Jie
Gong‡
a,
William H.
Butler
a and
Giovanni
Zangari
*b
aCenter for Materials for Information Technology, University of Alabama, Tuscaloosa, AL 35487-0209
bDepartment of Materials Science and Engineering and CESE, University of Virginia, 395 McCormick Rd., P.O. Box 400745, Charlottesville, VA 22904-4745, USA. E-mail: gz3e@virginia.edu; Tel: +1-434-243-5474
First published on 18th March 2010
Free-standing, highly ordered porous aluminium oxide templates were fabricated by three-step anodization in oxalic, sulfuric or phosphoric acid solutions, followed by dissolution of the aluminium substrate in HgCl2. Opening of the pore bottoms on the barrier layer side of these templates was carried out by using chemical or ion beam etching. Chemical etching is capable of achieving full pore opening, but partial pore opening occurs inhomogeneously. On the contrary, ion beam etching enables homogeneous and reproducible partial pore opening, with the pore size controlled through the etching time. By this method, pore openings as small as 5 nm can reliably be obtained.
AAO has been known for more than 50 years,16 but only after 1995, when Masuda and Fukuda reported the synthesis of a highly ordered pore array by two-step anodization in oxalic acid,2 its potential in nanofabrication was appreciated and related research efforts flourished. The self-organized pore growth process, leading to a densely packed hexagonal pore structure, is controlled through the electrolyte chemistry, applied voltage, and processing time. Self-organized growth has been achieved in oxalic,2,17–19 sulfuric,17–20 and phosphoric acids,18,19,21,22 and mixed oxalic and sulfuric acid solutions.23 Interpore distance and pore diameter can be varied between 50–500 nm and 5–300 nm, respectively, depending on the electrolyte and anodization voltage employed;18,24 in addition, the latter can be widened by chemical etching in diluted acids. Long range ordering of the porous structure can be achieved using a two or three-step anodization process2,25 or by pre-texturing the initial aluminium surface via nanoimprint,26,27 nanoindentation,28 or ion beam prepatterning.29
Many applications of AAO in nanofabrication require opening of the pore bottom; this has been achieved by chemical etching in acidic solutions,30 ion beam31 or plasma etching.3 Complete removal of the pore bottom results in an opening with the same size as the channel; however, in some cases, close control over the diameter of the opening, below the channel diameter, may be required. This is particularly important, for example, in the fabrication of nanoconstrictions for magnetic point contacts32 or for the control of magnetic domain wall motion33 in spintronic applications, as well as in the synthesis of molecular membranes.34,35
Partial pore opening of oxalic acid anodized AAO by wet chemical etching has been recently reported;30 however, no detailed comparison of this process with ion beam etching has been carried out. Furthermore, although ion beam etching methods have demonstrated the uniform engraving of holes on the hemispherical shaped surface of the barrier layer with diameter as small as 10 nm using oxalic acid AAO,31 no such process has been reported for phosphoric or sulfuric acid AAO, which could produce larger or smaller nanoholes, respectively, due to the different size of the porous structures. In this work, free-standing sulfuric, oxalic and phosphoric acid AAO have been synthesized by a three-step anodization process, and the effects of chemical and ion beam etching on the morphology of sulfuric, oxalic and phosphoric acid AAOs with different pore sizes have been compared. It is demonstrated that uniformly engraved holes with diameter down to 5 nm can be reliably obtained using ion beam etching of sulfuric acid AAO. Fig. 1 depicts a schematic diagram of the barrier opening processes discussed in this paper, and provides a qualitative illustration of the resulting AAO nanostructures.
![]() | ||
Fig. 1 Schematic view of the fabrication process of free-standing porous AAO with the barrier layer perforated by wet chemical or ion beam etching. (a) AAO after anodization; (b) chemical post treatment to remove the metallic Al support; (c) free-standing AAO before etching; (d) uniform controllable openings on the barrier side achieved by ion beam etching; and (e) non-uniform openings on the barrier side obtained by wet chemical etching. d1: constriction/opening size; tb: barrier layer thickness; dp: pore diameter. |
Highly ordered AAO templates were prepared using a multi-step anodization process according to the following scheme:25
1. A polished Al sheet was anodized for 5–15 min to morphologically “texture” the Al surface.36
2. The resulting oxide film was dissolved in a solution containing 0.2 M H2CrO4 and 0.4 M H3PO4 at 60–70 °C for 5–20 min.
3. The substrate was re-anodized for more than 12 h to create long-range ordering.
4. Step 2 was repeated, with a duration of more than 30 min to dissolve the thick oxide film formed in step 3.
5. The substrate was finally anodized for a varying period of time, depending on the thickness of the AAO film desired.
A Veeco Microetch RF-1201 Ion Beam Etching System was employed for IBE. An inductively-coupled RF ion source was used, together with a Plasma Bridge Neutralizer (PBN) to keep the ion beam collimated. The ion beam etching chamber was first pumped down to 7–8 × 10−7 torr, then Ar gas flowed into the chamber at 20 sccm (standard cubic centimetre per minutes), keeping the chamber pressure constant at 1.32–1.34 × 10−4 torr. During etching, a 450 V beam voltage and 450 mA beam current were employed. Suppressor voltage was 300 V, and incident RF power was 300 W. PBN Ar gas flow rate was 3 sccm, and the ratio of neutralizer current/beam current (K factor) was 1.1. The ion beam incidence angle was 45°; the sample fixture was rotated at 0.1 revolutions per second and water cooled. The etching rate, calibrated using a Si wafer, was 40 nm min−1. When converted to Al2O3 using a standard milling rate table, the etching rate was about 9 nm min−1.
Anodization was carried out under potentiostatic conditions in H2SO4, oxalic acid ((COOH)2) or H3PO4 solutions. Voltages were chosen in the range of values yielding long range order.18 Typical current density–time (I–t) curves in H2SO4 for the first and second anodization are shown in Fig. 2; the same plots for oxalic and phosphoric acid solutions are reported in Fig. S2 of the ESI.† All curves show four stages of pore formation (Fig. 2, 1st anodization). In stage I, current drops abruptly due to the formation of the planar barrier oxide layer. In stage II, current gradually increases and after different time intervals (dependent on the solution and voltage applied) reaches its maximum between stages II and III. The increase in current is characteristic for porous AAO formation, indicating that field-enhanced dissolution starts to concentrate at locally convex surface regions, nucleating the pores.25 In stage III, the current decreases slightly to a constant value, indicating that the first initiated disordered pores are becoming hexagonally ordered, and stationary pore growth has been attained (stage IV).25
![]() | ||
Fig. 2 Typical anodization curves (current density vs. time, I–t) for two-step aluminium anodization in a sulfuric acid electrolyte (0.3 M H2SO4, 26 V, 3 °C). |
Comparison of the 1st anodization on the electropolished flat surface and the 2nd anodization on the “textured” surface in the three electrolyte systems shows that the pore nucleation process (stage II) is greatly shortened on the “textured” surface. In addition, the lowest current on the flat surface is always much lower than on the “textured” surface. This implies that “pre-texturing” the surface accelerates the ordered pore initiation process. For both the 1st and 2nd anodization, the duration of regime II follows the trend H3PO4 (185 V) > (COOH)2 (40 V) > H2SO4 (26 V), which corresponds to the difficulty of achieving ordering and the barrier layer thickness.
The fabrication process and porous structure of AAO were studied in detail in H2SO4, (COOH)2 and H3PO4 solutions. The process parameters and extracted geometrical characteristics are summarized in Table 1. Fig. 3(a) and (b) compare the topography, as determined by AFM, of the AAO surface after the 2nd and 3rd anodization in oxalic acid, respectively. Fast Fourier transform (FFT) patterns of the images, giving information about the structural periodicity in the inverse (momentum) space are shown in the corresponding insets. Both the AFM topography and the FFT patterns show that the number of defects decreases and the pore ordering improves after the 3rd anodization step. The diffraction ring and its broadening observed in the FFT of Fig. 3(a) indicate that 2D ordering of the structure is imperfect and that the interpore distance varies across the area of the sample. The barrier side topographic image (Fig. 3(c)), obtained by separation of the AAO membrane after the 3rd anodization step in saturated HgCl2, shows highly ordered, hexagonal patterns of hemispherical domes. The FFT images of the front and barrier side of AAO surfaces anodized three times consist of six equidistant and distinct spots, indicative of a perfectly ordered hexagonal pore lattice. Additional characterization and images of the AAO obtained in oxalic, sulfuric and phosphoric acid are available in the ESI.†
Process steps | H2SO4 (26 V, 3 °C) | (COOH)2 (40 V, 15 °C) | H3PO4 (185 V, 3 °C) |
---|---|---|---|
1st anod. time/min | 5 | 5 | 15–20 |
1st oxide strip time/min | 10–15 | 10–15 | >30 |
2nd anod. time/h | >12 | >12 | 16–20 |
2nd oxide strip time/min | 30–45 | 30–45 | 120 |
Growth rate/μm h−1 | 3.5 | 6 | 7.5 |
Interpore distance/nm | 60 | 108 | 450 ± 50 |
Pore density/cm−2 | ∼3 × 1010 | ∼1 × 1010 | ∼5 × 108 |
![]() | ||
Fig. 3 AFM surface topography of oxalic acid anodized AAO and the fast Fourier transform (FFT) patterns: (a) after 2nd anodization; (b) after 3rd anodization; (c) highly ordered hemispherical domes on the barrier side after dissolution of non-anodized aluminium. |
Similar to anodization on pre-textured Al substrates either by lithography22 or by nanoimprinting/nanoindentation,26–28 a pore array pattern with double periodicity can, in principle, be obtained by first anodizing in oxalic acid, then in sulfuric acid. The resulting morphology of the pore opening and barrier layer are shown in Fig. 4. Obviously, the pore-to-pore distance in the AAO obtained by anodization at a different voltage (25–27 V) in sulfuric acid does not match the pitch of the pattern obtained in oxalic acid at 40 V. Therefore, the pores growing in the second stage will not follow the pre-patterned pits and a distorted pore arrangement will occur. The anodic pores produced in H2SO4 have a much higher density than the pre-existing pits, and are mostly located along the pattern ridges, with each pattern cell containing nearly the same number of pores. It is also interesting to note that, if the anodization sequence is reversed, i.e. anodizing in sulfuric acid followed by oxalic acid, the final pattern shows the same features as after anodization in oxalic acid alone.
![]() | ||
Fig. 4 Double periodicity obtained by anodizing in oxalic acid (1st and 2nd step), followed by anodization in sulfuric acid (3rd step): (a) front side and (b) barrier side AFM surface topography, (c) SEM large area morphology. |
![]() | ||
Fig. 5 SEM surface morphology of the barrier side of oxalic acid-anodized AAO after different etching times in 5% H3PO4 at room temperature (25 °C): (a) 25 min; (b) 30 min; (c) 35 min; and (d) 40 min. The inset in (d) shows an enlarged view of the pore opening. |
As a comparison, the results of CE after various etching times for H2SO4-anodized AAO (Fig. 6) and H3PO4-anodized AAO (Fig. 7) are also shown. As expected, due to the different barrier layer thickness in the three cases, the times necessary to fully open the barrier caps are quite different. For H2SO4-anodized AAO, 20–25 min of etching is sufficient (Fig. 6(d)); H3PO4-anodized AAO on the other hand needs 240 min to fully open the pores (Fig. 7(b)). Again, CE can conveniently open the barrier layer to a full extent, but it does not allow accurate control of the size and location of the openings. As shown by AFM topography in Fig. 6(a1–d1), CE generates a rough and irregular barrier side, resulting in inhomogeneous etching, and thus hindering prospective integration with other components. It is also interesting to note that, if the H3PO4-anodized AAO is over-etched, the walls of the AAO channels are etched at a much slower rate than the backbone, probably due to preferential anion incorporation at walls,15 or to differences in chemical composition27,30 or crystallinity15 between the inner and outer region. As a result, alumina fibers can be produced, as also reported in ref. 37.
![]() | ||
Fig. 6 AFM (a1–d1) and SEM (a2–d2) images of the barrier side of sulfuric acid-anodized AAO membranes after different etching times in 5% H3PO4 at room temperature (25 °C): (a) 5 min; (b) 10 min; (c) 15 min; and (d) 20 min. The insets in (b2), (c2) and (d2) show enlarged views of the corresponding images. |
![]() | ||
Fig. 7 SEM micrographs of the barrier side of phosphoric acid-anodized AAO membranes after different etching times in 5% H3PO4 at room temperature (25 °C): (a) 220 min (left: tilted angle view; right: planar view; same scale bar); (b) 240 min (left and right image scans are at different magnifications); (c) 260 min; and (d) 285 min. |
Inhomogeneous pore etching by CE may be explained in terms of faster etching at defect sites (grain boundaries of the initial Al surface or boundaries of the ordered pore domains) or at regions with a thinner barrier layer. An initial, small difference in etching rate at separate locations may cause a local depletion of the etchant and preferential diffusion of the etchant from the bulk solutions to these regions, resulting in positive feedback. This effect may explain the clustering of opened pores observed in Fig. 6 (b2) (probably a domain boundary), or Fig. 6 (c2) (probably a region with locally thinner barrier layer).
The AFM topography of oxalic acid-anodized AAO barrier side after different IBE processing times is shown in Fig. 8. Uniform and round holes (diameter 15–30 nm) can be obtained over a large area when the etching time is limited to 5–10 min (Fig. 8(b) and (c)). Round hemispherical caps are uniformly milled from the top due to the uniform flux distribution of the ion beam. Chemical composition15 or stress38 gradients that may be present in the barrier layer at the pore bottom are of no consequence, since the energy differences involved due to these gradients are much smaller than the ion beam energy. Long-time etching leads to the disappearance of the protruding caps and results in holes with shape and diameter similar to the nanopores on the opposite side (Fig. 8(d)). This indicates that the hemispherical barrier layer domes are completely etched away by the ion beam.
![]() | ||
Fig. 8 AFM surface topography of the barrier side of oxalic acid-anodized AAO membranes after different ion beam etching times. (a) no etching; (b) 5 min (inset shows the enlarged 3D AFM image of a single cap); (c) 10 min; and (d) 15 min. |
Much smaller constrictions can be obtained by IBE from the barrier layer side of H2SO4-anodized AAO in shorter times. This is due to the fact that the pore size and the barrier thickness of these AAO samples are smaller. The morphologies resulting from the etching process are shown in Fig. 9 and 10. As shown in Fig. 9, very uniform small openings over a large area can be engraved on the barrier side. The opening size can be accurately controlled down to about 5–8 nm by limiting the etching time to 2–4 min. Increasing etching time results in increased opening size up to more than 18 nm, as well as a flattened surface topography (Fig. 10).
![]() | ||
Fig. 9 AFM surface topography of the barrier side of sulfuric acid-anodized AAO after different ion beam etching times: (a) 2 min; (b) 4 min; (c) 8 min; and (d) 12 min. Insets in (a)–(d) show corresponding enlarged AFM images. |
![]() | ||
Fig. 10 AFM surface topography (top) and profiles (bottom) of the barrier side of sulfuric acid-anodized AAO membranes after different ion beam etching times: (a) no etching, top height 8.3 nm; (b) 2 min, 3.9 nm; (c) 4 min, 3.1 nm; (d) 8 min, 2.7 nm; (e) 10 min, 2.1 nm; and (d) 12 min, 0 nm, the surface is completely flattened by ion beam etching. |
IBE of H3PO4-anodized AAO produces larger openings, ranging from 100 to 200 nm, as shown in Fig. 11. The etching time necessary to open the hemispherical domes is much longer now since the barrier layer is thicker in these samples. Although every dome in the images can be opened, the shape and size of the openings are not uniform on the same sample. This is probably a consequence of the fact that pores and barrier domes on phosphoric acid-anodized AAO are not as ordered as those obtained using the other acids. As shown in Fig. 11(c) and (d), triangular, trapezoidal, square and round openings can be found.
![]() | ||
Fig. 11 AFM (a1–d1) and SEM (a2–d2) images of the barrier side of phosphoric acid-anodized AAO membranes after different ion beam etching times: (a) no etching; (b) 50 min; (c) 80 min; and (d) 100 min. Insets in (a2)–(d2) show tilted angle SEM views (same scale bar). |
The constriction size, defined as the full width at half depth of the AFM topographic scans across the barrier side openings, produced by ion beam etching on the three types of AAOs are extracted and summarized in Fig. 12. A broad range of constriction sizes can be obtained; in addition, the size distribution is narrow, with the exception of the H3PO4-anodized AAO. IBE processing of AAO, therefore, provides ideal templates to grow nanoconstrictions with closely controlled geometry, or to fabricate molecular membranes with closely controlled pore size in the deep sub-μm range.
Footnotes |
† Electronic supplementary information (ESI) available: Current transients during anodization, SEM and AFM images of aluminium oxide anodized in sulfuric acid and phosphoric acid. See DOI: 10.1039/c0nr00055h |
‡ Now at Seagate Technology, Research and Technology Development, 7801 Computer Avenue South, Bloomington MN 55435 |
This journal is © The Royal Society of Chemistry 2010 |