The effect of various dopants on the dielectric properties of barium strontium titanate

The effect of various dopants on the dielectric properties of barium strontium titanate

Materials Letters Node-Holland I5 ( 1993) 317-324 The effect of various dopants on the dielectric properties of barium strontium titanate S.B. Hemer...

730KB Sizes 0 Downloads 36 Views

Materials Letters Node-Holland

I5 ( 1993) 317-324

The effect of various dopants on the dielectric properties of barium strontium titanate S.B. Hemer,

F.A. Selmi, V.V. Varadan and V.K. Varadan

The Center&r the Engineering qf Electronic and Acoustic Materials. Department o~Engi~eerjng Science and Mechanics, The Penns~~lvaniaState University, ~~n~ve~si1.v Park, PA 16802, IK4 Received

16October 1992

Various acceptor and donor dopants were incorporated into B~.60Sr0.40MxTi,_,0j and their effect on dielectric properties for the material’s use as a phase shifting device was studied. The best combination of high E, low tan 5, and largest At with applied bias AK was found in the Fe-doped material. The addition of I mol% Ba or Sr to the Fe-doped material was seen to further improve the properties. Other dopants studied included Mn, Bi, Ga, Y, and Nb.

1. Introduction There has been recent interest in the use of perovskite ceramics as phase shifting devices in antenna and radar [ I-31. Desirable material properties include low-temperature Curie peak ( r,), high dielectric constant (e), low loss tangent (tan&, and large de with applied bias field (AL’), which corresponds to good tunability. Barium strontium titanate ceramics meet all of these properties. Barium titanate and strontium titanate can mix intimately with no apparent phase segregation [ 4,5]. It is known that the To of barium strontium titanate can be changed by adjusting the Ba:Sr ratio [6,7]. For phase shifting ceramics, it is desirable to operate the device in the paraelectric region, hence the desire for low T,. However, in general, the higher the dielectric constant, the more tunable the ceramic will be. By adding 40 mol% SrTi03 to BaTiOj, the Curie peak can be shifted from 120 to - 5°C but a relatively high dielectric constant can be maintained above Tc. Ba0.60Sr0.40Ti03 was chosen as the basic composition because of its good combination of low jrc, high dielectric constant, relatively low loss tangent, and good tunability. Heywang ]S ] developed a model for the positive temperature coe~cient effect above ‘f, that is widely accepted today. Heywang assumed the presence of a two-dimensional layer of surface acceptor states, 0167-577x/93/$

06.00 0 I993 Elsevier Science Publishers

either oxygen or acceptor ions, at the grain boundaries of BaTiO,. The acceptors act as traps, taking electrons from the interiors of the grains. This results in a potential barrier at the grain boundary. As the temperature rises above rc and permittivity declines, the trapped electrons gain enough energy to jump to the conduction band, increasing conductivity with temperature. In this study, stoichiometry was adjusted to favor dopant substitution on B sites by mixing the powders in proportion to give Bao.6,,Sr0.40MxTi r _,vO,, giving an A:B ratio of one, assuming all of the dopant occupies B sites. For capacitor ceramics of perovskite structure (AB03), doping in small amounts with acceptor ions on B sites can greatly affect dielectric properties [ 9- 131, Acceptor dopants are defined as ions with a lower valency than the ions they replace (e.g. Fe3+ for Ti4+), while donor dopants are ions with a higher valency (e.g. Nb5+ for Ti4+ ). Chan et al. [ 9 ] postulated that acceptor impurities are mostly compensated for by oxygen vacancies. We believe Fe, Mn, Ga, and Y act as acceptor dopants (M2+ or M3’ on Ti4+ sites) while Nb and Bi are donor dopants ( Nb5+ on Ti4+, Bi3+ on (Ba, Sr)2+ ). This would give oxygen vacancies in the manner: Mz03 +2(BaO,

SrO)

-+Z(Ba,Sr)+2M;,+50,+V,,

B.V. All rights reserved.

317

Volume 15, number

5,6

MATERIALS

assuming the ion was 3 +. The final electroneutrality is accounted for in the stoichiometry as: (Ba, Sr) (M;i)XTi,-X(V,

)X~LLX~~ .

Following the example of Chiang and Takagi [ 14 1, the ionic radii of Fe, Mn, Ga, Bi, Y, Nb, Ba, Sr, and Ti are listed in table 1 [ 15 1. We feel it is more reasonable to assume substitution of these ions on Ti4+ sites, with the possible exception of Y3+, and the probable exception of Bi3+, which may prefer to occupy (Ba, Sr)‘+ sites. However, our study did not allow a direct determination of A or B site occupation by these ions as donors or acceptors. The topic of dopants acting as donors or acceptors on A or B sites respectively, is of much debate, with various investigators concluding 2 + or 3 + ions occupy A sites as donors compensated by A site vacancies or Ti3+ ions, while others conclude they occupy B sites as acceptors [9-13, 15-261. A Ba:Ti> 1 in BaTiO, has been demonstrated by Hyatt et al. [ 271 to reduce the loss tangent particularly in the paraelectric region above the Tc.In our work, we have achieved better properties when (Ba, Sr) :Ti> 1, avoiding excess TiOz phase, which encourages rapid grain growth, as several other investigators have found [ lo- 131. All of the doped materials were compared to undoped material. While phase-shifting ceramics are essentially capacitors and are tested as such, they are used at genTable 1 Ionic radii (in A)

[IS] 6-fold coordinate

12-fold coordinate

A site ions Ba2+ srZ+ Bis+ 8) y,+ a)

1.36 1.16 I .02 0.89

1.44

B site ions Ti4+ Mn3+ Mn2+ Ga’+ Fe’+ YJ+ NbS+

0.61 0.54 0.67 0.62 0.49 0.89 0.64

1.60 _

a) No I2-fold coordinate data are available for these ions, so their 6-fold coordinate data are listed for comparison.

318

LETTERS

January

1993

erally higher frequencies than capacitors. Also, a low loss tangent and the ability to achieve a large Ae with applied bias voltage become important properties in phase shifting devices.

2. Sample preparation BaCO,, SrC03, TiOz, and dopant powders of high purity were mixed in the appropriate amounts for each composition. Each batch was wet ball milled in absolute ethanol with zirconia media for 6 h to thoroughly mix the powders and then dried in a ventilated oven at 80°C. The mixed powder was calcined at 800°C for 5 h and at 1150°C for 10 h. The calcined powder was wet ball milled in absolute ethanol with zirconia media for 24 h. The powder was dried and weighed. Polyvinyl alcohol (PVA) was then added as a binding agent in the amount of 2 wt% and mixed in distilled water with the powder for 2 h in the ball mill with zirconia media. The powder with binder was then dried and seived through a 230 mesh seive (63 urn opening), pressed under 600 MPa of pressure into discs 12.70 mm in diameter and z 3 mm in thickness. The discs were sintered in air at 1450°C for 1 h, after holding at 500°C for 1 h to allow for binder burnout. A heating rate of 180”C/h was used. Density measurements were taken on the sintered samples via Archimedes method. The samples were wet sanded to produce flat sides for electroding. Samples used for testing Ac with applied AV were sanded down to z 0.5 mm thickness to obtain a large current density. The samples were dried overnight in an oven at 80°C to remove water picked up during wet sanding. The samples were measured and electroded with a high temperature silver paint. The painted samples were tired at 350°C for 20 min and then 600°C and air-cooled. Wires were soldered to either side of the disc. The disc was coated in silicone rubber to prevent any water absorption during dielectric property measurements. Capacitance and loss tangent measurements were taken on a Hewlett-Packard 4192A impedance analyzer in the temperature region - 50 to 65 “C. Voltage was applied to the sample by a Kepco ABC 1500 direct current voltage supply. A blocking circuit designed to prevent large voltages from entering the impedance

Volume 15,number 5,6

January 1993

MATERIALS LETTERS

output Impedance the Supply

of

Voltage Supply

\

Sample

\ \ Power Rectifiers

Fig. 1. Modified circuit for measuring dielectric properties of materials as a function of dc biasing fields.

analyzer is shown in fig. 1 and was used when voltage was applied to the sample. Micrographs were taken with an ISI model SX-40A scanning electron microscope with Kevex EDS detector. SEM samples were chemically etched by a solution of 33O/6H20, 33% HNO,, 33% acetic acid and 1% HF for several minutes.

3. Results All compositions showed a density after sintering of approximately 95% of theoretical density. Table 2 lists the composition and a property summary.

Niobium

and yttrium

were added in the amount

of 0.5 mol%, and their dielectric constants varied little from the undoped material as shown in fig. 2. Their loss tangents, however, were substantially higher than that of the undoped material. SEM micrographs showed the grain sizes of these materials to be only slightly different from that of the undoped material, as seen in fig. 3. The grains of the undoped material are square-shaped with sharp edges, which indicates there was no liquid-phase sintering. The high loss tangents of the Nb-doped and Y-doped materials made them unsuitable for further study. It is interesting to note that these are both possible donor dopants (Y3+ on A*+ and NbSf on B4+ ). Other investigators have had success in doping BaTiC& and SrTi03 with Nb or Y to increase the dielectric constant, indicating a stoichiometry adjustment for occupation of A sites by these ions may produce better results [ 10,29,32,33].

Table 2 Summary of measurements of different compositions. Measurements are taken at 25°C and 1 MHz Composition

Grain size (Pm)

t

tan S

At (% & 16 kV/cm)

3403 3239 3294 2626 2757 3099 1708 2224 2182

0.0108 0.0269 0.0182 0.0108 0.0060 0.0040 0.0065 0.0055 0.0046

16.4 _ 23.3 12.4 19.8 10.8 9.4 8.2

319

Volume 15, number

56

MATERIALS

0.14

0.12

0.10

E

0 06

2k 5 I-

0.06

: -I

0.04

0.02

0.00 10000

J

I

LETTERS

January

1993

Fig. 4 shows the loss tangent for the Bi-doped material to be somewhat above that for the undoped material. The shorter and more diffuse Curie peak of the Bi-doped material is puzzling. SEM micrographs show the grain size to be much smaller ( FZ2 pm) than that of undoped material. Smaller and more uniform grain sizes have been linked to increased dielectric constant due to increased amount of interface barrier capacity, although this is more true in the ferroelectric region than the paraelectric region [ 8,23, 29,34-361. Bi203 is known to act as a liquid-phase sintering agent that also retards grain growth, as shown in the SEM micrograph. Although Bi5+ is a possible oxidation state which would act as a donor on Ti4+ sites, Bi3+ is a more likely oxidation state, as shown by Burn and Neirman [ 331. In general, the

6000

5

6000

ti

s

0

.c”

4000

z 2 ; 2000

0

l-

-40

-20

0

20

40

60

80

Temperature (“C) Fig. 2. Dielectric

properties

of the Y- and Nb-doped

materials

at

1 MHz: (0 1 B%..dhwTiO~, (0 1 B~.60Sro.40Yo.olTb.9903, (B)

Bao.soSr~.4~Nbo.olTi~.~~O~.

o! -50

-30

-10

10

30

50

70

90

Temperature (“C) Fig. 4. Dielectric Fig. 3. SEM micrograph

320

of undoped

barium

strontium

titanate.

properties

CM) Bao.&ro.wTi09,

of the Bi-doped

material

( •i ) Bao.60Sro.40Bio.o,Ti,.9903.

at

I MHz:

MATERIALS LETTERS

Volume 15. number 5.6

January 1993

more uniform the grain size, the sharper the Curie peak, as grains tend to undergo the tetragonal-to-cubit transition at the same time. 3.2. Acceptor dopants Doping with Mn and Fe showed the best results as far as reducing tan S with respect to the undoped material, as seen in fig. 5. Interestingly, both materials showed more rounded grains typical of liquid-phase sintered materials, but with grain sizes virtually unchanged from that of the undoped material (see figs. 6 and 7). No substantial amount of phase was seen at the grain boundaries in any of these materials in Fig. 6. SEM micrograph of Fe-doped barium strontium titanate.

005 -

004

-

E &

0.03 -

ii c * :: A

0.02 -

0.01 -



0.00 12000

t0000 -

Fig. 7. SEM micrograph of Mn-doped barium strontium titanate.

8000 -

E !!I

sE

6000 -

.z I5

4000 -

2 .g

2000

-

OJ

-60

-40

I

-20

0

20

40

60

Temperature (“C)

Fig. 5. Dielectric properties of the Mn-, Fe-, and Ga-doped materials measured at 1 MHz: (0) B&.noSr0,+,Ti03, (Bf Bao.60Sr~.4~Mno.o~Ti~.9903, Bao.soSro.4,Gaool*Tio.~O~.

(0)

Bao.soSro.4cFeo.o,Tiosu03,

CA

1

the SEM micrographs. Energy dispersive analysis of X-rays (EDAX) was inconclusive because of the overlap of the Ba and Ti lines as well as those of the Mn and Fe and background radiation caused by the chamber walls (Fe). Several investigators have shown the beneficial effects of adding Mn to capacitor ceramics [ 16,27,33,37]. While Mn did not substantially alter the dielectric constant, the loss tangent showed significant reduction. The Fe-doped material showed much the same behavior in the area of the loss tangent as the Mn-doped materia’l. The Fe-doped material has a diffuse peak in the dielectric constant of z 5000 at the r, of - 25cC. The undoped material had a much sharper peak in the dielectric constant of z 8500 at the T, of - 5°C. Dop321

Volume 15,number 5,6

January 1993

MATERIALS LETTERS

ing with Ga produced an increase in the dielectric constant and loss tangent around 7’,, but also a sharply decreasing loss tangent above T,, as shown in fig. 5. However, its high loss tangent around TC precluded the Ga-doped material from further consideration. 3.3. Ba and Sr additions to the Fe-doped material As a premium was placed on low loss tangent over a high dielectric constant, we attempted to improve further the loss tangent of the Fe-doped material by adding BaO and SrO. Interestingly, as fig. 8 shows,

this resulted in a large increase in the dielectric constant at T, with a lowering of the loss tangent, as well as a shift back to a T, closer to that of the undoped material. In general, the loss tangent will increase with the dielectric constant near the Curie peak. The effect of the A: B ratio in BaTiO, and SrTiO, has been studied by a number of authors [8,28,29,31,35,36, 38 ]. Excess Ba has shown fmer grain sizes, leading to a higher dielectric constant and lower tan6 in BaTiO, [24]. Syamaprasad et al. [ 391 achieved similar results in BaO,,,SrO,,,TiO, (largely unchanged dielectric constant with a slight decrease in tan 6). Also interesting is the divergence in loss tangent of the material with Sr added in the region 2565°C. The Sr-rich material’s loss tangent increases

0 03 0.20

0.02

0.15

E s

E

s k s

zil

s t

001

0.10

I-

3

0.05

0.00 10000 0.00 5ooo

8000 4000 E s

/

/

5 -

6000 E s

e s 0 ‘c z S .s P

tI

/

3000 -

2

4000

s .g

2000

-

iow

-

z a, g

ZOOQ

0 -I-60

-40

-20

0

20

40

60

80

Tempctrature (“C)

Fig. 8. Dielectric properties of the Fe-doped material with Ba of Sr additions at 1 MHz: (0) Baa.&0.40TiO~, (0 f Baa.~r0.&~.otTb.spo3~ (II) Bao.soSro.~eo.olT~.9903+ 1 mot% Ba, ( 0 ) B~.~~Sro.~oFe,,,Ti~,ggO~ + 1 mo% Sr.

322

0; -50

I

-30

-10

10

30

50

70

90

Temperature (OC)

Fig. 9. Variation of the dielectric properties of the Fe-doped material ( Bao,60Sro.40Feo.o,Tio.9903f at various frequencies: ( q ) I MHz,(m) IOOkHz, (0) lOkHz, (0) 1 kHz.

Volume IS. number

5.6

MATERIALS

only slightly, whereas the Ba-rich material’s loss increases to a larger extent with temperature above 25°C. 3.4. Frequent:,: Some of the more promising compositions were tested at lower frequencies for their suitability as capacitor materials. In general, the dielectric constant of these materials increased slightly with decreased frequency for all the samples tested as shown in fig. 9. The loss tangent increased greatly with lower frequency, except in the case of the Bi-doped material (see fig. lo), which showed little change with frequency, which is in agreement with the results of Burn and Neirman [ 331 for SrTi03 in the paraelectric region.

AETTERS

January

1993

3.5. Voltage The ability of this material to shift the phases of electromagnetic waves rests on its ability to change dielectric constant with applied bias voltage. Voltage was applied to thin samples and its effect on dielectric constant was measured. The addition of the voltage supply as well as a blocking circuit to prevent

1600

1000

!

,

,

,

,

,

,

,

,

,

,

,

0

2

4

6

6

10

12

14

16

16

20

22

Voltage (kV/cm) Fig. Il. Change in the dielectric constant at I MHz and 25°C with applied bias voltage in the Mn- and Fe-doped materials: (0 )

Bao.aoSro.loFen.ol~i0.9903, (A) (0) b.6&o.4JQ, Bao.60Sro.40Mn,,ITio.9903, (0 ) Bao.soSr,,Feo.~lTi0.9903+ 1mol% Ba, (w) Bao.~oSro.aoFeo.oITio.ssO~ + 1mol%Sr.

6000

E

s

r 8

.o

4000

L

ij

s! g 2000

0 -50

-30

-10

10

30

50

70

1000 ]

,

0

2

4

6

6

10

12

14

16

16

i0

90

Voltage (kV/cm) Temperature (“C) Fig. 10. Variation of the dielectric properties of the BI-doped material ( Bao.&ro.~Bi,.,,Tio.~~O~) with frequency: ( 0 ) 1 MHz, (m) 100 kHz, (0) 10 kHz, (0) 1 kHz.

Fig. 12. Change in the dielectric constant at 1 MHz and 25°C with applied bias voltage in the Bi- and Y-doped materials: (m) Ba0.60Sro.40TiO~, ( 0 ) Bao.~~Sr~.~oBio.o,Tio.~~O~.

Baa 60Sro.40Yo.olTio.9903r

(0 )

323

Volume 15, number

56

MATERIALS

large dc voltages from entering and damaging the impedance analyzer resulted in a lower dielectric constant than those measured without an applied bias field. The change in dielectric constant with applied bias voltage is considered to be accurate relative to each sample. In general, the higher the dielectric constant, the greater a AE was achieved for a given voltage (see table 1). Fig. 11 shows the change in dielectric constant for the Fe- and Mn-doped materials and fig. 12 shows the results for Y- and Bi-doped materials. The loss tangent also decreased with applied voltage.

LETTERS

[7] L. Zhou, Commun.

Z. Jiang and S. Zhang, 74 ( 199 1) 2925.

[8] W. Heywang,

J. Mater. Sci. 6 (1971)

January

1993

J. Am. Ceram.

Sot.

1214.

[9] N.-H. Chart, R.K. Sharma and D.M. Smyth, J. Am. Ceram. Sot. 65 (1982) 165. [ 101 A. Yamaji, Y. Enomoto, K. Kinoshito Am. Ceram. Sot. 60 ( 1977) 97.

and T. Murakami,

J.

[ 11I T. Ashida and H. Toyoda, Japan. J. Appl. Phys. 5 (1966) 269.

[ 121N.-H. Chan and D.M. Smyth, J. Electrochem. (1976)

Sot.

123

1584.

[ 13] T. Matsuoka, Y. Matsuo, H. Sasaki and S. Hayakawa, J. Am. Ceram. Sot. 55 (1972)

108.

[ 141 Y.-M. Chiang and T. Takagi, J. Am. Ceram. Sot. 73 ( 1990) 3286.

[ 151 R.D. Shannon and C.T. Prewitt, Acta Cryst. B 25 (1969) 4. Conclusions Doping with Fe produced the smallest loss tangent of all the dopants studied. Adding excess Ba or Sr to the Fe-doped material reduced the loss tangent further. The largest Ae with applied voltage was achieved in materials with a relatively high dielectric constant. Further study is needed to conclude whether acceptor or donor dopants reduce the loss tangent to a greater degree, and to determine the dominant defect chemistry.

Acknowledgement The authors wish to acknowledge the help of Dr. Deepak Ghodgaonkar in designing the blocking circuit used in voltage-capacitance measurements. This work was supported in part by the U.S. Army CECOM Laboratory contract No. BAA 07-90-C-A033 and the Department of Engineering Science and Mechanics.

References [ I] D.A.

Johnson, Microwave Laboratory of Stanford University, M.L. Report No. 825, July 196 1. [ 21 L.M. Sheppard, Bull. Am. Ceram. Sot. 7 1 ( 1992) 85. [ 31 F. Selmi, D.K. Ghodgaonkar, R. Hughes, V.V. Varadan and V.K. Varadan, SPIE, Vol. 1489. Structure, sensing, and control (1991) p. 97. [4]A. Basmajian and R.S. Devries, J. Am. Ceram Sot. 40 (1957) 373. [5] D. Kolar and L. Marsel, Sci. Ceram. 14 (1988) 921. [6] Y. Syamaprasad, R.K. Galgali and B.C. Mohanty, Mater. Letters 7 (1988) 197.

324

925.

[ 161 I. Bum, J. Mater. Sci. 14 (1979) 2453. [ 171Y.-M. Chiang and T. Takagi, J. Am. Ceram. Sot. 73 ( 1990) 3278.

[ IS] M. Drofenik, A. Popovic and D. Kolar, Bull. Am. Ceram. Sot. 63 ( 1984) 702.

[ 191T. Fukami, Y. Kashimoto

and H. Tsuchiya, Ferroelectrics 29 (1990) 126. [20] B. Huybrechts, K. Ishizaki and M. Takata, J. Am. Ceram. Sot. 75 ( 1992) 722. [ 2 1 ] S.B. Desu and D.A. Payne, J. Am. Ceram. Sot. 73 ( 1990) 3391. [22] S.B. Desu and D.A. Payne, J. Am. Ceram. Sot. 73 (1990) 3398. [23] S.B. Desu and D.A. Payne, J. Am. Ceram. Sot. 73 (1990) 3407. [24] S.B. Desu and D.A. Payne, J. Am. Ceram. Sot. 73 ( 1990) 3316. [ 25 ] Y.-M. Chiang and T. Takagi, J. Am. Ceram. Sot. 75 ( 1992) 2017. [26] S.B. Desu and D.A. Payne, J. Am. Ceram. Sot. 75 ( 1992) 2020. [ 27 ] E.P. Hyatt, S.A. Long and R.E. Rose, Bull. Am. Ceram. Sot. 46 (1967) 732. [ 281 F. Kulscar, J. Am. Ceram. Sot. 39 ( 1956) 13. [ 291 M. Kuwabara, J. Am. Ceram. Sot. 64 ( 198 1) Cl 70. [30] H. Mostaghaci and R.J. Brook, Trans. Brit. Ceram. Sot. 82 (1983) 167. [ 311 D.A. Tolino and J.B. Blum, J. Am. Ceram. Sot. 68 ( 1985) C292. [ 321 A.K. Maurice and R.C. Buchanan, Ferroelectrics 74 ( 1987) 61. [33] I. Bum and S. Neirman, J. Mater. Sci. 17 (1982) 3510. [ 341 T. Fukami and H. Tsuchiya, Japan. J. Appl. Physics 18 (1979) 735. [35] K. Kinoshita and A. Yamaji, J. Appl. Phys. 47 ( 1976) 371. [ 361 K.S. Mazdiyashi, Bull. Am. Ceram. Sot. 63 ( 1984) 59 1. [ 371 H. Ihrig, J. Am. Ceram. Sot. 64 ( 198 1) 6 17. [ 381 N.-H. Chan, R.K. Sharma and D.M. Smyth, J. Am. Ceram. Sot. 64 (1981) 556. [39] U. Syamaprasad, R.K. Galgali and B.C. Mohanty, Mater. Letters 8 ( 1989) 36.