On the mechanical properties of nanoporous anodized alumina by nanoindentation and sliding tests

On the mechanical properties of nanoporous anodized alumina by nanoindentation and sliding tests

Surface & Coatings Technology 206 (2012) 2115–2124 Contents lists available at SciVerse ScienceDirect Surface & Coatings Technology journal homepage...

3MB Sizes 0 Downloads 24 Views

Surface & Coatings Technology 206 (2012) 2115–2124

Contents lists available at SciVerse ScienceDirect

Surface & Coatings Technology journal homepage: www.elsevier.com/locate/surfcoat

On the mechanical properties of nanoporous anodized alumina by nanoindentation and sliding tests L. Vojkuvka a, A. Santos a, J. Pallarès a, J. Ferré-Borrull a, L.F. Marsal a,⁎, J.P. Celis b,⁎ a b

Departament d'Enginyeria Electrònica, Elèctrica i Automàtica, Universitat Rovira i Virgili, Avda. Països Catalans 26, 43007 Tarragona, Spain Department MTM, Katholieke Universiteit Leuven, Kasteelpark Arenberg 44, Heverlee B3110, Belgium

a r t i c l e

i n f o

Article history: Received 21 February 2011 Accepted in revised form 18 September 2011 Available online 24 September 2011 Keywords: Anodized aluminum Nanoindentation Scratch test Image analysis Plastic deformation Elastic behavior

a b s t r a c t Mechanical properties like hardness, Young's modulus, and plastic deformation of nanoporous anodized alumina fabricated in the most used acid electrolytes (i.e. sulphuric, oxalic and phosphoric) were extracted from nanoindentation and sliding tests. The dependence of the mechanical properties on the porosity level was analyzed for each type of samples. Nanohardness, Young's modulus and the coefficient of friction were related to the sample type and its porosity. Indents and scratches were inspected by FEG-SEM, which revealed that nanoporous anodized alumina undergo an elastoplastic behavior under the applied loads. The ductility is found to be higher for samples prepared in the phosphoric solution than for the ones obtained from oxalic and sulphuric solutions. This is associated with the high water content, the phosphate anions incorporated into the alumina structure during the anodization process and the aluminum nanopillar structure between those pores with irregular junctions located at the alumina–aluminum interface. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Since the development of the two-step anodization process to achieve self-ordered nanoporous anodized alumina (NAA) [1], the material science community has focused its interest on this material due to its potential applications as magnetic storage [2], photovoltaic solar cells [3], filters [4], chemical sensors [5], photonics [6], and metallic nanowires [7,8]. In addition, this synthesis process is relatively easy and cost-effective since no expensive laboratory equipment is required. Nanoporous anodized alumina obtained by a two-step anodization process consists of hexagonal and periodic pore arrangement caused by the self-ordering effect forming polydomains of 15 μm2, approximately. In applications where the mechanical interaction with other parts is expected, the mechanical properties of nanoporous anodized alumina must be known (e.g. micro/ nanoelectromechanical systems (M/NEMS), integrated biosensing/ analyzing nanodevices) [9]. In that respect, nanoindentation test is a commonly used technique to determine the mechanical properties of NAA [10]. In nanoindentation tests the investigated volumes of material are in the nanometer range. Usually, the indenter used for nanoindentation testing is a diamond having a Berkovich tip that has three-sided pyramid geometry. The tip penetration vs. the applied load is recorded. So, knowing the exact

⁎ Corresponding authors. E-mail addresses: [email protected] (L.F. Marsal), [email protected] (J.P. Celis). 0257-8972/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2011.09.040

indenter dimensions, some mechanical properties of the tested material as nanoindentation hardness (H) and Young's modulus (E) can be obtained [11]. The scratch test is another well-established technique used to characterize the superficial mechanical properties of thin films and coatings. In the scratch test, the applied normal force (Fn) and the tangential force (Ft) are measured and the friction coefficient (cf) of the tested material is calculated as c f = Ft / Fn . So far, only a few works report on the mechanical properties of nanoporous anodized alumina [12–20]. In this paper, we present an exhaustive and systematic study on the mechanical properties of nanoporous anodized alumina. To this end, multiple nanoindentation and scratch tests are performed. The indents and scratches on the NAA samples are inspected by field emission gun scanning electron microscopy (FEG-SEM). This makes it possible to relate the resulting morphology of the indents and scratches to the mechanical properties of each type of NAA. From this analysis, we have found out that the mechanical properties of nanoporous anodic alumina not only depend on the porosity level but also in the acid electrolyte used during the anodization process. To the best of our knowledge, this feature has not been reported previously and it could be very interesting for some applications such as electromechanical or biological nanodevices, in which special mechanical, physical and chemical characteristics (e.g. elasticity, flexibility, thermal stability, chemical inertness, electrical isolation, etc.) are required [9]. This paper is structured as follows: first, the experimental procedure to fabricate the NAA samples is described, second, the results extracted from characterization tests are presented, and finally, we discuss about these results and argument our conclusions.

2116

L. Vojkuvka et al. / Surface & Coatings Technology 206 (2012) 2115–2124

Table 1 Anodization conditions of the NAA samples studied (anodization voltage (V), type and concentration (C) of the acid electrolyte, temperature (T) and anodization time (tan)). Sample

I II III

First step

Second step

Acid

V (V)

C (M)

T (°C)

tan (h)

Acid

V (V)

C (M)

T (°C)

tan (h)

H2SO4 H2C2O4 H3PO4

20 40 170

1.0 0.3 0.2

1 17 1

1 1 1

H2SO4 H2C2O4 H3PO4

20 40 170

1.0 0.3 0.2

1 17 1

4 4 4

2. Materials and method 2.1. Anodization conditions Nanoporous anodized alumina samples were prepared by the two-step anodization process [1]. Three types of acid solutions were used: namely, 1.0 M sulphuric (H2SO4), 0.3 M oxalic (H2C2O4), and 0.2 M phosphoric (H3PO4) at applied voltages of 20, 40, and 170 V, respectively. The length of the second anodization step was selected to obtain a film thickness of about 25 μm on all samples. Then, the NAA samples were etched in an aqueous solution of 0.4 M H3PO4 at 35 °C to increase the sample porosity. Table 1 summarizes the anodization parameters used for the first and the second step. Both the pore widening time (tpw) after the second anodizing step and the calculated porosity (P) after etching, are shown in Table 2. Three different pore widening times were applied depending on the sample type: namely, i) tpw = 0, 5 and 10 min for samples prepared in the H2SO4 solution, ii) tpw = 0, 10, and 20 min for samples prepared in the H2C2O4 solution, and iii) tpw = 0, 20 and 40 min for samples prepared in the H3PO4 solution. NAA sample produced in sulphuric, oxalic or phosphoric acid was labeled as type I, type II, and type III, respectively. In addition, for each type of sample, three different degrees of porosity (i.e. pore diameters) were obtained by varying the length of pore widening time. In order to indicate the porosity level (i.e. pore diameter) of each type of sample, a subscript was included in their labels (i.e. low = small pore, medium = medium pore and high = large pore). 2.2. Preparation of nanoporous anodized alumina samples Before the two-step anodization process, commercial high purity aluminum (Al) foils (99.999%, Goodfellow Cambridge Ltd) were pretreated. The aluminum foils were degreased in acetone and cleaned in double-distilled water. Subsequently, they were annealed in nitrogen (N2) environment at 400 °C for 3 h to homogenize their crystalline phase and grain size. Then, the Al foils were electropolished in a 4:1 volume mixture of ethanol (EtOH) and perchloric acid (HClO4) at 20 V for 2 min at 6 °C under constant stirring. Finally, the Al foils were rinsed in double-distilled water, air dried, and stored in dry environment.

Table 2 Geometric characteristics of the NAA samples after FEG-SEM image analysis (dinterpore, dpore and P). Sample type I

II

III

dinterpore (nm) 60

105

455

tpw (min)

dpore (nm)

P (%)

Label

0 5 10 0 10 20 0 20 40

18 24 31 35 49 58 124 191 242

8 15 25 12 22 31 11 20 32

Ilow Imedium Ihigh IIlow IImedium IIhigh IIIlow IIImedium IIIhigh

The pre-treated Al substrates were anodized in a thermally isolated home-made electrochemical cell with a platinum wire as cathode and a copper plate as anode. The Al substrates were exposed to the acid electrolytes under stirring. The anodizing temperature was adjusted by a temperature controller from Polyscience (model 9106) during the whole process. The anodization voltage was controlled by a power supply from Delta Elecktronika (model SM 1500). The first anodization step consisted of applying directly the corresponding anodization voltage for 1 h. When the first anodization step was finished, the aluminum oxide (Al2O3) film was dissolved by wet chemical etching in a mixture of phosphoric acid (H3PO4) 0.4 M and chromic acid (H2Cr2O4) 0.2 M at 70 °C during 1 h. So, a pre-pattern was produced on the aluminum surface. Afterward, the second anodization step was carried out at the same anodization voltage maintained until the anodized layer reached a thickness of about 25 μm. In order to increase the porosity of the anodized samples, a pore widening was carried out by wet chemical etching in 0.4 M aqueous phosphoric acid at 35 °C. 2.3. Nanoindentation tests NAA samples of types I, II, and III were tested by nanoindentation tester (CSM-instruments equipped with a Berkovich tip). Two modes of measurement were applied: namely, a single indentation mode (SI) and a multicycled indentation mode (MI). In the SI mode, the applied loading force was varied from 1 to 250 mN, starting at F0 = 1 mN. At each indentation step, the force was increased accordingly to Fn = F(n − 1) + n (mN), where n = 1 to 24. Each indentation with its corresponding loading force (Fn) was repeated 5 times in a row, and each complete SI measurement matrix had 5 rows (i.e. 25 indentations). The MI mode was based on applying the maximum loading force of 250 mN in 10 separated load–unloading cycles. In each cycle, the loading force was gradually increased with steps of 25 mN, starting at F1 = 25 mN and finishing at F10 = 250 mN. In this way, each complete MI measurement consisted finally of 10 loading–unloading cycles repeated 5 times in a row. 2.4. Scratch tests Scratch tests on NAA samples with different porosities were performed using a multi specimen test machine (MUST tester from Falex Tribology, N.V. Belgium). The characteristics of the used scratching device (i.e. cantilevers and tips) are given in Table 3. A glass cantilever labeled G1 with a mounted Berkovich tip was used for applying forces (Fn) up to 10 mN. Steel cantilevers labeled S1 and S2 were employed to test at applied forces (Fn) from 20 up to 450 mN, since they are much stiffer than the glass cantilever. A cone tip on a steel cantilever labeled S3 was used for scratch tests at high loads. 2.5. Characterization Before nanoindentation and scratch tests, the geometric characteristics of the NAA samples consisting of the interpore distance (dinterpore) (i.e. center-to-center pore distance), the pore diameter (dpore), and the porosity (P) were estimated from images obtained Table 3 Characteristics of the cantilevers used for the scratch tests. Label

Cantilever

Tip

G1 S1 S2 S3

Glass Steel Steel Steel

Berkovich Berkovich Berkovich Cone

L. Vojkuvka et al. / Surface & Coatings Technology 206 (2012) 2115–2124

2117

by field emission gun scanning electron microscopy (FEG-SEM Philips XL 30) and environmental scanning electron microscopy (ESEM FEI Quanta 600) using a standard image processing package (ImageJ, public domain program developed at the RSB of the NIH, USA). The results obtained are summarized in Table 2. Gaussian fitting was used to calculate the average of each geometric characteristic, and the standard deviation was used as an estimation of the measurement dispersion. All the mechanically tested NAA samples (i.e. I, II and III) with different porosities were subsequently inspected by FEG-SEM to observe closely the indented areas and the scratches on the NAA surfaces. 3. Results 3.1. Nanoindentation tests analysis The representative load–unload curves obtained in the SI mode at a maximum applied force Fmax = 250 mN are recompiled in Fig. 1 for each NAA sample studied. Solid lines correspond to samples with low porosity (Ilow, IIlow, and IIIlow), dashed lines represent samples with medium porosity (Imedium, IImedium, and IIImedium), and dotted lines are samples with high porosity (Ihigh, IIhigh, and IIIhigh). At first glance and considering a very low applied force of 250 mN, we can see that the penetration depth increases with sample porosity in all the unloading curves for samples with low porosity (Ilow, IIlow, and IIIlow). However, the behavior of samples types I and II (Fig. 1a and b) differs from that of samples type III (Fig. 1c) as the porosity increases. For samples types I and II, the penetration depth increases clearly as porosity increases, but the shape of the unloading curve does almost not vary. However, for samples type III, there is a slight elastic shift as porosity increases, and the penetration depth increase is significantly shorter than for types I and II. Regarding the loading curves, samples types I and II behave in a different way than samples type III again. For samples types I and II, the loading curve bends as porosity increases, but this curve becomes almost less bent for samples type III with high porosity (IIIhigh). This means that the penetration of the tip into sample IIIhigh is rather linear. In addition, for samples with low (Ilow, IIlow and IIIlow) and medium (Imedium, IImedium and IIImedium) porosity, the load needed to reach the same penetration depth (i.e. 1000 nm) is approximately the same (i.e. 100 mN for samples with low porosity, and 75 mN for samples with medium porosity). Nevertheless, in the case of samples with high porosity, the load needed to achieve the same penetration depth (i.e. 1000 nm) is higher for sample IIIhigh (i.e. 55 mN) than for samples Ihigh (i.e. 25 mN) and IIhigh (i.e. 30 mN). FEG-SEM images of samples after nanoindentation are shown in Fig. 2. On samples Ilow, IIlow and IIIlow (Fig. 2a, d and g, respectively), the size of the indents follows the order IIIlow b IIlow b Ilow, being all rather comparable in size. If the porosity is increased in one degree (i.e. samples with medium porosity), the indent size increases according to the previous order (i.e. IIImedium b IImedium b Imedium). However, the difference in indent size between samples Imedium and IImedium is noticeably minor than between these and IIImedium (Fig. 2b, e and h). Finally, for samples with high porosity, the order of the indent sizes is maintained (i.e. IIIhigh b IIhigh b Ihigh), and the indent size on samples Ihigh and IIhigh becomes about twice wider compared to the indent size on sample IIIhigh (Fig. 2c, f and i). The representative load–unload curves in the MI mode for samples with medium porosity (i.e. Imedium, IImedium and IIImedium) are shown in Fig. 3. The curves obtained under MI mode with 10 cycles (i.e. solid lines) are compared to results extracted from the same samples tested under SI mode (i.e. dashed lines). As before, the sample type III (i.e. IIImedium) behaves differently from sample types I and II (i.e. Imedium and IImedium). Concretely, for load curves on samples Imedium (Fig. 3a) and IImedium (Fig. 3b), the

Fig. 1. Load–unload curves of the nanoindentation tests under single mode (SI) at the applied load of 250 mN for samples with low (solid lines), medium (dashed lines) and high (dotted lines) porosity. (a) Type I. (b) Type II. (c) Type III.

two SI and MI modes give a good fit up to the first and the second cycle, respectively. However, as the number of cycles increases on these samples, there is a slight shift of the load curve toward larger penetration depths for a given load. This shift increases as the number of cycles increases, what can be explained by a partial compression of the material remaining after each loading–unloading cycle. For samples type III (i.e. IIImedium), there is a good fit between the results obtained under SI and MI

2118

L. Vojkuvka et al. / Surface & Coatings Technology 206 (2012) 2115–2124

Fig. 2. Set of FEG-SEM images of the indents after nanoindentation tests for all samples and porosities. (a) Sample Ilow. (b) Sample Imedium. (c) Sample Ihigh. (d) Sample IIlow. (e) Sample IImedium. (f) Sample IIhigh. (g) Sample IIIlow. (h) Sample IIImedium. (i) Sample IIIhigh.

modes for all cycles, and the shift is practically unnoticeable along the entire load curve in contrast to Imedium and IImedium (Fig. 3c). Nevertheless, the unload curve of the last indentation cycle under MI mode on sample IIImedium slightly shifts toward shorter penetration depths. This differs from what is noticed on the unloading curves on Imedium and IImedium, which fit practically the results obtained under SI mode. This implies that IIImedium is less compressible than Imedium and IImedium. Hardness (H) and Young's modulus (E) of NAA samples were extracted from the nanoindentation tests using the software developed by the nanoindenter provider (i.e. CSM-instruments). This software uses the Oliver–Pharr method described in literature [10]. However, in line with previous studies [21], such a method was conceived for continuous solids and not for porous solids as NAA, since H is calculated by dividing the load by the real contact area of the indent. For this reason, actually, the hardness obtained from the nanoindentation software is the apparent hardness (Happ). In order to calculate the real hardness (Hreal) of NAA samples, it is necessary to find their real solid area (Areal). To this avail, FEG-SEM images were analyzed, and the real solid area was estimated (please see Appendix A). The average Areal, Happ, Hreal (calculated as −1 100·Happ·Areal (%)) and E for each sample and for both SI mode (performed indentations 5 × 25 = 125) and MI mode (performed indentations 10 × 5 = 50) were calculated.

For samples type I (i.e. Ilow, Imedium and Ihigh) the average Happ, Hreal, and E for SI mode and MI mode, are rather comparable. The difference between the average Happ, Hreal and E for SI mode and MI mode increases slightly for samples type II (i.e. IIlow, IImedium and IIhigh,) and that effect is more noticeable on samples type III (IIIlow, IIImedium and IIIhigh). In addition, the values of Happ, Hreal and E for all samples are equal or slightly higher for the SI mode than for the MI mode, what can be attributed to the compression of the porous material remaining after each load–unload cycle in MI mode. In order to study the influence of the porosity on the real hardness and Young's modulus, the average values of Hreal, and E were plotted versus the porosity values obtained previously (Table 2) in Fig. 4. The SI mode values are represented by solid lines while the MI mode values correspond to dotted lines. As a first result, it is observed that the real hardness (Hreal) decreases linearly as the porosity increases for samples I and II (Fig. 4a), and the trend of Hreal is practically the same under SI and MI modes. Nevertheless, for samples of type III, the dependence of Hreal on P is less linear under SI mode than under MI mode, and the values of Hreal under SI mode are fairly superior to for the ones recorded under MI mode when the porosity degree is low and medium. Concerning to the Young's modulus (E), it decreases linearly as porosity increases for samples types I and II for both SI and MI modes, as

L. Vojkuvka et al. / Surface & Coatings Technology 206 (2012) 2115–2124

2119

Fig. 4. Real hardness (Hreal) and Young's modulus (E) dependence on the porosity (P) for NAA samples types I, II and III. (a) Real hardness. (b) Young's modulus.

3.2. Scratch tests analysis

Fig. 3. Load–unload curves of the nanoindentation tests under single mode (SI-dashed lines) and multi-cycled mode (MI-solid lines) at the applied load of 250 mN for samples with medium porosity. (a) Imedium. (b) IImedium. (c) IIImedium.

shown in Fig. 4b. However, the behavior of samples type III differs from I and II. The evolution of E is clearly not linear under both SI and MI modes and the values of E (SI) for low and medium porosities are noticeably higher than the ones of E (MI). Another result that is worth noting is that, the Young's modulus of all the samples is rather comparable at any porosity level. Nonetheless, the values of Hreal for samples type III are generally higher than that for samples types I and II.

Scratch tests were performed on each type of sample (i.e. I, II and III) with different degrees of porosity (i.e. low, medium and high) using different cantilevers (Table 3). The scratch test results are shown in Fig. 5. The coefficient of friction recorded on NAA samples (cf) sliding against was calculated by dividing the tangential force (Ft) by the applied force (Fn). The soft glass cantilever (G1) was used to measure the coefficient of friction at applied forces (Fn) ranging from 1 to 10 mN. As Fig. 5a, b, and c shows (label G1), the cf varies abruptly between 0.07 and 0.20 on all types of samples and porosities. It is very difficult to establish differences between samples since they are very close and practically overlapped. The results obtained with the steel cantilever (S2) at Fn ranging from 20 to 100 mN, vary depending on the porosity degree, but the cf curve shape is similar on all samples. At low porosity (Fig. 5a label S2), the difference in cf is almost unnoticeable between all samples, and cf varies from 0.06 to 0.19. As porosity increases, the difference in cf is more noticeable, but the general shape of the curves is maintained (Fig. 5b label S2). Finally, at high porosity (Fig. 5c label S2), the coefficient of friction is higher on samples Ihigh (from 0.24 to 0.25) and IIhigh (from 0.21 to 0.24) than on sample IIIhigh (from 0.13 to 0.17). Finally, the results from scratch tests done with a cone tip mounted on a steel cantilever (S3) are shown in Fig. 5a, b, and c (label S3). Owing to the shape, dimensions and material properties of the cone tip (i.e. radius around 200 μm), the penetration depth of the cone tip (S3) into the NAA sample is much shorter than the penetration depth of the Berkovich tip (S1 and S2) at Fn ranging from 50 to 400 mN. The values of cf are rather lower than those obtained with the Berkovich tip (S1 and S2) on samples with medium and high porosity, but not on samples with low porosity (Fig. 5a label S3). In that last

2120

L. Vojkuvka et al. / Surface & Coatings Technology 206 (2012) 2115–2124

Fig. 6. Scratch width (Ws) dependence on the applied load (Fn) measured after the scratch tests using different steel cantilevers. (a) Results with the steel cantilever S1. (b) Results with the steel cantilever S2 (for low porosity samples only to 250 mN).

samples only to 250 mN), respectively. The measured width of the scratches increases linearly at increasing applied force on all the samples and all porosities. As the sample porosity increases, the scratch width increases with both steel cantilevers (S1 and S2). Nevertheless, this increase in Ws is more pronounced for samples types I and II than for samples type III. 3.3. Examination of nanoindents

Fig. 5. Friction coefficient (cf) dependence on the applied load (Fn) obtained after the scratch tests using different cantilevers and tips (G1, S1, S2 and S3). (a) Samples with low porosity. (b) Samples with medium porosity. (c) Samples with high porosity.

case, the curves overlapped to those corresponding to S1. In spite of this, the shape of the curves for S1 and S3 is practically the same as for Fn ranging from 50 to 400 mN. The trends on all these samples are practically the same, this is cf increases constantly between 50 and 300 mN, and remains almost constant between 300 and 400 mN. The scratch width (Ws) was measured after the friction tests with the steel cantilevers S1 (Fig. 6a) and S2 (Fig. 6b) at applied forces ranging from 50 to 400 mN, and from 20 to 100 mN (for low porosity

All the mechanically tested NAA samples (i.e. I, II and III) with low, medium, and high porosity, were inspected by FEG-SEM to derive the indent area. Fig. 7 shows a set of FEG-SEM images of indents made on the top side of samples IIlow, IImedium and IIhigh at an applied force of 250 mN. A cleavage is observed that increases in width as the sample porosity increases (i.e. 0.1 μm for IIlow, 1.0 μm for IImedium, and 4.5 μm for IIhigh). Similar results are obtained on samples types I and III. In all cases, no superficial cracks were observed at the indent edges or indent sides. The same cleavage appears inside the indents of samples type III with low porosity under an applied force of 250 mN (Fig. 8). The cleavage width increases as porosity increases, so that it is reasonable to assume that cleavage most probably occurs during the loading cycle of the nanoindentation test when the indenter tip penetrates into the samples. The shear forces between indenter tip and NAA are concentrated rather at the tip edges than at the tip sides. Therefore, the cleavages are also very uniform matching well the pyramidal

L. Vojkuvka et al. / Surface & Coatings Technology 206 (2012) 2115–2124

2121

Fig. 8. Set of top view FEG-SEM images of the indent under applied load of 250 mN on a sample type III with low porosity. (a) Before wet chemical etching. (b) After wet chemical etching.

3.4. Examination of scratches

Fig. 7. Set of top view FEG-SEM images of the indents on samples type II with different porosities under applied load of 250 mN. (a) IIlow. (b) IImedium. (c) IIhigh.

geometry of the tip. Fig. 8a shows an enlarged view of the indent on sample IIIlow. Compressed and deformed pores with irregular shapes (compared to the pores on the top surface of this sample) appear at the indent sides. A slight chemical etching in phosphoric acid solution of the same sample was carried out to analyze in detail the pore shapes at the indent sides (Fig. 8b). The deformed pore cells at the indent sides are distinguished by white borders, which are generated by stress accumulated between adjacent compressed pore cells after the nanoindentation process. During wet chemical etching, these borders of the compressed pore cells dissolve in a different way than the cell walls. In addition, it is observed that the cell deformation is much higher at indent cleavages than at indent sides. This also supports the previous assumption that shear forces are concentrated at the tip edges during the loading cycle.

A FEG-SEM image analysis was performed on scratches. Cracks were not observed in the scratched area on samples Ilow, IIlow and IIIlow. On the contrary, linear cracks are formed along the scratch side following the forward movement of the indenter tip throughout the surface of sample IImedium. The tip position and its movement direction are indicated in Fig. 9a. All cracks have the same angle with respect to the scratch bottom, which corresponds to the geometry of the pyramidal tip (i.e. half angle of the tip edges = 65°). It means that, during scratching, the cracks are simultaneously formed on both scratch sides along the scratch path. A magnified view of the crack area (see inset in Fig. 9a) reveals that the pore cells are deformed, and collapse following a direction parallel to the crack. This collapsing mode and the shape of the deformed pore cells are similar to the shear band formation observed in the stress simulations of large-scale porous metals and polymers with comparable hexagonal geometry [22,23]. FEG-SEM images of scratch area of sample IIhigh are shown in Fig. 9b. Besides the linear cracks present at the scratch sides, large circular cracks are generated at the scratch bottom. Those circular cracks can be produced by the excessive high load used during the scratch test (i.e. 400 mN). It is possible that the depth limitation was exceeded during the scratch test. For that reason, the compressed material is released behind the indenter tip forming circular cracks as the indenter tip is moved forward. A closer inspection of the scratch bottom revealed that the original porous structure is still preserved beneath the compressed material (see inset in Fig. 9b). Cracks were not observed after scratch tests on samples type III (i.e. IIIlow and IIImedium) at an applied force of 400 mN. FEG-SEM

2122

L. Vojkuvka et al. / Surface & Coatings Technology 206 (2012) 2115–2124

Fig. 9. Set of top view FEG-SEM images of the scratches performed under load of 400 mN for samples type II. (a) IImedium. (b) IIhigh. The insets are magnified views of the red squares on (a) and (b).

images of the scratch area on sample IIImedium are shown in Fig. 10a. A slight chemical wet etching on that sample was carried out in H3PO4 solution to observe the pore morphology after the scratch test (Fig. 10b). The scratch side structure before chemical etching appears to be almost completely non-porous. However, when the thin layer of non-porous Al2O3 which covers the tops of the pores is dissolved by chemical etching, the actual material structure emerges (Fig. 10b). The pore cells are collapsed and compacted all together in rows following the deformation direction of the indenter tip movement. The pore cell borders are clearly highlighted since, as was commented previously, the sites with accumulated stress present a different etching behavior. The formation of a thin layer of crushed Al2O3, which is spread over the sample surface during the scratch process, reveals that sample IIImedium is rather pliable. On analyzing the lateral view of the scratch (Fig. 10c), some linear cracks were identified on the scratch side, but these are less perceptible than on samples types I and II. The cracks present the same angle with respect to the scratch bottom (i.e. 65°). In addition, the penetration depth of the indenter tip is much shorter than on samples types I and II. No radial cracks at the scratch bottom were observed, and the cleavage on the scratch bottom was larger than on samples types I and II. 4. Discussion As we have shown throughout the Results section, generally, the mechanical behavior of samples anodized in phosphoric acid (type III) differs from those fabricated in sulphuric acid (type I) and oxalic acid (type II).

Fig. 10. Set of FEG-SEM images of the scratches performed under load of 400 mN for samples type III. (a) Top view of the scratched area of sample IIImedium before wet chemical etching. (b) Top view of the scratched area of sample IIImedium after wet chemical etching. (c) Lateral view of the scratched area of sample IIImedium.

From the nanoindentation tests, it is deduced that samples type III are more elastic than samples types I and II. This is reflected in both SI and MI nanoindentation modes. Samples type III recover relatively their shape after each load–unloading cycle. This behavior is more pronounced for tests in MI mode, because the loading force is not applied continuously and the material can relax between successive

L. Vojkuvka et al. / Surface & Coatings Technology 206 (2012) 2115–2124

indentation cycles, allowing the accumulated stress to release between successive cycles. After nanoindentation and scratch tests, the FEG-SEM image analysis confirms that the samples types I and II are rather brittle and stiff materials, but samples type III are more able to deform under nanoindentation and scratching without breaking up. This behavior can be due to several reasons detailed hereafter: i) Previous studies have verified that the walls between adjacent pores can be divided on two main regions [24,25]. The first one is a dark inner layer (pure alumina) and the other one is a bright outer layer (anion-contaminated alumina). The ratio of the inner to outer layer in disordered NAA depends on the acid electrolyte used for anodization and follows the order H2SO4 (0.05) b H2C2O4 (0.1) b H3PO4 (0.5). However, as Nielsch et al. reported [26], in ordered NAA the ratio of the inner to

2123

outer layer is the same for all acid electrolytes (i.e. 0.2). For that reason, owing to all samples have been fabricated under self-ordered regimens, it is not possible to relate the different mechanical behaviors of samples types I, II and III to the thickness of the anion-contaminated oxide layer. However, it is possible that the phosphate anions incorporated into the Al2O3 structure in the course of the anodization process confers more elasticity to the NAA fabricated in phosphoric than that produced in oxalic and sulphuric acids. ii) The water content in the NAA structure depends on the anodization conditions [24]. It is known that, during the anodization process, some voids are generated on the apexes of aluminum protrusions at Al2O3/Al interface [27,28]. This generation of voids takes place due to oxygen evolution and their size increases with increasing the anodization voltage. Those voids are preferential places for chemisorption of OH groups and

Fig. 11. Set of bottom view ESEM images of samples types I, II and III, (a) type I (H2SO4 — 20 V). (b) Type II (H2C2O4 — 40 V). (c) Type III (H3PO4 — 170 V) red circles indicate irregular junctions. (d) Type III (H3PO4 — 170 V) Al nanopillars on the Al pattern when the Al2O3 film is removed after the first anodization step. (e) Cross-section view diagram describing the possible effect of the Al nanopillar arrays on the mechanical properties during the nanoindentation tests (red arrows indicate the direction of forces).

2124

L. Vojkuvka et al. / Surface & Coatings Technology 206 (2012) 2115–2124

adsorption of water molecules from the acid electrolyte [29]. For this reason, the water content and OH groups in the NAA structure of those samples fabricated in phosphoric acid at 170 V is much higher than that of those produced in oxalic and sulphuric acids. All this would explain the behavior of samples III during the sliding tests. From the geometry point of view, taking into account that samples type III have the largest pore diameters, the movement of the indentation tip on these samples should be more rugged than for those samples types I and II. Nevertheless, the friction coefficients of samples type III are the lowest because the movement of the indentation tip is somehow cushioned during the sliding tests. This could be a result of the higher ductility of samples III conferred by the higher water and OH group content into their structure. iii) Finally, the presence of aluminum nanopillars between the pore cell walls at the bottom side of NAA samples could modify the mechanical properties of NAA [30]. As we can see in Fig. 11, for samples types I (Fig. 11a) and II (Fig. 11b) obtained at low anodization voltages of 20 and 40 V respectively, no irregular junctions between adjacent pore cells are observed. Nevertheless, for samples type III obtained at a high anodization voltage of 170 V, some irregular junctions connecting non-hexagonal pore cells are identified along the bottom side. These irregular junctions are generated by the high electric field concentrated at the pore bottoms, and give rise to holes. These holes are filled with Al during the anodization process (Fig. 11c and d). As a result, arrays of Al nanopillars grow throughout the NAA structure (i.e. between adjacent pores). This could modify the mechanical hardness of NAA at microscopic scale during the nanoindentation tests (Fig. 11e). For this reason, in spite of the Young's modulus of all the samples is rather comparable at any porosity level, the values of Hreal for samples type III are higher than that for samples types I and II. Another factor that could explain this higher hardness is that the pore size (i.e. pore diameter) of those samples fabricated in phosphoric is much larger than that of those samples produced in oxalic and sulphuric, which are rather comparable (Table 2).

Acknowledgments This work was supported by the Spanish Ministry of Education and Science (MEC) under grant number TEC2009-09551, CONSOLIDER HOPE project CSD2007-00007, AECID project A/024560/09 and Scientific Research Network on Surface Modification of Materials funded by the Flemish Science Foundation (FWO). Nanoindentation and scratch testing were performed at Dept. MTM-KULeuven with the help of Marc Peeters. Anodization was done at Universitat Rovira i Virgili. Appendix A. Real area calculations from FEG-SEM image analysis of samples types I, II and III Table A.1 Calculation of the number of pores (Np), the average area per pore (average Ap), the total porous area (Ap), the total area (At) and the real solid area (Areal) for all the samples and porosities (Ilow, Imedium, Ihigh, IIlow, IImedium, IIhigh, IIIlow, IIImedium and IIIhigh). (Both the number of pores and the average area per pore were estimated from FEGSEM image analysis). Sample

Np

Average Ap(nm2)

Ilow IIlow IIIlow Imedium IImedium IIImedium Ihigh IIhigh IIIhigh

222 478 ± 258 442 976 ± 168 105 12,548±4572 281 675 ± 265 437 1881 ± 330 158 29,319±8321 253 946 ± 273 444 2630 ± 304 97 46,419 ± 9214

Ap (nm2)

Ap (%)

At (nm2)

Areal (nm2)

Areal (%)

106,121 431,449 1,317,562 189,661 822,229 4,632,544 239,500 1,167,831 4,502,628

10.61 1,000,000 893,879 89.39 9.33 4,625,000 4,193,551 90.67 7.34 17,946,818 16,629,256 92.66 18.97 1,000,000 810,339 81.03 17.78 4,625,000 3,802,771 82.22 16.25 28,509,955 23,877,411 83.75 23.95 1,000,000 760,500 76.05 25.25 4,625,000 3,457,169 74.75 24.34 18,500,000 13,997,372 75.66

References [1] [2] [3] [4] [5] [6]

5. Conclusions

[7]

In summary, an extensive and systematic study on hardness, Young's modulus and friction coefficient of nanoporous anodic alumina has been discussed. The most commonly used types of NAA obtained from three different acid electrolytes (i.e. sulphuric (type I), oxalic (type II) and phosphoric (type III)) have been tested by nanoindentation and scratch testing. The effect of porosity has been analyzed by modifying this characteristic at three levels: namely, low, medium and high, for all types of samples. It has been demonstrated that the mechanical properties of NAA samples not only depend on the porosity level but also on the acid electrolyte used in the fabrication process. The mechanical behavior of samples types I and II are rather comparable and can be considered as elasticperfect plastic materials. However, samples type III show a more elastic behavior than those of types I and II which are rather brittle and stiff materials. The pore cells of samples type III deform under nanoindentation and scratch testing. This has been confirmed by FEG-SEM image analysis of indents and scratches. The origin of such a divergent mechanical behavior has been ascribed to the higher water content, the phosphate anions incorporated in the NAA structure and the intrusion of Al nanopillars between those adjacent pores with irregular junctions. This special behavior makes NAA fabricated in phosphoric acid a material with a unique characteristic suitable for some potential applications in nanotechnology (e.g. sensors, electromechanical and biological systems and so forth).

[8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30]

H. Masuda, K. Fukuda, Science 268 (1995) 1466. D. Al Mawlawi, N. Coombs, M. Moskovits, J. Appl. Phys. 70 (1991) 4421. N. Haberkorn, J.S. Gutmann, P. Theato, ACS Nano 3 (2009) 1415. G. Gorokh, A. Mozalev, D. Solovei, V. Khatko, E. Llobet, X. Correig, Electrochim. Acta 52 (2006) 1771. J.H. Holtz, S.A. Asher, Nature 389 (1997) 829. M. Deubel, G. von Freymann, M. Wegener, S. Pereira, K. Busch, C.M. Soukoulis, Nat. Mater. 3 (2004) 444. G. Sauer, G. Brehm, S. Schneider, K. Nielsch, R.B. Wehrspohn, J. Choi, H. Hofmeister, U. Gösele, J. Appl. Phys. 91 (2002) 3243. A. Santos, L. Vojkuvka, J. Pallarés, J. Ferré-Borrull, L.F. Marsal, Nanoscale Research Letters 4 (2009) 1021. F. Matsumoto, M. Harada, K. Nishio, H. Masuda, Adv. Mater. 17 (2005) 1609. S. Bueno, C. Baudin, J. Mater. Res. 21 (2006) 161. W.C. Oliver, G.M. Pharr, J. Mater. Res. 7 (1992) 1564. D. Choi, S. Kim, S. Lee, D. Kim, K. Lee, H. Park, W. Hwang, Nanotechnology 19 (2008) 145708. D. Choi, S. Lee, S. Kim, P. Lee, K. Lee, H. Park, W. Hwang, Scr. Mater. 58 (2008) 870. B.A. Latella, J. Mater. Sci. Lett. 19 (2000) 1127. B. Bhushan, S. Sundararajan, Acta Mater. 46 (1998) 3793. Z. Xia, L. Riester, B.W. Sheldon, W.A. Curtin, J. Liang, A. Yin, J.M. Xu, Rev. Adv. Mater. Sci. 6 (2004) 131. S. Ko, D. Lee, S. Jee, H. Park, K. Lee, W. Hwang, Thin Solid Films 515 (2006) 1932. T.H. Fang, T.H. Wang, C.H. Liu, L.W. Ji, S.H. Kang, Nanoscale Res. Lett. 2 (2007) 410. J.H. Jeon, W. Hwang, K.H. Lee, J. Compos. Mater. 43 (2009) 1155. K.Y. Ng, Y. Lin, A.H.W. Ngan, Acta Mater. 57 (2009) 2710. K. Zeng, C.-H. Chiu, Acta Mater. 49 (2001) 3539. S.D. Papka, S. Kyriakides, Int. J. Solids Struct. 35 (1998) 239. S.D. Papka, S. Kyriakides, Acta Mater. 46 (1998) 2765. A. Eftekhari (Ed.), Nanostructured Materials in Electrochemistry, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim, 2008, p. 1. G.E. Thompson, G.C. Wood, Nature 290 (1981) 230. K. Nielsch, J. Choi, K. Schwirn, R.B. Wehrspohn, U. Gösele, Nano Lett. 2 (2002) 677. S. Ono, H. Ichinose, N. Masuko, J. Electrochem. Soc. 138 (1991) 3705. S. Ono, H. Ichinose, N. Masuko, J. Electrochem. Soc. 139 (1992) L80. E. Palibroda, P. Marginean, Thin Solid Films 240 (1994) 73. W. Lee, K. Nielsch, U. Gösele, Nanotechnology 18 (2007) 475713.