is obs~erved below Td _500More over, in these samiples the msnacrodoomiois trai•slorm into

mictodoasiains over a wider temperature around Td-. Careful Z 40)0

tnal ysis of P vs. T1 bhelav itr pm ims out the insolaonta ce of psliser rprocessing, li series C samples. as the grains were growr frm i 3(0finer particles, tine grail si7e ClieCl is toinioliied.

The induced polarizatiot and the transversC s6rai2

characteristics of the fine graits samples trom I)(h1n hatches show 20 "2

degraidatin of properties. Illis otservatiolt is cleally (lettl'tlstaled -- t 15by plotting -dlt* vs. E field, as a functiots ol grasm size (Fig 5 a.aid 100 3 3 U B

6) These measutremsnests were madIe at 25"C. Whets the gmrinsize I .7

above 3 pmn, a maxinmuni -dII* (of about 50( p(C'/N is observed.

When the grain size is reduced, lhe ;ic n gisiude of maxitmum -dq* 0 4()()() 90010 12000

reduces with an inscrease in the correspontding electric Iield. E, Iield (%V/CN )"Tihe piezoelectric thickness coulnling ciseficiemns cAlcuilated

froin the resonance cusrves observed at dilierest i c. hiases are

ploated in Fig. 7 :snocl 8. 'these meastirements were made at 25"(C Fig.. F'flecive tra.,ssvet se piezielectric d co-tficictist of hatch B

willt increasing bias voltages. In hoilh hatches. coarse praiised r5ii)les calculated hom elcctrostrictiott data

samples showed itcreased coulpling at lower electric fields 'lihe

absolute values of coupling coefficients also increased as the grain

size is increased.

0.35

E21 B33 400

-U 3 U4

u0.2 44 300

I.6 13 C7

N. 536 ~ 200613

10 , ("6

,0 3000 (1000 90()() 012000

0.0 FE Fiel (V/CNM)-50 -25 0 25 50

"li'eiperature (°C)

Fig.6. Effective transverse piiezneleclric d coefficients of balsc C

Fig.3. Polarizatiots vs;ý iet perallne helasvior 1sf13a1101 B ) sasampless calculated from electrost riction data-

ISO

III thvis paipcs. tile diele lift. pyrti-lc~tftK.kt.elctomtti atne. mt

OA4 lIite i illciltincl Ilinolect Iic Ilmpsnt ics uif (90' VPMN -I I ft I' 1 seramiic'Il C uiiIadied 445 .i1w ii. In ot(f glai an size Conipalvat ivel y samlples

- p~~~Ifella red Itutiti coat Se i art is e Com tpact IShow hiigliver gra in size0. dept-ndallce. wellti the average grainl size k. arouild 3 S pit.

if resp1ctlive of life inlitial powder charIacter islims file cerailic Show

0.2 B2-- similar high electric field properties. Thruough) transverse strait,* 115 mlcamlitt-tletl. aI high -t I I of(1 50th pCIN at a d c bias of ON)(

0. 1 s-- 137 V/cu is oliserved When life qamples were preplared fronm very finelIwwdcrs. file ittagiiititodes (if K,, *dj I*. mitt i, teduces only

0(1 modmisleely w; five graini size is redisced to 1 5 lim. Since lthe effect0I 200011 400 (to 1I 0111 800 (1 000 il~)( l extritisic vat jahiles such aq dentsity muid tile gyaiii houvidary phase

F. Fietld (N II are mittiiiized through careful preparation. life observed grainl size

Fig .7. 1 Ibicktiess S muipi ing cot-I Init-icis fit halth it %mtiplt-s :it(ilhercii l. 1,C. hias fitt-da.Z CohulattIae I ton, rcsuilltic (III i'C;

Ackticm ledgefi eitts: Fitiart-icil 5tiipptt fromT Echo Ulhtra~sou nd II

Iteckville, PA mid lie flerv Franiklini Partnershiip l'ropaim of file

0.0tiittiwtiiveahili (if P'A arid I echitiical support f.ront Mingfoilg Sontgarc grateltflly as kiiowledged,

Refefeaics

0.4-III IN Y, Pali. W, Y (;u. 1). J. Taylor anid L. E. Croq% Lamge

P'iezoelectric effect iniduced bly l11rect Current B.Jias il

C I INhTT Ib-lAsor Ferrocleciric," Jpim. J. Appl. Phys.. 4.0.2 p053 (1 981)

A- C5 121 11I. lakeuch.,. It. Mvasuzalwa. C. Nakaya. and Y. Ito. "RelazorC7 Ferrorlectric Transducers," 1990) IEEE Ultrasonic

Sytuptis iuit Aommuic. 11697-70)5 (1990).

(LI 13 K.(Iciln. N. *aistiiii. 1. lIaasiii. and 11. 1 layashi.Jpnl. J.

0) 2000 -1000 I I (100 WOO$01 App~l. I'hys .. 24. 733-735 (1985).

E. Fiedd (Vicm 141 T. R. Shirout. U. Kutimm. MNi eghieri. N. Y.-ug. will S. J-Jmig." Graiti Size lDeh)Cmt(atce of Dielectric and

Fig .8 'llhcis stuIp)lung cut-h icict'.1, tih liatlclt C s-artiphesa Eleciroisitrition of PllbMg t1 Nb21i 103 -based ceramic,"difflf-if . C hi'; iels. alk lald lvil leillicecm~,;ýFerruehect rice. 7(),479-85 I 1997)

difer-utd.c. ia li-hs: ~aciiatd hoit Isol~tic cuse.. 151 Pi Peppet. irP. Dougherly alid T.R . Shrout."Panficle randGramn Size I~flecta# -it lthe Di electric Ijehiavior oif lthe Rela.,um

biscussioti leriticietrinc l'h( Mg l/tNli2/i0i. I."J Maier. Res., 511 2).21Ml2-2909 (199(m) (mid tile reference, lthre of),

A% meitteinied in file inttimbiwtiim sec~tion. tile granit size 161 It. C. Wang. and W. A. Schultze." lThe role of excesseffect onl tile prriwt itsti" i ma lie ailtrillmed~ I tII e stuiriic mid i hiSIC in, ag tes imvit oRxide mrid lead ox ide in de le tittittig tilevat iabllts. liit flie pit-stinai ligi ll'lt-. ti iii If cllii aunt-al lug tile n icto-inst amc li fe md 1 jiprfi qwr fie r leatd titgiteiesmi

CeCCl lec Csturis)IC variables1 atIC t111iiittiiiIc lot exanipic. 112. It 4. tijubate.- J, Amt Cet. Soc.. 73141 825-32 (19901) (and ltheB4 mnd C72. C-1 Of4 'ampleq w itli miitila~r w-eight hisses mind references ll~eve of)iherelrme Witlli %imnim Wa sulfld I il ase Viii t o$Tt at fi it-gia illho I Ins'dar. 17! S. Swart z andtI. It . SI itiitt "Falinicar int (if Perovsk ite Lead

tliow i it ice attie dii Icrmc~ts ill fIile 4m omx hn itghi electific f ie-ld Magnesia N inhate.' Mat. R"s DO-l 1.I7. 1 245~-1I251) (1982?pitipertlies. Az observcd it All live Itlilliet It- latge granti Sattillekt 191 U.. Kumtar. L.. E3. cross. andu A. I hliliyal." ryroelectric andWithi lower dleivgit V sbltw bettrlier ploefties as Cutttillared to fite Oralt electrostrictive piopieniea of ( --i y)PlZN.xs'l'.yl'T' cermiitcsaiilples with higher 4tietiitis solid solution systeit,- J. Amt. Ceran.. Soc.. 75181, p-2 155-

Care jutl Aan 0ysi.; fir live tile lows ail hi tig h electrtic field (A ( 1992).ptuilerties alniw cler dkistinction btwweeni tle proplerties if series It 191 "IlRE standards ott ptiezoelectric crystals: Measurtements ofandsre lape.Sil dpesii i i, nd*ll series C piezoelectric cerartit.l 961 ". Prtoc. IRE. 49. p-I 1161-1 169

suat111le1 nidy IV alt rtliated tuo file ItigIiver coilivetii:atiioll oif 7.rt 1 itils I 91m)0%; doIpaitI I II Uririider ing tile ;ivetag'e g aml .sis ofi .1ll ticse 110)1 D). F. R isstitan. nrid M-. A Strivens." The effectivesattliles. fralitimtg ftomi appilttimsiitely 01.7 Ji to ittly 3.5 jiii. it peniiittivity of two-phase systeniir," Proc. Phys. Soc. 59. p.appears that1 tile 5 illietitig tell )pce ra~ni lti le lml niWetce% tile propeltle. 1011- 10116 (I1947).itritn thle suerface aica ofi thle milled gIllmdterg, thfe e(~ilivaletit h fill 1)J. Vpas. S L. Swartz and T. R, Shrtout." Pytoeleciric and!Chlttical dialimctsr fiiffilet s-i ics It ;tld st-ties C ,tiiwders me. Ilectrn-siticlive tnopertlies of II -site mtodificatinits ottcakotlated a- 11)23 put. aml I 1)047 jtiii reslieclively Comitparinp file dfielec-riC anid Ce~letOStriCtiVe itroperties of lend magnesiumgraini size.;. it lia clear tlt.11tfile "let ts C Naltilil~e Olots at least 15 nitohlat cer~uties," Fersoeleclrics. -5(0. p-203-208 (1983).timites grotwthI (ti ritig slimtitinlg .Itt-cat ue uif lbiii growtmb, it ishypoiitlliset 411at Itt-se saitlliles ate tiii te holifntgetllcos m ciii llattedfto seriest It sartiile5 I Pemce. file lesser itiflinerice (if lthe average Fraintstze nil lthe low atnd htigh field piroperties ofi %eries C sampijles miay bealt ribilned to h ighier dcgce fil I it114itc iiigtihty.

I51

APPENDIX, 44-

Extrinsic contributions to the grain size dependence of relaxorferroelectric Pb(Mg1/3Nb2/3)03:PbTiO3 ceramicsC.A. RandallMaterials Research Laboratory. The Pennsylvania State University. University Park, Pennsylvania 16802

A. D, Hilton and D. 1. BarberDepartment of Physics. University of Essex. Wivenhoe Park. Colchester. Essex. United Kingdom

T, R, ShroutMaterials Research Laboratorv. The Pennsylvania State University. University Park, Pennsylvania 16802

(Received 16 July 1990; accepted 24 November 1992)

This paper addresses the observed grain size dependence of the dielectric behavior for

Pb(Mgl/3Nb2/3)O3:PbTiO3 ceramics grain sizes ;•l.0 A.m. A combined transmissionelectron inicroscop•' (TEM) analysis and dielectric characterization are modeled with amodified brick wall appioach. From this model, it is possible to extrapolate informationsuch as single crystal values of dielectric maximum, Kmn., the diffuseness coefficient, 5,and the average intergranular thickness for relaxor ceramics. The calculated intergranularthickness agrees well with TEM observations, =2.0 nm. This semi-empirical method maybe potentially useful in development work of relaxor ceramics to predict the optimizeddielectric properties obtainable within microstructural restrictions.

I. INTRODUCTION processing the columbite precursor, the poor disper-

Complex lead-based perovskites. with the general sion characteristics associated with magnesium carbon-

formula Pb(BIB 2)0 3 , exhibiting diffuse frequenc) de- ate powders were addressed using both steric hindrance

pendent dielectric permittivities are commonly referred (polyelectrolyte dispersant) and electrostatic repulsion

to as relaxor ferroelectrics. A typical example of a (pH adjustment by ammonia) in conjunction with high

relaxor from this complex lead perovskite family is the energy milling. Upon calcination the appropriate amount

Pb(Mg10 Nbm)O 3 (PMN) compound. Solid solutions of of PbO was added followed by mixing in a dispersant

relaxor PMN and the first order ferroelectric, PbTiO 3 and adjusting the pH to obtain minimum lead dissolution.

(PT), exhibit many attractive properties for dielectric Calcination at 700 *C for 4 h produces the desired ingle

and electrostrictive applications.' The high dielectric phase perovskite powder. Uniformity and reactivity of

constants, over broad temperature ranges. close to room PMN PT powder were further enhanced by milling.

temperature, make systems like (1 - x)PMN: xPT (x = Grain size variations in the PMN - PT ceramics were

0.07) very attractive for multilayer capacitor (MLC) and achieved by firing samples at different temperatures and

actuator applications. In several previous investigations times, as reported in Ref. 3. Significant grain growthfor the dielectric pioperties, including K,,m, diffuseness through longer firing times was not found to be effective.

(lie effective width of the maxima) and aging, have The dielectric measurements of ceramics were car-

been shown to be dependent on processing variations. ried out on an automated Hewlett-Packard System. The

in particular the role of grain size on these properties. dielectric constant and loss were measured on cooling

These earlier investigations did not lead to simple inter- at various frequencies (100 Hz, I K~z, 10 KHz. andpretation. partly because of the presence of second phase 100 KHz) at a rate of 4 °C/min.pyrochlore. The present paper successfully interprets the Transmission electron microscopy (TEM) character-structure-property relations in (1 -s )PMN .u P Te x t ization was performed after dielectric characterization.strutur -eratics, lation s good roesin g (Io -et xPMNxP = Slices were cut from fired disks using a diamond saw0.07) ceramics, using good processing control, dielectric followed by grinding to a thickness of -30 Atm usingcharacterization. as d transmission electron microscopy silicon carbide. A copper support grid was glued onto

(lie surface of each polished specimen using epoxyresin. Electron transparent regions .%ere obtained by ion-

11. EXPERIMENTAL beam milling the samples with an Ion Tech argon beam

Polycrystalline ceramic disks with the relaxor com- thinner operating at an accelerating voltage of 3 kV

position. (I - x)PMN:xPT (x - 0.07), were prepared and at an incidence angle of 15-30'. Finished TEM

via the B-site precursor method.2 Both reagent grade specimens were coated with a thin layer of carbon in

and high purity powders were used in this study. In order to pre ,nt charging within the electron microscope.

J. Mater. nies., Vol. 8. No. 4, Apr 1993 0 1993 Materials Research Society 1

C.A. Randall et at.. Extrinsic contribution•s to the grain size dependence

TEO observations were made with a JEOL-200CX 25000 ."microscope and a Philips 420 STEM. both equipped ""6with a LINK systems energy dispersive x-ray spectra- ` 20000meter (EDS). z 4L

3/3

Ill. RESULTS Z 15000-0A. Physical and dielectric properties U_

The eeneral characteristics of PMN:PT ceramnic E 10000,SI.- 2,usamples as a function of thermal history are summa- U

rized in Table 1. All the samples possessed densities -J 5000greater than 95% of the theoretical 8.11 g/cmn. The 0weight loss during sintering was confined to the range 0 . 1J 1 Iof 0.5--1.5 wt.% with slightly higher losses occurring -,53 -35 -15 5 25 45 65 85 1O5 125

in samples processed at higher temperatures. Annealed TEMPERATURE ('C)

samples and samples fired in BaZrO3 sand showed only FIG. t. Variation of the dielectric constant temperature dependencemarginal differences in weight loss. It should also be (10o liz) with a grain size in PMN:PT (0.93:0.07) ceramics.

mentioned that x-ray analysis of PMN:PT ceramicswere phase pure with no evidence of pyrochlore or PbO Figs. 2(a) and 2(b), respectively. A typical lattice fir-phases. ing image of the grain boundary phase in a coarse

The magnitude of dielectric constant K,., (0. 1 KHz) in img ofte ranbudyphsinacreT K grain reagent grade sample is illustrated in Fig. 2(c). Asversus grain size (gs) is presented in Table 1, and is shown. {100} planes (0.407 nm) have been imaged inshown in Fig. I (both from Ref. 3). For all ccratnic5 adjacent edge-on grains, clearly showing a discontinuitystudied, the K,,, values are relatively high compared between the two sets of lattice fringes. The thicknessesto those reported in the literature; this is believed to of those intergranular glassy phases were found to bebe owing to the overall care in processing. Clearly approximately -2 + I nm with little measurable vari-the value of K,,, for a given frequency increases and ations in thickness either along lengths of individualthe diffuseness of the dielectric temperature dependence boundaries or throughout each sample. This was alsodecreases with increasing grain size. found to be true for samples processed with excess

PbO addition. It is important to point out that withinB. Microstructural analysis the bonds of our processing variables a grain boundary

Intergranular boundaries examined in samples pre- phase was always found ubing these standard techniques.pared in this study used standard TEM under focus/over This includes the starting powder purity, namely fromfocus bright field Fresnel fringe contrast and lattice reagent grade to high purity powders (;?99.99%) andfringe imaging techniques.' Under focus/over focus the firing of stoichiometrically batched PMN:PT underFresnel images of the grain boundary are shown in conventional and hot-pressing and also after annealing.

TA BLE I. Processing conditions and general characteristics or 0.93 PMN : 0.07 PT ceramics.

Firing condition wt. loss % p (g/cm3) Grain size Krm T... C 6 (C)

950 "C - 0.5 h 0.5 7.4 1.5 '"m 11500 24 58- 41h 0.7 7.83 2.0 jam 13000 22.8 so-48 h 0.6 7.88 2.0 jam 20300 -,. 39

1050 "C - 0.5 h 0.8 7.88 2.0 Am 12600 20.6 57(Ztrconute sand) (0.97) (14800) (47)

- 4 h 0.5 7.91 3.0 pMm 15700 20.5 44

-20 h 1.1 7.89 3.0 Im 18800 22.2 43

1150 *C - 0.5 h 0.7 7.87 3 .0 gm 13500 19,7 48

(Zifconaie sand) (0.87) (15100) (47)- 4 h 1.0 7.88 3.0 gm 17800 ... 43

1250 'C - 05h 1.1 7.84 5.0 'm 17500 .. 44

- 4 h 1.2 7.82 6.5 Mum 20800 18.4 42

-20 h 1.45 7.76 9.0 Mm 22800 11.9 39

1300 "C - 0.5 Ii 1.1 7.82 6.0 Mum 21000 18.7 41

2 J. Mater. Res.. Vol. 8, No. 4. Apr 1993

C.A. Randall et al.: Extrinsic contr~butvons to thre grain size dependence

A direct chemical analysis of the grain boundary phasecould not be obtained owin; to the small scale of theboundaries, However, the phases are most likely Pb-richdue to the low melting pci. of this constituent and therecent SIMS study (Secondary, Ion Mass Specisoscopy)byNWang and Schulze. 6

,,F-. rSecond phases were found at all triple points in the- ~ coarse-grained materials prepared from reagent powders

regardless of firing condition or composition (excess upto 2 at. wt. % PbS). A representative example of triple

* **- ~junction phase is shown in Fig. 3(a). The EDS analysis* revealed that the second phase was rich in lead and

* ~contained impurity elements, Al, Si, P. and S, as shownin Fig,. 3(b). This result is consistent with the finding ofGorton ea. 7 As the triple points are continuous with

(a) the grain boundaries, similar compositions are expected_______in the grain boundaries. Similar "dirty" triple junctions

A 0

'50Onm

(b) *'17 l5Onm

e ~RQNDARY ______________________

P b

FIG 2ý . W~). Undcr-overfocus micrograph images of grain boundaryphase in FMN PT ceramics; reversal of cornrast implies presence of FIG. 3. (A) Glassy phase found at triple point in reagent gradegrain boundary phases. (c) L~attice fringe image of grain boundary PMN: PT ceramics. (B) EDS spectrum from triple point in reagentphase in PMN:PT ceramic. grade PMN: PT.

J. Mater. Res., Vol. 8, No. 4, Apr 1993 3

C, A. Randall et al: Extrinsic contributions to the grain size dependence

were observed in all reagent grade ceramic samples. We the relative importance of this effect compared to otherdraw attention to the rounded grain boundaries at the possible intrinsic effects, we need to consider the di.triple points, suggesting that a phase was liquid during electric permittiviry's temperature dependence at fixedsintering. Triple points of high purity ceramics were frequencies, as this allows us to also account for thefound to be clean and free of any second phase (Fig. 4). influence of both Kmu and diffuseness with grain size

in the relaxor systems.IV. DISCUSSION In some ferroelectrics such as BaTiO,. there is a

From the microstructural studies given here, we ex- spontaneous strain -1% at the phase transition. Wvithin

pect the observed continuous intergranular glassy phase a ceramic each grain stresses at its neighbor at the

of thickness -2 t 1 nm to reduce the dielectric constant transition and causes changes to the elastic boundary

of the 0.93 Pb(Mg,, 3Nb3)O :0.07 PbTiO ceramics. conditions of each grain. These self-induced stresses

There may be slight variations in the thickness of this have an effect on the dielectric perrittivity of thegrain boundary phase owing to different crystallographic BaTiO 3 grains. This is especially important in BaTiO 3grai bonday paseowig t difernt rysallgrahic ceramics with grain sizes -1.0/.,.m. as discussed inorientations between adjacent grains, but there is no ceraic in sifes -1. In as dicssed inevidence of an inhomogeneous distribution of a glassy detail in Refs. 9 and 10. In relaxor ferroelect~rics there islead phase. as found in modified Pb(Zr,iTi)on ceramicsg s very little spontaneous strain, and, hence, intergranularIn general, a ceramic in icrostructure is a mixture of stresses are assumed to be negligible. So a relaxor single

parallel and series intergranular boundaries. The series crystal dielectric permittivity is comparable to the idi-boundaries in this system will dominate the dielectric vidual grain electric permittivity for grains ?1.0 tm.constant owing to the small volume fraction of glassyphase existing and the amorphous nature, implying a So Kt - K= where K' is the crystallow dielectric constant relative to ferroelectric grains, dielectric constant. (2)

Hence, a series mixing law can be applied to thisproblem in the form: The temperature dependence of a relaxor single crystal

dielectric at fixed frequency can be described by theI + where R =-) (1) semi-empirical relationship of Smolenskii":K K' RK'b (tib) 1 1 (T - T,..) 2

and where KI = dielectric constant of grain. KZ K- Z 2+6K,-,

Kgb - dielectric constant of grain boundary,(0i) = mean grain size, and where K' = dielectric constant of the single crystal

(ib) _mean gra,, boundary thickness. relaxor,

Kinx = maximum dielectric constant of singleQualitatively. we can expect a relative change of dielec- crystal relaxor.tric constant owing to the variation of the ratio R with T"=. = temperature corresponding to dielectricgrain size according to Eq. (1). However, to demonstrate constant maximum, and

,= the diffuseness coefficient of single crystalrelaxor.

The temperature dependence of the inverse dielectricconstant in relaxor ferroelectrics does not follow theCurie-Weiss behavior in the region of the K,,., inrelaxors.10

Combining Eqs. (1), (2), and (3) allows us to modelthe average dielectric constant K of the ceramic relaxor

1W A as a'function of temperature:-. .+ ,,

11- ~ / 1 1 .1 +.IPT -Tmu)M- i- 4L (4)

J • , ,f : ,• By equating (3) and (4), we obtain the dielectric mixing". ~ . •,*-,." law of the ceramic Kin.1 as a relationship between the

, , .';.2OCtihi" series mixing of single crystal dielectric constant and

, , • grain boundary dielectric constants.

- .. . . 5- . - 1 1 1 1FIG. 4. Clear iriple point in high purity PMN:PT ceramics. Kgmu K;= R Kgb (5)

4 J. Mater. Res., Vol. 8. No. 4. Apr 1993

C.A. Randall et al.: Extrinsic contributons to the grain size dependence

In addition, we also note that there i, a constantdependence of the diffuseness Kma, product for agiven composition of relaxor ferroelectric at a constant 1.9-density. 1.8-

S8 =. 1.-.-(6)8x K=. : (6) 1.6 -"

Hence 82 K.,a is constant for all grain sizes (11t.), 1.5Tprovided the only variations in 8 and K=,, are owing 1.4to an extrinsic grain boundary phase. For the PMN:PT 1.3-(0.07) ceramics studied here, we can observe in Fig. 5, 1.2.that 62Km,, is constant for a broad range of grain g 1.Isizes, suggesting grain boundary thickness and grain size E 1.0dielectric constants extrinsically influence 82 Km1.. Using 0.9

a plot of this type, we can extrapolate the single crystal 08values of 8x0K.7-. O7-I,

To complement these results plotting I/K11 , vs 06 3j11(t.), as in Fig. 6, additional information can be ob- 20 X 103 KpmNtained from Eq. (5). The intercept of the y-axis gives O.4

l/K,',, and the slope corresponds to (t-h)/Ktb. Us- CA

ing typical values of lead glass for grain boundary 0.3-

dielectric constant Kb , 20 predicts an average grain Q2boundary thickness (tgb) =_ 1.8 nin. This theoretical pre- 0.1 1diction agrees very well with the TEM observations 1.0 2.0 3.0 4.0on these ceramic samples -2.0 nm ± 1.0 nm. Using I/GRAIN SIZE ,urm-the intercept of Kx,_=- 29000 (at 100 Hz). we canpredict a corresponding diffuseness coefficient 6, = 38 FIG. 6. r ariations of d/Ke, 1 and grain size for PMN:PT reagentfor single crystal PMN : PT (0.07). No single crystal dataof dielectric Kma, or 8 are known for single crystal ACKNOWLEDGMENTSPMN : PT (0.07).

A technique for characterizing K,,,.u, 6. and (gt') We wish to thank JoAnn Nlantz for typing this

from dielectric data and grain size information is demon- manuscript and Beth Jones for technical assistance. This

strated for an important relaxor ferroeectric ceramic, was partly supported by the Office of Naval Research

PMN : PT (x = 0.07). This methodology may be of and the Center for Dielectric Studies at The Pennsylvania

use to the industrial ceramic processing engineer when State University.

optimizing the properties of a given relaxor system in REFERENCESwhich no single crystal data are available.

I. Thomas R. Shrout and Arvind Halli'al, Am. Ceram. Soc. Bull.66 (4), 704 (1987).

2. S. L. Swanz and T. R. Shrout. Mater. Res. Bull XVIII, 663-6675.0o (1983).

3. T.R. Shrout. U. Kumar. M. Megherhi. W. Wang, and S.. 1ang,4.0 Ferroelectrics 76, 474-487 (1989).

4. D. R. Clarke. 3. Appl. Phys. 49. (4) 2407 (1978)."Ky- 5- 0. O.L. Kraneek. T.M. Shaw. and G. Thomas. I. AppI. Phys- SO.S 30 --- K a'~erll

_ "..a 6.(6) 4223 (1979)." " 6. H-. C. Wang and W. A. Schulze. J. Am. Ceram. Soc. 73 (4). 825I* 2.O0- (1990).

7. A.. Gonion, J!. Chen, H. Chan. D. Smyth, and M. P. Harmer,1.0 Proc. 6th IEEE Int. Syrup. Appl. Ferroelec.. 150-153 (1988)-

8.- . C. A. Randall, D. J Barber. and R. W. Whatmort, L Mater. Sci.

21. 925 (1987).0 1.0 2.0 3.0 4.0 9. W, R. Buessem, L.E. Cross. and A.H. Goswami. J. Am. Ceram.

I/GRAtN SIZE pm-1 Soc. .9, 33 (1966).10. G. Arlt. Ferroelectrics 104, 217-227 (1990).

FIG. 5. Variation of 6-Kma, and grain size for PMN :PT reagent 11, G.A. Smolenskii, J, Phys. Soc. Jpn 23. Suppl. 26 (1 70).grade ceramics.

J. Mater. Res., Vol. 8, No. 4. Apr 1993 5

APPENDIX 45.

journalBarrier Layer Capacitor Using Barium Bismuth Plumbate and

Barium Plumbate

Basavaraj V Hiremath"

AT&T Bell Laboratories. Princeton. New Jersey 085-*2

Robert E. Newnham" and Leslie E. Cross*

Materials Research Laboratory, The Pennsylvania State University, Universitm Park. Pennsylv.nia 16802

A new type of barrier layer capaciier is described utilizing selected for this investigation were barium plumbate and bar-

thin glass layers on highly conducting ceramics of barium ium bismuth plumbate. These compounds belong to a family ofplumbate and barium bismuth plumbate. The frequency conducting perovskites.*"

dispersion of the apparent dielectric constant has been It is clear from Maxwell-Wagner dispersion theory that a

explained using a modified version of the Maxwell-Wagner decrease of the grain interior resistivity will shift the capaci-model. These capacitors have a low temperature coefficient lance dispersion in thedirection of higher frequency. '" Because

orcapacitance and a high dispersion frequency in the mega- the conductivity of semiconducting strontium titanate is higherhertz range. Simple processing conditions together with low than that of semiconducting barium titanate ceramics, the cut-

firing temperature make it possible to produce the barrier off frequency of the barrier layer capacitors based on strontiumlayer capacitors inexpensively. titanate is 100 times higher than that of capacitors based on bar-

ium titanate." The work described in this paper has shown thai

I. Introduction the cutoff frequency can be further improved in barrier lavercapacitors fabricated from more conducting barium plumbate

T VIE demand for capacitors of small size and large capaci- and barium bismuth plumbate compositions.tance has increased recently because of the rapid miniatur- (1) Sample Preparation

ization of electronic circuits. This trend is reflected in the

development and widespread application of multilayer ceramic Barium carbonate (AR grade. J. T. Baker Chemical Co..

capacitors' and barrier layer capacitors.2 Because of the lower Phillipsburg, NJ). lead oxide (100Y Litharge Hammond Lead

operating voltage, barrier layer capacitors utilizing a thin Products Inc., Pottstown, PA). and bismuth oxide (AR grade.

dielectric layer have become feasible. A method of fabricating Fisher Scientific, Fairlawn. NJ) were used as starting materials

barrier layer capacitors with highly conducting ceramic grains to obtain various solid solutions in the series BatBiPb, )O,

is described in this paper. where 0 :5 x !5 0.25; the composition Ba(Bi,,,Pb, )O, is desig-

Barrier layer ceramic capacitors are normally fabricated by nated as BBP and BaPbO, as BO. The batched materials were

forming an insulating layer on the surface of the grains of semi- ball-milled for 10 h in a polyethylene jar using zirconia grind-

conducting ceramics such as barium titanate.' strontium titan- ing media ad deionized water. The slurry was dried at I I0°C

ate." and modified compositions of the titanates described in for 4 h. The dried powder was calcined in a flowing oxvgen

this paper-" They have been widely used because of their large atmosphere for 12 h at 850*C as recommended by Gilbert

apparent dielectric constant (10'). small temperature coefficient Oxygen was allowed to flow in a tube furnace at the rate of 0 07

of permittivity, and high dispersion frequency (101 Hz). standard cubic meter per hour (SCMH) or 2.5 standard cubic

In order to produce barrier layer capacitors having a large feet per hour (SCFH). This powder was found to be single-

apparent permittivity, it is necessary to sinter ceramic materials phase by X-ray diffraction. The calcined powder was again

so that the semiconducting ceramic contains grains ranging milled in a polyethylene jar for 8 h and the slurry was dried.

from 50 to 100 ;im in size.' To obtain disks of semiconducting The dried powder was mixed with a polyvinyl binder and

ceramic, a two-step process is often used! In the first step, a sieved through No. 100-mesh screen to obtain free flowing

semiconducting ceramic composition is sintered and subse- granules. The preweighed (about I g) granules were pressed

quently converted to a semiconducting state by heating in a into disks of I.4 x 1 -2 mdiameterand I x 10"'to2 X 100

reducing atmosphere. Then either the surface of the semicon- m thick. The average green density of these pellets was about

ducting disk is reoxidized. or an insulating oxide is diffused 50% to 60% of the theoretical density. The green pellets were

into the disk. The disk is then electroded and lead attached to heated at about 550"C for 6 h io burn out the binder. The pellets

iorm the finished capacitor product. were then sintered at 950°C for 30 min in an oxygen atmosphere

An important objective of the present work is to provide con- with a flow rate of 0.07 SCMH or 2.5 SCFH using a heating

ducting ceramic compositions for barrier layer capacitors which rate of 100*C/h. Cooling was carried out by turning off the

have high permittivity, low loss factor. stable temperature coef- power to the furnace.

ficient of capacitance, high insulating resistance, and high cut- In order to form a barrier layer. several glasses (see Table It

off frequency at relatively low cost. The conducting ceramics were diffused over the major surfaces of disks. The glasse'were powdered using a mortar and pestle and passed through a325-mesh screen. The fine powder was then mixed with ace-tone to form a paste. The paste was brushed onto one side of theceramic disks and dried in an oven at I000C for 30 min to driveout the acetone. The glass-coated disks were then heated in afurnace at various temperatures in the range 4000 to 900°C for

Manuscript No 196371 Received September9. 199t approved July 8. 1992 periods ranging from 5 to 30 min. After diffusing glass on oneSupported b• the Center tot Dielectrics Studies. which is a consortium of lead- major face of a disk, the other surface was also coated with

M9 manufacturer-, of ceramic capacitors in the United States

"Member. American Ceramic Society. glass powder and the process was repeated. Gold was sputtered

2953

Table t. Melting Points of Various Glasses Used in (he 1. 120 to'Fabrication of Barrier Layer Capacitors Z

Melting polt O 110Compo".iton IT) Z

0Li,O-3B.,O 800 0 100 -4LiO-5-B,O, 900 y SSolder glass* 560 ccPbGe.O,, 750 0 103EO' 750--- 80 0

r, ) PbO)ItI 0). B. ) 114 I). ZnI)t 10 5. A.O, I0 5,. B).O, W12 () CuO 12 I.and SiO. (2 0)1 'Supplied by Coming Gla.s, Wirks. Corning, NY.,70 t0 :and ipprox, maii: com~rvl-itor PW_)-G. O-8(,+i.0(, 70 - 0

10'60 106•.~~~ X-- 4106

onto the glass-covered disks to form electrodes on the major K 50faces. 05 0 5o 100 1SO 200

(2) Measurements TEMPERATURE ('C)The capacitance and dissipation factor of the barrier layer Fig. 1. Variation of the apparent dielectric constant with temperature

capacitors were measured over a wide range of temperatures and frequency of the disk barrier layer capacitor of composition BP(-55' to 150"C) and frequencies (l0W to 10'" Hz). To measurethe capacitance and dissipation factor at low frequencies (< 10'Hz). automated LCR bridges (HP4274A and HP4275A, Hew- glass. The variations shown in Figs. I and 2 are typical of thelett-Packard. San Diego. CA) were utilized. The lead wires of solder glass. At high temperature both the apparent dielectricthe sample were connected to a test fixture which was directly soldr gladss i pature both the e ie icconnected via a coaxial cable to the appropriate LCR bridges, conduction in the glass.

The lead capacitance was compensated by calibrating the

bridges under both the "shorted" and "open" conditions. (2) Model for the Barrier Layer Capacitor of Compo;iticnThe fixture connected to the sample was placed in a copper BP

container which both provided physical protection and acted as The classical Maxwell-Wagner model is used to explain thea heat sink. The temperaturt was controlled in an eiectrical variation of the apparent dielectric constant and the dissipationresistance box oven (Delta design 2300) interfaced to a desk- factor with respect to frequency at room temperature. As dis-top computer (HP9816A. Hewlett-Packard) %,hich controlled cussed by Von Hippel," the real (K) and imaginary (K) partsthe rate of cooling by regulating the flow of liquid nitrogen into of the dielectric constant are given by the follow, ing equations.the oven. The same computer was used for on-line control ofautomatic measurements through a HP6904B multiprogrammer KKand an HP59500A multiprogrammer interface. K' = K I + I + w-r.)

The dielectric constant and the dissipation factor for severalbarrier layer capacitors of composition BP and BBP were mea- 7 Kw.sured at high frequencies (10' to 10"' Hz) using the reflection K" += -: , (2)technique of an rf impedance analyzer (HP 4191A). "' This .WtI I + W-'r-method is based on the fact that an electromagnetic wave inci-dent on a dielectric is partially reflected. A standing waveresults from the interference of forward propagating and K,- K.'reflected waves, which can be related to the impedance of the K = K (3)sample. The dielectric constant and loss factor can be calcu-lated from the impedance by a lumped or distributed circuit The parameters listed in Eqs. (I) to (3) are defined in Table IItechnique. The lumped impedance method assumes the electric where values were obtained from Fig. 3. These parametersfield to be uniform throughout the sample. This is especially were used to calculate the real part of the apparent dielectricuseful for low-K materials in which the wavelength of the elec- constant of the barrier layer capacitor. The measured and calcu-tromagnet'c wave is much larger than the sample thickness. The lated values of the real part of the apparent dielectric constantdistributed impedance method accounts for a significant portionof the electromagnetic wave in the dielectric.

0.50 102

11. Results and Discussion

In this section dielectric properties and their modeling for •barrier layer capacitors of composition BP and BBP are dis- 0 .cussed. A modified version of the Maxwell-Wagner model was <applied to explain the behavior of the apparent dielectric con- U.

stant and dissipation factor with frequency. 0 0.25(i) Barrier Layer Capacitor of Composition BP 0to 0

a.tatThe dielectric properties described here were measured on a cc

disk barrier layer capacitor fabricated with solder glass (Table 5I). Intentionally, a thick layer of about I to 2 mm was coated on 103

the barium piumbate ceramic as a conductor. The apparent 101dielectric constant and the dissipation factor at room tempera- 0 4 x 10'

ture were 60 and 0.03. respectively, at 10 kHz measuring fre- .50 0 50 10o 150 200quency. The variations of the apparent dielectric constant and TEMPERATURE (-C)the dissipation factor with temperature are shown in Figs. I and2. respectively. The dielectric properties of the barrier layer Fig. 2. Variation of the dissipation factor with temperature and ire.capacitor of this composition are attributable to the diffused quency of the disk bamer layer capacitor of composition BP

Table II. Parameters of the Lumped Circuit for the 0.3Barrier Layer Capacitor of Composition BP

"L•ow.-frequency apparent dielectric constant. K = 60"Highfrfequency apparent dielectric constant. K: 30Cutoff frequencyf,.,,r = 2 x 10'Hz 0Relaxation time. 1T 7.95 x 10 ' s 0 0.2 .-- CalculatedRelaxation tine. T, = 1.77 x to - pp Measu0edRelaxation time. 'r. 9. 15 x 10 ': Z

a._S01

are plotted as a function of frequency in Fig. 3. There is a good 0.agreement between the measured and calculated values. Theincrease of the real part of the apparent dielectric constant mea-sured at frequencies greater than 10' Hz can be attributed to themeasuring apparatus. 0 .... . .....

By assuming the known value of the imaginary part of the 0 1 10' 10s i i07 10' 10' 1010101 102 0 o o o o lglg1

apparent dielectric constant (K") at the particular frequency, the FREQUENCY (Hz)relaxation ti, Y. is calculated. Using the known values for K,T. r,. and -r,, the dissipation factor tan 8 was calculated using Fig. 4. Calculated and measured values of the dissipation faotor •sEq (4). frequency at room temperature tot the disk barrier layer capacitor of

K" composition'B1tan KT4)

The measured and calculated values of the dissipation factor are than 1%. It is te be noted that the variation of capacitance "s tthplotted against the frequency as shown in Fig. 4. Here also the respect to temperature meets the requirement of Class I typecalculated values agree very well with the measured values. dielectrics. The variation of the apparent dielectric constantAgain the rise in the measured dissipation factor at high fre- with temperature is linear as in the case of the disk barrer laverquencies can be attributed to the instrumentation used. capacitors made with composition BR and is attributed to the

(3) Barrier Layer Capacitor of Composition BBP diffused glass.In an attempt to determine the reliability of BBP capacitors.

Disk barrier layer capacitors of composition BBP were fabri- the hot insulation resistance was measured at a temperature otcated using the glass EO (Table 1). Figures 5(a) and (b) are 85°C by applying an electric field of 100 V/m for 10 h. Thescanning electron micrographs of a sintered disk of composi- insulation resistance is plotted against temperature in Fig 7 Ittion BBP before and after glass diffusion. In Fig. 5(bl, the dark- is to be noted that the average resistanc, was 10' 11 for the dura-colored phases are the conducting grains and are surrounded by tion of the measurement runt( 0 h) The ins-!ati-n resistance ofbright grain boundaries. The grain boundary is probably the these bar, Ser Usyer capacitors did not decrease as rapidly as thatglass phase and its thickness is estimated to be I sm. of the capacitors prepared from barium titanate ceramics. The

At room temperature, the apparent dielectric constant of the dielectric properties exhibited by the barrier layer capacitors ofbarrier layer capacitor at I kHz was about 3000 with a dissipa- composition BBP are those of the diffused glass.tion factor of 0.02. The apparent dielectric constant (3000) ofthe barrier layer capacitor was much higher than that of the bar- (4) Model for the Disk Barrier Layer Capacitor ofrer layer capacitor fabricated with composition BP The varia- Composition BBPtion of the change of capacitance with respect to room As mentioned in the experimental procedure. several temper-temperature is plotted against the temperature measured at I atures and periods of glass diffusion were used to fabricate thekHz in Fig. 6. As we observe from Fig. 6. the barrier layer barrier layer capacitors. In order to estimate the thickness of thecapacitor has a very stable temperature coefficient of capaci- dielectric layer of the disk barrier layer capacitor. the two sur-tance; the variation in the measured temperature range is less faces of the disk were polished gradually and their surface con-

ductivities were measured using a multimeter (Micronta DigitalMultimeter Model No. 22-1984. fort Worth. TX). When

65 ....... , ............... .... .... approximately 100-1p.m thickness of the surface disk wasremoved, the polished surface became conducting, as indicatedon a multimeter by a short circuit. The other face also became

(n conducting when approximately the same thickness wasz removed by polishing. This result suggests that the thickness of

C glass diffusion on either side of the disk barrier laver capacitorU0---0 Calculated?.-. Measured is about 100 p.m.

1 j , The scanning electron micrograph shown in Fig. 5(b) is aUW 50 cross section of the dielectric layer which shows that the dark.Wua conducting grains of 10-p.m size are surrounded by I -p.m-thick

5dielectric grain boundaries. Hence, in the disk barrer layer- capacitor, we have two grain boundary dielectrics on either side

W •of the disk separated by the conducting bulk ceramic of compo-4 5 sition BBP. A schematic diagram of the barrier layer capacitor

depicting the grain boundary and surface barrier layers is showns in Fig. 8. Equivalent circuits for a combination of two grain

101 102 1 101 1 106 1 o l 101 boundary barrier layers adjacent to two surface barrier layers10EOUENY 10za 111which are separated by a conducting ceramic are shown in Figs.FREQUENCY (Hz) 9(a) and (b). Figure 9(b) is the equivalent lumped circuit dia-

Fig. 3. Calculated and measured values of the apparent dielectnc gram of the circuit shown in Fig. 9(a). in which the variable

constant vs frequency at room temperature for the disk barrier layer capacitance and resistor represent the dielectric phase of bothcapacitor of composition Bp. the grain boundary and the surface barrier layers. The fixed

IIc !5

f tj I 1 7 .

~I r t fit A !

it I i It t.4t'

i~ojvil-L .½wmw mM monI :ciic

"";I)RA U~ L!,1X W CC ,h>-1ll*C1" 1t !I

"------------------------------------V44 1~.''- - 4 4, ~'2'

- _________ -. ~~~"rh A I',4' K -''

3000 , ' , ,', .. S04-

Grain P-Boundary rnBarrier 0 0

2 0 0 0 o--- -- -Dl Oelectric Y..0 4~wn~ ~I' i

Phase - -0 K -iFr.q uenecy¶ '. 0 " 1 1,5LuL_

Surface OF Frequency 10'.10'

Oafrrier (t.-nd u c fin g 10 0 0 (n

-Phase W 0

4a.

GrainBoundary 4. 0 -,,0

Barrier (b) 10o 104 10' log 1010 1012 1014

FREQUENCY (Hz)

(a) Fig. II. Calculated ,alues of the apparent dielectric consant andi. .sipalrion factor .s trequcnc•. at room temperature Ior the dis•k barrier

Fig. 9. ta) Equivalent circuit of the disk barrier laser capacitor and K. er t .napacitor uf compositiu n BBP

IbN its Jumped circuit diagram.

phase is the same as that of glass 30 and considering only the assumingK. = 30, rr, = 6.67 (11-cm)". d, = I p.m. and -.

10 p.m. Similarly, the cutoff frequency for 1',e surtace barriercontributions of grain boundaries, the apparent dielectric con- laer at low frequency, is calculated to be - x 10"' Hstant of the boundaiv layers can be calculated as 300 using takingK, =. 300. (, = 6.67 (t.cm . 1( = IX) p.m. and d, =

Eq. (5). 18(X) p.m.At low frequency, the barrier layer capacitor behaves like a The real part of the apparent dielectric constant for the barrier

surface barrier layer w ith a dielectric thickness of 20() p.m . as Ther e capa rto f the a re nt diel e tri t a. r th rr h

earler.Appyin Eq (5 andusig d = 80lapy laer capacitor in the frequency. range 10: to 10` Hz. in vshichexplained d, = 700 (5) and usant di constant the surface barrier layer behavior is assumed to be predominant.andd, = 7(0~m and K, = 300. the apparent dielectnic constant was calculated as a function of frequency using Eq (2) assum-at low frequency is calculated to be 2700. which agrees very ing K = 2700. A: = 300. and T = 3.56 X 10- " s Simlarl,.

well with the measured apparent dielectric constant shown in the real part of the apparent dielectric constant m the frequen¢3Fig. 10. range 10" to 10" Hz. in which the grain boundary barrier laser

The calculated apparent dielectric constant and dissipation behavior is assumed to be contributing to the real pari of appar-factor are plotted against frequency for the barrier layer capaci- ent dielectric constant. was also calculated using Eq (2) andtor in Fig. I - Since the barrier layer capacitor has two contri- ,ssuming K, = 300 K, = 30. and T = 3.95 x 10 " sbutions to the frequency response of the apparent dielectric The dissipan3on factor for both the grain bxundar , and theconstant and dissipation factor, it is suggested that the fre- surface barrier layer is calculated as a function of trequencyquency response has two cutoff frequencies. The cutoff fre- using Eq. (3)and parameters A7. K.. and -t specified earler Thequency can be calculated using calculated apparent dielectric constant and dissipation factor

fid: are plotted as a function of frequency in Fig. I I At a frequenc ,f "f = 2T ,,K.J (6) of 10"' Hz the contribution of the surface barrier laser to the

"apparent dielectric constant drops off and then. at a irequencýwhere K, is the dielectric constant of the dielectric medium. F,, of 10" Hz. the grain boundary barrier layer contribution dropsis the permittivity of the free space, a, is the conductivity of the off. leaving only the contribution of the diffused glassconducting medium, and d, and d, are the thicknesses of con- Figure 10 shows the measured values of the apparent dielec-dueling and dielectric media, respectively. The cutoff fre- tric constant and the dissipation factor for the barrer layerquency for the grain boundary barrier behavior at high capacitors. Comparing Figs. 10 and I I. we find that at low Ire-frequency was calculated to be 4.4 x 10" Hz using Eq. (4) and quency both the measured and calculated values agree very

well. The difference between the measured and calculated val-ues at high frequency can be explained as follows.

3000 , 0.2 In our calculation, we assumed the dielectric constant of thez barrier layer to be the same as the dielectric constant of the dil-( fused glass (-30). The glass diffused in the barium bismuthiz M= plumbatc ceramic may have a dielectric constant different from0u 0 30. Secondly, the conductivity of the conducting ceramic phase

2 < is assumed to be constant in both the surface and grain bound-5 ary barrier layers. When the conducting barium bismuth

w2500 0.1 plumbate ceramic is subjected to the glass diffusion heat treat-4< ment. the conductivity may change. Thirdly. the inductance of'0-

the device becomes very important at high frequency (10' Hzi2 ") especially when the reflectance method of measurmi the

1 dielectric constant is employed. In the Maxwell-NagnerJ. model, the inductance of the device is not taken into consider-

<1ation when calculating the real and imaginary parts of the2000 0.0 apparent dielectric constant. In practice, even materials %kith

FREQ ENC (oI) r)0odielectric constants as low as I co exhibit enorm ous increases inFREQUENCY (Hz) dielectric constants at this frequency due to high inductance

FiR. 10. Measured values of the apparent dielectnc constant and dis- One would expect this effect to be more prominent when mea-

sipation factor vs frequency at room temperature for the disk barrier ,uring the dielectrc properties of the device which has an

layer capacitor of composition BBP apparent dielectric of 2780. The measured apparent dielectric

constant and dissipation factor shown in Fig. 10 increase enor- Referencesmously at high frequency ( 10 Hz) as a result of this effect. The 'Y Bakabe. -Recent Progress on Multiilar Ceramic C'apacimors pp 111j-29measured apparent dielectric constant does not decrease to a in Proceedings of the MRS International Meeting on A.,an.ed Materialsvery k~w value (30( predited the model F II). The �okyo, Japan. May

31-June I. 1488). Viil 10 MulauIseri Lditedt, 00 tiN'

Spredicted by (Fig. ama, S Simiya, R P H Chang, R Namatnito. and T Ohno 'Materiaiconsiderable increase in the measured dielectric at high Ire- RearchSociteis. Pittsburgh. PA 1999quency ( 10' HzI may have affected the value at low frequency R M Gtaister. "Barner-Layer Dtileitnc." pi Init L/ir tr fn, Part8(10' Hz). The peak in the measured dissipation factor (Fig. 10) 109B(2214[. 3-3ii lig~bImay be attributed to the instrumentation used. 'S Vaku. -Studies on the Bioundar, Layer Ceramic Capacitor" Re% Eiritrm',imnu Lab . 15 [9-I0168q-715 (1967

'M Fujim(itoandW D Kinrery. "Microitructures•, SrTiO. Internal Bund,I1l. Conclusion ary Layer Capacitors During and After Processing and Resultant Elevirical Prop.

This work showed that barrier laver capacitors can be fabri- enies, J Am Ceram S,( .6414 4) 1"71-73 198%,'R Wernicke. "Influence of Microstructure on the Elecrical Propertic, ol

cated using highly conducting ceramics barium plumbate and tntergranular Capacitors .- r Ditch Aeram Gev .55• 7 tSit..•Iiij7)bismuth plumbate. These capacitors had a high frequency dis- 'B V Hiremath. "Bamer Layer Ceramic Dielectric Capacitor Conaminin

persion (megahertz) of the apparent dielectric constant, a low Barium Plumhate,"U S Pat No 4761711 1Q89temperature coefficient of capacitance (TCC) (<1%), and 'H D Park and o A Pavne. "Charactenoation of Intemal Bisundar', Lasei

Capacitor*. pp 242-53 in Ad.a vnce, in Ceramic',,. (o) I Edited b, L %.I LeOmaintained a high insulation resistance under accelerated relia- inson Amencan Ceramic Society. Columhus. OH. i9. 1bility tests in wkhich the capacitors were subjected to high tem- 'R R Roup andC E Buler. 0 S Pat No 2520 ;7h. Augut 2Q IQSOperature and high field simultaneously. 'F Steele. "An Invextigaiiin of Banum Metaplumhate as a Ceramic Elecf'o

The frequency dependency of the capacitance was explained flr Ceramic Capacitors". M S Thesi, Penn,,yl'ania State Uni.ersitw. L nie.by the Maxwell-Wagner model and experimental data fitted sily Park. PA. 1985.

"'A Von Hippel. Dirlecrri Miterrati and Appht imoni Ms I I Pre-. Camfairly well with the predicted values: As predicted by the Max- brid•e. MA. 1966well-Wagner model. the dispersion frequency of the capacitors "G Gocxdman. "Capacitors Based on Ceramic Grain Bi'undrc Barrier La,was in the megahertz range. ers-A Review," pp 215-31 in Advances in Ceramics. %0'l I Edited bh L NsI

The simple processing and low firing temperatures make it Levinwn American Ceramic Society. Columbus. OH. 1981pos sibple manufacturessing te b rie g l perapaitores mnexe- it.L R Gilbert. "Bulk and Sputiiereed Thin Film Ba'PbBh;iO -A Ceranitpossible to manufacture the barrier laver capacitors inexpen- Superconductor": Ph.D Thesis PennslvaniaState Unirersii. Umniersif. Parksively. The high apparent dielectric constant and low TCC PA. 1979properties make these capacitors an alternative to conventional I1- W Dakin. "Microwave bielectric Measuremcnts."'J Appi Pn, - 18 (qbarium titanate Iow-TCC capacitors. 789-96(1947)

""aM A Stuchly. "Permitting Measurements at Microiasr Freciucnsics ILirnAcknowledgments: We would like to acknowledge the help of Paul Lumped Elemenis."IEEE irans. Instrum Meas . IM-23. 56-6.2 t19741Moses for the temperature coefficient measurements. Dean Anderson for the hot "R. Wermcke. "Form:r-ion of Second-Phase Layer, in SrTiO. Bounderinsulation resistance measurements, and Frank Steele for discussions (if ceramic Layer Capacitor,". pp 261-' t ,' Advances in Ceramics Viol I Edicd h, Lcomposition M L.vinson American Ceramic Society. Columbus. OH 1t141

APPENDIX 46-

GRAIN SIZE EFFEC-T O)N 11Wi DIELEC-1 RIC PROPERTIEIS OFSTRONTIUM BA.RIUM TITANATE

Uý Kumarr. S. F. Wang. S Varana~si. and)1. P. Dougherty

Center lot Digelectric StudiesMiaterials Research Laboratory

ilte Penrvsylvania State L',tiversinvUtniversity Park. PA 168(02

Ahtritlt this paper, the grainl size effect oberedont Ba(Zr 1,Ti l.,)0 etc . is sevesisltegental to tailor thle mnicros;tructute of fine

dielectric properties of Sri) 211.i) gTio, ISBTI ceraiitc i'4 reported. grain thin la e fvLC% and thui filing In thtsq paper we describeFpf this study.' SBT powvder Prepared by ; hydrutliertial procedure the grain size effect on the ditelectric properties of Sri, 28a,, RTtO1v,,,sed A porous ceramtic was prepared I,% xintering lthe povader ceralpic. We have used conventional and 'fa~st firing' chedules to

beI~pa tq fe(Wvl~ I 1010- I1 3 175flun /5 If))?. either h' VY prepare ceramics with differing denitsties. The size effects were

(1 oa -ieittCgt1 or by last firing'. When thle average gratt -studied through microstructure. XRD. dielectric. andi P vs Eoze of tile ceramtic xa belee 0 75-11,2 pin. tile room hseei eairtemperature dielecrric constant peaks to 55(H)-6001)t In porous fine,rato LCraJt1'C also, classic P ve E hysltersis behravior is obqvd: E.xoerimentl

Introduction For this study. Sit, 2Ba0 XTiOj powder. prepamred h%.Gran szeeffcts o te delctrc roprtes ~hydrothem,. :edure. (Cabot Cotporation. 8o% ertownui. PAt wa..kGrai Weeffctson iledieectic roprtis o Bai~ihas used. lite H E T. surf ace area of the powder is *14 6 rnigmt The

heel, jn actte area of research over thtree decadfes Ever RitiCe. SEM micrograph of the powder (Fig. It) clearly lin~sthe sizeKilickatnp' and Flevwatgigl reported rthe obsevattor oif unifornlitvyamnomalouslY htgh roomo tetnperature dielectric cttstantn in HaTiO, The powder was blended withi appropriate amount of PVAcerami1c %ith tiaverage grain size of 1 pll, several rescaicli solutioni and glycerol. Pellet preparatton and htnder e..:iporationarticles appeared in lthe literature. Careful literature review procedures are elaborated in Ref. 161 Binder ret-noved pellets %kereilldicate that the olsmervr grain size effect is a strong function of sintered between I IOW*C and I 350C for Senirm - lit InI a Oclat,e..eral exterimal antd a few internal varialbles. Effect of chetnical furnace. For conventinortl firing, heating and cooling rates were

prt.prepattatiott procedure, initial particle size distributiotn of tile maiintained as 40W)C/h. For 'fast firing,. lthe saimples were heted neiopo,,der. final grain size distribution. re!siqtivit', of ltie grain- atnd 950'C at a rate of 400'C/h. and then to the soak teomperature atthe grant boundaries aind thle density of lthe ceratnlic etc.. should be 100(r)C/h. They were cooled to moorn temperature with a sinmitlarcotnsideredl 1; exlersal coilttrihutiotts to the dielectric anotttaly. schedule. Bulk detisity was tmea~sured only when the destigtx ot lthe1hrougll careful preparattott it ts possible to minntitize the effect of sintered samples were greater than 90% theoretical. For thle rest nf

e.,Lerril vriabes.the santples. geometrical densities are reported. Procedure forWhile tfis;cusiog tltve intrinsic effect, it is necessary to dielectric and P vs. E measurements are elaborated in Ref [71Consider lthe boundary cottditions very. carefully. Bitesseni etal .121 prepared high density ceramic %%ith varying grain sizes.SENI anal 'svis of lite fine gratn ceramnic showedt fewer ferroelectricdlotnatos a- cotmpared to coarse grain ceramtic, It is well known -mpr7 ~ I.%.%

4that in a coarse erainl cerntuic. tile volutte change associated withthle cubic to retragrinal phase tratnsition is minimtized through ~Tdomiain fornataion. lIt fine grain ceratmic. grains without mnulti- 7 ~ - .-

domnains; will irtduce stress of complex nature otl other grains ., -6: I.' j * 1,Anomnalous inicreases in ltre dielectric consta~nt of fitne grain..~~.'~.cerainic are explained as a direct consequettce of tile -tress. 91

Arliet cal.. 131 on tite other hand, observed that the doinain .r 0 ssize is a strong function of grain size. Through SEM and TEM :1 a X-%olbservations. they observed an increase in tlte domain density ~ - - 1 - 4 *~Awhen the averatge graitn size of the ceramtc was between I pin andt ''~ *-* ,if) pmn Additionally. tile dotmain size reduced as a function oif stgrain size. limey proposed that the increase in room temperature Jdielectrtc con)stant ts at least partially due to tlte increase in domain Zk C... -Zwall density,' -jV

Shaikh et al.. 141 observed similar anotnaly ill a porou's " 3l' *lcerainic. Effect of density on the miagnituade of dielectric constant *~'~

is discussed in a paper by wa Gachigi et al..151 in this issue.In fine grain sa,6O3 ceramic, the tetragonal toV

orthorhombic transitioin occurs nearer to 25*C instead of 0 5'C.Some of the observations. such as the difference in domain size, 9V 2 . I 07629splitting of <200f> peak it XRD patterns. are expected to beInfluenced by the presence of tite orlhorhombic phase.

To minimnize the complications due to the orthorhombicPhase, it is desirable to investigate the grain size effect in BaTiOx- Fig. 1. SEM micrograph of 'as received' S8T powderbased solid solutions, Moreover, understanding the size effect incomtnercially imiportant solid solutions such as Srl.,,BaxTiO1.

CH3o8Uo7803-065-9/92.000EE 55

W 'Results and Discussions

Densities of the siniered pellet, are listed ini Table ITypical microstructures of the samples stotered helo%% I 251)C ptshown in Fig 2 As the parnicle size distribLilion of the sianrin

4 '. 'powder wa., narrow, the grain size dtstributionii of the Wtilered'k ceramic also remained narrow But when the sinterimg ,eirperatiue

was raised beyond 1250*C. abnormal grami gromw occured Jhut""b it was not possible to prepare samples with unifonn grain size u

the range of 3-10 pinIn Fig 3.. the temperature effect on the dielectric propenteg

are compared These K vs T plots are not cotrnenc:aed forA.- porosity Higher roomn temperature dielectric constant observed in

samples sintered at 12OWC with an average grain size of I0 Q pn.and 87% of theoretical density (T able 1), cleart% slior, the

.4k ~existence of grain size effect in SBT vcr% stinitar to thlat osrvicdL Iin pure BaTiO3 Dielectric constant values at room temperature

and at T, of all the samples are compared in Table I Ticompensate the values for porosity, Botcher s equation sAas used

- * I 20000 d

A1 ' * 40 . kf~zS% , , L _ • " -0000 c, oa "

, *. • "' p -r'l _ •... Jame C d

'A.)

% 0; 0 0Il a -125 -80 -35 to 55 10

Fig 3. D'ielectric properties of SitIcor ,.iiitrcr :rr'1l 2(N './21t. (b) I fl0C'/21t, ic) I •{t '/2h airi (,i r 1';loC./lth l,

solid line.: are dielectric constant and doited linle, aie (liele-Ltric los'Fig 2 1'•,'rc'I int i•|iiu litle of ,;tl'l ellilrtL 'mirlereh tt data measured at I kHztlirl ltttl'(/hhlhi arid• thbi 12)I 2ti"/1ih

"Table I Ph% ical imid dielectric propertje. of SBT ceramic

Simi lero|p "\\e D iell,, KRT KRT Kai Tc Kai Tc Curie -Tc.- 11.0 Pr EcoCh g . ri ros r (Cal) IOb, I (Cal.) Cons fCC) CCj fpC/ (K\i

pol ther' x]OI-5 cn2) cm)

I Vfi ,4 " S g 2 0,(t 4S40 4490,) 91"KO I 5ti b lII lj(,,rh (, 75 Ir .,21, 2XgRt .(•00 i 0 I()f(OtJ - -6

III 1 0 1 2 R 1 .J4 " 775 W44( f6 0 728(0 '12(8 i 1.54 (2 -20 '" l' " f'11 U ,'5 tP U 6h 8 T 310 50{10 7720 - 17?% 65 -24 -

115/0 5 ( " '700 19I7t(1"t i o 116(1 1 47 6 1 -22 '' . '.•1.111f(S. i' - Th IV V,5 509() 980M) 13360 I 49 62 -.1 1 98 148(112(0ii02 i, •' I1 -Ijr ;970 11160 138(K) 140 6' -2.4 38S 112.

2! 3u( 475 61 00 1 '.51 67 -2' 4, (, Ol~ .2 >I' P, l- U -'tI, 25u(i -bIO -g 4 -.t O -• ,(_______u_ (I-~i 10000'' "~l Ir~t 18740 1 ý5( 60 *2 4 7 (-1 5 A(i

1 2t, 1t 2,II '' 1 I6 '8i( 9 2890 15 0 1 157(q 1 147 ---125.5 '511nd I i 5 1 72 •70i 0( 1 1.70 '4 : Ti275,'.i-1l 2 71 7 :" i "ii 5 8.1 J(8175 14o-8 (1 - 49 64 4.4 87M "

, 13___'____, 2 . 1 •{ 32 •{t(i -10ff) I 158(1 16220 1 i 0 t;S -215 344 84(0

56

Coinparitig lthe pori',1o Corrected .iloes. it 1ý clelr that a peaktroom temperature diri t-iii (mostt 01i of 500-tt6000 is observedwhen the average gramt ;ize of the ceramic is between 0I 75- I 2 pinIn Fig. 4a.b. ajiid c. ti.e fretueti.fec effect on the dielectric constanitof cerantic witht differitig gramti sizes wec piotted Ahove T5. K vsT tichavior is nemzll independent of tiensuritig frequenicy in .0l The

- - - -- - - ' cases. lIn Table 1. Curie t onst.uit of thiese sa, le re cuoipazed(a' As seen the mtagnitude (A~ls, ithin ai narrow range of 1 4-I1 7 x

:Kj 2 Ig3 2 105( C) lit Fig 41, intti c. fine Fraiii ceratni showv smeringp ofdielectric peak ait all filter transaiit iotemperatures lIn tblie 1. is i

of the transition tetlperattires are listed. When the average grainsieof the cer~utiic is reducedI fronm 11) pot to 1) 7 pin. the cu itc to

tetragonal irajiisitiout teiiieraiture Oeci eases from 67TC to 62T.whereas the tel r gnoal ~ to mitbin thotmitc traitsit ion tem perat ure

Sratses froml -29*C to -19 ,C( Sinte the dielectric peakcorrespondIing to tlite oohrhtI. oinhii i to rhohottbote iral tranits itio

r occurs, over a large icoipetatute ranige fii fine grain cerunic, thesevalues are not tablulated.

lIn Fig 5. the P \-% F helaia for Mt three swunpler is compared

As seen, in fine graini cei;u~iocs. tile induced polarization reduces

same. only lihe getneral behavior should he eoin-idered for8 comiparisot. Even i a titie grain pmoiusceratitic, appearance of

Pr and Ec values calculated froim The hysteresis curves are listed200CC ~In Fig 6 a and IK tile polished and etched inicrostructures oif(b) rwo of the sintered ceramiics are qhown The samples were

125/5.(F) C 9,s.-(.2 u, polished slowly and carefully. so that the surface temperature ofthe samples are niaiiatited %l%%;:Ys tnearer to room temperature Toget a represetntative domatin otnjLture of the bulk, about 0I I1-0 2 tomtthick surface limvet Nta' removed before piolishintg lit Fig fib.domains 41t the fitne graini ceratitic is very dclariv seen Since the

(0000 density of thtesamtple- fired ait tettiipraitureq Weow 1 300*C wereE 1000 low, t wasnot possible toplinithem properly At present

eventhough te exstentce of ntitomi-dntn-ain- fitt fitne grain ceramtic isclearly detnotistrated. it is tiecessqary to prepare dense ceramic to

Y quantify the domain size anid the dotanam detnsity as a functimon ofgrain size.

0-150 0

20000 )I

L lteo 6 is='

Temaerature (00)

Fig 4. Effect of itoelsuring frequency on the dielectric properties Fig 5 P U. F It>steestiS behatior of SB3T ceramtic smntefCd afof S8T cera~nic qititered at (a) I 250*C/2h. h) I 275*1l/mnin fFast 0t I 350*C/lt0li. (2) I 21H41W. ind (3) 1 100*CI/t0hi.fired), and (c) I lO4)C/6h: measuring frequencies were 4) 1. 1. 101.and 10l(JkUz.

57

, -Conclusions

'~~"~* Based on the experimental obserations. the followingconclusions awe drawn.•;l|.L. ''• ., - •k..dl~V I When the grain size of the cerainic is b,-tween 0 75-1 2

t o y --% Am, the KRT values peaks to 5500-6t(XJ2 The Curie constant of the ceramic is relatively

, • ," . independent of the grain sizeIn fine grain ceramics. suppression of ic and

enhancement of low temperature transitions are clearly seen4. Fine grain ceramics also show classic P vs E hysteresis

"--��� " •ehavior.* 5. Etched and polished microsiructure of fine graw porous

- -"" ceramic show multi-domain characteristics"To quantify the resultl. ii k necrssar, to prepare dense

- , ceramics with differing average grain sizes ranging from 0 1 JIm it)

-. . -% A " •, I nm sith narrow distribuitun

-- "-Acknowledgement"s Author- gratefully acknowledge the finajici.-0aK support from Cabot Corporation of Bovenown. PA and the Ben

Franklin Partnership Program of the Commonwealth of PAAuthors also acknowledges the technical support provided by

~~~ , _R IF' efeteinces

(11 H- Kniekatop. and WV HeywingY" Depolarization effect,; inpolycrystalline BaliOi." Z Agnew Phys. Vol 6. p-315-390,

OP 121 W. R Buessein. L E Cross. and A K GosIA-aft. J Am

lkI Cerain Soc .vol d9. p 31l(11q66)

13 1 G Antli D) lentuingsý arind G deWith." Oie!.ctrtc properties offine grained BaTing.' J Appl Phys. Vol 58. p. l619-1625S~ (1994)

141 A.S Shaikh. R. W Vest, and G NI Ve,;t," Dielectric• • % properti" • of ultrafine grained cera mnic." IE E ra n

___W Ultrasonic. Ferroelectrics and Frequency Control. Vol 36..;b Pe I;P. nonp4. p407-412 (1989)15 I-1 K. wa Gachigi U Kuimar. and J P. Dougherty." Grain size

"L effects on BaTiOr" thih issue16) 1) Kumnar. S F -Wang. aid J P Doupheri•'.' Piepatiaio riof

dense ultra-fine grain barjum tatanate -bh.ed certaitics." thisissue

171 S F. Wang. ti Kuimar. "lk fuehiuer. P NIarsh. H Kunkel. idC 15aklen.' Grain cize effect ott the induced pieznelectric

Fig 6 Poli;hed ;tiid el red oticrowstructute of SAT ceramic prop•iies; of 090PNIN-ht I OPr ceruimc." this issuelittereil ai ! ill I (tl '(._/libh. nd ( h I 3lI-C/I n mit And the reference-: it 111-171

58

APPENDIX- 47

PREPARATION OF DENSE ULTRAFTINE GRAINBARIUM TITANATE-BASED CERANUCS

LI Kumar, S.F Wing. mid J.P. Dougherty

Mlteriais Research LahoraiorvThe Pennsylvania Slate University

University Park, PA 168012

mi Reetl..-ehv siee sdohna BaTiOl with 3-5 Results and DiscussionIBj'0) in IToery high dlensities at 8511*C. In this Paper. nrew

sisof, tilt effect of lower concentrations of Bi,01 nit the Phiysical characteristic,, of the as received powders used for

d.11lzificatioanud tile dielectric properties of the BaTiO1 ceramica~e ti td are lse nTbe1

di-OFse" As anl expainsion (if this work, hydrothernialS, Bi g ~~ w:as sitnlered with' ()-5 wtr BiO 5 hbetween 850)- Table 1. The properties of BaTi~i Powders.I5 C. When 5 wt'7 13,,)1 was "')addedl. thle powder comipacts

0oi~iered to a hulk density of >5.2 gills/cc. at 950'C The room ID Cmoiin Adtos SAL(,ltpe[,1lture dielectric constant oif this ceramtic wajs 150t). and K vs- (rn2 gni (wtpoitol Adiin I

T plot hollowed tife Z5UJ intdustrial specification. ___________ (%

bitroduction BTI BaTi~i ..... 8.54 1.8

in it recent piperl 11, by comparing thle effect of BisiOz on the B2 B~~ 10pt .6I-,,iotering behavior of high purity flaTiOi powd~ers piepared by of2 Nbix 2t~p 79

dlifferetit procedures, good sinterabilitv of hvdrothenitiall 13aiOl attemoperatures ats low as 851)C was denoistngratech By perforiltnin BT3 BaTi0% 0I 29%%tr 7 59 ...XRD and DTA mleisiremetis oil these mixtures, thle following B6i2Oreactioli sequence was identified:

1 aro,+t B~,B'r4 BaTiOi1 3 t% 8.17 ----

_4 5 l(Bi,01)n g(BaO)11,21 + Bit2110O10 (Eq. I)

As thre reaction prodJucts mnelt at 717'C uind 947C respectively. goo S2txklr10 ---- 1 1detisification occurs at low temsperatures through liquid phase Pat15ifliering The hydrothiental BaTi~s sinters to high deiiuitie at low

In a different paperl~l. thle effect of Bi,05 addition oil the temperatures In Fig. I , the shriiikage prof iles of h% droiliennal ,utdmicrosiructural development and thfe dielectric properties of the non-hydrothennal hight purity BaTiO5 powder compacts %%tth ';%%-sintered ceramic was (discussed. For this work. hydrothensial Bim)0 1 are compared. In this figure. good deinstficdtioii ofBali~i powder compacts with 3. 5. anid 7 wt'7 BiZOI were sintered hydrothennal powder compacts is clearly seen to occur at aboutibetween 750 -95(1C2h. 750'C. Typical microstructureq of the sintered compact ate slin%% it

In this paper. the results of file latest studies onl the effect of in Fig. 2. When the samples were sintered betueen 750t. 1250'CBi,Ot addition on the densificatioti microstructure development, with 3. 5. and 7 wt% Bi-,() , irrespective of the sioteringandý the dielectric properties of tile hydrothenriall BaTiO5 ceramic are temperature and the flux conc-entration, the averige grain size of theexplained The result and discussion section of this paper is divided ceramic remained as 1) 15-0.2 Witn. which was also thle initial panticleinto four pans. In the first part, soine of the physical and electrical size of BaTiO3 powder. Typical dielect-ric properties of the sinterettcharacteristics of tile ceramnic are reviewed. In the second part. the ceramics are given in Fig. 3 A broad trattsition at around I 10*C,effect of smialler quantities of 1360 3 on tile densificationt and aitd the suppression of other lower temperature transition- aredielectric proprieties of tile B'r ceramtics are discussed- lIt the third typical of thtis fiine grain ceramnic Though tile dielectric peAkpart. thie effect of Bi-additioti duritng the Imydrothennal process oi corresponding to cubic to teiragotmal transititon is clearly detected.the inicrostructural development is discussed- In the final parn, thle room temperature XRD patterns of these ceramic did not show cleareffect Of Bi2O1 on the properties of BillgSr0 .2TiO3 is analyzed. splitting in <2X)0> peaks.

ExprunenWThle surface area of thle as received hydrothermal powder%

tCabot Corporation. Boyer 1Townl. PA) were measured by BETCItechnique. To prepare the pellets. tire h'udrothernual powders wereMixed with (I- wt% of 1360 1 (Johnsoo Maithey. Seabrook. NH) inan agate imoner and~ pestle. Several 1/2" disks were prepared in aasteel die, by applying uniaxiaJ pressure (3IXXMI-15WM psi) at roomt n -totemiperatrwe. Pellets with about 53% green densities were preparedNby using 1.5-1.7 wt% PVA and 1.5-2,Owt% glycerol as binder andplasticizer After binder evaporatioin. tile green pellets were sinteredbetween 750-I 250*C/0.5-21i in a closed crucible. The bulk density_________________of thle well sintered disks were measured by noting the weight losstn Xyletme. Fractured microstructure oif the ceramiic was analyzed 0 S00 1000with, ati SEMI (IS1-SKI 1311. Akashi Beam Technology Corp.. Tempe.'ature IClokyn. Japan). The dielectric properties of the ceramnic dhisksamtples were mecasured using a comtputer cotntrolled LCR bridge Fig. I Shrinkage profiles of tat high purity BaTiOl (1 raos~eklu44274A, Hewlett Packard. CO) system. + 5wt%BiOi and lb) Br2 + 5wt% Bi2Oi

C11310xlJO7g)3-0465-99f2S3.00 QlEEE 70

In this section, the effect of smaller contcentration-z of B,-,Ojonl the prop-enties of BaTiOl ceramitc is discussed Toi achieve gooddensities fit thvese cesaitucs. it was necessary to rimter file compjlacts athigher temperatures Even, after sintering at I 25(IC. themicrostructure of the ceramic looked very similar to that shiown in,Fig 2 In Table 2. somec of the physical anid electrical charactensn"cof tife sintered ceramic are listed. The dielectric properties of theceramic showed qualitatively sunilar behavior as seen in Fig ISince the liquid phase volume is low, higher Tc and Kpr areobserved In these ceramic also, the dielectric peaks, correspondingto lower temperatures transitions are also suppressed

Table 2. Physical and dielectric properries of B*I' and BT2 ceramnicwith I wt% Bi 201

Fig 2 Ty~pical mnicrostructure of FIT sititered with B60O; This 1. D. Sint. wt.loss Bulk p Kmas Trva5 KlRT 'larim .icrostrucnjre represent the fracture surface of BT2 with Temtp % pit/cc CC)7%%t%~Bi,-03 sintered at 85(FCI2h. (C c~i) ± 2%,Q I 1k Hz

FITI 1150/2 1 09 499 3195 12-7 ll~ IN) 117tat 3000 1250/1 1.15 5.52 58201 j27 20(M 0 033

BT-2 115-0/2 1,43 s561 58110 121 2It 1 0(tli I

45 5 1250/2 1 .50 5,79 7075 123 2650 0.019

420003

1000 In two hatche" of BaTiO,, Hi-ions, were introduced duringthle hydrothennal processing T nrdc i~n ntesrcueair appropriate amount of Bismuth fin the nitrate form was addeddurting the formnation of Titanium livdioxidei'll To this mixture.Ba- -oluiion was added for the BaTiO, fomiiatioti Due to the

0 ------ cotnplexiiv of file reaction, thie nature of Bil'additiott is unknown-200 02 0 lI the XRD patterns of the reacted powder. few additional0 200 diffractcion peak- other thani the imajor Bit rio, peaks were detected

These peaks with fcable intens-ities, could not be mtatched to wi,, ofTemperaeture (0CC) thle Bismuth Titatiate oir Barium 13kn',uih Titanium Oxsidecomipounds, unamibiguously The concentratilo of Si-ion list(ed inTahle I.- was, calculated through atontic ah;nrti ion specrroscopv

(b) .05 The powder compacts were siniered between 9f~t'- 1250t'00t 5-3IiThe density anid the weight loss data of thfe sintered compacts; arelisted in Table 1. It is necessary to increase the sinie-ing temperatureabove 1000*C to ad' eve sufficiently good densities, When filesamples were sintered over a platitutn sheet. comparatively betterdensities were observed at lower iemperatures But the bottomsurface of the ceramnic, especially in BT4. wasý bright yellow~ in

.13color, indicating the segregation of liquid phase. 'Ibisobevin.02 indicates that the viscosity of the liquid is low.0 .05The microstructure and the dielectric properties of 3 -Ai% Bi-

1 \ 2 added samples are very similar to that shown in Fig 2 and Fig 3.But the properties of 0.3 wt%- Bi- containing samples showedmarked diffevenres Thermicrositrucuralevaluation a a funiction ofsintering temperature is compared in Fig. 4a, h. c and d When thesamtples were qintered be~yond 1050'C, the avrage grain size of fileceramic increased to >h10g.in tin K vs T vplots, dielectric peaks;

__________________________ corresponin~ig to three transitions could he detiected ver-, clea rlv tIn0 Tahle 3. sonic of the dielectric properties are listed Thle dielectric-200 0 200 properties of these samples varied as a fulictinil of 'ticasuroing

frequenicies at all temperatures; Careful analysýis of thleTemp e rat ur e (OC) microstnucture. density, and the dielectric prnper~ ies of theceai

as a functioti of sititeringz temtperature and time tvidicaie thr grainFig I The dielecttri properties of BT2 ceratmic with 5%;w17r grot mechanism is dominantly diffusion conirolled in ndtu e

13,201i %itiered at various temperatures (a) K vs TI mid(h) tart 6 vs T: the properties wecre ritastneld at 1ktizSmittritig temperatures of ceramlic-, are (Ill 750'C. (2t1RtltC'. () RS5l'C, (4) tM)O'C.( and 0)t 950'C Thentagnilude are not corrected for porosiiv

71

04 Table 3: Physical propen ies of BT3 and BT4 ceramitc

I. D. Sint. Seter Wi loss Bulk pTemp

I . %C, gm/cc

S , BT3 IO0TO(-5 ZrO, 0.78 4 40)

I 1(00/1I Zro 2 058 4 54O ! 0()/3 ZrO' 0) gI 5,28

I000/2 ZrO2 0 52 5 1()9- t1050/I ZrO 2 0.55 5.28

1(150/I1.5 ZrOi 0 R4 5 361050/2 ZiO2 1 07 5 601 I M/2 ZrOl 1.181150/2 ZrO2 I. 1-7 5 h51200/2 ZrO2 I 37 5 681250/2 ZYO2 1.34 5S68

SBT4 8501/2 Pt1 0645S'•,9 4, . ."50/2 Zr0 2 ()92 4 76

A A y- 900/2 P1 0163 4 9995012 Pt 066 5 3!

"",q t Ii 1(80/ I ZrOl 092 5 22210/ ZO 1.36 5.82

Sa 1200/2 Zrlo 1.76 5.89

,12502 ZiG 2-03 5.91

i~~ a, ' " ., ,..~ •" Pan4.

In the Bail gSro 2TiO, (BST) powder used for thisinvestigation, small amount of carbonates were (leteced throughXRD pattems In a separate paper presented iII this-conferemne. heeffect of grain size on the dielectric properties of pure poroti,

,'- ) "ceramic is discussed In that stud%, the rtuallemt average gram sire"achieved through conventional sintering %k as 0 2 ptm. aid in this

V A. cerait i-. dielectric peaks corresponding to three transiltios are* - 'clearth identified

d% 1 The densification heha'ior of BST a- a function of flux" .. concentration and the sintering teinperature- are represented in Ftg

." 5 Pritnaily because of the presetice of strontium, it i- neces5sar% to"R -, ' •inter the ceramic with higher flux connentrattotns atd at hliher

t A' to- temperatures K vs-. T plot of the flux added cerauic are givet III*., , "•" ". , .i Fig 6 and 7. Suppression of dielectric peak,; at the tramttmtt

temperatures are clearly seen XR[) patten.: of these ceramtic failedto shom. splitting in <2(0> peaks due ioseiragonalit%

700 X0O 9010 14)(l0 1 ll00 1200 1300

Sint. "Temp C

Fig 4 Fyat hwie(l lmicrnsimlt turles of Bl' I eatl(, 1ql e .1t (:j) Fig 17fe. f|u.t vt'iVlvllIlted mlc l. • |i ~IlIl /h.(b) 110501C11h. 1,.) 111111C11hi. and id) of Sf(I2HBalR'iOj q'he'.e celalllc', %kete 'Itleled lot

115 *) C/2h 2hr

"72

4000 .2 odst.

Z I When smaller aniouni of BiOi was added in4. I~.'bh%,drothermal BaTi0j. to achieve high density, it was necessar) to

Ln sinter the compact at higher temperaturesU1 2. Addition of Bi,0 1 duruig tire hydrothemial preparation o(

200 BaTiO~.1 produced an unidentifiable Bi-based second phase Toui achieve high densities, even with 3 9A ,% Bi-)Cb. it was ntcessar) tf,

- rie the compacts beyond IIM)(tYC This ob~seration indicates theimportance of processing procedure.

Li 3. When the Bi2O1 concentration was reduced to 0).29 At7,LJ ~good dlensification. accompanied with grain growth: occurs heyond

IO(JO0*C. Careful consideration of weight loss observed duringim .......... .sintering. grain growth and the dielectric properties suggest that the

01. ...... ........ . ............... 0 densificafion is accomiplishted domninantly by grain boundar-y and- 100 0 100 bulk diffusion rtechantsmns

TEMPERRTURE (C) 4. When 5 w1% 130*i wsaddt ri,1alJ~powders. the compacts Sumter to hulk densities >5 2 gils/cc at95(YC. Percentage changes in TCC' calculated from C "s T plots of

Fig 6 K --s T plots of SBT`2( with 2wt% Bi-20i. which were this ceramic show Z513 industrial specificatinns, with a roomiineasured at I kHz. These ceramics were siniered at 1. temperature dielectric constwit of 1500) By changing the particie1250'C, 2- 12(0 0'C, 3. 1 15f0'C, and 4. 1I 100C for 2hr size and the concentration of the Bi,0 1 , it appears that it is posibte

to reduce the sintering temperature to 850*C. which 13 the iyptcalthick film processing temperature.

Acknowledgenments: Authors gratefully acknowledge the financialsupport from Cabot Corporation of Bovertown. PA and the BenFranklin Partnership Program of the- Commonwealth of PA,Authors; also acknowledges the technical support provided by V

r, Laubscher and Mingfong Song.

fu

S1000MC: References

III Ui. Kunmar. and J. P. Doug hertvSioteifo$g behaVIOr Of highpurity BaTiO, in the presence of Bi-)O1 as tire liquid phtase".Coi,,m,,oucatecl tn J, Airi Cer. Sce-

-121 Ui Kuniat. and)I P, Dotigherty,1)iclectnic propentes of fine. . ... .grain hydrothermal BaTIOI sintered vwith Bt2 O i.

O10 010 Commutticated to.1 Am Cer. Soc.-10 013J Cabot Patent

Temperature (AC)

Fig '7 K v" 1 plots of S131'20 with 5wtq7r Bi'() 'I. Which weremieaysured at I kHz. Thei ceramtics were sintered at 1.850C., 2 950fiC and 3. 1050'C

73

THIN- FILM- FERROELECTRICS

APPENDIX 48-

'111uN FILMNS AT 1ll11ll l~FRQULNCIE-1S

J. Clien., K, R_ I dayakumiin, K. G. BIrooks, and L. Ei CrossMaterials Research Laboratoryl'cimsvlvaiid Slate UntiversityUttiveisity Park, PA 16802

AblstractA Shifting the focus from lthe confines of tile bulkmaterials to tile dielectric behavior of ferroelectric

TIhe high frequency dielectric response of sol-gel thin filmts at low-amrplitude a.c. field of high(derived lead Z.irconlate litaitate (P7,1) thlin Wills has fiequeticies, as in the present study, thie dominantbeen investigated. Conceptnuali~iit lihe pitesenice of factors that i14fluetice relaxaticin appear to be connectediniterface layeriws cliticad ill cypllaillilg tilt kliclcctic to tile Presence of interface layers anld grainlmteasuieiienits. II v a carvlul -conti ni of lthe process-ing intperfections. For, first, these filmns have beenparameters, ii(led by i apid lhicima! annealing, lthe low characterized by very small grain sizes, in lthe 0.1-0.2frequenlcy dielectric chala cter sistics could he msiisaiied piln ranige 11,21, thiat has thie effcct of displacinig thieutito a G;117 range. P ao 5 Sio 51 i 0 whtich i is threshold freqluency due 10 the piezoelectric clamnpinigpmat clecifric at lonil tenjmpestm arllpci s to IWk a (rf Vrains or dontaiiis to higher levels. mnd second.potenlt i al ca id id ateitC 111Ciai br igI forl ig eqIueO (i) alr~i In f thle filmis through the sol -gel processapplicationls, entsures itaxitnal chemical purity. This study will.

consequeittly. argue that thie dielectric dispersion inl thle

fliti)(111diillI Mllz to several GJliz ranige stems front thie presentceI it t ttitti 1111of barrier layers mnd graini imperfections. buith of

Ili hes modrn imes may an vaied re lie which call be squarely linked to thle film processing.

ap1pl icat ions of hi gh hiequellc y d ielecti ics; inl highresolutionf digital coin mittncation deviceseiucom pass iit g cello I ar phonec mnd smte IIite Reisults alld Viscussioun

clittutcatittits, iii nticrm'elcctiomiics and packagingfor high slieed swilclri og mtode llo cr suppily,' ill The l'ZT films used inl thle study were of thlefreqtueitcy sensors for tnicrowvave dlefectioni. t plainly inorphotr-opic pltasc boundary comipositioni (with alist hut a few. Ijielectrics. (,Iialilvitig for such Zr[1'i mole ratio oif 52/41R), fabricated by thle sol-gel

applcatonsmoit pissss ighdieect icpettlitivty. spilt-oll technlique. The details of tile flilim falm icationllow dIissipationl loss, and1( Illow WTeollraltioc coctticicut as, well as lthe stiuctural and electropttysicalof elect~ophys1 ~icatl propce itic. I-eri telec iic materijals, character izattion of these filmts have beemi outlliited #inwhich constitute a uniquesitIC SlasslW (If diClcciriCS. while Refs. I and 2: ltme dielectric mid ferroelectric prop)ertyvet y promlisingp ill satisfyingp these genci al alto ihutes. mneasurements were limkited to tile radio frequencysuffer front a dmopl ill (Ike dielcktt ic coltstaat l a1 ranige inl these earlier studies, For dielectriccltaracteristic fit eqatecy . nscsi ihed to lthe pie7.tcleciric mneasu remineats iitll (is study, a Itight frequencytesotmance of thie ciySiafllio' as, %acll ais file iniettia oif fit ittpedmitce analyser (l11P 4191 A) was used. All file(II ltillt houlidam ie-s; oil lii ptawOI usi bl c xtaiamoits fillins were rapii 1Ithermally anne1)aled at telitperaturesadvaniced for tile pltimenvoite imtcludc lthe existence of and tittes as specified.mtierlace lavers at thle Ii lii-c let. tooe boundary.in pur ities itt the filim. mitd pranti imlpei fedions ars also Fg I i a plot of thle dielectric pertni ti vity as agraini boutidaries. At mtill higher fretlueticirs. ;in fmnction of frequenicy for a 0.395 ptll thtick PZT film.,addit iotia drop in lthe tel alive pti timiti vi ty may ()(ccut. rapid thermallIy anotealed inl the temperature range ofassigited genetrallv ito dielecitic tel as atiot. Il 60t0t-800"WC (or 6t0 Seconds. 1t is patent front the figureparaclecti ic maaterials. lo' pic ;i 'elect lic Cs 1Ctan11ce (1an thiat with iticre asitig amiteal iing temper ature . theoccur if lthe phasl-e is; fett,~nteril.te clamnped and reIlaxatimit Irequeticy decreases sysleitiatical ly oftcec dicicled tic ollstant. s' hit.h teltc to thic dicct.ICI i1c gt caete imimaon is. thie precipintous tumble iii Permtnitivttyco 'mstant at Itetlueticies above: ;tmd Ire Ii t pie zoelec it t to almost a ciphler at thit, ftequenicy. While this lattereStOnance meslpctivcly, ac te "'h' to titne another., At i pointi is seetitingiy haflfIing at first glanice. lite

givenl high It lelecv . dii' cc Itic tela sationl canl still Is equency respionse of thle measuved ohmelecitoic constahilcauise a tall[ ill ltme diecickto c 11cotmilivity inl thisý lioll anld tile cortespowtduitg imtpedatice spectruil from thlepiolar phase. etfui valent circuit model, based oit propierties of lthe

CiO318luxo 781)3 96.9N5 ~2.$3(mOQl(,ltl 182

film core, the electrode, and a series resistance 131.lends a modicunt of utideistanding to the attoitialouisdielectric behavior. Sayer et al. 131 postulated theformation of an interface layer at (lie top electrode- , 10filim boundary and (ie fihm-suhstale bounsdary; by 1varying the internal barri, layer thickness Irom (It)0 lo -pm to 0.5 tim (ilhe di, arbed layer at boilh the 0 0

interfaces are lumped) on a fil, (i of I tin total 'thickness, the dielectric constant was computed, which , , I [3has been reproiuced here as Figure 2. Scrutiny off this / )figure, and its juxtaposition with Fig. I, reveals a 06striking similarity, templing tile spc'ulation of the R F 0presence of interface layers of incieasing Ihickness R.- CF itio,

with increasing severity of annealing o tihe films. As • 10observed by the autlhors 131, the calcihaled higher 500 600 700 800dielectric constant for filmns with thicker barrier layers Annealing Temperature (C)might he related to the assumption of tie lullthickness of the film in. t(ie computations, wheti il

1000 -Fig. 3 Resistivity plot ted as a Iluctiom ol annealingtenm perauture.10000

C "50,C 1 04S 500

1z W

&AJ tO

10o 104 10' 10' 10, 10 o

Frequency (1iz)

Fig. I Htigh frequency dielectric response of P17T thin i

films, auuealed 61 lO-80)WC at a cimnstan dw 'll time of uL

61) seconds- Note the systematic change in tiherelaxati on frequeucy w i th annealing tem perature. 1 0 4 5 . 7 -J k,

1 0 1 0 1 0 1 0 10Relaxation Frequency (Hz)

2000 O.5,m Fig. 4 Relaxatioti fretric'icies detertoined for films of

varying Ihick ness shows a1 d(I p lot Ilhimler filtins.

1500

O. 1 ¢m rt ;ality, il stur hilt be soma lt -, by a thickiiess equal Io, fhal1of tile disrttleld layeis. In outr earlier study I 1, flhe

1000 o.oi'Um high Ircuimeitely dlielecilits relamtiot, Wats altluhile I ifiteO l lnltitn I)1 low lesi' ivily surface layers whoseresistivity changes with tile aniealtig conditions; tromn500 Fig, 3, it is apparetit thai higher amicahiug teniperalure

results in lower resistivity of this anomalous layer.0 i28aiintainmg the saiie Iiocessiug patmamneters, dielectric

0 2 4 6 relaxaion I ietlueicies of tilms vavyinig in IthickinessIog( ) from 0.15 i,, to 0.75 p'i, were determined (Fig. 4 )

Ihrititter films aic privy io lower rt laxationFig, 2 Co minputed dielectric contisl a tt of a 1 11m filit fr'lhrenncies, i 11dica 1iltp htei lllelled setisitivity to the

with varying iuternal barrier Ihickmess; the 1111t: value tteitace lae ialimi When tile thermal budgetwas Fedlited dintilg atntcralillg by lowerirg fIle dwello f r '=lI OW (repl o d uc ed fro n S ay e r ci at. 3I 1) 1 1 11i o to (it) S o.l,, , •s , I 11 secot i"., . th e re w a, s ,il

183

I 04 e vidlence (of ielaixaiioi Iin films anrtcale(I at 500t to6(WC '1 : (FPig. 5);. initer face l ayer gi owth mlay thus, he

10 seconids controlled to elm natI.Ie low, frequency *dielectricdisper~sionl. FIXtendingp this theme. b% inautipulatingp tile

1! 1 03~________----- 60"C p csilig ;iai anitet s, tile low fictlucncy dielci ti itpier ilt i vity and loss oif Ilite IITF filins could hie8O001C 7000C .s ~ ssaneil upno a (017,,. as shown Ii Fig, (6a. A li

U5506C sollitiogi svStCniI elnht acing cotlirposit ions thiat is

t I.STi slsem- iii: cciically. thinl filmns illi(tie 1U5Tsysitem, ()I Compolirsitiont coirespoiidiiig to a, Ba/Sr trode

atmi of 50l/50l, fabricated by ltie sol -gci chiemical10 ***-.~..'~.~~... 0 techiqicue. revealed dfispelrsion-free dielectric

10' 10, 10, 10 10' IQ* charactersistics (Fig. (il) bor iieasuiemntiis upto aFrequency (11z) GIl~z.

P~ig. 5 Dielectric pcIi IIIiItiýi jilottedI as a fiiti"oii (Ifrequeincy fur vat iiols allneal Ing teitilpela ,ittes but, witha reduced (well l ittle (if W (Iseconlds. S u 1ta ri

Earlier stud~ies of both bulk cei aiiics and dniiifiliirs has shuowii thlat file utility of feiroclectrics for

007a0.5 Ihigh liequenlcy applicirt totis is linmited onl account of0 zji tile nielectrilc (isliersioit iat I M117z to aI few, hundred

~ 80-0.4 N1117, dlependlinrg onl the specific material, ' lire present-enideaivor ha;s, stniwii that 11Y aI delibeuiiie interplay (if

600 o.(lie piocessilig pai aineiers. dielectric relaxation iii thielIZ*] flmhts, calli be 1 irevented upto aI G117. Fimn tha.

U ~~~0.2 aic cterocetlciic :ii loom lemnlerautni are i(Ii'l

sItIhc(.ted to relauxaition due to tile piez.oelectricclanrrpitig of graiiirs or dmhitais. and lthe inertil

Zoo0 0.1 respurinse rif doinaiti wall iiiovernenlt. Ii wounldIlierefore appewai prtitlent ito examinrle tile leasibiihtv of

0 0.0 filmis tflat Zile palaeclsric at tooml tcitiheratui e for10 10(0 1000 high Imirequircy alrplicatiriis * Butt 5Srtr 5iOi falls, Iii

Fre~fiueliCy (Mt l17) tIhir category, and tilie first results are i nideedpirrittsing.

300 -II I Clieii, K. R. i.1daakomr.iir K. (I Brooks, and L.F.Cross. "Rapid theroral annealing (of sol-gel derived

PZTthtif ______,_ ))1 Ph vs. ppl. 4456,May

0.2 ~ 19

.~ It), 0.1121 1K. W& tIdavakumma. J. Clein. S. It. Krullaiitlln. miod

,%kimchiing applhicationsý, hiloc. Sevenltilhil mt5yntp.

0 0.0 AppI. I:ctio)clctUic5,t( 1990 pp 741-431010 1000

Urequency 011l 7) N I 1Sayer A. Mnitsinglr, A~ K Ailmia. aind Ak Lo,"Ijiclecir ic reson)Ise oif ferrrciccletic (liiii filmos on Iioin-

itetili ectri~c fhitgiae le polci cs, 'voh 1Fig (I 'lite hicqoueircN chmaact:et istics of Iciroelectnic pp1. 129-140, 19Q_142hf 1 Ifilmis 1.1), aiid pataiche Ii(i K1ST I iliIIS (h).Ir otethie fooi-fi-dipcrsive Iiclavirr ripto tile Ilicasrired

ieicivifI(4 1 G11 liy aI c;uicfnl 1.0ijiid o1fithe growthII(if tile iiit~ci ace la yer.

18)4

"APENDIX .. 49.

Journal of the CetanlfC Society of Japan 100 [ 9 1 1091 - 1093 (1992) Paper

Changes in the Crystal Structure of RF-MagnetronSputtered BaTiO 3 Thin Films

Kenji UCHINO, Nam-Yang LEE, Tamaki TOBA, Narikazu USUKI, Hideaki ABURATANI and Yukio ITO

Deparinient oj'Phlrsics. Sophia Universtiv. 7-1, Ktoi-cho, Chiyoda.ku. Toklo 102

RF-7 .'., I] .,';,/• ;:C o BaTiO 3 01MIRR

Q*A* -T'~••• -

[Received December 11, 1991: Accepted May 21, 19921

The crystal structure of BaTiO3 thin films fabricated by line samples of BaTiO3, and clarified that both theRF-magnetron sputtering has been investigated. As- tetragonality, c/a, and the Curie temperature, To,sputtered films exhibited a cubic structure with a small are decreased drastically at a particle size of 0.12grain size of about 6-Snm. After annealing at a tern- pn.6 Fine particles less than this critical grain sizeperature above 1100'C, the crystal structure changed will not exhibit ferroelectricity because the crystalfrom cubic to tetragonal, because the annealing lattice becomes a cubic structure.process caused grain growth. The critical grain size of in orer t exlin th pen

the thin films which provided the cubic structure exist- In order to explain this phenomenon, we have

ed in the range of 0.1-0.2 pm. This value agreed well proposed an 'effective surface tension" model.6' 7 ' Awith the critical grain size of BaTiO 3 fine particles, fine particle with R in radius experiences a hydrostat-0.12pm. ic pressure of 2y/R (y: surface tension), and this

hydrostatic pressure decreases the Curie tempera-Key-u, orda : RF.iagneron sputering. BaTiO3 thin flm, ture of the particle down b.low room temperatureAnncaling procss. C, itical grain size. Phase transition and changes the structure into cubic. It is interesting

that the revealed in perovskite ferroelectrics ((Pb.Ba)TiO3 , (Ba, Sr)TiO3) shows an almost constant

1. Introduction value of 50 N/rn. This extraordinarily large valueWith increasing the demand for ferroelectric thin has not been explained yet, however, it may be at-

films for the applications to optical wave guides, non- tributed to the surface layer generated by the perma-volatile memories and so on, many studies have been nent dipoles.attempted using RF-sputtering and other thin filmfabrication techniques.)A4 However, ferroelectric 2. Experimentsfilms fabricated by these methods frequently have 2.1 Sample preparationnot shown so good ferroelectr:&ty as seen in bulk sin- The BaTiO 3 film was deposited by using an RF-tered ceramics or single crystals. Although many magnetron sputtering system (Anelva, SPF-researchers have attempted to give explanations to 43011S). Corning 7059 and fused quartz substratesthis phenomenon, they have been insufficient so far. with 15 mm x 15 mm x I mm in size were used.

Nagatomo et al. have reported that the ferroelec- All the substrates were cleaned by usingtric BaTiO3 thin film could be prepared at a relative- trichloroethane(TCE) and acetone, and then placedly low substrate temperature of 700"C by using RF- in a stainless steel holder by clips prior to film deposi-planar-magnetron sputtering, however, the ferroelec- tion. The target had a stoichiometric composition oftricity of those films was weak or absent for the fine- BaTiO3 with purity of 99.99%. The size was 100 mmgrained film, and it was necessary to anneal at in diameter and 5 mm in thickness. The sputtering1000'C in order to obtain ferroelectric BaTiO3 films conditions are summarized in Table 1. The thick-with high quality. 1 .4) They have explained these ness of the sputtered fihns was more than 2 pm,phenomena by the 90* domain model suggested by much thicker than the grain size.Arlt et al.-5l However, the reason why the fine- In order to control the grain size the as-sputteredgrained filn exhibits cubic symmetry and non-ferroe-lectric properties has not been discussed. Table 1. Summary of -puttering conditions.

The purpose of this study is to clarify why the as- v . ..... . ..grown BaTiO3 thin films do not show ferroelectrici- .ty. Our discussion is based upon the "critical grain "Too..'. d0,t. 4) -

size' niodel. [We have reported the particle/grain size depen- )..s

dence of ferroelectricity for powder and polycrystal- -"......

thin films were annealed at the temperature range of pared to the other planes with low indices.3"4 The600'-1200'C for 12 h. Chemical analysis of the com- decrease in the X-ray intensity and the peak broaden-position was made by the ICPS(Inductively Coupled ing for the substrate tempciature of 800"C seems toPlasma Spectroscopy) method and the Ba/Ti ratio be caused by the slight mis-orientation due to higherwas proved to be nearly 1.0. subitrate temperature.

2.2 Measurement The samples were annealed at various tempera-An X-ray diffractomneter (JEOL, J DX--1 I"A) was tures to increase the grain size so that they were corn-

employed to examine as-grown and annealed thin pared to the fine grain ceramic BaTiO3 . Figure 2films. The crystal symmetry anid the consequent lat- shows the XRD patterns of (200} reflection of filmstice constants of samples were calculated by averag- annealed at 1000', 1100%, and 1200'C. The diffrac-ing the {100} and {200} reflections of the BaTiO3 tion pattern of the film annealed at 1000'C shows aperovskite cell. Average grain size in the range of symmetrical pattern. On the other hand, the peak ofsub-micron was determined from SEM (Hitachi, S-900) nmicrographs using a line intercept method. Inthe range of grain size of nano-meter Scherrer'sequation 8) was used for {(100} peaks. The grain sizesdetermined by SEM and X-ray diffraction werealmost the same (-=0.1 pm) for the sampleannealed Týo0"C

at 1000°C; this suggests that the crystallite sizealmost corresponds to the grain size.

3. Results and discussionsFigure 1 shows the X-ray diffraction patterns of

as-grown films sputtered at the gas pressure of 1.0Pa for various substrate temperatures. The film sput-tered at room temperature showed an amorphousphase, hut the film fabricated above 500*C showed a Tz,'00occrystalline phase. The lattice constant decreasedwith the substrate temperature from 4.15A at 500 C 43 44 45 46 47

down to 4.07A at 800'C, but it was still larger than 2P ldci)

that of sintered BaTiO3 (a : 3.989A, c: 4.029A). Fig. 2. XRD peaks of the 1200} reflections of BaTiO3 annealedOnly two peaks due to the preferential orientation at various temperatures.

along J100} appeared and the preference did notchange significantly with the substrate temperature.This result is different from previous researches. It a) b)has been reported that if the mobility of atoms dur-ing deposition is high (i.e.. at a high substrate tem-perature), the films grow with the orientation of

MMI) due to its highest occupatinm density corn-

( tool %420C'-

0o. 5pr 0. 5•im

c: d)

"-.--___-. 600oC

c: .... .7001C. . . . _ _ao ' : -. . . .

20 if) 40 4L1 O. 5xn 0. 5ýxf

Fig-3. SEM photographs of BaTiOj thin films annealed at vari-Fig. 1. XR[) p.tterns of films sputtered on the suhstrate at vari- otus temperatures.ous temperatures. (a) 600'C, (b) 1000-C. (c) I100-C. (d) 1200"C.

Kenji l'CH1NO et al. J0107111 OfMCW (c'iMOf F &X ICA ofjalp t O 100 9 Q 942 11093

the filn annealed at 1100'C or 1200'C skews to the evaporated Ba'I'iOA thin films ranges over 35-70nmright side from the center, revealing an additional and the flin without heat-treat nient did not show anypeak at a lower angle. This seems to be attributed to ferroelectric properties.9 ' Although there are somethe appearance of the (002) peak, although the peak differences among these values due to fabricationwas not entirely separated from the (200) peak. In methods, both of them are below the critical grainconclusion, the phase transition from cubic to size, 0.12,pm. Accordingly, it can be thought that thetetragonal was caused by thermal annealing at the cubic structure and non-ferroelectric properties oftemperature above 1 100°C. The microstructure of an- as-sputtered films are attributed to the small grainnealed samples was observed by using SEM and the size.results are shown in Fig. 3. By the annealingprocess, the grains grew from about 4Onm to 350nm. 4. Conclusions

Figure 4 plots the relationship between the grain The results of this study are summarized.size and the annealing temperature, where the crys- (1) As-sputtered films prepdred on the substratetal structure is also indicated. The crystal structure of 500'-800'C exhibited a cubic crystal structure.was cubic below 0.1 pim in grain size, and tetragonal (2) The crystal 'structure of sputtered filmsabove 0.2 pm. Therefore, it is shown that the critical changed from cubic to tetragonal after annealed at agrain size might exist between 0.1 and 0.2 im. This temperature above 1100°C.result agrees exactly with that obtained by the study (3) The critical grain size separating cubic andon the particle size dependence of the crystal struc- tetragonal phases existed in the range of 0.1-0.2 pnm,ture of fine powder BaTiO3,6' where the critical and was in good agreement with the critical graingrain size was indicated to be 0.12 jini. The grain size of 0.12 pm determined from the fine powder ofsize of as-sputtered thin films ranged over 6-8nm in BaTiO3 .this study lijima has reported that the grain size of

References

11 T. Nagatomo and 0 Omoto, ]pn J Appl Phis. 26, Suppl,ra.:r .-- 26-2. 11-14 ý1987i-

0 Cbic Str-uct-e 2) K. Sreenivas. M. Sayer and P. Garrett, Thin Solid FdimzS :172. 251-67 (19891,

31 C H. LeeandS. I park.] Mfat Se Mat w Elect 1,219-24_cc - ' 1990).

4t T'. Nagatomo, T. Kosaka. S. Omon and 0. moto. Ferror'c-trics, 37, 681-84 11981).

2500 5) G. Arlt, D. Hennings and G de With, ] Appi Phys. 58,16i9-25 (1985).

"6) K. Uchino, E. Sadanaga and T. Ilirose, J. Am Ccram Soc..looc, 72, 1555-58 (1989).

7) T. Yarnakawa and K, Uchino. Proc 7th Internation. Syrmp

0 _ Jon Appl. Ferroelectrics. IEEE, pp. 610-12 (1990. Illinois,

600 800 1000 120C USA).

kvwealir Ttrý,rature (lC) 8) B. D. Cullity, "Elements of X-Ray Diffractions. Second Edi-tion, Ed. by Addison-Wesley Publishing Co., Inc., Mas-

Fig. 4. Relationship between grain size and annealing tempera- sachusetts (1979) pp. 99-106.lure. The crystal symmetry is also indicated. 9) Y. lijima.Jpnj Appti. I"os., 24, Suppl. 24-2, 401-03 (1985).

"APENDIX 50ý

36

SIZE EFFECTS IN FERROELECTRICTHIN FILMS

ROBERT E. NFWNIIAM. K It. Ill )AYAKUMAIt,AND SUSAN TROLIER-McKINSTRY

INTRODUCTION

i Beginning with work on the melfing behavior of metals, it has been reportedthait many plase transitions are susceptible ((I size cllects. The moIning poiiit ofbulk gold, for example, is 1337.58 K, but this temperature drops rapidly for grainsizes below 100 A (rig. 36. 1). This decrease in the phase transition temperaturehas been attributed to the change in the ratio of surface eneigy to volume energyas a function of par ticle size. Thus, for spherical particles of radius r, (hle Inctingtemperalure can be predicted from [Eq. 136.1)

AU dV- AS 'IdV- adA = 0 (36.1)

where At/ and AS are the changes in internal energy and entropy on melting, a is?rta: u1r ae1'z,:a), '1:, -'" cwect Ic e liquid and the solid, and "1I is the meltingIcmperature of the particlc. If AS and AU re t41`2 pcraQtIl teICpeICttt.ltLii'iIldilfercuce between the bulk and siall -article Inchting Iempcratures is inverselyproporlional to the particle radius according to

"l• - "i 2a" ;' = 2.y (36.2)

c'hronicl Prvr,•esmoigi cif Adivinco'f Alelarrial.s,

I~dicil by 1.arry t.. Mitcnch mid Jon K. Wc(:.

ISOIN 0-471-54201-6 kip 199)2 Jdulm Wilcy -mid Sms, | ,

379

380 CHAPTER 36: FERROELECTRIC THIN FILMS

~200

.41

* •

a00 200 S00

Figure 36.1. Melting Iemperaturc or gold parlictes as a function of size. Taken from Ref. [I].

where L is the heal of fusion, p the density, and T, the bulk melting point.It is interesting that many types of equilibrium phase transitions display

comparable size effects. In addition to the data on geld particles, the meltingtemperatures of copper [3), tin [4], indium [5], lead, and bismuth [2] particlesand thin films have all been shown to be size dependent. Similarly, thesuperconducting transition depends on size both intrinsically [6, 7] and extrinsi-cally if the stress exerted on the superconducting phase is different at differentsizes [8j. Superfluid transitions in he-iu _ .,tied powder compacts and thinfilms also depcnd on either the film thickness or the size of the pore diameter [9).

In fcrroic materials, both the prescnce of domain walls and the ferroictransition itself arc influenced by crystallite size. While this has not previouslybccni critical to most applic.tions, with the growing importance of thin-film- andsmall-particle-based devices, it is becoming important to understand the sizeeffects expected in ferroic materials.

REVIEW OF FERROIC SIZE EFFECTS

In brief, we expect four regions in the size dependence of ferroic properties (Fig.36.2). In large crystallites. multidomain effects accompanied by hysteresis takeplace. Reductions in size lead to single-domain particles and, at yet smaller sizes,to destabilized ferroics with large property coefficients. Finally, at sufficientlysmall sizes a reversion to normal behavior is expected at the point where therearc simply too few unit cells to sustain cooperative behavior. Similar transitionswith size arc expecled in secondary ferroics. As an introduction to the intrinsicsize effects in ferroelectric films, it is instructive to review what is known aboutthe transitions between regions in ferroic particles.

Beginning with the larger end of the size spectrum, it is well known that large-

REVIEW OF FERROIC SIZE EFFECTS 381

grained ferroic ceramics exhibit complex domain structures that are bordered byseveral types of domain walls. As the size of the system decreases, however, thevolume free energy necessarily decreases as well, and it becomes increasinglydifficult to support the free-energy term associated with domain walls [10].Consequently, the number of domains is expected to diminish as first one andthen the other types of domain walls are eliminated. The transition frompolydomain to single-domain behavior is well documented in a number offerromagnets. Pure iron suspended in mercury, for example, shows a critical sizefor conversion to single-domain behavior at -.23 nm and Feo.,Coo.6 a criticalsi7e of -~28 nm [I I]. This is in good agreement with the calculations of Kittel,who suggested 20 nm as the minimum size for multidomain behavior in magnets[10]. Results for acicular agglomerates of y-Fe,2 3 particles separated bynonmagnetic grain boundaries are also consistent with these estimates; Berkow-itz et al. [1 2] report that the stable single-domain range at room temperature iscentered at ,--40nm.

Although the loss of multidomain behavior in ferroelectric ceramics is knownto occur for much larger grain sizes (approximately several tenths of amicrometer), it is also clear that the stress state in a monolithic body containingdomains is considerably more complex than that for isolated particles. Conseq-uently, the changes in ferroic properties as a function of size in ferroclectricpowders are expected to follow the ferromagnetic analog more closely. This isborn out in experiments on 0-3 PbTiO 3/polymer composites in which the tiller

F iii, o ne I, m Inf .e It am I& -. F ft Ial, Ir sittw

Single domain

SV00008affmag,'e4C supsepa'aelecific Sup*'Pata.Iaam.c

li10 i @ B 89 00

P'iamfflag 'r.C PA-a8'sI,,lo Paraol ectfic

F 3ra

Figure 36.2- Transitions in ferroic behavior as a function Of Mie.

382 CHAPTER 36: FERROELECTRIC THIN FILMS

particle size was varied. Lee et al. [13] report that there is a significant drop inthe ability to pole such composites-for filler particles smaller than -200nm.Assuming both that the small particles were well crystallized and that theparticle size distributions were narrow, this implies that transition frommultidomain to single domain occurs in PbTiO3 particles near 200nm.

At still smaller sizes, ferroic materials undergo a phase change to the high-temperaturc prototype group. In the case of ferromagnetic particles, this hasbeen correlated with the size at which the decrease in volume free energyaccompanying magnetization is on the order of the thermal energy [14]. As aresult, the spin direction is randomized with lime, leading to an unmagnetizedbut highly orien table single-domain crystal. Thus, a magnet in this size regime ischaracterized by a zero net magnetization, the disappearance of a magnetichysteresis loop, extremely high magnetic susceptibilities, and a symmetry that,on average, is higher than that of the ferromagnetic phase. Iron exhibitssuperparamnagnctic behavior at particle sizes near 7nm [11], V-Fe 20 3 at-30 nm [12] and BaFc,1 2- 2,TiCoO 1 9 at 15-35 nm depending on the stoich-

iornetry and the degree of particle shape anisotropy [15]. The loss of the ferroichysteresis loop coupled with the ability to respond strongly to the presence of amagnetic field has beef, utilized in a variety of applications, including fer-romagnetic fluids and high-frequency transformers where eddy current lossesare a problem.

A similar mechanism has been proposed to explain the dielectric and elasticproperties of relaxor fcrroclcctrics L16, 17]. Compositions including many of theA(I3/ 2, Bj';2)O and A(B'2t., l,';;.t)O, exhibit microdomains (typically 2-30nm insize) of I : I ordering oti the B sublattice dispersed in a disordered matrix. It hasbeen suggested that as a result of this nanostructure, the spontaneous polarizal-ion in these matlerials is also subdivided into very sm.;ll local regions. Thus, alead magicsiuin niobate ceramic can be regarded as a collection or disorderedbut highly orientable dipolcs. Thc result, much like the case of superpara-magnetism, is a high dielectric permittivity over a broad temperature range eventhough the net spontaneous polarization is zero. Because of the long range of t(ieelectric fields induced by tife local dipoles, however, there is more interactionbetween the local electric dipoles than was present in the ideal superparamagnet.Consequently. the superparaclectric behavior in relaxor ferroelectrics is modu-lated by coupling between local moments, so that a spin glass model is necessaryto describe the phase transition behavior [17]. Evidence for the importance ofthe size of lhe microregions is given in experiments on materials that can beordered by heat treatment. As the scale of the ordered regions grow beyond acertain size, the material reverts to ordinary ferroelectric (or artiferroelectric)behavior with a well-defined transition temperature and a nondispersivedielectric response.

As yet, direct observation of superparaelectric behavior in particulateordinary ferroelectrics has not been documented. Recently, however, manyinvestigators have attempted to determine the critical size for reversion to thehigh-temperature prototype symmetry [18-20] 'ýs in the case of the melting of

REVIEW OF FERROIC SIZE EFFECTS 383

PbTIO3 di•nwte, (rnm)

0 100 200 300520

4800*. 440

460

* * 420

S120U110

100

so

100 200 400 00 800 1000

81TiO3 size (..m)

Figure 36.3. Transition teriperature for ferroclectric powder. Data from Refs. [ 18 19].

macial particles, as the particle size is decreased, tihe transition lemperature dropsmarkedly (Fig. 36.3). While at larger sizes thie high-symmetry phase is theordinary paraclectric plhase, as the size of the ferroelectric particles becomessmall enough that thermal energy can disorder hlie dipoles, there should he it

transition from paraclectric to superparaelectric particles. The reported resultsindicate that unconstrained Bal'iO 3 particles show the transition to ia 'ubtcphase near 80- 12Onom [19], whereas PbTiO 3 is stable in the letragonal form to-10-20tim [13, 18]. Blecause it is difficult to characterize the electrical

properties of such small particles, it is not known if and when the high-symmetryparticles actually become superparaelectric. On the other hand, NaNO 2 showsonly some broadening in the differential thermal analysis (DTA) characteristicfor the ferroelectric phase transition with no change in the transition tempcr-ature for particle sizes down to 5 nm. While it is an order-disorder ferroelect ric,and so might have different size dependence for the properties than would adisplacive ferroelectrc, it is interesting that there is no evidence for su!crpara-electricity even at particle sizes of 5 nm.

It is also critical to note that it is possible to shift the critical size for reversionto high-temperature symmetry with changes in the processing. Residual strailus,in particular, have been shown to drastically affect the propertiLý of lBaTiO,[21]. 1Thus, it is not surprising that heavily milled BlaTiO, powders %,tIh anaverage raditis of -•n( liu have been shown to possess permanent dipolemoments [22].

Saegusa el al. [20] examined the solid solution between llaTiO 3 and Plb'FiO,to determine thie critical size for stabilization of the cubic phase to room

384 CHAPTER 36: FERROELECTRIC THIN FILMS

feniii pratm Li Ces a function of composition. Assuming that crystallinity andstOIcLieonietrV were inaititaitncd for the smiallest sizes, their data suggest that tliecc tical sj.'e is not it Ii ca r functioll of comicposition. It would be interesting tofollow Ole cuagnit ode of (lhe polarization as a function of temperature in) such;

po%% ders to see how its mnagnitude is affected by particle size.Althbough thle gecicratl outline of size effects expected in films are similar to

those demionstrated in ferromagnetic particles. it would not be surprising for thehi iinda tics biet weco differcii t regions to shift along t lie size axis for differentgeoliret ices. K it tl %asonice of the first in ex pinie (illrs possibility in ferromagneticiiialeik e1101I) . 1 Ic cabrlatl~tcd (that itn a Iiliic where thie preferred maginetiz..rtioisliciliciiidictilmi to thre nmajor sue faicc the cintilt Idoiniiiii to single-domain transitionl:ýhoitld lwct1 lcill neai30lnl. an oi der of maginitude larger thcan tIre saine tranisitioniill ia paitl iciila te Ilintetfial. If, howe%%ver. ( here was cithter nio a nisot ropy or the eais%liiagilcl tiatifoil i (11ctioii fell ill the film plane. ncw types of domain walls mayappear anid persist down to very smiall size%.

Thie tiiiinsitioec to supcr-paraiitagiietic behavior is apparently suppressed incomiipac at i rel perfect thcicc lilins produced iii higic-vactiune systems. probablybeca use thec large two-dimiensioneal area raises thie volume einergy above tliea~ aila ble I licemal energv. even for very thin filmis- Coiscsqucntly. regularlcr roenageictic bhelriior persists in much thinner films (downl to -0.5cmi for NiIlahins sandss nehied between nonniagnetic layers [23)). let such samples. thieeiciomnagiet ic (ran-wict ii tecmperat ure also drops rapidly for filmns < I neil thick.

[-iliis that grow as discrete islands of magnetic matcrial rather thancon tleintios Ia sees or thi~it consisjt of mnagnetic particles isola ted from cacti ot leerby niommingnetic hydride or oxide layers, on tile other hand, should behave likepali ctlat Ce s-stencs (und1ergoing stiperparaciiagiietic t raesitions appropriate forthc prlinicary nmagnetic poirticle size). This type of bcliavior thas been confirmed inferrocinagneiic filins that became supcr-paraniagrictic at apparent thickenesses of2.7 nin twhcre thie lilto actually consisted of 5-1I0-nm islands [24]). It isinteresting that no change in the Curie temperature was noted before Chie onsetof iecrineal ranidomization of the spitns.

Gixeci this ienforemation, whcat can be predicted about thie behavior offerroelectrie thin filnms*? First, for ane uncileetroded film it scems likely thlatferroclect ricity will remaien stable at least down to thick nesses of - 100 cil for

RaliO ' aned possibly considerably lower for PbTiO, films. This is in accordwit h several theoretical predictions for size effect., ie thini ferroelectric wafers. Illthat work, it was shown that when space charge effects (whlich will be consideredicc tire following section;) are discounted, thle onset of intrinsic size effects isproijccted to fall in the range of I - t0 nell [25].

Multidoinairi confliguirations should also remaien stable down to very smallsizes. C'orroblorating evidence for this is suggested by transmission clectron

encicoscopyi (TiM ) studies of thlinnied ferr-oelectric cecramlics anid single crystalswhere, provided thce grain size is large enough, domains can be detected in foilsbelow 10(X1 ci thick L261. When, however, the grain size in thie film falls below

THIN-FILM SIZE EFFECTS 385

some critical limit, the density of domains will probably decrease in a mannersimilar to that shown in ceramics [27].

Finally, as in the case of the magnetic films, unless there is some extriiiscmechanism for forming discrete polar microregions, the onset of superpara-electricity should be depressed by the relatively large volnie of ferroelectric.

THIN-FILM SIZE EFFECTS

The experimentally observed size effects reported in the ferroelectric thin-filmliterature fall into four major categories: an increase in the coercive field, a

decrease in the renianent polari/atioll, a depression in tihe dielectric constant,and a smearing of the paraelectric--ferroclectric transition over vallucs expeciedfronm bulk materials of the same composition. Typic.,, d,.i on tle lirsI IWOobservations are listed in Table 36.1, the third is depicted in Fig. 36.4, alnd ihefourth point is discussed by Ifiryukov et ail. [281]. One point that is immnediatctyappalrent from Table 36.1 and a perilsal I•o literalurc data is I[hial deviations from

bulk properties are nearly universal, bul the thickness at which the propertiesbegin Io diverge and the magnitude of the disparity are stiongly dependent on

200

-00 -- 00 0 fO0 6OO07, OC

Figure' .16.4. Temperature dlependenuccn OIft la tan;d Ian A (b) fo~r (11a, Sr)'rio)] lims Filhn Ihiicknes% d

tn~cYOmeletTS): 11) O, 1; (2) 0,31,;13) (0.5; 14) 1.,0,15) 2.0;,16) 5.0117') 81.. Data from Ref. 1521.

386 CHAPTER 36: FERROELECTRIC THIN FILMS

I ARIl 36.1. Reinangtint Iolari/,ation and Coercive Field in a Number of Perovskite Films

M ci l" l',(pC/cm 2) E, (k V/cm) Reference

Singlc-crysull I'htx()j -55 75 6.75 35So) gel PhTi., 3.37 548 36SlpUlicrCd (l)I 0 I') I'i(i 55 75 37Spulttered 1(X)W) Pli( ), 35 1601 38Slomt rerd o(r CVI) PIliiO) 12 250 39(*VI) (WIl) I'l lIio 14.1 20.16 40hulk PIi 58/42 45 17 41sI) gel PZ'i 36 42Sot-ge Ilr 53/47 12 150 43Sol) gel I'ZT 52148 35 34So) gel 11'r 18-20 50-60 44Sol -gel I1ZT 40410 6.6 26.7 45Spullcied I'ZI 90/1,l) 13.9 60.0 46Sltlltered PIT 58/42 30.0 25 41Sputtccd 'Z'r 65/35. weak (100) 12.5 90 47Spittesed PZ'F 65/35 3.6 33 47SIl gel PLZT 2154/46 28.5 190 48Single-crystal IaTiiO• 26 I 49Pol•cr stalline Ila iO, 8 3 49Sput, leted IBaito, 0.8 3 50Scren-printcd lani, ,,,Sn,, O. 1.7-2.8 25 51"Spot leled ((X)I 111a I00, 7 60 33Spullered iBaTiO 3, weak (101) 16 20 33

"Ahhrc.t.ahton; (.VI. chermical varx)r deposition; PZT, lead zirconale titanate PLZT. leadlallhal olll l i~l'ic',llate litarile)I

the picparatioit conditions. Consequently, films produced at one laboratorymay display marked size effects, while others of the same thickness andctnmposition possess bulk properties.

ihe reasons for this type of discrepancy lie in the variety of mechanismscausing the apparent size effect. Included among these are microstrucluralhcterogencitics, variations in crystalline quality, mechanical stresses imposed onthe lilm by the substrate, space charge effects, and finally intrinsic size effects Itis critical to note that the first two of these, which in the opinion of the authorsaccount for the majority of the "size effects" observed in thin-filn properties, are,m fact. si7.C indcpcndcnt. It is fortuitous, then, that many film preparationtechniques produce ilims that are defective and would remain defective even ifmacro)scopic samplcs could be fabricated.

Inhomogencily in the film microstructurc can take the form of incorporatedporosity, surface and interface roughness, or variations in the grain size. Inparticflar, utany furroclectrics grown by vapor deposition processes arecolunmar and should be expected to have low densities (Fig. 36.5). This, in turn,

THIN-FILM SIZE EFFECTS 387

Dense Tissue Columnn Particle

poros Coumn trucu eRecryst. Particle

H.0zone~ 0.9

0.0

30 2001~0.0.

Ar ~ ~ 0 Pr0s (mTrr' Substrate temp. T/Tm

Figuire 36.5. Sirticlure of spullered film as a function of gais prcssuie irnd normistizcd st.bsrarietemiperature 1371.

could appreciably lower the dielectric constant and, if there was poor couplingbetween lthe grains, would increase Ole coercive field as well [29]. Eiven ill fuisthat appear dense, I3udikcvich) et al. hiave shown that tlic nmicros[ rucl tre maychange continuously as a function or lilm thickness 30.1. -1 lits, for sputotercdfla0 85Sro I TiO3 tilins. thiinner films tend to he composed of small particles(15 onm for a 4-nmn-thick filin) where Ithicker films show a distribution of grainlsizes raniging from thie very line particles deposited next to lthe substrate to largergrains at the film surface (2CO-300nm for a film 20WX lint thick). Given thlis typeofnticrostructural heterogeneity, it is no wonder that many properties appear todepend on film thicknecss. As the absolute grain size at any givenl thickness isexpected to be a sensitive function of both the deposition conditions and anlyposwlianneang, samlplcs prepared at differentI lahora torics should 1(bhavedifferently.

A second significant influence on thin-film propertics is the crystalline qualityof lthe ferroelectric material. It is known thiat the toss of clear X-ray diffractionpeaks is coupled to (fie disappearance of the paritclectric-ferroclectric phasctransition. Unfortunately, many film deposition techniques also result in poorcrystallinity. During sputtering, for example, the growing film is subjected tobombardment hy high-energy ions. While this can be advantageous in terms ofproviding additional energy to (the deposit and increasing the surface mobility,heavy bombardment, particularly at low substrate temperatures, can alsointroduce high defect concentrat ions. In chemiically prepared thin films, on theoilher hand, low annealing templeratures can be insufficient to crystallize tilefierroieletric phase fully.

Fast neutron irradiatoiiu I,' Il;'TiO3 !ziigle crystals (neutron energy _> SO keV),for example, eventually introduces sufficient defects thiat the material It 1sfoi'rmsinto a metastable cubic state withi an expanded lattice parameter [3 1]. Havingundergone this transition, no displacive transformation to the teiragonat,ferroclectric, slate can be detected at temipera(tures above 78 K. Although the

388 CHAPTER 36: FERROELECTRIC THIN FILMS

bulk still displays somte long-range order, thce X-ray peaks are broadened by afactor of 5 oser thie unirradiated crystal, and it is not unlikely that hie surfacewas conpi prc ainot ph i~cd- A nucalhug to I N(W-C is requited before thle latticep11at ,rreltr relases hack to its mimial value. Unufortunately, in many thini filmls,arciniehliig teniper itiircs are kept ats low as possible to minimliize changes InIstulciolirunt!%. While thlis mlay help Ilainitainl thle perovsk-iie structure, it is

h1111111 11;1 SIN 70 is inrt-ficieent- to- crystarlli7,e amorphous -or badlytlamaii~get hllin Ii flly NI an v .iaut hors wotking withi cit hcr vacuum-deposited orchenrlically pi cpmrid thbin lilins ieport miodilicatioris of thle pcrovskitc structurewith .1 slighlyt ex~panded cubic unit cell 132].

III Lcharactcci jng the crystallinity of feisoelectrie films, Surowiak ci al [133ircistiiecd tat tie strains inl spurtfer-deposiled la~ijO filmns. They found dilt

Willsi' that cx percieiced lrcavy bomtbardicriir during growth tended to have mnorelric~~l .cu5 dfoinied crxsrallites tic., thle mean inicrodef'ormiatioll Ad/d was as largeas 0,01 OVAIS)S anld smiall coherent scalttering Si7C5. L~arger, less defectivecryst all i tes Ad d < O.tW451 could be formied whet' the gr out hcoridil ions were not

asrgot ots 1 33]. IDillerences between thie twAo typcs of filmns were readIlyappal en in lthe electrical proper-ties: lower A11/4 values were associated with)11101 le mi ncl t polar I uat ionls. piezoelectric coisia nts close to single-crystal%aluecs, and ielati~cly niarrow phase I ia sitlions. This, last pointi was exatmined byBit 'vuko% el al. 1 281. w~ho demonstrated that films, with large lattice Strainls"slionid be eicc perd to show diffuse phase I ratisitions. Poorly crystallized filmisfrom any pirepar atioii method will probably display low remianenit polarizationl,lowered dielectric arid elect romnechianical coupling contstants, and diffuse phaserransritols-

It i mr% dlo stir plm ingta r1.1 mechanical stresses should ailu afflect thin-film1) p~i plies A% iii most ferroic narerital. lite ;ippearamince of tile order paramecterait r lie hiai,~iiioii tenmpicatrume is accompaniled inl perovskitc l'erroclectrics by aspoiitaineons strainl. I orlirai structure should he irillueniced by lthe types ofstrains pieserirt ii the film. In) a simiilar way, stresses inl letcroepitaxial filmis havebeeni showni to alter tile equilibrium domain strueture. Two-dimnensional stressesCall Stabiliie lthe ferroelectric phase to higher temperatures in bulk materials andthis 1irciecainsin could operate ill th1in films whent there is good cohesion betweentime substrate arid tive film.

'Ilie primary difference between size: effects iii ferromagnectic materials andferroeleetric materials is that, in thle celctrical analog, it is necessary tocomlpenlsate: tle polarizationl at thle surface of the material. InI a fcrroelectrrc thatis slightly coniductinig or elctiroded with a material with a low carricr denisiry.tremendours depolarization fields or space charge migration can be genecratedC~cii ill comparatively thick lInts ( -I pmi). These carl shift tile phase transitiontemperature, lower thie magnitude of thie spontaneous polarization, and evenldestabilie thle ferroelectric phase in lthe film.

IDcspitc all of these opportunities for extrinsic siz.e effects,, therec is evidenceihar Will% thlar are prepared carefully can displary near bulk properties to very

THIN-FILM SIZE EFFECTS 389

1400 0.08

-0.071200.

1 1000 .0510 C6O.04

. 800-"V' . - . - ,#~ 0.023

600 0.020 2000. 4000' 6000'

Thickness (A)

Figure 36.6. Dieleciric constant and dispersion factor as a function of filn thickness

small thicknesses. In work on sol-gel l'ZT filins. Udayakum;ir t34], forexample, showed that roon temperature :lielCctric contstanlts of - 3(g) could hemaintained for films exceeding 3()0im thick 11'ig. 36.6). The bulk remanientpolarization was retained to 4503nrn and remained finite, though reduced. iifilms !90nm (hick (Fig. 367). The high breakdown strength of these films (Fig.37.8) will also be critical in device applicalions [34].

In work on RF-sputlered BalTiO, films, Dudkxvich c( al. [30] showed ihatthe site of thie coherent scattering region,. D, within their filns was moreimportant than the thickness in determining the microscopic electrical pro-pertics. As seen in Fig. 36.9, tie dielectric constant increases markedly as the

PZT So#-Uel Thin Film

$ 40-

c 20- S120 A

,a 30 4-4012.-U--e- 3340A

------ No00A

1010 100 200 300 400

Electric Field (kV/cm)

Figure 36.7. Remanent pol'rii;ilion in PZT sot gel hin filins as a funo•niii of' lcciric ficld.

390 CHAPTER 36: FERROELECTRIC THIN FILMS

E 1.5

- 1.20 0

0

• 0.9en

6. 0.6

" 0.3

V

"-• j0.0,1000 2000 3000 4000 5000 6000

Thickness (A)

Figtire 36.9. Brcakdnwn strength a.,. a funclion of film Ihickness.

coherent scaltCeing sie grows Iarg•cr than - 30 n,. and when 1) reached 50nm,rtoom empcrature dielectric conslants exceeding l00) wcrc achievcd, and someindication of ai dielectric constatit peak could bc detcctcd at the Curietctieperature [3(1.

As shown in Table 36.1. scvc al othcr recent, papers have demonstrated that itit tsmshlc It achieve bulk or near bulk properties for several members of the

lcrovsk itC family.

1200 " 1 1 1 600

t1000 f 5D

800- 400

-600 * 300

o 0400 . 200

200 • 100

o , , o

100 200 300 400 500

fI (10-

I:ig.lr 36.9. Sinc of1 he Coherent scattering region icure I tand the ititectric permittivity icurve 2) of

II.11a I• film-. as a fuliction of subsitate tclflpetalutC. 'I aken I htrm Ref 30).

REFERENCES 391

CONCLUSION S

As the fmild of ferroelectir th inn filns grows, it becomes increasingly important 1o

examine the role of size effects on the expected properties. While intrinsic sizeeffects similar to those demonstrated in fcrrornagnctic analogs will act as lowerlimits to the ,~ize of ferroclectric devices, in many cases extrinsic, processing-induced con tribuitions overshtadow (fie fundamental size rest rict ions. Consecq-uently, careful characterization of films to determine thie role of extrinsic effects(i.e., internal micrnstructitrc, interface layers, and poor crystallinity) are neces-sary to understand relationships between processing and properties in fer-roelectric films.

REFERENCES

L. Ph. Bflutat and J. P. Borel, Size [ffect on the Melting Tenipcrai tre of Gold Pajrticles, Phys Rev.A 1.1(6), 2287 -2298 11976).

2. M. lak agi. Iilectron-tDiffraciioai S1tidy of Liquid -Solid I raiatioiin inIi hill Metal I tittas, J1 Pill-Soc. Jfn., 9(3), 3 59 -363 (19 54).

3. Ný T- il~adkich, R. Niederrnayer, and K, Spiegel. Nahwlias grotkr %chaniclipuiiiktsernictiri-gtiingen hei diinnen rnetallschclfien, I'ii 1. Smt al.No., 15, 181 191 109661

4. Mt. Iflack man and A. F Curo~nt. O n tile Size tDepeaide nce of tilie Melting a nil Soi di tic;itii i i

Temperatures of Sinall P~articles of Tin. Sltliuniire anid l'o'perne 4 I'j a iiiilpmI,i. Wile), Nei%York (1959).

5 K. T. Boiko, A. T. Pugiacliev, and Y. N11. Bralsykluiin. Melting of Sul-criticallly lIon CoindenseidFilmns of Indium, Sot'. Phys. Soll. State. 11812). 2832 28A4 I1969).

6. M. Isto, Fluletuto ion Effect oii the KnIIighit Shift of Slapercilndaictimiig Al Fine Part icic. J On ik

Soc Jpn, 41(2), 412 -418 H119761.

7 1. FIllanik ei al., Properfies. of Stipercoiiiuctii'ag NhTi Suiperfine Filament Ciomposites %viilitOianseiers < - 0,1 lni. Crtlinit-rt s. 25. 559 - 565 (1985)

S1 11 P. Watson, Criiicail MIagiilic Fieluf and Transition Teloperatitre (if Synthetic Ifigh-finedSuipercoinductors, I'Il)5 Rci' 14140l 221 2310 f19661,

9 T. Chaainer, Y. NMorii, and1( If. Kojiina. Size [tfects iii Supricfluid 'lie. Il'hyNs R4,1,. 11, 2111)1. 34413944 (198(11.

It0 C. K ittei. Theory of the Structure of Ferromagrietic foniaiin% in Films and Small Particles. Phl is

Rel.. 71X 11, 1 2j, 905) 971 (10461,f

11. E' F. K le~ller adt 17- 17,. Luluorsky, Parli~le Size D epenidence of ('urcrivity and ltcnimicnce %ifSiitgk.Dorn~ain I'artucles, J. Arpp. I'hY... .4 3). 656 65H 119(4)

12. A. E. 11crkowiiz W. i. Sclaticc. arid P. J. Flanders, Influence of Crysiaultte Size ol thie MiagneticProperties of Aciesilar T-Fe2O, Particles. J. Appi. Piqs.. 3912). 1261- 1263 419681.

13 MI. 11. Lee. A. lluIliyal. and R. F. Newiaham, Poling Studies (if 1'acocelciric (ioilpo'sitcsPrepared by Copreciptialed PhTiOl Powder, Fg'rrocieiizc~s. 87, 71-801 (19X81

14. C. P. Bhean and 1. S. Jacobs, J. Appl. Plays.. 27(12), 1448 (19S61.IS. 0. Kubo, T. Ido. 11, Yokoytana. and Y. Koike, Particle Size Effects on Magnettc Properties (if

13dFe 1z 21 .Ti.Co.O 19 Finte Particles, J. Appi. Pltys., 57(11. 428(0-4282 11985).16. L 1E. Cross, Relaxor Ferroelecirics, Ferroeluectri -s. 7613/4). 241- 267 (19871.

17. D. Vieltlatad, S. J. lang. L. E. Cross. and M, Wuttig. Freezing of the Poilatrization Fluctuations inLead Magnesiunt Niohiite Relaxors, J- Appl. I'Iays, 6*16), 2918 - 292! (91990)

392 CHAPTER 36. FERROELECTRIC THIN FILMS

P'h I If), I Itrailiie l' hvices 16-1s.ki It, 311101. 5852 5X5819I'89)

11) K I ticini'. preprinti. Ainericon (*eraiiiic society annual nieetiii; f 191(9)

'it K. isegiiso. %V F, %N hine. and I I 1K Ilo%%efii lreparalnnl i f .b Itj it) anid lilt eflre of(ii opi isi I on aiid [ lie Sim (if filie Cr Ystilli Ic of' tlie Cr s al k se. j pnicti t, Arter ican Ce ramuic

Society tinitna meeting 11989)

21I S I W110iii i . I Nial s~it n . A tid G I oda. Fffeci- of lil Lapped Sorface Lav ers on (lIke 1)ielcotniPincritic, iof I ci roietetc Ceramitcs. J. AtppI. IlJis %, Mil1l). 5839 5R44 1 1980)

22 R Ba;clitia tin and K Ilait er. Si ablec Stisp~eisti in of Ferroc levi nc taTio, Parlicl". Sal State( niiittinii . 6901. 1(65 8 69 119918)

23 P' Iltct ii lai . c (lit pert . P. 1'. Renard. C. Miarlicre, and 1) R enard. NI agiie I ii.Interaction

Ilcisseeti I Ilirtlliii Nickel I ilim 'I lirioigl a Gonld literlayer. tii Afmqinviii 1Thin FI-Ati, 1996. Rl

K rislii an d . les tafltn itin c phy sique. I frtive iII 98,l

24 NI l'ruili'ii 1 )ilt I er'romitrtifir' I ibis. jivilomsorthi and ('t Ltd. Washington j194641

25 K Ilintidet. Skulacec effcts (1in Phase Tratisitions in i-crmoelectinies arid Aniliferroekcori.s

Verroelecirics 35, 99- 10)4 (1981)

26v NI I a nak a and 6 1 loott i.I:clron Opt11ical Studies of Blartumi I i anate Single Cr ystal I Ifni$s. JP'h I Sm .11mn. 191(f1. 954 9 70 11964),

27 (i Ant., 1) lletinitgs. td 6, de W~illi Dielectric Propetties tof Fine-Graitied Barium TitanuileCeraiiiics- .1 ..l;pi ('l1Ii .RA')1. 1619 1625 (I 985).

28, S V Ilir% ikit. V NI Niduk orlo'.. A. M. Margolin. Yu 1. Golityko. I N Zakharchrrnko, V PDu)tdkes icli. aiii I:. (;. I esetiki. Phlase iransinioits i Piolvcrystallnie arid IleiciocpioarisiI eirroelecl nc I-IillSi. I triocltiotis. 56. 11 5-11 I1984),

21) 0 Jiuii. Ilie lleitiityl'aiii State UlitertIty privalte cntnmnuioi~cation (199(11

ItII V V' DoI )kcsIi~h. V. A IIuk tees VI NI hNI ihrkl tnYu i I ( otlivkoi. Yo iG S itidees. V NINhIi khlor lii. andir 1. G F<esc ii. liter hal Size Efflect ti Critde usd la ijO, I- erroiute1v tic F (im.

Ph(is Se "'l 4it). 6'5 461, 46t7 tlqgl

I1 Ni C . V'ttils iaid I A Shierrill, I ast Neultron Eff~ects in Teiragitial Blariumn Titanate. J Art)

2 'i Na~l, a t N,%alsi;,Wi N 'mwa~ss,. mids l%1 tmlplvo I, 1 liaige inm tNt etista.Iale Cubtic to 5 ~I cilaignitil I i'rli of siimltictivri Iloriiiii 'I tinaome. B/ull Chiem. S,,i Jpti. 47(51. I1169 1171 (19741

33 / Sntomits.a k. A NI NI arg. ini. 1. N. Zak harehteikoi. tand S V Ilir ukov. Tile Inifluence ofi'Snliitioireo fittlie Pic7t 'ecl nc I'm pen tesofi Ba Ti), and I IlaSr11, 'th~ tin Ftiliims ' i Iii a IDiffusk:Ilmc'liaae i ntution. Th/iin SobSd 1 ihnts 176, 227- 246 (19891(9.

14 K W. I (la.yakmtioa . J. ('licit. S It Krkitpa nd hi. an 1,iI 1. (r,"%. St I- Gel D erimve P ZT ThinI ilint (tir S~kt iclinp Alpliectittiom. piper presented at ilic loetuietifriiuii Symiposiunm on

Appitaitons tit Ferroelectics 199) 11991M))

15. V 6 ( a5 nilvachietkat. R_ I Spittkoi NI A Marlyncikno. atid U3 G. 1)eseiiki. .Sponiancou;

linlarizalitio and Coercive Fields of Lead Tiianate. Sou. P'Iii.. Sol. Stiare, 12451.,I1203 - 12(M

(1970).

30, A. X. Ktiag. L. S Wut. aid Q. F. Zhou, rbTiO. Thin Film Prepared by Sol- Gel Process, ISAI/994) /Ii'.rr_. X.34 (199101.

17 1'. Yarniakt 11. Waitanabte. 11 Kimiura, 11 K anava, and 11I. Okltitma. Stnuctunal, Fernoelecttic.anid Ps kroelecinic IProperties (if H ighly c-Axis Orieniled Pb, .CajTiO, Thtin Film (;rcimi hiRadtiti-frequtency Magneiron Sptiiieninit. J. tire, Sui. 7iuhp~iuI A. 6(5), 2921-2928 (1988)

38. 1K. Iiti~ -v. Tinunila. R. Takayatira. I. Ueda. Preparation (if, -Axis Oriented PbTiO 3 Thin Filmsand Their Crystallographic. Dielectric. and Pynoelecirie Prroperties. J. App!. Phy~s , 60( II, 361 -367 1191(6).

39( NI ( ) atiia and 1'. Ilanak awl., P reparna (tion and Bastic I'ripertie% oif PbTi(), I erroelecliticI ltim 1 0 mtid 'Ihliir Dcyice Applications., icrrvi,wh rni.s, 63, 243 .252 1191(5)

REFERENCES 393

4( S-( r'nI Yt~ean IIliKik(topt I~t Aril't fIer m Imin I flrio hNAnger Llect ron Spemt r oscop) anid i heir ;iAct i cal P ropmer s, I hint ~Nol Itt dmin 171. 2 S I 62tt

I i9ml

41 K. Sreenmivaand Ni. Sayer, (itaracierizatimnn of lPt$Zr. NiO, I hin I dlins Depositcd fromt MNiii-

clemnien Metal I1 argcls, J ..lp~i Phi.%. 64J.1). 10$4 14911 I98MI

42 1) A. Payne, Integration oft I errodecetrie Nialermms onl Sernicinoduktors by Sol (jet MN~lhods

ISAF M199 Ibw . 12 1 (199011

4311I. Magma, MI Watanabie. 11 Sakmi. r, I Y Krmhomta,T Filmn I altolrrmclettric~of ('71 Systent hN

Sot - GeCl Processing. Is~f 1> 9901 4/r-e X II 99(l)

44ý S K. IDey anid H4 Zisleeg. Sol Gecl [erroelectfic I lnt I-tilni Mefrged viilt (mjAs Pi I I MnimorN

Tecltnoltig). ISA! 1990 Iw ,41o. 12 211 9911

45 G Yi. Z Wit, and MI Saiier. Preparaitont of I'ktt. FiI( ) '11 1it I tmt IIs Sol ( m~lIincstmI:Iecrical, O)ptical. and] JIcIcirut~opltit I'rttpertie J Iprl P111., . 641 ýi 2717 2724 ji'i

46. T. Okamura. N. Adieht, 1. Shiosaki. anti) A Kasthltl, i pitldikl ( ;r,,%&lOt Of I MrOeILIMrtPNZr,,i, q

1 0, 1 ifinis by Reaciiiec Spuitcring of Multi Melat' Iargcf. 1,SA. IY 90t Ihstr , X

11990).

47. A. (roteami. S. NMaistibara. Y kfivanaka, nd u N Shoimdta. I efroclecnrtc PW/i/.- 1t) V~(imT I tilm'

Prepared hy Meilt Targel Spuitterintg, J~rt J ApprI I'/t . 261srnppl 26- 2u 18 21 119s'IX

43. Z Wut and) NM. Sayer. Pritperlies or Sot- Gel Protcessed PLZI (2'54 41ni1 ttttn Iuints. IS -It /99(1

.4/nor. Xi 6 190

49. R J affe, W. R. Coork, Jr., anmd It i J tte, I'we:i trIms ot- (raontm s, Acadcromn In. Iitt . (1)7 1)

VI) T, Nagalottto and 0) Onittin. I'rep.mturminu of Bal if), I tinu hN if Snutltcring midI 1licir(ryslaltoi'grapluc DI Ictcir mu and ()j1 nI1n;W mIOPurt ic. .1 1,01 .1 111 q'l I 'I 2t* Sm rpjt 26,2). 1 1 14

(198 7!.51 MI. Lisposzko. NI Pawetcuyk. N1 Urtitniki, amnd Z Snrossnal. Dif)tlmtc Phtase lran.,tttns ttt

Thick rerriteteciric I1alTi.Sn, .)()3 1itms of l'erovitv'k ne4pe Siruct urC. IltImi Slit I IIt .,. 69.

339 -345 (1990)t

52 A. A. Guel'soii, A Ni tLerer. V S Nlikhnttesknt. V NI Nltkhtrt~io%. aind S V ( tilii. Phmtt.dctProper miv of I 11a Sr)Tim (.) er* roielecl r e Thtin Filmis min Weak E'lect ric I telds sm PII s'n .5 ltdStare, 19j7), 1121 - 1124 (1977)

"APENDIX 51

SPECTROSCOPIC ELLIPSOMETRY STUDIES ON ION BEAM SPU, TER D'.POSITED

Pb(Zr,Ti)0 3 FILMS ON SAPPHIRE AND PT-COATED SILICON SUBSTRATES

S. Trolier-McKinstry, H. Hu, S. B. Krupanidhi,

P. Chindaudom*, K. Vedam, and R. E. Newnham

Materials Research Laboratory

The Pennsylvania State University

University Park, PA 16802

'P Chindaudom is now with NECTECH, Ministry of Science, Bangkok, Thailand

Abstract

Spectroscopic ellipsometry has been utilized to non-destructively depth profile

multi-ion-oeam reactively sputtered lead zirconate titanate films. Some degree of

inhomogeneity (in the form of low density layers or surface roughness) was found in all

of the films examined. The evolution in the structure and microstructure of such films

during post-deposition annealing was investigated with in-situ spectroscopic

ellipsometry. It was found that the onset of microstructural inhomogeneities was

assoc Jted with the crystallization of the perovskite phase, and that the final film

microstructure was dependent on the details of the annealing process.

A model was developed to approximate the effect that local density variations play

in determining the net electrical properties of ferroelectric films. Depending on t,,e

configuration of the embedded porosity, it was demonstrated that microstructural

inhomogeneities can significantly change the net dielectric constant, coercive field, and

remanent polarization of ferroelectric films. It has also been shown that low density

regions near the film/substrate interface can result in apparent size effects in

ferroelectric films.

Introduction

Ferroelectric thin films are potentially uselul as pyroelectric sensurs,

feroelectric memory elemer:s, electrooptic switches, and miniature electromechanical

transducers. For these applications, high quality films with reproducible electrical,

optical, and electromechanical propertiEs are required. Several growth techniques have

emerged in the recent past for the deposition of multi-calion oxide films. In all cases,

the final properties seem to be closely connected to the nature of the deposition process

[1]. For this study, multi-ion-beam reactive sputtering (MIBERS) was chosen for the

growth of optical quality perovskite films. Films for optical applications should have a

uniform smooth surface, should be free from defects or inclusions, and should possess

minimal absorption and low scattering over the wavelengths of interest. With the

MIBERS technique the composition, microstructure, and physical morphology of the

films can be controlled by careful choice of the growth conditions [2,3].

Unfortunately, as is the case for transparent films used in optical coatings, non-

destructive characterization of the film (including any variations in the microstructure

with depth in the sample) has in the past been difficult to perform. To alleviate this

problem, this paper describes spectroscopic ellipsor-try (SE) as a non-destructive,

non-invasive tool for depth profiling ferroelectric thin films on both transparent and

absorbing substrates.

The success of SE for depth-profiling samples was demonstrated in the 1970's

(e. g. [4]). Following the initial demonstration, careful studies were undertaken to

demonstrate that the SE results are consistent with other characterization tools 15-8].

SE has since become widely used for characterizing thin film systems where at least one

component is strongly absorbing. Transfer of the process to either monolithic

transparent samples or transparent films on transparent substrates, however, has been

hindered by inherent difficulties in obtaining accurate SE data for such samples. This

difficulty has recently been overcome via the development of a series of systematic

3

calibrations for a rotating analyzer spectroscopic ellipsometer [9,10). The information

derived from this technique is particularly useful in correlating the role of the

processing procedure in controlling the physical structure and net electrical properties

of ferroelectric thin films.

Film Preparation

Pb(Zr,Ti)03 (PZT) films with a Zr/Ti ratio of about 50/50 were deposited by

multi-ion-beam reactive sputtering (2]. Tlas technique, as established earlier, offers

a highly controllab!e, reproducible deposition process with excellent uniformity in

composition and thickness for large surface area samples. As substrates for the

deposition of oriented films, both (1102) and (0001) sapphire were used (One side

polished, Union Carbide). The interest in (1102) sapphire slems from the fact that this

orientation can be grown epitaxially on silicon, leading to the possibility of integrating a

PZT elec!rooptic component on to a silicon substrate prepared with a sapphire

intermediate layer [111. In addition, platinum-coaled silicon wafers were also utilized

as substrates for some films. These tatter substrates had a thick Si0 2 barrier layer to

prevent reaction between platinum and silicon at elevated temperatures, and a thin Ti

film between the SiO2 and Pt to improve adhesion.

The depositions were carried out on unheated substrates without low energy ion

assistance. II was observed that the substrate temperature rose to about 1000C during

deposition due to intrinsic bombardment by the sputtered species. While the films on

sapphire were deposited with a Pb content very close to the stoichiometric composition

of the perovskite phase, the films on Pt-coated silicon were deposited with 3 - 6%

excess Pb to promote crystallization of the correct phase on annealing [2]. Some films

were reserved for in-situ ellipsometric studies. The remaining films were heated ex-

situ at 10°C/min to 6500C and soaked at this temperature for 2 hrs. This procedure

yields well-crystallized perovskite films. All films deposited on sapphire showed highly

4

oriented X-ray diffraction patterns after annealing. Films on (1102) substrates, as

seen in Fig. 1, were almost exclusively (101) oriented. Similarly, with (0001)

substrates, the films were also (101) orientated, although to a lesser degree. Films on

Pt-coated silicon had a random polycrystalline orientation. Ferroelectricily was

established by measuring hysteresis loops for the films on Pt-coated silicon substrates.

Ellipsometer configuration and calibration

A schematic of the rotating-analyzer spectroscopic ellipsometer used in these

experiments is shown in Figure 2. In addition to standard calibrations to eliminate first

order errors in the optics [121, two additional steps were required to achieve accurate

data for transparent samples. First, the problem of detector non-linearity was

minimized by calibrating the polarization detection system both for the presence of

ambient light and for variations in the gain of ac and dc components of the signal with

changes in the overall signal level [13]. Secondly, inherent difficulties in accurately

measuring the near 00 (or 180°) phase change on reflection from a transparent sample

[14] were overcome through the use of an achromatic compensator.

In contrast to the standard Babinet-Soleil compensator, the three-reflection

achromatic compensator [15] used in these experiments produces approximately a 90'

phase retardation for all wavelengths between 300 and 800 nm. To correct for the

remaining wavelength dependent errors introduced by the compensator, a two-

measurement, "effective source" calibration was utilized. With this technique, the

polarization state of the light emerging from the compensator is first measured at each

wavelength with the ellipsometer detection arm in the straight-through position. Then,

the sample is aligned at the desired angle of incidence and measured over the same

wavelength range to obtain a spectrum containing lumped information on the sample

properties and the source polarization. The change in the light polarization due to

reflection from the sample itself, the quantity of interest, can then be calculated.

5

With these calibrations in place, the ellipsometric parameters A and 41- can be

measured for transparent samples to within 0.030 and 0.010, respectively, over the

spectral range 300 - 800 nm. This corresponds to an accuracy in the real and

imaginary parts of the refractive index of 0.001 [9,13] for a bulk specimen of vitreous

silica [161. As shown in Figure 3, this is an order of magnitude improvement over the

accuracy which can be obtained without these data correction procedures. It should be

noted that the additional procedures utilized constitute a calibration of the polarization

detection system, and as such are sample independent.

Measurements made as a function of temperature were performed in a

windowless electrical resistance furnace. As shown in Figure 4, a kanthal-wire

wrapped alumina tube was used as the heat source. The temperature was monitored by a

thermocouple placed within half an inch of the sample, and controlled by computer. The

outer shell of the chamber consisted of a capped, monolithic brass cylinder machined

with two 1/2" x 1/4" holes for the entrance and exit beams. Both the cylinder and the

baseplate were subsequently electroplated with nickel to minimize oxidation of the

copper and vaporization of the zinc at elevated temperatures. Because the chamber is

windowless, no additional corrections for the ellipsometer optics were required.

Experimental Procedure

Measurements were made on films on sapphire substrates at an incidence angle of

800 using the compensator and the effective source calibration described above. For

films on Pt-coated Si, measurements were performed at an angle of incidence of 700

without the compensator. SE data were collected both on films which had been annealed

at 650°C for 2 hours and on as-deposiled films which were heat-treated in-situ in the

ellipsomeler. For the in-situ experiments, data were taken at different annealing

temperatures in 500C intervals between 25 and 60( JC. Between room temperature and

3500C, data were collected at the annealing temperalure. Above 4000C, however, the

6

imrn was oeated to the desired temperature, soaked for half an hour, and cooled below

3000C for measurement to eliminate errors associated with glow from the furnace. In

all cases, the film was heated to the previous annealing temperature at 5CC/min, and

then raised from there to the new annealing temperature at 20C/min. Cooling was done

at 5°Cimin until the furnace could no longer follow. After this treatment, the same film

was annealed in a conventional furnace at 6500C for 2 hr in order to match the

maximum annealing temperature experienced by the ex-situ annealed samples.

Following data collection, the SE data were modelled in order to analyze the film

thickness, optical properties and the degree of inhomogeneity present within the film.

Modelling of the experimental data was done under the assumption of planar

interfaces between layers with all layers parallel to the surface of the film.

Inhomogeneities in the film density were described by subdividing the film into a series

of layers with different volume fractions of air present in each. Bruggeman effective

medium theory was used to calculate the effective dielectric functions of two phase

mixtures. Variables in the fitting procedure included all of the layer thicknesses, the

volume fraction of air present at any depth in the film, and the dispersion relation

describing the optical properties of the film itself. The equation used to describe the

film refractive index was:2

2 A(]) X(n + i k) = I +

2 2

X- i 2XA(2) (eq.l)

where (n + Ak) is the complex refractive index, X is the wavelength in nm, X0 is the

oscillator position, and A(1) and A(2) are constants. Both X0 and the AO) were

determined during the fitting. To reduce the number of variables, the optical properties

of the film were assumed to be isotropic.

Reference optical property data were used to describe the substrate dielectric

functions. For sapphire, reference data were taken from Malilson [17] and Jeppesen

"7

[18] for the ordinary and extraordinary indices, respectively. Because the literature

values for dn/dT are so small (on the order of 10"5/°C), room temperature data for the

refractive indices of sapphire were used at all temperatures. For the Pt-coated silicon

substrates, ellipsometric spectra were collected for the bare substrate, and were

directly inverted to provide an effective dielectric function for the exposed metal.

Modelling of the data for these substrates showed that the surfaces consisted of roughened

platinum. Unfortunately, the degree of roughness was found to change as the substrates

were heated, so there is considerable residual uncertainty in the effective dielectric

functions for the Pt-coated silicon substrates.

For the modelling, the data sets were truncated to between 400 and 800 nm in

order to eliminate the onset of the absorption edge in PZT. Output from the filling

program included values for the "best-fit" parameters, 90% confidence limits for each

variable, a correlation coefficient matrix describing the interrelatedness between

variables, the unbiased estimator, a [4], of the goodness of fit, and calculated A and T

spectra for the final model. All of these factors were examined to determine the

appropriateness of a given model.

With the exception of the films on (0001) sapphire, all of the fitting was done

assuming that both film and substrate could be treated as isotropic materials. The

routine which handled the propagation of light through anisotropic materials was

written by Parikh and Allara 1191 following the 4 x 4 matrix formalism of Yeh

(20,21]. An anisotropic substrate model was not used for the film on (1102) sapphire

as the angle between the inclined optic axis of the alumina and the plane of incidence of

the light could not be determined accurately.

Analysis of Crystallized MIBERS Films on Sanohire and Pt-coated Silicon

Results from the ellipsometric modelling of the SE data for the films annealed at

6500C for 2 hr are shown in Table 1. All of the films were determined to be between

8

400 and 600 nm thick, which is consistent with profilometry measurements on the

same films. In most cases, a low density interfacial layer between the substrate and the

film was essential to properly match the peak heights in both the A and P1' spectra. In

addition, layer of surface roughness improved the fit for some films. As seen in Figure 5,

the final fits resulted int very good matches to the experimental data.

The high refractive indices obtained for the middle layer of the films suggest that

this region is reasonably dense. Figure 6 shows the modelled refractive indices for

several films on sapphire (determined with A(2) from eq. 1 assigned to zero) in

comparison with reference data for 2/65/35 and 16/40/60 PLZT ceramics [221. As

the refractive index of PLZT ceramics is largely controlled by the Zr/Ti ratio [22], the

film on (0001) sapphire in particular shows excellent agreement with the values which

would be expected for a dense PZT 50/50 material. The lower n values for the films on

1f02) sapphire could be due either to the presence of residual porosity distributed

throughout the densest portion of the film or the assumption of isotropic behavior for

both the film and the substrate. When these films were modelled with reference data

from the film on (0001) sapphire, 5.7% additional residual porosity was required in

each layer of the film to fit the experimental data.

Unless the restriction A(2) = 0 was imposed during the fitting, however, finite

values for the imaginary part of the refractive index in the films were obtained in the SE

modelling. In part, the larger k values may reflect the fact that transparency in lead-

based perovskites decreases as the lanthanum content is reduced. The opacity of dense,

bulk PZT ceramics has been attributed to light scattering at refractive index

discontinuities such as domain walls and grain boundaries consisting of a second phase

[23]. It is likely that the same mechanisms are present in the films. In addition, there

are probably "microstructural" contributions to the effective k which arise from

inhomogeneities in the film not accounted for in the modelling (i.e. scattering from

distributed porosity in the bulk of the film, the presence of a lossy layer associated with

9

either space charge formation or lead loss during firing, or the presence of additional

porosity in the film not properly accounted for in the modelling). It was also found that

a non-zero value of A(2) resulted in an increase in the dispersion found for the film

refractive indices and a somewhat improved fit for the SE data on the films. The reason

for this additional dispersion is unknown. The values reported in Table 1 are for fits

with a non-zero A(2). Because of the uncertainties in the optical properties of the Pt-

coated silicon substrates, it was not possible to determine the exact refractive indices

for PZT films on such substrates.

Figure 7 shows a comparison of the ex-situ annealed PZT films on Pt-coated Si

and (0001) sapphire substrates. Both are densest throughout the bulk of the films, with

low density layers near the substrate and some surface roughness. This type of depth

profile is common in films with a columnar or cluster morphology 1241. It is also

consistent with SEM observations on the same films [2].

For the same annealing conditions, films on Pt-coated silicon possess a much

thicker surface roughness layer than those on sapphire. That could be a result either of

the higher degree of roughness of the sputtered PI substrate or the higher initial PbO

content in the film on platinum. As discussed by Yang et al. [251, the roughness of a film

with a columnar microstructure is dependent on the smoothness of the substrate: with

rougher substrates leading to rougher films. Fox el al. [31 also suggest that

crystallization of the perovskile phase and PbO vaporization are important in

determining the microstructural features of annealed lead lanthanum titanate films.

Evolution of Structure Durin9 In-Situ Annealing

Films deposited at room temperature on (1102) sapphire and Pt-coaled Si

substrates were annealed in-situ in the ellipsometer to determine whether the

inhomogeneities seen in films crystallized conventionally were created during the

deposition, or whether they were generated during the annealing. For these studies,

10

sapphire substrates were preferable as all changes in the spectra with temperature

could be attributed unambiguously to the film. For the films on Pt-coated silicon,

temperature-dependent changes in the effective substrate optical properties obscured

interpretation of the SE data. Nevertheless, the results for the film on Pt-coated silicon

seem consistent with those reported for the sapphire substrates.

Figure 8 shows the ellipsometric spectra collected during the annealing of the

film on sapphire. The data can visually be divided into three regimes. At low

temperatures the interference oscillations are damped strongly at short wavelengths. As

shown in Figure 9, however, this additional damping disappears between 4500 and

5000C. Little change occurs from this temperature until 6000C. which appears to mark

a lriiiiliuoil olworJltIm O :w :,cuild aind lliid lujicnm... I l I lm0lu j, '10- illu-lialud wilh li1u dula

after the 650 0C anneal, is characterized by higher T values (See Fig. 8).

It was not possible to model the full spectra at low temperatures with a single

oscillator, as that did not mimic the abrupt decrease in damping below 500nm. A much

better fit could be achieved by mixing the contribution of the oscillator with reference

data for rf sputtered lead oxide [26]. For all low temperature data, reference data for rf

sputtered PbO provided a better fit than did data for evaporated PbO, largely because the

band gap was shifted 0.6 eV lower in energy. No explanation for the disparity in the

dielectric functions of the two was given by Harris et al. [261, though one possibility

would be the presence of mixed valence states in the sputtered Pb 1271. Fits to as-

deposited films on sapphire substrates were not improved significantly by the addition of

surface roughness or a gradient in the PbO content to the model. Due to the strong

correlation between the void volume fraction and the PbO content, it was not possible to

achieve reproducible fits with low density regions near the film-substrate interlace. N,

evidence for low density regions near the substrate were found in Fox's work on as-

deposited MIBERS lead lanthanum titanale films 131.

11

The volume fraction of PbO remained approximately constant at lower

temperatures, but beginning at 4500C, successively less PbO was required to match the

ellipsometric spectra. Figure 10 shows the best-fit models for 450 and 5000C; the

modelling is improved by allowing the lead "loss" to begin at the film surface and move

progressively through the film thickness. Over the same temperature range, the film

changed from an orange color to pale yellow.

Two possibilities would account for these changes. First, following composition

analysis of as-deposited and annealed films, Fox et al. [3] concluded that as-deposited

MIBERS lead lanthanum titanate thin films contain excess PbO which is removed during

high temperature annealing. Comparable results were obtained on PZT films 121. The

onset temoeraure at which this was projected to occur was 490 ± 500C [3]. While this

temperature is lower than that reported for PbO loss from bulk PbTiO 3 samples [281, it

is in good agreement with the SE experiments. Secondly, the lead oxide could at -450°C

revert to a less lossy species. Thus, the increase in transparency at low wavelengths

could be associated either with a homogenization of the lead oxidation state or with the

incorporation of the lead species in to a more transparent phase (like that of evaporated

PbO, the perovskite, or a pyrochlore phase). It is not possible to distinguish between

these two mechanisms on the basis of the SE data and both may be operative. Similarly,

although it is likely that a pyrochlore phase was formed during these lower temperature

anneals (at least for the film on Pt-coated Si), this could not be unamibiguously

identified on the basis of the SE modelling.

At 550 0C the experimental data were fit well with a two layer model consisting of

a "dense" underlayer with a thin layer of surface roughness. As shown in Figure 11, the

extent of the inhomogeneities becomes more pronounced for anneals above 5500C. In

contrast to the samples annealed only at 6500C for 2 hrs (i.e. those not exposed to

"extended periods at intermediate temperatures), no low-density layer near the

film/substrale interface was required to model the ellipsometric spectra.

12

Between the 500 and 5500C anneals, the film refractive index also increased

markedly (see Figure 12). This is most likely associated with the crystallization of the

perovskite phase. X-ray diffraction patterns following the 600 and 650°C anneals

confirmed that the films on sapphire had converted to the perovskile structure with a

very high degree of <110> orientation. Figure 12 also shows that the refractive index

remains reasonably constant for all fits above 550°C; there is a slight drop at 650CG

(not shown) that is probably associated with the fact that the "dense" bottom layer of the

PZT contains some residual porosity. In summary, Figure 13 shows the proposed

reaction scheme for changes occurring during the annealing of a MIBERS film deposited

at room temperature.

Role of Annealing Profile in Controlling Inhomogeneities in MIBERS Films

In comparing the microstructures of films annealed in different ways, it is clear

that while the inhomogeneity profiles are consistent for samples given identical

annealing schedules, they are strongly dependent on variations in the heating cycle. This

is apparent in Figure 14, which shows the SE-determined depth profiles for films

annealed at 6500C for 2 hrs with and without extended lower temperature soaks. It is

interesting that for samples annealed in-situ in the ellipsometer the major changes in

the film inhomogeneities were coincident with the crystallizalion of the perovskile

phase. This suggests that while the initial microstructure of a vapor-deposited film is

controlled by factors such as substrate temperature, gas pressure, and adatom mobility

[29], the final appearance is also a function of any post-deposition processes involving

diffusion. Thus, in ferroelectric films elimination of excess PbO. reaction with the

substrate, crystallization of either the pyrochlore or perovskite phases, and grain

growth could successively alter the film microstructure. Additional support for this

hypothesis is given by Fox et al. [3], who used scanning electron microscopy to follow

the microstructural evolution of MIBERS lead lanthanum titanate films.

13

It is demonstrated in the foluwing section that inhomogeneilies in the

iicrostructure, especially those associated with low density regions, alter the net

coercive field, dielectric constant and remanent polarization of ferroelectric films.

Given the dependence of the microstructure on the details of the post-deposition

annealing profile observed here, it is clear that two films of the same composition,

crystal structure, and thickness, which were annealed at the same peak temperature,

could nevertheless still possess considerably different electrical properties.

Consequently, post-deposition annealing should be considered an important variable in

both the microstructure and property development of ferroelectric films.

It is also clear that the reasons for the discrepancies between the properties of

the two films in Figure 14 would not be detected by X-ray diffraction. This clearly

establishes the need for microstructure-sensitive characterization techniques in the

study of ferroelectric films for device applications. Both spectroscopic ellipsometry and

electron microscopy should be useful in this regard.

Influence of Inhomoleneities on the Electrical Properties -of Films

Some ferroelectric films grown with the MIBERS technique possess low density

regions near the film/substrate interface which are consistent with the appearance of

either clustering or columns in the annealed film. Such a low density layer should be

expected to alter the net electrical properie.s of the film. Indic•tion.; of this cnn bh .rnn

experimentally in the hysteresis loop of a film which was shown to have an

inhomogeneous depth profile by spectroscopic ellipsometry. The coercive field and net

dielectric constant of the film was 75 kV/cm and -850, respectively. By contrast, a

homogeneous film of the same composition and the same grain size (chemically

prepared) had a lower coercive field (-40 kV/cm) and a higher net dielectric constant.

Both films have higher Ec values than would be expected for bulk ceramics of the same

composition, probably as a result, in part, of the fine grain size of the films and stress

14

exerted on the film by the substrate [30]. Nevertheless, the inhomogeneous vapor-

deposited iilms possess a higher coercive field and a more severe tilt to the "vertical"

sides of the loop than do the homogeneous films. An attempt was made to model the

hysteresis loop of the films to determine wh'lIher these differences could be attributed to

the ellipsometrically characterized inhomogeneities.

To model the effect that embedded porosity of this type would have on the apparent

electrical properties of the film, the following approximation was considered. If a

representation of a columnar microstructure is subdivided into elements vertically,

then most segments ccntain dense PZT in series with a low dielectric constant layer (see

Fig. 15). It is assumed here that the upper electrode is conformal, so that the surface

roughness does not strongly influence the net electrical properties of the film. Given

these conditio:is, the net capacitance of one element can be expressed as1- E 2 (d - x) + F_1x

C E1l 2

where El and E2 denote the dielectric constants of the dense and the defective regions

respectively, d is the total film thickness, and x is the thickný._, of the low density

material. For this geometry,the ratio of the actual coercive field of the dense PZT to the

apparent coercive field of the element isE V1d

ECO) d-x 2

Ec V rc2 (d - x) + Fxd

It is also assumed that there are no forces (such as stress) acting to depo!e the film.

While some depolarization will occur in actual samples, tilting the lines bounding the

top and bottom of the hysteresis loops, that will not alter the basic conclusions of this

argument.

In vapor-deposited materials, the dielectric constant and the thickness of the

defective material would be expected to vary locally as nucleation and growth

commenced, leading to a distribution in the capacitances and coercive fields of the

different subelements. The absence of a well-defined coercive field, in turn, leads to a

loss in the squareness of the hysteresis loop, which is one of the desired attributes for

the model. As a first approximation, the dielectric constant and coercive field for the

dense fine-grained PZT were assigned the values from the sol-gel film, i.e. el = 1300,

and Ec(1) = 40kV/cm [31]. To mimi- the slope of the sides of the hysteresis loop for a

600 nm thick MIBERS film, the coercive fields for the elements were divided into equal

steps between 40 and I 10kV/cm. This gives the right slope and correctly predicts the

average value for Ec (-75kV/cm). The shape of the derived hysteresis loop is also ir,

reasonable agreement with the experimental data.

To further check the validity of this model, the net dielectric constant for a film

600 nm thick with a 60 nm thick low density layer near the substrate was also

calculated using a parallel model to average the capacitance of the individual vertical

elements. The coercive field distribution described above requires dielectric constant

values for the bottom layer to vary between 70 and 1300. This leads to a net dielectric

constant for the film of 770, reascnably close to the experimentally observed value of

-850 [21. The volume fraction of air in the bottom layer of each element needed to cause

the assumed coercive field distribution was also calculated using the logarithmic and

series approximations for the dielectric constant of a composite. These values bracket

the average volume fraction of air derived from the SE determination. This is also

reasonable. Consequently, this model provides a good approximation for the o-served

electrical properties of some types of inhomogeneous ferroelectric films. In all cases

heterogeneities within the film should be expected to increase the coercive field value.

When a distribution of coercive fields is present, the hysteresis loop should be both

significantly broadened and tilted to the right, as is generally observed in reported data

on ferroelectric thin films.

16

Moreover, depending on the distribution of porosity in the film, the coercive

field of some regions could be so high as to make them practically U switchable,

lowering the apparent polarization for the film. Thus it is possible that a film of well.

crystallized ferroelectric could display apparent remanent polarizations significantly

below that expected for the bulk ceramic m3terial. This would occur despite the fact that

if the film could be fully switched, the remanent polarization would be 99.2% of the

bulk value for the case where the bottom 60 nm (of a film 600 nm thick) was, on

average, 92% dense.

Finally, defective regions in the film could serve as pinning sites for domain wall

motion. This, in turn, could influence both aging and fatigue in ferroelectric films.

Consequently, deposition techniques which facilitate production of films which are

highly homogeneous, in addition to being highly crystalline, may be essential in the

preparation of optimized films for device applications.

It is clear from the above discussion that inhomogeneities in the microstructure

can significantly affect the observed low and high field electrical properties of

lerroelectric films. This explains much of the variability reported in the literature for

these properties. As the degree and types of inhomogeneities present in the film are

controlled by the preparation conditions, it is not surprising that films prepared under

different conditions display widely disparate dielectric constants and hysteresis loops.

However, the role of inhomogeneities has often been ignored. For vapor deposited films,

low-energy ion bombardment during growth and rapid thermal annealing have been

shown to be useful in improving the film microstructure [32,331.

Role of Inhomogeneities on Apnarent Size Effects in Ferroelectric Films

Any film which contains low density regions near the substrate should also be

expected to display marked extrinsic size effects. Thus, for the films discussed above,

ana for other vapor-deposited films grown under low adatom mobility conditions, as the

total film thickness is decreased, the defective layer contributes a larger traction to the

overall properties, and the observed properties demonstrate an extrinsic dependence on

film thickness.

Consider, for example, the simplified model for the MIBERS film discussed in the

previous section, where the first 60 nm of perovskile has a lower density than the

remainder of the film. As before, the dielectric constant of the defective region will be

distributed between 70 and 1300, so that the hysteresis loop of the film has the proper

shape. If this initial layer is kept constant while the total film thickness of the film is

varied, then the hysteresis loops become progressively broader and more tilted as d is

decreased (see Fig. 16). At the same time, the net dielectric constant drops off

markedly, even though E for the solid phase is assumed to remain unchanged. Thus, an

infinitely thick film would be essentially undisturbed by the anomalous layer, and would

have a well-defined coercive field and dielectric constant equivalent to the values for a

bulk, fine-grained ceramic. For thinner films, however, both E and Ec diverge from the

bulk values. The decrease in dielectric constant for thinner films shown in Figure 16 is

also in agreement with a variety of experimental studies [31,34].

Figure 16a is drawn with even the thinnest films showing the maximum value

for the remanent polarization. If the low density layer is, on average, 92% dense (i.e.

the ellipsometrically determined value), this is a good approximation, even for the 100

nm thick film, since the effective remanent polarization is calculated to be 95.2% of the

bulk value. As the filmn thickness increases, the maximum Pr value rapidly approaches

100% of the bulk value. However, this marks the limiting value for Pr assuming that

full switching of the film could be achieved. In practice, it is more likely that given the

presence of high local coercive fields in the thinner films, some areas of the film would

become unswitchable, and the measured remanent polarization would be lowered.

The sensitive dependence of the apparent size effects on the initial stages of the

film microstruclure would also explain why films prepared under different conditions

18

demonstrate different properties as a function of thickness. Spectroscopic ellipsometry

offers one means through which depth profiles of inhomogeneilies in thin film samples

can be characterized non-destructively, and so should be useful in resolving many of the

residual questions about structure-microstructure-property relationships in

ferroelectric films.

Conclusions

It has been shown that spectroscopic ellipsometry can be utilized to characterize

the microstructural inhomogeneilties in ferroelectric thin films. Some degree of

inhomogeneity (in the form of low density layers or surface roughness) was found in

each of the films examined; in all cases these were more important in modelling the

ellipsometric spectra than were intrinsic changes in the properties of the ferroelectric

phase as a function of film thickness. The presence of microstructural inhomogeneilies

is not necessarily linked to the existence of a poorly crystallized film. As a result,

well-crystallized, and even well-oriented films can display poor microstructures.

A model was developed to approximate the effect that such local density variations

should have on the net electrical properties of an otherwise perfect film. Depending on

the configuration of the embedded porosity, it was demonstrated that microstructural

inhomogeneilies can profoundly alter the dielectric constant, coercive field, and

remanent polarization of ferroelectric films. Effects of this type are expected to be

especially pronounced in vapor-deposited films with columnar or cluster

microstructures and low density sol-gel films. Therefore, optimization of processing

conditions to produce dense, homogeneous microstructures will be critical in the

development of high quality devices.

Due to the relationship between the film microstructure and the net electrical

properties, any systematic variations in the density with film thickness, like those

associated with columnar growth, will cause changes in the net film properties as a

19

function of thickness. This mechanism is expected to be responsible for the majority of

apparent size effects reported in the literature. Consequently, in examining the

properties of ferroelectric films, and especially in considering the variation in

properties with film thickness, it is imperative that some microstructure-sensitive

technique, such as microscopy or spectroscopic ellipsometry, be utilized.

It has also been demonstrated that spectroscopic ellipsometry can track the

evolution of crystallinity and structural inhomogeneity during annealing of as-deposited

MIBERS films. One of the advantages of performing these studies on transparent

materials is that the entire depth of the film can be sampled (and characterized) at once.

In this work, crystallization of the perovskite phase was shown to be largely complete

after half an hour at 5500C for MIBERS films on sapphire substrates. For the prolonged

heating cycles utilized during in-situ annealing of the ferroelectric films, roughening

of the film surface was coincident with this crystallization, and can probably be

attributed to the growth of crystal nuclei. In addition, lower temperature phenomena

like changes in the lead species present could be identified.

It was found that the final microstructures of MIBERS PZT film- is dependent on

the details of the annealing process. Thus, while the depth profile of film

inhomogeneities was consistent for films given the same annealing schedule, changes in

the annealing resulted in considerable modification of the final density distribution.

Even films annealed at the same peak temperature, but which were exposed to

intermediate temperatures for different lengths of time displayed this type of behavior.

Consequently, while the final film microstructure will be influenced by the as-deposited

state, it will not necessarily be controlled only by the deposition parameters.

This has several important consequences in terms of processing ferroelectric

films for device applications. First, the post-deposition annealing schedule can be as

important as the deposition conditions in controlling the film microstructure, and thus

the film properties. Second, evaluation of annealing schedules should be conducted in

20

light of structural information for the film (i.e. from microscopy or spectroscopic

ellipsometry) in addition to X-ray diffraction information. Third, as the net electrical

properties depend on the inhomogeneities present in the film, some limited property

tuning may be possible given proper control of the annealing process. In particular,

porosity profiles could be tailored to permit control or grading of the film properties

(i.e. strucluial or optical) through the thickness. One area in ceramic materials in

which this type of control might be interesting is in the preparation of functionally

gradient materials or in ceramic membranes.

References

1 S. B. Krupanidhi, J. Vac. Sci. & Technol. A, 10 (1992) 1569 .

2 S. B. Krupanidhi, H. Hu, and V. Kumar, J. Appl. Phys., 71 (1992) 376.

3 G. R. Fox, S. B. Krupanidhi, K. L. More, and L. F. Allard, J. Mat. Res., 7 (1992).

4 D. E. Aspnes and J. B. Theeten, Phys. Rev. Lett., 43 (1979) 1046.

5 K. Vedam, P. J. McMarr, and J. Narayan, App!. Phys. Lett., 47 (1985) 339.

6 P. J. McMarr, K. Vedam, and J. Narayan, J. Appl. Phys., 59 (1986) 694.

7 J. Narayan, S. Y. Kim, K. Vedam, and R. Manukonda, Appl. Phys. Left., 51 (1987)

343.

8 N. V. Nguyen, K. Vedam, and J. Narayan, J. Appl. Phys., 67 (1990) 599.

9 P. Chindaudom, Characterization ot Inhomogeneous Transparent Thin Films on

Transparent Substrates by Spectroscopic Ellipsometry, Ph.D. Thesis, The

Pennsylvania State Univerity (1991).

1 0 P. Chindaudom and K. Vedam, Opt. Lett., 17 (1992) 538.

1 1 S. Krishnakumar, S. C. Esener, C. Fan, V. H. Ozguz, M. A. Tille, C. Cozzolino, and S.

H. Lee, MRS Symposium Proceedings Vo!. 200: Ferroelectric Thin Films E. R.

Myers and A. I. Kingon, eds. MRS: Pittsburgh, PA 1990 91.

1 2 D. E. Aspnes, J. Opt. Soc. Am., 64 (1974) 812.

21

1 3 P. Chindaudom and K. Vedam, AppL Opt. 1993 (submitted).

1 4 J. M. M. de Nijs and A. van Silfhout, J. Opt. Soc. Am., 5 (1988) 773.

1 5 R. J. King and M. J. Downs, Surf. Sci., 16 (1969) 288.

1 6 I. H. Malitson, J. Opt. Soc. Am., 52 (1962) 1377.

1 7 I. H. Maitson, Opt. Soc. Am.,55 (1965) 1205.

1 8 M. A. Jeppesen, J. Opt. Soc. Am., 48 (1958) 629.

1 9 A. N. Parikh and D. L. Allara, J. Chem. Phys., 96 (1992). 927.

20 P. Yeh, J. Opt. Soc. Am., 69 (1979) 742.

21 P. Yeh, Surface Science, 96 (1980) 41.

22 P. D. Thacher, Appl. Opt., 16 (1977) 3210.

23 C. E. Land, P. D. Thacher, and G. H. Haertling, in Applied Solid State Science:

Advances in Materials and Device Research Vol. 4, Ed. by R. Wolfe. Academic Press

Inc., New York, 1974.

24 B. Yang, Spectroscopic Ellipsometry and Computer Simulation Studies of Thin Film

Morphology Evolution, Ph.D. Thesis, The Pennsylvania State University, 1988.

25 B. Yang, B. L. Walden, R. Messier, and W. B. While, SPIE Proc. 821: Modelling of

Optical Thin Films (1987) 68.

26 E. P. Harris, P. S. Hauge, and C. J. Kircher, Appl. Phys. Lett., 34 (1979) 680.

27 A. E. Ennos, J. Opt. Soc. Am., 52 (1962) 261.

28 R. L Holman and R. M. Fulrath, J. AppL. Phys., 44 (1973) 5227.

29 R. Messier, A. P. Girl, and R. A. Roy, J. Vac. Sci. Technol. A2 (1984) 500.

30 G. H. Haertling, J. Vac. Sci. Technol. A, 9(1991) 414.

3 1 K. R. Udayakumar, J. Chen, S. B. Krupanidhi, and L. E. Cross, Proc. 1990 IEEE 7th

Int. Symp. Appl. Ferroelec. (1990) 741.

32 H. Hu and S. B. Krupanidhi Appl. Phys. Lett. 61 (1992) 1246.

33 H. Hu, L. Shi, V. Kumaj, and S.B. Krupanidhi, Ceram. Trans. 25 ed. by A. S. Bhalla

and K. M. Nair, The American Ceramic Society, Westerville, Ohio (1992) 113.

22

3 4 A. A. Gitel'son, A. M. Lerer, V. S. Mikhalevskii, V. M. Mukhortov, and S. V. Orlov,

Sov. Phys. Sol. State, 19 (1977) 1121.

23

List of Fiaures

Fig. 1: X-ray diffraction patterns of the PZT films on (a) (1102) sapphire and

(b) Pt-coated Si. Films were annealed at 650°C for 2 hours.

Fig. 2: Schematic of the rotating analyzer ellipsometer utilized in this work.

Fig. 3: Comparison of uncorrected and corrected SE data for vitreous silica.

Fig. 4: Electrical resistance furnace and sample mount (a) Side view. (b) Top view of

the baseplate and sample holder. The entire baseplate can te translated along y

and rotated about z to permit alignment of the sample at any temperature.

Rotation of the sample about the x axis was performed with the worm gear and

screw shown in the figure. This could also be adjusted at any temperature.

Fig. 5: Fit to the film on (0001) sapphire.

Fig. 6: Comparison between the refractive index of PZT films and PLZT ceramics. (Note:

Thacher has demonstrated that the refractive index of PLZT ceramics is largely

controlled by the Zr/Ti ratio (22]. Consequently, the film refractive indices

should fall between those for the 2/65/35 and 16/40/60 PLZT ceramics).

Fig. 7: Depth profile of the inhomogeneities in films deposited on (a) Pt-coated silicon

and (b) (0001) sapphire.

Fig. 8: SE data for the in-situ annealing of a film on sapphire.

Fig. 9: Annealing of a film showing a decrease in the high energy damping at 500°C. The

lower temperature data was very close to that shown for 450°C.

Fig. 10: Best fit models for intermediate temperature anneals of an as-deposited film on

sapphire. Note that the refractive index of the "a-PZT" changed with annealing

temperature as shown in Fig. 12.

Fig. 11: Evolution of microstructure with temperature during high temperature in-situ

annealing of an as-deposited film on sapphire. Note that the refractive index of

the "PZT" material in the film changed continuously during these anneals as

24

shown in Fig. 12. These changes are probably associated with the crystallization

of the perovskite phase.

Fig. 12: Refractive index of an as-deposited film on sapphire as a function of annealing

tpmperature.

Fig. 13: Reaction scheme for the in-situ annealing of a PZT film on sapphire (Note that

as a pyrochlore phase could not be identified from the SE data alone, it is not

shown on the reaction scheme. It is possible, however, that pyrochlore phase

formation is concurrent with the removal of the lossy lead oxide phase).

Fig. 14: Final microstructures of films on sapphire annealed at 650°C (a) with and (b)

without extended annealing at lower temperatures. The annealing profile shown

for (b) is simplified.

Fig. 15: Geometry used for modelling the effect of inhomogeneities on the electrical

properties of an inhomogeneous ferroelectric film. (a) A two-dimensional

representation of columnar growth subdivided vertically. (b) One element of

the above structure.

Fig. 16: The effects of a 60 nm thick low density layer near the substrate on the

electrical properties of PZT thin films. (a) Variation in the hysteresis loop with

the total film thickness. (b) Variation of the dielectric constant anu the coercive

field with film thickness. Limiting values are c , 1300 and Ec - 40 kV/cm for a

homogeneous fine-grained film on a Pt-coated Si substrate.

25Table 1: Best-fit parameters for the modelling of films annealed at 6501C for 2 hours.

A ir

PZT + Air t f, (Air)1

'7: PZT R.t

:PZT + Air f v (Air)

fv(Air)l tj (nm) t2 (nm) fv(Air) 3 t3 (nm)

PZT on 0.07 ± 0.02 18.3 ± 5.2 478.5 ± 9.8 0.08 ± 0.01 83.9 ± 3.8(0001)Sapphire

PZT on *- -- ---- - - - -- 548.1 ± 6.6 0.08 ± 0.01 62.4 ± 4.7

(1102)Sapphire

PZT on . .. .. - --- - 580.3 ± 2.7 0.12 ± 0.01 53.8 ± 1.3

(1102)Sapphire

PZT on Pt- 0.21 ± 0.03 109.2 ± 2.6 557.5 ± 18.8 - - - - ---..coated Silicon

PZT on Pt- 0.15 ± 0.02 107.7 ± 3.9 522.0 ± 25.9 0.14 ± 0.06 56.3 ± 18.5coated Silicon

Pt

101

- (b)100• 200

111t 112 PtL 102 202

.1-7

(a)

I I I I

20 30 40 50 60 70

2 0 (deg.)

SXe Lami 3 Filter

Polarizer

CompensatorN,,%b

Rotating HeatingRotatng /Chamber

Alignmenttelescope

PM

Uncorrected

-- -------- Corrected-- Reference Data

22.0 1.49

21.8 14- 1.48

21.6 CD.

1.47't (0) 21.4

"2 , - 1.4621.2 C""L.•

2! .01.4521.0

2 .8 1.44

300 350 400 450 500 550 600 650 7C0Wavelength (nm)

Alumina Disk

: Kanihal Wire Wra pped: Alumina Tube

thermocouple

W sample

Clip~' Mirror

Mount

Slot for ElectricalConnections andThermocouple

Shell Seating Slot

Alumina TubeSeating Slot

2 - Model

0 Experimental Data

26--- --

24 -

.J. • ] 22-

20-

16-

400 450 500 550 600 650 700 750Wavelength (nm)

15-

10. - -

5-A (0)

0-

-5 -

400 450 500 550 600 650 700 750Wavelength (nm)

2.9I

Reference data 2/65/35 ceramic

2.8 Reference data 16/40/60 ceramic

. Film on (0001) sapphire

2.7"... Film on (1102) Sapphire

. 2.6-

• . .. .. .... .... ...

2.5-

2.4- "_

400 450 500 550 600 650 700 750Wavelength (nm)

PZT+Ai t=107.7 ±3.9 nmf~ (Air) 0. 15 ±0.02

t 522.0 ±25.9 nm

Ai t56.3±18.5nm

PZT + A'rf (Air) 0. 15 ± 0.06- Pt Coated Si

--- E ~ M----- - - -

PZT + Air

A t18.3 ±5.2 nmf (Air) =0.07 ±0.02

IIIt =478.5 ±9.8 nm

PZT+ irt =83.9±3.8 nmPZT + Airf, (Air) = 0.08 ± 0.01

(0001) Sapphire

I

*4

34-.- 4000C

-• - 450 0C29-

T (0) ~U

24

19 . -. -.-,,

14

400 450 500 550 600 650 700 750 800

Wavelength (nm)

25-

20

S. . .....

15-2

400 450 500 550 600 650 700 750 800Wavelength (nm)

f (PbO) =0.26 ±0.03a-PZT + PbO t =662.7 ±3.8 nm 4000 C

Air-~~~ ~ -Z~~ ~ t33.7±2.6 nm

t- 619.1 ±5.8 nm 4500 Ca-PZTt1+ PbO f v (PbO) =0. 14 ±0.01

Air

-. .......

_______500PZTt =602.6 ±19.4 nm

_ _ a-PZT' +PbO. t=63.0±22.9 nmf (PbO) =0. 12 ± 0.06

Sapphire=

5500 C A ir

iPZT'+ Air t =82.3 ±3.4 nm f, (Air) =0.06 ±0.01

S PZT' t =492.1 ±3.6 nm

--- - - - - - - - - - - - - - - - - - - -

6 00OC Air

PZT + Air t = 80.8 ± 2.3 nm f, (Air) = 0.26 ± 0.01P7-T Air-t =47.7 ±2.8Bnm t, (Air) =0.08 ±0.01

650 0CAir

PZT + Air t =723 ±0.4 nm f, (Air) =0.56± 0.01

PZ -Art =178.6 ±1.9 nm f (Air) =0.15 ±0.01

SPZT + Air t 57.6±1.5 nm fv (Air)=0.07 0.01

PZ ~ t323.7±5.4 nm

3.0

400 C2.8. --- 450 C

".". .... 500 C,......... 550 C

Refractive 600C

Index

". .......... ......... ......... .....2.4 -... .. "................... .

22 -.

2.0 -400 500 600 700 800

Wavelength (nm)

~ o7

a) U)

a)) 1

I- rr-

800

000 10000C 2000C 3000C 40000 50000 600"C

Cl) )~3 ~ JN N NI

>~ >~ - > >

(D -

CD,.... ......

.....

. . ,- -.

-- CC)

II1 <� II • II II • I

-4 Ln0I1+ II 1+ H- IH- C

-

1 o-j 00 9l c. - D

:3 1+3 '+4= 1+333 C a3 o 03

Temperature (°C) Temperature (0C)o o0•C>-

U-,

3 0

-1r3CD

-3Ci

=r

CA-

0

(a)

d(b)

821

Film Thickness

--6oooA

2.5 r -4oooA-jý -2000A

_L4(a)

0~

S-.

a. -2.57

-10 -5 0 5 10Coercive Field *Ec (0)

300 1000

900250

-gE

-10~0 -50) 1

200 7000

Si- 8600 o15

500

1 00 4 400

50 300

0 400 800 1200 1600Film Thickness (nm)