Changes in macroscopic behaviour through segregation in niobium doped strontium titanate

Changes in macroscopic behaviour through segregation in

Niobium doped Strontium Titanate

M Bäurer1, L F Zagonel2, N Barrett2 and M J Hoffmann1

1Institut für Keramik im Maschinenbau (IKM), Universität Karlsruhe (TH) , Haid-

und-Neu-Straße 7, 76131 Karlsruhe, Germany 2Commissariat à l’Energie Atomique, CEA-DSM/DRECAM-SPCSI,CEA-Saclay,

91191 Gif-sur-Yvette, France

E-mail: [email protected]

Abstract. The electrical properties of electroceramics are largely governed by their complex

microstructure which must be precisely controlled for applications (as, for example, sensors,

actuators and capacitors). In this study, the dependence of niobium segregation in strontium

titanate on different sintering parameters is shown. 1.2 mol% Nb doped SrTiO3 was sintered at

1420°C in air for different times and with different post sintering cooling rates. Scanning

electron microscopy images provided evidence of morphological differences among the

prepared samples. X-ray diffraction revealed a decrease in lattice parameter due to niobium

leaving titanium substitutional sites and going to grain boundary zones. X-ray photoemission

spectroscopy has shown substitutional or segregated niobium has a 5+ chemical state.

Impedance spectra analysis permitted estimation of changes in grain boundary thickness in

agreement with other results. The analysis establishes that niobium segregation happens during

cooling and is diffusion controlled, depending on grain size.

1. Introduction

Perovskite ceramics are widely used for passive components in electronic industries. Despite of the

small volume fraction of grain boundaries in the material, these internal interfaces have a huge

influence on the macroscopic properties of the ceramic. The segregation of acceptors and donors in

SrTiO3 and BaTiO3 has been described by Desu and Payne [1]. In their study it is shown that dopants

which are in solution at high temperatures segregate to grain boundaries during slow cooling.

Fang and Gu [2] showed that niobium segregates to grain boundaries in air sintered samples,

Chung and Kang [3] on contrary showed that there is no enrichment of the dopant at the grain

boundaries. In these two studies the influence of the cooling rate is not further investigated. In the

latter work the sample material was cooled to room temperature in a few minutes, in the former the

samples were furnace cooled over a time range of several hours. The comparison of these results fits

well to the theory of segregation presented by Desu and Payne [1]. But as the preparation conditions

and raw materials of these two studies are different from each other it is not clear if the results are

comparable. Chiang and Takagi [4] showed that segregation of Nb is detectable only for high Nb

concentrations with STEM even in quenched samples. In this study niobium doped SrTiO3 has been

chosen to examine whether there is segregation on slow cooling from a sintering temperature slightly

Interfacial Nanostructures in Ceramics: a Multiscale Approach IOP PublishingJournal of Physics: Conference Series 94 (2008) 012015 doi:10.1088/1742-6596/94/1/012015

c© 2008 IOP Publishing Ltd 1

lower than the eutectic temperature of 1440°C. An additional effect at this temperature is the

occurrence of abnormal grain growth.

2. Experimental Procedure

Samples were prepared from high purity strontium carbonate (99.9+%, Sigma Aldrich), titania

(99.9+%, Sigma Aldrich) and niobium oxide (99.9%, ChemPur) powders. The components were

weighted in according to

0.994 SrCO3 + 0.006 Nb2O5 + 0.988 TiO2 (1)

under the assumption that the Nb ions occupy B-sites of the perovskite lattice and that their

additional charge is completely compensated by doubly charged strontium vacancies according to

Sr0.994Ti0.988Nb0.012O3 (2)

The powder mixture was milled in an attrition mill with zirconia milling balls. After milling it was

calcined for 6h at 975°C to form a single phase perovskite, which was confirmed by x-ray powder

diffraction within the measuring accuracy. The calcined powder was again milled in a planetary mill to

break up agglomerates formed during the heat treatment. The impurity levels have been controlled by

ICP-OES (Inductively Coupled Plasma Optical Emission Spectrometry) after both milling steps. The

elemental analysis after the second milling is given in table 1. The powder was uniaxially pressed into

discs of 15mm diameter and 5mm thickness in a steel die and subsequently cold isostatically pressed

at 400MPa. The discs were heated with 20K/min in air to a temperature of 1420°C then kept at

temperature for 1 h and 20 h, respectively. After sintering the samples were either cooled with

10K/min to room temperature or quenched with more than 200K/min by removing it from the hot zone

in the furnace.

Table 1. Impurity levels after milling procedure

Trace element Impurity level [µg/g] Trace element Impurity level [µg/g]

Al, Ni, Mn < 5 Na 50

Mg, Fe, Cr < 10 Ba 60

Y 10 Si < 150

Cu < 15 Ca, Zr 200

The sintered material, having the shape of a rod with approximately 13mm in diameter and 4mm in

thickness, was ground down in diameter to 10.5mm and then cut into discs of 1mm thickness. By this

only the centre of the sample was analysed so that there are no effects from the sintering surface.

Impedance spectroscopy was performed with a Hewlett Packard LCR-meter (HP 4284A) in the

frequency range from 20Hz to 1MHz with sputtered gold electrodes and an applied electrical field of

0.1V/mm. The spectra were then interpreted by using a brick layer model consisting of the RC

elements in a row fitted with a least squares approximation. Additionally, a 2-point DC resistance

measurement was performed on the samples to get an estimate for a nearly 0 Hz value.

Lattice parameter measurements by X-ray diffraction (Siemens D500) were made on polished

samples using monochromatic Cu K! radiation. The samples have been previously annealed for 5h at

500°C in order to reduce stresses introduced during polishing. A doublet pseudo-Voigt function, for

Cu-K!1 and K!2 radiation, was fitted to each peak in the range from 2" = 12- 75°. A least means

squares method was then used to analyse the pattern and calculate the lattice parameter according to

Cohen [5]. The method has been used by Desu and Payne to show dopant segregation in BaTiO3. As

the ionic radius of niobium is larger than that of titanium by approximately 6% [6], the amount of

niobium in solution should correlate with the lattice parameter.

X-ray photoemission spectroscopy (XPS) measurements were performed at UHV (P <10-7 Pa)

using monochromatic Al K! radiation (h# = 1486.7 eV). Photoelectrons were detected at normal

emission (unless otherwise stated) using a hemispherical analyzer (radius 125 mm) with an angular


2

acceptance of ±8° and a pass energy of 25 eV. The overall energy resolution, including the

monochromator, was 0.6 eV. Binding energies are measured with respect to the Fermi level as

measured on a gold foil in electrical contact with the sample. The sample surfaces have been cleaned

in situ by heating for 90 minutes at 600°C. After this time, the carbon contamination is reduced to

negligible amounts. Also, as observed on a single-crystal reference, low energy electron diffraction

(LEED) confirms that the resulting surface is well ordered and therefore stoichiometric and apparently

without significant reconstruction. This procedure shows advantages with respect to (0.5 to 1 keV)

Argon ion sputtering which eroded oxygen preferentially (results not shown) as already observed in

literature and revealed by ion collision simulations [7].

3. Results

3.1. Microstructure

The microstructural evolution during the difference of 19h in sintering time in between the samples is

not continuous. As shown in figure 1, the initial and the final microstructure follow a normal grain size

distribution whereas at an intermediate time of 4h an abnormal distribution with some grains clearly

growing much faster than the others is visible. After 20h only abnormally grown grains are present.

Between these grains a small number of pockets (figure 1(d)) with a titanium rich phase are detectable.

The appearance of these pockets is a clear indication that the original assumption of strontium vacancy

compensation for the additional charge of the donors is not true and that there is an effective B-site

excess.

As a pure size effect is not enough to let a grain grow abnormally [8], a change in grain boundary

mobility for some boundaries is needed to accelerate grain growth. For that reason it is likely that the

boundaries of the impinged, formally abnormal, grains in the final microstructure have structurally

different grain boundaries. The grain size of the samples has been measured by a linear intercept

method. The grain size after 1h sintering time is 2.1µm after 20h it is 46 µm.

Table 2. Sample preparation parameters and measured physical properties

Sample

Sintering

Time (h) Cooling method

Grain

size

(µm)

lattice parameter

(±0.00002 nm)

thickness/boundary

(nm)

Nb 3d5/2 BE

(±0.5 eV)

20 quenched 46 0.39050 9.23 207.5

20 furnace 46 0.39050 9.20 207.4

1 quenched 2.1 0.39049 0.57 207.5

1 furnace 2.1 0.39043 1.24 207.5

Single crystal 104 - - 207.5

3.2. X-ray

The lattice parameters derived from X-ray diffraction for the four sample types are shown in table 2.

The samples sintered for 20h, containing large grains do not show a significant difference in lattice

parameters between quenched and furnace cooled whereas it differs between the two cooling modes

for the samples with small grains, sintered for 1h. This difference is evidence for segregation of the

dopants towards the grain boundaries: a smaller lattice parameter in the furnace cooled sample

indicates that it contains less substitutional niobium in the bulk ceramic compared to the quenched

one. The samples sintered for 20h do not show this difference in lattice parameter because the total

grain boundary area is much smaller and the diffusion paths towards the boundaries are

correspondingly longer because of the large grain size. However, from these results it is not clear

whether the internal sink is a boundary or a triple pocket.


3

3.3. Electrical properties

For the bulk material an increase of the DC conductivity with the temperature is expected [9] as the

number of charge carriers is not constant. With the quench conserving a state closer to the high

temperature state there is the possibility that the bulk properties differ considerably. The measured

spectra and the curve calculated from the fitted values are shown in figure 2. From the fitted values of

the capacitances for the grain boundaries an equivalent thickness of the grain boundary region has

been calculated according to

C=$0$rA/d (3)

assuming that the dielectric constant $0 of the boundary region is equal to the bulk value [10]. The

results of this equivalent thickness calculation are also shown in table 2, normalized to grain size.

There is a clear thickening of the boundary during furnace cooling for the 1h sample. The 20h samples

do not show a significant difference in the calculated thickness. Comparing the 1h and the 20h

samples there is a large difference in grain boundary thickness which could be related to abnormal

grain growth e.g. a pile up process or microstructural differences allowing or generating grain

boundary thickening.

3.4. XPS

The Sr 3d, Ti 2p and O1s core electron level spectra display no appreciable energy shift for the

different preparation treatments and are close to the values of a single-crystal SrTiO3: 133.6 eV, 459.1

eV and 530.2 eV, respectively, in agreement with the literature [11]. The Nb 3d core levels, see Figure

3, on the other hand, depend slightly on the preparation procedure and have binding energies close to

that of the single-crystal reference [12]. Table 2 displays the center of the Nb 3d5/2 core level peaks for

studied samples. The binding energies observed for both sintering times and both cooling rates

(a)

(b)

(c)

(d)

Figure 1. Microstructure of the samples (a) 1h sintering time, (b) 20h, (c) intermediate stage after 4h

and Ti-rich pocket after 20h.


4

correspond to a +5 oxidation state, 207.5 eV [13]. Therefore, the niobium segregating at grain

boundaries forms Nb2O5 under the sintering atmosphere containing oxygen, the same oxidation state

as in Ti (substitutional) sites, where it acts as a donor. Niobium sub-oxides like NbO2, at BE equal to

~205.5, were not observed.

(a)

0 5000 10000 15000 20000

0

5000

10000

-Im

Z/!

Re Z/!

(b)

0 5000 10000 15000 20000

0

5000

10000

-Im

Z/!

Re Z/!

204 206 208 210 212

Nb 3d1/2

No

rma

lize

d I

nte

nsity (

a.u

.)

Binding Energy (eV)

Single Crystal

1h - FC

1h - Q

20h - FC

20h - Q

Nb 3d3/2

Figure 2. Impedance spectra for the samples (a)

1h quenched (x) and furnace cooled (+) and (b)

20h quenched (x) and furnace cooled (+). The

line represents the calculated values from the fit

procedure.

Figure 3. Niobium 3d core levels for studied

samples. The binding energies correspond to a

5+ oxidation state for the single crystal and for

the sintered poly-crystals in which some

segregation should be present.

4. Conclusions

Scanning electron microscopy, impedance spectroscopy and X-ray diffraction suggest that there is Nb

segregation towards the grain boundary region in SrTiO3 during slow cooling in the furnace whereas

more dopant is in solution at high temperatures which can be partly preserved by quenching. XPS

results confirm that niobium is occupying the lattice in a +5 oxidation state. The segregation observed

for shorter sintering times and low cooling rates preserves this oxidation state, indicating that Nb2O5 is

formed at grain boundaries.

Acknowledgements

We would like to acknowledge financial support by the European Commission under contract Nr.

NMP3-CT-2005-013862 (INCEMS). Additionally we are very grateful for the ICP measurements

conducted at the Max-Planck-Institute, Stuttgart.

References [1] Desu S B and Payne D A 1990 J. Am. Ceram. Soc. 73 [11] 3391-3421

[2] Fang P and Gu H 2003 Key Eng. Mat. 247 323-326

[3] Chung S-Y, Kang S-J L 2002 J.Am.Cer.Soc. 85 [11] 2805-2810

[4] Chiang Y-M and Takagi T 1990 J.Am.Cer.Soc. 73 [11] 3278-3285

[5] Cohen M U 1935 Rev. Sci. Instrum. 6 [3] 68-74 and erratum

[6] Shannon R D and Prewitt C T 1969 Acta Cryst. B 25 925-946

[7] Adachi Y, Kohiki S, Wagatsuma K and Oku Masaoki 1999 Appl Surf. Sci. 143 272-276


5

[8] Srolovitz D J, Grest G S and Anderson M P 1985 Acta Metallurgica 33 2233-2247

[9] Moos R and Härdtl K H 1997 J. Am. Ceram. Soc. 80 [10] 2549-2562

[10] Denk I 1995 Ph.D. thesis MPI-FKF Stuttgart

Mao Z and Knowles K M 1997 Brit. Ceram. Trans 96 50-56

[11] van der Heide P A W, Jiang Q D, Kim Y S and Rabalais J W 2001 Surf. Sci. 473 59-70

[12] Shibagaki S and Fukushima K 1999 J. of the European Ceramic Soc. 19 1423-1426

[13] Miller C F, Simmons G W and Wei R P 2000 Scripta mater. 42 227-232

10µm


6

Changes in macroscopic behaviour through segregation in niobium doped strontium titanate

Documents

Transcript of Changes in macroscopic behaviour through segregation in niobium doped strontium titanate