E L S E V I E R Microelectronic Engineering 29 (1995) 285-288

MICROr.I~C'TROI~I¢~ EH(~IHEEKUqG

Texture, structure and domain microstructure o f ferroelectric PZT thin films

J. Hector a,b, N. Floquet a, J.C. Nlepce, P. Gaucher b , J.P. Garme b.

a Laboratoire de Recherches sur la R~activit6 des Solides - U.R.A.C.N.R.S.23 - Universit6 de Bourgogne -

Facnlt~ des Sciences Mirande - BP 138 - 21004 Dijon Cedex - France.

b Laboratoire Central de Recherches - Thomson-CSF - Domaine de Corbeville - 91404 Orsay Cedex- France.

Ferroelectric thin films with different compositions from PZT 50/50 to PbTiO3 were prepared by a modified sol-gel process and deposited by spin-coating on Pt/Ti/SiO2/Si(100)wafer. The ferroelectric and dielectric properties of the thin films dependent on thin film composition are presented and compared with bulk ceramic properties. The results correlated with the crystal structure and ferroelectric domain microstucture of PZT crystallites determined by X-ray diffraction. The influence of the sol-gel film process on the ferroelectric properties, the crystallite texture, structure and domain m i c r o b e is discussed.

1. INTRODUCTION

Many research groul~ are working on thin films based on Pb(Zr,Ti) compositions, aiming at ferroelectric, piezoelectric and pyroelectric sensing devices. They have reported on the difference of electrical properties of otherwise similarly processed PZT thin films that were deposited on different substrates. Several factors such as phase purity, microstructural quality, crystaUine orientation, grain size, type of substrate used for deposition are shown to influence the film electrical properties. In this paper, we attempt to bring out the relationships among crystallite structure, domain microstmcture and electrical properties for a series of sol-gel prepared PZT thin films by varying the composition.

2. EXPERIMENTAL PROCEDURE

Ferroelectric thin films with different compositions from PZT 50/50 to PbTiO3 were prepared by a modified sol-gel process [1] and deposited by spin-coating on Pt/Ti/SiO2/Si(100) wafer. The Pt bottom electrode was obtained by sputter deposition of a thin Ti adhesion layer followed by a Pt layer of about 100 nm thickness on thermally oxidized (100) Si wafers. The Pt electrode is made of (I 11 ) oriented ~ l i t e s with sizes from 30 to 50 nm. Films were dried and pyrolyzed at 350°C for 10nm. Films were then rapid thermally annealed at 700°C for 2 nm. Films of 300rim thickness were prepared by one coating.

Electrical properties were measured and ferroelectricity demonstrated by Sawyer-Tower circuit. The texture of the films was observed by a Jcol 6400F scanning electron microscope. The structure, domain microsmacmm and orientation were analyzed by X-ray diffraction using a curved sensitive detector (Inel CPS 120 using Cu K~1 radiation). Using a (0,20) diffractometer, only the crystallographic planes parallel to the silicon wafer are analyzed, whereas with a curved detector the orientation of the diffracting planes can be chosen by varying the incidence angle. This set-up is very useful to record PZT film diffractograms free of substrate peaks, such as (400)Si and (111)Pt peaks.

&RESULTS

3.1. Ferroelectric measurements All synthetized films from VZT50/50 to PbTiO3

exhibit ferroelectric properties. Table 1 gives ferroelectric data obtained from hysteresis curves measured on the PZT films, for comparison, ferrcelectric data ofbnlk ceramics are given [2].

PZT epsilon Pr C/m2 Ec kV/cm2 Zr/Ti :film bulk film bulk film bulk 50/50 700 1000 0.30 0.35 50 15 20/80 350 300 0.35 60 0/100 150 200 0.35 0.75 70 20

Table 1 With decreasing Zr/Ti ratio compositions, the

shape change of ferroelectric hysteresis loops is

0167-9317/95/$09.50 © 1995 - Elsevier Science B.V. All rights reserved. C C/"~ I 01 ~ 7 _ O q 17(QS~f)~ 161 - 1

286 J. Hector et al. /Microelectronic Engineering 29 (1995) 285-288

similar to that observed in bulk ceramics: decrease of the dielectric constant and increase of the remanent polarization and the coercive field. However, for the same PZT composition, differences between the film and the bulk ceramic are observed: the dielectric constant is lower and the remanent polarization and the coercive field are higher for the films. These differences could be explained by a variety of effects, and generally by the presence of pyrochlore phase and composition and size inhomogeneities of the crystallites within the film.

3.2. Structural analyses All the PZT flms are characterized by a high

density of crystallites with size below 0.5gm as shown on the SEM micrograph (fig. 1).

Figure 1. SEM PZT 50/50 film rnigrograph

The crystallites are phase-pure perovskite. No pyrochlor¢ phase was detected.

Intensity. (cp~) 1000

800

600

400

200

0

Pt(111)

20 30 20 (degrees) 50 60

Figure 2. Diffractogram of a as-grown PZT50/50 film

As shown in fig.2, a diffractogram of PZT 50/50 film exhibits single peaks defining a pseudo-cubic structure with a unit cell volume close to the known tetragonal cell volume.The parameter values were

obtained by refinement from all the (hkl) PZT diffraction peaks. The (111) and (311) Pt diffraction peaks were used as internal standard. By comparison, diffractogram of PZT 50/50 powder synthetized by the same sol-gel route, shown in fig.3, identifies a tetragonal structure with the values of a and c parameters known for a PZT 50/50 bulk ceramic.

2000 Intensity (cps)

~ 1 ~ ~ ~" 1500

1000

500

20 30 20 (degrees) 50 60

Figure 3. Diffractogram of a sol-gel PZT 50/50 powder

This result proves that the chemical composition of our sol-gel process is well controlled, and, as the crystallites are shown without preferential orientation, the crystallite growth is not controlled. Sol-gel powder and film diffractograms have broadened peaks of the same order, due to the same crystallite size and structural defect effects. In consequence, the decrease of tetragonality between the sol-gel powder and the film could not be due to the sol-gel process. This structural effect is more likely due to film stresses induced by the substrate From structural analyses, the PZT 50/50 films arc perovskite phase with unmeasurable tetragonal distortion, while from electric measurements, films exhibit ferroelectric properties.

Intensity (cps) 1000 ,

800

600

400

200

0

~E

Pt(111 )

20 30 20 (degrees) 50 60

Figure 4. Diffractogram of a as-grown PZT 20/80 film (0.8gm thick).

J. Hecwr et al. / Microelectronic Engineering 29 (1995) 285-288 287

To clear up these inconsistent results, the same structural analyses were carried out on PZT 40/60, 30/70, 20/80, 10/90 and PT films the structure of which is of higher tetragonality (1.035 < c/a <1.065). For these compositions, the film diffractograms are characteristic of tetragonal structure, the splitting of XRD peaks is well observed and is increasing as the Zr/Ti ratio is decreasing. The PZT 20/80 diffractogram is shown in fig.4. Compared to known values of a and c parameters (1.035<¢/a<1.064), film c/a ratio is smaller for all compositions (1.012<c/a<1.053), corresponding to a decrease of c parameter ( Ac ~ - 0.007 nm) and an increase of a parameter (Aa 0.004 nm). These results are reported in fig.5.

4.15 =

4.05 | : . c f i lm E I " t =

¢w 3.9 ~ .~ i tetragonal a cer

3.85 i

0 0.5 1

P b Z r 0 3 PbZrO.5TiO.503 PbT i03

Figure 5. a and c parameters for films (this work) and bulk ceramics [2].

These results proves the PZT film composition is well-controlled, the tetragonality change is constant for all synthetized compositions and in consequence is not related to the film composition but likely to the film-substrate stress relationships.

3.3. Ferroelectric domain microstructure of the as-grown thin film

An another important result concerns the ferroclectric domain microstructure of the crystallites in a difl~ctogram by the intensity ratio of the double peaks such as (001)/(100), (101)/(110),...[3]. The (001)/(100) ratio as measured from the film diffractograms for all the synthetized compositions is ten times less ( - 0.05) than for an unpoled bulk eeramic. This strong decrease of intensity ratio of the double peaks expresses the ferroelectric domain microstructure of the PZT crystallites : each PZT crystallites has 180 ° and 90 ° domains and for each PET crystallite, the ratio a-domains/c-domains is much greater than 1.

(an a-domain has its c axis parallel to the substrate whereas a c-domain has its c axis perpendicular to the substrate). This domain configuration of the as- grown crystallites clear up the inconsistent results obtained from PZT 50/50 films. The tetragonal structure of these films exists but does not appear on X-ray diffractograms due to the a-domain configuration of the PZT crystallites.

3.4. Ferroelectric domain microstructure of the thin film after poling

The presence of 90°domains in the PZT crystallites and of a-domain configurated thin films is confirmed by the X-ray diffractograms of the remanent poled thin films as shown in fig. 6:

1000 Intensity (cps)

800

600

400

200

0

Pt (111)

g E 5"

20 30 20 (degrees) 50 60

Figure 6. Diffractogram of a PZT 20•80 fi lm (0.glum thick) after poling under 40V the integrated intensity of the (001), (101), (002), (102) and (112) diffraction peaks is equal or greater than the one of their (100), (011), (200), (210) and (211) counterparts. The intensity ratio change between the as-grown film (fig.4) and the remanent poled film (fig.6) proved that: the as-grown PZT crystallites have multi-domain microstructure with a majority of a-domains, whereas the multi-domain microstructure of the poled crystallites has a majority of c-domains, which is expected for a rcmanent polarization state. This result proves that the remanent state is obtained by electrical 90 ° domain switching.

The hysteresis response for an a-domain configured thin film should be different from a remanent poled thin film, which is ¢-dom_ain configured. This effect is well observed by comparing the (~ first cycling polarization curves ~> obtained by increasing the voltage after each cycle up to 40V and the following ¢~ cycling polarization c u r v e s >L

As shown in figure 7, the remanent polarization values for the ~, first cycling polarization curves ~>

288 J. Hector et al. / Miclvelecovnic Engineering 29 (1995) 285-288

are increasing continuously up to a maximum polarization. For the second and following cycling polarization curves, the remanent polarization values increase abruptly from the same minimum applied voltage.

E 0.2

~- 0.'1

0 I~ 0

g -o.1 E ID n." -0.2

m

x X X ~ X x x ~ x X :

-X-X-~. i I 1~)" = . 20

Xxxxxx x x

V app l ied (Volt)

Figure 7. Variation of remanent polarization with voltage for PZT 20/80 film ; - first and x second polarization cycling curves

Further, the remanent polarization values for the first cycling polarization curves indicates that the switching of the as-grown 90 ° domain is occuring gradually as the field is increasing to a maximum polarization field. On the other hand, for the second and following cycling polarization curves, the remanent polarization curve evidences that all a- domains which are reversible switching, are almost switching for the same applied minimum field. This abrupt switching could be related to a decrease of the switching stress within a crystallite seeing that the a-domain population was decreasing after the first 40V polarization.

4. DISCUSSION

Ferroelectric properties are usually connected to the ferroelectric domain microstructure of the perovskite crystallites, which in turn express their strain state within the film. Many factors are involved in the determination of the ferroelectric domain configuration.Thus, in a study on a series of sol-gel PZT films, Tuttle and al. [4] reported that PZT film stress in the vicinity of the Curie point controls the crystallite 90 ° domain arrangement within the film• According to this transformation stress concept, PZT films that are preferentially a- domain oriented, should be under tension at the Curie temperature. From this thesis, the sol-gel PZT film crystallized on a Pt(lll)/Ti/SiO2/Si(100) substrate at 700°C during 2 mn should be under slight tension or free of tension. There is no significant change in the domain microstructure

through an hour annealing at 700°C, indicating no significant change in the film tension.

Grain size and domain configuration are usually connected. This study proves that 90 ° domains are present in PZT crystaUites with size below 0.5~m. This result is in good agreement with data reported by Arlt [5]. The critical grain size which marks no 90 ° domain formation is reported to be about 0.3 to 1.0 gin for PZT ceramics.

Relationships between tetragonal distortion of the perovskite structure and domain configuration are not extensively reported, whereas they are closely connected because both occur to minimize overall film stress. From our result, it appears that the tetragonality decrease could not be related to the sol-gel process but more likely to the film-substrat strain relaxation because it is constant for the same deposition process, whatever the PZT composition. The low tetragonal distortion and the a-domain configuration, which appear for randomly oriented crystallites, correspond to a smaller tension relaxation between the PZT crystallites and the substrate than if only an a-domain configuration was occuring.

5. SUMMARY

Crystallite structure, domain microstructure and electrical properties for a series of sol-gel prepared PZT thin films with different compositions were presented and their relationships were discussed. Results evidenced the dependence of the crystallite structure and domain microstructure on the electrical properties of the films.

REFERENCES

1. P.Gaucher, J.Hector, J.C. Kurfiss, NATO ASI Series E, Vol.284 (1994) p. 147.

2. Landolt-B6rnstein, Ferroelectrics and related substances, Group III, Vol. 16a (1981) p.77 and 119.

3. C.M.Valot, N. Floquet, P. Perriat, M. Mesnier, J.C. Niepce, 3 rd Int. Symp. on Domain structure of ferroelectrics and related materials. Zakopane (Poland) (1994).

4. B.A.Tuttle, T.J.Garino, J.A.Voigt, T.J.Headiey, D Dimos and M.O.Eatough, NATO ASI Series E, Vol.284 (1994) p.l17.

5. G. Arlt Ferroelectrics 104 (1990) p. 217.