J O U R N A L OF M A T E R I A L S S C I E N C E L E T T E R S 5 (1986) 667-670

Structural characterization of yttrium oxide thin films using transmission electron microscopy B. W. K R A K A U E R , J. S. G A U * , D. J. S M I T H

The Institute of Optics, and *Department of Mechanical Engineering, University of Rochester, Rochester, New York 14627, USA

The usefulness of materials for thin films of optical interference filters depends on the chemical composition and structure obtained as a result of the deposition process [1]. A columnar microstructure results in the potentially deleterious properties of unstable spectral reflectance [2], intrinsic stress [3], and form birefringence [4]. Changes in the chemical composition or structure, in a direction perpendicular to the substrate surface, have explained the undesirable property of optical inhomogeneity [5]. Yttria (57203) films are important for use as an integral part of optical filters that reflect or transmit light at a wavelength of 351 nm, the tripled Nd : glass laser wavelength of multistage amplifiers used for inertial confinement fusion studies. To date, however, little research has been reported that relates the yttria film structure to its optical properties [6 9]. We, therefore, report the first structural characterization of reactive electron-beam-evaporated yttria thin films, as a function of film thickness, using the normal modes of transmission electron microscopy (TEM) and the cross-sectional replication technique [10]. Then this structure is related to the spectral reflectance properties of the films. Yttria was deposited onto carbon-coated copper T E M grids and 1 in. diameter quartz pieces. F o r T E M observation, three different film thicknesses (20, 50 and 100 nm) were grown in the same evaporation run. To determine the film thicknesses in situ, the optical transmittance of a quartz substrate was monitored as yttria was deposited on to it [1 II. This transmittance,

therefore, oscillated with a regular period such that the minima corresponded to film thicknesses of 2/4n, 2/2n, 2/n . . . . , where n, the refractive index of yttria, was about 1.8 and 2, the wavelength of light, was 400nm. Quartz lamps were used to heat the substrates (27, 200 and 250°C) and the source-to-substrate distance was 50cm. The residual gas pressure and the oxygen backfill pressure were 2 x l07 and 5 x 10-Storr, respectively. 99.99% yttria sintered tablets were evaporated with an electron beam and the deposition rate at the substrate was determined by an lnficon IC 6000 crystal quartz monitor to be 0.15nmsec -j. A Jeol 100B TEM, operated at 100kV, was used for microstructural analysis and structure determination, and a doublebeam Cary 2300 spectrophotometer was used to determine spectral reflectance data. The selected-area electron diffraction pattern (SAD) for the 5 0 n m yttria film (Fig. 1) has been indexed as cubic bixbyite by comparison with the powder diffraction file data (PDF) [12] for bulk yttria (Table I). For a positive match, we require that the relative intensities match and that d,/d,

=

(1)

D,/D,

where di is the lattice plane spacing of the ith reflection of bulk yttria and Di is the ring diameter of the ith reflection of the yttria film. A match also occurs for the 20 and 100 nm films. Hence, for all film thicknesses (deposited at 200 °C): the crystal structure does not change; no second phase or precipitates of, for

T A B L E I Diffraction data for yttria films (deposited at 200 ° C)

Ring

Bulk

No.

di (nm)

1 2 3 4 5 6 7 8 9 10 I1 12 13 14 15 16

0.4340 0.3060 0.2652 0.2500 0.2372 0.2261 0.2165 0.2080 0.1936 0.1874 0.1818 0.1769 0.1720 0.1677 0.1636 0.1599

0261-8028/86 $03.00 + .12

Film Int. 16 100 30 7 1 8 1 12 3 46 2 <1 5 1 4 31

hkl 2 11 222 400 4 11 420 332 422 43 1 52 1 440 433 600 6 11 620 54 1 622

© 1986 Chapman and Hall Ltd.

d,/d i

1.0 1.418 1.637 1.736 1.830 1.920 2.005 2.087 2.242 2.316 2.372 2.453 2.523 2.588 2.653 2.714

D i (cm)Di/D 1

+ 0.025

+ 0.02

Int.

5.95 8.36 9.68 10.15

1.0 1.41 1.63 1.71

M VS+ S VW

11.30

1.90

W

12.32

2.07

M

13.70

2.30

VS

14.95

2.51

VW-

16.05

2.70

S 667

Figure 2 Centred dark-field micrograph of a 100nm yttria film (deposited at 200 °C), formed by imaging a portion of the 22 2 reflection in Fig. I.

Figure 1 Selected-area electron diffraction pattern of a 50 n m y t m a thin film (deposited at 200 ° C) indexed as cubic bixbyite.

example, yttrium, exist; no amorphous phase exists; and there is no preferred crystallite orientation. Also, the crystallite diameter is constant with thickness, as the broadness of the rings does not vary. Therefore, the grains are equiaxed in a plane parallel to the substrate surface, but not necessarily in the perpendicular direction. The only differences between the patterns, due to the different scattering volumes of the films, are that the 211, 411, 332, 431, and 611 reflections do not appear for the 20 nm film. From centred dark-field microscopy (Fig. 2) the average crystallite diameter is found to be 3 nm. In the cross-sectional replica (Fig. 3a), each fracture runs continuously from the substrate surface to the film surface and in a direction perpendicular to these surfaces. The surface replica (Fig. 3a) indicates that the fractures are actually intercolumnar. Bright-field microscopy (Fig. 4c) also indicates that the films consist of columnar grains, but that the regions along which the fractures occur (Fig. 3a) are of a large void density and that the average column diameter is 20 nm. This columnar microstructure compares well with that observed in other thin films formed by physical vapour deposition (PVD) [13]. If the substrate temperature is lowered, the columnar microstructure becomes much less pronounced (Fig. 3b). Also noted is the evidence of Moir~ fringes as revealed by the bright-field electron micrograph of the 100 nm film (Fig. 4c). Therefore, each column consists

of many small single crystals stacked on top of one another [14]. This is also evidenced by comparing Fig. 4a to c, where the number of diffracting grains is increasing with thickness. Thus, as the film grows, nucleation sites form. Also, the grains may be equiaxed in three dimensions, which needs to be determined by forming an electron diffraction pattern of the cross-section of the film. Centred dark-field microscopy (Fig. 2) shows that the columns are not single grained in the lateral direction. The model (Fig. 5) summarizes the yttria film microstructure and confirms the characteristics of zone I of the zone model that has been developed [13] and simulated with computers [15]. For zone I, film growth has been explained by a lack of atomic mobility, or surface diffusion, and a shadowing effect. The different zones are defined by the ratio of substrate temperature, Ts, during deposition, to the melting point of the evaporant, Tin, and for zone I Ts/Tm

<

0.3

(2)

Indeed, our films have ratios of 0.18 (Fig. 3a) and 0.19 (Fig. 3b). The plot of transmittance against wavelength (Fig. 6) indicates that the refractive index of the 100 nm yttria film changes because of water adsorption on exposure to air, since the curve minimum, labelled A, does not occur at 400 nm, the wavelength used to monitor film thickness in situ. That water adsorption is the source of this shift can be explained by first considering an approximate equation for the average refractive index, n, of a thin film n2 =

[(1 -- P)n 4 + (1 + p)n2vnZv]/[(1 + P)n2v + (1 - P)n~l

(3)

Figure 3 Pt/C cross-sectional replicas of 350 n m thick films formed on substrates at: (a) 250 ° C, and (b) 27 ° C. Shown are the film surface, S, the film cross section, C, and the quartz substrate, Q.

668

!i~ ~

o

Figure 4 Bright-field micrographs o f yttria thin films (deposited at 200 ° C) of different thicknesses, but grown in same evaporation run: (a) 20 nm; (b) 50 nm; (c) 100 nm. The ring in Fig. 3c indicates Moir6 fringes.

where P = (density of voids)/(density of film), n v is the index of refraction of voids, and ns the index of refraction of the columns [16]. Thus, if the voids fill with water, nv changes from 1.0 to 1.33, and n increases. Since the position of the minimum, labelled A, is defined by

4 dn/2 =

1

(4)

where n is the refractive index of the yttria film, d the film thickness, and 2 the wavelength of light; as n increases, 2 must also increase. Thus, the position of the minimum and the whole curve shift to higher values of 2. The capillary action of the void regions has explained why the films adsorb water [17]. Such films have unstable spectral reflectance properties, because as the humidity changes, so does the index of refraction [2]. Optical inhomogeneity causes the peak (labelled B in Fig. 6) to lie higher than the value of transmittance for an uncoated quartz substrate, because the refractive index, n, of the film varies in a direction perpendicular to the substrate surface [5]. From Equation 3, for such a gradient to occur, either P, nv, or ns, which are structure sensitive, must vary with thickness. Therefore, structural characteristics that have been eliminated as sources of this gradient are: a changing column diameter with thickness [17]; a changing crystalline phase with thickness [18]; and a changing crystallite size with thickness. We have not, however, eliminated stoichiometry changes, which frequently occur in oxides grown by PVD [5].

It can be concluded that yttria thin films, produced by reactive electron-beam evaporation under the conditions of our study, are single phase with the cubic bixbyite structure independent of film thickness (up to 350nm), when deposited on to substrates heated between 200 and 250 ° C. The microstructure can be described as columnar with an average diameter of 20 nm that does not change with increasing film thickness. The columns are composed of many grains, which have a random distribution in crystallographic orientation, have an average diameter of 3 nm that does not change with thickness, and are possibly equiaxed in three dimensions. This structure can be classified as a type I in the zone model for film growth. Optical inhomogeneity can only be attributed to changes in film stoichiometry with film thickness. Surrounding the columns are voids that adsorb water; this results in films with unstable spectral reflectance characteristics.

Acknowledgements We thank Paul Vianco for his many valuable discussions on the experimental aspects of this study and gratefully acknowledge the Materials Science Program at the University of Rochester for the use of their Electron Microscopy facilities.

References 1. E. R I T T E R , Appl. Opt. 15 (1976) 2318. 2. S. O G U R A and H. A. M A C L E O D , Thin Solid Films 34 (1976) 371. 3. A. E. ENNOS, Appl. Opt. 5 (1966) 51. 4. M. R. J A C O B S O N , F. H O R O W I T Z and B. LIAO, Proe. Soc. Photo-Opt. Instrum. Engng 505 (1984) 228. 5. J. P. B O R G O G N O , B. L A Z A R I D E S and E. PELLETIER, Appl. Opt. 21 (1982) 4020. ,,,

/

B~

0.3 0.4 0.5 0.6 0.7 0.8 0.§ 1.0 1.1 112 WAVELENGTH (#m)

I Figure 5 Model of yttria thin film microstructure, where average diameter of crystallites, G, is 3 nm, average void width, V, is 3 nm, and average column diameter, C, is 20 nm.

Figure 6 Relative transmittance against wavelength. Curve 1, transmittance of a 100 n m yttria film deposited on to quartz (at 200 ° C) relative to an uncoated quartz substrate. Curve 2, transmittance of an uncoated quartz substrate relative to another uncoated quartz substrate. Optical inhomogeneity occurs when curve 1 rises above curve 2. Water adsorption occurs when curve 1 shifts to higher wavelengths, as indicated by A. 669

6. 7. 8. 9. 10. 11. 12. 13.

670

A. F. A N D R E E V A and I. YA G I L ' M A N , Inorg. Mater. 19 (1983) 443. T A D A S H I T S U T S U M I , Jpn J. Appl. Phys. 9 (1970) 735. W. H E I T M A N N , Appl. Opt. 12 (1973) 394. M. C O L E N and R. F. B U N S H A H , J. Vac. Sci. 13 (1976) 536. J. M. P E A R S O N , Thin Solid Films 6 (1970) 349. D. H. H A R R I S O N , Proc. Soe. Photo-Opt. Instrum. Engng 50 (1974) 1. Powder Diffraction File, Card No. 15-1200, Joint Committee on Powder Diffraction Standard, Pennsylvania (1974). P. H. L I S S B E R G E R and J. M. P E A R S O N , Thin Solid Films 34 (1976) 349.

14. 15. 16. 17.

18.

C. R. M. G R O V E N O R , H.T.G. HENTZELL and D. A. SMITH, Acta Metall. 32 (1984) 773. LIAO B A N G J U N and H. A. M A C L E O D , Proc. Soc. Photo-Opt. lnstrum. Engng 505 (1984) 236. H. A, M A C L E O D , ibid. 325 (1982) 21. M. H A R R I S , H.A. MACLEOD, S. O G U R A , E. P E L L E T I E R and B, VIDAL, Thin Solid Films 57 (1979) 173. R. E. K L I N G E R and C. K. C A R N I G L I A , Appl. Opt. 24 (1985) 3184.

Received 6 December and accepted 9 December 1985