Use of Ionic Liquid in Fabrication, Characterization, and Processing of Anodic Porous Alumina
Two different ionic liquids have been tested in the electrochemical fabrication of anodic porous alumina in an aqueous solution of oxalic acid. It was found that during galvanostatic anodization of the aluminum at a current density of 200 mA/cm2, addition of 0.5% relative volume concentration of 1-butyl-3-methylimidazolium tetrafluoborate resulted in a three-fold increase of the growth rate, as compared to the bare acidic solution with the same acid concentration. This ionic liquid was also used successfully for an assessment of the wettability of the outer surface of the alumina, by means of liquid contact angle measurements. The results have been discussed and interpreted with the aid of atomic force microscopy. The observed wetting property allowed to use the ionic liquid for protection of the pores during a test removal of the oxide barrier layer.
KeywordsPorous alumina Anodization Galvanostatic Ionic liquids Wettability Roughness
Anodic porous alumina (APA [1–4]), also called porous anodic alumina (PAA) [5–9]), anodic aluminum oxide (AAO [10, 11]), or alumite [12, 13], is a form of Al2O3, which is deposited onto an aluminum (Al) foil working as the positively biased pole of an electrolytic cell [14–16]. Whereas in basic or neutral electrolytes (ELs) a compact alumina layer is grown, called “barrier” layer, in acidic ELs that can dissolve the oxide a porous alumina layer is grown, on top of a thin (10–100 nm ) barrier-type layer. Depending on the applications, pore ordering in APA can be necessary either on both sides such as for photonic crystals made by using APA as a lithographic mask , or only on one side such as for in situ photonic crystals made by incorporating materials on one APA surface [19–21], or on none of the two sides such as for filtering membranes [22–24], biosensor electrodes with enhanced surface area , and templates for the growth of separated metal nanowires [13, 25, 26], or supported oxide or polymer nanotubes [27, 28]. However, for several applications, it is desirable that the APA thickness h is comparatively high, h ≥ 100 μm. This can give the film the required robustness for use as either a standalone membrane in case of, e.g., battery separator  or lithographic etching mask [18, 29], or the required high aspect ratio (a.r. = h/d, where d is the pore diameter) when using the layer as a template for, e.g., nanowire electrodeposition (where a.r. ≥ 1,000 can be required [21, 30]). However, the film growth rate vg reported so far for conventional mild anodization (MA) of Al has been in the order of 0.03–0.1 μm/min , which makes the growth of thick APA quite time consuming. This slow growth hinders the advantage of the relatively cheap setup for the fabrication of APA, discouraging its use for both the development of prototype structures in the research academy, and for possible industrial fabrication processes. The current interest in speeding up the APA growth has recently found a viable way in the application of a single-step process combining MA with industrial hard anodization (HA) conditions , made possible by the protecting oxide layer grown during the preliminary MA phase.
Another possible route for fast APA fabrication could be the identification of proper additives, which can conveniently change the environmental conditions for anodization. Ionic liquids (ILs) are a class of solvents that have recently attracted a renewed interest as chemical additives in a number of reactions [32, 33]. To our knowledge, the only use of APA and IL system has been so far as an additive in the fabrication of cobalt nanowires inside an APA template . In this work, we report on the use of ILs in the fabrication and characterization of APA, starting from a well-known APA fabrication EL such as oxalic acid. The idea behind using IL additives in this system is that the convective flow of the IL component species can facilitate the displacement of the EL ions useful for anodization, and avoid formation of strong temperature gradients between the anode and the beaker walls, which are in direct contact with the refrigerating bath.
Materials and Methods
We used 0.25-mm thick foils of polycrystalline Al (Goodfellows, 99.999% purity). The foils were cut with scissors into rectangular pieces of single face areaS ~ 15 × 3 mm2, and flattened back to roughly planar surface, after scissors curling, by pressing each of them between two new glass slides.
Degreasing was performed by hand brushing (>10 s) with acetone-wet lens paper, 3 min sonicating in warm (60 °C) acetone, rinsing in de-ionized (DI) water, sonicating another 3 min in warm (60 °C) ethanol, and thoroughly washing (>30 s) in running DI water.
After degreasing, an electropolishing (EP) step was performed on the Al foil, which was partly dipped in a 250-mL beaker filled up to 200 mL with a 1:5 v/v HClO4:C2H5OH mixture, and kept inside a refrigerating bath set atTbath =+7 °C. The cathode was a Pt plate, also partly dipped in the EL, kept at a gap distance ofg ~ 11 mm from the Al anode. The process was run for 7 min without stirring, at constant currentiEP.
In our setup, both sides of the dipped Al foil come into contact with the EL, the anodic contact being provided from the top, outside the EL. The dipped single face surface area for EP wasSEP ~ 12 × 3 mm2, and as a result the constant current density wasJEP = iEP/2SEP ~ 170 mA/cm2. During the process, hydrogen gas evolution could be observed at the cathode. The final Al surface looked mirror-like.
Since in our setup immersion of the beaker in the refrigerating bath was not compatible with stirring onto a magnetic plate, the temperature close by the Al anode was probably higher thanTbath(~+10 °C difference has been measured in several cases). We tried to minimize this effect during anodization as compared to EP, by keeping the anode as close as possible to the external temperature controlled bath, using in this case a much smaller beaker (50 mL, filled up to 30 mL).
As the starting anodization EL, we chose an aqueous (DI water) solution of oxalic acid ((COOH)2, Sigma-Aldrich, Italy), and decided to run this process as well as the EP atTbath =+7 °C without stirring. We have only run single anodization processes, and considered the inner APA surface (in contact with the Al substrate) as the test surface for the layer quality, that is the regular pore arrangement. The outer APA surface (in contact with the EL) has been checked as well, soon after anodization, for an estimation of the outer pore mouth diameterdout.
The anodizations were all run in galvanostatic mode, changing as a parameter the anodization currenti and so the current densityJ = i/2Sanod, whereSanodis the dipped anode surface. AnSanod ~ 10 × 3 mm2was used, significantly smaller thanSEP, to be sure to anodize only electropolished Al surface, and to avoid that possible side effects occurring at the ambient air–EL meniscus insist on the same region during the two consecutive processing steps, (namely EP and anodization).
As ILs to test we chose two different commercially available room temperature (RT) water-soluble ILs, namely 1-butyl-3-methylimidazolium 2-(2-methoxyethoxy) ethyl sulfate (C13H26N2O6S, “IL1”) and 1-butyl-3-methylimidazolium tetrafluoborate (C8H15BF4N2, “IL2”), both from Sigma-Aldrich. The choice was driven by the former being particularly rich in oxygen, possibly taking place in the anodization reaction in spite of the oxalic acid and/or water oxygen, and the latter being very easily soluble in water.
The APA outer and inner surfaces were imaged by means of atomic force microscopy (AFM) with a MFP-3D instrument (Asylum Research, USA), operating in Tapping mode with gold-coated silicon cantilever probes NSG10 (NT-MDT, Russia). The probes had a nominal resonance frequency ~250 kHz and standard tip (apex diameter ~10 nm, aspect ratio ~2.6). Apart from surface quality inspection, the AFM images of the outer APA surface have also been used for quantitative determination of the sample roughness by means of the root mean square (RMS) of the distribution of sample features height. The RMS values of at least three AFM images acquired in different regions with 10 μm scan size have been averaged for each APA sample.
The APA top surface wettability was measured with different solvents by means of sessile drop method, using a DataPhysics OCAH 200 at laboratory conditions (temperature 17–20 °C, relative humidity 40–60%). Droplets of ~1 μL volume (drop diameter ~1 mm) have been used in all cases. Similarly to the RMS measurements, for the liquid contact angles (CAs), the values on at least three different regions on each sample have been averaged as well. For both used liquids and especially for the IL2 solution, the CAs were soon decreasing in time after touching the APA surface. The reported values have been measured immediately after contact (t < 5 s).
For removal of the Al substrate, which was necessary to determine the APA thickness h, we adopted a non-standard technique. The reason was that in our samples the Al was sandwiched between two adjacent APA films, and in these conditions, we found the standard dissolution in either saturated HgCl2 [1, 4, 6, 34–38] or CuCl2 [7, 8] not to be so effective as for one-side APA films on Al. Therefore, we decided to run a second EP-like process in much harder conditions than during the Al smoothing step, namely at RT and current density ~10 JEP. The hard Al etching was accompanied by a strong hydrogen bubbling at the cathode and by a typical noise, while quasi-periodic (1–2 Hz) Al spitting off between the two APA films was visible through the beaker walls. We stopped the process when we could see that a significant loss of Al had already occurred at the bottom of the dipped sample, sufficient for optical inspection (typically after a time t ~ 30 s). The APA thickness h was then measured by optical micrographs acquired in reflection perpendicular to the film sections, with a ± 1 μm resolution uncertainty.
Results and Discussion
Anodization in Bare Oxalic Acid
Independent of the anodization mode, if either potentiostatic or galvanostatic, in the steady state both i and V should be constant over the process time t, as for ionic conduction it is i ~ eαV, with α an appropriate constant [5, 14]. However, V(t) is linearly increasing in Fig. 1a after the initial transients, with different rates increasing in turn with J. This behavior shows that we were not in a condition of equilibrium for the different anodization reactions. The reason can be that in our setup immersion of the EL beaker in the cooling bath was not compatible with stirring onto a magnetic plate. Therefore, a local depletion of ions in the EL close to the anode, along with formation of a stable ion concentration gradient, can have occurred over time, which leads to a progressive increase in V in order for the power supply to keep J constant.
The h values obtained at tend = 30 min as a result of the corresponding anodizations have been plotted in Fig. 1b. If the same current efficiency of the process was maintained in all the anodizations and for all the process time period, a linear relationship between J and h was expected [5, 22]. However, after the initial increase in h with increasing J a tendency to saturation is observed in Fig. 1b. Obviously, a progressive reduction of the current efficiency has occurred, probably due to the appearance of side reactions different from the anodization ones .
Actually, side reactions can also occasionally lead to catastrophic events, such as shown in Fig. 1a for the curve at J = 200 mA/cm2. In that case after reaching a critical value Vcrit ~ 95 V in tcrit ~ 12 min, V started to decrease with some fluctuations, and finally increased up to the maximum power supply voltage. The reason for the latter increase was that the Al foil was cut at the air–EL meniscus, and the piece of anode dipped in the EL fell on the bottom of the beaker, opening the circuit. The APA thickness measured for this sample was hcrit = 11 μm, such that the respective critical electric field was Ecrit = Vcrit/hcrit ~ 8.6 MV/m. This is ~36% lower than the dielectric strength of compact alumina, Ebreak = 13.4 MV/m . Whereas this could be partly due to the porous nature of our alumina, we do not think that the origin of this behavior is the dielectric breakdown of the oxide due to the high V reached. Instead, we assign the discontinuity in the curve to a temperature activated fast etching of the Al at the air–EL meniscus, where on fluctuations of the interface the bare Al can locally come into contact with the EL. In fact, when using a 10-fold diluted oxalic acid we could reach a Vcrit ~ 160 V before that any similar catastrophic event occurred, with a respective hcrit ~ 13 μm, which gives Ecrit ~ 12.3 MV/m, much closer to Ebreak. Furthermore, in the latter case, the final steep change of voltage was toward the zero (closed circuit with virtually no resistance), and there was no anode resection of the air–EL meniscus.
The limited improvement inh obtained in the considered time periodtendon increasingJ in bare oxalic acid resulted in a maximum (non-linear) mean growth ratevgmax = hcrit/tcrit ~ 0.83 μm/min (obtained forJ = 200 mA/cm2).
All the quantities in Fig. 2a have been plotted versus the anodization current density J. One can see that dout is approximately constant within the errors, as expected, since it should depend only on the type of EL and on its concentration. The weak increase actually observed can be due to a local rise of the EL temperature and thus of the oxide dissolution rate, probably occurring during anodization. On the contrary, Dout is clearly increasing with J. Indeed, Dout should increase with V [5, 14, 17], and our V(t) curves in Fig. 1a showed a V that increased during each anodization. As a consequence, the roughness is almost constant for the higher J values (i.e., the larger Dout), whereas it is significantly depressed for the lowest J (i.e., the smallest Dout). We attribute this effect to the spatial “low-pass” characteristic of the AFM probe tip, which can hardly penetrate the smaller pores and thus senses the respective APA as an almost continuous, smooth surface.
Anodizations in Oxalic Acid–IL Solutions
The general effect of addition of an IL in the EL, as compared to all the possible different anodization conditions for our setup, can also be seen in Fig. 3a. With respect to the dotted line curve, obtained for anodization run with external cooling and without stirring, the dashed line curve beneath it was obtained in the same EL also without stirring yet at RT. This curve presents a quite constantV level (after the initial transients). Obviously the higher EL temperature allowed for maintaining a higher ionic mobility as well, and no local ion depletion at the anode occurred, different from the cooled EL condition, (Figs. 1a,2a dotted line). A similar effect of approximately constantV was observed when the anodization was run at RT and stirring was also activated, as shown by the continuous line curve beneath the dashed one. In this case, theV level was even lower, as probably ion exchange and transport was further eased by the mechanical agitation, which was added to the thermal one. As a drawback obviously some instabilities were generated in the system, which resulted in strong fluctuations ofV. The situation of both cooled and stirred 0.03 M oxalic acid EL is not shown, as it was not experimentally accessible in our setup, but we would also expect a similar situation of approximately constantV(t), with intermediateV level lying between the high (RT, stirring) and the low (cooling, no stirring) ion mobility conditions. TheV(t) curves for anodization with cooled EL and no stirring (i.e., low ion mobility condition) but with IL contentc = 0.5% are also reported in Fig. 3a (thick black line for IL2 and thick gray line for IL1, respectively). One can see that theV level at regime was constant also in the latter cases, but significantly lower than for all the no-IL containing EL conditions. This was obviously due to the increased electrical conductivity of the EL after injection of the IL ions.
From Fig. 3a, it is clear that by keeping V low via the IL additive one can run anodization at comparatively high i as compared to standard values reported in the literature for bare oxalic acid EL, and still operate in MA condition. This should allow for avoiding detrimental effects such as the barrier breakdown observed in Fig. 1a. Furthermore, in case of two-step anodization it would help to keep the conditions as close as possible to the V desired for optimal ordered APA growth, which for potentiostatic process in oxalic acid is in the 40–60 V range [5, 10, 17].
The resulting conductivity of the IL1-added EL was four times as much for the IL2-added EL. This could make one think that the performance of IL1 solutions in APA growth would be better. However, APA films were observed after anodizations run with IL1 solutions only for the lowest relative concentration values explored, namelyc = 0.01%, and with comparatively lowJ = 20 mA/cm2. In those conditions,h was quite low, as expected (hIL1 = 5–10 μm). For higherJ and/or higherc, no APA film was obtained at all, and on the contrary black pits were always observed on the Al substrate. Atc = 0.5% andJ = 600 mA/cm2, in particular, anodization resulted in complete dissolution of the anode during successive rinsing in DI-water. Obviously the result of the high oxygen content, associated with the quite high conductivity (i.e., ion mobility), makes the dominating effect of IL1 to be a heavy ion bombardment of Al, rather than a support to the flow of the EL ions.
For the IL that on the contrary demonstrated to provide APA after most preliminary test anodizations, namely IL2, we decided to systematically investigate the results of the processes run for different combinations ofc andJ parameters, in the rangec = 0.01–2% andJ = 20–600 mA/cm2. In particular, in Fig. 3b, a part of the (c,J) “phase” space for all the combinations of the valuesc = 0.1%, 0.5%, and 1.0% andJ = 100, 200, and 400 mA/cm2is shown, with respect to the resulting APA growth rate. It turns out that a local maximum ofvgwas found for the point (c,J) = (0.5%, 200 mA/cm2), with valuevgmax(IL2) = 1.1 μm/min. This growth rate is higher than for the same acidic EL with 10-fold higher concentration (see Fig. 1b) and more than three times higher than for the EL solution with the same acid concentration (V(t) dotted line curve in Fig. 3a,vg ~ 0.3 /min).
Obviously, some improvement due to IL2 ions occurs only at intermediate c and J values. For J, side effects can be imagined to negatively affect APA formation on excessive increase of this parameter, such as an extraordinary local EL heating, with a consequent loss of current efficiency. On the other hand, too many IL ions in solution can overwhelm the other EL ions, and decrease the current efficiency in turn. We can work out the molar ratio of the IL–oxalic acid species in the experimentally observed best condition of 0.5% v/v for IL2. The numbers of moles for each species can be calculated as nox = MoxVox, where Mox is the molarity and Vox the volume of oxalic acid, and nIL2 = mIL2/MWIL2, where mIL2 is the mass and MWIL2 the molecular weight of IL2, respectively. Therefore, the molar ratio is nox/nIL2 = MoxMWIL2/ρIL2c, with ρIL2 the mass density of IL2. Since it is MWIL2 = 226.03 g/mole and ρIL2 = 1.21 g/mL, for c = 0.5% it turns out nox/nIL2 ~ 1.1. Therefore, the best improvement in vg on addition of IL2 is obtained for a ~ 1:1 ratio of the IL2 moles with respect to the moles of the oxalic acid. When this ratio was increased of a factor two, it was not possible to grow APA any more even with IL2. On the contrary, anodization run with the same ratio, obtained for example by doubling both concentrations (Mox = 0.06 M and cIL2 = 1%), produced APA with consistent h values. The color of the respective outer APA surfaces was also quite similar, pale yellow in all cases, as usually observed due to inclusion of oxalate ions [5, 14].
The molar ratio for the same relative concentrationc = 0.5% in the case of the other IL can also be calculated. For IL1, being MWIL1 = 338.42 g/mole and ρIL1 = 1.19 g/mL, it turns outnox/nIL1 ~ 1.7. This value is of the same order of magnitude as for IL2. Anyway even forc = 0.1%, no APA was obtained in the case of IL1, such that this negative result can only be assigned to an inherent chemical difference between the interaction of the two ILs with the oxalic acid.
IL Aided Characterization and Processing
After using the IL2 as an additive in anodization, we have also tried to take advantage of its properties in the characterization of the system. In Fig. 2b, the liquid CAs θ measured on the APA surfaces described in Figs. 12a have been reported. The filled squares represent the CAs obtained with DI-water as the wetting phase. One can see that all the respective APA films look rather hydrophilic (θwater < 90°). Most samples showed quite similar values (θwater ~ 29°), whereas only the sample with smaller pores showed a significantly higher θwater. The reason is probably that for that sample the pores were too small to be filled by the water, and the drop was actually sitting on a mixed APA–air interface . In practical terms, the water “probe” did not allow for high enough resolution to sense the smallest APA pores. A similar resolution limit affected also the topographic measurements by AFM, as observed in the RMS roughness plot of Fig. 2a (see the previous subsection for discussion). In Fig. 2b, the open squares report instead the CAs obtained with a c = 10% v/v solution of IL2 in the DI-water wetting phase, θIL2. In this case, the CA values were all quite similar to each other, and lower than for the bare water, θIL2 ~ 16°. This means that the IL2 solution had higher wetting power than water, and could wet even the smallest APA pores. Actually a similar behavior is expected from any IL, which should work such as a highly polar solvent that can easily penetrate voids of a few nanometer diameter only . Therefore, IL2 can also be used successfully for this kind of characterization of the porous APA surface morphology.
The effect of addition of ILs into an oxalic acid aqueous solution commonly used for the fabrication of APA has been investigated. Two different ILs have been used for the first time as additives in this anodization process. By adding one of them, namely 1-butyl-3-methylimidazolium tetrafluoborate, in an approximately 1:1 molar ratio with the solution acid, and properly tuning the current density, we could obtain a growth rate of APA of 1.1 μm/min. This growth rate is comparable to the value normally obtained in the industrially applied HA conditions, but has been obtained in MA conditions in our case. Therefore, our process should make it possible to obtain thick APA layers in comparatively short times (order of few hours) and with ordered pore arrays also on the outer surface, after two-step anodization in the appropriateV range. The high-anodization current in itself does not guarantee fast APA growth, as demonstrated when the other IL was used as the EL additive. Therefore, a better understanding of the chemical mechanisms underlying the observed increase in growth rate has to be pursued, and is currently the subject of further research activity in our group. However, the presently reported preliminary results hold promise for the development of a technologically viable procedure for the fast growth of APA. In this application perspective, the possible use of ILs in characterization of the porous film and in its subsequent processing has also been explored. As a result, the selected IL has been demonstrated to be useful also as a pore wall protection medium during pore opening of APA, a process step that is often taken when passing membranes are fabricated out of the supported porous surfaces.
The authors would like to thank Mr. Romeo Losso for providing the original idea of using ionic liquids in the preparation of anodic porous alumina and for the recommendations, thereafter, and Mr. Claudio Larosa for technical support and useful discussions on the topic.