Home

Search

Collections

Journals

About

Contact us

My IOPscience

Universal, geometry-driven hydrophobic behaviour of bare metal nanowire clusters

This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2008 Nanotechnology 19 075709 (http://iopscience.iop.org/0957-4484/19/7/075709) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 161.122.101.135 The article was downloaded on 17/09/2010 at 01:40

Please note that terms and conditions apply.

IOP PUBLISHING Nanotechnology 19 (2008) 075709 (5pp)

NANOTECHNOLOGY doi:10.1088/0957-4484/19/7/075709

Universal, geometry-driven hydrophobic behaviour of bare metal nanowire clusters

Received 31 August 2007, in nal form 30 November 2007 Published 31 January 2008 Online at stacks.iop.org/Nano/19/075709 Abstract A parallel array of isolated metal nanowires is expected to be hydrophilic. We show, however, that a clustering of such nanowires brought about by vacuum drying produces a dual-scale roughness and confers a strongly hydrophobic property to the surface. The mean size of the nanowire clusters as well as the contact angle are both found to be related to the wire length, and the critical wire length above which the surface becomes hydrophobic is 10 m. Surface roughness is generally known to enhance water-repellent properties, but this is the rst report of roughness-induced hydrophobicity on a bare (uncoated) metallic surface.

1. Introduction
Techniques for making a surface hydrophobic (non-wetting) are of great technological importance [1, 2], with applications in glass coating, microuidics, pesticides and other areas [3, 4]. This has, in recent years, led to a study of the water repellent nature of certain naturally occurring surfaces [5, 6]. There are several reviews of the current status of this eld [7, 8]. Since hydrophobic surfaces do not favour the formation of a hydrogen bond network with water molecules, they orient themselves so as to minimize the area of contact with such surfaces, which are therefore not wetted by water. On the other hand, a typical metal such as copper allows the formation of hydrogen bonds and thus favours the wetting process. Previous attempts at conferring hydrophobic properties to metallic surfaces (such as copper) involved coating them with organic compounds [9] or chemically modifying the metal surface via complex formation [10]. The wetting property of a surface is characterized by the contact angle ( ) between the liquid meniscus and the solid surface (see gure 1, inset). According to convention, the liquid is said to wet the surface when < 90 and to not wet the surface if > 90 . For a smooth surface, the contact angle can be expressed in the form of Youngs equation: cos = (SV SL )/ , where the surface energies of the solidliquid, solidvapour and liquidvapour interfaces are given by SL , SV and , respectively. The nature of the surface roughness inuences the contact angle in different
1 Present address: Indian Institute of Technology Roorkee, India. 2 Author to whom any correspondence should be addressed.

ways. In the context of hydrophobicity, the following two types of surface microtexture have been studied in detail. In the rst case, the liquid is in contact with the surface at all points and the hydrophobicity is enhanced merely due to an increased solid/liquid interfacial area arising from surface roughness. In this case, the effective Young contact angle is described by the Wenzel equation: cos = r cos , where r (>1) is the solid roughness (i.e. ratio of true and apparent surface areas). In the second case, the surface contains voids and air traps into which the liquid is unable to penetrate, such that the single solid/liquid interface is replaced by a composite solid/liquid and liquid/vapour interface. The effective contact angle in this case is described by the CassieBaxter equation: cos = 1 + fS (cos + 1), where f S is the fraction of the solid actually in contact with the liquid. Generally, a pattern of tall, smooth pillars would behave like a Cassie surface, while a pattern of short, squat pillars would act like a Wenzel surface. The CassieBaxter model is usually more appropriate for distinctly hydrophobic solids, whereas the Wenzel model is more suitable for slightly hydrophobic or marginally hydrophilic surfaces ( 90 ). In this paper, we study a special type of microstructured surface consisting of a parallel array of bare (i.e. uncoated) metal nanowires growing out of a substrate of the same metal. We show that a particular type of self-organization of the metal nanowires occurs during controlled drying (in vacuum), provided the wires are taller than a certain critical height. Such a clustering of the nanowires makes the surface strongly hydrophobic (gure 1). Even though we present data mainly from a copper nanowire surface, we have observed similar
1
2008 IOP Publishing Ltd Printed in the UK

0957-4484/08/075709+05$30.00

Nanotechnology 19 (2008) 075709

P Bhattacharya et al

method. Droplets of distilled water (average volume 2 l) were placed on the surface using a syringe and the static contact angle was measured with a horizontal microscope equipped with a CCD camera. Images were processed and analysed using the ImageJ software [11, 12] in order to measure the angle of the liquidsolid interface.

3. Results
We observe that the free standing Cu nanowires, produced by completely etching out the template and drying in vacuum, exhibit a tendency to cluster due to the capillary force of the liquid acting between them. The nature and extent of clustering depends directly on the average length of the exposed parts of the nanowires. Figure 2 shows scanning electron microscope (SEM) images of the type of clustering observed on drying three nanowire samples with different average wire lengths (l ). It is clear that the shorter nanowires (e.g. l = 9 m) show hardly any clustering, while the longer ones (l = 17 m and 45 m) form well-dened, tapered clusters. We now show that well-clustered nanowire surfaces exhibit pronounced hydrophobic behaviour, while poorly clustered surfaces favour wetting. Column (iii) in gure 2 shows photographs of a 2 l water drop on the nanowire surfaces of different lengths. We clearly see that the surface of the 7 m long nanowire sample, in which there is almost no clustering, is relatively hydrophilic (row A). However, the nanowire samples with l = 17 m (row B) and 45 m (row C) show strong clustering tendencies and their surfaces are denitely hydrophobic. It is important to point out that, in samples in which the template is only partially etched, the nanowires are strongly anchored to the base and are not allowed to bend and form clusters: these samples therefore always exhibit very low contact angles. In order to obtain an idea about the dimensions of the nanowire clusters, we performed a 2D fast Fourier transform (FFT) of the SEM images shown in column (i) of gure 2. Figure 3 shows the radially averaged FFT intensity for all three samples. The broad peak observed in the scattering intensity at k = kp corresponds to a mean cluster size: m = 2/k p , where k is the wavevector associated with the clustering process. We nd that the mean cluster size m scales linearly with the length of the nanowires (gure 4). This is signicant in the context of an earlier study of hydrophobicity induced by clustering and cellular pattern formation in carbon nanotubes [13]. By balancing the mechanical bending energy and capillary forces it was predicted that l . The exponent was calculated to be 2 by Journet et al [14] and 1.5 by Zhao and Fan [15]. However, gure 4 clearly shows that the value of obtained from our observations is very close to 1, and the data cannot be tted to either = 1.5 or 2. Finally, we plot the measured contact angle against the length of the nanowires (gure 5) and show that the surface becomes progressively hydrophobic with an increase in the length of the nanowires (which, as shown earlier, leads to increased clustering). However, the value of the contact angle tends to saturate with wire length (above 35 m) and the maximum contact angle observed by us is 141 . 2

Figure 1. Photograph of a water drop on a free standing array of clustered copper nanowires. Inset shows a schematic diagram dening the contact angle ( ). (This gure is in colour only in the electronic version)

effects in other metal nanowire surfaces, such as those of silver and cobalt. Thus, the observed phenomenon appears to be controlled by the surface microtexture rather than the nature of the material. To our knowledge, this is the only report of a bare metallic surface showing appreciable hydrophobicity.

2. Experimental details
The copper nanowire arrays were electrochemically grown within commercially available porous anodic alumina (PAA) templates (Whatman Anodisk), with a diameter of 1.3 cm and a thickness of 60 m, containing parallel cylindrical pores with an average diameter of 200 10 nm. The PAA templates were back-plated with a 200 nm copper layer that served as a working electrode during electrodeposition. The nanowire array was electrodeposited potentiostatically in an acidic copper sulfate bath (pH = 2.0) using a conventional electrochemical cell with three-electrode geometry, comprising a working electrode (copper back-plate), a counter electrode (99.9% pure copper plate) and a reference electrode (standard calomel). The PAA template was subsequently removed in a basic etchant (1 M NaOH). The samples were rinsed gently but repeatedly in deionized water to remove traces of the etchant, and were then dried (under identical conditions) in an oil-free vacuum environment. The diameter ( D ) of the electrodeposited nanowires is dictated by the pore size of the PAA template (200 nm), while the wire length (l ) was varied by controlling the time of deposition. The mean spacing (d0 ) between the edges of two adjacent nanowires as grown in the PAA template is about 50 nm. During electrodeposition, a 50 m thick copper layer gets further deposited over the copper back-plate of the template, and this provides a rm substrate for the nanowires. The adhesion of the nanowires to the substrate is quite good, and even after completely etching out the PAA template, we observed no appreciable detachment of the nanowires from the substrate. The contact angle ( ) of water on the surface of the nanowire samples was measured using the sessile drop

Nanotechnology 19 (2008) 075709

P Bhattacharya et al

Figure 2. Scanning electron micrographs (SEM) of an array of exposed and vacuum-dried, free standing copper nanowires with 200 nm diameter. Columns (i) and (ii) show SEM images at different magnications from samples with different average lengths (l ) of 9, 17 and 45 m, arranged respectively in rows A, B and C. Column (iii) shows photographs of a 2 l water droplet sitting on the top surface of the three nanowire samples of different lengths. The hydrophobic nature of the surface can be seen to increase with the extent of clustering.

Figure 3. Radially averaged intensity of a 2D fast Fourier transform (FFT) of the SEM images shown in gure 2 (column (i)). The squares, circles and triangles represent FFT results for l = 9, 17 and 45 m, respectively. The peaks in the scattering intensity (marked by arrows) occur at k = kp and determine the mean sizes of the clusters, m = 2/kp .

Figure 4. Relation between the mean cluster size and the length of the nanowires, plotted on a loglog scale. The least squares t straight line passing through the data points has a slope that is very close to unity.

We now refer to the schematic diagrams in gure 6 to understand our results. Figure 6(a) shows a liquid surface supported on parallel nanowires separated by a distance 3

d0 , which determines the surface radius of curvature ( R ) of the supported liquid surface. The pressure difference across the liquidvapour interface is given by the Laplace equation: p = 2 /R . A decrease in the distance between the surface features decreases the radius of curvature and

Nanotechnology 19 (2008) 075709

P Bhattacharya et al

Figure 5. Relation between the measured contact angle of water on the nanowire surface and the mean length of the nanowires. The horizontal line at = 90 demarcates the hydrophilic and the hydrophobic regions. The crossover from one type of behaviour to the other occurs at l 10 m.

leads to an increase in the total pressure the structure can support. Figure 6(b) shows a schematic representation of the unclustered copper nanowires obtained by partial removal of the template from the sample. The presence of a part of the template prevents clustering of the nanowires and leads to a parallelly aligned array. In this case the distance between the nanowires (d0 50 nm) is such that the radius of curvature of the liquid surface on the metal wires is too large (i.e. p is too small) to support the liquid surface. This results in a total wetting of the surface. The entry of water into the crevices between the nanowires increases the effective contact area of the liquid with the metal surface and leads to a moderately small contact angle [7]. However, when the template is completely removed, the free standing nanowires can come together at the tips during etching and drying (see gure 2). Such nanowire clusters are schematically shown in gure 6(c). The clustering of the nanowires causes a reduction in the effective value of d0 , which leads to a reduction in R and an increase in p . Hence, water does not enter the crevices within each nanowire cluster, resulting in a large decrease in f S , the fraction of water in contact with the metal. This results in an increase in the effective contact angle. With an increase of the wire length and, consequently, the clustering tendency, there is, in effect, a crossover from the Wenzel scenario (gure 6(b)) to the CassieBaxter one (gure 6(c)). To substantiate our conjecture that the air trapped within the voids in the nanowire clusters leads to an enhancement of the hydrophobicity of the metal surface, we studied the effect of thermal cycling on the contact angle of water drops (2 l) placed on the metal nanowire surface. We rst lowered the temperature of the metal nanowire sample (with the water drop on it) by establishing thermal contact with a liquid nitrogen bath (at 100 K). The water drop on the surface freezes but the contact angle of the solidied drop remains greater than 90 , as seen in gure 7(A). The sample is then isolated from the low temperature bath and allowed to attain room temperature. 4

Figure 6. Schematic diagram showing (a) a liquid surface with a local radius of curvature R , supported on a nanowire array (CassieBaxter situation) in which the nanowires are separated by a distance d0 , (b) nanowire array geometry in which the water drop completely wets the surface (Wenzel situation) and (c) model for hydrophobic behaviour in the system of nanowire clusters of dimension .

While warming up, the solidied water drop melts and the contact angle gradually decreases to lower values (gures 7(B) and (C)). This may be ascribed to the condensation of moisture on the surface of the individual nanowires, which leads to the expulsion of the trapped air. This allows the external water drop to penetrate into the crevices within the nanowires while the system warms up to room temperature and leads to hydrophilic behaviour. When a bare metal surface (such as Cu) is brought in contact with the atmosphere, a layer of surface oxide is formed on it. What is the role of the surface oxide on the hydrophobic properties of the metal nanowires? At this stage, we have no direct evidence to either prove or negate the effect of the ubiquitous surface oxide. However, it is quite clear that such an oxide layer would be present both on the sharp and somewhat conically clustered structure (shown in rows B and C in gure 2) as well as on the comparatively less clustered nanowires shown in row A. Since the former structure is hydrophobic while the latter is hydrophilic, we may conclude that the observed water repellence is controlled by the prevalent geometry rather than the presence of a surface oxide layer. In summary, we have demonstrated that, under specic conditions, a bare (i.e. not deliberately coated or treated) metallic surface can exhibit hydrophobicity when the surface in question consists of clusters of nanowires. The magnitude of the water repellence depends on the nature of the clustering

Nanotechnology 19 (2008) 075709

P Bhattacharya et al

Figure 7. Photograph of (A) a water drop on a hydrophobic nanowire surface just after freezing. In (B) and (C), the surface shows progressively hydrophilic behaviour during the melting of the water droplet, due to condensation of moisture within the crevices of the nanowires.

of the nanowires, which in turn has been shown to depend on the wire length. We observed that, below a critical length (10 m), the clustering is either non-existent or random and the nanowire surface favours wetting. On increasing the length of the nanowires, the cluster size increases and the surface develops water-repellent properties. Apparent contact angles for water greater than 140 have been observed. We have shown that the observed hydrophobicity is a purely geometrical phenomenon related to the nature of clustering and is therefore not affected by the inherently hydrophilic nature of the metallic (copper) surface. We may therefore expect to observe this phenomenon for geometrically similar clustered nanowires surfaces of any metal. We nd that the cluster size ( ) scales linearly with the length ( ) of the nanowires. This observation cannot be understood from a simple model based on balancing the mechanical and capillary energies [14, 15] and requires further theoretical work.

References
[1] Feng X, Feng L, Jin M, Zhai J, Jiang L and Zhu D 2004 J. Am. Chem. Soc. 126 62 [2] Yang Y H, Li Z Y, Wang B, Wang C X, Chen D H and Yang G W 2005 J. Phys.: Condens. Matter 17 5441 [3] Patankar N 2003 Langmuir 19 1249 [4] He J L B and Patankar N 2003 Langmuir 19 4999 [5] Cheng Y and Rodak D E 2005 Appl. Phys. Lett. 86 144101 [6] Cheng Y, Rodak D E, Angelopoulos A and Gacek T 2005 Appl. Phys. Lett. 87 194112 [7] Callies M and Quere D 2005 Soft Matter 1 55 [8] Quere D 2002 Physica A 313 32 [9] Shirtcliffe N J, McHale G, Newton M I, Chabrol G and Perry C 2004 Adv. Mater. 16 1929 [10] Wang S, Feng L and Jiang L 2006 Adv. Mater. 18 767 [11] Rasband W 2007 ImageJ, US National Institutes of Health, Bethesda, Maryland, USA http://rsb.info.nih.gov/ij/ [12] Stalder A, Kulik G, Sage D, Barbieri L and Hoffmann P 2006 Colloids Surf. A 286 92 [13] Li W and Amirfazli A 2005 J. Colloid Interface Sci. 292 195 [14] Journet C, Moulinet S, Ybert C, Purcell S T and Bocquet L 2005 Europhys. Lett. 71 104 [15] Zhao Y-P and Fan J-G 2006 Appl. Phys. Lett. 88 3123

Acknowledgment
The authors are grateful to Ms Bhagyashree Chalke for her help with the SEM measurements.