Optical loss distribution in anodically oxidized aluminawith a 2-D structure
Mitsunori Saito, Masayuki Kumagai, Mitsunobu Miyagi, and Kenji Wada
Optical losses have been measured for anodized alumina films with 100-200-A size microstructure. Asignificant loss increase is seen near the surface of the film, and such a loss distribution disappears by treatingthe film with hot water or glycerin. The optical loss also depends strongly on the direction of polarization.An alumina microstructure has been modeled based on electron microscope observations, and optical losseshave been calculated theoretically. From a comparison of experimental and theoretical results, it wouldappear that distribution and anisotropy of optical losses are caused by conical micropores and aluminummicrocolumns. Key words: Artificial dielectrics, alumina, anodization, optical loss.
1. Introduction
Artificial dielectrics perform a variety of opticalfunctions which cannot be achieved with naturallyexisting materials.' In particular, artificial materialswith 1-D (layer), 2-D (wire), or 3-D (dot) microstruc-tures are attracting great interest because of their opti-cal nonlinearity and quantum size effects.2-4 Al-though artificial materials have long been used inmicrowave technology,5 construction of an artificialmicrostructure is difficult for optical uses since its sizeshould be of the order of 100 A, i.e., much less than awavelength of light. Recently, multilayered structureoptical materials have been fabricated by sputtering orvapor phase epitaxy.6-8 However, more complicatedmicroprocessing techniques such as x-ray lithographyand molecular beam epitaxy would be required to con-struct microstructures in two or three dimensions.
A possible much simpler process is anodic oxidation.Anodically oxidized alumina (A1203) contains manycylindrical micropores arranged in parallel (Fig. 1).9The diameter and the spacing of the pores are -100and 400-600 A, respectively-much less than a wave-length of light. The pores may be filled with variousmaterials (metals, semiconductors, and dielectrics) by
Kenji Wada is with National Institute for Research in InorganicMaterials, Tsukuba 305, Japan; the other authors are with TohokuUniversity, Department of Electrical Communications, Sendai 980,Japan.
simple processes such as immersion and electroplatingin a suitable solution. Therefore, an anodized alumi-na film becomes a favorable host for 2-D microbodies.
In previous papers the authors demonstrated boththeoretically and experimentally that an anodized alu-mina film acts as a polarizer or a birefringent platedepending on the material implanted into the pores.1013
In the course of the experiment, we noted that refrac-tive index and transmitted light intensity vary with theposition in the film; the refractive index graduallydecreases toward the surface of the film and light tendsto converge around the center of the film. The intensi-ty distribution of light may originate partly from therefractive index profile of the film. However, thereseems to be another origin which yields the optical lossdistribution, since the variation of the light intensity isobserved even in a film of a short optical length wherethe refractive index profile does not deflect a light pathso much.
In this work we have measured loss distributions inthree kinds of alumina film treated in different waysafter anodization. We have also calculated the opticallosses theoretically and compared them with the ex-perimental results. In the following sections, we dem-onstrate that the optical loss distribution can be un-derstood by considering the scattering loss of a conicalpore structure.
II. Sample Preparation
The fabrication process of anodized alumina isshown in Fig. 1. The starting material is a 99.99%aluminum plate 500 Asm thick and 25 X 8 mm2 in area.First, the aluminum plate was polished mechanicallyand electrochemically to a thickness of -300 ,im.Then the plate was immersed in a 20-wt% sulfuric acid
solution and the dc voltage was applied by two carboncathodes placed on each side of the aluminum plate.In this way, the aluminum plate was oxidized simulta-neously from two opposite surfaces. The applied volt-age was controlled between 20 and 24 V to keep theelectric current density at 0.5 mA/mm2 on the surfacesof the aluminum plate. With this current density, thealuminum plate was oxidized completely in -100 min.
Alumina films prepared in this manner were rinsedwith pure water to remove sulfuric acid from the pores;consequently, the pores were filled with water at thisstage. We designate the film of this stage as sample A.Another film, sample B, was treated further in boilingwater for 1 h. By this treatment (hot water sealing),the pores were filled with hydroxide of aluminum andas a result the film became homogeneous.'4 A thirdfilm, sample C, was immersed in glycerin to replace thewater in the pores. Finally, these three alumina filmswere sliced and polished to a thickness d of 100-220,gm. The prepared samples were stuck on to glassslides for mechanical support in the optical measure-ment.
111. Experiment
The transmissivities of the alumina films were mea-sured with the optical system shown in Fig. 2. Alinearly polarized He-Ne laser of 0.63-gm wavelengthwas focused on a sample attached -to a glass slide.Transmitted light was detected by a silicon photodi-ode through a 50-,gm diam pinhole. Since an image ofthe sample was magnified six times by lenses, thespatial resolution of measurement was evaluated to be-8 (=50/6) gim. The intensity distribution of thetransmitted light was measured by moving the samplestep by step at 10-/Am intervals. The light intensity
Fig. 1. Fabrication process of an anodized alumi-na film.
transmitted through a glass slide without a sample wasalso measured as a reference.
Optical power loss 2a was calculated from transmit-ted light intensities I0 for a glass slide and I, for a glassslide with a sample. By taking account of Fresnelreflection losses on surfaces, I1110 for a sample of opti-cal path length d is expressed as
Ij1I = TOT1T2 exp(-2ad)/Io, (1)
where To, T, and 2 are transmissivities at air-glass,glass-alumina, and alumina-air interfaces, respective-ly. From the refractive indices of air (1.00), glass(1.46),15 and alumina (1.62),16 To, T, and 2 are calcu-lated to be 0.965, 0.997, and 0.944, respectively.' 7
Figure 3 shows the results for three kinds of aluminafilm. Open circles indicate optical losses for polariza-tion whose electric vector is vertical (V) to the pore axis(Fig. 1). Closed circles correspond to polarization hor-izontal (H) to the pore axis. In all the films, opticallosses are larger for H-polarization than for V-polar-ization, and abrupt peaks are seen at the center of thefilm or the bottoms of the pores (x = 0 gim). Thesephenomena may be caused by residual aluminum,which is described later. In sample A, the optical lossincreases gradually from the center to the surface. Insample B, the optical loss is almost uniform although itincreases in the regions close to the surfaces. SampleC also shows a uniform loss distribution over the wholefilm.
Fig. 2. Experimental setup for measuring the op-tical loss of an alumina film.
Fig. 3. Optical losses measured at various positioifilms: (a), (b), and (c) correspond to samples A, B,
tively.
IV. Structural Model
To make a structural model for theoreticfracture sections of alumina films were obEscanning electron microscope (SEM). E:SEM pictures are shown in Fig. 4. As rimany researchers,9"14"1820 anodized aluminamany tubular cells. In Fig. 4(a), a pore iscenter of each cell. Pore diameters were 8the middle of the film (x O 0 Am) and 150-
100 150(a) 1000 A
Cb) i a
Val I Fig. 4. SEM pictures of anodized alumina films taken (a) near thesurface (x 100 um) of sample A, where pores are seen at the centers
100 150 of tubular cells, and (b) at the center of sample B, where a residualaluminum layer is seen.
( ) the surface (x - 100 gim). The spacing between poreswas 420-540 A, independent of the position. By con-trast, pores were not observed in sample B since thefilm was homogenized by hot water sealing. As exem-plified in Fig. 4(b), a layer of residual aluminum wasobserved at the center in all the films, which resultedfrom incomplete oxidation in the anodization process.
Based on the results of our SEM observations andother researchers' work,1420 we made the structuralmodel shown in Fig. 5. At the beginning of anodiza-
V tion, aluminum surfaces are oxidized and pores areformed as designated by 1. Diameter 2ao and spacing
00 150 b of pores are determined by anodizing conditions suchas applied voltage and type of solution.9 As oxidation
is of alumina' advances deep into the aluminum substrate, poresand C, respec- grow successively as designated by 2, 3, and 4. At the
same time, the diameter of the pores is enlarged sincepore walls (alumina) are eroded by a sulfuric acidsolution.20'2' As a result, pores of a conical shape areformed in the film. At the center of the film, anunoxidated layer of aluminum remains, whose thick-
al analysis, ness 6 is measured to be -1 gm by the SEM observa-served by a tion. Further, we assume that unoxidated aluminumxamples of columns of diameter 2 am also remain between pores.eported by Although this assumption was supported by our recent
a consists of work,22 the columns were too thin (probably <10 A) toseen at the be observed by the SEM.30-120 A in Table I lists dimensions of the pore structure evalu-.270 A near ated by the SEM observations. Existing published
Fig. 5. Structural model of an anodized alumina film. Conicalpores grow successively from 1 to 4 as oxidation advances from the
surface to the center of the aluminum substrate.
Table 1. Fabrication Conditions and Pore Size of Anodized Alumina
This work Ref. 9
Sulfuric acid 20% 15%Oxidation voltage 20-24 V 20 VFilm thickness 150 + 150 ,um (both sides) <1 ,umPore diameter (2a) 80-120 A (x = 0 im) 120 A
150-270 A (x = 100 im)Pore spacing (b) 420-540 A 440 A
data for thin alumina films9 are also shown for compar-ison. Our data almost coincide with the data in theliterature except for the pore diameter near the surface(x 100 gim). Unlike thin films, thick films suffermuch erosion since thicker films are exposed to sulfu-ric acid for a longer time during anodization. There-fore, dissolution of alumina plays an important role indetermining shapes and sizes of pores in thick films.For the samples prepared in this work, the dissolutionrate is calculated to be -0.8 A/min from oxidation timeand variation of pore diameters. This value is in goodagreement with the value of 0.75 A/min reported inRef. 21.
In the following theoretical calculation, we take av-erages of the observed values and assume the porediameter 2a (A) at position x (,gm) to be
2a = 100 + 1.1x, (3)
and the pore spacing b to be 480 A. Also, we assumethat pores are completely filled with water and glycerinin samples A and C, respectively, and that there are nopores in sample B.
V. Theoretical Analysis and Discussion
An expression for the optical loss in a porous alumi-na film was derived theoretically in Ref. 12. Whenpores are filled with a transparent material, we canapproximate Eq. (14) in Ref. 12 into a simple form, andoptical power losses 2aH and 2av for H- and V-polar-ization are expressed as
27r5a4 (nl -n2)2 X3a I4ir5a4 n3(n2 -n2)2
b2x 3 (n 2 + n2)2
where X is the wavelength of light and n, and n2 denoterefractive indices of alumina and a material in thepores, respectively. Note that here we use notations2aH and 2 av instead of aH and av in Ref. 12 to distin-guish power losses from attenuation constants of theelectromagnetic fields.
Using Eqs. (3) and (4), we calculated optical losses ateach position x. We set refractive indices to be 1.62 foralumina, 1 6 1.33 for water, and 1.47 for glycerin.2 3 Re-sults are shown for samples A and C with thin lines inFigs. 6(a) and 6(c). In sample A, losses become largernear the surfaces since scattering losses increase withpore size. As Eq. (4) indicates, optical losses dependon the difference of refractive indices, ni - n2, andhence the loss increase is not significant in sample C.For sample B, thin lines are not drawn since there is noscattering loss induced by pores. Calculated values(thin lines) coincide well with measured values (dottedlines) for V-polarization. For H-polarization, howev-er, measured values are larger than the calculated val-ues by -20 dB/mm in all kinds of film including sam-ple B. Therefore, although a loss distribution isexplained well by pore-induced scattering, thereshould be another factor that yields a loss of -20 dB/mm only for H-polarization.
As mentioned earlier, unoxidated aluminum col-umns may possibly remain in the film. Such metalliccolumns yield an optical loss that depends seriously onthe direction of polarization. Here we assume tenta-tively that 3-A diam aluminum columns are located at480-A intervals, i.e., an aluminum column remains inevery space between pores. With a complex refractiveindex 1.36-j7.59 for aluminum,' 5 optical losses werecalculated using Eq. (14) in Ref. 12. The bold lines inFig. 6 show total losses induced by both pores andaluminum columns. By taking account of aluminumcolumns, the losses greatly increased for H-polariza-tion but not for V-polarization. The theoreticalcurves (bold lines) are in good agreement with experi-mental values both for H- and V-polarization. Sincethe optical loss is almost proportional to a4 and b 2 ,similar theoretical curves are obtained when the diam-eter and the spacing of aluminum columns are as-sumed to be 6 and 2000 A, respectively, i.e., aluminumcolumns are assumed to remain only in those regionswhere the spacing of adjacent pores is large enough.
Consequently, an optical loss in the alumina film isthought to originate from two main factors: conicalpores which cause a loss distribution, and residualaluminum columns which cause a loss anisotropy. To
Fig. 6. Theoretically calculated optical loThin lines indicate losses induced by pores Eq. (4), and bold lines indicate total losseaccount of residual aluminum columns. Dmental results taken from Fig. 3; (a), (b),
samples A, B, and C, respec
characterize the features of aluminafurther experiments are necessarybution, anisotropy, and wavelengthand other optical properties.
VI. Conclusion
The optical loss distributions infilms have been measured at the 0.(
All the films show higher transmittance for vertically(a) polarized light than for horizontally polarized light
with regard to pore axis. In as-grown alumina, theoptical loss increases gradually from the center to thesurface of the film. Such a loss distribution is not seenin a film treated in hot water or glycerin. SEM obser-vations have revealed that pores in alumina films are ofa conical shape whose diameter varies from -100 to"-'200 A. By assuming a conical pore shape and residu-al aluminum microcolumns, a theoretical analysis wasachieved and experimental results were simulated suc-cessfully. In this manner, our study demonstratedthat the conditions of anodization and postanodiza-
100 150 tion treatment are important factors which determinethe optical loss of anodized alumina.
Hereafter we shall study more precisely the effectsof microstructure on various optical properties, whichwill help us not only to understand the features ofanodized alumina films but also to derive novel func-tions from the films.
This study was supported by a Grant-in-Aid forScientific Research (02750308) from the Ministry ofEducation, Science, and Culture and also by the Hoso-Bunka Foundation.
V i - References.....,...... ; 1. 0. Wiener, "Die Theorie des Mischk6rpers fur das Feld der
Plm) 2. Y. Arakawa and H. Sakaki, "Multidimensional Quantum WellLaser and Temperature Dependence of Its Threshold Current,"Appl. Phys. Lett. 40, 939-941 (1982).
(C) 3. R. K. Jain and R. C. Lind, "Degenerate Four-Wave Mixing inSemiconductor-Doped Glasses," J. Opt. Soc. Am. 73, 647-653(1983).
4. F. Hache, D. Ricard, and C. Flytzanis, "Optical Nonlinearities ofSmall Metal Particles: Surface-Mediated Resonance andQuantum Size Effects," J. Opt. Soc. Am. B 3, 1647-1655 (1986).
5. R. E. Collin, Field Theory of Guided Waves (McGraw-Hill, NewH York, 1960), Chap. 12.
6. M. Kitagawa and M. Tateda, "Form Birefringence of SiO2/Ta 2O5 Periodic Multilayers," Appl. Opt. 24, 3359-3362 (1985).
7. T. Yokogawa, M. Ogura, and T. Kajiwara, "ZnSe-ZnS Strained-. , ; Layer Superlattice Grown by Low Pressure Metalorganic Vapor
Phase Epitaxy Using Methylalkyls," Appl. Phys. Lett. 49,1702-50 100 150 1704 (1986).Pim) 8. K. Baba, K. Shiraishi, K. Obi, T. Kataoka, and S. Kawakami,
sses of alumina films. "Optical Properties of Very Thin Metal Films for Laminatedwhich are calculated by Polarizers," Appl. Opt. 27, 2554-2560 (1988).s calculated by taking 9. F. Keller, M. S. Hunter, and D. L. Robinson, "Structural Fea-otted lines are experi- tures of Oxide Coatings on Aluminum," J. Electrochem. Soc.and (c) correspond to 100, 411-419 (1953).tively. 10. M. Miyagi, Y. Hiratani, T. Taniguchi, and S. Nishida, "Alumi-
num Anodizing Technique for Fabricating Optical Thin-FilmWaveguides: a Proposal," Appl. Opt. 26, 970-971 (1987).
films more clearly, 11. M. Saito, M. Kirihara, T. Taniguchi, and M. Miyagi, "Micropo-concerning distri- larizer Made of the Anodized Alumina Film," Appl. Phys. Lett.dependence of loss 1.55, 607-609 (1989).
12. M. Saito and M. Miyagi, "Micropolarizer Using Anodized Alu-mina with Implanted Metallic Columns: Theoretical Analy-sis," Appl. Opt. 28, 3529-3533 (1989).
13. M. Saito and M. Miyagi, "Anisotropic Optical Loss and Birefrin-anodized alumina gence of Anodized Alumina Film," J. Opt. Soc. Am. A 6, 1895-38-gm wavelength. 1900 (1989).
14. H. Ginsberg and K. Wefers, "Zur Struktur der AnodischenDeckschichten auf Aluminiumoberfldchen," Metallurgie Berlin17, 202-209 (1963).
15. E. D. Palik, Ed. Handbook of Optical Constants of Solids (Aca-demic, Orlando, FL, 1985), Part 2.
16. G. Hass, "On the Preparation of Hard Oxide Films with Precise-ly Controlled Thickness on Evaporated Aluminum Mirrors," J.Opt. Soc. Am. 39, 532-540 (1949).
17. M. Born and E. Wolf, Principles of Optics (Pergamon, NewYork, 1985), Chap. 1.
18. K. Huber, "Pore Volume of Electrolytically Produced Protec-tive Coatings on Aluminum," J. Colloid Sci. 3, 197-206 (1948).
19. R. W. Franklin and D. J. Stirland, "Studies on the Structure ofAnodic Oxide Films on Aluminum, II," J. Electrochem. Soc. 110,262-267 (1963).
20. G. Paolini, M. Masoero, F. Sacchi, and M. Paganelli, "An Inves-tigation of Porous Anodic Oxide Films on Aluminum by Com-parative Adsorption, Gravimetric, and Electronoptical Mea-surements," J. Electrochem. Soc. 112, 32-38 (1965).
21. M. Nagayama and K. Tamura, "Dissolution of the Anodic OxideFilm on Aluminum in a Sulphuric Acid Solution," Electrochim.Acta 12, 1097-1107 (1967).
22. M. Okano, T. Taniguchi, M. Saito, and M. Miyagi, "ResidualAluminum and Its Dichroism in Anodized Alumina ThinFilms," Opt. Lett. 15, 253-254 (1990).
23. Tokyo Astronomical Observatory, Ed., Rika Nenpyo (Chrono-logical Scientific Tables), (Maruzen, Tokyo, 1987), p. 509, inJapanese.
