Preparation of transparent anodic alumina with ordered nanochannels by through-thickness anodic oxidation of aluminum sheet

Materials Chemistry and Physics 104 (2007) 306–311

Preparation of transparent anodic alumina with ordered nanochannels by through-thickness anodic oxidation of aluminum sheet M. Mehmood ∗ , A. Rauf, M.A. Rasheed, S. Saeed, J.I. Akhter 1 , J. Ahmad, M. Aslam National Centre for Nanotechnology, Pakistan Institute of Engineering and Applied Sciences (PIEAS), Islamabad 45650, Pakistan Received 31 May 2006; received in revised form 16 February 2007; accepted 16 March 2007

Abstract Through-thickness anodic oxidation of aluminum sheet has been possible to form 100–400 ␮m thick transparent alumina in 0.3 M oxalic acid at 30–70 V. Transmission at 900–1100 nm was 20–60%, the higher being at lower anodizing voltages. Two-step anodizing after electropolishing in Brytal solution resulted in best hexagonal order at 50 V. The alumina layers with ordered porosity growing from the opposite faces met at the middle of the sample without changing the orientation of the pores, except at the last stage when a number of pores meet to form Y-, S- and I-type junctions. Three-dimensional compressive stresses around aluminum nanowires (nanoparticles) seem to be responsible for re-orienting the pores. In case of thicker barrier layers formed at higher anodizing voltage, the pores growing from opposite side do not meet and are separated by a barrier type compact oxide at the middle of the sample. © 2007 Elsevier B.V. All rights reserved. Keywords: Ordered porosity; Nanochannels; Anodic alumina; Transparency; Interface

1. Introduction Nanostructured materials exhibit interesting properties in a wide range of spectrum including catalytic activity [1], corrosion resistance [2], optical properties [3] and magnetic properties [4], etc. Since 1995 [5], nanoporous anodic alumina has significantly attracted the attention of scientists and researchers. This is because of the self-organization of vertical (cylindrical) pores to form hexagonal array that has provided controlled and narrow distribution of pore diameters and inter-pore distances in addition to the possibility of forming the pores with extremely high aspect ratio [5–9]. This has opened a wide range of applications of anodic alumina, particularly as a template to form nanowires, nanotubes, nanodots, and composites for catalysis, emitters, rechargeable batteries, magnetic storage devices, etc. [10–16]. The superlattice of pores and nanowires in anodic alumina has also been found extremely useful to exploit and study magnetic interactions, dielectric properties, and optical interference [17–19]. The other useful properties are photo- and electro-luminescence, and transparency with absorption wavelengths higher than corundum [20]. ∗ 1

Corresponding author. Tel.: +92 519290273; fax: +92 519223727. E-mail address: [email protected] (M. Mehmood). Address: PINSTECH, PO Nilore, Islamabad, Pakistan.

0254-0584/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.matchemphys.2007.03.019

The anodic porous alumina is composed of two layers, i.e., the top thick layer with vertical pores and a barrier layer that is in turn supported by aluminum [21]. For obtaining selfstanding alumite with ordered porosity, anodizing has always been performed from one side of aluminum substrate; various post-treatments have been necessary to dissolve remaining aluminum with or without the barrier layer. However, the chemicals used are either hazardous to the environment or aggressive to alumina. Hence, special care is required, which may be cumbersome and sometimes less reliable [22–26]. Therefore, obtaining alumite with ordered porosity by anodizing aluminum sheet from both sides may be interesting. This may reduce the time required for anodizing to half in comparison to the case, in which, anodizing is performed from one side of the aluminum substrate. In addition, it may be of fundamental interest to observe the structure formed when the two alumite layers growing from opposite sides meet each other. Investigations have also been performed to determine the degree of ordering as a function of anodizing voltage. 2. Experimental High purity aluminum sheet (99.99%, 0.5 mm thick) was used as the starting material. C2 H2 O4 ·2H2 O (Riedel, 97.5%), Na2 CO3 (Panreac, 99.5%), Na3 PO4 12H2 O (Riedel, 98%), CrO3 ·2H2 O (Merk, 99%), and H3 PO4 (BDH, 98%) were purchased from commercial resources and used without further processing.

M. Mehmood et al. / Materials Chemistry and Physics 104 (2007) 306–311 After annealing at 500 ◦ C for more than 100 h, the aluminum samples were degreased in acetone ultrasonically for 15 min. Electropolishing and anodizing were performed using Stabilized Power Supply (FARNELL, TSV70 MK.2) with two electrode configuration; the counter electrode being a platinum plate. Electropolishing was carried out in Brytal solution (15 wt.% Na2 CO3 and 5 wt.% Na3 PO4 ) at 80 ◦ C at a voltage of 2 V for 1 h or more, which resulted in a smooth surface. The electropolished samples were anodized at constant voltage, ranging from 30 to 70 V, in 0.3 M oxalic acid at 0–1 ◦ C. First anodizing was performed for about 3–12 h. The resulting anodic oxide was selectively dissolved in a mixture of chromic acid (0.2 M CrO3 ·H2 O) and phosphoric acid (0.4 M H3 PO4 ) at 70 ◦ C for 3 h. Then second anodizing was performed under the same conditions as the first anodizing until a transparent region appeared indicating through-thickness oxidation over about 0.5–1 cm2 area of the sample for necessary measurements. The samples were characterized using Scanning Electron Microscope (SEM, LEO-441-I, UK), Field Emission Scanning Electron Microscope (FE-SEM, Keck, LEO 15150) and UV–visible spectrometer.

3. Results and discussion Fig. 1 shows typical top view SEM images of the surfaces of anodized samples prepared at different voltages. The pores tend to arrange in a hexagonal pattern forming the domains of long-range order in case of anodizing at 40 and 50 V, as

307

shown in Fig. 1(a) and (b). It may be noticed, however, that the ordering is better at 50 V than 40 V, as revealed by the parallel arrangement of hexagonal cells, larger domain size and narrow domain boundaries in Fig. 1(b). The sample prepared at a still higher anodizing voltage, i.e., 60 V (Fig. 1(c)), exhibits mixed character and ordered domains are dispersed in the order-less matrix. The ordered domains cover approximately 40% area. The order is totally lost accompanied by a wide dispersion in pore sizes in case of the sample prepared at 70 V, as shown in Fig. 1(d). Fig. 2(a) shows cell size as a function of anodizing voltage. The average cell size is approximately 103, 123–125, and 169 nm after anodizing at 40, 50 and 60 V, respectively. The ratio of cell size to anodizing voltage is shown in Fig. 2(b). The ratio is minimum at 50 V. This behavior has also been confirmed at 10 ◦ C for the anodizing voltage of 40 and 50 V. It seems worthmentioning here that the order was also better at 50 V both in the case of 10 and 0 ◦ C, as partly shown in Fig. 1. It is known that elastic strains (or stresses) develop at oxide–metal interface due to volume expansion during oxidation. The oxide–metal interface has protrusion below the pore tips. This stress distributes in such a way that it concentrates

Fig. 1. SEM images of the surfaces, after preparing the samples at: (a) 40 V, (b) 50 V, (c) 60 V and (d) 70 V; the temperature was ∼0 ◦ C.

308

M. Mehmood et al. / Materials Chemistry and Physics 104 (2007) 306–311

Fig. 2. Hexagonal cell size (a), and ratio of cell size to anodizing voltage (b), as a function of anodizing voltage.

below the pore tips and relaxed below the pore wall [27]. These regions of high stress (below the pore tips) seem to repel each other depending on the magnitude of stress associated with the individual pore and the distance between them, i.e., cell size. This repulsive force should be proportional to the strain associated with the individual stress/strain center at the pore tip which in turn proportional to the barrier layer thickness and thus the anodizing voltage. On the other hand the repulsive force should be strengthened by decrease in the cell size, i.e., the distance between the stress centers. Therefore, the repulsive force should increase with a decrease in the ratio of cell size to the anodizing voltage. We consider that the repulsive forces are responsible for the hexagonal ordering, and the best order should be attained when we have the strongest repulsive forces, which should be attained at the lowest ratio of the cell size to the anodizing voltage. It appears, therefore, that the best self-ordering at 50 V in comparison with other anodizing voltages is attributable to stronger interfacial strain interactions as a result of smaller ratio of interpore distance to anodizing voltage. Furthermore, electropolishing in Brytal solution seems to produce more conducive conditions for better self-ordering at 50 V in 0.3 M oxalic acid, in contrast to perchlorate solution that is known to favor 40 V for the best ordering under similar anodizing conditions. The exact reason for the role of electropolisihing solution is under investigation. The samples were exposed to the radiations in the wavelength range of 900–1100 nm for estimating transmission through the samples. Fig. 3 shows the percentage transmission, as a function of anodizing voltage. Transmission has been possible between 20 and 60%, although it varies with anodizing voltages. Aluminum is known to be optically very active metal for extremely high absorbance. Therefore, the presence of extremely thin aluminum film (a few nanometers) should be very effective in reducing the transmission. Accordingly, the aluminum has been mostly consumed through the thickness, although an extremely thin aluminum film (few nanometer-thick) may still be present resulting in some loss of transmission. Decrease in transmission with an increase in anodizing voltage may suggest that more efficient consumption of aluminum takes place at smaller anodizing voltages.

Typical cross-sectional views of the sample anodized at 50 V have been shown in Fig. 4. As shown in Fig. 4(a), almost 100 ␮m thick alumite layers have grown from opposite sides to meet each other at the middle of the sample. Aluminum has been apparently consumed completely through the thickness except at the islands, which exhibit thin unconsumed aluminum. The loss in transmission, as mentioned in Fig. 3 may partly be attributable to these islands. Fig. 4(b) shows the cross-section at higher magnification. The terraces in the image show longitudinal view of pores, while the edges exhibit the cross-sectional view. Uniformity of the inter-pore spacing and extremely parallel arrangement of high aspect ratio pores is clearly evident. Fig. 4(c) is a high magnification image of the rectangular region highlighted in Fig. 4(a). The position of the interface formed between the alumite layers have been marked in the figure. It may be noticed that pores remain extremely parallel even close to the last stage of anodization. Fig. 4(d) and (e) present two regions selected from Fig. 4(c) to show the interface at larger magnification. The parallel pores growing from both sides tend to join the pores of the other side and a change in the orientation of verti-

Fig. 3. Average percentage transmission of light at a wavelength range of 900–1100 nm through the transparent regions of the sample anodized for through-thickness oxidation.

M. Mehmood et al. / Materials Chemistry and Physics 104 (2007) 306–311

309

Fig. 4. (a–e) Typical cross-sectional views of a sample prepared at 50 V; (a and c–e) The middle position of the section in order to reveal the interface of the meeting sides of the alumite, while (b) focuses the longitudinal and cross-sectional view of the pores.

cally growing pores in order to facilitate this joining can also be seen in Fig. 5(d). A few pores from one side are joining with two pores on the opposite side forming Y-junctions. Such pores have been explored and found very interesting for preparing physical nanodevices [10,14,17]. In contrast to Fig. 4(d), a few aluminum particles (or the sections of aluminum wires) are still present in Fig. 4(e). This region is closer to the island of unconsumed aluminum. Accordingly, aluminum is consumed gradually and it forms particles (or wires) at the last stage, which are also consumed as the growth of the continuous oxide proceeds. The continuity of a few pores in proximity to the aluminum particle can also be seen indicating that the pore joining occurs before complete consumption of aluminum, i.e., as a result of anodic oxidation. Furthermore, I-type pore junctions can also be seen in Fig. 5(e). These junctions have been formed between the pores already heading towards each other. The formation of numerous Y-junctions, at a specified location with same diameter of the pores has not been reported earlier, as far as our knowledge is concerned. Using such pores

Fig. 5. A schematic diagram of the way the pores from the opposite sides tend to meet each other depending on their relative position (a number of pores from opposite sides join each other to form Y-type, S-type and I-type junctions.).

310

M. Mehmood et al. / Materials Chemistry and Physics 104 (2007) 306–311

As far as our knowledge is concerned, the formation of an interface between alumite layers grown from opposite sides has not been explored before. The porous layers joining each other in this way may be explored for physical devices based on filling the pores with a material of choice to form, e.g., Y-junctions with equal diameter of wires; the wires separated at the end by a dielectric material, i.e., compact (barrier layer of) alumina depending on the preparation conditions of the alumite. 4. Conclusions

Fig. 6. FESEM cross sectional view of the sample prepared at 60 V, showing two alumite layers grown from opposite sides separated by barrier layer (a few rounded white spots at the middle of the barrier layer, as indicated by arrow) may be particles of unconsumed aluminum).

as template can extend the applications of porous alumina for making nano-devices. The growth of pores and their ordering is believed to depend on complex interaction of dissolution and formation conditions, non-uniform electric filed across the barrier layer, and stress state in the barrier layer and the underlying aluminum [27–29]. Aluminum with an inhomogeneous thickness due to protrusions below the pores tips is finally converted to a mesh of thin wires at the thicker sections of aluminum below the cell walls. The arrangement of these wires (or the nanoparticles when viewed along a given cross-section) should be complex because two different hexagonal cells of the opposite sides meet each other with various degrees of coherency from one position to another, as schematically shown in Fig. 5(a) and (b). When these wires (or particles) are oxidized by anodizing, the three-dimensional stress state, electric field distribution, formation and dissolution conditions of alumina must be totally different from that on the bulk aluminum due to change in geometrical features at the alumina–aluminum interface, which may alter the path of the pore tips. In our opinion, the expansion around the aluminum nanoparticles due to the conversion of aluminum to alumina seems to exert the compressive stresses that push the barrier layer. Thus, a pore tip changes its path when it is heading towards an aluminum particle; otherwise it keeps itself on a straight path, as schematically shown in Fig. 5(a) and (b), respectively. The former phenomenon results in Y-type or S-type junctions and the later one favors the I-junction (straight) joints, as indicated in Fig. 4(d) and (e). Fig. 6 shows a cross-sectional view of the samples prepared at anodizing voltage of 60 V. The barrier layer is still present at the middle of the section. This suggests that thicker barrier layer remain intact and is not pierced-through by the pores. A few particles of aluminum can be seen sandwiched between the barrier layers of the opposite side. Although not revealed by SEM, the presence of an extremely thin layer of aluminum cannot be ruled out in other parts of the barrier layer, as well.

(i) Anodizing of aluminum sheet from opposite sides has been performed in oxalic acid to achieve through-thickness oxidation to form 100–400 ␮m thick transparent alumina. Transmission in the near infrared range, i.e., 900–1100 nm, ranges between 20 and 60%. The higher transmission is achieved at lower voltages, which may be due to better consumption of aluminum at lower voltages. (ii) Electropolishing at 2 V in Brytal solution at 80 ◦ C after heat treatment and subsequent two-step anodizing in oxalic acid has resulted in the best hexagonal order at 50 V. The pores grow straight from opposite sides and meet at middle of the sample. A number of pores join to form Y-type, S-type and I-type junctions. The three-dimensional compressive stress state around the nanoparticles, at the last stages of anodizing, seems responsible for pushing the pore tips away and diverting their path to ensure smooth joining. (iii) In case of thicker barrier layers at higher anodizing voltage, e.g., 60 V, the pores do not meet and are separated by a barrier type oxide. Acknowledgements The authors are grateful to Prof. Hiroki Habazaki (Hokkaido University) for useful comments and suggestions. References [1] H. Habazaki, M. Yamasaki, A. Kawashima, K. Hashimoto, Appl. Organomet. Chem. 14 (2000) 803–808. [2] M. Mehmood, B.P. Zhang, E. Akiyama, H. Habazaki, A. Kawashima, K. Asami, K. Hashimoto, Corros. Sci. 40 (1998) 1–17. [3] A.G. Cullis, L.T. Canham, P.D.J. Calcott, J. Appl. Phys. 82 (1997) 909–965. [4] G. Herzer, IEEE Trans. Magn. 25 (1989) 3327–3329. [5] H. Masuda, K. Fukuda, Science 268 (1995) 1466–1468. [6] H. Masuda, F. Hasegwa, S. Ono, J. Electrochem. Soc. 144 (1997) L127–L130. [7] H. Masuda, K. Yada, A. Osaka, Jpn. J. Appl. Phys. Part 2: Lett. 37 (1998) L1340–L1342. [8] H. Masuda, M. Satoh, Jpn. J. Appl. Phys. Part 2: Lett. 35 (1996) L126–L129. [9] T. Ohmori, T. Kimura, H. Masuda, J. Electrochem. Soc. 144 (1997) 1286–1288. [10] T. Gao, G. Meng, J. Zhang, S. Sun, L. Zhang, Appl. Phys. A: Mater. Sci. Process. 74 (2002) 403–406. [11] J. Li, C. Papadopoulos, J. Xu, Nature 402 (1999) 253–254. [12] J.Y. Liang, H. Chik, J. Xu, IEEE J. Selected Topics Quant. Electron. 8 (2002) 998–1008. [13] K. Nielsch, F. Muller, A.P. Li, U. Gosele, Adv. Mater. 12 (2000) 582–586. [14] Y.T. Tian, G.W. Meng, S.K. Biswas, P.M. Ajayan, S.H. Sun, L.D. Zhang, Appl. Phys. Lett. 85 (2004) 967–969. [15] K. Kim, M. Kim, S.M. Cho, Mater. Chem. Phys. 96 (2006) 278–282.

M. Mehmood et al. / Materials Chemistry and Physics 104 (2007) 306–311 [16] [17] [18] [19] [20] [21] [22]

A. Saedi, M. Ghorbani, Mater. Chem. Phys. 91 (2005) 417–423. H. Chik, J.M. Xu, Mater. Sci. Eng. R: Rep. 43 (2004) 103–138. C.R. Martin, Science 266 (1994) 1961–1966. C.R. Martin, Chem. Mater. 8 (1996) 1739–1746. S. Tajima, Electrochim. Acta 22 (1977) 995–1011. J.W. Diggle, T.C. Downie, C.W. Goulding, Chem. Rev. 69 (1969) 365–405. G.Q. Ding, M.J. Zheng, W.L. Xu, W.Z. Shen, Nanotechnology 16 (2005) 1285–1289. [23] X. Mei, D. Kim, H.E. Ruda, Q.X. Guo, Appl. Phys. Lett. 81 (2002) 361–363.

311

[24] G.R. Xu, F.L. Ren, S.H. Si, Q.F. Yi, Acta Physico-Chim. Sinica 22 (2006) 341–344. [25] T.T. Xu, R.D. Piner, R.S. Ruoff, Langmuir 19 (2003) 1443–1445. [26] Y.C. Zhao, M. Chen, Y.N. Zhang, T. Xu, W.M. Liu, Mater. Lett. 59 (2005) 40–43. [27] O. Jessensky, F. Muller, U. Gosele, Appl. Phys. Lett. 72 (1998) 1173– 1175. [28] F.Y. Li, L. Zhang, R.M. Metzger, Chem. Mater. 10 (1998) 2470–2480. [29] V.P. Parkhutik, V.I. Shershulsky, J. Phys. D: Appl. Phys. 25 (1992) 1258–1263.