linking grain boundaries and grain growth in ceramics

5
DOI: 10.1002/adem.201000214 Linking Grain Boundaries and Grain Growth in Ceramics By Michael Ba ¨urer * , Heike Sto ¨rmer, Dagmar Gerthsen and Michael J. Hoffmann Perovskite ceramics are widely used in electronic devices as capacitors, [1] positive-temperature-coefficient-resistors (PTCR) [2] or ferroelectric actuators. [3] The performance of such devices are governed by grain size or grain boundary proper- ties. Prominent examples are PTCR ceramics which indicate highly resistive grain boundaries above the Curie temperature and a drop in resistivity of several orders of magnitude below Curie temperature or the grain size dependence of high strain multilayer actuators based on lead zirconate titanate (PZT). [4] All devices are usually produced by sintering of ceramic powders. Therefore, a better understanding of the link between the local structure of the grain boundaries and the grain growth behavior in the ceramic would help to tailor the microstructure for better device performance. Grain boundaries (GB) can generally change their proper- ties with temperature [5,6] as the GBs add additional degrees of freedom to Gibbs phase rule allowing for grain boundary phases which are thermodynamically not stable as isolated phases. The reduction in GB energy can lead to structural changes such as the formation of intergranular films (wetting liquids at GBs) [7] multilayer adsorption of dopants [8] or roughening transitions. [9] Monitoring such changes in poly- crystalline materials is challenging as the selection of the right processing conditions for further examination is difficult. However, grain growth experiments are a convenient way of identifying changes at GBs occurring at different tempera- tures [10] as the process of grain growth itself is insignificant during the heating and cooling phase of a furnace cycle. Samples can then be selected from grain growth experiments for further examination by transmission electron microscopy (TEM). In the present paper we used SrTiO 3 as a model system for grain growth in perovskite ceramics. Grain growth can usually be described by d 2 d 2 0 ¼ kDt¼ 2ag MDt (1) with k as growth factor, d as the average grain size at time t þ Dt, d 0 as the initial grain size at time t, and g as the average COMMUNICATION [*] Dr. M. Ba¨urer, Prof. M. J. Hoffmann Institut fu ¨r Keramik im Maschinenbau, Karlsruher Institut fu ¨r Technologie Kaiserstraße 12, 76131 Karlsruhe, Germany E-mail: [email protected] Dr. H. Sto¨rmer, Prof. D. Gerthsen Laboratorium fu ¨r Elektronenmikroskopie, Karlsruher Institut fu ¨r Technologie Kaiserstraße 12, 76131 Karlsruhe, Germany The macroscopic properties of most materials are strongly influenced by grain size. In ceramic materials the microstructure usually results from the sintering process. Understanding the basic mechanisms of grain growth on an atomic length scale in ceramics would be beneficial to tailor the microstructure for improved macroscopic performance of devices. A method is presented using grain growth experiments to select samples for closer examination of grain boundaries with transmission electron microscopy. The growth experiments are used to identify temperatures were changes at grain boundaries occur at high temperature. Subsequently samples of interest are investigated using transmission electron microscopy (TEM) methods. The correlation between TEM results and changes in grain growth behavior can be used to gain closer insight into the processes occurring during grain growth at an atomic length scale. Strontium titanate is used as model system to demonstrate the combination of growth experiments with TEM results. Normal grain growth shows two distinct drops in growth rate in the temperature range between 1 300 and 1 425 8C, independent of the A-site to B-site stoichiometry of the perovskite. In previous studies a high preference for grain boundary planes oriented parallel to the 100 direction of one of the adjacent grains was found in the high temperature regime. This study shows that the preference does not exist in the low temperature regime possibly explaining the change in grain growth rate. 1230 wileyonlinelibrary.com ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ADVANCED ENGINEERING MATERIALS 2010, 12, No. 12

Upload: michael-baeurer

Post on 06-Jun-2016

228 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Linking Grain Boundaries and Grain Growth in Ceramics

COM

MUNIC

ATIO

N

DOI: 10.1002/adem.201000214

Linking Grain Boundaries and Grain Growth in Ceramics

By Michael Baurer*, Heike Stormer, Dagmar Gerthsen and Michael J. Hoffmann

The macroscopic properties of most materials are strongly influenced by grain size. In ceramicmaterials the microstructure usually results from the sintering process. Understanding the basicmechanisms of grain growth on an atomic length scale in ceramics would be beneficial to tailor themicrostructure for improved macroscopic performance of devices. A method is presented using graingrowth experiments to select samples for closer examination of grain boundaries with transmissionelectron microscopy. The growth experiments are used to identify temperatures were changes at grainboundaries occur at high temperature. Subsequently samples of interest are investigated usingtransmission electron microscopy (TEM) methods. The correlation between TEM results and changesin grain growth behavior can be used to gain closer insight into the processes occurring during graingrowth at an atomic length scale. Strontium titanate is used as model system to demonstrate thecombination of growth experiments with TEM results. Normal grain growth shows two distinct dropsin growth rate in the temperature range between 1 300 and 1 425 8C, independent of the A-site to B-sitestoichiometry of the perovskite. In previous studies a high preference for grain boundary planes orientedparallel to the 100 direction of one of the adjacent grains was found in the high temperature regime.This study shows that the preference does not exist in the low temperature regime possibly explainingthe change in grain growth rate.

[*] Dr. M. Baurer, Prof. M. J. HoffmannInstitut fur Keramik im Maschinenbau, Karlsruher Institut furTechnologieKaiserstraße 12, 76131 Karlsruhe, GermanyE-mail: [email protected]

Dr. H. Stormer, Prof. D. GerthsenLaboratorium fur Elektronenmikroskopie, Karlsruher Institutfur TechnologieKaiserstraße 12, 76131 Karlsruhe, Germany

1230 wileyonlinelibrary.com � 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ADVANCED ENGINEERING MATERIALS 2010, 12, No. 12

Perovskite ceramics are widely used in electronic devices

as capacitors,[1] positive-temperature-coefficient-resistors

(PTCR)[2] or ferroelectric actuators.[3] The performance of such

devices are governed by grain size or grain boundary proper-

ties. Prominent examples are PTCR ceramics which indicate

highly resistive grain boundaries above the Curie temperature

and a drop in resistivity of several orders of magnitude below

Curie temperature or the grain size dependence of high strain

multilayer actuators based on lead zirconate titanate (PZT).[4]

All devices are usually produced by sintering of ceramic

powders. Therefore, a better understanding of the link between

the local structure of the grain boundaries and the grain growth

behavior in the ceramic would help to tailor the microstructure

for better device performance.

Grain boundaries (GB) can generally change their proper-

ties with temperature[5,6] as the GBs add additional degrees of

freedom to Gibbs phase rule allowing for grain boundary

phases which are thermodynamically not stable as isolated

phases. The reduction in GB energy can lead to structural

changes such as the formation of intergranular films (wetting

liquids at GBs)[7] multilayer adsorption of dopants[8] or

roughening transitions.[9] Monitoring such changes in poly-

crystalline materials is challenging as the selection of the right

processing conditions for further examination is difficult.

However, grain growth experiments are a convenient way of

identifying changes at GBs occurring at different tempera-

tures[10] as the process of grain growth itself is insignificant

during the heating and cooling phase of a furnace cycle.

Samples can then be selected from grain growth experiments

for further examination by transmission electron microscopy

(TEM).

In the present paper we used SrTiO3 as a model system for

grain growth in perovskite ceramics. Grain growth can

usually be described by

d2�d20 ¼ kDt¼ 2agMDt (1)

with k as growth factor, d as the average grain size at time

tþDt, d0 as the initial grain size at time t, and g as the average

Page 2: Linking Grain Boundaries and Grain Growth in Ceramics

COM

MUNIC

ATIO

N

M. Baurer et al./Linking Grain Boundaries and Grain . . .

Fig. 1. Arrhenius plot of growth factor as function of the inverse temperature showingtwo distinct drops in grain growth with rising temperature. Adapted from ref.[12].

Fig. 2. Grain with grain boundaries parallel to the 100 plane in SrTiO3. Reproducedwith permission from ref.[14]. 2010, Materials Research Society.

grain boundary energy. M is the effective grain boundary

mobility, the proportionality factor between average GB

velocity during growth and the driving force for growth; a is a

geometrical factor in the order of unity.[11] Intensive grain

growth studies[12] show that the growth factor 2agM is

reduced in SrTiO3 at two distinct temperatures by orders of

magnitude with increasing temperature. This behavior clearly

deviates from the expected Arrhenius type behavior (Figure 1)

and is independent of the Sr/Ti-ratio of the ceramic.

In order to understand this interesting phenomenon, we

analyzed the grain boundaries formed during grain growth in

the low temperature regime (1 300 8C) and compared them

with previously published data of the GB morphology taken

from samples processed at 1 425 8C.[13] In the high temperature

regime a high fraction of the GBs is parallel to a 100-type plane

of one of the adjacent grains. The microstructure is dominated

by these facets, as shown in Fig. 2.[14] For an evaluation of the

influence of the GBmorphology on grain growth the GBs have

been classified as ordered flat (A), disordered flat (B), stepped

(C), or curved (D). The first two classes (AþB)were parallel to

a 100-type plane. Despite of the high fraction of GBs with 100

facets, the growth rate is not dominated by the morphology

distribution. In both, Ti-rich and Sr-rich material approxi-

mately 50% of the GBs are of the types A and B but with a

different growth rate.

Experimental

Ceramic powder with a nominal Sr/Ti-ratio of Sr/

Ti¼ 1.005 was prepared by a conventional mixed oxide route.

TiO2 (Sigma–Aldrich, 99.9þ ) and SrCO3 (Sigma–Aldrich

99.9þ ) were attrition milled with 2mm zirconia milling balls

in isopropanol. The powder slurry was dried in a rotary

evaporizer and subsequently in a vacuum furnace at 80 8C for

24 h. After sieving it has been calcined at 975 8C for 6 h in air to

form the perovskite. In order to break up agglomerates

formed during calcination, the powder was milled again in a

ADVANCED ENGINEERING MATERIALS 2010, 12, No. 12 � 2010 WILEY-VCH Verl

planetary mill, dried and sieved as described above. Green

bodies were prepared by uniaxial pressing of the powder in a

steel die and subsequently cold isostatically pressed at

400MPa. Samples were sintered at 1 300 8C for 10 h in oxygen

and quenched to room temperature after soaking by removing

it from the hot zone of the furnace. Further details on powder

characteristics[15] and sample preparation[12] are published

elsewhere.

TEM plan-view samples were prepared conventionally by

grinding, polishing, and Ar-ion etching. TEM characterization

by means of conventional and high-resolution TEM (CTEM/

HRTEM) imaging as well as selected-area electron diffraction

(SAED) was performed with an aberration-corrected FEI

Titan3 80–300 microscope.

For comparison with the results from the high temperature

sintered ceramics [13], eight grains were aligned with the 100

direction of the crystal parallel to the electron-beam direction

(Figure 3) which gives approximately 50GBs to neighboring

grains with different misorientations and GB-planes. Addi-

tionally four grainswere alignedwith the 110 or 112 directions

parallel to the electron beam (Figure 4).

Results and Discussion

None of the GBs surrounding the grains oriented in 100

direction is aligned parallel to a 100-type (i.e., 010 or 001)

direction (Figure 3 and 5) after sintering at 1 300 8C. Taking the

value of 55% of the GBs being parallel to a 100-type plane in

one of the adjacent grains for ceramics with the same

composition (Sr/Ti¼ 1.005) at 1 425 8C, approximately a

quarter of the GBs should be parallel to this crystallographic

plane if the preference for 100-type planes as GB planes would

be similar at 1 300 and 1425 8C. As pointed out by Saylor

et al.[16] the grain boundary energy can be approximated by

ag GmbH & Co. KGaA, Weinheim http://www.aem-journal.com 1231

Page 3: Linking Grain Boundaries and Grain Growth in Ceramics

COM

MUNIC

ATIO

N

M. Baurer et al./Linking Grain Boundaries and Grain . . .

Fig. 3. Left-hand side: bright-field TEM micrograph of grain aligned with the 100 direction parallel to the electron beam. Right-hand side: high-resolution TEM images of selectedgrain boundaries of the aligned grain with 100-type (010 and 001) directions marked with white lines.

averaging the energy of the free surfaces of the crystals.

Therefore the GB energy is a function of the orientation of the

GB plane relative to both adjacent grains and only indirectly

linked to the misorientation of the grains. The high frequency

of 100 parallel GB planes inmaterial sintered at 1 425 8C[13] had

been explained with the highly anisotropic surface energy of

SrTiO3 crystals, leading to a minimum in GB energy if the GB

plane is parallel to 100 in one of the grains, the plane with the

lowest surface energy.

The temperature dependence of the surface energy g may

be described by

g ¼ E�TS (2)

where E is the surface enthalpy and S is the surface entropy,[17]

with different values of E and S for every orientation. From

this the very pronounced existence of 100 parallel facets at

Fig. 4. Left-hand side: bright-field TEM micrograph with electron beam aligned parallel toimage of GB in the lower part of the image on the left side.

1232 http://www.aem-journal.com � 2010 WILEY-VCH Verlag GmbH & C

1 425 8C and their absence at 1 300 8C seems counter intuitive

as the total magnitude of the surface energies will be lower at

the higher temperature. A possible explanation is, that with

rising temperature local minima in the GB-energy landscape

flatten out and that therefore the global minima comprising a

100 plane of one of the grains are more likely to reduce the GB

energy despite of enlarging the GB area by faceting (Figure 6).

Possible evidence of the existence of facets parallel to planes

other than 100 are steps on GBs as shown in the bottom part of

Figure 5. This fits well to a study on the faceting behavior of aP

5 bicrystal that showed that the faceting behavior of the

GB changes with rising temperature, loosing facets in this

process.[18]

After these energetical arguments for the existence of

different types of grain boundaries a link to the effective

mobilities of the grain boundaries has to be established. It has

the 112 direction in the grain with dark contrast. Right-hand side: high-magnification

o. KGaA, Weinheim ADVANCED ENGINEERING MATERIALS 2010, 12, No. 12

Page 4: Linking Grain Boundaries and Grain Growth in Ceramics

COM

MUNIC

ATIO

N

M. Baurer et al./Linking Grain Boundaries and Grain . . .

Fig. 5. Top right: bright-field TEM image of a grain (with dark contrast) with the 100 direction aligned parallelto electron beam. The grain shows macroscopically curved GBs. Left: magnified images of selected GBs with 100directions marked by white lines. Bottom: GB containing a step.

been shown that the 100 parallel facets in the high temperature

range exhibit a higher mobility than the average grain

boundaries[13] possibly due to the high number of Ti and O

vacancies at such grain boundaries.[19] This is in agreement

with the findings of Sursaeva et al.[20] that a facetted boundary

can move faster than a curved boundary. Despite of this, the

effective mobility is lower at 1 425 8C than at 1 300 8C when

normalized to the thermal energy (Figure 1). From the current

experimental data it is hard to explain this unusual grain

growth behavior. An explanation solely due to the mobility of

Fig. 6. Arbitrary GB energy function for fixed misorientation between neighboring grains and tilt of GB plane aindicates the direction between two triple lines fixing the global direction of the GB. Although the depth of all minminimum at T2. The change in faceting changes the length of the GB. Note that the energy function is not relat

ADVANCED ENGINEERING MATERIALS 2010, 12, No. 12 � 2010 WILEY-VCH Verlag GmbH & Co. KGaA

single facets is not possible because the

behavior of the facets is closely linked to

the surrounding GBs and themobility itself is

hard to access experimentally. A good

possibility to understand the dominant

mechanism is to use grain growth simula-

tions that allow a systematic and indepen-

dent variation of the GB-energy and mobi-

lity-landscape.[21]

Another important point when analyzing

grain boundaries is the possibility that the

identified effects are superposed or even

caused by impurities present in the material.

In the case of perovskites a convenient way

of doing this is to vary the A to B site

stoichiometry of the ceramic, as most impu-

rities have an affinity to occupy either the A-

or B-site of the crystal structure. Experimen-

tally observed effects that occur in both, A-

and B-site rich materials are not related to

impurities. This can be illustrated by the

example of Si4þ in strontium titanate. In the

absence of TiO2 excess and in the case of

SrO-excess Si4þ is incorporated on the Ti-site

as it is close in ionic radius and has the same

charge. By addition of TiO2 it can be expelled

from the lattice to form intergranular films at

GBs[22,23]. In other materials without possi-

bility of changing the cation ratio, the

detection of impurities and their impact

might be much more difficult.

In the case of chemical changes at grain boundaries such as

formation of liquid films, careful control of cooling rate is

necessary as the films can recede during cooling. Again this

possibility is general for all types of materials, but can also be

illustrated in strontium titanate. Waser et al.[24] stated that ‘‘A

TEM investigation of the ceramics did not reveal any

enrichment of TiO2 at the boundary between perovskite

grains’’ and that ‘‘this indicates that the wetting of the grains

recedes during cooling and the TiO2 accumulates in triple

points’’. For a quenched material Stenton and Harmer stated

round an arbitrary direction u for two temperatures T. Red lineima decreases, faceting changes with a preference for the globaled to the crystallography and GB energy of SrTiO3.

, Weinheim http://www.aem-journal.com 1233

Page 5: Linking Grain Boundaries and Grain Growth in Ceramics

COM

MUNIC

ATIO

N

M. Baurer et al./Linking Grain Boundaries and Grain . . .

that a crystalline TiO2 phase was present between the grains

after quenching from above the eutectic temperature of

1 440 8C in TiO2 rich material.[25] In the present study,

although the sample was quenched from sintering tempera-

ture, the cooling rate is not as important as the structural

features of the GBs such as the general GB orientation changes

in the timescale of grain growth which is much longer

compared to the overall cooling time.

Conclusions

Grain growth experiments are a convenient way of

selecting samples for further examination in TEM to gain a

better understanding of the influence of structural features of

the grain boundaries on grain growth. In the case of strontium

titanate it has be shown that the faceting behavior of the grain

boundaries is highly different at 1 300 8C compared to 1 425 8C.The high frequency of GB-planes parallel to 100 in one of the

adjacent grains at high temperature does not persist at low

temperature, giving a possible explanation for the significant

reduction in grain growth with rising temperature in SrTiO3.

Received: July 16, 2010

Final Version: August 27, 2010

Published online: November 10, 2010

[1] M. Fujimoto,W. D. Kingery, J. Am. Ceram. Soc. 1985, 8, 169.

[2] J.-H. Jeon, J. Eur. Ceram. Soc. 2004, 24, 1045.

[3] M. J. Hoffmann, H. Kungl, Curr. Opin. Solid State Mater.

Sci. 2004, 8, 51.

[4] H. Kungl, M. J. Hoffmann, J. Phys. 2010, 107, 54111.

[5] J. Luo, Curr. Opin. Solid State Mater. Sci. 2008, 12, 81.

[6] M. P. Harmer, J. Am. Ceram. Soc. 2010, 93, 301.

[7] H. Luo, J. Wang, Y.-M. Chiang, J. Am. Ceram. Soc. 1999,

82, 916.

[8] S. J. Dillon,M. P. Harmer, J. Am. Ceram. Soc. 2007, 90, 996.

1234 http://www.aem-journal.com � 2010 WILEY-VCH Verlag GmbH & C

[9] S.-Y. Chung, D. Y. Yoon, S.-J. L. Kang, Acta Mater. 2002,

50, 3361.

[10] S. J. Dillon, M. Tang, W. C. Carter, M. P. Harmer, Acta

Mater. 2007, 55, 6208.

[11] J. E. Burke, D. Turnbull, Prog. Met. Phys. 1952, 3, 220.

[12] M. Baurer, D.Weygand, P. Gumbsch, M. Hoffmann, Scr.

Mater. 2009, 61, 584.

[13] M. Baurer, S.-J. Shih, C. Bishop, M. Harmer,

D. Cockayne, M. Hoffmann, Acta Mater. 2010, 58,

290.

[14] S. J. Shih, S. Lozano-Perez, D. J. H. Cockayne, J. Mater.

Res. 2010, 25, 260.

[15] M. Baurer, H. Kungl, M. J. Hoffmann, J. Am. Ceram. Soc.

2009, 92, 601.

[16] D. M. Saylor, B. E. Dasher, T. Sano, G. S. Rohrer, J. Am.

Ceram. Soc. 2004, 87, 670.

[17] M. McLean, H. Mykura, Surface Sci. 1966, 5, 466.

[18] S. B. Lee, W. Sigle, W. Kurtz, M. Ruhle, Acta Mater. 2003,

51, 975.

[19] S. von Alfthan, N. A. Benedek, L. Chen, A. Chua,

D. Cockayne, K. J. Dudeck, C. Elsasser, M. W. Finnis,

C. T. Koch, B. Rahmati, M. Ruhle, S.-J. Shih, A. P. Sutton,

Ann. Rev. Mater. Res. 2010, 40, 557.

[20] V. G. Sursaeva, B. B. Straumal, A. S. Gornakova,

L. S. Shvindlerman, G. Gottstein, Acta Mater. 2008,

56, 2728.

[21] M. Syha, D. Weygand, Modell. Simul. Mater. Sci. Eng.

2010, 18, 015010.

[22] C.-J. Peng, Y.-M. Chiang, in Ceramic Microstructures ’86:

Role of Interfaces Proceedings of the 22. University Con-

ference on Ceramics, and the International Materials

Symposium, Plenum Press, New York 1987, p. 555.

[23] C.-J. Peng, Y.-M. Chiang, J. Mater. Res. 1990, 5, 1237.

[24] R.Waser, T. Baiatu, K.-H. Hardtl, J. Am. Ceram. Soc. 1990,

73, 1645.

[25] N. Stenton, M. P. Harmer, Adv Ceram.: Addit. Interfaces

Electron. Ceram. 1984, 7, 156.

o. KGaA, Weinheim ADVANCED ENGINEERING MATERIALS 2010, 12, No. 12